Gri 2011 7244 PDF

ARISTOTLE UNIVERSITY OF THESSALONIKI
SCHOOL OF ENGINEERING – CHEMICAL ENGINEERING DEPARTMENT

SECTION OF UNIT OPERATIONS AND APPLIED THERMODYNAMICS
LABORATORY OF THERMOPHYSICAL PROPERTIES & ENVIRONMENTAL PROCESSES
MEASUREMENT OF
THE THERMAL CONDUCTIVITY OF
COMPOSITE SOLID MATERIALS
by
KONSTANTINOS D. ANTONIADIS
DIPL. CHEMICAL ENGINEER
Thesis submitted for the degree of
Doctor of Philosophy
in Chemical Engineering
THESSALONIKI 2011
ΚΟΝΣΤΑΝΤΙΝΟΣ Δ. ΑΝΤΩΝΙΑΔΗΣ
ΜΕΤΡΗΣΗ ΤΗΣ
ΘΕΡΜΙΚΗΣ ΑΓΩΓΙΜΟΤΗΤΑΣ
ΣΥΝΘΕΤΩΝ ΣΤΕΡΕΩΝ ΥΛΙΚΩΝ
ΔΙΔΑΚΤΟΡΙΚΗ ΔΙΑΤΡΙΒΗ
Υποβλήθηκε στο Τμήμα Χημικών Μηχανικών, Πολυτεχνική Σχολή,

Τομέας Τεχνικής των Φυσικών Διεργασιών
& Εφαρμοσμένης Θερμοδυναμικής
Ημερομηνία Προφορικής Εξέτασης: 13/04/2011
Εξεταστική Επιτροπή
Καθηγητής Μ.Ι. Ασσαέλ, Επιβλέπων

Καθηγητής Κ. Παναγιώτου, Μέλος Τριμελούς Συμβουλευτικής Επιτροπής
Αν. Καθηγητής Ε. Γ. Καστρινάκης, Μέλος Τριμελούς Συμβουλευτικής Επιτροπής
Καθηγητής Γ. Σακελλαρόπουλος, Εξεταστής

Καθηγητής Μ. Σταματούδης, Εξεταστής
Καθηγητής Π. Γιαννακουδάκης, Εξεταστής
Επικ. Καθηγητής, Ε. Βουτσάς, Εξεταστής
© Κωνσταντίνος Δ. Αντωνιάδης
© Α.Π.Θ.
Μέτρηση της Θερμικής Αγωγιμότητας Σύνθετων Στερεών
ISBN:
«Η έγκριση της παρούσης Διδακτορικής Διατριβής από το Τμήμα Χημικών

Μηχανικών της Πολυτεχνικής Σχολής του Αριστοτελείου Πανεπιστημίου
Θεσσαλονίκης δεν υποδηλώνει αποδοχή των γνωμών του συγγραφέως»
(Ν. 5343/1932, άρθρο 202, παρ. 2)
to my parents
Abstract
The scope of the present work was the performance of very low
uncertainty measurements of the thermal conductivity of solid and
composite materials to be employed in thermal conductivity reference
correlations. The technique employed for the measurements is the transient
hot–wire method, which is established as an accurate absolute technique for
thermal conductivity measurements of solids.
A new experimental device was designed and several refinements
were made in relation to the already existing one. More specifically, a new
silicone paste, of higher thermal conductivity was used in the sensor and
two stainless steel–spacers were employed to define exactly the distance
between the solid specimens that were being measured. Moreover, Comsol
Multiphysics finite element software was employed to accurately describe
the geometry of the sensor and reproduce the temporal temperature rise of
the hot wire from the beginning of the time scale. The refinements applied
to the hot–wire instrument allowed the thermal conductivity measurement
of solids with an uncertainty better than  1 %.
In details, in the first chapter, the importance of the knowledge of
material properties and the accurate measurement of them is pointed out.
Moreover, the international terminology for reference materials is analysed
and the major worldwide material certification institutions are presented.
In Chapter B, a brief overview of the most significant and widely used
methods for the measurement of the thermal conductivity of solids is
presented. The Fourier’s heat conduction law is presented and the
dominating equations are analysed. Following that, a brief description of the
fundamental principles of the available steady–state and transient methods
for measuring the thermal conductivity of solids is shown. A critical
ii Abstract
evaluation of the methods is made by presenting and discussing three

interlaboratory measurement programs for Pyroceram 9606. The chapter
ends with a detailed historical review, starting from 1780, of the
development of the transient hot–wire method.
The third chapter presents and analyses the fundamental theoretical
and computational background needed for performing measurements of
thermal conductivity with the transient hot–wire technique. Firstly, the
“ideal” theoretical model on which the transient hot–wire technique is based
on, and the equations for its analytical solution are given. In addition, the
most significant corrections that have to be applied to this “ideal” solution
are briefly mentioned. Continuously, the finite element method and its use
together with the transient hot–wire technique is presented. The
characteristics of Comsol Multiphysics finite element software are analysed,
and an evaluation model representing a simple geometry, i.e. the
experimental setup for measuring the thermal conductivity of liquids, is
created. The comparison of the simulation results with the analytical
solution of the model are in perfect agreement. Consequently, the finite–
element model created for representing the actual experimental setup for
solids is described analytically, and the procedure followed for obtaining the
thermal conductivity of both liquid and solid samples from experimental
data is shown in details. The chapter ends with a sensitivity analysis of the
effect that the finite–element model’s parameters have on the calculated
thermal conductivity value of solids.
In Chapter D the newly designed and constructed transient hot–wire
sensor is analytically described and its main improvements compared to the
older one are illustrated. A large section of the chapter is attributed to the
electronic components consisting the experimental setup. An historical
review of previously used Wheatstone bridge circuits is presented and the
current configuration together with the data acquisition system and the
working equations for calculating the experimental temperature rise of the
hot wire are analysed in detail. Moreover, there is a description of the
procedure followed to derive the temperature rise of the hot wire from the
acquired voltage response from the sensor. The calculated temperature rise
is therefore employed together with Comsol Multiphysics software to
calculate the thermal conductivity value of a material. Consequently, the
validation of the sensor is performed by using it for measuring the thermal
conductivity of toluene, which is a reference liquid. The last part of the
Abstract iii
chapter is dedicated to the analysis of the sources of uncertainty introduced

during the experiment and the overall uncertainty of the thermal
conductivity is calculated to be better than  0.5 % for liquids, and better
than  1 % for solids.
The following chapter is dedicated to the presentation of the thermal
conductivity measurements of solid and composite materials. Firstly, the
developed transient hot wire sensor is employed for the measurement of the
thermal conductivity of the silicone pastes used in the sensor. Following
that, the technique is used for measuring the thermal conductivity of the
reference materials Pyroceram 9606 (designated as glass ceramic BCR–724)
and Pyrex 7740 (designated as BCR–039) over a temperature range from
315 K to 440 K. Subsequently, the sensor is used for studying candidate
reference materials for thermal conductivity. The candidate glassy
materials measured from ambient temperature up to 440 K were
Polymethyl Methacrylate (PMMA) and Borosilicate Crown Glass (BK7). The
present thermal conductivity measurements are compared with the data of
other investigators, and in all cases it was found that the deviations are
within the mutual uncertainty of the measurements.
In the last part of Chapter E, the hot wire sensor is used for
measuring the enhancement of the thermal conductivity of an epoxy–resin
reinforced initially with plies of plain weave glass fabric, then by carbon
multi–walled nanotubes (C–MWNT), and finally with both these two
macroscopic and nanoscopic reinforcements at hand. The uncertainty of the
measurement increases to ±2 % due to inhomogeneities in the samples
preparation. The results reveal that for the epoxy reinforced with glass
fibres, with volume fraction of 28 %, the thermal conductivity increase was
27 % compared to plain epoxy resin. When reinforced with C–MWNTs the
maximum enhancement observed was about 20 % at a concentration of
1.2 % by weight of C–MWNT. Similarly, when reinforced with both the
C–MWNTs and glass fibres the maximum thermal conductivity
enhancement observed was about 60 % at a concentration of 1.2 % by weight
of C–MWNTs.
The present thesis ends with a summary of the innovations achieved
and present the future research challenges on the transient hot wire
technique.
Περίληψη
Σκοπός της παρούσας εργασίας ήταν η πραγματοποίηση μετρήσεων

θερμικής αγωγιμότητας στερεών και σύνθετων υλικών, πολύ χαμηλής
αβεβαιότητας, για χρήση σε συσχετίσεις αναφοράς θερμικής αγωγιμότητας. Η
μέθοδος που εφαρμόστηκε για τις μετρήσεις είναι η μέθοδος του
θερμαινόμενου σύρματος σε μη μόνιμη κατάσταση, η οποία καθιερώνεται
πλέον ως μια απόλυτη και ακριβής τεχνική για τη μέτρηση της θερμικής
αγωγιμότητας στερεών.
Στο πλαίσιο της διατριβής, σχεδιάστηκε μία νέα πειραματική διάταξη
και έγιναν αρκετές βελτιώσεις–τροποποιήσεις σε σχέση με την ήδη
υπάρχουσα διάταξη. Πιο συγκεκριμένα, στον αισθητήρα χρησιμοποιήθηκε μια
νέα σιλικόνη υψηλότερης θερμικής αγωγιμότητας, καθώς και δύο μεταλλικά
στηρίγματα από ανοξείδωτο χάλυβα για τον πλήρη καθορισμό του πάχους της.
Ακόμη, το λογισμικό πεπερασμένων στοιχείων Comsol Multiphysics
χρησιμοποιήθηκε για την ακριβή μοντελοποίηση της γεωμετρίας του
αισθητήρα και την πλήρη αναπαραγωγή της θερμοκρασιακής αύξησης του
θερμαινόμενου σύρματος από πραγματικό χρόνο μηδέν. Οι βελτιώσεις που
έγιναν στη διάταξη του θερμαινόμενου σύρματος σε μη μόνιμη κατάσταση
επιτρέπουν τη μέτρηση της θερμικής αγωγιμότητας στερεών με αβεβαιότητα
καλύτερη του  1%.
Αναλυτικά, στο πρώτο κεφάλαιο επισημαίνεται η σημασία που έχει η
γνώση της θερμικής αγωγιμότητας των υλικών και η αξία της μέτρησής της με
χαμηλή αβεβαιότητα. Ακόμη, αναλύεται η διεθνής ορολογία που
χρησιμοποιείται για τα πρότυπα υλικά αναφοράς και παρουσιάζονται οι
κυριότεροι διεθνείς φορείς πιστοποίησης πρότυπων υλικών.
Στο Κεφάλαιο Β γίνεται μία σύντομη επισκόπηση των πιο σημαντικών
και ευρέως χρησιμοποιούμενων μεθόδων για τη μέτρηση της θερμικής
vi Περίληψη
αγωγιμότητας στερεών. Παρουσιάζεται ο νόμος του Fourier για τη μεταφορά

θερμότητας με αγωγή και αναλύονται οι βασικές εξισώσεις. Στη συνέχεια
γίνεται μία σύντομη περιγραφή των θεμελιωδών αρχών των υπαρχόντων
μεθόδων μέτρησης θερμικής αγωγιμότητας στερεών υλικών σε μόνιμη και μη
μόνιμη κατάσταση. Ακόμη, γίνεται κριτική αξιολόγηση των παραπάνω
μεθόδων μέσω της παρουσίασης και του σχολιασμού τριών διεργαστηριακών
προγραμμάτων μέτρησης της θερμικής αγωγιμότητας του στερεού κεραμικού
υλικού Pyroceram 9606. Το κεφάλαιο ολοκληρώνεται με μια λεπτομερή
ιστορική αναδρομή από το 1780, της εξέλιξης της μεθόδου του θερμαινόμενου
σύρματος σε μη μόνιμη κατάσταση.
Στο τρίτο κεφάλαιο παρουσιάζεται και αναλύεται το θεωρητικό και
υπολογιστικό υπόβαθρο που απαιτείται για την πραγματοποίηση μετρήσεων
θερμικής αγωγιμότητας με τη μέθοδο του θερμαινόμενου σύρματος σε μη
μόνιμη κατάσταση. Αρχικά περιγράφεται το “ιδανικό” θεωρητικό μοντέλο στο
οποίο στηρίζεται η μέθοδος και δίνονται οι εξισώσεις για την αναλυτική του
επίλυση. Ακόμη, αναφέρονται εν συντομία οι πιο σημαντικές διορθώσεις που
πρέπει να εφαρμοστούν στην “ιδανική” αναλυτική λύση. Ακολούθως,
παρουσιάζεται η μέθοδος των πεπερασμένων στοιχείων και η ταυτόχρονη
χρήση της με τη τεχνική του θερμαινόμενου σύρματος σε μη μόνιμη
κατάσταση. Αναλύονται τα χαρακτηριστικά του λογισμικού πεπερασμένων
στοιχείων Comsol Multiphysics και κατασκευάζεται ένα μοντέλο
αξιολόγησης, το οποίο αντιπροσωπεύει τη γεωμετρία της πειραματικής
διάταξης στη περίπτωση μέτρησης της θερμικής αγωγιμότητας υγρών. Τα
αποτελέσματα της προσομοίωσης είναι σε πλήρη συμφωνία με αυτά που
προκύπτουν από την αναλυτική επίλυση των εξισώσεων του μοντέλου. Στη
συνέχεια περιγράφεται αναλυτικά η εφαρμογή του μοντέλου πεπερασμένων
στοιχείων στη νέα πειραματική διάταξη για τη μέτρηση στερεών και
αναλύεται η μεθοδολογία που ακολουθείται για τον υπολογισμό της τιμής της
θερμικής αγωγιμότητας από τα πειραματικά δεδομένα, τόσο στη περίπτωση
μέτρησης υγρών όσο και στη περίπτωση μέτρησης στερεών. Το κεφάλαιο
τελειώνει με την ανάλυση ευαισθησίας της επίδρασης που έχουν οι
παράμετροι του μοντέλου πεπερασμένων στοιχείων στην υπολογιζόμενη τιμή
θερμικής αγωγιμότητας των στερεών.
Στο επόμενο κεφάλαιο δίνεται η λεπτομερής περιγραφή της νέας
πειραματικής διάταξης και αναλύονται οι βελτιώσεις–τροποποιήσεις που
πραγματοποιήθηκαν σε σχέση με την προηγούμενη υπάρχουσα διάταξη. Ένα
μεγάλο μέρος του κεφαλαίου είναι αφιερωμένο στην ηλεκτρονική διάταξη που
Περίληψη vii
χρησιμοποιείται στις μετρήσεις. Γίνεται ιστορική ανασκόπηση των

ηλεκτρονικών διατάξεων που έχουν χρησιμοποιηθεί μέχρι σήμερα και
παρουσιάζεται το κύκλωμα της παρούσας αυτοματοποιημένης γέφυρας τύπου
Wheatstone, καθώς και οι εξισώσεις που χρησιμοποιούνται για τον
υπολογισμό της πειραματικής θερμοκρασιακής αύξησης του θερμαινόμενου
σύρματος. Επιπρόσθετα, στο κεφάλαιο περιέχεται και η μεθοδολογία που
ακολουθείται για τον υπολογισμό της πειραματικής θερμοκρασιακής αύξησης
του σύρματος από τη μετρούμενη μεταβολή της τάσης στα άκρα του. Η
πειραματική αυτή θερμοκρασιακή αύξηση χρησιμοποιείται σε συνδυασμό με
το λογισμικό Comsol Multipyhysics για τον υπολογισμό της θερμικής
αγωγιμότητας του υλικού. Ακολούθως, ο έλεγχος της ορθής λειτουργίας του
νέου αισθητήρα πραγματοποιείται με τη μέτρηση της θερμικής αγωγιμότητας
τολουολίου, το οποίο είναι ένα πρότυπο ρευστό. Το τελευταίο τμήμα του
κεφαλαίου είναι αφιερωμένο στην ανάλυση των πηγών αβεβαιότητας που
υπεισέρχονται κατά τη διάρκεια του πειράματος και η συνολική αβεβαιότητα
των μετρήσεων της θερμικής αγωγιμότητας είναι καλύτερη από  0.5 % για τα
υγρά και καλύτερη από  1 % για τα στερεά.
Το Κεφάλαιο Ε είναι αφιερωμένο στην παρουσίαση των μετρήσεων
θερμικής αγωγιμότητας στερεών και σύνθετων υλικών. Αρχικά ο νέος
αισθητήρας θερμαινόμενου σύρματος σε μη μόνιμη κατάσταση
χρησιμοποιείται για τη μέτρηση της θερμικής αγωγιμότητας των σιλικόνων
που χρησιμοποιήθηκαν στον αισθητήρα. Έπειτα, η τεχνική χρησιμοποιείται
για τη μέτρηση της θερμικής αγωγιμότητας των πιστοποιημένων πρότυπων
υλικών αναφοράς θερμικής αγωγιμότητας Pyroceram 9606 (γνωστό με την
ονομασία BCR-724) και του Pyrex 7740 (γνωστό με την ονομασία BCR-039)
στη θερμοκρασιακή περιοχή από 315 Κ έως 440 Κ. Στη συνέχεια, ο
αισθητήρας χρησιμοποιείται για τη μέτρηση υποψήφιων υλικών αναφοράς
θερμικής αγωγιμότητας. Τα υποψήφια υαλώδη υλικά που μετρήθηκαν από
θερμοκρασία περιβάλλοντος μέχρι τους 440 Κ είναι το Polymethyl
Methacrylate (PMMA) και το Borosilicate Crown Glass (BK7). Τα
αποτελέσματα των μετρήσεων θερμικής αγωγιμότητας που έγιναν στην
παρούσα εργασία συγκρίθηκαν με δεδομένα άλλων ερευνητών και σε όλες τις
περιπτώσεις οι αποκλίσεις ήταν μέσα στα αμοιβαία όρια αβεβαιότητας των
χρησιμοποιούμενων μεθόδων.
Στο τελευταίο τμήμα του Κεφαλαίου Ε, η συσκευή θερμαινόμενου
σύρματος σε μη μόνιμη κατάσταση χρησιμοποιείται για τη μέτρηση της
αύξησης της θερμικής αγωγιμότητας μίας εποξικής ρητίνης ενισχυμένη
viii Περίληψη
αρχικά με ύφασμα από ίνες γυαλιού, έπειτα με πολυστρωματικούς

νανοσωλήνες άνθρακα (C–MWNTs), και τέλος και με τις δύο αυτές
μακροσκοπικές και νανοσκοπικές ενισχύσεις μαζί. Η αβεβαιότητα των
μετρήσεων αυξάνεται στο ±2 % λόγω ανομοιογενειών που υπάρχουν στα
δείγματα που παρασκευάζονται. Τα αποτελέσματα δείχνουν ότι η εποξική
ρητίνη ενισχυμένη με ίνες γυαλιού σε ποσοστό 28 % κατ’ όγκο έχει θερμική
αγωγιμότητα 27 % μεγαλύτερη από αυτή της ρητίνης. Ακόμη, όταν η ρητίνη
ενισχύεται με C–MWNTs η παρατηρούμενη αύξηση της θερμικής
αγωγιμότητας είναι 20 % για συγκέντρωση C–MWNTs 1.2 % κατά βάρος.
Ομοίως, όταν η ρητίνη ενισχύεται και με τα δύο, ίνες γυαλιού και
C–MWNTs, η αύξηση στη θερμική αγωγιμότητα είναι περίπου 60 % για
συγκέντρωση C–MWNTs 1.2 % κατά βάρος.
Η παρούσα διατριβή ολοκληρώνεται με μία σύνοψη των καινοτομιών
που επιτεύχθηκαν και παρουσιάζονται οι μελλοντικές ερευνητικές
προκλήσεις πάνω στη μέθοδο του θερμαινόμενου σύρματος σε μη μόνιμη
κατάσταση.
Acknowledgements
Behind each postgraduate student, there are always few persons that
contribute to the successful completion of the research. I would like to
express my deepest thanks to my supervisor, Professor Marc J. Assael, for
his guidance, cooperation and kind encouragement throughout the course of
my postgraduate research. His advices were vital for solving problems and
surpassing obstacles that appeared during my research. His experience and
expertise on the field of thermophysical properties essentially contributed
on the completion of my thesis. Especially, I would like to thank him for his
friendship and the confidence that he has showed on me all these years.
I would like also to express my thanks to Dr. Ulf Hammerschmidt, for
hosting me for three months in his laboratory at the Physikalisch–
Technische Budesanstalt (PTB), in Braunschweig, Germany. His help and
cooperation were valuable all along my staying there. He gave me the
opportunity to get in contact with several thermal conductivity instruments
of his laboratory, and especially to operate the Transient Hot Bridge sensor.
Moreover, I would like to thank Dipl.-Ing. Vladislav Meier from PTB for his
precious guidance and collaboration on the operating aspects of the
Transient Hot Bridge sensor and for the endless hours we have spent on
productive discussions over it.
Moreover, I would like to thank the members of my supervising
committee, Professor Costas Panayiotou and Associate Professor Eleftherios
Kastrinakis, from the Chemical Engineering Department of Aristotle
University of Thessaloniki (A.U.Th.), for their cooperation and help all along
my PhD project.
In addition, I would like to thank the Assistant Professor Eleni
Pavlidou from the Physics Department of A.U.Th., for the measurement of
x Acknowledgements
the tantalum wire diameter with Scanning Electron Microscope, and Dr.
Dimitrios Tzetzis for his essential help in the preparation of the reinforced
epoxy–resin specimens.
Furthermore, I would like to mention the valuable help provided on
construction issues by Mr. Vassileios Goutsios and Mr. Triantafyllos
Tsilipiras from the Chemical Engineering Department of A.U.Th. Moreover,
I would like to thank Mr. Michael Mprintakis for his precious advices on
electronic matters.
I also cannot miss out to thank my friends and colleagues within the
Laboratory of Thermophysical Properties and Environmental Processes for
their cooperation and the productive discussions that we had. In particular,
I want to thank Dr. Katerina E. Gialou and Dr. Ifigeneia N. Metaxa for their
help on my first steps in the use of the transient hot wire technique, and Dr.
Konstantinos E. Kakosimos for his guidance on the use of Comsol
Multiphysics software and his help on computer’s software issues.
I want also to thank the undergraduate students that I had the
pleasure to work with during their diploma thesis, Mr. Gregory
Mantziaroglou and Mr. Vasileios Charalampidis.
Last but not least, I would like to thank my family and all my friends
for their constant encouragement all these years. Especially, I want to
deeply thank my parents for their multilateral support in my decisions and
for consisting the most important motive for my personal development. A
special thank to Vana for her patience and love that she showed and keeps
showing to me.
Marie Curie has quoted: “I am among those who think that science
has great beauty. A scientist in his laboratory is not only a technician: he is
also a child placed before natural phenomena which impress him like a fairy
tale”. My personal tale is written in the following pages.
Contents
Abstract ____________________________________________________________ i
Acknowledgements _________________________________________________ ix
A Introduction _______________________________________________________ 1
A1. The Need for Material Properties ______________________________________ 2

A1.1. The Importance of Accurate Measurements __________________________________ 3
A2. Reference Materials __________________________________________________ 5
A3. Composite Materials __________________________________________________ 8

A3.1. Effective Thermal Conductivity ____________________________________________ 9
A4. Scope ______________________________________________________________ 10
B Methods for Measuring the Thermal Conductivity ___________________ 11
B1. Other methods ______________________________________________________ 12

B1.1. Steady State Methods ____________________________________________________ 13
B1.1.1. Guarded Hot Plate Method ___________________________________________ 14
B1.1.2. Heat Flow Method ___________________________________________________ 16
B1.1.3. Calorimeter Method __________________________________________________ 18
B1.1.4. 3ω Method __________________________________________________________ 20
B1.2. Transient Methods _______________________________________________________ 22
B1.2.1. Transient Hot Wire technique _________________________________________ 23
B1.2.2. Transient Plane Source Method _______________________________________ 23
xii Contents
B1.2.3. Transient Hot Strip Method ___________________________________________ 25

B1.2.4. Pulse Transient Method _______________________________________________ 26
B1.2.5. Transient Hot-Bridge Method _________________________________________ 27
B1.2.6. Dynamic Radial Heat Flow Method ____________________________________ 28
B1.2.7. Laser Flash Technique ________________________________________________ 30
B1.3. Critical Evaluation _______________________________________________________ 31
B1.3.1. Round-Robin Tests ___________________________________________________ 32
B1.3.2. Conclusion ___________________________________________________________ 36
B2. Transient Hot Wire Technique ________________________________________ 38

B2.1. Historical Development ___________________________________________________ 39
B2.1.1. The period from 1780 to 1888 __________________________________________ 39
B2.1.2. The period from 1888 to 1971 __________________________________________ 43
B2.1.3. The period from 1971 to today _________________________________________ 47
B2.1.4. Brief Presentation ____________________________________________________ 53
B3. Summary ___________________________________________________________ 55
C Theoretical ______________________________________________________ 57
C1. Ideal Model _________________________________________________________ 58

C1.1. Correction to the Ideal Model ______________________________________________ 60
C2. Finite–Element Analysis _____________________________________________ 65

C2.1. Fortran Finite Element Model _____________________________________________ 67
C2.2. Comsol Multiphysics Software _____________________________________________ 70
C2.2.1. Evaluation of the Heat Transfer Model _________________________________ 72
C2.2.2. Finite Element Model for Solids________________________________________ 80
C2.3. Sensitivity Analysis of the Finite Element Model ____________________________ 90
C3. Summary ___________________________________________________________ 95
D Experimental Procedure __________________________________________ 97
D1. Experimental Configuration __________________________________________ 98

D1.1. Sensor Design ____________________________________________________________ 99
D1.1.1. Sensor Assembly/Construction ________________________________________103
D1.1.2. Temperature Coefficient of Resistance _________________________________104
Contents xiii
D1.2. Temperature Controller _________________________________________________ 106
D2. Electronic Components _____________________________________________ 107

D2.1. Historical Development __________________________________________________ 108
D2.2. Present Bridge Circuit and Data Acquisition System _______________________ 111
D3. Working Equations _________________________________________________ 116

D3.1. The Bridge Equations ___________________________________________________ 116
D3.1.1. Constant Resistance Fraction [RLo(0)/RSo(0)] _________________________ 121
D3.1.2. Calculation of Initial Resistances’ Values at zero time __________________ 122
D3.2. The Temperature Rise Equations _________________________________________ 124
D3.3. The Heat Flux Equations ________________________________________________ 126
D4. Calculation of the Thermal Conductivity Value ________________________ 127
D5. Validation of the Technique _________________________________________ 132

D5.1. Measurements of Toluene ________________________________________________ 132
D6. Uncertainty Analysis _______________________________________________ 136
D7. Summary __________________________________________________________ 139
E Thermal Conductivity Measurements ______________________________ 141
E1. Solid Materials _____________________________________________________ 142

E1.1. Silicone Pastes __________________________________________________________ 142
E1.1.1. Thermal Conductivity Measurements of Silicone Paste AS1803 _________ 143
E1.1.2. Thermal Conductivity Measurements of Silicone Paste BORO, Type 650 _ 145
E1.2. Pyroceram 9606_________________________________________________________ 147
E1.2.1. Thermal Conductivity Measurements _________________________________ 148
E1.3. Pyrex 7740 _____________________________________________________________ 156
E1.4. Polymethyl Methacrylate (PMMA) ________________________________________ 164
E1.5. Borosilicate Crown Glass BK7 ____________________________________________ 172
E2. Composite Materials ________________________________________________ 179

E2.1. Epoxy Resin ____________________________________________________________ 180
xiv Contents
E2.1.1. Samples Preparation ________________________________________________180

E2.1.2. Thermal Conductivity Measurements _________________________________180
E2.2. Epoxy Resin Reinforced with Glass Fibers _________________________________ 182
E2.3. Epoxy Resin Reinforced with Carbon Multi-Walled Nanotubes _______________ 185
E2.4. Epoxy Resin Reinforced with Glass Fibres and Carbon Multi-Walled Nanotubes189
E3. Summary __________________________________________________________ 193
F Innovation – Future Work ________________________________________ 195
F1. Innovations ________________________________________________________ 196
F2. Future Work _______________________________________________________ 197
References ________________________________________________________ 199
List of Publications ________________________________________________ 219

A
Introduction
The chapter constitutes a brief introduction on the thermal

conductivity. In the beginning of the chapter the need for knowledge of the
material properties is described while the importance that accurate
measurements plays in many aspects of daily life is shown. Consequently,
international terms employed for defining reference materials are explained
and the major suppliers of reference materials are mentioned. The chapter
ends with the presentation of the scope of this thesis.
2 Introduction
A1. The Need for Material Properties

The measurements of material properties take place in many
occasions of the daily lives, e.g. during weighting of food and raw materials,
measuring the time to go from one place to other, monitoring the air quality
and ambient temperature, e.t.c. Almost, everything needs the performance
of a measurement in order to be accurately described. A measurement can
range from huge (distance between earth and sun) to tiny (nanoparticles
detected by microscopy).
Lord Kelvin in 1889 quoted [Thompson, 1889]: “When you can
measure what you are speaking about and express it in numbers you know
something about it; but when you cannot measure it, when you cannot
express it in numbers, your knowledge is of a meagre and unsatisfactory
kind: It may be the beginning of knowledge, but you have scarcely, in your
thoughts, advanced to the stage of science, whatever the matter may be”.
The measurement of material properties has a major impact on the
well being of people, the security of national infrastructure and the
effectiveness of world trade, as without accurate testing there cannot be any
guarantee of the ability to deliver the specified products to meet all the
individual needs. Moreover, governments require accurate and reliable data
in order to create regulations and laws that will protect the public health
and welfare, as well as the environment.
Furthermore, the material properties are crucial for the optimum
design of industrial equipment or of a simple product. Nowadays, the
production of new products includes extensive design and computational
modelling, in order to reassure that it will have the desired properties. This
procedure is facilitated by the continuous development on computer
software’s capabilities. The design consists in predicting the behaviour of
the product, during and after the manufacture process, by solving
mathematical models. However, in order to get a realistic model of the
product, the precise material properties (at specified conditions) should be
included in the models.
A recent survey in the United Kingdom investigated the need for
material property data on engineering materials. The survey covered all
material types, all applications and industrial sectors, and all sizes of
companies [Bennet & Sims, 2010]. In Figure A.1, the sources of material
data are shown. It is notable that no source dominates over the others but
Introduction 3
on the contrary there are numerous and varied sources.
*MSDS: Material Safety Data Sheet

Figure A.1 Sources of material data [Bennet & Sims, 2010].
It should be noted that the thermal properties of materials are among

the most important material properties governing industrial heat transport
processes. More specific, the thermal conductivity and thermal diffusivity
are the main thermal properties in the optimum design of industrial
equipment used for heat transfer applications. Moreover, the
thermophysical properties of a material determine its use, i.e materials with
low thermal conductivity can be used for insulation purposes (in heat
exchanger for avoiding heat losses), whereas materials with high thermal
conductivity, such as metals, can be employed in electronic applications.
A1.1. The Importance of Accurate Measurements

The available materials’ property data should be characterised by low
uncertainty in order to be useful. In the same survey mentioned before
4 Introduction
[Bennet & Sims, 2010], all the participants declared that the required data
in all phases, from design to manufacture of the final product, is expected to
have an accuracy of better than 10 %, while 50 % of the companies that
participated in the survey declared that they need data with an accuracyof
lower than 5 %. However, these requirements are far away from the already
available data, and therefore it is necessary the newly available data to rely
on best metrology practice, that will reassure the high accuracy demanded
by industry.
From an economic point of view, the accurate measurement of
material properties can have a considerable effect on the production
efficiency of a company and on the price of the final product. The lack of
quality material data makes a company liable to accidents due to the
destruction of the equipment (equipment breakdown due to fatigue,
corrosion, operation at extreme conditions, et.c.), or/and to the production of
defective products. Thus, the company reduce its production capacity in
order to avoid the destruction of its equipment. This company’s inefficiency
to fully exploit its equipment capacity has impact on the final price of the
product, which is higher. As an example, it is mentioned that the economic
loss in Japan due to accidents and defects of industrial products, is
estimated to be as high as 4 % its gross domestic product (GDP) [Bennet &
Sims, 2010].
Therefore, the need for accurate measurements is evident and an
international metrology system is crucial for ameliorating the quality of the
provided material data. This worldwide metrology network is established
through the National Metrology Institutes of each country. The central
coordination of the network is undertaken by the Bureau International des
Poids et Mesures (BIPM) in Sevres, Cedex France.
In order to be comparable across borders and over time,
measurements need be traceable to appropriate and stated references.
Therefore, there are certified reference materials that enable metrological
traceability, allow the calibration of test equipment and ensure the
confidence in measuring and testing data. The metrological traceability is
defined in the International Vocabulary of Metrology as “the property of a
measurement result whereby the result can be related to a stated reference
through a documented unbroken chain of calibrations, each contributing to
the measurement uncertainty” [International Organization for
Standardization, 2007].
Introduction 5
A2. Reference Materials

According to the ISO Guide 35: 2006 the following definitions are
given [International Organization for Standardization, 2006]:
a) Reference Material (RM)

It is a material, sufficiently homogeneous and stable with respect to
one or more specified properties, which has been established to be fit for its
intended use in a measurement process. The properties can be quantitative
or qualitative, e.g. identity of substances or species. Moreover, uses may
include the calibration of a measurement system, assessment of a
measurement procedure, assigning values to other materials, and quality
control. However, a single RM cannot be used for both calibration and
validation of results in the same measurement procedure.
b) Certified Reference Material (CRM)

It is a reference material characterized by a metrologically valid
procedure for one or more specified properties, accompanied by a certificate
that provides the value of the specified property, its associated uncertainty,
and a statement of metrological traceability.
Except from the above two terms, there are two more defined by the
National Institute of Standards and Technology (NIST) [National Institute
of Standards and Technology, 2000]:
c) Standard Reference Material (SRM)

It is a CRM issued by NIST that also meets additional NIST–specified
certification criteria. NIST SRMs are issued with Certificates of Analysis or
Certificates that report the results of their characterizations and provide
information regarding the appropriate use(s) of the material.
d) NIST Traceable Reference Material (NTRM)

It is a commercially produced reference material with a well–defined
traceability linkage to existing NIST standards. This traceability linkage is
established via criteria and protocols defined by NIST.
The ISO Committee on Reference Materials (ISO/REMCO) has

pusblished a series of Guides related to the reference materials (Table A.1).
6 Introduction
The objective of this committee is to carry out and encourage a broad

international effort for the harmonization and promotion of CRMs, their
production and application.
Table A.1 ISO reference material guides

[International Organization for Standardization, 1997, 2000a, 2000b, 2006, 2009].
Guide Title
Terms and definitions used in connection with reference
ISO Guide 30
materials
ISO Guide 31 Reference materials – Contents of certificates and labels
Terms and definitions used in connection with reference
ISO Guide 32
materials
ISO Guide 33 Uses of certified reference materials
General requirements for the competence of reference material
ISO Guide 34
producers
Reference materials – General and statistical principles for
ISO Guide 35
certification
Reference materials with certified material properties are

commercially available from several sources. The most known suppliers are:
i. The Institute for Reference Materials and Measurements (IRMM)

IRMM was found in 1960, and its mission is to promote a common and
reliable European measurement system in support of EU policies.
IRMM offers over 600 certified reference materials for applications in
the fields of food and feed analysis, environmental analysis, engineering
and health. IRMM reference materials laboratory is one of the largest
and most sophisticated in the world. It hosts a unique multi–functional
and flexible production laboratory, with cryo–grinding and freeze drying
equipment, high purity milling and levitation melting devices. The
Reference Materials unit of IRMM was the first European reference
material producer to obtain accreditation according to ISO Guide 34
[International Organization for Standardization, 2009].
Today, IRMM has teamed up with other two European reference
material producers, Bundesanstalt für Materialforschung und prüfung
(BAM) and LGC (UK), in a European Reference Material (ERM®)
partnership. The ERM concept guarantee to use the most advanced
principles currently available, described in ISO Guides 34 and 35
[International Organization for Standardization, 2006, 2009], for the
Introduction 7
production of certified reference materials. Moreover, they assure to

demonstrate rigorously homogeneity and stability for all materials and
guarantee the certified value for every single unit over the complete
shelf life of the materials. In other words, the ERM is a trademark for
certified reference materials, similar to the SRM trademark used by
NIST.
ii. The National Institute of Standards and Technology (NIST)

NIST, founded in 1901, is a non–regulatory federal agency within the
U.S. Department of Commerce. It supplies industry, academia,
government and other users with over 1300 Standard Reference
Materials of the highest quality and metrological value. These materials
are used to perform instrument calibrations in units as part of overall
quality assurance programs, to verify the accuracy of specific
measurements and to support the development of new measurement
methods.
iii. The Korea Research Institute of Standards and Science (KRISS)

KRISS was founded in 1975 and its mission is to promote the industrial
competitiveness of Korea by advancing measurement standards,
science, and technologies in ways that enhance the nation’s economic
performance and secure a better quality of life for all. It supplies
domestic industry with more than 500 Certified Reference Materials.
Concluding, the importance of the Certified Reference Materials

should be pointed out once more. CRMs consist an essential part of the
international metrology science. They are used in the calibration of
instruments, establishing traceability of measurement results, method
development, method validation and quality control of measurements and
laboratories. Without the use of CRMs it would not be possible to compare
material data measurements acquired with different measuring techniques.
Although there are CRMs for every property, there is still a long way to go
until having materials with certified properties for every possible measuring
conditions. Moreover, there is still work to do in order to ameliorate the
uncertainty of the provided certified material properties.
As far as thermal conductivity of solid material is concerned, there
are already two CRMs, Pyroceram 9606 and Pyrex 7740. Thermal
conductivity measurements of these two materials are presented in Chapter
8 Introduction
E, along with measurements of two candidate reference materials,

Polymethyl Methacrylate (PMMA) and Borosilicate Crown Glass BK7.
A3. Composite Materials

In recent years, there have been an increasing number of applications
requiring more efficient and lightweight thermal management such as
communication satellites, high–density electronics, and advanced aircraft.
The fibre–reinforced composites are considered as ideal candidates for many
of these applications and knowledge of the mechanical, as well as physical
properties, pertinent to these materials, has become a very crucial design
input.
Moreover, fibre–reinforced composites are commonly used as a
replacement for metals, in cases where they have better mechanical
properties. Composites have the advantage of higher corrosion resistance
and being lighter than metals. However, one significant drawback compared
to metals is the anisotropic low thermal conductivity, which in some
applications may prove to be significant, e.g in the case of electronic
applications. Generally, the thermal conductivity of polymer composites is
about 10 to 100 times lower than the thermal conductivity of metals. In
order to overcome this hindrance, carbon nanotubes are employed as
reinforcement of the fibre–reinforced composites. Carbon nanotubes have an
exceptionally high axial thermal conductivity and therefore contribute to the
increase of the effective thermal conductivity of the composite. The
production of nanocomposites represent a new frontier in material science,
as the reinforcement scale has change from micrometers, with traditional
carbon and glass fibres, to nanometers.
Despite the fact that there is a plethora of established standards
and/or procedures for the measurement of various mechanical properties of
composites, the same does not apply to the measurement of a number of
physical properties, including the effective thermal conductivity. As a result,
the lack of reliable thermophysical data may hinder the full utilization of
composites. It should be mentioned, that the thermal conductivities of a
fibre–reinforced composites can be calculated from various theoretical,
empirical, as well as numerical methods. However, each of these methods
have certain assumptions applicable for certain specific cases and condition
Introduction 9
ranges [Hasselman et al., 1980; Agari et al., 1991; Gorthala et al., 1994;
Ramani & Vaidyanathan, 1995; Amazouz & Gauvin, 1997; Ning & Chou,
1998; Pantano & Averill, 2002]. Thus, the determination of reliable
experimental data for validation purposes is of paramount importance.
Nowadays, the use of nanoparticles for the polymer matrix
reinforcement becomes a highly desirable procedure in order to enhance
both the physical and the mechanical properties of the composite [Paradise
& Goswami, 2007; Zhou et al., 2008; Wang & Qiu, 2010]. It should be noted
that the numerical and the analytical methods and assumptions regarding
the calculation of the effective thermal conductivity, developed for
conventional composites, have to be reanalysed in order to be applied for
nanocomposites. Clearly, such analyses also necessitate very accurate and
reliable experimental data for comparison.
A3.1. Effective Thermal Conductivity

It should be noted that the term “thermal conductivity” is referred
only for homogeneous materials. However, it is conventional to speak of the
“thermal conductivity” of various types of composite materials, glass–fibre
insulation, carbon–fibre composites, or polymer blends. In the case of these
materials, the thermal conductivity is taken to be the empirical constant of
proportionality in the linear relationship between a measured heat
transport per unit area and the temperature difference over a prescribed
distance in the material. The “thermal conductivity” is not then, strictly, a
property of the material, since it can often depend on a large number of
parameters, including the history of the material, its method of
manufacture, and even the character of its surface [Wakeham & Assael,
2000]. In other words, the measured property is the “effective thermal
conductivity” of the material.
Hence, when performing thermal conductivity measurements, it is
important always to keep in mind the distinction between homogeneous and
inhomogeneous materials and the terms used in reporting the results.
Therefore, in the present work the term “effective thermal conductivity”
should strictly be employed, when it is used for measurements of reinforced
polymers. However, for simplicity, the term “thermal conductivity” is also
used for the measurements of composite materials, meaning though the
10 Introduction
effective thermal conductivity of the specimens.
A4. Scope
It is well known that there are significant differences in the measured
thermal conductivities values of a solid material obtained by different
techniques. Therefore, the aim of the present work is to measure the
thermal conductivity of solid materials and composite polymers with very
low uncertainty. The measuring method employed for the measurement of
the thermal conductivity of the specimens is the transient hot wire
technique, which is described in details in the following two chapters. The
experimental device designed and employed here consist an improvement of
the already available transient hot–wire instrument for the measurement of
the thermal conductivity of solids of Assael et al. [Assael et al., 2002]. The
refinements made on the new improved thermal conductivity sensor
confront the drawbacks of the older one and allowed the performance of low
uncertainty measurements. More specifically the lower uncertainty is
achieved by:
i. The use of stainless–steel spacers to define exactly the distance

between the two solid samples.
ii. The use of a new silicone paste of higher thermal conductivity.
iii. The use of Comsol Multiphysics FEM software to accurately describe
the geometry of the sensor (for measurements from 20 μs to 10 s).
The developed transient hot wire sensor is employed for the low
uncertainty measurement of the thermal conductivity of Pyroceram 9606
and Pyrex 7740, which are certified reference materials. Moreover, the
thermal conductivity of two candidate reference materials, Polymethyl
methacrylate (PMMA) and Borosilicate Crown Glass BK7, is also measured.
The last part of the work is devoted to employment of the transient
hot wire method in the measurement of the thermal conductivity of
composite materials. Specifically, the hot wire sensor is used for measuring
the enhancement of the thermal conductivity of an epoxy–resin reinforced
initially with plies of plain weave glass fabric, then by carbon multi–walled
nanotubes (C–MWNT), and finally with both these two macroscopic and
nanoscopic reinforcements at hand.
11 Methods for Measuring the Thermal Conductivity
B
Methods for Measuring
the Thermal Conductivity
In this chapter, a brief overview of the most significant and widely

used methods for the measurement of the thermal conductivity of solids is
presented. The chapter begins by presenting the Fourier’s heat conduction
law. Consequently, the thermal conductivity measurement methods are
divided into two main sections, the steady state methods and the transient
ones, and a brief description and the fundamental principles of each method
are given. Following this, a critical evaluation of the methods is made by
presenting and discussing three interlaboratory measurement programs for
Pyroceram 9606. The last section of the chapter is focused on the transient
hot wire technique, which is established as an accurate technique for
thermal conductivity measurements of solids. As this technique is used in
this thesis, a detailed historical review of its development, starting from
1780, is presented and the main modifications of the technique are
described.
B1. Other methods

The heat transfer through a material can take place with three
different mechanisms: conduction, convection and radiation. Heat
conduction is the transfer of thermal energy in a molecular level, which is
caused by the combination of the molecules’ vibrations in a lattice, their
collisions during their random movement (in case of fluids) and the energy
transfer by free electrons. On the other side, convective heat transfer, which
occurs only on gases and liquids, include the heat transfer due to the
circulation of currents from one region to another, while radiation is the
heat transferred by the energy emitted or absorbed by solids, liquids or
gases, in the form of electromagnetic waves as a result of their temperature.
In order to perform an accurate thermal conductivity measurement
the heat energy should be transferred only by the conduction mechanism,
and therefore the experimental setup should ensure conditions that are in
accordance with the above mentioned requirement. However, in the case of
heat transfer in solids, which is the research issue of the present work,
there is no convection and the effect of radiation can be considered
negligible for small measurement times, and thus the heat transfer depends
mainly on the thermal conductivity of the material. Specifically, the thermal
conductivity, λ, is an intrinsic physical property of a material and expresses
the facility with which heat is propagated in passing from one internal
molecule to another [Fourier, 1822], whereas thermal diffusivity, α,
indicates how fast heat is conducted.
The differential conductive heat transfer equation for a viscous,
isotropic, and incompressible material, can be obtained from the energy
conservation equation [Wakeham et al., 1991]:
 DT
Q  g h =  C P  , (B.1)
Dt
where ρ and C P represents the material’s density and heat capacity


respectively, Q denotes the heat flux vector, g h is the internal energy
increase due to viscous dissipation, and the notation D Dt represents a
substantive derivative.
The first term of equation (B.1) refers to heat transferred by
molecular means, the second one to the heat generation and the right hand
Methods for Measuring the Thermal Conductivity 13
term refers to the change of the internal energy of the material. A general
solution of (B.1) is not possible and therefore some restrictions need to be
applied. Thus, the produced temperature gradients should be small so as to
have a near-equilibrium state, while material (in case fluids and gases)
movement should be avoided so that g h  0 . Under these conditions, the
substantive derivative is replaced by the partial derivative.
Moreover, the heat flux vector, for conductive heat transfer in an
isotropic material, is defined by the Fourier’s law equation [Fourier, 1822]:

Q     T , (B.2)
where T is the temperature gradient.

Therefore, combining equations (B.1) and (B.2), and using the above
analysed constraints it is
T
 C p  =    2T , (B.3)
t
which is the fundamental equation for the calculation of the thermal

conductivity of a material.
The developed experimental techniques for the measurement of the
thermal conductivity are divided into two broad categories:
- the steady state methods, mainly based on Fourier’s law, and
- the transient methods, mainly based on equation (B.3).
In the following sections, each category is analysed in detailed for the case of
solid sample measurements.
B1.1. Steady State Methods

The steady state methods are based on the measurement of the heat
flux necessary to maintain a temperature difference, constant in time,
between two surfaces of a measured solid sample. The thermal conductivity
is calculated as a function of the heat flux, the temperature gradient, and
the properties and geometry of the sample. In general, the working
equations of the steady state methods are much simpler than the one that
describe the transient methods. In the case of solids, the difficulties in the
application of a steady state technique derive from a need of extremely high
experimental times and samples of large dimensions. Many steady state

methods can be applied to solid, such as:
i. Guarded Hot Plate method iv. 3 omega (3ω) method

ii. Heat Flow method v. Comparative technique
iii. Calorimeter method vi. Longitudinal Heat Flow technique .
In the following subsections the fundamental principles of the main

steady state methods for the measurement of the thermal conductivity of
solids will be briefly presented. It is obvious that there are steady state
methods that cannot be applied to solids and therefore are not mentioned in
this work.
B1.1.1. Guarded Hot Plate Method

Guarded hot plate is a versatile and widely used steady state method
for measuring the thermal conductivity of solids. The guarded hot plate
method is covered by ASTM test method C177 [American Society for Testing
and Materials, 1997] and ISO 830 [International Organization for
Standardization, 1996], and many instruments based on this method are
commercially available. This method is applicable to the measurements of a
wide variety of specimens, ranging from opaque solids to porous or
transparent materials, and a widely range of environmental conditions
[Filla, 1997]. Moreover, the guarded hot plate method can be used in
apparatus operated with either vertical or horizontal heat flow.
A typical measurement arrangement for determining the thermal
conductivity according to ASTM C177 is shown in Figure B.1. The main
components of this arrangement are two isothermal cold surface assemblies
and a guarded hot plate. The guarded hot plate is composed of a metered
section thermally isolated from a concentric primary guard by a definite
separation gap. This arrangement uses two almost identical specimens of
the under measurement material and the hot plate is sandwiched between
them. Electrical energy, which is measured accurately, is dissipated through
the guarded hot plate and causes heat flow in the top and bottom sample.
The primary guard is used to reduce the radial heat flow, while a
temperature controlled secondary guard, in the form of edge insulation, is
necessary in order to restrict heat losses from the outer edge of the primary
guard. The secondary guard can be omitted in systems designed for ambient
conditions. In some guarded hot plate apparatus [Sheffield & Schorr, 1991]
optional auxiliary heaters are used between the specimens and the cold
surface assembly in order to control the sample temperature, independently
of the heat flux or the temperature gradient that the hot plate causes. The
cold surfaces assemblies, which are adjusted to the same temperature, are
isothermal heat sinks used for removing the energy generated by the
heating units. The temperature gradient within the specimen is usually
measured with thermocouples.
Figure B.1 Typical Guarded Hot Plate apparatus arrangement

[American Society for Testing and Materials, 1997]
Under steady state conditions, Fourier’s equation (B.2) for the above
geometry can be written:
Q   T   T 
A         , (B.4)
 Hot Plate  x top  x bottom
where Q and A denote the heat flow and the total metered surface of the hot
plate respectively, and ΔT and Δx represent the temperature drop across the
specimens (top or bottom) and their thickness. Moreover, the final
calculated thermal conductivity, λ, of the samples is the average of the two
pieces and thus it is important to use nearly identical specimens.
Usually deviations from the above described configuration can be
caused by samples’ inhomogeneities, or temperature differences between the
metered section and the guard, or temperature differences between the

outer edge of the assembly and the surrounding controlled environment.
Other errors affecting the experimental measurements are the edge heat
flow at the periphery of the specimens, and the heat flow across the gap
between the hot plate and the primary guard. Moreover, it was noted that
the uncertainty of a guarded hot plate apparatus depends on the thickness
of the specimen being measured [Salmon, 2001].
Filla [Filla, 1997] made a typical application of the method for the
measurement of the thermal conductivity of ceramics with an estimated
uncertainty of  5 %, whereas Salmon et al. in their review [Salmon, 2009],
claimed that the uncertainty level of the method is better than  3 %, for
measurements of thermal insulations in the region of 253 K to 353 K.
B1.1.2. Heat Flow Method

The heat flow method provides a rapid mean of determining the
steady-state transmission properties of thermal insulations and other
materials. There is the ASTM C518 test method that describes the use and
construction of a heat flow meter apparatus [American Society for Testing
and Materials, 2010]. The principle of the technique is that known heat
input is applied to the one end of the sample and removed at the other,
through a heat sink. This method is very similar to the guarded hot plate
method and their only difference is that the temperature drop along the
sample is measured by thermocouples immersed in the specimens and not
placed in the heating plates. The main difference between the heat flow
meter instrument and the hot plate apparatus is that in the first one the
thermal conductivity value depends on the heat flux reference standards,
while the second one operates in an absolute way. Therefore, the
uncertainty of the measurements made with ASTM C518 test method can
not be better than those achieved with the guarded hot plate.
In general, heat flow meter apparatus consist of one or two
specimens, two isothermal plate assemblies, one or more heat flux
transducers, and equipment to control the environmental conditions (if
needed). The technique can be divided into two categories according to the
geometry of the used instrument, i.e. radial and axial heat flow method.
Salmon and Tye made an intercomparison of seventeen heat flow meter
apparatus used for measuring the thermal conductivity of thermal

insulation materials at 283 K and 296 K [Salmon & Tye, 2000]. It shown
that in a total of 154 data points, 69 % were within  3 % and 50 % were
within  2 % of the values acquired with the guarded hot plate apparatus of
the National Physical Laboratory, at the United Kingdom.
a) Radial Heat Flow Method

In the radial heat flow method, a steady current of electricity is
passed through a long cylindrical bar of the material, establishing a zone of
uniform axial temperature. A radial temperature gradient is created by
allowing the surface of the sample to radiate to a cooled environment. The
thermal conductivity can then be obtained from the measurement of the
energy dissipation at the chosen portion of the bar and the temperature
difference between the axis and the surface [Angell, 1911; Powell &
Schofield, 1939]. The primary concern for accurate thermal conductivity
measurements with this method is to eliminate the axial heat flow. For a
resistance-heated cylindrical solid specimen the radial heat flow is
[Carslaw & Jaeger, 1959]:
1 d  dT  q o
 r  0, (B.5)
r dr  dr  
where qo is heat generation rate, λ the specimen’s thermal conductivity, T

the temperature and r the radius. For independent λ along the specimen
radius, the solution of equation (B.5) is
qo  r 2
 , (B.6)
4  Ta  T 
where Ta denotes the temperature on the axis of the cylinder, and T the
temperature at distance r.
A representative application of the method was done by Donaldson
and Taylor [Donaldson & Taylor, 1975] for the measurement of the thermal
diffusivity of Armco iron and the values found were within 5 % of the
accepted ones.
b) Axial Heat Flow Method .

The axial heat flow method was firstly described in a series of papers
in 1954 [Jain & Krishnan, 1954a, 1954c, 1954d, 1954b] and since then used
by many researchers for the measurement of molten and solid metals
[Goldratt & Greenfield, 1978; Hemminger, 1989]. In these papers, the
theoretical and the experimental part of the method was developed and
applied to a thin circular rod of tungsten and Acheson graphite. In general,
the thermal conductivity is computed from the axial temperature gradient
and certain parameters. The differential equation, in steady state, for the
thin rod, and neglecting the effects of the radial temperature gradient, can
be written [Carslaw & Jaeger, 1959]:
2T p     4 I2 
x 2

 
T  T0
4
 
  2
0, (B.7)
where T is the temperature at distance x of the under consideration point,

To is the temperature of the walls of the chamber and therefore the
temperature of the ends of the rod, p is the perimeter of the rod’s cross-
section, ρ its specific resistance. Moreover, ε is the total emissivity of the
surface, ω is the cross-sectional area of the rod, σ is Stefan’s constant of
radiation, I is the heating electric current, and λ is the thermal conductivity
of the specimen.
Following the mathematical procedure described elsewhere [Wagner,
1969], it is obtained:
2 T   T 0 
4 4
p   
  L  x   , (B.8)
2  TL  T
where T is the temperature of a rod of infinite length and TL is the

temperature at the midpoint of a rod whose length is 2L.
A good application of the method for the measurement of solid and
molten lead was done by Hemminger, who claimed an estimated
uncertainty of 2.5 % for the solid phase and 3 % for the liquid one
[Hemminger, 1989].
B1.1.3. Calorimeter Method

The calorimeter method is a steady state method for the
measurement of the thermal conductivity of refractory materials which is a
standard test since 1945 [American Society for Testing and Materials,
1993]. The principle of the method is depicted in Figure B.2. Heat flows
through the under measurement specimen, which is surrounded by guard
bricks, into a water-cooled calorimeter used for measuring the quantity of
heat flowing through it. Under steady state conditions, the thermal
conductivity of the solid material can be obtained from the temperature
gradient through the sample and from the heat flow through it.
It should be mentioned that the calorimeter and the guards are
independently supplied with cooled-water from a central water-circulating
system, which maintains the provided water at a constant temperature,
pressure, and flow rate. Calibrated differential 10-junction thermocouples
are used to measure the temperature difference between the calorimeter
and the inner guard, while a second set of differential 10-junction
thermocouples is used for recording the temperature of the incoming and
outgoing cooling water. Finally, the temperature of the test specimen is also
measured with calibrated thermocouples embedded in the sample.
Figure B.2 Typical testing section set up for a thermal conductivity apparatus using
the calorimeter method [American Society for Testing and Materials, 1993]
It is important to mention that under steady state conditions, when

the hot face (the face of the sample that undergoes the heating) temperature
is constant and the temperature difference between the calorimeter and the
inner guard is around 0.03 K or lower, the thermal conductivity, λ, of the
sample can be calculated from
Q L
 , (B.9)
A  T 2  T1 
where L denotes the distance between the two thermocouple junctions

where temperatures T1 and T2 are measured, and A represent the
calorimeter area. Moreover, T1 and T2 are respectively the lower and higher
test temperatures measured in the test specimen. The heat rate flowing
through the specimen, and consequently into the calorimeter, Q, can be
expressed as
C P  T  m
Q , (B.10)
t
where C P is the water heat capacity, ΔΤ is the temperature difference of

the cooling water between the calorimeter’s outlet and inlet, and m denotes
the mass of water flowing in the calorimeter during time t.
Therefore, combining equations (B.9) and (B.10), the specimen
thermal conductivity can be obtain from experimental variables as
C P  T  m  L
 . (B.11)
A  t  T2  T1 
The uncertainty of the method depends on the design parameters of

the apparatus and it increases in temperature [Salmon et al., 2009]. A
typical calorimeter apparatus, created and used by Barth et al. [Barth et al.,
2007] for the measurement of thermal insulations up to 1,923 K, has a
relative expanded uncertainty of 6.2 %.
B1.1.4. 3ω Method
The 3 omega (3ω) method was firstly proposed in 1987 [Cahill & Pohl,
1987] and is a reliable and accurate method widely used for the
measurement of the thermal conductivity of thin films [Cahill & Pohl, 1987;
Cahill, 1990; Kaul et al., 2007; Lee, 2009]. The principle of the 3ω method is
the application of an alternating current of angular frequency ω through a

thin metal heater line that has been directly deposited on an electrically
insulated specimen. Due to Joule heating, this current heats the sample at
an angular frequency of 2ω and produce temperature oscillations at the
same frequency, with amplitude ΔΤ and phase difference φ. Since the
resistance of the metal heater depends linearly on the temperature, the
temperature variation causes also resistance oscillations to the metal line. It
was noted that the resistance oscillation at 2ω, with the source current at
angular frequency ω, generates a small oscillating voltage signal across the
metal heater at 3ω [Birge & Nagel, 1987]. Therefore, the thermal
conductivity of the sample is evaluated by measuring the amplitude and
phase of the resistance changes of the metal film as a function of the
frequency. The amplitude ΔΤ is given by the equation [Carslaw & Jaeger,
1959]:
P
T   K o q r  , (B.12)
l   
where P l is the amplitude of the power per unit length generated by a

heater current of angular frequency ω passing through the narrow metal
line, λ is the thermal conductivity of the sample, and K o q r  is the zeroth-
order modified Bessel function.
By measuring the third harmonic signal at two different frequencies,
f1 and f2, the thermal conductivity of the sample can be obtained (without
knowing the its density and specific heat) by the equation [Cahill & Pohl,
1987]:
f2
V 3  ln
f1 dR
  , (B.13)
4    l  R  V 3,1 V 3,2  dT
2
where R denotes the average resistance of the metal line, V is the voltage
across the line at frequency ω, V3,1 is the voltage at the third harmonic for
frequency f1, V3,2 is the voltage at the third harmonic for frequency f2, and
dR dT is the slope of the calibration of the metal line source at the
measurement temperature.
In the 3ω method a single element is used both as heater and
thermometer, as it is done in the hot strip and hot wire methods (see
subsection B1.2.3 and Section B2). However, their main difference is the use
of frequency dependent temperature oscillation instead of the time
dependent temperature response. A typical application of the method, for
the measurement of thin solid polymer specimens was presented by Gu et
al. [Gu et al., 2009], in which the uncertainty for the thermal conductivity is
3 %.
B1.2. Transient Methods

Transient methods for the measurement of thermal conductivity
and/or thermal diffusivity are based on the generation of a non-stationary
temperature field inside the specimen. They are widely used as they have
many advantages compared to the steady state methods. Some of the
advantages are the simple sample configuration, the short time duration of
measurements and mainly low measurement uncertainty. In addition, they
can be used for the measurement of materials with significant heterogeneity
or multilayer materials. In the case of liquids, the transient methods have
the advantage that provided the small measurement times involved,
convection does not affects significantly the measurement due to the inertia
of the fluid.
There are several transient techniques, which differ on the method of
heat generation in time, or/and on the geometry of the employed
instruments and heat source. Moreover, there are methods in which there is
contact between the heat source and the sample, and others which are non-
contact methods. The most used transient methods for the measurement of
solids or thin films are:
i. Transient Hot Wire technique vii. Dynamic Radial Heat Flow method
ii. Transient Plane Source method viii. Laser Flash technique
iii. Transient Hot Strip method ix. Force Rayleigh Scattering method
iv. Pulse transient method x. Thermal Waves Analysis
v. Step-wise Transient method xi. Interferometry
vi. Transient Hot Bridge method xii. Photothermal Deflection method .
In the following subsections, the principles of the main transient

methods used for the measurement of the thermal conductivity of solids will
be briefly presented.
B1.2.1. Transient Hot Wire technique

The transient hot wire method, which was used in the present work,
is presented in detail in Section B2 and Chapter C, and therefore it is not
discussed here.
B1.2.2. Transient Plane Source Method

The transient plane source method is one of the most commonly used
methods for the measurement of the thermal conductivity of liquids and
solids [Gustafsson, 1991; Gustavsson et al., 1994; Boumaza & Redgrove,
2003; Assael et al., 2004a; Malinaric, 2004; Jannot et al., 2006; Jannot &
Acem, 2007; Malinaric, 2007]. Its principle is very similar to that of the
transient hot wire and transient hot strip method (see subsections B1.2.3
and B2). In these three methods, a resistive element is used both as a heat
source and as a temperature sensor. Moreover, the assumption that the heat
source is placed among an infinite medium is made. The heat source should
be thin and its electrical resistance as large as possible, in order to provide
applicability to small samples and high sensitivity to temperature
measurements.
In the transient plane source method, which was proposed by
Gustafsson [Gustafsson et al., 1979a; Gustafsson, 1991], the planar heat
source is sandwiched between two distinct specimens of the same material
or fully immersed in a single specimen for powders, pastes, and liquids. A
constant current is applied to it, so as to sufficiently increase the sensor
temperature by 1 to 2 K. Thus, as the sensor’s temperature varies so does its
resistance. Consequently, monitoring the resistance change versus time, the
thermal conductivity of the sample can be obtained. Despite the fact that
the resistive element of the heat source can have any form, for reason of
experimental and theoretical convenience, the arrangements used are in the
form of “hot plate/square” or “hot disks” (see Figure B.3). The patterns of the
resistive element are usually created by deposition techniques or by using
thin metal foils. In both cases, the resistive element should be enclosed in
thin electrical insulating layers. Moreover, there is the Gustafsson-probe
which resembles to a hot-disk with the difference that it consists of a spiral
metal foil enclosed inside two Kapton electrical insulating films (see Figure
B.3) and its use is defined by ISO 22007-2 [International Organization for
Although the method is widely used for the measurements of a great
variety of materials, it is not an absolute technique and it does not have a
theoretical analytical solution of the heat transfer model. The majority of
the commercial plane source devices are accompanied with the suitable
software, which numerically solves the partial differential heat transfer
equations. A typical application of the method for the measurement of the
thermal conductivity of solids by Gustafsson [Gustafsson, 1991], showed an
uncertainty of is 3 %. The main disadvantage of such techniques is that
there can be no exact theory that will correspond to a spiral disk.
(a) (b)
(c)
Figure B.3 Schematic illustration of different transient plane source heat sources:
(a) Hot plate/square sensor, (b) Hot disk sensor, and (c) Gustafsson probe.
B1.2.3. Transient Hot Strip Method

The transient hot strip method, which was firstly proposed in 1979
[Gustafsson et al., 1979b], consists the development of the transient plane
source method (see subsection B1.2.2). The main difference of the two
methods is the shape of the resistive element, which is in a thin metal strip.
The planar metal strip used, acts both as a resistive heater and as a
temperature sensor. The principle of the method is that the thin metal strip
is placed between two electrical insulating specimen slabs and it is heated
by supplying a constant current. By monitoring the voltage change across
the strip, and therefore the variation of its electrical resistance, the
temperature rise of the strip can be calculated using its premeasured
temperature coefficient of resistance. The method has been widely used for
the measurement of non electrical conducting solids and fluids, as well as
for conducting materials by applying a thin insulating layer on each side of
the metal strip [Gustafsson et al., 1979b; Log & Metallinou, 1992; Sabuga &
Hammerschmidt, 1995; Hammerschmidt & Sabuga, 2000; Hammerschmidt,
2003; Stosch & Hammerschmidt, 2003; Gustavsson et al., 2006].
As no exact theory exists for such an arrangement, Gustafsson
adapted the already existing theory for the hot wire to the strip equivalent
and applied modifications on the working equations. The working equations
of the method, in the ideal case of neglecting the end effects due to strip’s
electrical contacts and finite length (the strip length to width ratio should be
kept larger than 20 – 30 in order to minimize the effects), are [Gustafsson et
al., 1979b]:
a Vo2  I
V t    f   , (B.14)
2   L  
where
f      erf  1  
2
4 

 1  e 
2
 1
4 
 
 Ei  2 . (B.15)
Here Vo denotes the voltage at time t =0, V t  and I are the voltage drop
and the current at time t, a is the temperature coefficient of the strip’s
electrical resistance, τ is the dimensionless time, L is the strips half length
and λ is the thermal conductivity of the material. The error function and the
exponential integral are denoted by erf and Ei respectively.

The transient hot strip method has the advantage that a strip makes
better contact to the solid specimen than a hot wire and it is handled much
more easily. However, one of the main drawback of the method compared to
the hot wire one (see subsection B2) is that a strip has smaller electrical
resistance than the wire, and thus the change of the temperature dependent
voltage signal is much smaller. Moreover, the working equation (B.14) is
nonlinear and the measurands cannot be determined analytically, but have
to be calculated numerically.
Hammerschmidt and Sabuga, in their uncertainty assessment for the
transient hot strip method [Hammerschmidt & Sabuga, 2000], determined
that in the case of solids the uncertainty of the method for the measurement
of the thermal conductivity is 3 % and for thermal diffusivity is 11 %.
B1.2.4. Pulse Transient Method

In the pulse transient method the specimen consists of three parts (I,
II and III). A planar heat source, clamped between part I and part II,
produces a heat pulse due to Joule heating of its electrical resistance. A
junction of a thermocouple that is placed between the second and the third
part measures the temperature response to the heat pulse. An experimental
setup and the principle of the method is depicted Figure B.4.
The ideal model of the pulse transient methods is [Kubicar et al.,
2005b]
h2
Q
T  h ,t   e 4  t , (B.16)
C P       t
where T  h ,t  represents the transient temperature at the thermocouple

junction and Q is the energy of the heat pulse, whereas ρ, C P and α are
respectively the density, the specific heat and the thermal diffusivity of the
specimen. In addition, h depicts the distance between the heater and the
thermocouple.
The thermophysical properties referred in equation (B.16) can be
determined either by the one-point procedure, where the maximum of the
temperature response is taken as the input, or by the fitting procedure,
where the function of equation (B.16) is fitted to the observed data within
the time window of the temperature response [Kubicar et al., 2005b]

The uncertainty of the method depends on the measuring time during
which the temperature field is developed in the specimen, the geometry of
the specimen, and the properties of the heating source. Kubicar et al.
applied the method for the measurement of homogenous (BK7, PMMA,
stainless steel) and heterogeneous (composites, concrete) solid materials,
and estimated an uncertainty of about 4 % for thermal conductivity and
about 6 % for thermal diffusivity measurements [Kubicar et al., 2005a;
Kubicar et al., 2006].
Figure B.4 Experimental setup of the transient pulse method

(part I is “cut-out” to show the heat source structure) [Kubicar et al., 2005b].
B1.2.5. Transient Hot-Bridge Method

An evolution of the transient hot strip method is the transient hot
bridge technique [Hammerschmidt & Meier, 2006]. In this technique, which
was developed at Physikalisch-Technische Bundesanstalt in Germany, the
novel thermoelectric sensor is a printed circuit of nickel foil between two
polyimide sheets. Its layout consists of four tandem strips connected in
parallel. Each tandem strip is segmented into two parts of different lengths.
The eight strips are connected symmetrically in order to make an equal
resistance Wheatstone bridge. A typical transient hot bridge sensor can be
seen in Figure B.5.
At constant and homogeneous temperature profile, the bridge is
initially in balance and therefore there is no need to balance it prior to a
measurement. Applying a heating electric current to a pair of unequally

spaced strips, an inhomogeneous temperature profile is created and
therefore the bridge becomes unbalanced. Thus, the high sensitivity and
time dependent output voltage signal, produced by the sensor, is recorded
and analysed in order to provide the thermal diffusivity, the thermal
conductivity, and the volumetric specific heat of the under measurement
material. Except of the advantages of the transient hot strip method, the
transient hot bridge method has the advantage that as the specimen
surrounds all the bridge’s resistors, there is no error introduced due to the
wiring of the bridge. Moreover, the use of tandem strips segmented into two
parts compensates for the perturbing end effects. According to
Hammerschmidt and Meier the preliminary estimated uncertainty of the
hot bridge instrument, in the case of solid measurements, is determined to
be about 2 % for the measurement of thermal conductivity and 8 % for the
thermal diffusivity.
Figure B.5 Hot Bridge sensor [Hammerschmidt & Meier, 2006].
B1.2.6. Dynamic Radial Heat Flow Method

Dynamic radial heat flow method is a transient method for
determining the thermal diffusivity of a solid sample, and consequently its
thermal conductivity. It was first proposed in 1935 and has already been
used for the determination of the thermal diffusivity of ZrO2, ZrC, and TiC
and recently of various ceramic materials [Sheffield & Schorr, 1991].
In this method, the sample is subjected to continuous heating (and
cooling) at constant rate and its temperature gradient is determined by the
use of thermocouples that are embedded inside the sample and connected to
a computer controlled acquisition system. Moreover, the outer surface of the

specimen should be sufficiently large in comparison to the diameter of the
heated area so that the geometry can be considered radially infinite
[Donaldson & Taylor, 1975]. The constant heat or cooling rate, which should
be instantaneous and uniform, is achieved with the use of a silicon
controlled rectifier (SCR) which in turn supplies furnace power. In Figure
B.6, a schematic diagram of a radial heat flow system is presented. Finally,
the sample should be an isotropic material with properties constant at the
temperature range of interest. The primary concern for accurate thermal
conductivity measurements using this technique is to eliminate axial heat
flow, which is never entirely eliminated.
Figure B.6 Schematic block diagram of the dynamic radial heat flow method
(TC1 and TC2 are the thermocouples used) [Sheffield & Schorr, 1991].
For a solid cylindrical sample with length l, diameter d, and l d  2 ,

the thermal diffusivity equation is
dT
R2 
  dt , (B.17)
  T
where R denotes the distance over which temperature gradient ΔΤ is

measured, dT dt the heating or cooling rate, and Γ represent a geometry
parameter which is equal to 4 for an infinite length cylinder, to 6 for a

sphere and to 3 for an infinite plate. Having calculated the thermal
diffusivity, α, of the sample, its thermal conductivity, λ, can be calculated
through
     C P , (B.18)
where ρ and C P are the density and the specific heat of the sample.
B1.2.7. Laser Flash Technique

The laser flash method, which was firstly proposed in 1961 [Parker et
al., 1961], is a transient non-contact technique used for the measurement of
the thermal diffusivity of solids, and rarely of liquids. The method has been
proposed as a standard method for thermal diffusivity measurement since
1992 under the ASTM designation E1461 [American Society for Testing and
Materials, 2001] and later also under the EN 821-2 [European Commitee for
Standardization, 1997] and the ISO 18755 [International Organization for
Standardization, 2005]. The basic principle of the method is that the front
surface of a small, thin disc specimen of the under measurement material,
initially at constant temperature, is irradiated by a short-duration, high-
intensity uniform thermal pulse from a high power laser source. The heat
absorbed diffuses through the sample and the resulting temperature rise of
the rear face versus time, can be monitored using a suitable detector and
acquisition system. The temperature rise detector can be an infrared
detector, an optical pyrometer or even a thermocouple. The block diagram of
the laser flash method can be seen in Figure B.7.
Figure B.7 Schematic block diagram of the laser flash method.

The thermal diffusivity, α, of a sample of uniform thickness, L, can be

calculated by the equation:
L2
  1.38  2 , (B.19)
  t1 2
where t1 2 is the time required for the rear face to reach half of the
maximum temperature rise.
The main advantages of the method are the lack of contact between
the heat source, the simple specimen geometry and its small size, the
rapidity of the measurement and the ease for measurements over a large
temperature range. The laser flash method has been used widely for the
measurement of the thermal diffusivity of solids [Ohta et al., 1991; Perovic
& Maglic, 1991; Baba & Ono, 2001; Vozar & Hohenauer, 2003; Gaal et al.,
2004; Blumm et al., 2007b; Min et al., 2007]. According to ASTM E1461, the
method has uncertainty of  5 % for the measurement of the thermal
diffusivity of a variety of materials, which is primarily associated with the
measurement of the temperature. However, there are researchers, using
laser flash apparatus claiming an estimated uncertainty of  3% [Gaal et
al., 2004; Min et al., 2007].
It should be noted that the laser flash method has been extended to
the measurement of liquids at high temperatures [Maeda et al., 1996;
Blumm et al., 2007a]. However, the accuracy of the method at these
conditions is under question, as there is a chance of convective flow within
the sample or even misleading measurements from the infrared detector, as
it could sense the intermediate surface between itself and the fluid’s
container.
B1.3. Critical Evaluation

Summarizing the above mentioned methods for the measurement of
the thermal conductivity or thermal diffusivity, it can be deduced that the
majority of the steady state and transient methods seems to be suitable for
accurate (uncertainty lower than  3 %) measurements of solids under a
wide range of conditions. Therefore, a reasonable question is created:
“Which method should be used?”.
In order to answer this question, many interlaboratory measurements
projects has been performed [Perovic & Maglic, 1991; Williams & Shawyer,
1991; Tye & Salmon, 2002; Salmon et al., 2003; Rudtsch & Hammerschmidt,
2004; Salmon et al., 2007; Wulf et al., 2007; Martin et al., 2009]. The
objective of these interlaboratory programs (round-robin projects) was to
compare and evaluate the uncertainty of the different measurement
techniques for the determination of the thermal conductivity. In a project
like these, the thermal conductivity of a specific batch of a certain material
is measured, using different available techniques, and the results are
compared. It should be noted that the round-robin projects constitute the
basis for establishing a material as a reference.
In the following subsections, three representative interlaboratory
projects regarding the measurement of the thermal conductivity of solid
specimens of Pyroceram 9606 are presented and the results are discussed.
B1.3.1. Round-Robin Tests

The material selected to show the differences between the different
thermal conductivity measurement techniques is Pyroceram 9606, which is
a glassy ceramic. Since May 2007, Pyroceram 9606 is supplied by the
European Commission Institute for Reference Materials and Measurements
(http://www.irmm.jrc.be/) as a certified thermal-conductivity and thermal-
diffusivity reference material (designated as glass ceramic BCR-724) up to
1025 K [Salmon et al., 2007]. Moreover, Pyroceram 9606 constitutes one of
the materials measured during the present work (see Chapter E).
In order to establish Pyroceram 9606 as a reference material three
major interlaboratory projects were realized:
a) Project A
In 1966, Powell et al. [Powell et al., 1966] proposed Pyroceram 9606
as a thermal conductivity reference solid with an uncertainty of  5 % over
the temperature range from 200 K to 1000 K (no confidence level was
specified). In this work, six laboratories took part with only one set from
direct thermal-conductivity measurements [Robinson & Flynn], whereas all
other measurements were of thermal diffusivity performed by the laser-
flash or radial heat-flow techniques.
b) Project B
In 1988, Hulstrom et al. [Hulstrom et al., 1988], following an
interlaboratory measurements program reported new recommended thermal
conductivity values of Pyroceram 9606 up to 573 K, with  5.6 %
uncertainty at the 95 % confidence level. In referring to the previous values
by Powell et al. [Powell et al., 1966], Hulstrom quotes “It would appear that
the overall state of the art in thermal-conductivity measurements for
materials in this range has changed little over the last 18 years”. In this
round-robin project there were seven participants:
i. South Dakota School of Mines, U.S.A.

ii. Dynatech R/D Company, Cambridge, Massachusetts, U.S.A.
iii. Properties Research Laboratory, U.S.A.
iv. Sandia National Laboratories, U.S.A.
v. Austrian Research Centre at Seibersdorf, Austria
vi. General Electric Company, U.S.A.
vii. University of Manchester, U.K. .
The techniques employed by these participants were:
- Guarded hot-plate technique.

- Laser Flash technique.
- Comparative technique.o
The guarded hot-plate technique was operated as an absolute

technique for the direct measurement of the thermal conductivity of
Pyroceram 9606 by two laboratories with an uncertainty of better than
15 %. Measurements between the two laboratories differed by 14%. On the
other hand, laser flash thermal-diffusivity measurements, obtained in an
absolute way, were found to differ in excess of 20 %. Finally, measurements
performed by the comparative technique, were found less susceptible to
errors and a spread of the values of about  5 % was observed. To obtain the
recommended values, an iterative algorithm was adopted according to which
all data that deviated by more than two standard deviations from the
master fit were excluded.
c) Project C .
As already mentioned, since May 2007, Pyroceram 9606 is supplied
by the European Commission Institute for Reference Materials and

Measurements (http://www.irmm.jrc.be/) as a certified thermal-conductivity
and thermal-diffusivity reference material (designated as glass ceramic
BCR-724) up to 1025 K [Salmon et al., 2007]. This certification was the
outcome of a research project, funded by the European Union under the
“Competitive and Sustainable Growth” program (“HTCRM - High
Temperature Certified Reference Materials,” Contract SMT4-CT98-
2211/2003 [Salmon et al., 2003]). The uncertainty of the certified thermal
conductivity value is  6.5 %, while that of the thermal diffusivity is  6.1 %
at the 95 % confidence level.
In this project the following laboratories participated:
(A) Austrian Research Centre GmbH, Materials Technology, Seidersdorf,

Austria.
(B) CeramRes, Queen’s Road, Stoke-on-Trent, United Kingdom.
(C) Corus R&D, Ceramics Research Centre, Ijmuiden, the Netherlands.
(D) Constantine the Philosopher University, Physics Department, Nitra,
Slovakia.
(E) Forschungsinstitut für Wärmeschutz, Gräfelfing, Germany.
(F) INSA, Centre de Thermique de Lyon, Vilieurbanne cedex, France.
(G) Forschungsinstitut für Kerntechnik und Energiewandlung, Stuttgart,
Germany.
(H) Institute for Reference Materials and Measurements, European
Commision, Geel, Belgium.
(I) Laboratoire National d’ Essais, Trappes Cedex, France.
(J) Netzsch Gerätebau GmbH, Selb/Bayern, Germany.
(K) National Institute of Science and Technology, Building and Fire
Research, Gaithersburg, U.S.A.
(L) National Physical Laboratory, CBTM, Teddington, U.K.
(M) Physikalisch-Technische Bundesanstalt, Dept. 3.102, Braunschweig,
Germany.
(N) Sociétè Francaise de Céramique, Ceramique Industrielles, Paris,
France.
The thermal conductivity measurements included in the project were

obtained by the following techniques:
- 3 sets by the guarded hot-plate technique (G, LCase1, MCase1),

- 4 sets by the parallel transient hot-wire technique (B, C, LCase2, N),
- 1 set by the resistive transient hot-wire technique (LCase3), and
- 1 set by the transient hot strip (MCase2).
In addition, 5 more sets of measurements (A, G, I, J, LCase4) were included
for the thermal-diffusivity certification.
In this round-robin project, for the first time, the transient
techniques for the measurement of the thermal conductivity dominate the
picture. Unfortunately the situation has not improved. The uncertainty has
increased to  6.5 % at the 95 % confidence level.
It is important to make some points about the thermal conductivity
data used for determining the certified values. In Figure B.8 the percentage
deviations of the thermal conductivity, λ, measured at individual
laboratories from the interlaboratory mean thermal conductivity value,
λmean, are shown for each laboratory, for each investigated temperature.
Figure B.8. Percentage deviations of the thermal conductivity of each laboratory, λ,

from the interlaboratory mean value, λmean [Salmon et al., 2007].
It is obvious from Figure B.8 that around 300 K, the deviations of

seven sets of measurements from the interlaboratory mean value are
between -2 % and -4 %, whereas the other two sets (laboratories LCase2 and
N) are between +5 % and + 11%. Moreover, the guarded hot-plate
instruments seem to be properly designed for measurements between 325 K
to 475 K, whereas the hot–wire instruments operated in the parallel mode
can not provide low uncertainty measurements.
B1.3.2. Conclusion
In Figure B.9, the deviations of the recommended values of Powell et
al. [Powell et al., 1966], and of Hulstrom et al. [Hulstrom et al., 1988] from
today European Comission certified values [Salmon et al., 2007] are shown.
In the first case, the maximum calculated deviation is 4.3%, while in the
second one is 3.6 %, which both are well within the mutual uncertainties
This comparison is quite interesting as in 1966, techniques were not as
advanced as today.
Figure B.9. Percentage deviations of the thermal conductivity of each laboratory, λ,

from the interlaboratory mean value, λmean [Salmon et al., 2007].
Following the aforementioned discussion and presentation of the

results of three representative round-robin projects, in which different
steady state and transient thermal conductivity measurement methods were

used, the following points can be depicted:
- The apparatus used in the round-robin projects for the measurement of

Pyroceram 9606, claimed to comfort with the available (standard or not)
test methods. However, the results showed a broad range of uncertainties.
- Devices based on the same measuring method gave values that differ up
to 14 % (see Figure B.8. values of laboratory B and N). However, these
values are within the uncertainty of the method.
Therefore, it is the authors’ belief that techniques employed in

proposals for reference thermal-conductivity materials should be further
developed so as to minimize errors. The technique used should be absolute,
based completely on theoretical background and should not require
calibration or any other empirical constants. Moreover, the accuracy of the
technique should not depend on the measurement conditions.
A technique covering the above requirements is the transient hot wire
technique, operated in the resistive mode and employing two thin wires.
This technique is considered to be a well-established absolute technique for
the measurement of the thermal conductivity of gases, liquids, nanofluids,
melts, and solids [Wakeham et al., 1991; Assael et al., 2004b] and has been
already successfully applied to a very wide range of temperatures and
pressures. Its great variety of applications makes it a unique technique.
Thus, taking into all the above into account and considering the great
experience of the Laboratory of Thermophysical Properties &
Environmental Processes on the design and operation of transient hot-wire
instruments, it was decide to design a new transient hot wire sensor of
lower uncertainty (lower than 1 %) for the measurement of solids and
composite materials.
B2. Transient Hot Wire Technique

The fundamental principle of the transient hot wire technique is
illustrated in Figure B.10. In this method, the temporal rise of a thin
vertical wire immersed in a test material, initially at thermal equilibrium,
is observed following the application of a step voltage across the wire.
Following this change, the wire appears as a line source of heat with
constant magnitude per unit length, and the time evolution of the
temperature of the wire is determined by the thermal conductivity of the
material. The wire acts as both, a heat source and as a temperature
recorder. Therefore, the wire material must be pure and it should have very
thin diameter (< 50 μm), as it acts as a line source. Furthermore, nowdays,
in order to avoid end effects, two wires identical except for their length, are
employed. Thus, if arrangements are made to measure the resistance
difference of the two wires as a function of time, which corresponds to the
time dependent resistance change of a finite section of an infinite wire (as
end effects being very similar, are subtracted), the wire’s temperature rise
can be determined.
Figure B.10 Schematic illustration of the transient hot wire technique.
In the ideal design of a transient hot wire device (see Chapter C), heat
transfer is considered to happen only by means of conduction and the wire is
considered to be an infinite continuous heat source, which produces constant
heat flux per unit length. Moreover, the material to be measured is
considered to be incompressible, to have constant density, specific heat and

thermal conductivity, and to extent infinitely in the radial direction.
The transient hot wire technique is widely used, as it has many
advantages compared to other methods. Firstly, it has a precise fully
developed theoretical background, which can be adequately modified to
describe any experimental setup. Secondly, it is an absolute method for the
measurement of the thermal conductivity and so neither empirical
constants, nor calibration setup, are required. Finally, if it is applied
properly, very low uncertainty, not affected by the experimental conditions,
can be achieved, i.e. well below 1 % for gases, liquids and solids [Wakeham
et al., 1991; Assael et al., 2008], and below 2 % for nanofluids and melts
[Assael et al., 2004a; Peralta-Martinez et al., 2006b].
Today's instruments employ highly accurate automatic electronic
bridges allowing more than 1000 measurements from times lower than 1 ms
up to 1 s (or 10 s, in the case of solids) coupled with finite elements methods
to establish a very low uncertainty.
B2.1. Historical Development

It is of great interest to follow up the evolution of the technique from
its beginning up to its present setup. The development of the transient hot
wire technique will be presented here through three chronological periods of
time: from 1780 to 1888, from 1888 to 1971, and from 1971 to today. The
review includes the development of the technique, not only for the
measurement of solids but also of fluids, as it was first used for the
measurement of the thermal conductivity of gases and liquids. It should be
noted that this historical review comprises a typical selection of
investigators that contributed to the evolution of the hot wire technique and
certainly does not include all investigators associated with this.
B2.1.1. The period from 1780 to 1888

It was about 1780 when Joseph Priestly [Priestly, 1781] performed
probably the first systematic experiments in order to measure the “power of
gases to conduct heat”, which however turned out to be the specific heat
capacity, which was unknown at that time. At that year, Joseph Priestley
became a minister in Birmingham, where he continued his scientific

researches and also his theological ones. He expressed his views quite
forcibly, and, in 1785, his “History of the Corruptions of Christianity” was
publicly burned.
Similar experiments were also carried out by Count Rumford (Sir
Benjamin Thomson) in 1786 [Thompson, 1786, 1799; Burr, 1934]. He
examined “the conductive power of artificial airs or gases” and while doing
so he stumbled on a completely new mode of heat transfer, convection. Thus,
in 1799, Count Rumford concluded that in gases heat transfer took place
only through convection, and consequently that a gas was unable to conduct
heat. Furthermore, he postulated that “Heat is incapable of passing through
a mass of air, and that it is to this circumstance that its non-conducting
power is principally owing”. This conclusion was not challenged for many
years and was universally accepted because of Count Rumford’s reputation
within the scientific community.
In 1804, Sir John Leslie in “An Experimental Inquiry into the Nature
and Propagation of Heat” [Leslie, 1804], commented about Count Rumford:
“A late ingenious experimenter, who by the perspicuity and useful tendency
of his writings, is deservedly a favourite of the public, has advanced the
paradoxical conclusion that fluids are non-conductors of heat;… If the
proposition, however, be taken in its strict sense, it is more palpably
erroneous. Were fluids absolutely incapable of conducting heat, how could
they ever become heated?... How could water, for instance, be heated by the
content of warm air? But the question really deserves no serious discussion”.
These comments ignited a length controversy cornering the heat conduction
in gases.
Forty four years later, the discussion about conduction on gases was
still going on, when Sir William Robert Grove presented an interesting
communication [Grove, 1845, 1848]. The paper was illustrated by an
experiment, in which it was shown, that a platinum wire, ignited by a
voltaic current, was cooled far below the point of incandescence when
immersed in an atmosphere of hydrogen gas. Sir William Robert Grove
wrote: “It was found not to be due to specific heat, nor to the conducting
powers of the gases. Convection did not explain the fact; and considerable
difficulty was found, upon examination, to exist, if it was attempted to refer
it to the greater mobility of the particles of hydrogen gas, the tightest
known, than of either oxygen, nitrogen, or carbonic acid. It was found that
this peculiar property also belonged, but to a less extent, to all the
compounds of hydrogen and carbon”.
Some years after, in 1861 Professor Gustav Magnus reported in
Poggendorf's Annalen, immediately being translated into English and
appearing in full in the Philosophical Magazine [Magnus, 1861a, 1861b]: “A
platinum wire is less strongly heated when surrounded by hydrogen than by
atmospheric air or any other gas...It cannot hence be concluded that gases
do not conduct heat, but only that they do this on so small a degree that the
action of conduction is cancelled by their diathermancy... Hydrogen conducts
heat better than all other gases”.
Nevertheless, the debate about conduction in gases had not seen its
end yet. John Tyndall [Tyndall, 1863] in his book “Heat considered as a
Mode of Motion” writes: “The subject of gaseous conduction has been
recently taken up by Prof. Magnus, of Berlin, who considers that his
experiments prove that hydrogen gas conducts heat like a metal… Beautiful
and ingenious as these experiments are (referring to experiments performed
by Magnus & Grove), I do not think they conclusively establish the
conductivity of hydrogen…. Theories are indispensable, but they sometimes
act like drugs upon the mind. Men grow fond of them as they do of dram-
drinking…”.
The gas conduction debate ended in essence with the wide acceptance
of the groundbreaking paper of James Clerk Maxwell [Maxwell, 1860a,
1860b] on the dynamic theory of gases and his calculation of a theoretical
value of the thermal conductivity of a gas. In the same paper, he also
showed the dependence of the thermal conductivity on temperature and
pressure. Furthermore, two years later, Rudolf Julius Emanuel Clausius
[Clausius, 1862a, 1862b] backed up Maxwell's theories, demonstrating that
the thermal conductivity of ideal gas depended on temperature, but was
independent of pressure.
Ten years later, in 1872, Josef Stefan [Stefan, 1872; Crepeau, 2007]
with his experimental expertise (he is well known for the fourth power
temperature to radiation law), set out to verify Maxwell's words. He thus
made the first absolute measurements of the thermal conductivity of air
with an instrument he called “diathermometer” (Figure B.11), and the
measured value was only 7 % off the theoretical value calculated using
Maxwell’s model. It should be mentioned that Stefan’s measuredvalue of the
thermal conductivity of air was 23.4 mW.m-1.K-1, when today's accepted
value is 26.34 mW.m-1.K-1 at 300 K. The diathermometer's principle was

based on the energy balance
equation, rearranged to be expressed
as a measure of pressures. The
thermal conductivity of the gas,
placed in the space between a
cylinder and the outer wall, was
obtained by the difference in
pressures. This was one of the first
thermal conductivity instruments.
Moreover, Stefan’s experimental
data also showed that the thermal
conductivity of air did not vary with
the pressure, verifying once again
the predictions of Maxwell and
Clausius [Crepeau, 2007].
All these discussions and
experiments with heated wires, in a
sense prepared the ground for the
work of August Schleiermacher
[Schleiermacher, 1888]. Following
the work of Josef Stefan and his
predecessors, August Schleier-
macher set out to measure the
thermal conductivity of gases by Figure B.11 Schematic reproduction of the
using a Pt hot wire (probably this is original diathermometer used by Joseph
Stefan [Crepeau, 2007].
the first real Hot-Wire instrument).
The instrument, shown in shown in Figure B.12, was composed of a
horizontal Pt wire of 0.4 mm diameter and 32 cm length, placed in the
centre of a glass cylinder of 2.4 cm diameter and kept taught by a metallic
spring. Potential leads were soldered at points A and B. The resistance of
the portion of the wire between points A and B, was obtained by measuring
the voltage and the current through it. The temperature inside the glass
was obtained from the resistance of the wire, while the outside temperature
was registered by a wound resistance thermometer. Equilibrium was
reached in minutes.
Figure B.12 Schleiermacher's hot wire instrument [Schleiermacher, 1888].
The thermal conductivity, λ, of the gas was obtained from the

expression [Tsederberg, 1965]
q
= A , (B.20)
T1 - T2 
where T1, and T2 are the inside and outside temperatures, q, the supplied
heat per unit length and A, is a constant obtained from the geometry of the
instrument or by measuring other gases of known thermal conductivity.
This instrument was employed for measurements up to 1,273 K.
B2.1.2. The period from 1888 to 1971

During this period, the transient hot wire evolves as a full instrument for
measuring the thermal conductivity of fluids and solids. Sophus Weber
[Weber, 1917a, 1917b] investigated the sources of errors involved in the
determination of the thermal conductivity of gases by the method of
Schleiermacher, and proposed a modified form of apparatus in which the
errors due to convection are greatly reduced. The main novelty of this
instrument, shown in Figure B.13, was that it was placed vertically. With
this improved instrument, the thermal conductivity of a large number of
gases was measured.
However, the first “transient” hot wire instrument was proposed by
Bertil Stâlhane and Sven Pyk in 1931 [Stâlhane & Pyk, 1931], in order to
measure the thermal conductivity of powders. The wire was simply placed in
the powder, located between two solids that were kept at constant
temperature. They further investigated the relationship between time and
temperature rise, for a fine straight

wire subjected to a step change in
the wire’s heat input. The
experimental results led to the
conclusion that at a short interval
after the initiation of the step
change, the following empirical
relation holds for the temperature
difference, ΔT:
A q  ro2 
T   ln   B , (B.21)
  t 
where q, is the heat per unit length,

ro the wire's radius and t the time.
Constants A and B were obtained
from fluids of known thermal
conductivity.
Figure B.13 The hot-wire instrument
Some years later, in 1938,
proposed by Weber
Pfriem [Pfriem, 1938] developed the [Weber, 1917b, 1917a]
first absolute technique, which was
consequently used by Eucken and Englert [Eucken & Englert, 1938]. They
both improved the design of Stâlhane and Pyk, and designed an absolute
transient hot-wire instrument for low temperatures. They employed a
cross-like arrangement where the 0.1 mm-diameter Pt wire was in the
middle of the cross, two wires at its side served as potential leads and the
remaining two, as support wires. In addition, the temperature was obtained
from the wire's resistance.
In 1949, van der Held and van Drunen [van der Held & van Drunen,
1949] developed an apparatus for the measurement of the thermal
conductivity of liquids. Moreover, they mathematically deduced the
empirical expression of Stâlhane and Pyk by solving the Fourier’s equation,
and they further advanced it closer to what will become later known as the
“ideal solution” (see Chapter C). Thus, they were able to perform absolute
measurements. Specifically they obtained
q  4   t 
T   ln  0.5772  , (B.22)
4     ro2

or for two distinct times t1 and t2
q t 
T1  T2  ln  2  , (B.23)
4      t1 
where α, is the thermal diffusivity of the medium.

Three years later, de Vries [de Vries, 1952] designed a transient hot
wire sensor for measuring the thermal conductivity of solids. The 0.1 mm
diameter hot wire, made of enamelled constantan, was folded and placed
inside a glass capillary of 0.4 mm outside diameter. Outside the capillary, a
fine monel gauze formed a socket of 1.4 mm outer diameter. The glass
capillary and an additional thermocouple were introduced into a gauge
cylinder, and the empty space was filled with paraffin wax. The whole
sensor had a length of 10 cm. Resistance was measured with a sensitive
D.C. Wheatstone bridge, while heating times of 3 min were employed. The
results of this method showed that daily trends in moisture content of soil
reflected on the measurement of the thermal conductivity.
In 1955, Gillam et al. [Gillam et al., 1955] employed the ideas of
Stâlhane and Pyk, and of Eucken and Englert to produce a simpler
apparatus of lower uncertainty, 0.3%, for liquids and solids. As far as solids
are concerned, the apparatus could measure solids of low melting points by
pouring the melted solid into the cell and let it solidified there. They
employed a 0.1 mm Pt wire with potential leads. The temperature was
obtained from the wire's resistance, and an optimized Kelvin bridge was
employed to record accurately the resistance change. The recorded times
were between 1 - 60 s.
New instruments at this period were characterized by:
i. A wire with potential leads, from its resistance the temperature rise was
obtained,
ii. Heating times of up to 1 min, and
iii. The appearance of new bridges for resistance measurement. .
Accordingly more instruments were developed. In 1960, Grassman

and Straumann [Grassmann & Straumann, 1960] publish a paper
presenting a transient hot wire instrument for gases and solids which had
an uncertainty of  1 %, whereas Haupin [Haupin, 1960] developed an
instrument for solids in which a thermocouple with its junction, placed in
the middle of the sample and acting both as heat source and thermometer,
was heated with alternating current. A year later, Turnbull [Turnbull, 1961]
developed a theory and an instrument for measuring the thermal
conductivity of molten salts in the liquid and solid states, while in 1963,
Horrocks and McLaughlin [Horrocks & McLaughlin, 1963] developed a
technique for measuring the thermal conductivity of liquids over an
extended range of temperature (295 K to 519 K).
In 1964, Mittenbühler [Mittenbühler, 1964] used a heating wire of
one metal, with a thermocouple welded to the heating wire in the form of a
cross, in order to measure the thermal conductivity of solids. The developed
technique constituted the basis for the German DIN 51046 Hot Wire
standard test method [Deutsches Institut für Normung, 1976], the
European EN 993-14 Cross-Array Hot Wire method [European Commitee
for Standardization, 1998a], and the ISO 8894–1 Hot–Wire method
[International Organization for Standardization, 2010], still in use today.
Besides that, in 1968, Burge and Robinson [Burge & Robinson, 1968]
proposed an instrument for gases employing lower times (Figure B.14),
while Hayashi et al. [Hayashi et al., 1971] propose an instrument for solids
up to 1,473 K.
Figure B.14 Schematic of line-source thermal conductivity cell proposed

by Burge and Robinson [Burge & Robinson, 1968].
It is noteworthy that during 1971 many changes took place. Davis et

al. [Davis et al., 1971] proposed a very small transient hot-wire instrument,
where the observation of the traditional galvanometer spot was substituted
by a digital voltmeter. Davis et al. employed a very thin Pt wire of 6.4 μm
diameter and 13 cm length. At the same year, McLaughlin and Pittman
[McLaughlin & Pittman, 1971a, 1971b] also proposed an innovative
transient hot-wire instrument with a 25 μm diameter and 15 cm length Pt
wire, with a weight under it to ensure verticality. The instrument was
designed to operate from 100 to 450 K and up to 10 MPa pressure. In
addition, following the ideas of Mani [Mani, 1971], a digital voltmeter was
employed to record the wire’s resistance change with time, and thus the
temperature change of the wire.
However, the most important breakthrough is attributed to the work
of Haarman [Haarman, 1971]. His equipment used an automatic
Wheatstone bridge to measure the resistance difference of two wires. Both
wires were identical except for their length. Hence, the end effects of the
wires were subtracted. Additionally, the bridge employed was equipped with
an electronic potential comparator and it was capable of measuring the time
required for the resistance of the hot wire to reach six predetermined
values, with the aid of high-speed electronic switches and counters. This
new bridge’s arrangement led to a ten-fold reduction of the duration of each
experimental run, eliminating in this way the convection effects and
reducing all other time-dependent errors.
B2.1.3. The period from 1971 to today

The pioneering work of Haarman, (i.e. the use of two wires with a
faster electronic bridge and the reduction of the experimental time to few
seconds), in addition to a paper published by Healy et al in 1976 [Healy et
al., 1976], which summarized the modernized theory of the transient hot
wire method and the most important corrections that needed to be made
during the measurements, were the start of a new series of transient hot-
wire instruments developed after 1971. Moreover, in the new instruments
the wire acts as both, a heat source and as a temperature recorder. It is
noteworthy that the new instruments were similar, as the research groups
cooperated through projects. Nowadays, many research groups on the
transient hot wire method originated from or are connected to J. Kestin
(Brown University, U.S.A.) or W. A. Wakeham (Imperial College &

Southampton University, UK).
In Figure B.15 some important changes that the design of the
transient hot wire has undergone are shown. The first design (Figure
B.15a), used for the measurement of the thermal conductivity of gases [de
Groot et al., 1974] and liquids [Nieto de Castro et al., 1976], consisted of two
5 μm or 10 μm diameter Pt wire, fixed at both ends. These were Wollaston
processed wires, i.e they were initially drawn covered by silver. The
drawback of this instrumental setup was that the thermal expansion of the
wires during the transient heating causes the wires become slack, and thus
convection problems to arise.
In order to overcome the thermal expansion problems, a gold spring
was employed. This design (Figure B.15b) was a logical improvement step.
The gold spring ensured that the wires were always taut. The contraction of
the gold spring was able to absorb the wire extension during heating.
Moreover, the Wollaston processed Pt wires had 10 μm diameter Pt. This
design was successfully used for measurements at room temperatures
[Kestin et al., 1980; Assael et al., 1981; Haran & Wakeham, 1982], but at a
wide temperature range its use showed some limitations due to gold spring
properties. An alternative arrangement, used in order to eliminate the
thermal expansion problems, was to introduce a weight instead of the gold
spring. The cell shown in Figure B.15c includes a platinum weight, and this
experimental setup was successfully used for the measurement of the
thermal conductivity of gases [Maitland et al., 1986] and liquids [Menashe
& Wakeham, 1981; Roder, 1981]. In this type of cells, the electrical
continuity was achieved by introducing a loop of soft gold between the
weight and the bottom support.
However, the aforementioned designs were not suitable for the
measurement of the thermal conductivity of electrically conducting liquids,
thus a new design was needed. In 1982, Alloush et al. [Alloush et al., 1982]
proposed the use of wires made of tantalum. Tantalum, upon oxidation,
forms a thin layer of tantalum pentoxide, which is an electrical insulator. It
is noted that a small correction to the calculated temperature should be
applied [Nagasaka & Nagashima, 1981]. In Figure B.15c an arrangement
with all the weights and electrical connections made of tantalum is
illustrated. Thus, the application of the previous modification permitted the
Figure B.15 Some schematics of transient hot-wire instruments for fluids.

use of the transient hot wire method for the measurement of electrically
conducting liquids [Wakeham & Zalaf, 1985; Assael et al., 1988]. It should
be noted that this arrangement is still in use today.
The use of tantalum wires led to designs (d) and (e) of Figure B.15. In
this case, instead of a weight, the wire supports are also made of tantalum,
so they expand similarly to the wire when the temperature rises, and
consequently the heat expansion problems are minimised [Jawad et al.,
1999; Assael et al., 2004a]. Furthermore, the most recent design of the
transient hot wire instrument (Figure B.15e) comprises tantalum wires with
length much shorter than the previous ones used [Assael et al., 2004a].
Until now, the majority of instruments mentioned were suitable only
for liquids and gases. However, a great development has been also made
concerning the measurement of the thermal conductivity of solids and melts
with the transient hot wire method. Today, there are already two transient
hot wire standard test methods for the measurement of thermal
conductivity of solids, the American ASTM C1113-99 Resistive Hot Wire
test method [American Society for Testing and Materials, 1999] and the
European EN 993-15 Parallel Hot Wire test method [European Commitee
for Standardization, 1998a]. The EN 993–15 is similar to the ISO 8894-2
Hot–Wire method (parallel) [International Organization for
In the case of the resistive mode of the transient hot wire method
[American Society for Testing and Materials, 1999] a pure platinum wire is
placed between two solid specimens of the material and a constant electrical
current is applied to it. The wire acts as both a heat source and as a
temperature recorder. However, the heating wire is usually 0.35 mm - 0.50
mm in diameter, with a length of about 20 cm, resulting in a resistance of
about 0.1 Ohm to 0.2 Ohm, which in turn requires high operating power.
The large power and the low resistance can easily produce larger
uncertainties. Moreover, the use of a single wire instead of two enhances the
errors associated with the end effects of the wire. Another point that should
be added is that, in this resistive form of the transient hot wire method, only
a segment of the temperature versus time curve is used instead of the whole
curve, in order to obtain the thermal conductivity. Hence, the choice of this
“segment” affects the uncertainty of the method, which is also affected by
the appearance of contact resistance when the wire is embedded in the solid.
The contact resistance still exists even if powders are employed.
On the other side, in the parallel mode of the transient hot-wire

instrument, instead of the wire acting both as sensor and a thermometer, a
thermocouple is placed usually 1.5 cm from the heating platinum wire to
register the temperature [European Commitee for Standardization, 1998b].
However, the errors that exist in the resistive mode of the hot wire, also
appears in this one.
Therefore, in order to overcome the difficulties rising from the contact
resistance and the end effects of the wire, the experimental setup with two
tantalum wires was adjusted for the measurement of solids and melts. The
most recent designs can be seen in Figure B.16. In the first design (a), firstly
used by Assael et al.[Assael et al., 2002], the two wires are placed inside a
soft silicone layer, which is squeezed between the two solids. The soft
silicone layer eliminates the possibility of having an air layer between the
wire and the solid, and thus it greatly enhances the contact between wire
and silicone. In this arrangement, measurements of the temperature rise at
very short times provide the properties of the silicone layer, while from
measurements at longer times the properties of the solid are acquired. The
design shown in Figure B.16b is based on the same idea, but here the wires
are placed between soft alumina layers which are consequently fired under
pressure to become one solid layer. This arrangement has been successfully
employed for measurements of the thermal conductivity of molten metals
[Peralta-Martinez et al., 2006a, 2006b] up to 1,200 K.
A recent modification of the transient hot wire method is the short
hot wire technique [Xie et al., 2006], in which a short metal wire of about
10 mm is used to measure the thermal conductivity and the thermal
diffusivity. This arrangement has the advantage of a smaller sample cell but
its main drawback is that the wire cannot be considered an infinite heat
source, and therefore, the accurate analytical theory of the “ideal solution”
(see Chapter C), for the temperature evolution of the transient hot wire,
cannot be applied.
Lately, the use of the transient hot wire technique is expanded on the
measurement of the thermal conductivity in the moderate to dilute gas state
[Wakeham, 2009]. Research is mainly done on the estimation of the
uncertainty of the method under this thermodynamic region, as there are
experimental data with significant discrepancies. A possible low uncertainty
will make the technique a reliable way to reconcile the differences.
(a) (b)
Figure B.16 Some transient hot-wire instruments for solids.
It should be mentioned that the development of the transient hot wire

technique was boosted by the tremendous development of the automatic
electronic bridges, which followed the advances in electronics and
computers. The new electronic equipment gives the possibility to record a
large number of experimental measurements (1000 experimental points)
starting from very small times (20 μs) and it is totally computer controlled.
The detailed description of the transient hot wire automatic bridges, as well
as the working equations, are presented in Chapter D.
Moreover, during the last decade, the use of the finite elements in the
measurement procedure of thermal conductivity is being developed. The
finite elements are used to fully describe the exact geometry of the
experimental sensors and to ameliorate the accuracy of the technique. The
recent advancement in computer technology enables the solution of large
system of equations and the display of the results quickly and conveniently.
Therefore, the initial Fortran programming language used to create the
mesh of the geometry, today, is replaced by the numerous finite element
software packages (Comsol, Ansys, etc.). In Chapter C, the differential
equations that describe the transient hot wire technique are presented, and
the use of COMSOL Multiphysics finite element software is analysed.
B2.1.4. Brief Presentation

In Table B.1 a brief concentrated presentation of historical review of
the transient hot wire technique is shown.
Table B.1 Brief presentation of the historical development of

the transient hot wire technique.
Measured
Year Researcher Brief review
material
Period 1780 to1888

1780 J. Priestley Gas First experiments in conduction
1786 Count Rumford Gas “Gases do not conduct”
1804 J. Leslie Gas “Can not be so”
W.R. Grove &
1848-1861 Gas First experiments with heated wires
G. Magnus
1863 J. Tyndall Gas Still doubts about conduction
J.C. Maxwell &
1860-1862 Gas Thermal conductivity theory
R.J.E. Clausius
1872 J. Stefan Gas Thermal conductivity measurement
1888 A. Schleiermacher Gas First hot wire-horizontal
Period 1888 to 1971

1917 S. Weber Gas Vertical hot wire
B. Stâlhane &
1931 Liquid / Solid First transient hot wire, & theory
S. Pyk
A. Eucken & Gas / Absolute transient hot wire
1938
H. Eglert Liquid T  f R 
E.F.M. van der Held & Transient hot wire, time 3 min,
1949 Liquid
F.G.van Drunnen T  f t c 
Transient hot wire theory, low T,
1952 D.A. de Vries Solid
time: 10 s to 20 s, T  f t c 
D.G. Gillam,
Transient hot wire,
L. Romben, Liquid /
1955 time 1-60 s,
H.-E. Nissen, Solid
T  f R 
O.Lamm
P. Grassman & Two transient hot wires,
1960 Gas / Liquid
W. Straumann time: 1 s to 20 s
1960 W.E. Haupin Solid Transient hot wire
Melt and
1961 A.G. Turnbull Transient hot wire measurement
Solid
J.K. Horrocks & Transient hot wire, analysis of
1963 Liquid
E. McLaughlin errors, time 30s, T  f  R 
Transient hot wire & t/c cross
1964 A. von Mittelbühler Solid (basis for German DIN 51046 and
European EN 993-14 standards)
Table B.1 (continued) Brief presentation of the historical development of

the transient hot wire technique.
Measured
Year Researcher Brief review
material
Period 1888 to 1971

H.L. Burge & Transient hot wire theory, time 30 s,
1968 Gas
L.B. Robinson T  f t c 
K. Hayashi,
Transient hot wire theory,
1971 M. Fukui, Solid
up to T  1,473 Κ , T  f t c 
I. Uei
E. McLaughlin & Transient hot wire, time 20s,
1971 Liquid
J.F.T. Pittman Digital Voltmeter T  f  R 
P.S. Davis,
Small Transient hot wire,
F. Theewes,
1971 Liquid time 10s,
R.J. Bearman,
Digital Voltmeter
R.P. Gordon
Two wires,
1971 J.W. Haarman Liquid
new - faster - electronic bridge
Period 1971 to Today

Two Pt short wires,
1976 C.A. Nieto Castro Gas
fixed at two ends
J. Kestin,
W.A. Wakeham Two Pt short wires,
1980 Gas
M.J. Assael connected to gold spring
E.N. Haran
W.A. Wakeham
J. Menashe Gas / Two Pt short wires,
1981-1986
G.C. Maitland liquid connected to platinum weigh
H.M. Roder
A. Alloush
Electrical
Y. Nagasaka
1982-1988 conducting Use of two tantalum wires oxidised
W.A. Wakeham
liquids
M.J. Assael
Parallel hot wire
1998 European Commission Solids
(European Standard EN 993-15)
Resistive hot wire
American Society for
1999 Solids (American Standard ASTMC1113-
Testing and Materials
99)
Use of two wires enclosed in silicon
2002 M.J. Assael Solids
paste, time 20s
Use of two tantalum wires oxidised,
2004 M.J. Assael Nanofluids
time 1 s
M.V. Peralta Martinez Two wires placed inside alumina
2006 Melt
M.J. Assael layer
B3. Summary
In this chapter, the main steady state and transient methods for the
measurement of the thermal conductivity of solids was presented. Moreover,
a critical evaluation of the methods was done by using the results of three
round–robin projects for the measurement of Pyroceram 9606. From this
brief introduction to the principles of the methods and the interlaboratory
measurement projects the following conclusions can be deduced:
i. It can be easily distinguished that the steady state methods are not
suitable for measurements of uncertainty lower than 3 %. Moreover,
they require long experimental times compared to the transient ones.
ii. Round–robin projects showed that devices based on the same measuring
method gave values that differ up to 14 %.
iii. From the available transient techniques, the only one with a precise
fully developed theoretical background, is the transient hot wire
technique, which has been already applied to solids with uncertainty of
2% [Assael et al., 2004b]
iv. Finally, the transient hot wire technique is the only technique that does
not need calibration, nor the use of empirical constants. .
Therefore, taking into account the above mention conclusions it has

been decided to apply a technique that offers accuracy superior than the
others on the measurement of the thermal conductivity of solids, and that
has been developed through the last years. This method is the Transient
Hot Wire.
C
Theoretical
The main objective of this chapter is to present and analyse the

fundamental theoretical and computational background needed for
performing measurements of thermal conductivity with the transient hot–
wire technique. The chapter consisted of two major sections.
The first one describes the “ideal” theoretical model on which the
transient hot–wire technique is based on, and the equations for its
analytical solution are given. In addition, the most significant corrections
that have to be applied to this “ideal” solution are briefly mentioned.
The second part of the chapter is devoted to the finite element method
and its use together with the transient hot–wire technique. Initially some
basic information on the finite element method is provided and the
previously developed finite element Fortran model for solids is discussed.
Following that, the main characteristics of the commercially available finite
element software, Comsol Multiphysics, are presented. An evaluation model
representing a simple geometry, i.e. the experimental setup for measuring of
the thermal conductivity of liquids, is created and the simulation results are
compared to the analytical solution. The evaluation model is also employed
for acquiring the thermal conductivity of a liquid from experimental data.
Consequently, the finite–element model created for representing the actual
experimental setup for solids is described analytically, and the procedure
followed for obtaining the thermal conductivity of both liquid and solid
samples from experimental data is shown in details. The chapter ends with
a sensitivity analysis of the effect that the finite–element model’s
parameters have on the calculated thermal conductivity value of solids.
58 Theoretical
C1. Ideal Model

The transient hot wire method is based on the observation of the
temporal temperature rise of a thin vertical wire immersed (or embedded) in
the test material. The wire is initially at equilibrium and a step voltage is
applied to it. In this way, electrical current flows through the wire and heats
it up. Therefore, the wire acts as a line heat source of constant heat flux per
unit length, producing a time dependent temperature field inside the test
material. However, the evolution of the wire’s temperature depends on the
thermal conductivity of the test material. Thus, the thermal conductivity of
the test material can be calculated by monitoring the temperature rise of
the wire and taking into account the properties of the wire and of the
surrounding material, as well as the geometry of the wire’s enclosure.
In the ideal model, an infinitely-long, vertical thin wire of zero heat
capacity and infinite thermal conductivity is immersed in an infinite
isotropic material of temperature–independent thermal diffusivity α, and
thermal conductivity λ, and which is initially (at t = 0) at thermodynamic
equilibrium with the wire at temperature To. Furthermore, the heat transfer
from the wire, when a stepwise heat flow per unit length, q, is applied, is
considered to be only conductive (radiation is considered negligible). Thus,
the temperature field around the wire can be described by Fourier’s partial
differential heat conduction equation:
T
 C p  =   2T (C.1)
t
T  1 T  2T 
 C p  =     2 , (C.2)
t  r r r 
where C P and ρ, are the specific heat and the density of the material around
the wire, and r is the radial distance from the wire. In addition, T r and
2
 2T r are the first and the second partial differential of the temperature
T . The following initial and boundary conditions are employed:
i. Initial condition
For t  0 and for any r T  r ,t   0 (C.3)

Theoretical 59
ii. Boundary conditions
 T  q
For t  0 and for r  0 lim  r    (C.4)
r 0
 t  2   
For t  0 and for r   lim  T  r ,t    0 . (C.5)

r 
Therefore, the problem is standard and the analytical solution of the

above fundamental equation (C.2) has been given by Carslaw and Jaeger
[Carslaw & Jaeger, 1959]
q  r2 
T  r ,t   T  r ,t   To   E1  , (C.6)
4     4   t 
where E1 is the exponential integral with the expansion [Carslaw & Jaeger,
1959]
 x 
 n
e x 
E1  x    dx    ln x   (C.7)
x x n 1 n  n !
 x x2 
E 1  x     ln x     ...  , (C.8)
 1  1! 2  2! 
where α is the constant thermal diffusivity, γ is the Euler constant equal to

0.5772157 and
r2
x . (C.9)
4   t
At the wire radius r  ro equation (C.6) can be written
q   4t  r 2 1  r 2 2 1  r 2 3 
T  ro ,t     ln  2    o    o     o   ... (C.10)
4       ro e  4t 4  4t  18  4t  

However, if the wire radius ro, and the time t, are chosen so that
parameter x has small value (typically 10-6), then the second and the
following terms on the right side of equation (C.10) can be considered
negligible (less than 0.01% of ΔΤ ) and it can be written
60 Theoretical
q  4t 
T  ro ,t    ln  2  . (C.11)
4     ro e 
Equation (C.11) is the fundamental working equation of the transient

hot–wire technique, and it gives the possibility to obtain the thermal
conductivity of the measured material from the slope of the line ΔT versus
lnt, while at the same time, the thermal diffusivity can be obtained from its
intercept.
C1.1. Correction to the Ideal Model

In practice, the design of the experimental setup for the measurement
of the thermal conductivity of a material introduces a number of departures
from the ideal solution. Therefore, essential corrections have to be made to
the acquired experimental data to adjust it to the ideal model described by
equation (C.11). The corrections applied to the experimentally measured
temperature rise follow the equation
Tid  ro ,t   Texp t    i Ti , (C.12)
where T id  ro ,t  is the ideal temperature rise at the wire radius ro at time t,

Texp t  is the temperature rise measured experimentally and  i T i is a
sum of the applied corrections.
The corrections to the ideal solution have been presented explicitly
elsewhere [Healy et al., 1976; Kestin & Wakeham, 1978]. In Table C.1, a
summary of the most important currently available corrections is presented
and their contribution to our type of experiment is commented. It should be
mentioned that some corrections are eliminated due to careful experimental
design and some others are considered negligible (effect to T id  ro ,t  no
more than 0.01 %).
Theoretical 61
Table C.1 Corrections to the ideal model.
Correction Contribution
This correction is needed only in the case of fluid measurement (gas or liquid) and it depends on the duration of the
experiment. The convection effect can be minimized with careful design and operation of the experimental equipment,
Convection as well as using a small temperature rise and a time scale smaller than the time required for the onset of significant
free convection. One of the main advantage of this instrument is that should convection be present, it will be
recognized immediately.
In the case of solid measurement heat is transferred only by conduction, as is it assumed in the ideal model, and
therefore no correction due to radiation is needed. However, in the case of gas or liquid measurement, a part of the
electric energy passing through the wire is radiated to the fluid. The correction that has to be done in this case is
[Healy et al., 1976]:
Radiation
8    ro   To3 2
T R    T  r ,T   (C.13)
q
where σ is the Stefan-Boltzmann constant and To the initial temperature.
This correction takes into account the finite radius of the wire ro, its finite thermal conductivity λW, and its heat
capacity   CP W per unit volume.
The correction that has to be applied in this case is given by [Carslaw & Jaeger, 1959; de Groot et al., 1974; Healy et
al., 1976]:
 r 2   Cp    Cp  
Finite wire TW  r ,t  
q
 
o W   ln  4    t   ro2   ro2   
 2     (C.14)
properties 4     2   t  r e   2   t   4  W  t  2  W 
  o      
4   t
which is valid for 1 .
ro2  e 
Usually the first term of equation (C.14) constitutes the major correction due to finite properties of the wire. In
practice, this correction has a larger effect at small times, and it no case it should be larger than 1 % of Tid .
62 Theoretical
Table C.1 (continued) Corrections to the ideal model.
The employed wire has a finite diameter, ro and thus the boundary of equation (C.2) changes to:
T q
For t  0 and for r  ro  (C.15)
t 2      ro
For the above boundary condition, the solution of the differential conduction equation, given by Carslaw and Jaeger
[Carslaw & Jaeger, 1959], is
Finite wire
 
 J o ur  Y1 uro   Yo ur   J 1 uro  
diameter q
1  e 

T  r ,t    2 
 u 2 t
   du (C.16)
  ro   0
 u 2  J 12 uro  Y12 uro   
   
where Jo and J1 are the first kind Bessel functions of first and second order respectively, whereas Yo and Y1 are the
second kind Bessel functions of first and second order.
Equation (C.16) cannot be solved analytically, but in case of small values for x [equation (C.9)], the ideal solution of
equation (C.10) is recovered.
The variation of the thermal conductivity and the heat capacity per unit volume of the under measurement material
with temperature, produces a correction that should be applied to the ideal solution. According to the analysis done by
Healy et al. [Healy et al., 1976], who considered a linear perturbation of these properties, the temperature, Tr, to
which the observed thermal conductivity refers is
Variable
1
fluid Tr  To    T t1   T t 2   (C.17)
2 
properties
where To is the initial equilibrium temperature of the material, and T t1  and T t 2  are the temperature rise
measured experimentally at time t1 and t2 respectively.
This correction is not negligible and is performed to the acquired experimental data.
Theoretical 63
Table C.1 (continued) Corrections to the ideal model.
In the ideal model, it is considered that there is an infinite outer boundary to the heat wave. However, in practice it is
not possible, as the fluid or solid is contained in a vessel of radius b. Therefore, the boundary of equation (C.2)
changes to:
For t  0 and for r  b T b ,t   0
which implies that the outer wall temperature does not change. A general solution to this problem is given by Carslaw
and Jaeger [Carslaw & Jaeger, 1959].
Outer b 4   t
However, Fischer [Fischer, 1939] proposed a limiting form of the general solution for  1 and 2   1 :
boundary ro r0  e
  g2  t 
q   4   t   2
TOB  r ,t   4      ln  2 
 b e
 
  1
e b2   Yo  g   
 
where gν represents the consecutive roots of the modified Bessel function J  g   0 .
In practice, this correction, contrary to the correction due to wire’s finite properties, has larger effects at large
measurement times, and in no case, it should be larger than 1 % of Tid , by design.
64 Theoretical
Summarizing, in the case of measurements of solids, the main

corrections that have to be applied to the experimental temperature
increase Texp , are the correction due to the finite diameter and properties
of the wire [equations (C.14) and (C.16)], and the correction due to the finite
outer boundary of the measurand container [equation (C.19)]. Moreover, the
calculated thermal conductivity is refered to the corrected equilibrium
temperature Tr [equation (C.17)].
However, the analytical solution of the equations of the ideal model has
some deficiencies, that have been analysed explicitly by Karagiannidis
[Karagiannidis, 1998] and Gialou [Gialou, 2004]. In the case of solids, the
drawbacks of the analytical solution are:
i. The mathematical difficulty to simulate the exact complete experimental

setup, as it needs the solution of a complex system of partial differential
equations that describes the heat transfer from the source to the
material measured.
ii. The weakness to apply the analytical solution to short experimental
times, t  0.1 s , as the corrections applied to the ideal model require
small wire diameter and not very short experimental times.
iii. In the ideal model it is considered that the material to be measured is
isotropic. However, for the measurement of solids a multilayer sensor is
used, with the wire encapsulated inside a silicone layer. Therefore the
ideal model with the small analytical corrections is no longer sufficient to
provide an accurate description of the experimental measurement. Thus
it is needed to employ another method that uses the experimental data
starting from t  0 in order to obtain the properties of the silicone layer.
Taking into account the above drawbacks of the analytical solution of

the equations describing the experimental setup, a numerical method was
used in this work. The numerical method used is the Finite Element Method
(FEM), which has been applied successfully to the measurement of gases
and liquids [Karagiannidis, 1998], of nanofluids [Metaxa, 2006], of melts
[Peralta-Martinez, 2000; Bilek, 2006] and of solids [Gialou, 2004]. The finite
element analysis used until now included the use of computer programs
written in Fortran language [Karagiannidis, 1998; Peralta-Martinez, 2000;
Gialou, 2004; Metaxa, 2006]. The main difference of this work in relation to
the above mentioned applications is the use of the finite element commercial
Theoretical 65
software package Comsol Multiphysics, which gives much higher flexibility

to the employed finite element analysis and thus it improves the
uncertainty of the measurements.
It should be noted that the numerical solution obtained with the use
of the finite element method is based on an inverse mathematical problem,
in which the thermal conductivity of the material is derived from the
comparison of the characteristic experimental temperature response of the
sensor with the simulated one.
In the following section, the fundamental principles of the finite
element method will be briefly presented and its use with the transient hot–
wire technique will be described.
C2. Finite–Element Analysis

The essential ideas of the finite element method appeared in the first
half of the twentieth century by three different groups of scientists
(mathematicians, physicists and engineers), at different times and for
different reasons. However, the term finite element method appeared in
1960 and since then, it has been widely accepted in engineering, resulting in
thousand of papers and hundreds of books on the subject [Huebner et al.,
1995]. In the last decade, the application of the finite element method in
many complex engineering problems has been improved by the rapid growth
in computer technology, e.g. the use of parallel processors, e.t.c. Hence, only
a very brief description will be given here.
The finite element method is a numerical analysis technique for
acquiring reliable approximate solutions to a wide variety of problems. The
basic idea of the method is to built up a given actual domain with an
assemblage of simple, small, interconnected geometric shapes, called finite
elements. The elements used can be triangular or quadrilateral, according to
their shape, and all together consist the mesh of the domain. Hence, since
these simple elements can be assembled together in many ways, an
extremely complex geometry can be represented.
Contrary to the finite difference method, which considers the solution
region as an array of grid points, the finite element method gives a
piecewise approximation to the governing equations, that is, the complex
partial differential equations (PDEs) are reduced to either linear or
66 Theoretical
nonlinear simultaneous equations. Moreover, the finite element

discretization procedure reduces a problem of infinite unknown field
variables, to one of a finite number of unknowns by dividing the solution
region into elements, and by expressing the unknown field variable in terms
of assumed approximating functions. Therefore, for each element, the
needed approximation functions that describe the problem are generated
and used in the solution of the PDEs. The approximation functions are
usually obtained using ideas of interpolation theory and hence they are also
called interpolation functions. The interpolation functions are defined in
terms of the values of the field variables at specified points called nodes.
Nodes usually lay on the element boundaries, where adjacent elements are
connected.
One of the main advantages of the finite element method, is its ability
to formulate solutions for single individual elements before assembling them
together to create the entire geometry of the problem. Hence, a complex
problem is divided into many simplified problems. In addition, there are
three different approaches to follow in order to obtain the properties of
individual elements [Huebner et al., 1995; Reddy & Gartling, 2001]:
- The direct approach, which can be use in relatively simple problems. This
approach is based on the direct stiffness method of structural analysis and
can be used to formulate the element properties for the simplest element
shapes.
- The variational approach, which relies on the theory of the calculus of
variations, that states that the function that extremises the variational
integral corresponding to the governing differential equation is the
solution of the original governing differential equation and boundary
conditions. It can be employed for both simple, and more sophisticated
element shapes.
- The weighted residuals approach, which can be used to represent complex
problems, where no variational formulation can be written. It starts with
the governing equations of the problem and proceeds without relying on a
variational statement. It is widely used in non–structural applications
such as heat transfer and fluid mechanics.
The successful application of the finite element analysis of a problem

needs to follow some fundamental steps which are [Huebner et al., 1995;
Theoretical 67
Reddy & Gartling, 2001; Lewis et al., 2004]:
i. Discretization of the problem’s domain into a set of non–overlapping

finite elements, in order to create the mesh of the problem. There can be
a variety of element shapes, e.g. triangles, quadrilaterals, e.t.c.
ii. The next step is to assign nodes to each element and choose the proper
interpolation function to represent the variation of the field variable over
the element.
iii. Development of the finite element model of the problem (i.e. element
equations) using one of the above mentioned approaches for obtaining
the elements properties (i.e. direct, variational or weighted residuals
approach).
iv. Following this the elements of the mesh are assembled together to obtain
the global system of simultaneous equations.
v. Consequently, the boundary conditions are imposed.
vi. Once the boundary conditions are determined and imposed, the system of
equations is solved.
vii. Finally, the solution is postprocessed in order to acquire the quantities of
interest.
The first finite element models created for heat transfer problems,
and specifically for the calculation of the thermal conductivity, were written,
as already said, in Fortran programming language. However, the
development of the computer engineering technology led to the creation of
commercial finite element software packages, which provide much higher
flexibility in the finite element analysis, have user–friendly interface, and
provide the user with the needed tools to describe and simulate any
geometry. Such packages are ANSYS and Comsol Multiphysics software.
In the next sections, the application of the Fortran finite element
model in the measurement of the thermal conductivity of solids is briefly
discussed, and consequently the use of Comsol Multiphysics software is
analysed. In the end of the chapter, a sensitivity analysis of the modelling
tool is also performed.
C2.1. Fortran Finite Element Model

As already mentioned, in the case of the measurement of solids, the
68 Theoretical
finite element method has been applied successfully by Gialou [Gialou,

2004]. In this application, the code of the finite element programme used for
the solution of the partial differential equations was written in Fortran
language. The developed code was able to represent the experimental setup
(i.e. a wire encapsulated inside a silicone layer and the sample) using only
rectangular elements. In Figure C.1, a typical representation of a quadrant
of the mesh used is shown (due to symmetry the solution is the same for the
other three quadrants). The circular hot wire was represented by a square of
side α, the silicone layer has thickness 2.δ and the width of the sensor is b.
The created mesh was quite dense in the wire, in the intermediate silicone
layer and in both interfaces.
Figure C.1 Schematic representation of the mesh used by Gialou [Gialou, 2004].
However, in this way the real circular cross–section of the wire was
represented by a rectangular element and therefore could not be defined
properly, introducing a source of uncertainty to the solution especially at
short times (t  0.1 s ). As mentioned before, the use of a multilayer sensor
makes crucial the application of a numerical method that gives the
possibility to exploit the experimental data starting from t  0 s . Data at
short times is used for the calculation of the thermal properties of the
Theoretical 69
silicone layer, while data at longer times are used for obtaining the
properties of the sample under measurement. In Figure C.2, a typical
experimental temperature response for Pyroceram 9606 measured using
silicone paste and the predicted with finite element method temperature
rise [Assael et al., 2002], can be seen.
It can be noted that at short times (t  0.1 s ) the two curves do not
coincide and therefore this leads to an erroneous calculation of the unknown
thermal properties of the silicone and consequently a small error is
introduced in the calculation of the properties of the solid. The discrepancy
between the two curves at small times can be explained considering that
there is still thermal contact resistance between the wire and the silicone
layer, which however has not be included in the finite element program.
Figure C.2 Experimental and predicted temperature rise in Pyroceram 9606

(sensor with silicone paste) [Assael et al., 2002].
Besides that, the whole code of the old Fotran model should be re–
written, if further improvements were to be included. Additionally, it had
not great flexibility in terms of suitable size of finite elements, time–
stepping options, implementation of new layers in the model and post–
processing of the results. Therefore, the logical step forward was the use of a
commercially available finite element software package (Comsol
70 Theoretical
Multiphysics), which provides all the needed tools to describe and simulate
any geometry with a user–friendly interface.
C2.2. Comsol Multiphysics Software

Comsol Multiphysics software, whose first version appear in 1999
under the name Femlab, consists a powerful interactive environment for
modeling and simulating all kinds of scientific and engineering problems
based on partial differential equations (PDEs). The software does not need
an in–depth knowledge of mathematics or numerical analysis in order to
model, simulate and solve a problem. It provides an integrated graphical
user interface in order to build and solve models by using predefined physics
modes, PDE modes, or a combination of them i.e. multiphysics modeling.
The large number of built–in physics modes included in the software, allows
the user to create a model just by defining the relevant physical quantities
and not the underlying equations. However, advanced users have the
flexibility to set up constants, variables, as well as mathematical and logical
functions. In addition, users have the possibility to access the model by
using script programming in the Comsol’s script language or in the Matlab
one, for defining properties, sources, boundary conditions, e.t.c.
In Comsol Multiphysics, models of one (1D), two (2D) or three (3D)
dimensions can be created. The procedure to design the geometry of a model
is quite simple. The user creates simple solid objects, already available in
the software (e.g. rectangles, circles, blocks, cones, spheres, e.t.c.), and then
he combines them by using Boolean operations, like union, intersection and
difference, in order to build composite solid objects. A great feature of
Comsol Multiphysics software is its flexibility to mix–use subdomains of
different space dimensions in the same problem, simplifying in that way the
modeling procedure and decreasing the execution time. Moreover, the
geometry of the model can be imported from .DXF and .GDS files for 2D
models and from .STL and .VRML files for 3D models.
All the solid objects combined to create the model, consist the model’s
main domain. Each separate object is called a subdomain and to each one of
them, different physical properties can be attributed to. The properties’
values of each subdomain can be set up, either by determining the
coefficients that defines the PDE on the subdomain, or by selecting a
Theoretical 71
material from the program’s database (in this case the software
automatically uses the corresponding coefficients in the PDEs). Comsol
Multiphysics program has a large extensible material library including a
broad collection of physical, mechanical, electromagnetic and thermal
properties materials.
After the design procedure is over, the model’s geometry is
automatically meshed. By default, Comsol Multiphysics software generates
an unstructured mesh. More specifically, for 2D models, the unstructured
mesh consists of triangular elements, whereas there is the possibility to
create a mapped mesh of quadrilateral elements. For 3D models the
generated mesh can be also unstructured consisted of tetrahedral elements,
or structured when it is generated from the extrusion or revolution of the 2D
mesh. It should be noted that in order to used a mapped mesh the geometry
of the subdomain should be fairly regular in shape.
In the case of a 2D unstructured mesh, there are a large number of
control parameters that allow the user to interfere in the mesh generation
process and to acquire the desired mesh. The mesh parameters can be
controlled either for the whole model or for each subdomain separately. For
more advanced users there are options to control separately specified
boundaries segments and nodes. The control parameters include among
others, the specification of the maximum element size, the maximum
element size scaling factor, the resolution of narrow regions where more
elements should be created, e.t.c. A quick review of the quality of the
elements, and the statistics of the created mesh is provided by the
corresponding option.
The next and most important step is the solution of the PDEs. In
order to solve the PDEs, the software employs the finite element method
and provides to the user a wide variety of solvers for stationary, time–
dependent and eigenvalue problems. Moreover, for the solution of linear
systems the software features both direct (UMFPACK, SPOOLES and
TAUCS) and iterative (GMRES, Conjugate Gradients and Geometric
Multigrid) solvers. The solver is set up automatically, depending on the
application mode but the user can intervene and select a more appropriate
solver for his model. More details for the above mentioned solvers can be
found in the program’s User Guide ["COMSOL Multiphysics User's Guide,
Version 3.2," 2005].
It is important to note that the solvers always break down each
72 Theoretical
problem into the solution of one or several linear systems by approximating

the given problem with a linearized one. Hence, in all cases (stationary,
time–dependent or eigenvalue models) the solver spends the most of the
solution time solving systems of linear equation systems. Therefore, one of
the most important solver’s parameter, that affects the solution time and
memory requirements, is the choice of the most effective linear (direct or
indirect) system solver. In brief, the direct solvers, which are mainly
suitable for 1D and 2D models, use the Gaussian elimination method for the
solution of the linear system. For the 3D models the iterative solvers are
preferred as the direct ones in this case require large memory resources and
long computing times.
Finally, for post–processing the software provides tools for plotting
and recording any model quantity or parameter. Moreover, it does not limit
post–processing just on predefined quantities. The user is able to insert the
corresponding equation of a function and to combine user defined variables
with the software’s ones. The visualization options of the solution include
among others surface plots, slice plots, contour plots, cross – section plots,
e.t.c.
After this brief introduction to the capabilities of Comsol Multiphysics
software, a detailed description of the heat transfer model created and used
during the present work is given.
C2.2.1. Evaluation of the Heat Transfer Model

Since the introduction of the finite element method in the accurate
measurement of thermophysical properties, a discussion upon the reliability
of the acquired results started and it is still going on. Therefore, each time
that the finite element method is applied, the acquired results should be
evaluated. Moreover, as the finite element techniques have greatly
developed since their first use together with the transient hot wire
technique, and as it is the first time that the Comsol Multiphysics software
is used in this area, it is crucial to verify the new finite element solution.
The governing equation in Comsol Multiphysics’ General Heat
Transfer application is ["COMSOL Multiphysics User's Guide, Version 3.2,"
2005]
Theoretical 73
T T  p
ts   C P       k   T   Q    C P  u  T   : S    (C.20)
t  T p t
where ts is a time–scaling coefficient, ρ is density, C P is the specific heat

capacity, T is the absolute temperature, k is the thermal conductivity, u is
the velocity vector, p is the pressure, S is the strain rate tensor, τ is the
viscous stress tensor, and Q includes heat sources other than viscous
heating.
The last three terms in the right hand side represents convective heat
transfer, viscous heating and pressure work respectively. For the
experimental setup used in the present work these terms are considered
zero and equation (C.20) becomes
T
ts   C P       k  T   Q (C.21)
t
In order to justify the modeling procedure and the reliability of the

finite element solution, a simple evaluation test model was created and the
results were compared to the analytical solution. The model consists of two
subdomains: one which is a part of the circular cross–section of the hot wire
and the other which is considered to be a fluid sample in which the wire is
immersed. For the verification procedure the wire subdomain has been
given the properties of tantalum and the fluid subdomain has the properties
of liquid toluene [Ramires et al., 2000] (see Table C.2). Moreover, the
assumption that the hot–wire diameter is constant in its whole length and
equal to 25 μm is also made.
Table C.2 Properties used in the test model subdomains
λ  C P q
Subdomain
 W  m1  K 1   kJ  m 3  K 1   W  m1 
Hot wire (Tantalum) 57.50 2,498 0.87
Fluid (Toluene at 303 K) 0.1291 1,475 -
In Figure C.3 a schematic representation of the model can be seen.

The dimensions of the test model can be virtually considered infinite as at
the edge points of the sample the temperature rise, at the end time of the
run, is zero. The wire is placed cylindrically symmetric inside the sample.
This symmetry condition allows the use of only one quadrant of the
74 Theoretical
geometry and therefore the computer potential gained by this simplification

is exploited in creating finer mesh.
Figure C.3 Schematic representation of the evaluation test model.
In the finite element analysis, it is significant for the created model to

be able to represent accurately in the solution, the local variation of
variables. The basic parameters responsible for this are the density of the
mesh, and the time steps used in the selected numerical solver. The
verification of the mesh density can be achieved either by comparing the
acquired results with measured experimental values or, if that is not
possible, by doing a convergence test. A convergence test is the procedure in
which the mesh of the model is refined so many times until the solution
converges to a stable value. Determining the correct coarseness of the mesh
is of great importance, especially to transient runs, as it influences the
solution mainly on very short times. The same procedure can be followed for
determining the time step.
Theoretical 75
In the present work, these two parameters were verified by

comparing the finite element results to that acquired by the analytical
solution. It is annotated that the analytical solution is calculated using the
first three terms of equation (C.10) and taking into consideration the
correction of equation (C.14). It is reminded that the analytical solution is
not accurate at very short times and therefore cannot be considered for
validation for time lower than 0.01 s. After, defining the mesh and the
solver’s time steps, the evaluation model is used for calculating the thermal
conductivity value from experimental data of a liquid.
a) Mesh determination
The first part of the validation procedure was to determine the
suitable density for the mesh. The mesh should have higher density in
regions where a higher temperature gradient is expected, e.g. the wire
subdomain, so that the local variation of temperature can be accurately
recorded. Comsol Multiphysics software provides by default nine predefined
mesh options, which automatically determine the parameters needed for
creating the mesh, e.g. the maximum element scaling factor, the element
growth rate, the mesh curvature factor, and mesh curvature cutoff. Each
option creates a finer mesh than the previous one.
In Figure C.4 a comparison between different mesh coarseness is
shown. Specifically, for the typical set of conditions, the temperature
increase acquired with the finite element method is substracted from the
value obtained from the analytical solution and the residuals are plotted
versus time. Apparently the values should be close to zero. In both solutions
the same heat flux input and material properties were used (see Table C.2)
and the temperature rise was calculated at distance ro  rw  12.5 μm from
the center of the wire. The calculations were done up to 10 s in order to have
one more time region in the logarithmic time axis and the overall
temperature rise of the wire for the properties used is 5 K.
It is easily noted from Figure C.4 that, as expected the difference
between the two solutions, becomes smaller as the mesh gets finer. The
comparison of the two temperatures rises is made for times greater than
0.01 s (where the ideal solution is more accurate) and for the very coarse
mesh, it can be seen that the finite element temperature rise, significantly
departs from the ideal solution (about 1%). As the mesh gets finer the model
converges more quickly to the analytical solution and for the finer meshes,
76 Theoretical
the temperature departure becomes almost constant and lower than 5 mK.
The mesh finally selected to be used in the model is the one that
consist of about 1,000 elements and has a maximum absolute deviation from
the ideal solution lower than 0.2 % for t  0.01 s . Although finer mesh could
give slightly better results, it was decided that the selected one was
sufficient for validating an experimental measured response, which is not of
higher accuracy. Additionally, in this way computational power is saved and
the model equations are solved in shorter times. It is noted that despite the
fact that the validation was not done for very small times, it is the author’s
belief that the accuracy of the model is also consistent for that time scale.
Figure C.4 Effect of the model’s mesh density on the calculated temperature rise.
Another important characteristic of the created mesh is the quality of

the elements from which it is consisted. The modeled geometry consists of
subdomains that have large differences in size, i.e thin wire and thick
sample, and therefore the quality of the elements has to be carefully
evaluated. The quality measured is mainly related to the aspect ratio, which
means that anisotropic elements can get low quality even though the
element shape is reasonable. In Comsol Multiphysics software the mesh
Theoretical 77
quality, q E , for triangular elements is calculated from
4 3A
qE  (C.22)
h  h22  h32
1
2
where A is the area of the triangular element, and h1, h2, and h3 are the
lengths of the triangle’s sides.
The element quality can be a number between 0 and 1. However, if
q E  0.3 , then the mesh quality should not affect the solution quality
["COMSOL Multiphysics User's Guide, Version 3.2," 2005]. For the created
model (see Figure C.5) the minimum element quality is q E  0.76 .
Figure C.5 2D mesh of the evaluation model, together with

illustration of the elements’ quality.
b) Determination of solver’s time–stepping method

The next part of the validation procedure of the finite element
78 Theoretical
solution was to verify the time stepping of the numerical solver. There are
three options related to the time steps used by the solvers:
i. The free option (default), for which the solver chooses its time steps
arbitrarily without taking into consideration the output times that the
user specifies.
ii. The strict option, for which the solver must use each time step
introduced by the user as output times and it adds extra time steps in
between if necessary.
iii. The intermediate option, for which the solver must take at least one time
step in each subinterval of the time list introduced by the user. It is
similar to the strict option, but gives somehow more freedom to the
solver.
Figure C.6 Comparison between different time stepping options of Comsol Multiphysics.
Hence, using the previously selected mesh, the model was solved for
each one of the three different time–stepping options and the resulting
temperature rise differences from the analytical solution are plotted in
Figure C.6. It is noted that all the three options converge to the same
Theoretical 79
values. This is expected since the problem is not complicated, and its
geometry is quite simple. However, the free and intermediate options show
that, at equally spaced time intervals, the deviation from the analytical
solution has a wave behavior, which can influence the thermal conductivity
calculation. On the contrary, the strict option results in a monotonically
decreasing deviation throughout the whole response and therefore it does
not affect the slope of the temperature rise, which is crucial for the
calculation of the thermal conductivity. Thus, the strict time step option was
selected for the present work.
Furthermore, it should be added that as the geometry of the created
model is simple, the accuracy of the solution do not depend on the linear
solver used. Thus, the solution of the PDE was performed using the time
dependent solver together with the direct linear solver UMFPACK, which is
a highly efficient and stable solver for symmetric and non–symmetric
systems of PDEs. For controlling the error in each integration step, the
relative and absolute tolerance of the linear solver was set to 10-5 and 10-4
respectively.
c) Application of the Finite Element Model on Liquid Measurement

The last validation check of the viability and consistency of the
created model was its employment for the calculation of the thermal
conductivity of liquid toluene from experimental data acquired by the
transient hot wire method. The unknown properties of the fluid are obtained
by means of inverse analysis. Specifically, in the test model the thermal
properties of the tantalum wire remained identical as before, whereas in the
sample’s material properties the experimental heat flux per unit length and
the ambient temperature were used. The remaining needed thermal
properties i.e. thermal conductivity and specific volumetric heat capacity of
the liquid, were given an estimated value for the specified temperature.
Then the model’s heat transfer equation is solved for each element and the
simulated temperature rise at the wire is compared to the experimental one.
Finally, the unknown thermal properties are calculated by iterative
adjustments of the estimated properties values until the Comsol
Multiphysics calculated temperature rise is superimposed satisfactory to the
experimental one, i.e. typically the slope of the linear trendline crossing the
residuals points of the temperature rise is of the order of 10-5. In Figure C.7
both (Comsol’s and experimental) temperature rises as function of time are
80 Theoretical
shown. It can be noted that the superimposition of the two curves is

excellent and the obtained value for the thermal conductivity of toluene
deviates 0.3 % from the value proposed by the Subcommittee on Transport
Properties of the International Union of Pure and Applied Chemistry
[Ramires et al., 2000].
Figure C.7 Typical temperature rise as a function of time for toluene.
From the above validation procedure it is obvious Comsol

Multiphysics can be combined successfully with the transient hot wire
technique and provide accurate results for the measurement of the thermal
conductivity of an homogeneous material surrounding the wire e.g. a liquid.
C2.2.2. Finite Element Model for Solids

The experimental setup used in this work for the measurement of the
thermal conductivity of solids consists of a multilayer sensor and therefore
new extra layers should be introduced to the above described model (a
detailed description of the experimental device is given in Chapter D). As
already mentioned, at the introduction of this subsection, the finite element
method has already been used to describe multilayer sensors, such as for the
measurement of solids [Gialou, 2004] or of melts [Peralta-Martinez, 2000;
Theoretical 81
Bilek, 2006].
In Figure C.8 a schematic representation of a cross–section of the
structure of the actual experimental arrangement is given. It can be easily
deduced that the temperature response of the wire will depend on the
thermal properties of each material layer. Therefore, the finite element
model should include each one of the shown layers. In general, the thermal
conductivity of a material is responsible for the rate at which the
temperature increases, while the specific volumetric heat capacity (density x
heat capacity product) is responsible for the magnitude of the increase.
Figure C.8 Schematic representation of the experimental setup

used for the measurement of solid samples.
In Figure C.9 a typical temperature response for that geometry is

shown. The temperature rise curve can be easily divided into three distinct
time regions, which depend on the thermal properties of the material
through which heat is propagated. Thus, at very short times the
temperature rise is affected by the properties of the wire, while in the
second time region the heat wave transfer depends on the properties of the
82 Theoretical
silicone paste used. The very thin Kapton film that exists over the silicone
layer has not any obvious effect on the slope of the temperature rise curve.
Finally, at longer times the thermal response is mainly affected by the
properties of the under measurement solid sample.
Figure C.9 Typical transient temperature response of the multilayer hot–wire sensor.
However, during the modeling procedure of the above setup it was

found that in order to perfectly describe, with finite elements, the
experimental temperature rise acquired by the hot wire, it is necessary to
introduce in the model two interface layers:
i. one between the wire and the silicone paste, and

ii. one between the Kapton film and the solid sample.
The interface layers are typical contact resistances (temperature

discontinuities) that inevitably appear in every contact between two solid
materials with quite different thermal properties, when a heat wave is
transferred from one solid to the other. The same behavior was also
observed in the case of measuring melts. Peralta-Martinez [Peralta-
Martinez, 2000] modeled the interface introducing a thin film with the
properties of air, while Bilek [Bilek, 2006] used interface coefficients, f, that
are calculated only from the ratio of the interfacial thermal conductivity to
Theoretical 83
the thickness of the layer and do not depend on the specific volumetric heat
capacity of the layer, as it is considered to be too thin.
In Comsol Multiphysics these discontinuities, which occur at
molecular level, are also modeled with two thin interfacial layers of very low
thermal conductivity. The effect of the interfacial layers can be
characterized by interface coefficients. It was found, as in the case of melts,
that for very thin interface layers the effect is independent of the product
density and specific heat capacity of the layer and depends only on its
thermal conductivity and thickness. It was also verified that the exact
nature of the properties of the interface layers does not affect the optimal
reproduction of the experimental curve, but only modifies the thickness of
the layers. Thus, the extra two thin layers used in the present work have
interface coefficients f w between the wire and the silicone paste, and f s
between the sensor and the solid sample. The values of the interface
coefficients depend on the material in contact.
W SP K S
fW  (C.23) fS  (C.24)
dW SP d K S
where W SP and dW  SP is the thermal conductivity and the thickness of the
interface layer between the wire and the silicone paste, while K S and
d K  S is the thermal conductivity and the thickness of the interface layer
between the Kapton film of the sensor and the solid sample.
In Figure C.10 a schematic representation of the Comsol Multiphysics
finite element model for solids is presented. As in the evaluation model the
cylindrical symmetry gives the possibility to use a quadrant of the cross–
section of the experimental setup. Moreover, the wire’s diameter is assumed
to be constant in its whole length.
The model consists of the following six subdomains–layers and
corresponding unknowns:
i. The hot wire, with the thermal properties of tantalum i.e. W and
  C P W .
ii. The interface layer between the wire and the silicone paste, which has
an interface coefficient f w .
iii. The silicone paste, with thermal properties SP and    C P SP .
iv. The Kapton film, with thermal properties K and   C P K
v. The interface layer between the experimental sensor and the solid
84 Theoretical
sample, which has an interface coefficient f s .

vi. The solid, with thermal properties S and    C P S .
Figure C.10 Schematic representation of the finite element model for solids.
In the above model there are seven unknown parameters:
- The interface coefficients f w and f s .

- The thickness of the silicone paste (which, in the newly designed sensor, is
set exactly equal to 1 mm – See Chapter D).
- The paste’s thermal properties [thermal conductivity, SP , and specific
volumetric heat capacity,    C P SP ].
- The solid’s thermal properties [thermal conductivity, S , and specific
volumetric heat capacity,    C P S ].
Furthermore, in Figure C.11 the mesh of the created Comsol

Multiphysics model for solids, together with the quality of the elements is
shown. The mesh density was selected following the aforementioned
described procedure. The 2D mesh consists of about 37,000 triangular
elements and has a minimum element quality of 0.6. Obviously, the mesh
Theoretical 85
density is higher close to the wire and at the interface layers than at the
edge of the solid sample.
Figure C.11 Full 2D mesh of model for solids, together with

illustration of the elements’ quality
The procedure to extract from experimental data the unknown

thermal properties of the paste and mainly of the solid sample is similar to
the one followed in the case of a liquid. The values are obtained using the
iterative inverse analysis in each region of the temperature rise curve. The
temperature rise is calculated at distance ro  rw from the center of the
wire. Therefore, from measurement points at low times t  0.5 s  the
interface coefficient f w [using equation (C.23)] and the silicone paste’s
thermal conductivity and specific volumetric heat capacity are obtained. The
accepted values are those for which the simulated temperature rise, T sim ,
is superimposed to the experimental one, Texp , and the trendline crossing
their residuals has a slope of the order of 10-5. Having obtained those values,
it is easy to apply the same procedure for longer times (up to 10 s or longer
times depending the material measured) in order to calculate the interface
coefficient f s [from equation (C.24)] and the thermal conductivity and the
product (density x heat capacity) of the solid. In Figure C.12 and Figure
86 Theoretical
C.13, the superimposition of the experimental to the simulated temperature

rise, versus the logarithm of time can be seen for both time regions for the
measurement of polymethyl methacrylate (PMMA) at 333 K.
It was mentioned that the beginning of the transient temperature rise
i.e. t  0.01 s , is affected by the wire–silicone paste interface, whereas the
influence of the sensor–solid sample interface can become apparent in the
temperature response, at the moment the heat wave leaves the silicone
paste and the Kapton film i.e. t  0.5 s . The effect of the Kapton film can
not be distinguished in the shape of the temperature rise curve, as its
thermal properties are close to those of the silicone paste and it is very thin.
The perfect agreement of both curves over the whole time scale can be seen
in Figure C.12 and Figure C.13.
Theoretical 87
Figure C.12 Temperature rise as a function of time (up to 1 s)

for polymethyl methacrylate (PMMA) at 333 K
and plot of the residuals between the experimental and simulated curve.
88 Theoretical
Figure C.13 Temperature rise as a function of time (up to 10 s)

for polymethyl methacrylate (PMMA) at 333 K and
plot of the residuals between the experimental and simulated curve.
Theoretical 89
The aforementioned procedure was repeated for a variety of

experimental parameters and settings, and the results reinforced the
conclusion that the created finite–element model accurately represents the
actual experimental setup. A typical screenshot of the temperature profile
obtained with Comsol Multiphysics for the case of PMMA can be seen in
Figure C.14. There are also included the temperature contours at the end of
the transient run, i.e. for t  10 s .
Figure C.14 Typical temperature profile obtained with Comsol Multiuphysics model
for PMMA at t = 10s.
90 Theoretical
C2.3. Sensitivity Analysis of the Finite Element Model

In the previous subsection, it has been shown that the created finite
element model can accurate describe and provide a very close simulation of
the experimental temperature response (to better than  0.05 K). However,
it is crucial to evaluate the effects, that the design parameters of the model
have on the behavior of the temperature rise, and therefore to the calculated
thermal conductivity value. Thus, in order to perform the sensitivity
analysis a typical solution of the model for solids was considered, using the
geometrical characteristics of the actual sensor i.e. tantalum wire diameter
25 μm, silicone paste thickness 1 mm and 50 mm width and Kapton film
thickness 25 μm and 50mm width. The multilayer sensor was considered to
be sandwiched between two solid samples of 20 mm thickness and 50 mm
width. The material properties used in the model are shown in Table C.3.
Table C.3 Material properties used in the model subdomains at 333 K
λ  C P f q
Subdomain
W  m
1
K 1
  kJ  m K3 1
 W  m 2
K 1
  W  m1 
Hot wire (Tantalum) 57.500 2,498 – 1.10
Wire – Silicone Interface – – 3.0.106 –
Silicone paste 0.172 1,800 – –
Kapton film 0.370 1,550 – –
Kapton – Sample Interface – – 7.7.103 –
PMMA 0.198 1,814 – –
The sensitivity analysis was performed by altering each one of the

above parameters (geometrical or material properties), and the new solution
 T mod  was compared to the initial one. The observed discrepancies on the
slope of the temperature rise curve allowed the evaluation of their influence,
on the calculated thermal conductivity value of the solid.
The effect of some parameters can be derived directly through the
equations of the analytical solution. Therefore, from equation (C.11) it is
obvious that the accuracy of the thermal conductivity of the sample around
the wire depends mainly on the accuracy of the heat flux q and slightly on
the wire radius ro . As the finite element model follows the “ideal” one, the
magnitude of the effect of these two parameters can be easily proved
through simulation. In Figure C.15, it can be seen that a  5 % change on
the wire radius ro just causes a temperature offset at the first time decade
Theoretical 91
and does not affect the slope of the temperature rise for later times.
Moreover, an applied -3 % change on the heat flux, results in an equal
decrease on the calculated thermal conductivity value of the silicone paste.
Figure C.15 Influence of the wire diameter and the heat flux on the temperature response
However, it is not feasible to assess the influence of the uncertainty of

all the parameters through the analytical solution. Thus, the effect of the
remaining model parameters is examined just through the finite element
model. The sensitivity analysis is performed in two time regions:
- up to 1 s, where it is analyzed the effect of the wire – silicone interface

coefficient, f w , and the previously mentioned influences of the heat flux
and the wire radius, and
- up to 10 s, where the effects of the paste’s thermal properties and of the
interface coefficient between the silicone and the sample, f s , are
evaluated. Moreover, in this time region the influence of the solid thermal
properties is also studied.
In Figure C.16 the effect of the interface coefficient f w is illustrated.

It is easily noted that it has a similar influence to the wire radius and
applies a temperature offset at very small times without affecting the
overall slope of the temperature curve.
92 Theoretical
Figure C.16 Influence of the wire – silicone interface coefficient and

of the silicone paste’s thermal conductivity on the temperature response
On the contrary, the thermal properties of the silicone paste lead to

significant alterations on the temperature rise curve on the whole time
region. As it can been seen from Figure C.16, each thermal property
parameter imposes a unique change on the temperature rise curve and
therefore it can be easily identified. The thermal conductivity of the paste
affects mainly the first two time decades and a change of  1 % results in a
deviation of the slope of the temperature rise of  0.4 %. The characteristic
peak between the second and third time region is attributed product of
density and specific heat capacity of the silicone and can not be suppressed
by another design parameter.
The next step is to examine the interface coefficient between the
sensor and the solid sample. It can be seen in Figure C.17 that the interface
coefficient is an important factor for fitting the temperature rise curve at
the end of the second and at the beginning of third time decade. The
departure of  5 % in the interfacial coefficient leads to a minor deviation of
 0.01% on the calculated thermal conductivity of the solid. At this point, it
should be mentioned that the influence of the Kapton film, that exist
between the silicone paste and the sample, has a negligible effect in the
Theoretical 93
fitting procedure, as it is too thin and its effect is compensated by the

interface layer.
Figure C.17 Influence of the solid’s thermal properties and

of the Kapton – Solid interface coefficient on the temperature response
The last part of the temperature response is mainly influenced by the

thermal properties of the solid. This statement is sustained by Figure C.17
where it can be observed the properties of the solid affects only on the last
time decade and it does not interfere on the heat transfer process inside the
sensor. Although the absolute temperature change is much lower than those
resulting from the alteration of the paste parameters, they are still easily
recognized, as they modify the slope of the temperature rise.
It is also important to mention that despite the fact that the slope of
the temperature curve in the last time decade significantly depends on the
thermal conductivity of the solid, the product of density and specific heat
capacity of the solid introduces also a deviation (see Figure C.17). Thus, a
change of  5 % in the value of the specific volumetric heat capacity of the
solid, results in  0.8 % difference on the slope of the temperature response.
Figure C.18 illustrates the residuals plot from the fitting of the
simulated temperature curve to the experimental data, for a measurement
94 Theoretical
of PMMA at 333 K. A change of  0.5 % on the thermal conductivity of the

solid can be easily discerned.
Figure C.18 Differences between experimental and simulated temperature rise for 0.5 %
change in the optimum thermal conductivity value for PMMA at 333 K.
Concluding, through the above presented sensitivity analysis it has

been demonstrated that the created model can be used to closely represent
the temporal temperature response of the transient hot wire device over
four order of time magnitude from the beginning of the experiment, during
which the heat wave crosses through four materials. Moreover, it was shown
that each model parameter affects a specific time region and has a unique
influence on the slope of the temperature rise curve and therefore can be
identified and calculated. Hence, the above analysis provides evidence that
employing the created finite element model with the experimental data, can
calculate the thermal conductivity of the solid sample with an uncertainty
better than  0.5 %. However, this uncertainty refers only to the finite
element method and has to be taken into account together with the
uncertainty of the experimental procedure (described on Chapter D) in order
to have the overall uncertainty of the method.
Theoretical 95
C3. Summary
In this chapter, the fundamental theoretical “ideal” model on which
the transient hot wire technique is based, was presented together with its
analytical solution and its corresponding corrections. Furthermore, a large
part of the chapter is attributed to the finite element method and its
application with the transient hot wire technique for calculating the
thermal conductivity of materials. The Comsol Multiphysics finite element
software was used to create a model to describe the actual experimental
setup and a sensitivity analysis for its parameters was performed.
The main concluding points of the chapter that have to be highlighted
are:
i. In the case of solid measurements, the transient hot wire device consists
of a multilayer sensor and therefore the analytical solution of the “ideal”
model cannot describe the experimental measurement. Thus, the
numerical method of finite elements was employed for the solution of
the complicated partial differential equations that describe the heat
transfer in the experimental setup.
ii. Older finite element model written in Fortran programming language
shown several deficiencies as it lacks of flexibility in terms of suitable
size of finite elements, time–stepping options, implementation of new
layers in the model and post–processing of the results. Moreover, it can
not accurate simulate the temperature rise for time t  0.1 s .
iii. The created Comsol Multiphysics model has been used successfully for
the calculation of the thermal conductivity of a liquid (Toluene) and a
solid (PMMA). The experimental temperature rise was closely
reproduced (  50 mK) from the beginning of the time scale.
iv. Finally, the sensitivity analysis shown that the thermal conductivity
values obtained from the model have an uncertainty better than  0.5%.
.
D
Experimental Procedure
The purpose of this chapter is to present all the stages of the

procedure followed for measuring the thermal conductivity of solids using
the transient hot wire method. Initially the newly designed, constructed
transient hot–wire sensor employed is analytically described and its main
improvements compared to the older one are illustrated. A large section of
the chapter is attributed to the electronic components consisting the
experimental setup. An historical review of previously used Wheatstone
bridge circuits is presented and the current configuration together with the
data acquisition system and the working equations for calculating the
experimental temperature rise of the hot wire are analysed in detail.
Another important section of the chapter is focused on how the
experimentally acquired voltage response from the sensor is processed to
derive the temperature rise of the hot wire and consequently employed
together with Comsol Multiphysics software to calculate the thermal
conductivity value of a material. Following this, the validation of the
technique is presented. The validation procedure is based on the fact that
the same transient hot wire sensor and methodology can be used for
measuring the thermal conductivity of reference liquids, such as toluene.
The last part of the chapter is dedicated to the analysis of the sources of
uncertainty introduced during the experiment and the overall uncertainty of
the thermal conductivity is calculated.
98 Experimental Procedure
D1. Experimental Configuration

The experimental configuration employed in the present work is an
improvement–modification of the setup used by Gialou for the measurement
of the thermal conductivity of solids [Gialou, 2004]. Based on the pioneering
idea of Harman [Haarman, 1971], the transient hot wire sensor consists of
two thin tantalum wires of 25 μm diameter, which are identical except for
their length, in order to compensate the end effects of the wires. The wires
act both:
a. as a heat source, producing a time–dependent field within the material,

and
b. as a thermometer, registering the temporal resistance change and thus
its temperature rise.
Moreover, in order to minimize the contact resistance between the wire

sensor and the solid, the wire sensor is placed inside a soft silicone paste.
According to the analysis of the “ideal” model presented in Chapter C,
the hot wire was selected to be as thin as possible in order to minimise the
corrections that need to be applied. Thus, taking also into account the
investigation on the wire material performed by Charitidou [Charitidou,
1990], a tantalum wire of high purity (99.9 %) and with nominal diameter of
25 μm was supplied from GoodFellow Metals Ltd., U.K. The wires actual
diameter was measured with Scanning Electron Microscopy (SEM) and it
was found to be 25.83 μm.
In order to further improve the uncertainty of the instrument, the
following modifications were carried out in the original experimental setup
of Gialou [Gialou, 2004].
i. The support (leading) wires, on which the thin hot wires are spot
welded, are entirely made of tantalum instead of the previous assembly
of tantalum wire welded on Chromel – Nickel wires. The advantage of
this modification is that there are no welded joints of different
materials, which can influence the uncertainty of the measurements
(contact resistance, extra e.m.f., etc).
ii. The new sensor is equipped on the top and the bottom of the hot wires
with two constant 1 mm–thick–SS spacers that ensures that the solid
samples are kept precisely 1 mm apart, and hence the paste’s thickness
Experimental Procedure 99
is also constant and equal to 1 mm. In this way, one prime source of
uncertainty (paste’s thickness) is eliminated.
iii. The use of continuous leading wires without any joint, as well as the
construction and use of the sensor connector support (see Figure D.1
and Figure D.2), make the whole sensor more robust than before.
The above mentioned differences will be made clear in the following

subsection, where the details of the design of the new sensor are presented.
D1.1. Sensor Design

In Figure D.1 and Figure D.2 the new sensor is illustrated. The new
transient hot wire sensor consists of two 25–μm–diameter Ta wires of
lengths 20 mm and 50 mm respectively (1), faced one over the other, as
shown in Figure D.1. The 25 μm–diameter is the thinnest Ta wire produced.
The lengths of the wires were measured with a cathetometer. According to
theory of the “ideal” model the wires should be placed vertically inside the
under measurement material. Thus, both hot wires are spot welded on three
1 mm Ta leading wires which are flattened at their ends (2). These thick
tantalum wires act both as electrical contacts and supports. The three
flattened ends of the leading wires form a plane in which the hot wire rests.
The leading wires, whose lengths are measured before the sensor assembly,
continue inside the 7–mm–diameter SS support tube (3) and end up to the
connecting plug of the sensor. In order to avoid electric contact between the
leading wires and the support tube, four hole round ceramic tube insulators
of 4 mm diameter and 5 cm length are used (4). On the support tube three
holes of 2 mm are made on the specific points from where a length of 30 mm
of the leading wires comes out (2). The spot welds of the thin Ta wires on the
thick ones was performed using a tantalum probe, in order to avoid the
unnecessary presence of other material than Ta in the welding.
As already mentioned, another important innovation applied was the
use of two stainless steel strips of 1 mm thickness on both ends of the hot
wire sensor (5). The strips have a length of 30 mm and are placed at a
distance of 15 mm from the first support wire and at 15 mm after the last
one, as not to disturb the heat flow. These strips are used to ensure that the
thickness of the silicone layer of the sensor (6) will be always equal to
exactly 1 mm. Although this idea seems too simple, it solved completely the
problem of having a uniform silicone layer of known thickness.

On the upper part of the supporting tube a special 25–mm–diameter
stainless steel mounting support (7) for the connecting plug (8) of the sensor
is used (see Figure D.2). This support ensures that the sensor can be
plugged or unplugged to the cables connecting to the electronic components,
without stressing the connections of the thick Ta wires on the connecting
plug. The connecting plug is stabilized on the mounting support with a 3.5–
mm–diameter screw, whereas the whole assembly is fixed on the supporting
tube with a similar screw (9).
Figure D.1 Schematic representation of the new experimental sensor with wires.
Under the above mentioned mounting support, there is circular cap

(10) of 910 mm diameter. The cap has five through all holes; one for the
supporting tube (3) and the rest are used only in the case of measuring
solids (two for the thermocouples (11) and two for the extra supporting
metallic rods (12)). The cap is fixed on the supporting tube by using a
5–mm–diameter screw.
It should be noted that the sensor at this stage of construction (i.e.
support tube and bare Ta wires) is employed for the measurement of the
thermal conductivity of liquids (see Section D5).
Figure D.2 Schematic representation of the whole experimental assembly.

Figure D.3 Construction details of the experimental sensor.
For the measurement of solids the stainless steel cylindrical shell of

100 mm length and 77 mm diameter, used by Gialou, was modified and
used. The shell consists of two half cylinders (13), each one of which holds a
solid block sample of dimensions 10x5x2 cm3 (14). The two half parts are
kept tightly together with screws at three points (15), whereas at the bottom
part of the shell a 3 mm thick cap (17) is used to sustain the samples in
place before tightening them. The only modification made on the shell was
the widening of the track that existed for the supporting tube to a depth of
50 mm. Finally, in order to control the vertical temperature gradient on the
cylindrical shell, one of the two half cylinders has two holes (11), where two
PT–100 thermometers are embedded at different depths. In Figure D.3 the
dimensions of the cylindrical shell can be seen.
D1.1.1. Sensor Assembly/Construction

The procedure followed for assembling the new sensor for measuring
the thermal conductivity of solids has three distinct phases.
a) Hot wire connections

Assuming that the support tube with the thick 1–mm–diameter Ta
leading wires is prepared, the next thing to be made is to spot weld the thin
tantalum hot–wires on the three flattened end points of the tantalum
supports. The hot–wires should be kept stretched in order to form a vertical
line source and to avoid the influence of convection during measurement in
liquids. At this stage, the lengths of the hot wires are measured using the
cathetometer and the operation of the sensor is validated by measuring the
thermal conductivity of toluene.
b) Silicone paste layer

After the successful employment of the sensor on the measurement of
the thermal conductivity liquids, it is left to dry up. Then, in a specially
made base (see Figure D.4), two 25.4–μm–thick polyimide films (Kapton 100
MT, DuPont®, Luxembourg) 100x50 mm2 are held while silicone paste is
laid carefully and uniformly over them. Two different silicone pastes were
used during the present work depending on the measured material. The
first one was AS1803 silicone paste (ACC Silicones), which was found to
have a thermal conductivity value of 1.16 W  m 1  K 1 at room temperature,
and the second one was a high–temperature red RTV silicone (Type 650,
BORO, VersaChem, U.S.A.) which was found to have a thermal conductivity
value of 0.177 W  m 1  K 1 at the same conditions. The measurement of the
thermal conductivity of the silicone pastes is described in section E1.1.
Following that, the sensor with the wires is placed on top of one of the
silicone–paste–covered polyimide films and the other film is placed over it.
Then the specially made base is squeezed tight exactly to 1 mm because of

the 1–mm–spacers (see Figure D.4). After about some days, or even a week
(depending on which silicone paste is used), the silicone paste hardens but
still retains its elastic properties. This produces an elastic layer of exactly 1
mm thickness with Kapton film on the outside, and inside the silicone paste
with the two wires. Furthermore, this assembly ensures that the Ta wires
are in the middle of the silicone layer.
c) Solid measurement
In order to perform a measurement of the thermal conductivity of a
solid material, the hot wire sensor with the silicone layer is squeezed
between two specimens of the solid. Specifically, each specimen should have
at least a thickness of 5 mm and dimensions of 100x50 mm2. One solid
sample is placed on the half cylinder which is connected to the upper
circular cap, then the hot wire sensor is placed over it and it is squeezed
with the other half cylinder of the shell containing the second specimen.
Figure D.4 Schematic representation of the special made base for the silicone paste.
D1.1.2. Temperature Coefficient of Resistance

During the heating of the tantalum wire its resistance varies
depending on temperature. Thus, for a more accurate calculation of this
resistance change a calibration of the tantalum wire was performed using a
standard platinum resistance thermometer (Type 5187SA, No 250293,
Tinsley and Co. Ltd., U.K.) with Class A calibration to ±1 mK for the
temperature measurement and a type KB5 four terminal resistance bridge
(Cropico Ltd., U.K.) with uncertainty of ±0.01 % in the resistances
measurement. The calibration was made from room temperature up to
573 K. It should be noted that the calibration procedure was performed
twice, one for the thin tantalum wire of 25 μm and one for the thick
tantalum wire of 1 mm used in the sensor. The least–squares regression
analysis led to the following correlations:
i. Thin tantalum wire of 25 μm
R25 m T 
 1  3.366 103  T  To   1.64747 107  T  To  .
2
(D.1)
R To 
ii. Thick tantalum wire of 1 mm .
R1mm T 
 1  3.435 103  T  To   4.88444  107  T  To  . .
2
(D.2)
R To 
where To denotes the reference temperature (273.15 K) and R To  is the

resistance of the tantalum material at the reference temperature To
(17.7491 Ohm), which is calculated from the equation
L
R To   Ta  , (D.3)
S
where Ta represents the specific resistance of tantalum, L the length of the
wire, and S the cross–sectional area of the wire. Using equation (D.3), the
specific resistance of tantalum acquired from the above procedure is
Ta  13.65 μOhm·cm, which is in accordance with the value proposed by the
supplier company (Goodfellow Metals Ltd., U.K.) Ta  13.5 μOhm·cm.
According to the calibration results [equations (D.1) and (D.2)], it is
obvious that the first and second temperature–resistance coefficients of the
thin (25 μm) and the thick (1 mm) wire have a small difference, which is
attributed to the different material batches or to small differences in the
purity of the tantalum wires. Consequently, it was considered necessary to
use in each case the measured temperature–resistance coefficients.
Finally, it should be noted that the temperature rise of the hot wire is
calculated using equation (D.1) with an uncertainty better than ±20 mK, as
it results from the measurements. The uncertainty in the resistances of the
thick wires, although is of similar magnitude, is immaterial, as the wire is
thick and thus its resistance is very small compared to the hot wire’s.
D1.2. Temperature Controller

In order to perform thermal conductivity measurements in a wide
temperature range, the whole experimental arrangement was placed in the
center of an accurate, vertical three zone tubular furnace (Carbolite, TVS
12). According to the manufacturer, the furnace temperature can reach up
to 1,473 K, and it is equipped with controllers for checking and regulating
the temperature in each one of the three zones separately. In Figure D.5 the
furnace with the control system is depicted.
Figure D.5 Three zone temperature furnace.
The temperature stability across the furnace was evaluated using two
platinum resistance thermometers (PT-100) embedded in the top and
bottom part of the cylindrical experimental setup. The thermometers were
also calibrated versus a standard platinum resistance thermometer (Type

5187SA, No 250293, Tinsley and Co. Ltd., U.K.) to a maximum uncertainty
of less than  80 mK. The standard platinum resistance thermometer used,
has a Class A (  1 mK) certification from the National Physics Laboratory,
U.K. All temperatures refer to International Temperature Scale of 1990
(ITS-90) [Preston-Thomas, 1990]. Moreover, during the calibration
procedure the thermometers’ resistances were measured by a KB5 accurate
(±0.01 %) four terminal resistance bridge (Cropico Ltd., U.K.). The acquired
working equations of the thermometers are:
i. Thermometer at the top part
RT T 
 1  3.90975 103  T  To   6.29836 107  T  To  .
2
(D.4)
RT To 
ii. Thermometer at the bottom part .
RB T 
 1  3.9080 103  T  To   6.27837 107  T  To  . .
2
(D.5)
RB To 
where To  273.15 K denotes the reference temperature and

RT To   100.62 Ohm and RB To   100.73 Ohm are the resistances of the
top and bottom thermometer respectively at the reference temperature To.
Preliminary measurements with the platinum resistance
thermometers placed in the position of the wires, showed a 0.4 K maximum
difference between the top and the bottom of the sensor support cylinder
and a 0.05 K maximum difference radially. These differences have no
significant effect in the final quoted uncertainty of the thermal–conductivity
measurement. The thermometers’ resistances during the experimental
measurements were recorded by a digital high precision thermometer
(Tempmaster–100, Labfacility, U.K.).
D2. Electronic Components

As has been already mentioned, in the transient hot–wire technique,
the temporal rise of a thin wire placed in the test material, initially at
equilibrium, is observed following the application of a step voltage across
the wire. From equation (C.10) of the “ideal” solution, the experimental data
needed for the calculation of the thermal conductivity are: the temperature
rise at the wire inside the under measurement material,   ro ,t  , the total
experimental time, t, during which the temperature rise takes place, and
the heat flux per unit length, q, that is transferred from the wire to the
surrounding material. Thus, it can be easily deduced that the data
acquisition system of the experimental configuration should be able to
transmit across the wire a constant per unit length heat flux and to record
its temporal resistance change over a specified time.
Furthermore, an important feature of the electronic circuit employed
is that it should be able to minimize or eliminate the end effects of the wire.
Solution to this problem was given by the pioneering work of Haarman
[Haarman, 1971], who used a Wheatstone bridge to measure the resistance
difference of two wires which were identical except for their length. In that
way, the end effects of the wires were compensated. Since then, and
following the advances in electronics and computers, the electronic bridges
employed together with transient hot wire instruments have tremendously
been developed. In the following sections, a brief presentation of the
historical development of the electronic acquisition system is presented, and
the present bridge circuit is analysed in detail.
D2.1. Historical Development

We should start by saying that all electronic bridges that are
described here and cover more than 40 years of discussions and modelling,
were the ingenious work of Malkom Dix at Imperial College, U.K.
One of the first Wheatstone bridges developed is the one shown in
Figure D.6. This bridge was successfully employed in the 1970's, by many
scientists [de Groot et al., 1974; Nieto de Castro et al., 1976; Kestin et al.,
1980; Assael et al., 1981]. As it can be seen from the bridge’s circuit, the
long wire of resistance, RL, and the short of resistance, RS, were placed on
the two left arms of the bridge. On the right side, the upper arm of the
bridge consisted of resistances R3, R6 and five other parallel resistances,
whereas the lower right arm included the resistances R4 and R5. The bridge
balance was measured between points A and F. The values of resistances R1
and R2 were selected so that the bridge was on the verge of reaching
equilibrium. Applying a voltage to the circuit, the five parallel resistances

were introduced to the circuit. As current started passing through the wires,
their resistance increase and the bridge reaches the first balance point.
Upon reaching equilibrium, time is monitored and one of the resistances of
the upper right arm is disconnected (through a relay), increasing in that
way the arm’s total resistance and throwing the bridge out of balance.
However, as the heating of the wires continues, the bridge reaches a new
equilibrium point. As before, at this point, time is recorded and a second
relay of the upper right arm is opened extracting one more resistance from
the circuit. The process is repeated six times and then the bridge is
automatically returned to its rest position.
Figure D.6 Wheatstone bridge circuit used in the 1970’s

[de Groot et al., 1974; Nieto de Castro et al., 1976; Kestin et al., 1980; Assael et al., 1981].
Therefore, the above–mentioned electronic bridge allowed the

measurement of six equilibrium points (wires’ resistances versus time) for a
measuring circle of 1 s. For more points, the same procedure is repeated
after the whole setup reached a thermodynamic steady state. The resistance
R8 allows the change of the bridge balance times during the iterative
measuring circles. The main drawbacks of this bridge were the complexity of
the electronic circuit, the need for many high accuracy resistances and the
limit to only six equilibrium points per measuring circle.
Figure D.7 Improved Wheatstone bridge design [Maitland et al., 1986; Assael et al., 1988].
Thus, in order to be able to obtain more resistance measurements at

one experimental run, a new improved bridge’s design was employed in
1980’s and 1990’s [Maitland et al., 1986; Assael et al., 1988]. This design
was also based on the idea of altering a bridge’s arm during the measuring
procedure. However, instead of changing the resistances of the upper right
arm, the voltage of the lower right arm was changed to predetermined, by a
computer, values. It can be noted in Figure D.7, that the left part of the
bridge is similar to the previous design. On the right arm of the bridge, the
upper branch consisted of resistance R3, while the bottom one complied
resistances R4 and R5. Resistances R1 and R2 were selected so that the
bridge was close to its equilibrium. The bridge’s balance was detected
between points A and B by a sensitive high–impedance comparator. Prior to
a measurement, a sequence of voltage values VE was stored to the computer
controlling the bridge. The measuring circle started (and therefore the time
sequence began) by applying to the circuit a voltage Vo through a high–
precision source of DC voltage. The current passing through the wires
increased theirs resistances and then at some time the bridge was balanced.
At the balance point, the time and the resistances of the wires were
registered to the computer and the first value of the step voltage VE was
applied to the bridge by 12-bit D/A converter. The established voltage VE
lead the bridge out of equilibrium, but as the wires kept heating, rapidly the
bridge reached a new balance point. The procedure was repeated and this
electronic setup was able to register up to 32 equilibrium points for a run up
to 1 s.
D2.2. Present Bridge Circuit and Data Acquisition System

The automatic electronic bridge used in the present work was
employed by Gialou for the measurements of solids [Gialou, 2004] and has
also been used by Metaxa for the measurement of nanofluids [Metaxa,
2006]. Although a detailed description of the electronic circuit can be found
elsewhere [Assael et al., 2002; Gialou, 2004], for completeness reasons, it is
included also in the present work.
The fundamental principle of Wheatstone bridge is the same as the
ones previously mentioned, but differs in the calculation of the wires’
resistances. In the present circuit the change of the wires’ resistances over
time is acquired through a known standard resistance Rstd , which is placed
on the same bridge’s arm with the wires. In Figure D.8 a schematic
representation of the automatic Wheatstone bridge is illustrated.
The main characteristics of the bridge include the direct
measurement of the out of balance voltage across the circuit during a
transient run and the ability to monitor and register experimental
measurement values from 20 μs from the initiation of the wires heating at
t  0 , leading to a large number of data points. Furthermore, the time
resolution is  1 μs.
Figure D.8 Schematic of the automatic Wheatstone bridge circuit used in the present work
[Gialou, 2004; Metaxa, 2006].
As before, the short and the long wire, of resistance RS and RL

respectively, are placed on the two left branches of the bridge. On the same
branches two variable resistors R1 and R2 (Resistance Box RBC5-A, Cropico
ltd., U.K.) are included, whereas on the right branches of the bridge there
are two identical, high accuracy, resistors R3. Resistances RA, RB and RC are
the lead resistances of the two wires, and Rstd is a standard resistor of
10 Ohm (Tinsley).
In order to perform an experimental transient run, a first balance of
the bridge should be achieved by adjusting the variable resistors R1 and R2,
under a low voltage (0.5 V), so that typically R1  RL and R2  RS . For better
resolution during the initial balance of the bridge, a second variable resistor
R11 (Resistance Box RBB6-B, Cropico Ltd., U.K.) has been connected in
parallel with resistor R1 (due to figure simplification the resistance R11 is
not shown). In Table D.1 the values and the uncertainty of the resistors
used in the bridge are shown. All the resistors are of very low uncertainty,
as the uncertainty of the temperature rise of the wires depends on the
accurate measurement of their resistances.
Table D.1 Details of Wheatstone’s bridge resistors.
Resistance Uncertainty
Resistor Manufacturer – Type
Ohm  %
 5 at the first 3 digits
R1 11,111.00 Cropico - RBC5-A
 10 at the last 2 digits
R11 1,122.11  0.05 Cropico - RBB6-B
Rstd 10  0.05 Tinsley
After achieving the initial bridge balance, two MOFSET switches

activate the bridge and a constant voltage, VS, is applied across its circuit
from a DC power supply (TSP3222, Thurlby Thandar Ltd., U.K.). The
output signal is then amplified and consequently converted by an analogue–
to–digital (A/D) converter (PC30AT, Amplicon Liveline Ltd., U.K.) and
finally stored to the computer workstation controlling the experimental
assembly. The A/D converter has a maximum conversion rate of 50 kHz
(20 μs) and time accuracy  1 μs, allowing the acquisition of a large number
of experimental data points. Typically, for measurement of liquids 500 data
points are acquired, with a sampling rate of 2 ms, while for solids 1,000 data
points are registered, with the sampling rate depending on the overall time
duration of the transient run.
For voltage measurements a seven–digit high–resolution digital
multimeter is used (HP 34401A, Multimeter Quick Reference, Hewlett
Packard, U.S.A. The voltmeter is employed to measure the supply voltage
across the bridge during the transient run, and during the steady state
measurements to measure the voltage across the bridge resistances,
including the standard resistance Rstd.
It should be mentioned that all the electronics are operated
automatically through a computer controlled data acquisition program
written in QuickBasic language. The computer program controls the data
acquisition card (PC30AT, Amplicon Liveline Ltd., U.K.) and via a controller
card (Model 488–PC2 IEEE 488.2, Amplicon Liveline Ltd., U.K. ) controls
the electronic bridge, the digital multimeter and the power supply. The
Wheatstone bridge with the data acquisition system is shown in Figure D.9.
Finally, it is noted that the delay introduced from the switching of the
devices is much less than few microseconds and therefore minor difference
between the starting points do not influence the measured thermal
conductivity value.
Figure D.9 Wheatstone bridge and data acquisition system.

D3. Working Equations

In this section all the equations needed for the calculation of the
thermal conductivity from the experimental data are given. As mentioned
earlier, in order to calculate the thermal conductivity three variables must
be measured or calculated: the temperature rise at the wire inside the
material under measurement,   ro ,t  , the experimental time, t, during
which the temperature rise takes place, and the heat flux per unit length, q,
that is transferred from the wire to the surrounding material. From these
variables only the experimental time is acquired directly from the
experimental data. The other two variables are calculated from the working
equations of the experimental setup. From the experimentally obtained
resistance differences of the long wire from the short one  RLo  RSo  , the
resistance change of the middle portion of an infinite long wire can be
calculated, eliminating in this way the end effects. This resistance change is
then used for obtaining the temperature rise at the wire, and the heat flux.
D3.1. The Bridge Equations

The bridge circuit is shown in Figure D.8. The automatic Wheatstone
bridge has been designed to impose on the wires, a known magnitude
constant current flow, ensuring the generation of a constant heat flux, and
to record the evolution of the resistance change of the wires. Moreover, the
end effects of axial heat transfer in the wires are automatically compensated
by the bridge and the resistance change of a finite segment of an infinite
wire can thus be calculated. It should be noted that the end effects refer to
small unaccounted resistance arising from the connections of the wires to
the terminals of the sensor.
The end effects elimination is based on the assumption that the
resistance of the long wire at a time t, RLo t  , can be expressed as the sum
of the short wire’s resistance, RSo t  , and the “working” resistance, RWo t  ,
of a finite portion of an infinite wire at a time t :
RLo t   RSo t   RWo t  . (D.6)
Thus, using within the Wheatstone bridge two identical wires except
for theirs length (the wires’ supports are also identical), the end effects are
eliminated and the working resistance refers to the middle section of one of
them, i.e. refers to a wire with no ends.
The resistances’ change of the wires after a time t, during which a
constant heat flux per unit length passed through them, are calculated from
the following equations:
 RLo t   RLo t   RLo  0 (D.7)
 RSo t   RSo t   RSo  0 (D.8)
 RWo t   RWo t   RWo  0 (D.9)
 RLo t    RSo t    RWo t  , (D.10)
where  RLo t  , and  RSo t  represent the resistance change of the long and
short wire respectively after time t, and  RWo t  denotes the resistance
change of the “working” portion of a wire of infinite length after time t.
Furthermore, RLo  0  , and RSo  0  represent the resistances of the long and
the short wire at equilibrium condition prior to measurements, and RWo  0 
denotes the working resistance of a finite portion of an infinite wire at the
same time.
According to the bridge circuit (Figure D.8) the total equivalent
resistance of the left arm of the Wheatstone bridge at a time t, R t  , can
be expressed as:
R t   Rstd  R2  RB  RLo t   RSo t   RA  R1 , (D.11)
where R1 and R2 are two variable resistors of high accuracy, whose

specifications are mentioned in Table D.1, RA and RB are the lead
resistances of the two wires, and Rstd is a standard resistor of 10 Ohm
(Tinsley).
It should be noted that the lead resistances RA, RB and RC includes
both, the resistances of the thick tantalum wire (1 mm) supports on which
the hot wire are spot welded (Figure D.1) and the resistances of the cables
connecting the sensor to the Wheatstone bridge. The last ones are measured
during the sensor assembly and have a constant resistance value as they are
not placed inside the temperature furnace (see section D1.2), whereas the
resistances of the supporting tantalum wires changes with temperature

according to equation (D.2).
From the circuit of Figure D.8 it is obtained:
V AB t   I 3 t   Rstd (D.12)
V AC t    I 3 t    R std  R 2  (D.13)
V AD t   I 3 t   Rstd  R2  RB  RLo t   I 5 t   RC (D.14)
V AE t   I 3 t   Rstd  R2  RB  RLo t    I 1 t   R A  RSo t   (D.15)
V AF t   I 3 t   Rstd  R2  RB  RLo t    I 1 t   RA  RSo t   R1  , (D.16)
where V t  denotes the voltage between the mentioned terminals and I t 

the electric current at a time t.
Applying Kirchoff’s law to the same circuit it is acquired:
I t   I 1 t   I 2 t  (D.17)
I 1 t   I 3 t   I 5 t  (D.18)
I 4 t   I 2 t   I 5 t  . (D.19)
It is obvious that when the bridge is in balance then
I 5 t   0 , (D.20)
and from equation (D.18) it is
I1 t   I 3 t  . (D.21)
Assuming that at time t = 0 a VS voltage is applied to terminals A and

F and using equations (D.11), (D.14) and (D.21) it is
V AD  0  Rstd  R2  RB  RLo  0 
 (D.22)
VS R  0 
and through equation (D.6)
V AD  0  Rstd  R2  RB  RWo  0   RSo  0  

 . (D.23)
VS R  0 
After time t, the above mentioned ratio becomes
V AD t  Rstd  R2  RB  RLo t 
 . (D.24)
VS R t 
Moreover, as it will be show in section D3.1.1, the ratio of the long

wire’s resistance to the resistance of the short one can be considered, to a
very good zero approximation, constant during the experimental sampling,
and thus the following equation for the resistance change of the short wire
is acquired
RSo  0 
 RSo (t )   RWo (t )  . (D.25)
RWo  0 
Therefore, employing equations (D.6) to (D.11) and (D.25), equation

(D.24) becomes
 R 0 
Rstd  R 2  R B   RWo  0   R So  0     RWo t    So 1
V AD t   R 0  
Wo  . (D.26)

VS  R 0 
Rstd  R 2  R B  2R So  0   RWo  0    RWo t    2  So 1
 R 0 
 Wo 
Hence, the voltage change at the ends A and D, δVAD(t), at time t is

calculated from
V AD t  V AD t  V AD  0 
  . (D.27)
VS VS VS
From the previous equation and using equations (D.23) and (D.26) the
change in the resistance of a hypothetical segment of an infinite “working”
wire,  RW t  , at a time t can be expressed as a function of the ratio of the
out–of–balance voltage to the supply voltage at time t, V AD t  /V S , the total
resistance of the left arm of the Wheatstone bridge at zero time, R  0  ,
and the ratio of the short to the “working” wire resistance RSo  0  / RWo  0  at
zero time,
V AD t 
 R 2  0 
VS
 RWo t  
 R 0  V t   RSo  0    R 0 
R  0    So  1   R  0   AD 2  1   RF  0    2  So 1
 R 0  VS  R 0   R 0 
 Wo   Wo   Wo 
(D.28)
where RF  0  denotes the sum of the bottom left branch resistances at zero
time
RF  0  Rstd  R2  RB  RLo  0 . (D.29)
According to equation (D.28) one of the main goals of the Wheatstone

bridge is achieved, i.e. the evolution of the resistance change of the middle
portion of a wire, which acts as a finite segment of an infinitely long wire,
can be recorded.
Furthermore, as the Wheatstone bridge should supply constant heat
flux per unit length to the wires and the resistances of both wires change
with time, then consequently the electric current of the circuit will vary as a
function of time. Therefore at a time t, the current I 1 t  of the circuit of
Figure D.8 is
VS
I 1 t   , (D.30)
R t 
and using (D.7), (D.8) and (D.11) and replacing  RSo t  from equation
(D.25) it is
VS
I 1 t   . (D.31)
 R 0 
R  0    RWo (t )  1  2  So 

 RWo  0  
In addition, the heat flux per unit length at the same time t for the
“working” wire is
RWo (t )
q t   I 12 t   , (D.32)
LL  LS
and therefore
2
 
 
 VS  RLo (t )  RSo t 
q t      . (D.33)

 R  0    R (t )  1  2  R  0   L  LS
 
So L


Wo 
 RWo  0  

D3.1.1. Constant Resistance Fraction [RLo(0)/RSo(0)]

For the calculation of the resistance change of a hypothetical segment
of an infinite “working” wire,  RW t  , at a time t the hypothesis that the
ratio RLo  0  / RSo  0  is constant was made. Despite the fact that the proof of
this hypothesis is analyzed elsewhere [Gialou, 2004], it was considered
necessary to be repeated here in this work for completeness reasons.
It is obvious that at zero time the fraction of the long wire’s resistance
to the resistance of the short one is constant,
R Lo  0 
 constant , (D.34)
RSo  0 
and it can be considered equal to the ratio of the lengths of the wires
RLo  0  LL
 . (D.35)
RSo  0  LS
After time t, the resistances’ ratio becomes
RLo  0    RLo (t )
, (D.36)
RSo  0    RSo (t )
where, as mentioned before,  RLo t  , and  RSo t  represent the resistance

change of the long and short wire respectively after time t.
The above resistances’ changes can be expressed using a pseudo–
linear temperature coefficient as
 RLo t   L T ,Tinit   RLo  0  TL (t ) (D.37)

 RSo t   L T ,Tinit   RSo  0  TS (t ) , (D.38)
where  L T ,Tinit  is a pseudo–linear temperature coefficient calculated from

the calibration of the tantalum wires, and TL t  and TS t  denote the
experimental temperature rise from the equilibrium temperature prior to a
measurement, Tinit , of the long and short wire respectively at a time t.
Therefore, from equation (D.36) it is obtained
R Lo  0    L T ,Tinit   R Lo  0   TL (t ) R Lo  0   1   L T ,Tinit   TL (t )

 , (D.39)
RSo  0    L T ,Tinit   RSo  0   TS (t ) RSo  0   1   L T ,Tinit   TS (t )
and using equation (D.35) it is
RLo  0    R Lo t  LL 1   L T ,Tinit   TL (t )

  . (D.40)
RSo  0    RSo t  LS 1   L T ,Tinit   TS (t )
From the above equation it can be deduced that if the temperature

rise for both of the wires is the same then the resistance ratio is constant
during the whole experimental run and it is independent from the
temperature.
Furthermore, if the resistance ratio is constant then it can be written
RLo  0  RLo  0    RLo (t )

 , (D.41)
RSo  0  RSo  0    RSo (t )
and using equations (D.6) and (D.10), the relation employed in the
calculation of the change in the resistance of a portion of an infinite
“working” wire,  RW t  , at a time t, is obtained.
RSo  0 
 RSo (t )   RWo (t )  . (D.25)
RWo  0 
D3.1.2. Calculation of Initial Resistances’ Values at zero time

From the aforementioned equations it can be seen that in order to
calculate the resistance change of the “working” portion of a wire of infinite
length after time t,  RW t  , the initial values (at zero time) of the long and
short wire resistances, RLo  0  and RSo  0  must be known. Also, the initial
values of the total equivalent resistance of the left arm of the Wheatstone
bridge, R  0  , and of the equivalent resistance of the bottom arm of the
bridge, RF  0  , must be known.
The calculation of the above quantities cannot be achieved with a
single transient measurement, but an initial steady–state measurement is
needed. Thus, the experimental procedure to measure the thermal
conductivity of a sample material always starts by applying a very small
voltage to the sensor and balancing the Wheatstone bridge by adjusting the
resistances R1 and R 2 , i.e. minimizing the voltage output of the bridge.
Following that the transient measurement is performed by applying to the
bridge the predefined voltage VS and after its end a steady–state
measurement is performed by applying to the bridge terminals A and D
predefined voltages of 0.5 V, 1.0 V, and 1.5 V. The measured values of the
voltage change across the terminals of the bridge and the sensor are
monitored and used for further processing. Specifically the voltage is
measured across terminals AB, AC, AD, AE and AF. Therefore, using the
equations that describe the electronic circuit (Figure D.8) of the bridge
[equations (D.42) to (D.46)]
V AB R std
 (D.42)
VS R
V AC Rstd  R 2
 (D.43)
VS R
V AD R std  R 2  R B  R Lo R F
  (D.44)
VS R R
V AE R std  R 2  R D  R Lo  R So  R A
 (D.45)
VS R
V AF VS , (D.46)
the three steady state values for each one of the resistances RLo , RSo , R ,
and RF are calculated. Subsequently, using linear extrapolation the initial
values of the resistances at time zero RLo  0  , RSo  0  , R  0  , and RF  0 
can be calculated.
D3.2. The Temperature Rise Equations

The experimental change of the wires’ resistances can also be
evaluated using pseudo–linear temperature coefficient of resistance as
mentioned in equations (D.37) and (D.38). Applying the same methodology
to the “working” portion of the hot wire, its experimental temperature rise,
TW t  , can be expressed as
 RWo t 
TW t   , (D.47)
 L T ,Tinit   RWo  0 
where, RWo  0  denotes the resistance of a “working” finite portion of an

infinite wire at zero time, and  RWo t  denotes the resistance change of the
“working” wire after time t. In addition,  L T ,Tinit  is a pseudo–linear
temperature coefficient calculated from the calibration of the tantalum
wires.
In order to calculate the resistance change of the working wire after
time t,  RWo t  , the values of the “working” wire resistance at time zero,
 RWo  0  , and at time t,  RWo t  , should be calculated [see equation (D.9)].
Thus, using temperature coefficients of resistance for the evaluation of the
above mentioned resistances the following equations are obtained:
RWo (0)  RWo To   1    Tinit  To     Tinit  To  

2
i. (D.48)
 
RWo (t )  RWo To   1    T  To     T  To  
2
ii. (D.49)
 
where RWo To  denotes the “working” resistance at reference temperature

To  273.15 K , and the temperature resistance coefficients of the wire
material, α and β, are calculated from equation (D.1). Moreover, Tinit and T
represent the initial and final temperature respectively.
Using equations (D.9), (D.47), (D.48) and (D.49) the pseudo–linear
temperature coefficient,  L T ,Tinit  , can be calculated as
    2  T To   T Tinit  

L T ,Tinit   , (D.50)
1    Tinit To     Tinit To 
2
where α and β are as mentioned the first and second temperature resistance
coefficients of the material of the wires, which were calculated through the
calibration procedure and are included in equation (D.1).

Furthermore, from equations (D.9), (D.48) and (D.49) the change of
the hot–wire resistance is given as
 RWo t   RWo To   T  Tinit     2    Tinit  To      T  Tinit   .

2
(D.51)
 
In equation (D.51), replacing TW t   T  T init , and RWo To  from

equation (D.48) a quadratic equation is obtained:
RWo  0   TW t     2    Tinit To      TW2 t  

 RWo t   (D.52)
1    Tinit To     Tinit To 
2

  TW2 t     TW t   C  0 (D.53)
where
     2    Tinit  To  (D.54)
 RWo t 
C
RWo  0  
 1    Tinit  To     Tinit  To  ,
2
 (D.55)
which can be solved and the experimental temperature rise can be

calculated.
Since the calculations are straightforward, the uncertainty of the
temperature rise TW t  depends on the uncertainty of the temperature
coefficients of resistance, on the uncertainty of the temperature, on the
resistance measurements at steady state and finally on the uncertainty of
the resistance change calculations.
In the ideal case the two wires should have identical cross–sectional
area and differ only on their length. However, in practice there are always
small geometry differences in their average diameters. These diameter
differences can be assigned to changes in the production line of the wire roll
and it is proved by measuring the wire’s diameter using Scanning Electron
Microscopy (SEM). In each case the SEM measured diameter value differs
some microns from the value given by the supplier. These geometrical
inhomogeneities result in different electric resistance per unit length and
therefore in a different temperature rise and heat flux from each wire.
According to the ideal model (see section C1) the temperature rise for
the short  TS t   and the long  TL t   wire after time t can be written as
qs   4   t 
TS t    ln  2   (D.56)
4       rS  e  
qL   4   t  
TL t    ln  2    , (D.57)
4       rL  e  
where α and λ denote the thermal diffusivity and the thermal conductivity
of the material under measurement, γ is the Euler constant, rS and rL
represent the short and long wire radius, respectively, and q S and q L are
the short and long wire heat flux per unit length.
Using equations (D.10), (D.37), (D.38), (D.56), and (D.57) in equation
(D.47) and assuming that the end effects contribution to the temperature
rise can be neglected, the corrected temperature increase of the long wire is
acquired from [Kestin & Wakeham, 1978]:
TW t 
TL t   (D.58)
1
where
  4   t  
 1  ln  2   
 RSo  0     rS  e    1  RSo  0   l L 
   
   .
 R Lo  0   RSo  0    ln  4    t    RLo  0  l S
  

    
 rS  e  
2

Here l S and l L denotes the short and long wire length respectively.
D3.3. The Heat Flux Equations

In order to complete the study of the design parameters that fully
describe the experimental setup, the constant heat flux input applied on the
hot wire during the experiment should be calculated. The bridge
arrangement of Figure D.8 ensures that equal currents flow in both wires,
but this alone does not imply an equal rate of heat dissipation per unit
length in the two wires. A detailed analysis and description of the

calculation of the heat flux per unit length of the middle portion of the long
wire can be found in previously published work [Kestin & Wakeham, 1978].
According to it the heat flux per unit length, q  , can be expressed in terms
of experimental quantities as:
q t 
q  t   , (D.59)
1  1  1  2 
2
where
2
 
 
 VS  RLo (t )  RSo t 
q t      , (D.60)
 R  0    R (t )  1  2  RSo  0   
LL  LS


Wo 
 RWo  0   
and
RLo  0  RSo  0 l L 
2  1  
 R  0 l   l L  l S
lL  Lo S 
1  , (D.61)
 l L  l S    R1  RA  RB  R2  Rstd   RLo  0  RSo  0   l L  l S 
 RSo  0  l L 
1     l S
R  0  lS
2   
Lo
. (D.62)
lL  lS 
D4. Calculation of the Thermal Conductivity Value

In the present section the methodology followed for the calculation of
the thermal conductivity value of the measured material is presented. A
similar procedure has been used in the Laboratory of Thermophysical
Properties and Environmental Processes by Gialou [Gialou, 2004] and
Metaxa [Metaxa, 2006]. The calculation of the thermal conductivity value of
the material includes three phases. It starts with the acquisition of the
experimental resistance rise data of the wire, then the temperature rise is
obtained from these data, and finally, with the use of Comsol Multiphysics
software, the thermal conductivity is obtained through a trial and error
procedure.
a) Acquisition of wires’ resistance change data

Before initiating an experimental run, the supplied voltage, VS, the
experiment duration, t, and the number of data points to be acquired, are
defined, depending on the material to be measured. The next step is to
perform a steady state run in order to balance the Wheatstone bridge. This
measurement is carried out by applying a small voltage of 0.5 V to the
bridge and balancing its arms by altering resistances R1 and R2. The
following step is performing the transient measurement, which takes place
by applying to the wires the predefined voltage VS. Both wires are heated
and their resistances increase with time. Thus, recording the bridge’s output
voltage V AD t  (see Figure D.8), the change of the resistance of the
“working” portion of a wire of infinite length  RWo t  can be calculated
[equation (D.28)]. The last stage of the experimental run includes the
measurement of the circuit’s resistances at three different applied voltages
(0.5 V, 1 V and 1.5 V). From these measured values, the initial values of the
resistances at zero time and voltage are derived through linear
extrapolation.
The results of an experimental run are recorded in two different files,
a “.dat” file and a “.res” one. The “.dat” file includes all the needed data
(steady state resistances values and time evolution of the voltage across the
bridge) for the calculation of the temperature rise of the wires and of the
heat flux per unit length that has been transferred through the wires. The
“.res” file contains once more the time evolution of the voltage across the
bridge and the calculated resistance change [equation (D.28)] of the
“working” wire, and is used for the calculation of the resistance rise of the
wire, and therefore of the temperature rise.
b) Analytically and numerical calculated experimental temperature rise

The experimental temperature response of the working section of the
wires is calculated through the use of a QBASIC computer program, which
is based on the analytical solution of the transient hot wire method (Section
C.1). More specifically the computer program use the data stored in both
“.dat” and “.res” files and using the equation of the ideal model, with the
corresponding corrections (finite diameter and properties of the wires, the

finite outer boundary of the measurand container, and the variable material
properties), calculates the corrected experimental temperature rise of the
wires. The program also takes into account the compensation of the end
effects of the wires and the constant heat flux per unit length that is
transmitted through them. Another parameters used in the QBASIC
program are the density and specific heat of the tantalum wires and of the
measured material, the temperature resistance coefficients of tantalum and
the lengths of the wires.
From the acquired corrected temperature rise curve, it is possible to
get the thermal conductivity value of the measured material by calculating
the slope of the curve (according to the ideal model). However, this solution
can be accurately applied on the measurement of isotropic materials, such
as fluids, but it can not provide an accurate description of the experimental
measurement in the case of the anisotropic measurement, e.g. the
measurement of solids with a multilayer sensor (see Section C1.1). Thus, the
use of a numerical solution based on the finite element method is employed.
As already analyzed, Comsol Multiphysisc is the finite element
software that was used for the numerical simulation of the experimental
setup. In order to get the numerically calculated temperature curve, the
estimated properties of each layer constituting the sensor should be inserted
in the created finite element model, i.e. for liquids, the wire and liquid
properties are required, while for solids the properties of the wire, the
silicone paste, the Kapton film and the specimen are required. Afterwards
the software solves the heat transfer partial differential equations and the
numerical temperature response is acquired. More details on the use of
Comsol Multiphysics were given on Section C2.2.
c) Calculation of the thermal conductivity value

The procedure to extract the thermal conductivity value of the
measured material is based on iterative inverse analysis. The unknown
thermal properties are calculated by iterative adjustments of the estimated
values until the Comsol Multiphysics calculated temperature rise is
superimposed satisfactory to the analytically calculated one from the start
of the experimental time. In order to compare both curves and get the
thermal conductivity value, the residual of the temperature rises is
calculated at each time point and then a chart is created. If the estimated
values employed in Comsol Multiphysics are correct then in the residuals

chart there is no systematic deviation and the linear trendline crossing the
residuals is parallel to the time axis (slope of the order of 10-5). However, if
the estimated values are wrong, then the trendline of the residuals crosses
the time axis and the Comsol Multiphysics numerical solution must be
recalculated using different material thermal properties until getting no
deviation.
In the case of solids, the temperature response can be divided in three
time regions depending on which layer mainly affect its. Thus, the iterative
inverse analysis is performed for each region separately, one after another
in order to get the correct values for each layer material.
The major employed differences of the methodology used in the

present work from the one by Gialou and Metaxa, are:
i. the use of Comsol Multiphysics software for simulating the experimental

temperature rise curve instead of the older finite element model written
in Fortran language, and
ii. the accurate representation of the experimental temperature rise from
time t  0 s .
In Figure D.10 the superimposition of the analytical to the numerical

calculated temperature rise versus the logarithm of time can be seen for the
measurement of the glass Pyrex 7740 at 353 K.
Figure D.10 Temperature rise as a function of time for Pyrex 7740 at 353 K
and plot of the residuals between the analytical and
the numerical calculated temperature rise.
D5. Validation of the Technique

The thermal conductivity of solids is known best to an uncertainty of
2  6 % (see Chapter E). As the present instrument is capable of achieving
a much better uncertainty of i.e. 1 % , it had to be able to be validated by
measuring the thermal conductivity of liquids (as liquids thermal
conductivity is known to 0.5 % uncertainty). Hence, the present
instrument was built with this capability in mind, and in essence this
makes it a unique instrument in measuring the thermal conductivity of
solids.
It should be mentioned that the establishment of a fluid as reference
needs some prerequisites. The most important is the stability of the fluid (in
the physical phase of interest) in a wide range of temperature and pressure
conditions, in which experimental measurements can take place, and the
availability of low uncertainty density data at these conditions range.
Moreover, it should be commercially available on high purity and with low
cost, and to be as possible less toxic and corrosive. In the last decade
responsible for the proposal of standards for transport properties for
calibration of instruments is the International Association for Transport
Properties (I.A.T.P.), which was formerly known as the Subcommittee on
Transport Properties (1981–2001) of the International Union of Pure and
Applied Chemistry (I.U.P.A.C.). Also, there is the International Association
for the Properties of Water and Steam (I.A.P.W.S.), which is responsible for
evaluating the thermophysical and transport properties of water and steam.
D5.1. Measurements of Toluene

Toluene is a liquid with all the needed characteristics for being a
thermal conductivity standard. It is a clear, colorless, flammable liquid
aromatic hydrocarbon, with a sweet, pungent odor and water insoluble. It is
widely used in industry for the manufacture of other chemicals or as a
solvent. Its liquid phase can cover a wide range of temperature and
pressure, and therefore can be used for calibration purposes.
Liquid toluene has been proposed by the Subcommittee on Transport
Properties of the International Union of Pure and Applied Chemistry as a
standard thermal–conductivity liquid for the temperature range 189 K to
440 K. The proposed equation for the calculation of the thermal conductivity
value of saturated liquid toluene with an uncertainty of ±1 % is [Ramires et
al., 2000]:
 *  0.420919  3.629457 T *  5.348298 T *2  2.8189482 T *3  0.519700 T *4

(D.63)
  T
where  *  ,T *  and   298.15   0.13088 W  m 1  K 1 ,
  298.15  298.15
while the value of the thermal conductivity of toluene at room temperature

(298.15 K) is given with an uncertainty of ±0.6 %.
There are more reference fluids that could be used for the validation
procedure such as water, benzene, and n–heptane. However, water
molecules are polar and can not be measured without modifying the actual
experimental setup. On the contrary, n–heptane was rejected, as its thermal
conductivity is not know with so low uncertainty as toluene’s, and benzene
was rejected as being more toxic.
Before performing a thermal conductivity measurement of solid, the
experimental device was always validated by measuring liquid toluene of
high purity (99.9%) supplied by Riedel de Haën company. The
measurements took place at room temperature (298.15 K) and at
atmospheric pressure. The wires in their support were placed in toluene and
500 temperature versus time points were obtained. To calculate the thermal
conductivity of toluene the procedure described in D4 has been followed,
and, the Comsol Multiphysics software was employed. The Comsol
Multiphysics model created for solids can be successfully used also for
liquids by setting in the program instead of the properties of the silicone
layer, the Kapton film, and the solid, the properties of liquid toluene. In
Table D.2 some indicative measurements of toluene performed during a
period of three years is shown. The results shown very good repeatability
and the deviations from the aforementioned recommended value is always
lower than ±0.5 %.
Table D.2 Deviations of the measured thermal conductivity value of toluene

from equation (D.63).
Deviation
Date
(%)
04/2005 0.02
08/2005 0.04
10/2005 -0.25
01/2006 -0.04
05/2006 -0.15
02/2007 -0.28
03/2007 -0.33
06/2007 0.15
03/2008 0.30
In Figure D.11 a typical experimental temperature response curve

together with the simulated one, are shown as a function of time. The values
calculated with Comsol Multiphysics software are 0.129 W  m 1  K 1 for
thermal conductivity and 1,475 kJ  m 3  K 1 for the product (density x
specific heat capacity) which deviates by 0.3 % and 0.4 % from literature
recommended values. Moreover, in Figure D.11 the differences between the
experimental and the simulated temperature rise are shown.
Therefore, the proposed configuration of the transient hot wire
instrument can be easily employed to measure the thermal conductivity of
liquids with very low uncertainty (better than ±0.5 %). Thus, its application
can be expanded to the measurement of the thermal conductivity
measurement of solids and composite materials.
Figure D.11 Temperature rise as a function of time for toluene at 303 K

and plot of the residuals between the analytical and
the numerical calculated temperature rise.
D6. Uncertainty Analysis

In every experimental procedure, it is important to know the
uncertainty of the method and therefore the uncertainty of the measurands.
Usually the terms uncertainty and accuracy are considered to be the same.
However, there is an important difference between them. According to their
definitions [Joint Committee for Guides in Metrology, 2008], accuracy shows
the “closeness of the agreement between the result of a measurement and a
true value of the measurand” and is a qualitative concept, while uncertainty
is “a quantitative parameter, associated with the result of a measurement,
that characterizes the dispersion of the values that could reasonably be
attributed to the measurand”. Furthermore, in metrology the terms
repeatability and reproducibility are also confused. They both refers to the
closeness of the results of successive measurements of the same measurand,
but their main difference is that repeatability suggest keeping the same
experimental conditions, while for reproducibility the conditions should be
altered.
In the present work, the overall uncertainty of the method used for
the calculation of the thermal conductivity value of solids consists of two
separated parts: firstly the uncertainty arising from the experimental
devices and the electronic configuration used, and secondly the uncertainty
due to the use of the finite element method. For the second one, a detailed
analysis has been presented in section C2.3. As far as the uncertainty of the
experimental setup is concerned, it has been shown above that the
calculated thermal conductivity value depends on the temporal temperature
difference between the wires, and therefore on the difference of their
resistances. Thus, the parameters to be evaluated are those that affect the
wires resistances and the time measurement. In the following sections, each
parameter is analyzed separately, and then using basic rules of propagation
of uncertainty, the overall uncertainty of the experimental setup is
calculated.
a) Supply voltage
The voltage applied to the Wheatstone bridge, and its evolution
during a transient experimental run, is monitored by a digital high–
resolution multimeter with uncertainty better than  1 μV. As it was shown,
the recorded voltage differences are transformed into resistances changes
and consequently on temperature differences of a wire with no ends. The

effect of this parameter on the temperature rise of the wire is estimated to
be of the order of 10-4 %.
b) Experimental time
The experimental time is measured and registered through the A/D
card (PC30AT, Amplicon Liveline Ltd., U.K.) with a resolution of 16–bit and
an uncertainty of  1 μs. However, in the calculation of thermal conductivity
value, the logarithm of time is used and therefore its influence on the
uncertainty of the obtained value is lower than  10-3 %.
c) Temperature coefficients of resistance

In order to minimize the error introduced in the calculations by the
temperature coefficients of resistance of the tantalum wires, both thin and
thick (of nominal diameters 25 μm and 0.5 mm respectively), were
calibrated (see D1.1.2). The calibration was made with a platinum
resistance thermometer (Type 5187SA, No 250293, Tinsley and Co. Ltd.,
U.K.) with accuracy ±1 mK for the temperature measurement and a type
KB5 four terminal resistance bridge (Cropico Ltd., U.K.) with uncertainty of
±0.01 % for the resistances measurement. This has no measurable effect in
the temperature rise of the wire.
d) Sensor thickness
The experimental sensor is constructed so that the thickness of the
silicone paste film is defined and equal exactly to 1 mm. Thus, it does not
affect the uncertainty of the technique.
e) Other parameters
There are more variables that affect the measurement of the
temperature rise of the wire, but the error introduced is considered very
small. Such parameters are the measurement of the furnace temperature,
which is made with two platinum resistance thermometers with uncertainty
better than  80 mK, and the small differences in the cross–sectional area of
the hot wire.
According to the Joint Committee for Guides in Metrology [Joint

Committee for Guides in Metrology, 2008], the combined uncertainty uC  y 
of the quantityY  x  , is the positive square root of the combined variance
uC2  y  obtained from:

2
N
 Y  x  
uC  y      u  x i 
2 2
(D.64)
i 1  x i 
where u 2  x i  are the variances of the input quantity xi. The partial
derivatives of Y  x  are called sensitivity coefficients and describe how the
output estimate y varies with change in the values of the input estimates x1,
x2, ..., xN. As the thermal conductivity value, obtained from the equation of
the ideal model, is obtained only from the multiplication or division of the
input quantities, equation (D.64) can be written as a linear addition by
using relative uncertainties [Joint Committee for Guides in Metrology,
2008]
 i 
2 2
 uC  y   N u x
  
  (D.65)
 y  i 1  x i 
u  x i  
2
where  are the relative uncertainties of each input estimate xi.

 x i 
Therefore, employing in equation (D.65), the relative uncertainties of
the parameters affecting the measured thermal conductivity value (i.e.
paragraphs a, b, c, d, and e), it is estimated that the uncertainty of the
experimental procedure is no more than  0.1 %. However, the overall
uncertainty of the transient hot wire technique for the measurement of the
thermal conductivity of solids, described in the present work, is calculated
by taking into account both experimental uncertainty and the uncertainty
due to finite element method (  0.5 %). Thus, the overall uncertainty is
estimated to be lower than  1 % for the measurement of the thermal
conductivity of solid materials. This uncertainty originates mainly from the
inability to accurately measure the temperature and the thermal and
geometrical properties of the hot wire.
D7. Summary
The chapter was devoted to the detailed analysis of the experimental
configuration employed. The design of the new transient hot wire sensor for
the measurement of the thermal conductivity of solid was presented and the
major ameliorations made in relation to the older one, are depicted.
Moreover, the working equations of the electronic Wheatstone bridge are
analysed and the methodology for acquiring the thermal conductivity value
by using the measured data and the Comsol Multiphysic software is shown.
Finally, the method and the sensor are validated through the measurement
of saturated liquid toluene and the overall uncertainty of the method is
calculated.
The most significant concluding points of the chapter are:
i. The new design, except from the robustness that provide to the sensor,
ensures that the thickness of the silicone paste of the hot wire sensor is
exactly 1 mm. In this way, the uncertainty source related to the paste
thickness is compensated.
ii. The experimental configuration, through the working equations, allows
the calculation of the thermal conductivity in an absolute way through
the measurement of the resistance change of a finite portion of an
infinite wire.
iii. The designed transient hot wire sensor can be used also for the
measurement of the thermal conductivity of liquids with an uncertainty
better than  0.5 %.
iv. The overall uncertainty, including both experimental and numerical
errors, is estimated to be better than  1 % for the measurement of the
thermal conductivity of solid materials.
E
Thermal Conductivity
Measurements
In this chapter the thermal conductivity measurements of the solid

and composite materials studied during the present work is shown and
discussed. In the first part of the chapter, the developed transient hot wire
sensor is first employed for the measurement of the thermal conductivity of
the silicone pastes used in the sensor. Following that, the technique is used
for measuring the thermal conductivity of the reference materials
Pyroceram 9606 (designated as glass ceramic BCR–724) and Pyrex 7740
(designated as BCR–039) over a temperature range from 315 K to 440 K.
Subsequently, the sensor is used for studying candidate reference materials
for thermal conductivity. The candidate glassy materials measured from
ambient temperature up 440 K were Polymethyl Methacrylate (PMMA) and
Borosilicate Crown Glass (BK7). The second part of the chapter focuses on
the accurate measurement of the thermal conductivity of composite
materials. Specifically, the hot wire sensor is used for measuring the
enhancement of the thermal conductivity of an epoxy–resin reinforced
initially with plies of plain weave glass fabric, then by carbon multi–walled
nanotubes (C–MWNT), and finally with both these two macroscopic and
nanoscopic reinforcements at hand.
142 Thermal Conductivity Measurements
E1. Solid Materials

The actual transient hot wire instrument was employed for the
measurement of the thermal conductivity of solids at elevated temperatures.
The materials studied are two thermal conductivity reference materials,
Pyroceram 9606 (designated as glass ceramic BCR–724) and Pyrex 7740
(designated as BCR–039), and two possible candidate glassy solids,
Polymethyl Methacrylate (PMMA) and Borosilicate Crown Glass (BK7).
However, in order to acquire the thermal conductivity values of the above
solids as a function of temperature, the properties of the silicone paste, in
which the hot wire is enclosed, need to be found first.
The procedure followed for the measurement and determination of
the thermal conductivity of the specimens is described in details in the
previous chapter (Section D4).
E1.1. Silicone Pastes

As described in Chapter D, the tantalum hot wires are placed inside a
silicone layer of exactly 1 mm thickness, in order to minimize the thermal
contact resistance between the samples and the sensor, and to attribute
elastic properties to the sensor. The thermal conductivity and the
volumetric specific heat of the silicone layer are found from measurements
at short times t  0.5 s .
It should be mentioned that two different silicone pastes were used
during the present work depending, on the thermal properties of the
measured material. The first one is AS1803 silicone paste (ACC Silicones),
which has relative high thermal conductivity value was used for the
measurement of Pyroceram 9606 and the second one, which is a high–
temperature red RTV silicone (Type 650, BORO, VersaChem, U.S.A.), was
employed for the measurement of the rest of the solid samples and for the
measurements of the composite materials.
The choice of the silicone paste is of high importance. When
measuring relative good thermal conductors (such as Pyroceram 9606), the
silicone paste should be of high thermal conductivity and still retain its
elastic properties. A silicone paste with low thermal conductivity would
prevent the heat transfer from the wire to the solid and therefore the solid’s
Thermal Conductivity Measurements 143
thermal conductivity would barely affect the evolution of the temperature of

the wire. This would reduce the sensitivity of the technique and may affect
its uncertainty.
Moreover, the elasticity of the paste is also important for the correct
measurement of the samples thermal conductivity. The use of pastes with
low elasticity, resulted in the appearance of air gaps between the sensor and
the specimens and consequently the appearance of high thermal contact
resistance.
It should also be mentioned that the temperature range of the
experiments was mainly specified by the silicone pastes limits. The use of
the silicone above the temperature where it looses its elasticity, results once
more on the appearance of high thermal contact resistance. Hence, the
temperature range of the measurement is limited to 440 K.
E1.1.1. Thermal Conductivity Measurements of Silicone Paste AS1803

The AS1803 silicone paste was used in the present work due to its
high thermal conductivity, which allows the measurement of good thermal
conductors, such as Pyroceram 9606. The thermal conductivity
measurements of the paste are shown in Table E.1 .
Table E.1 Measurement of properties of silicone paste AS1803

as a function of temperature
T λ Δλ   C P     C P 
(K)  W  m1  K 1  (%)  kJ  m3  K 1  (%)
314.18 1.130 -0.24 2,503 0.38
333.74 1.100 0.41 2,445 -0.04
353.46 1.060 0.18 2,395 -0.13
373.05 1.020 -0.09 2,340 -0.44
392.90 0.975 -0.83 2,290 -0.53
413.209 0.950 0.57 2,270 0.77
The thermal conductivity, λ  W  m1  K 1  , values shown in Table

E.1, were fitted to a function of the absolute temperature T (K) with an
equation:
i
 T 
    298.15 K     i   , (E.1)
i  298.15 
where the coefficients  i and the value of the thermal conductivity,

λ(298.15Κ)  W  m1  K 1  is given in Table E.2. The deviation of the
experimental thermal conductivity values from the one obtained from
equation (E.1), λfit, is calculated from:
   fit 
  100   . (E.2)
 fit 
Moreover, the volumetric specific heat capacity values   C P  are

also shown in Table E.1, and were fitted to a function of the absolute
temperature T (K) with an equation:
i
T 
  C P     C P   298.15 K    bi   , (E.3)
i  298.15 
where the coefficients bi and the value of   C P   298.15 K  are given in

Table E.2. The deviation of the experimental volumetric specific heat
capacity values, from the one obtained from equation (E.3),   C P fit , is
given by:
   C P     C P fit 
   C P   100    . (E.4)

   C P fit 
Table E.2 Coefficients and standard deviations of equations (E.1) and (E.3).
Coefficient/Parameter Values
Equation (E.1)
λ(298.15Κ)  W  m1  K 1  1.1632
0 (-) 1.48704
1 (-) -0.48691
σ (%), at 95 % confidence level ±0.81
Equation (E.3)
  C P   298.15 K   kJ  m3  K 1  2,532.6
b0 (-) 1.28635
b1 (-) -0.28635
The maximum deviations of the experimental points presented in

Table E.1., from equations (E.1) and (E.3), are 0.83 % for thermal
conductivity and 0.77 % for the volumetric specific heat. At the 95 %
confidence level, the standard deviation of the thermal conductivity

measurements is 0.81 %, while for the volumetric specific heat it is 0.80 %.
Both deviations are within the standard uncertainty of the technique.
E1.1.2. Thermal Conductivity Measurements of Silicone Paste BORO, Type 650

The RTV silicone paste BORO, Type 650 was employed for the
measurement of the thermal conductivity of all the solid samples, except for
Pyroceram 9606. Due to its great elastic properties is was also implemented
for the measurement of composites materials. The thermal conductivity
measurements of the paste are shown in Table E.3.
Table E.3 Measurement of properties of silicone paste BORO, Type 650

as a function of temperature.
T λ Δλ   C P     C P 
(K)  W  m1  K 1  (%)  kJ  m3  K 1  (%)
314.44 0.1735 -0.62 1,775 -0.12
315.87 0.1743 -0.20 1,770 -0.47
333.70 0.1716 0.11 1,796 0.14
333.91 0.1722 0.48 1,805 0.63
353.15 0.1690 0.47 1,815 0.27
353.55 0.1683 0.09 1,812 0.09
372.70 0.1653 0.19 1,830 0.18
372.87 0.1650 0.02 1,825 -0.10
393.26 0.1611 -0.31 1,844 -0.27
392.40 0.1619 0.09 1,838 -0.29
412.55 0.1586 0.11 1,851 -0.51
413.02 0.1574 -0.60 1,855 -0.31
432.81 0.1547 -0.26 1,880 0.13
433.04 0.1557 0.41 1,890 0.65
Once again, the thermal conductivity, λ  W  m1  K 1  , values shown

in Table E.3, were fitted to a function of the absolute temperature T (K)
with an equation:
i
 T 
    298.15 K     i   , (E.5)
i  298.15 


   fit 
  100   . (E.6)
 fit 

i
T
  C P     C P   298.15 K    bi  
 , (E.7)
i  298.15 
where the coefficients bi and the value of   C P   298.15 K  are given in

capacity values from the one obtained from equation (E.7),   C P fit , is
given by:
   C P     C P fit 
   C P   100    . (E.8)

   C P fit 
Equation (E.5)
λ(298.15Κ)  W  m1  K 1  0.1773
0 (-) 1.27653
1 (-) -0.27676
Equation (E.7)
  C P   298.15 K   kJ  m3  K 1  1,763.4
b0 (-) 0.85654
b1 (-) 0.14345

Table E.3., from equations (E.5) and (E.7), are 0.62 % for thermal

E1.2. Pyroceram 9606

Pyroceram 9606 is an opaque glassy ceramic, originally developed by
NASA, and since it is particularly well defined and thermally stable, it was
proposed as a standard reference material for thermal conductivity by the
National Institute of Standards and Technology (NIST), U.S.A. [Powell et
al., 1966]. Moreover, since May 2007, Pyroceram 9606 is supplied by the
European Commission Institute for Reference Materials and Measurements
(http://www.irmm.jrc.be/) as a certified thermal–conductivity and thermal-
diffusivity reference material (designated as glass ceramic BCR–724) up to
1025 K [Salmon et al., 2007]. This certification was the outcome of a
research project, funded by the European Union under the ‘Competitive and
Sustainable Growth’ program (“HTCRM - High Temperature Certified
Reference Materials,” Contract SMT4-CT98-2211/2003). The uncertainty of
the certified thermal conductivity value was ±6.5 %, while that of the
thermal diffusivity was ±6.1 %. Therefore, measurements of very low
uncertainty are still in need today.
The typical composition of Pyroceram 9606 is shown in Table E.5.
Table E.5 Typical chemical composition of Pyroceram 9606.
Composition
Component
(%)
SiO2 56.80
Al2O3 19.40
MgO 13.90
TiO2 8.80
K2O 0.08
CaO 0.2
Na2O 0.2
Miscellaneous traces 0.62
The two specimens of Pyroceram 9606 employed in the present work

were made by Corning Inc., New York and purchased from Anter
Corporation, Pittsburgh, PA, U.S.A. The samples have dimensions of
5x10x2 cm3 and their density was found by weighing them and determining
their volume equal to 2,596 kg.m-3 at 293.15 K.
E1.2.1. Thermal Conductivity Measurements

Many investigators have performed measurements of the
thermophysical properties of Pyroceram 9960. A brief review of these
measurements is presented in Table E.7. As already mentioned in Chapter
B, three major interlaboratory projects were realized in order to establish
Pyroceram 9960 as a reference material:
i. one coordinated by Powell et al. in 1966 [Powell et al., 1966],

ii. one reported by Hulstrom et al. in 1988 [Hulstrom et al., 1988], and
iii. one conducted under the auspices of the European Commission for
Reference Materials and Measurements [Salmon et al., 2007].
The present thermal conductivity measurements performed covered

the temperature range from 315 K to 420 K and the values are shown in
Table E.6.
Table E.6 Measurement of properties of Pyroceram 9960 as a function of temperature
T λ Δλ   C P     C P 
(K)  W  m1  K 1  (%)  kJ  m3  K 1  (%)
314.18 3.76 -0.20 1,986 0.03
333.74 3.71 0.35 2,103 -0.30
353.46 3.63 -0.05 2,233 0.61
373.05 3.58 0.20 2,310 -0.17
392.90 3.50 -0.54 2,388 -0.27
413.21 3.48 0.29 2,466 0.19

equation:
i
 T 
    298.15 K    i   , (E.9)
i  298.15 

Table E.7 Thermophysical properties measurements of Pyroceram 9606.
Temperature range
Reference Measurement technique Property Uncertainty
(K)
Gathering of existing 200 – 1,000 λ ±5.0 %
[Powell et al., 1966]
experimental data 100 – 1,400 λ ±10.0 %
Round-Robin interlaboratory λ ±5.6 % at 95 % confidence level
[Hulstrom et al., 1988] 323 – 573
comparison CP Not quoted
[Cahill, 1990] 3ω method 30 – 750 λ Not quoted
[Perovic & Maglic, 1991] Laser flash technique 473 – 1,473 α ±5.5 %
λ Not quoted
[Gustafsson, 1991] Transient plane source 295 – 473
α and C P Not quoted
[Ohta et al., 1991] Laser flash technique 298 – 976 α ±5.0 %
[Matsumoto & Ono, 1992] Radiative heat exchange method 298 – 393 λ ±2.5 %
[Suliyanti et al., 1995] Laser Flash 298 – 1,273 α ±3.0 %
[Hisano & Placido, 1998] Thermal Radiation Calorimetry 673 – 1,273 λ 10 % - 20%
[Filla, 1997] Guarded Hot-Plate 400 – 1,400 λ ±5.0 %
Dynamic measurement access
[Zawilski & Tritt, 2001] 10 – 300 λ Not quoted
technique
Parallel thermal conductance
[Zawilski et al., 2001] 10 – 300 λ Not quoted
technique
[Gaal et al., 2004] Flash method 290 – 1,273 α ±3 % at 95 % confidence level
λ ±1.5 %
[Assael & Gialou, 2003] Transient Hot – Wire 293 – 590
  C P  ±5.0 %
Combined Guarded hot Plate and
[Tleoubaev & Brzezinski, 2005] 298 and 318 λ Not quoted
Heat Flow meter method
λ ±6.5 % at 95 % confidence level
[Salmon et al., 2007] Certification of BCR–724 298 – 1,025 α ±6.1 % at 95 % confidence level
CP ±7 % at 95 % confidence level
   fit 
  100   . (E.10)
 fit 

i
T
  C P     C P   298.15 K    bi  
 , (E.11)
i  298.15 
where the coefficients bi and the value of   C P   298.15 K  is given in

given by:
   C P     C P fit 
   C P   100    . (E.12)

   C P fit 
Equation (E.9)
λ(298.15Κ)  W  m1  K 1  3.83
0 (-) 1.485
1 (-) -0.6596
2 (-) 0.1745
Equation (E.11)
  C P   298.15 K   kJ  m3  K 1  1,873
b0 (-) -1.078
b1 (-) 2.990
b2 (-) =0.912

Table E.7, from equations (E.9) and (E.11), are 0.54 % for thermal
In Figure E.1, the present thermal conductivity measurements and

their percentage deviations from the values calculated by equation (E.9) are
shown. In the same figure deviations the thermal conductivity values of
other investigators from equation (E.9) are also illustrated.
Figure E.1 Thermal conductivity measurements of Pyroceram 9606 as a function of

temperature and their deviations from the values calculated from equation (E.9).
As it was mentioned, since May 2007 the European Union of

Reference Materials and Measurements has issued a certificate for the
ceramic material Pyroceram 9606. The certified thermal conductivity
values, λ, over the temperature range from 298 K to 1,025 K is given by the
equation
515.1
  2.332  , (E.13)
T
where T is temperature in K. The uncertainty of the above certified thermal

conductivity values is ±6.5 % at the 95 % confidence level. In Section B1.3 a
detail analysis of the Round–Robin project performed in order to obtain the
certified equation (E.13) is given. It should be mentioned that in this project,
for the first time, the transient techniques for the measurement of thermal
conductivity dominated over the steady state. However, the uncertainty of
the obtained values remained high. In Figure E.1, it can be seen that the
deviations of the certified values by IRMM from the values of equation (E.9)
are within mutual uncertainties.
Furthermore, in Figure E.1 the recommended values of Hulstrom et
al. [Hulstrom et al., 1988], from a round–robin project, characterized by a
±5.6 % standard deviation at the 95% confidence level, are shown. The
recommended thermal conductivity values, λ, over the temperature range
from 323 K to 573 K is given by the equation
  4.0733875  3.620446 103 T  4.269643 106 T 2 , (E.14)
where T is temperature in oC. The maximum deviation of these values from

the present set is 5.3 %, which is still within the mutual uncertainties of the
values.
The thermal conductivity values for Pyroceram 9606, reported by
Powell et al. in 1966 [Powell et al., 1966], have a maximum uncertainty of
±5 % and deviate from values calculated from equation (E.9) by 4.3 %. This
comparison is quite interest since in 1966, techniques were not as advanced
as today. Hulstrom et al. in referring to the values of Powell et al. for
Pyroceram 9606 quotes: “It would appear that the overall state of the art in
thermal–conductivity measurements for materials in this range has change
little over the last 18 years”.
In Figure E.1 more recent measurements are also shown:
i. The measurements of Cahill in 1990 were performed by the 3ω

technique in the temperature range from 30 K to 750 K [Cahill, 1990].
Although no overall uncertainty for the measured data is quoted, the
error due to black–body radiation is found less than 2 %. These values
differ by up to 3.8 % from the present values calculated by equation
(E.9).
ii. Gustafsson employed the transient plane source technique for
measuring the thermal conductivity of Pyroceram 9606 [Gustafsson,
1991]. He used the “Hot disk” arrangement for the measurements and
the thermal conductivity values acquired shown a 3 % spread and
deviate up to 3.2 % from those of equation (E.9).
iii. Matsumoto and Ono performed thermal conductivity measurements of
Pyroceram 9606 using a method based on radiative heat exchange
[Matsumoto & Ono, 1992]. The measurements have a claimed
uncertainty of 2.5 %, and the proposed equation from 298 K to 393 K is
  3.9118  2.9646 103 T , (E.15)
where λ is the thermal conductivity in W  m1  K 1 and T is the

temperature in oC. The values calculated from equation (E.15) are in
perfect agreement with the present measured values.
iv. The measurements of Filla where performed with a steady state
guarded hot plate and have an quoted uncertainty of ±5 % [Filla, 1997].
The measurements are mainly performed in high temperatures (400 K
to 1,400 K) and they agree with values calculated from equation (E.9).
v. The previous measurements of Assael and Gialou [Assael & Gialou,
2003], performed with an uncertainty of ±1.5 %, agree very well with the
present measurements performed with the lower uncertainty of ± 1 %.
In Figure E.1 two measurement sets, differing in the silicone paste
employed in the sensor, are illustrated.
vi. Finally, Tleoubaev and Brzezinski used a combined guarded hot plate
and heat flow meter method to measure the thermal conductivity of
Pyroceram 9606 [Tleoubaev & Brzezinski, 2005]. Although the
uncertainty of the measurements is not quoted, the deviation from the
values calculated from equation (E.9) is up to 3.1 %.
In Figure E.2 the volumetric specific heat measurements of

Pyroceram 9606 as a function of temperature and the percentages
deviations from the values calculated from equation (E.11) are shown. In the
same figure the previous published data of Hulstrom et al. [Hulstrom et al.,
1988], Gustafsson [Gustafsson, 1991] and Assael and Gialou [Assael &
Gialou, 2003], are also shown. Moreover, it is included the data published
in the certification of Pyroceram 9606 as a reference material for thermal
conductivity and thermal diffusivity [Salmon et al., 2007].
Figure E.2 Volumetric specific heat measurements of Pyroceram 9606 as a function of

Hulstrom et al. used the differential scanning calorimeter method for

the measurement of the specific heat capacity, without however quoting the
uncertainty of his measurements [Hulstrom et al., 1988]. It can be seen in
Figure E.2 that the deviations from the values calculated from equation
(E.11) are within the uncertainty of the employed transient hot–wire
technique.
The data published by Gustafsson shows large deviations from values
of equation (E.11) in low temperatures (up to 9.7 %), which however are
minimized in higher temperatures. The data published by Gustafsson has
no quoted uncertainty and is claimed to have a spread of 3 % [Gustafsson,
1991].
The previous measurements of volumetric specific heat capacity by
Assael and Gialou, were performed with an uncertainty of ±5 % [Assael &
Gialou, 2003]. The agreement with the present set is excellent and the
deviations are within the uncertainty limits.
The volumetric specific heat capacity included in the certification of
BCR 724 is calculated from the measured specific heat capacity values, as
measured from six different research groups participating in the Round–
Robin project, and the mean density value of Pyroceram 9606 measured at
298.15 K (2602 kg  m3 ) [Salmon et al., 2007]. The specific heat capacity,
C P  J  g 1  K 1  , is represented with the polynomial function:
C P  0.2334  2.923 103 T  3.878 106 T 2  2.497 10 9 T 3  6.1295 1013 T 4

(E.16)
where T is the temperature in K. The uncertainty of the values at a 95 %

confidence level is 7 %.
Therefore it can be seen in Figure E.2, that although the deviations of
the volumetric specific heat from the one calculated from equation (E.11) are
high at low temperatures (up 11 %), they are still in within the mutual
uncertainties of the measurements.
E1.3. Pyrex 7740

Pyrex 7740 is a well-known Type I, Class A borosilicate glass, which
conforms to ASTM E438 [American Society for Testing and Materials, 1992]
and since 1990 it is considered as a certified reference material for thermal
conductivity, BCR 039, by the European Union of Reference Materials and
Measurements [Williams & Shawyer, 1991]. The use of Pyrex 7740 as a
reference material has its roots back to the mid–1960s, when Powell et al.
recommended thermal conductivity values for it [Powell et al., 1966].
Pyrex 7740 has a low coefficient of expansion, which allows to be
manufactured in relative heavy walls giving it mechanical strength, while
retaining reasonable heat resistance. Moreover, it is highly resistant to
chemical compounds such as strong acids, alkalis, e.t.c., and can withstand
temperatures up to 760 K.. Therefore, due to its excellent thermal and
mechanical properties, Pyrex 7740 is used in many laboratory and
industrial applications. Its typical composition can be seen in Table E.9.
Table E.9 Typical chemical composition of Pyrex 7740.
Composition
Component
(% approx.)
SiO2 80.60
B2O3 13.00
Na2O 4.00
Al2O3 2.30
Miscellaneous traces 0.01
The two samples of Pyrex 7740 employed in the present work were
supplied by Anter Corporation, Pittsburgh, PA, U.S.A. The samples have
dimensions of 5x10x2 cm3 and their density was found by weighing them
and determining their volume equal to 2,227 kg.m-3 at 293.15 K.

thermophysical properties of Pyrex 7740. A brief review of these
measurements is presented in Table E.10
Table E.10 Thermophysical properties measurements of Pyrex 7740.
Temperature range
(K)
Collection of existing
[Powell et al., 1966] 100 – 700 λ ±5.0 %
experimental data
Round-Robin interlaboratory λ ±10.3 % at 95 % confidence level
[Hulstrom et al., 1988] 50 – 300
comparison CP Not quoted
[Cahill, 1990] 3ω method 30 – 750 λ Not quoted
Steady – State axial heat flow
[Pillai & George, 1991] 350 – 1,250 λ ±2.0 %
comparative method
[Matsumoto & Ono, 1991] Radiative heat exchange method 303 – 363 λ Not quoted
[Williams & Shawyer, 1991] Certification report of BCR 039 198 – 468 λ ±1.7 % at 95 % confidence level
[Log & Metallinou, 1992] Transient Hot Strip method 299.15 λ ±3.0 %
[Yang et al., 1992] AC techniques 20 – 310 α and C P Not quoted
λ ±5.0 %
[Miller & Kotlar, 1993] Transient method 296
α ±5.0 %
λ ±3.2 % at 95 % confidence level
[Sabuga & Hammerschmidt, 1995] Transient Hot–Strip 191 – 473
α ±17.0 % at 95 % confidence level
[Hisano & Placido, 1998] Thermal Radiation Calorimetry 523 – 673 λ 10 % - 20%
Transient Hot-Wire and ±5.8 % with transient hot wire
[Hammerschmidt & Sabuga, 2000] 198 – 468 λ
Transient Hot Strip ±2.5 % with transient hot strip
[Milano et al., 2001] Pulse method 290 – 350 λ and C P Not quoted
[Zawilski & Tritt, 2001] Dynamic Measurement Access 10 – 300 λ Not quoted
[Gaal et al., 2004] Flash method 298.15 α ±3.0 % at 95 % confidence level
Combined Guarded Hot Plate
[Tleoubaev & Brzezinski, 2005] 298 and 318 λ Not quoted
and Heat Flow Meter method
λ ±1.5 %
[Assael et al., 2004] Transient Hot – Wire 293 – 570
  C P  ±5.0 %
[Longo, 2008] Steady – State heat flow method 276 – 333 λ ±3.8 %

Table E.11
Table E.11 Measurement of properties of Pyrex 7740 as a function of temperature
T λ Δλ   C P     C P 
(K)  W  m1  K 1  (%)  kJ  m3  K 1  (%)
314.44 1.150 -0.18 1,770 -0.54
333.70 1.170 -0.52 1,797 0.19
353.55 1.210 0.76 1,815 0.40
372.87 1.230 0.42 1,826 0.26
393.26 1.250 -0.02 1,837 0.07
413.02 1.270 -0.39 1,848 -0.08
432.81 1.300 0.03 1,860 -0.23

equation:
i
 T 
    298.15 K    i   , (E.17)
i  298.15 

   fit 
  100   . (E.18)
 fit 

i
T 
  C P     C P   298.15 K    bi   , (E.19)
i  298.15 

given by:
   C P     C P fit 
   C P   100    . (E.20)

   C P fit 
Equation (E.17)
λ(298.15Κ)  W  m1  K 1  1.1318
0 (-) 0.6718
1 (-) 0.3282
Equation (E.19)
  C P   298.15 K   kJ  m3  K 1  1,769
b0 (-) 0.8806
b1 (-) 0.1192
The maximum deviations of the experimental points, presented in

Table E.11, from equations (E.17) and (E.19) are 0.76 % for thermal
conductivity and 0.54 % for the volumetric specific heat. At the 95%
their percentage deviations from the values calculated by equation (E.17)
are shown. In the same figure deviations the thermal conductivity values of
As it was mentioned before, since 1990 the European Union of
Reference Materials and Measurements has issued a certificate for Pyrex
glass material. The certified thermal conductivity values, λ, over the
temperature range of 198 K to 468 K is given by the equation
  1.1036  1.659 103    3.982 106  2  6.764 109  3 , (E.21)
where Θ is the mean specimen temperature in oC.

However, it should be noted that this certificate refers only to a Pyrex
glass and not specifically the 7740 grade. These certified values,
characterized by a ±1.7 % standard deviation at the 95 % confidence level,
are presented in Figure E.3. As it can be seen, the present measurements

are in excellent agreement with these values.
Figure E.3 Thermal conductivity measurements of Pyrex 7740 as a function of temperature

and their deviations from the values calculated from equation (E.17).
Moreover, in Figure E.3 the recommended values of Hulstrom et al.

[Hulstrom et al., 1988], from a round–robin project, characterized by a
±10.3 % standard deviation at the 9 5% confidence level, are shown, together

with the previously reported values of Powell et al. in 1966 [Powell et al.,
1966], of 5 % maximum uncertainty. The agreement of the present
measurements with all these sets is excellent. In addition, there is good
agreement between these two sets an the certified values for Pyrex glass.
In Figure E.3 more recent measurements are also shown:
i. The measurements of Matsumoto and Ono, were performed by applying

a perpendicular heat flow to a thin–plate sample [Matsumoto & Ono,
1991]. Although the technique is described as an absolute technique, no
uncertainty is quoted. These values show a spread of ±2.5 %, and differ
by up to 5% from the present values calculated by equation (E.17).
ii. The measurements of Pillai and George were performed in a steady–
state axial heat flow comparative instrument with an uncertainty of 2%
[Pillai & George, 1991]. The deviations of these measurements from the
present values calculated by equation (E.17), are within the mutual
uncertainties of the two instruments.
iii. The thermal conductivity measurement of Log and Metallinou, at
322.15 K, performed with the transient hot–strip method, and with a
claimed uncertainty of 3% (no confidence level was specified) [Log &
Metallinou, 1992], are in excellent agreement with the present values
calculated by equation (E.17).
iv. Miller and Kotlar used an experimental device for the simultaneous
measurement of the thermal conductivity and thermal diffusivity
[Miller & Kotlar, 1993]. The device was operated in transient state and
the thermal conductivity value at 296 K was measured 1.06 W  m1  K 1 ,
with an estimated uncertainty of ±5 %. Thus, the measured value is
within the mutual uncertainties of the two instruments.
v. The measurements of Milano et al. were performed in a transient
double–plate instrument [Milano et al., 2001]. However, no uncertainty
is quoted, but the measurements deviate about 1.6 % from the present
values calculated by equation (E.17).
vi. The previous measurements of Assael et al. [Assael et al., 2005],
performed with an uncertainty of ±1.5 %, agree very well with the
vii. Finally, Tleoubaev and Brzezinski used a combined guarded hot plate
and heat flow meter method to measure the thermal conductivity of
several reference materials, among which was Pyrex 7740 [Tleoubaev &
Brzezinski, 2005]. Although the uncertainty of the measurements is not
quoted, the deviation from the values calculated from equation (E.17) is
about 1.8 %.
From the above presentation it is apparent that the present set of

thermal–conductivity values agree well with the previous sets of
measurements, while at the same time enjoy a lower degree of uncertainty.
Furthermore, in Figure E.4 the volumetric specific heat
measurements as a function of temperature and the percentages deviations
of the values calculated from equation (E.19) are shown. In the same figure
the previous published data of Assael et al. [Assael et al., 2004] and
Hulstrom et al. [Hulstrom et al., 1988] are also shown.
It should be noted that Hulstrom et al. used the differential scanning
calorimeter method for the measurement of the specific heat capacity,
without however quoting the uncertainty of his measurements. Thus, it can
be seen in Figure E.4 that although at low temperature the values agree,
when the temperature increases the deviations from the values calculated
from equation (E.19) go up to 12.5 %.
Figure E.4 Volumetric specific heat measurements of Pyrex 7740 as a function of

E1.4. Polymethyl Methacrylate (PMMA)

PMMA or Perspex is an amorphous, colourless thermoplastic material
of excellent optical transparency and a luminous transmittance of about
92 %. It has good abrasion resistance and dimensional stability but is brittle
and notch sensitive. Its water absorptivity is very low in comparison with
other polymer materials. PMMA was proposed by the National Physical
Laboratory (NPL), U.K, as a possible candidate for thermal conductivity
reference standard in 2001 [Tye & Salmon, 2001]. However a more recent
intercomparison between 17 laboratories organised by Physikalisch-
Technishe Bundesanstalt (PTB), Germany, showed uncertainties in the
thermal conductivity values between 8 % to 13 %, which far exceeded the
laboratories quoted uncertainties. Hence, the employment of PMMA as an
acceptable thermal conductivity standard is still under consideration.
The two samples of PMMA employed in the present work were
produced by casting and supplied by Degussa Rohm Plexiglas GmbH and
were made available by PTB. The samples have dimensions of 5x10x2 cm3
and their density was found by weighing them and determining their
volume equal to 1,200 kg.m-3 at 293.15 K.

thermophysical properties of PMMA. A brief review of these measurements
is presented in Table E.13. It should be noted there are more studies on
thermal conductivity of PMMA but it was chosen not to present studies that
only one measurement and that at “room temperature” quoting no exact
reference temperature, and no uncertainty in their measurement.
the temperature range from 315 K to 353 K and the values obtained are
shown in Table E.14.
Table E.13 Thermophysical properties measurements of PMMA.
Temperature range
(K)
[Eiermann & Hellwege, 1962] Steady state and transient methods 83 – 363 λ ±4.0 %
[Cahill & Pohl, 1987] 3ω technique 30 – 300 λ Not quoted

[Miller & Kotlar, 1993] Transient method 296 λ and α ±5.0 %
[Log & Gustafsson, 1995] Transient plane source 293.15 λ ±5 %
Simultaneous measurement of λ ±2.0 %
[Lima E Silva et al., 1998] 306
thermal conductivity and diffusivity α ±3.7 %
[Tye & Salmon, 2001] Guarded Hot Plate 273.15 – 353.15 λ ±1.0 % at 95 % confidence level
Transient Plane Source and
[Boumaza & Redgrove, 2003] 293 – 343 λ Not quoted
Guarded Hot Plate
λ ±3.6 %
[Malinaric, 2004] Dynamic Plane Heat Source method 293.15
α ±2.7 %
λ Not quoted
Guarded hot-plate and guarded
[Rudtsch & Hammerschmidt, 2004] 200 – 353 α ±4.4 % at 95 % confidence level
heat-flow meter techniques
CP ±1.5 %
λ ±1.5 %
  C P  ±5.0 %
[Kubicar et al., 2005a] Pulse Transient Method 298.15 λ, α and C P Not quoted
λ ±2 %
[Hammerschmidt & Meier, 2006] Transient Hot Bridge 293.15
α ±8 %
[Franco, 2007] Hot wire apparatus 293.15 λ ±5 %
Intercomparison with transient 2 % - 6 % (depending on the
[Rides et al., 2009] 293 – 453 λ
methods method used)
[Malinarič & Dieška, 2009] Dynamic Plane Heat Source method 296.15 λ and α ±1.5 %
Table E.14 Measurement of properties of PMMA as a function of temperature
T λ Δλ   C P     C P 
(K)  W  m1  K 1  (%)  kJ  m3  K 1  (%)
314.61 0.192 -0.11 1,728 0.45
324.32 0.194 -0.10 1,764 -0.51
333.89 0.197 0.43 1,814 -0.58
343.94 0.198 -0.10 1,896 0.88
353.67 0.200 -0.10 1,927 -0.27

equation:
i
 T 
    298.15 K    i   , (E.22)
i  298.15 

   fit 
  100   . (E.23)
 fit 

also shown inTable E.14, and were fitted to a function of the absolute
i
 T 
  C P     C P   298.15 K    bi   , (E.24)
i  298.15 
where the coefficients bi and the value of   C P   298.15 K  is given

inTable E.15. The deviation of the experimental volumetric specific heat
given by:
   C P     C P fit 
   C P   100    . (E.25)

   C P fit 
Equation (E.22)
λ(298.15Κ)  W  m1  K 1  0.1889
0 (-) 0.6771
1 (-) 0.3226
Equation (E.24)
  C P   298.15 K   kJ  m3  K 1  1,631
b0 (-) 0.0071
b1 (-) 0.9928
The maximum deviations of the experimental points, presented

inTable E.14, from equations (E.22) and (E.24) are 0.43 % for thermal
As it was mentioned before, in 2001, Tye and Salmon after a very
careful investigation aiming to produce a candidate thermal conductivity
reference polymer, proposed an equation for the thermal conductivity of
PMMA with an uncertainty of ±1 % at the 95% confidence level [Tye &
Salmon, 2001]. The proposed equation that was valid over the temperature
range of 273.15 K to 353.15 K is
  0.18774  1.3396 104   , (E.26)
where Θ is the mean specimen temperature in oC.

The deviations of this equation, also shown in Figure E.5, from
present values calculated by equation (E.22) are in excellent agreement with
the present measurements.
Moreover, in 2004, Rudtsch and Hammerschmidt, also aiming to
propose a thermal conductivity reference material candidate, coordinated in

PTB an intercomparison project for PMMA involving 17 European
laboratories. Unfortunately the thermal conductivity values produced even
at 303 K ranged from 0.16 to 0.21 W·m-1·K-1. It was argued that the most
possible reason for this discrepancy was the not–properly–treated effect of
contact resistance. To independently prove that the thermal contact
resistance was adequately taken into account, at the same paper, values
obtained by PTB by two different methods (guarded–hot plate, transient–
hot strip) were also reported [Rudtsch & Hammerschmidt, 2004]. The
deviations of these values, shown in Figure E.5, from the present values
calculated by equation (E.22), are well within the mutual uncertainty of the
instruments.
There are more other investigators who have measured the thermal
conductivity of PMMA. Boumaza and Redgrove in 2003, employed a
transient plane–source (Gustaffson probe) and a guarded hot–plate
instrument to investigate the thermal conductivity of PMMA with
temperature [Boumaza & Redgrove, 2003]. Their result, characterised by
5% reproducibility, shown in Figure E.5, agree very well with the present
measurements.
It should also be mentioned the very interesting comparisons of
various methods of measurements employed for the thermal conductivity of
PMMA, done by Kubicar and Bohac [Kubičár & Boháč, 2002], Lockmuller et
al. [Lockmuller et al., 2003] and Rides et al. [Rides et al., 2009].
In Figure E.5, PMMA thermal conductivities measurements of other
investigators are also shown:
i. Miller and Kotlar used the same experimental device used for the
measurement of Pyrex 7740, for the simultaneous measurement of the
thermal conductivity and thermal diffusivity [Miller & Kotlar, 1993].
The device was operated in transient state and the thermal conductivity
value at 296 K was measured 0.18 W  m1  K 1 , with an estimated
uncertainty of ±5 %. Thus, the measured value is within the mutual
uncertainties of the two instruments.
ii. Log and Gustafsson employed the transient plane source technique for
measuring the thermal conductivity of PMMA with an uncertainty of
±5 % [Log & Gustafsson, 1995]. They reported a value at 293.15 K which
agrees with the thermal conductivity value calculated from equation
(E.22).
Figure E.5 Thermal conductivity measurements of PMMA as a function of temperature and

their deviations from the values calculated from equation (E.22).
iii. Lima e Silva et al. measured simultaneously the thermal diffusivity

and the thermal conductivity of PMMA with a single surface sensor
[Lima E Silva et al., 1998]. They claim an uncertainty of ±2 % for the
thermal conductivity, and as it can be seen from Figure E.5 the reported
value is within the mutual uncertainties of the instruments.
iv. The previous measurements of Assael et al. [Assael et al., 2005],
performed with an uncertainty of ±1.5 %, agree very well with the
v. Hammerschmidt and Meier employed the new transient hot–bridge
sensor for measuring the thermal conductivity of PMMA with an
estimated uncertainty of ±2 % [Hammerschmidt & Meier, 2006]. The
reported at 293.15 K a value of 0.1934 W  m1  K 1 , which lies in the
mutual uncertainties of both methods.
vi. Finally, Malinarič and Dieška used the dynamic plane source method
for measuring the thermal conductivity and the thermal diffusivity of
PMMA. They claimed an uncertainty of ±1.5 %, and their measured
value is in excellent agreement with the value calculated from equation
(E.22).
In Figure E.5 the volumetric specific heat measurements as a

function of temperature and the percentages deviations of the values
calculated from equation (E.24) are shown. In the same figure the data
acquired by Rudtsch and Hammerschmidt through differential scanning
calorimeter with an uncertainty of ±1.5 % are shown [Rudtsch &
Hammerschmidt, 2004]. Moreover, the previous published data of Assael et
al. [Assael et al., 2005] are also illustrated. It can be seen that the measured
values are in perfect agreement with both previous reported values.
Figure E.6 Volumetric specific heat measurements of PMMA as a function of temperature

and their deviations from the values calculated from equation (E.24).
E1.5. Borosilicate Crown Glass BK7

BK7 is a Borosilicate Crown Glass that is commonly used material for
optical components and can be manufactured with outstanding
homogeneity. It has isotropic thermophysical properties with an excellent
long–term stability. In 2002, an intercomparison between 11 European
laboratories was organized by Physikalisch–Technische Bundesanstalt
(PTB) aiming to qualify it as a possible candidate reference material for
thermal conductivity in the temperature range 173 K to 773 K [Rudtsch et
al., 2002].
The two samples of BK7 employed in the present work were
manufactured and supplied by Schott AG, and were made available by PTB.
The samples have dimensions of 5x10x2 cm3 and their density was found by
weighing them equal to 2,504 kg.m-3 at 293.15 K. Further properties of this
specific material can also be found in literature [Rudtsch et al., 2002]. The
typical composition of BK7 glass can be seen in Table E.16.
Table E.16 Typical chemical composition of BK7
Composition
Component
(%)
SiO2 60 – 70
B2O3 10 – 20
Na2O 1 – 15
K2O 5 – 15
BaO 1 – 10
Miscellaneous traces <1

It should be noted that there are not many investigators that have
performed measurements of the thermal conductivity of BK7. In Table E.17
a brief review of the main thermophysical properties measurements
performed on BK7 is presented.
Table E.18.
Table E.17 Thermophysical properties measurements of BK7.
Temperature range
(K)
Laser flash, Hot – wire and
[Ebert, 2002] 223 – 773 λ Not quoted
double plate techniques
λ Depending on the technique used
[Rudtsch et al., 2002] Interlaboratory comparison 173 – 773 α Depending on the technique used
CP Depending on the technique used
λ ±1.5 %
  C P  ±5.0 %
λ ±5.0 %
[Kubicar et al., 2005b] Pulse Transient Method 298.15 α ±3.7 %
CP ±2.0 %
λ ±2.0 %
[Hammerschmidt & Meier, 2006] Transient Hot Bridge 293.15
α ±8.0 %
α ±5.0 %
[Hemberger et al., 2010] Laser Flash method 80 – 300
5.0 % - 10.0%
CP
(depending on temperature)
Table E.18 Measurement of properties of BK7 as a function of temperature
T λ Δλ   C P     C P 
(K)  W  m1  K 1  (%)  kJ  m3  K 1  (%)
315.87 1.080 -0.67 1,970 0.22
333.91 1.120 0.90 2,013 -0.12
353.15 1.135 0.42 2,061 -0.45
372.70 1.146 -0.08 2,134 0.30
392.40 1.155 -0.38 2,184 -0.14
412.55 1.164 -0.34 2,259 0.43
433.04 1.178 0.49 2,309 -0.26

equation:
i
 T 
    298.15 K    i   , (E.27)
i  298.15 

   fit 
  100   . (E.28)
 fit 

i
T 
  C P     C P   298.15 K    bi   , (E.29)
i  298.15 

given by:
   C P     C P fit 
   C P   100    . (E.30)

   C P fit 
Equation (E.27)
λ(298.15Κ)  W  m1  K 1  1.0617
0 (-) 0.11921
1 (-) 1.32874
2 (-) -0.44795
Equation (E.29)
  C P   298.15 K   kJ  m3  K 1  1,919
b0 (-) 0.7026
b1 (-) 0.1877
b2 (-) 0.1096
The maximum deviations of the experimental points, presented in

Table E.18, from equations (E.27) and (E.29) are 0.90 % for thermal
It should be mentioned that the results of the interlaboratory
comparison on BK7, coordinated by PTB in 2002, showed typically an 8 %
spread in the low temperature range, while at higher temperatures this rose
up to 40 % [Rudtsch et al., 2002]. Thus, it was concluded that further
investigation was necessary, and therefore the corresponding data is not
included in Figure E.7.
The previous measurements of Assael et al. [Assael et al., 2005],
performed with an uncertainty of ±1.5 %, agree very well with the present
measurements performed with the lower uncertainty of ± 1 %. Only in one
temperature the measurements are just over the mutual uncertainties of
the two instruments.
Kubicar et al. [Kubicar et al., 2005b] employed a pulse transient
method with a quoted uncertainty of 5%. The deviation of his value, shown
in Figure E.7, from that calculated by equation (E.27), is well within the
mutual uncertainties. Ebert employed a hot wire instrument and a laser
flash one [Ebert, 2002]. No uncertainty was specified. His results for both
methods, shown in Figure E.7, deviate up to 4.3 % from the values
calculated by (E.27).
Figure E.7 Thermal conductivity measurements of BK7 as a function of temperature and

In 2006, Hammerschmidt and Meier employed the new transient hot–

bridge sensor for measuring the thermal conductivity of BK7 with an
estimated uncertainty of 2 % [Hammerschmidt & Meier, 2006]. The
reported value at 293.15 K is higher than all the other published data and
therefore the deviation from equation (E.27) is 5.3 %, higher than the
mutual uncertainties of both methods.
In Figure E.8 the volumetric specific heat measurements as a
function of temperature and the percentages deviations of the values
calculated from equation (E.29) are shown. In the same figure the previous
published data of Assael et al. [Assael et al., 2005] and Kubicar et al.
[Kubicar et al., 2005b] are also shown. The measured values have excellent
agreement.
Hemberger et al. have also measured the thermal conductivity and
the specific heat of BK7 with the laser flash technique [Hemberger et al.,
2010]. However, the measurements are in the temperature range of 80 K to
300 K, and therefore are not included in Figure E.7 and Figure E.8.
Figure E.8 Volumetric specific heat measurements of BK7 as a function of temperature and
E2. Composite Materials

Various techniques have been used in the past in order to measure
experimentally the thermal conductivity of conventional fibre reinforced
materials. Among others, the techniques used for the measurement of
composites are the guarded hot-plate technique [Oram & Wolff, 1998], the
transient plane source [Kalaprasad et al., 2000; Wang & Qiu, 2010], the
comparative heat flow method [McIvor et al., 1990; Gowayed & Hwang,
1995; Sweeting & Liu, 2004], the differential scanning calorimeter method
[Kalogiannakis et al., 2004; Cecen et al., 2009], and the hot wire method
[Yen et al., 1991]. Moreover, several attempts has been made to measure the
thermal conductivity of composites reinforced with carbon nanotubes [Song
& Youn, 2005; Gojny et al., 2006; Moisala et al., 2006; Thostenson & Chou,
2006; Yang et al., 2010; Zhou et al., 2010]. Nevertheless, no single technique
exists for measuring the thermal conductivity of composites that can provide
the basis for establishment of a standard. The reliability of a specific
method depends on various factors, such as the speed of operation, the
required accuracy and execution under various environmental conditions,
and the size and shape of the available specimen.
Therefore, it is of interest to use the developed transient hot–wire
device for the measurement of the thermal conductivity of a solidified epoxy
polymer along with certain reinforcements. To this respect, in this work the
transient hot–wire technique is employed in the measurement of the
thermal conductivity of conventional glass fibre reinforced composites, as
well as nanotube reinforced composites. Moreover, the technique is used to
study the enhancement of the thermal conductivity of glass fibre composite
laminates in the presence of carbon multi–walled nanotubes (C-MWNT),
which has not been considered yet.
In the present work, plain epoxy–resin blocks were initially measured
in order to obtain the thermal conductivity of the matrix and serve as the
control measurements. Then specimens of epoxy resin were manufactured
and measured using as reinforcements:
i. glass fibres (GF),

ii. carbon multi-walled nanotubes (C-MWNTs), and
iii. glass fibres and carbon multi-walled nanotubes (GF + C-MWNTs). .
The experimental procedure followed for the measurement and

calculation of the thermal conductivity of the prepared composites samples

is the same as the one described in details in Chapter D. For each
measurement two composite specimens were used. Each specimen had at
least 5 mm thickness and dimensions 10x5 cm2. The measurements were
performed in the available three zone temperature furnace and each
measurement had a duration of 10 s.
E2.1. Epoxy Resin
E2.1.1. Samples Preparation

In order to measure the thermal conductivity of the epoxy-resin
polymer, different sets of epoxy–resin blocks were prepared. Each set
consisted of two blocks. The epoxy resin used was the EPIKOTETM 816 LV
and the curing agent was EPIKURETM 207, available from Hexion Specialty
Chemicals, Inc., Ohio, U.S.A. The resin has a very low viscosity, low
volatility, excellent pigment/filler and substrate wetting and is commonly
employed in the building and civil engineering industries, e.g. flooring
compounds, adhesives, mortars and grouts. The epoxy was formed from a
mixture of resin and curing agent (which is a modified polyamine) at a ratio
100/47 by weight respectively. The two components of the epoxy resin were
mixed and the mixture was stirred mechanically. The mixture was poured
into moulds with dimensions of 10x5 cm2 and cured at a temperature of
293 K for a time of 24 h followed by post–curing for 2 h at a temperature of
363 K. After the post–curing process, both sides of the samples were
polished in an attempt to reduce contact thermal of the measurements.

In order to evaluate the homogeneity of the samples, seven different
sets of two specimens of epoxy resin were prepared by weighing as described
above. The measured values of the thermal conductivity of these samples at
307.5 K are shown in Table E.20 and plotted in Figure E.9.
Table E.20 The thermal conductivity of different sets of

epoxy–resin polymer at 307.5 K.
Thermal conductivity, λ
No of run  W  m1  K 1 
1 0.260
2 0.252
3 0.255
4 0.251
5 0.260
6 0.250
7 0.255
Figure E.9 Thermal conductivity of different sets of epoxy–resin polymer at 307.5 K.
The above values indicate that although the transient hot–wire

technique used is accurate to ±1 %, the actual values measured differ by
±2 % because of inhomogeneities in the samples, probably attributed to
factors like imperfect mixing or non–identical composition.
Consequently, the thermal conductivity of the epoxy–resin polymer as
a function of temperature was measured. The results are shown in Table
E.21. The thermal conductivity, λ  W  m1  K 1  , values shown in Table
E.21, were fitted to a function of the absolute temperature T (K) by the
following equation:
  T 
    298.15 K  0.1118  0.880    , (E.31)
  298.15  
where the value of the thermal conductivity, λ(298.15Κ)  W  m1  K 1  is

given in Table E.21.
Moreover, the deviation of the experimental thermal conductivity
values from the one obtained from equation (E.31), λfit_1, is calculated from:
   fit _1 
  100   . (E.32)
 fit _1 
It can be seen in Table E.21 that the maximum deviation of the

experimental points from equation (E.31) is -1.07 % and the calculated
standard deviation at the 95% confidence level is 0.69 %.
Table E.21 Thermal conductivity epoxy–resin polymer

T λ Δλ
(K)  W  m1  K 1  (%)
298.15 0.252 0.83
307.46 0.255 -0.72
320.32 0.267 0.22
335.84 0.275 -1.07
348.78 0.290 0.84
E2.2. Epoxy Resin Reinforced with Glass Fibers

For the preparation of the reinforced epoxy–resin samples, an
elongated–glass (e–glass) plain weave fabric of thickness 0.15 mm was
obtained from Hexcel, CT, U.S.A. Its nominal weight is 206 g  m2 , and its
nominal construction is 7.4 fils–yarn.cm-1. The plain weave fabric was
chosen because it is characterized by the simplest repetitive geometry
among all woven fabrics and would not create great measurement
uncertainties. The unit cell of a balanced plain weave consists of two yarns
in both the fill and warp directions.
To fabricate the laminate the wet lay–up method was employed. The
laminate thickness was about 5 mm, with 20 layers of elongated–glass plain
weave fabric placed in a stacking sequence (0º/90º)20. The fiber volume
fraction was 28 %, in order to measure the minimum effect that the fibres
may have on the enhancement of the thermal conductivity. The laminate
was left to cure at room temperature for a time of about 48 h and then it
was post–cured at a temperature of 363 K for a time of 2 h. After the post–
curing process 9 smaller samples (5x10 cm2) were cut out of the laminate
using a diamond tipped rotary saw. It should be noted that 5 of the samples
had the 0o direction of fibres in parallel with the longer axis of the sample,
while the rest of them were cut so as to have an angle of 60º with that axis.

The measurements of the thermal conductivity of the glass fibre
reinforced epoxy resin as a function of the temperature are shown in Table
E.22. The thermal conductivity, λ  W  m1  K 1  , values shown in Table
E.22, were fitted to a function of the absolute temperature T (K) by the
following equation:
  T 
    298.15 K  0.3784  0.6254    , (E.33)
  298.15  
where the value of the thermal conductivity, λ(298.15Κ)  W  m1  K 1  is

given in Table E.22.
Moreover, the deviation of the experimental thermal conductivity
values from the one obtained from equation (E.33), λfit_2, is calculated from:
   fit _ 2 
  100   . (E.34)
 fit _ 2 
It can be seen in Table E.22 that the maximum deviation of the

experimental points from equation (E.33) is 0.61 %, and the calculated
standard deviation at the 95% confidence level is 0.29 %.
Table E.22 Thermal conductivity epoxy–resin polymer reinforced with glass fibres
Material T λ Δλ
(K)  W  m1  K 1  (%)
Epoxy–resin polymer + 298.15 0.320 -0.38
glass fabric (0o/90o)20 312.20 0.330 -0.20
320.12 0.338 0.61
333.81 0.347 0.54
348.50 0.353 -0.57
Epoxy–resin polymer +
glass fabric (0o/90o)20 298.15 0.324
(laminate cut at 60o from 306.93 0.332
principal axis) 320.27 0.345
It should be noted that the first five values in Table E.22 were
obtained by placing the samples with the glass fibres parallel to the sample
axis, thus parallel to the wires of the sensor. Measurements on samples
with the 60º angle show that the orientation of the fibres had no real effect
on the thermal conductivity. It seems that the isotropic nature resulting
from the amorphous character of the glass fibres probably nullifies the effect
of orientation of the fibres in the composite.
In order to evaluate the thermal conductivity enhancement caused
from the presence of glass fibres in the epoxy resin, the above thermal
conductivity measurements are plotted in Figure E.10 together with the
previous measurements of the thermal conductivity of the epoxy resin
(Table E.21).
It can be seen in Figure E.10 that reinforcing the polymer with 28 %
volume fraction of glass fabric, resulted approximately in a 27% increase in
the thermal conductivity. That increase may be a result of the relatively
good conducting nature of the glass fibres compared to the epoxy resin
matrix. The observed increase in thermal conductivity with temperature is a
linear relationship that has resulted from the strong dependence of the
epoxy matrix with the temperature. A similar (linear) relationship was also
obtained between the thermal conductivity and the temperature in the
measurements of the plain epoxy blocks.
Gowayed and Hwang have studied the influence of the fibre volume
fraction on the thermal conductivity of the epoxy–resin composite [Gowayed
& Hwang, 1995]. They have reported a thermal conductivity enhancement
of 37 % for a fibre volume fraction of 34.6 % and an enhancement of 93 % for
a volume fraction of 55.7 %. This results is in accordance to the one acquired

in the present work, where a 28 % glass fibre volume fraction leads to a
thermal conductivity enhancement of 27 %. Moreover, it is expected that the
fibre influence becomes higher as the volume fraction increases, and thus
the thermal conductivity would be expected to increase as the conductive
effect would become more significant.
Figure E.10 Thermal conductivity measurements as a function of temperature of

epoxy–resin polymer and epoxy–resin reinforced with glass fibres.
E2.3. Epoxy Resin Reinforced with Carbon Multi-Walled Nanotubes

The Carbon Multi–Walled Nanotubes (C-MWNTs) employed in this
work were produced by XINTEK Inc., North Carolina. The specimen
macroscopically appeared as large individual formations with a portion of
surrounding dust. Transmission electron microscopy information was
supplied by the producer, indicating that the mean outer diameter of the
C-MWNTs was smaller than 10 nm (Figure E.11). The pristine material was
examined by Scanning Electron Microscopy (SEM, JEOL Model JSM-6300,
20 keV), Figure E.12, which revealed a homogeneous sample of well–defined
nanotubes with lengths greater than 10 μm. Unfortunately, the
arrangement of the specimen did not permit the determination of mean tube
length. These C-MWNTs dimensions and type were chosen because of the
results obtained in previous experiments with liquids [Assael et al., 2006].
Three different sets of carbon multi–walled nanotubes (C-MWNTs)
reinforced epoxy–resin samples were prepared. They consisted of epoxy
resin with 0.25 %, 0.5 % and 2% by weight C-MWNTs respectively. The
C-MWNTs were added in the epoxy liquid and the mixture was
homogenized by ultrasonic vibration (Model HD 2200 Bandelin Electronics).
Following homogenization, the hardener was added and the mixture was
mechanically stirred in order to achieve better homogeneity of the samples
and to minimise agglomeration of the nanotubes.
It should be mentioned that a preliminary study has shown that
increasing the weight percentage of the C-MWNTs in the epoxy–resin
formulation was limited by the dramatic increase in viscosity, making the
fabrication of composites infeasible. It is known that the C-MWNTs are
typically held together as bundles with intrinsic van der Waals forces in
most solvents [Paradise & Goswami, 2007]. When mixed with the epoxy
resin, the C-MWNT exist as entangled aggregates, which lead to an
increased viscosity in the C-MWNT/epoxy system. Surfactant–assisted
methods are reported to achieve a good dispersion, but surfactants are
difficult to remove from the composites [Gong et al., 2000]. Also, some
studies reported that functionalization of the C-MWNTs could improve the
dispersion [Zhu et al., 2003; Hu et al., 2010; Yang et al., 2010] but in this
work non–functionalised nanotubes were used.
Figure E.11 Typical transmission electron microscopy SAD images of the C-MWNTs used.
Figure E.12 Typical scanning electron microscopy of the C-MWNTs used


The three different sets of reinforced epoxy–resin samples were
measured at room temperature (298.15 K), in order to evaluate the effect
that the C-MWNTs’ concentration has on the thermal conductivity of the
composite. The measured thermal conductivity values as a function of the
C-MWNTs mass fraction are shown in Table E.23. In the same table, the
enhancement of the thermal conductivity is also shown.
The enhancement of the thermal conductivity of epoxy–resin
composite, Δλenh, is calculated from:
   o 
enh  100   , (E.35)
 o 
where   0.252 W  m 1  K 1 is the thermal conductivity of plain epoxy–

resin at 298.15 K (Table E.21).
Table E.23 Thermal conductivity epoxy–resin polymer reinforced with C-MWNTs

as a function of the C-MWNTs concentration at 298.15 K.
C-MWNTs’ mass fraction λ Δλenh

(%)  W  m1  K 1  (%)
0.00 0.252 –
0.25 0.265 5.1
0.50 0.290 15.1
2.00 0.275 9.1
Similar results on the enhancement of the epoxy–resin thermal

conductivity are reported elsewhere. Thostenson and Chou reported a linear
dependence of the epoxy–resin thermal conductivity on the nanotube
concentration [Thostenson & Chou, 2006]. In the same paper, it is reported
an enhancement of about 11 % for a mass fraction of 2 % of carbon
nanotubes. Moreover, Gojny et al. studied the influence of the type of carbon
nanotube on the epoxy–resin thermal conduction mechanism and noted that
the use of 0.3 % mass fraction of C-MWNTs lead to a thermal conductivity
enhancement of about 4 % [Gojny et al., 2006]. In the same year, Moisala et
al. published thermal conductivity values of epoxy composites reinforced
with single–walled or multi–walled carbon nanotubes [Moisala et al., 2006].
Among others, it was reported an increase of about 14 % for a mass fraction
of 0.5 % C–MWNTs.
E2.4. Epoxy Resin Reinforced with Glass Fibres and Carbon Multi-Walled
Nanotubes

The samples of epoxy resin reinforced with both glass fibres and
C-MWNTs were prepared in a similar way like before. The C-MWNTs were
added in the epoxy liquid and the mixture was homogenized by ultrasonic
agitation (Model HD 2200 Bandelin Electronics). Following homogenization,
the hardener was added and the mixture was mechanically stirred in order
to achieve better homogeneity of the samples and to minimise
agglomeration of the nanotubes. Once more three different sets of samples
were prepared, containing 0.25 %, 0.5 % and 2% by weight C-MWNTs
respectively, and glass fabric with volume fraction of 28 %.
Sections cut from the composites were studied using scanning
electron microscopy, SEM (JEOL model JSM–6300, operating at 20 keV),
and the results are shown in Figure E.13 and Figure E.14. The scope of the
SEM analyses was to obtain information on the morphology of the
composites. Multiple areas were examined, as well as different samples, and
the concentration of C-MWNTs was found to be homogeneous and similar in
all specimens studied.
In the sample with mass fraction of C-MWNTs of 0.25 %, shown in
Figure E.13, two areas can be identified. Firstly, the woven layers of the
glass fabric are easily distinguished; they are equally spaced throughout the
sample. Secondly, a homogeneous blend of the resin with the carbon
nanotubes is detected between the woven layers of the glass fabric. In
addition, it is possible to observe air bubbles (well–defined circles), the
formation of which is associated with the method of manufacture (wet lay–
up) of the composites, that leave behind a small void fraction. Figure E.14,
shows a typical high–magnification SEM image that reveals some
agglomerated nanotubes at certain locations in the middle, while e–glass is
shown on both sides. It is of importance to stand on the uniformity and the
stability of the complex formation throughout the specimen.
Figure E.13 SEM image of epoxy–resin samples reinforced with C-MWNTs mass fraction of
0.25 % and glass fibres.
Figure E.14 Magnified SEM image of epoxy–resin samples reinforced with C-MWNTs mass
fraction of 0.25 % and glass fibres.
It should be noted, that the aforementioned behavior of the

C-MWNTs reinforced epoxy, having high viscosity, singled out the choice of
the fabrication method to the wet lay up technique. Clearly, the
impregnation of the resin in advanced composite fabrication techniques such
as the Resing Transfer Molding (RTM) and the Vacuum Infusion (VI)
processes is a complex process and is greatly influenced by the resin
viscosity. Also, in such processes the phenomenon of filtering the nanotubes
has been observed [Lin et al., 2006]. This is due to the fabric structure that
contains gaps of different scales that are blocked and lead to a very slow or
even insufficient resin impregnation.

The three different sets of epoxy–resin samples reinforced with both
glass fibres and C-MWNTs were measured at room temperature (298.15 K),
in order to evaluate the effect that the C-MWNTs’ concentration has on the
thermal conductivity of the composite. The measured thermal conductivity
values as a function of the C-MWNTs mass fraction are shown in Table
E.24. In the same table, the enhancement of the thermal conductivity of the
epoxy–resin composite, Δλenh, (calculated from equation (E.35)) is also
shown.
Table E.24 Thermal conductivity epoxy–resin polymer reinforced with glass fibres and
C-MWNTs, as a function of the C-MWNTs concentration at 298.15 K.
C-MWNTs’ mass fraction λ Δλenh

(%)  W  m1  K 1  (%)
0.00 0.320 27.0
0.25 0.367 45.6
0.50 0.399 58.3
2.00 0.373 48.0
In figure the enhancement on the thermal conductivity is plotted

versus the C-MWNTs’ mass fraction. In the same figure, the enhancement
values of Table E.23 are included for comparison purposes.
Figure E.15 Thermal conductivity enhancement as a function of

the composition of C-MWNTs. Black lines show trends only.
From the trend in the above figure, it seems that reinforcing the
polymer only with C-MWNTs resulted approximately in a 20% increase in
the thermal conductivity with 1.2 % mass fraction C-MWNTs. Similarly,
reinforcing the epoxy resin with both C-MWNTs and glass fibres produced a
thermal conductivity enhancement of about 60 %, at the same weight
concentration of C-MWNTs. The apparent maximum observed in both
curves can be attributed to the following:
i. The conductive path threshold limit between plies of the glass fabric. A
recent study that has investigated nanotube reinforced glass fibre
composites reported that the C-MWNTs are repelled by the glass–fibres
[Wichmann et al., 2006].
ii. The breakage of the length of the nanotubes, as a result of a stronger
mixing in order to achieve uniformity, and
iii. The difficulties observed in proper mixing at the high concentration of
2%. .
Moreover, a similar maximum in the enhancement of the thermal

conductivity has been also observed with C-MWNTs in liquids [Assael et al.,
2006], and it was shown that the critical parameter was the length to
diameter ratio of the nanotubes.
Finally, it should be mention though, that the measured effective
thermal conductivity of the composites might be affected by the quality of
the nanotubes, the preparation method, and the measuring technique. It is
noted once more that although the employed transient hot–wire technique
has an uncertainty of 1 % for the measurement of solids, when used for the
measurement of composites, the uncertainty of the results become 2 % due
to inhomogeneities in the samples.
E3. Summary
In this chapter the developed transient hot wire instrument is
employed for the measurement of the thermal conductivity and of the
volumetric specific heat, as a function of temperature, of:
i. The intermediate silicone layer, in which the hot wire is enclosed. The
silicones measured are the AS1803 and the BORO Type 650.
ii. Two certified thermal conductivity reference solid materials,
Pyroceram 9606 (BCR–724) and Pyrex 7740 (BCR–039)
iii. Two candidate reference materials. The glassy material measured are
Polymethyl Methacrylate (PMMA) and Borosilicate Crown Glass
(BK7).
iv. Epoxy–resin composites reinforced initially reinforced with plain
weave glass fabric and then with carbon multi–walled nanotubes
(C–MWNTs). In the last part, the influence of both reinforcements at
hand is studied.
The measured thermal conductivity and volumetric specific heat

values of the above solid materials were fitted in two equations
i
 T 
    298.15 K     i   , (E.36)
i  298.15 
i
T
and   C P     C P   298.15 K    bi  
 . (E.37)
i  298.15 
The uncertainty of the measurements for the silicone pastes and the
solid materials is ±1 %, whereas for the measurement of the composite

polymers it increases to ±2 % due to inhomogeneities in the samples.
Moreover, for each solid material, the present measurements are
compared with the data of other investigators and in all cases it was found
that the deviations are within the mutual uncertainty of the measurements.
Finally, as far as the composites materials are concerned, the results
revealed for the epoxy reinforced with glass fibers, with volume fraction of
28 %, that the thermal conductivity increase was 27 % compared to plain
epoxy resin. When reinforced with C–MWNTs the maximum enhancement
observed was about 20 % at a concentration of 1.2 % by weight of C–MWNT.
Similarly, when reinforced with both the C–MWNTs and glass fibres the
maximum thermal conductivity enhancement observed was about 60 % at a
concentration of 1.2 % by weight of C–MWNTs.
F
Innovation – Future Work
The chapter analyses the innovations achieved in the present thesis

and describes the future challenges facing the transient hot wire technique.
Moreover, the possible subsequent research directions are also shown.
196 Innovation – Future Work
F1. Innovations
In this thesis the thermal conductivity of solid materials and
polymers composites is measured with the transient hot wire technique. The
main innovations achieved are:
i. The new sensor design, except from the robustness that provides to the
sensor, ensures that the thickness of the silicone paste of the hot wire
sensor is exactly 1 mm. In this way, the uncertainty source related to
the paste thickness is compensated.
ii. The overall uncertainty, including both experimental and numerical
errors, is estimated to be better than  1 % for the measurement of the
thermal conductivity of solid materials. Moreover, the designed
transient hot wire sensor can be used also for the measurement of the
thermal conductivity of liquids with an uncertainty better than  0.5 %.
iii. The created and employed Comsol Multiphysics finite element model
accurate described the experimental setup. The model allowed the
closely reproduction (  50 mK) of the temporal temperature rise on the
transient hot wire over four order of time magnitude, from the
beginning of the time scale. As the heat wave firstly crosses through the
silicone paste, data at short times (20 μs to 0.5 s) are used for the
measurement of the paste’s thermal properties. Data at longer times
(>0.5 s) are used for the calculation of the thermal conductivity of the
solid specimen. The same model was employed in the case of measuring
liquids.
iv. The thermal conductivity of the certified reference materials Pyroceram
9606 and Pyrex 7740, and of the candidates reference materials
Polymethyl Methacrylate (PMMA) and Borosilicate Crown Glass BK7,
were measured with an uncertainty of  1 %.
v. The new transient hot wire sensor was used for the measurement of the
thermal conductivity of composite materials for first time. The
measurements of the epoxy–resin composites were performed as a
function of temperature, and the enhancement caused by reinforcing
them with glass fibres or/and carbon multiwall nanotubes was also
studied. The uncertainty of the technique in this case increased to ±2 %,
due to inhomogeneities in the sample preparation, being though the
lowest measuring uncertainty reported for this type of materials.
Innovation – Future Work 197
F2. Future Work

Despite the improvements and the innovations achieved in the
present work, there are still several research challenges to be faced. First of
all, the use of elastic silicone paste in the sensor design, imposes an upper
temperature limit on the thermal conductivity measurements, and make
impossible the high–temperature measurements. Thus, the challenge is to
substitute the silicone paste with other material that will combine both, the
elasticity of the silicone paste with the high temperature performance.
Another, challenge concerning the silicone paste, is the employment
of a paste with higher thermal conductivity in order to be able to measure
good thermal conductors, such as metals. Metals have thermal conductivity
in the order of 10 to 100 and therefore the measurement of their thermal
conductivity with the present device is not possible, as the paste prevents
the heat transfer from the wire to the specimen and the solid’s thermal
conductivity barely affects the evolution of the temperature of the wire.
Thus, further research should be concentrated in order to expand the
thermal conductivity application limits of the device.
An interesting research challenge is the modification of the present
device in order to be able to measure thin films. The actual device can
measure the thermal conductivity of specimens with at least 3 mm
thickness. Therefore, the experimental sensor needs to be modified, and the
intermediate silicone layer to be substituted by a material of higher thermal
conductivity.
Finally, improvements can be made in the iterative inverse analysis
followed for the calculation of the thermal conductivity of the measured
material. Currently the material properties are manually adjusted in
Comsol Multiphysics model. However, there is the possibility to use Comsol
Multiphysics Script language in order to make the procedure automatic and
reduce the post–processing needed time.
References
A
Agari, Y., Ueda, A., & Nagai, S., 1991, “Thermal conductivity of a
polyethylene filled with disoriented short-cut carbon fibers”, Journal
of Applied Polymer Science, 43: 1117-1124.
Alloush, A., Gosney, W. B., & Wakeham, W. A., 1982, “A transient hot-wire
instrument for thermal conductivity measurements in electrically
conducting liquids at elevated temperatures”, International Journal
of Thermophysics, 3: 225-235.
Amazouz, M., & Gauvin, R., 1997, “Solving heat dissipation problems in
unidirectional composite structure”, Journal of Reinforced Plastics
and Composites, 16: 686-697.
American Society for Testing and Materials, 1992, Standard Specification
for Glasses in Laboratory Apparatus, ASTM Designation E438-92.
Philadelphia, U.S.A.
American Society for Testing and Materials, 1993, Standard Test Method
for Thermal Conductivity of Refractories, ASTM Designation C201-
93. Philadelphia, U.S.A.
for Steady-State Heat Flux Measurements and Thermal
Transmission Properties by Means of the Guarded-Hot-Plate
Apparatus, ASTM Designation C177-97. Philadelphia, U.S.A.
for Thermal Conductivity of Refractories by Hot Wire (Platinum
Resistance Thermometer Technique), ASTM Designation C1113-99.
for Thermal Diffusivity br the Flash Method, ASTM Designation
E1461-01. Philadelphia, U.S.A.
for Steady-State Thermal Transmission Properties by Means of the
200 References
Heat Flow Meter Apparatus, ASTM Designation C518-02.

Angell, M. F., 1911, “Thermal Conductivity at High Temperatures”, Physical
Review (Series I), 33: 421.
Assael, M. J., Antoniadis, K. D., Kakosimos, K. E., & Metaxa, I. N., 2008,
“An improved application of the transient hot-wire technique for the
absolute accurate measurement of the thermal conductivity of
pyroceram 9606 up to 420 K”, International Journal of
Thermophysics, 29: 445-456.
Assael, M. J., Botsios, S., Gialou, K., & Metaxa, I. N., 2005, “Thermal
conductivity of polymethyl methacrylate (PMMA) and borosilicate
crown glass BK7”, International Journal of Thermophysics, 26: 1595-
1605.
Assael, M. J., Charitidou, E., Georgiadis, G. P., & Wakeham, W. A., 1988,
“Absolute Measurement of the Thermal Conductivity of Electrically
Conducting Liquids”, Berichte der Bunsen-Gesellschaft für
Physicalishes Chemie, 92: 627-631.
Assael, M. J., Chen, C. F., Metaxa, I., & Wakeham, W. A., 2004a, “Thermal
conductivity of suspensions of carbon nanotubes in water”,
International Journal of Thermophysics, 25: 971-985.
Assael, M. J., Dix, M., Gialou, K., Vozar, L., & Wakeham, W. A., 2002,
“Application of the Transient Hot-Wire Technique to the
Measurement of the Thermal Conductivity of Solids”, International
Journal of Thermophysics, 23: 615-633.
Assael, M. J., Dix, M., Lucas, A., & Wakeham, W. A., 1981, “Absolute
determination of the thermal conductivity of the noble gases and two
of their binary mixtures as a function of density”, Journal of the
Chemical Society, Faraday Transactions 1: Physical Chemistry in
Condensed Phases, 77: 439-464.
Assael, M. J., & Gialou, K., 2003, “A Transient Hot-Wire Instrument for the
Measurement of the Thermal Conductivity of Solids up to 590 K”,
Assael, M. J., Gialou, K., Kakosimos, K., & Metaxa, I., 2004b, “Thermal
conductivity of reference solid materials”, International Journal of
Assael, M., Metaxa, I., Kakosimos, K., & Constantinou, D., 2006, “Thermal
Conductivity of Nanofluids – Experimental and Theoretical”,
B
Baba, T., & Ono, A., 2001, “Improvement of the laser flash method to reduce
uncertainty in thermal diffusivity measurements”, Measurement
Science and Technology, 12: 2046-2057.
References 201
Barth, G., Gross, U., & Wulf, R., 2007, “A new panel test facility for effective
thermal conductivity measurements up to 1,650°C”, International
Bennet, S., & Sims, G., 2010, “Evolving needs for metrology in material
property measurements - the conclusions of the CIPM Working Group
on Materials Metrology”, Metrologia, 47: 1-17.
Bilek, J., 2006, Sensors for the Thermal Conductivity at High
Temperatures, Phd. Thesis, University of Southampton,
Southampton
Birge, N. O., & Nagel, S. R., 1987, “Wide-frequency specific heat
spectrometer”, Review of Scientific Instruments, 58: 1464-1470.
Blumm, J., Lindemann, A., & Min, S., 2007a, “Thermal characterization of
liquids and pastes using the flash technique”, Thermochimica Acta,
455: 26-29.
Blumm, J., Lindemann, A., Niedrig, B., & Campbell, R., 2007b,
“Measurement of selected thermophysical properties of the NPL
certified reference material stainless steel 310”, International Journal
of Thermophysics, 28: 674-682.
Boumaza, T., & Redgrove, J., 2003, “Use of the Transient Plane Source
Technique for Rapid Multiple Thermal Property Measurements”,
Burge, H. L., & Robinson, L. B., 1968, “Thermal conductivities of He, Ne, Ar,
and of their mixtures”, Journal of Applied Physics, 39: 51-53.
Burr, A. C., 1934, “Notes on the History of the Experimental Determination
of the Thermal Conductivity of Gases”, Isis, 21: 169-186.
C
Cahill, D. G., 1990, “Thermal conductivity measurement from 30 to 750 K:
The 3Ο method”, Review of Scientific Instruments, 61: 802-808.
Cahill, D. G., & Pohl, R. O., 1987, “Thermal conductivity of amorphous
solids above the plateau”, Physical Review B, 35: 4067-4073.
Cecen, V., Tavman, I. H., Kok, M., & Aydogdu, Y., 2009, “Epoxy- and
polyester-based composites reinforced with glass, carbon and aramid
fabrics: Measurement of heat capacity and thermal conductivity of
composites by differential scanning calorimetry”, Polymer
Composites, 30: 1299-1311.
Carslaw, H. S., & Jaeger, J. C., 1959, “Conduction of Heat in Solids”, (2nd
ed.), Oxford: Oxford University Press.
Charitidou, E., 1990, Absolute Method for the Measurement of the Thermal
Conductivity of Liquids, Phd. Thesis, Aristotle University of
Thessaloniki, Thessaloniki (in Greek).
Clausius, R., 1862a, “LIX. On the conduction of heat by gases”, Philosophical
Magazine Series 4, 23: 417 - 435.
202 References
Clausius, R., 1862b, “LXXI. On the conduction of heat by gases”,

Philosophical Magazine Series 4, 23: 512 - 534.
COMSOL Multiphysics User's Guide, Version 3.2. (2005). Stockholm:
COMSOL AB.
Crepeau, J., 2007, “Josef Stefan: His life and legacy in the thermal sciences”,
Experimental Thermal and Fluid Science, 31: 795-803.
D
Davis, P. S., Theeuwes, F., Bearman, R. J., & Gordon, R. P., 1971, “Non-
steady-state, hot wire, thermal conductivity apparatus”, The Journal
of Chemical Physics, 55: 4776-4783.
de Groot, J. J., Kestin, J., & Sookiazian, H., 1974, “Instrument to measure
the thermal conductivity of gases”, Physica, 75: 454-482.
de Vries, D. A., 1952, “A Nonstationary Method for Determining Thermal
Conductivity of Soil in Situ”, Soil Science, 73: 83-90.
Deutsches Institut für Normung, 1976, Bestimmung der Wärmeleitfähigkeit
bei Temperaturen bis 1600 oC nach dem Heiβdraht-Vefahren, DIN 51
046.
Donaldson, A. B., & Taylor, R. E., 1975, “Thermal diffusivity measurement
by a radial heat flow method”, Journal of Applied Physics, 46: 4584-
4589.
E
Ebert, H. P., 2002, “Vergleich von Messverfahren zur Bestimmung der
Wärmeleitfähigkeit”, ZAE Bayern - Tätigkeitsbericht 31.
Eiermann, K., & Hellwege, K. x. (Writer) (1962). Thermal conductivity of
high polymers from −180°C. to 90°C: Interscience Publishers, Inc.
Eucken, A., & Englert, H., 1938, “Die experimentelle Bestimmung des
Wärmeleitvermögens einiger verfestigter Gase und
Flüssigkeiten”, Zeitschrift für die gesamte Kälte-Industrie, 45:
109 - 118.
European Commitee for Standardization, 1997, Advanced technical
ceramics - Monolithic ceramics - Thermo-physical properties - Part 2:
Determination of thermal diffusivity by the laser flash (or heat pulse)
method, EN 821-2.
European Commitee for Standardization, 1998a, Methods for Testing Dense
Shaped Refractory Products - Determination of thermal Conductivity
by the Hot-Wire (Cross-Array) Method, EN 993-14.
European Commitee for Standardization, 1998b, Methods for Testing Dense
Shaped Refractory Products - Determination of thermal Conductivity
by the Hot-Wire (Parallel) Method, EN 993-15.
References 203
F
Filla, B. J., 1997, “A steady-state high-temperature apparatus for
measuring thermal conductivity of ceramics”, Review of Scientific
Instruments, 68: 2822-2829.
Fischer, J., 1939, “Zur Bestimmung der Wδrmeleitfδhigkeit und der
Temperaturleitfδhigkeit aus dem Ausgleichvorgang beim
Schleiermacherschen Meίrohrverfahren und beim Plattenverfahren”,
Annalen der Physik, 426: 669-688.
Fourier, J. B. J., 1822, “Théorie Analytique de la Chaleur”, (1rst ed.), Paris:
F. Didot.
Franco, A., 2007, “An apparatus for the routine measurement of thermal
conductivity of materials for building application based on a transient
hot-wire method”, Applied Thermal Engineering, 27: 2495-2504.
G
Gaal, P. S., Thermitus, M. A., & Stroe, D. E., 2004, “Thermal conductivity
measurements using the flash method”, Journal of Thermal Analysis
and Calorimetry, 78: 185-189.
Gialou, A., 2004, Thermal Conductivity of Solid Materials, Phd. Thesis,
Aristotle University of Thessaloniki, Thessaloniki (in Greek).
Gillam, D. G., Rombén, L., Nissen, H.-E., & Lamm, O., 1955, “Accurate
Determination of Thermal Conductivities”, Acta Chemica
Scandinavica, 4: 641 - 656.
Gojny, F. H., Wichmann, M. H. G., Fiedler, B., Kinloch, I. A., Bauhofer, W.,
Windle, A. H., et al., 2006, “Evaluation and identification of electrical
and thermal conduction mechanisms in carbon nanotube/epoxy
composites”, Polymer, 47: 2036-2045.
Goldratt, E., & Greenfield, A. J., 1978, “New method for measuring the
thermal conductivity”, Review of Scientific Instruments, 49: 1531-
1536.
Gong, X., Liu, J., Baskaran, S., Voise, R. D., & Young, J. S., 2000,
“Surfactant-assisted processing of carbon nanotube/polymer
composites”, Chemistry of Materials, 12: 1049-1052.
Gorthala, R., Roux, J. A., Vaughan, J. G., & Donti, R. P., 1994, “Comparison
of processing parameters for pultruded graphite/epoxy and
fiberglass/epoxy: a heat transfer and curing model”, Journal of
Reinforced Plastics and Composites, 13: 288-300.
Gowayed, Y., & Hwang, J. C., 1995, “Thermal-conductivity of composite-
materials made from plan weaves and 3-D weaves”, Composites
Engineering, 5: 1177-1186.
Grassmann, P., & Straumann, W., 1960, “Ein instätionares verfahren zur
messung der wärmeleitfähigkeit von flüssigkeiten und gasen”,
204 References
International Journal of Heat and Mass Transfer, 1: 50-54.

Grove, W. R., 1845, “LXXI. On the application of voltaic ignition to lighting
mines”, Philosophical Magazine Series 3, 27: 442-446.
Grove, W. R., 1848, “On the peculiar cooling effects of hydrogen and its
compounds, in cases of voltaic ignition”, Journal of the Franklin
Institute, 46: 358.
Gu, M., Wang, J.-l., & Zhang, X., 2009, “Thermal-Conductivity
Measurements of Polymers by a Modified 3ω Technique”,
Gustafsson, S., Hamdani, A. J., & Karawacki, E., 1979a, “New method for
measuring thermal properties of transparent solids”, Journal of
Physics E: Scientific Instruments, 12: 387-390.
Gustafsson, S. E., 1991, “Transient plane source techniques for thermal
conductivity and thermal diffusivity measurements of solid
materials”, Review of Scientific Instruments, 62: 797-804.
Gustafsson, S. E., Karawacki, E., & Khan, M. N., 1979b, “Transient hot-
strip method for simultaneously measuring thermal conductivity and
thermal diffusivity of solids and fluids”, Journal of Physics D: Applied
Physics, 12: 1411-1421.
Gustavsson, M., Karawacki, E., & Gustafsson, S. E., 1994, “Thermal
conductivity, thermal diffusivity, and specific heat of thin samples
from transient measurements with hot disk sensors”, Review of
Scientific Instruments, 65: 3856-3859.
Gustavsson, M., Wang, H., Trejo, R. M., Lara-Curzio, E., Dinwiddie, R. B., &
Gustafsson, S. E., 2006, “On the use of the transient hot-strip method
for measuring the thermal conductivity of high-conducting thin bars”,
H
Haarman, J. W., 1971, “A contribution to the theory of the transient hot-
wire method”, Physica, 52: 605-619.
Hammerschmidt, U., 2003, “A New Pulse Hot Strip Sensor for Measuring
Thermal Conductivity and Thermal Diffusivity of Solids”,
Hammerschmidt, U., & Meier, V., 2006, “New transient hot-bridge sensor to
measure thermal conductivity, thermal diffusivity, and volumetric
specific heat”, International Journal of Thermophysics, 27: 840-865.
Hammerschmidt, U., & Sabuga, W., 2000, “Transient Hot Strip (THS)
method: uncertainty assessment”, International Journal of
Haran, E. N., & Wakeham, W. A., 1982, “A transient hot-wire cell for
thermal conductivity measurements over a wide temperature range”,
References 205
Journal of Physics E: Scientific Instruments, 15: 839-842.

Hasselman, D. P. H., Zdaniewski, W. A., Swearengen, J. C., & Beauchamp,
E. K., 1980, “Effect of alumina dispersions on the thermal
conductivity/diffusivity and thermal stress resistance of a borosilicate
glass”, Journal of Materials Science, 15: 518-520.
Haupin, W. E., 1960, “Hot Wire Method for Rapid Determination of Thermal
Conductivity”, Ceramic Bulletin, 39: 139 - 141.
Hayashi, K., Fukui, M., & Uei, I., 1971, “Determination of Thermal
Conductivity of Solid Materials by means of the Hot Wire Method”,
Memoirs of the Faculty of Industrial Arts, 20: 81-103.
Healy, J. J., de Groot, J. J., & Kestin, J., 1976, “The theory of the transient
hot-wire method for measuring thermal conductivity”, Physica B+C,
82: 392-408.
Hemberger, F., Göbel, A., & Ebert, H. P., 2010, “Determination of the
thermal diffusivity of electrically non-conductive solids in the
temperature range from 80 K to 300 K by laser-flash measurement”,
Hemminger, W., 1989, “Thermal conductivity of lead in the range -180 to
500Β°C”, International Journal of Thermophysics, 10: 765-777.
Hisano, K., & Placido, F., 1998, “Quasistatic thermal conduction
measurement by thermal radiation calorimetry”, High Temperatures
- High Pressures, 30: 297-305.
Horrocks, J. K., & McLaughlin, E., 1963, “Non-Steady-State Measurements
of Thermal Conductivities of Liquid Polyphenyls”, Proceedings of the
Royal Society of London Series a-Mathematical and Physical
Sciences, 273: 259-274.
Hu, Y. Z., Shen, J. F., Li, N., Ma, H. W., Shi, M., Yan, B., et al., 2010,
“Comparison of the thermal properties between composites reinforced
by raw and amino-functionalized carbon materials”, Composites
Science and Technology, 70: 2176-2182.
Huebner, K. H., Thornton, E. A., & Byrom, T. G., 1995, “The Finite Element
Method for Engineers”, (3rd ed.), New York: John Wiley & Sons, INC.
Hulstrom, L. C., Tye, R. P., & Smith, S. E., 1988, “Round robin testing of
thermal conductivity reference materials”, Proc. Thermal
Conductivity 19, Cookevelli, Tennesse.
I
International Organization for Standardization, 1996, Thermal Insulation -
Determination of steady-state thermal resistance and related
properties - Guarded hot plate apparatus, ISO 8302. Geneva,
Switzerland.
International Organization for Standardization, 1997, Calibration in
analytical chemistry and use of certified reference materials, ISO
Guide 32. Geneva, Switzerland.
206 References
International Organization for Standardization, 2000a, Reference materials

-- Contents of certificates and labels, ISO Guide 31. Geneva,
Switzerland.
International Organization for Standardization, 2000b, Uses of certified
reference materials, ISO Guide 33. Geneva, Switzerland.
International Organization for Standardization, 2005, Refractory Fine
ceramics (advanced ceramics, advanced technical ceramics) -
Determination of thermal diffusivity of monolithic ceramics by laser
flash method, ISO 18755. Geneva, Switzerland.
International Organization for Standardization, 2006, Reference materials -
- General and statistical principles for certification, ISO Guide 35.
Geneva, Switzerland.
International Organization for Standardization, International
2007,
vocabulary of metrology -- Basic and general concepts and associated
terms (VIM), ISO/IEC Guide 99. Geneva, Switzerland.
International Organization for Standardization, 2007, Refractory materials
-- Determination of thermal conductivity -- Part 2: Hot-wire methods
(parallel),, ISO 8894-2. Geneva, Switzerland.
International Organization for Standardization, 2008, Plastics -
Determination of thermal conductivity and thermal diffusivity - Part
2: Transient plane heat source (hot disc) method, ISO 22007-2.
Geneva, Switzerland.
International Organization for Standardization, General
2009,
requirements for the competence of reference material producers, ISO
Guide 34. Geneva, Switzerland.
International Organization for Standardization, 2010, Refractory materials
-- Determination of thermal conductivity -- Part 1: Hot-wire methods
(cross-array and resistance thermometer)” ISO 8894-1. Geneva,
Switzerland.
J
Jain, S. C., & Krishnan, K. S., 1954a, “The Distribution of Temperature
along a Thin Rod Electrically Heated in vacuo. I. Theoretical”,
Proceedings of the Royal Society of London. Series A. Mathematical
and Physical Sciences, 222: 167-180.
Jain, S. C., & Krishnan, K. S., 1954b, “The Distribution of Temperature
along a Thin Rod Electrically Heated In vacuo. II. Theoretical
(Continued)”, Proceedings of the Royal Society of London. Series A.
Mathematical and Physical Sciences, 225: 1-7.
Jain, S. C., & Krishnan, K. S., 1954c, “The Distribution of Temperature
along a Thin Rod Electrically Heated In vacuo. III. Experimental”,
Proceedings of the Royal Society of London. Series A. Mathematical
References 207

Jain, S. C., & Krishnan, K. S., 1954d, “The Distribution of Temperature
along a Thin Rod Electrically Heated In vacuo. IV. Many Useful
Empirical Formulae Verified”, Proceedings of the Royal Society of
London. Series A. Mathematical and Physical Sciences, 225: 19-32.
Jannot, Y., & Acem, Z., 2007, “A quadrupolar complete model of the hot
disc”, Measurement Science and Technology, 18: 1229-1234.
Jannot, Y., Acem, Z., & Kanmogne, A., 2006, “Transient hot plate method
with two temperature measurements for thermal characterization of
metals”, Measurement Science and Technology, 17: 69-74.
Jawad, S. H., Dix, M. J., & Wakeham, W. A., 1999, “Vapor-phase thermal
conductivity measurements of refrigerants”, International Journal of
Joint Committee for Guides in Metrology, 2008, Evaluation of Measurement
Data - Guide to the Expression of Uncertainty in Measurement
(GUM), JCGM-100:2008
K
Karagiannidis, L., 1998, Thermal Conductivity of New Alternative
Refrigerants, Phd. Thesis, Aristotle University of Thessaloniki,
Thessaloniki (in Greek).
Kalaprasad, G., Pradeep, P., Mathew, G., Pavithran, C., & Thomas, S.,
2000, “Thermal conductivity and thermal diffusivity analyses of low-
density polyethylene composites reinforced with sisal, glass and
intimately mixed sisal/glass fibres”, Composites Science and
Technology, 60: 2967-2977.
Kalogiannakis, G., Van Hemelrijck, D., & Van Assche, G., 2004,
“Measurements of thermal properties of carbon/epoxy and glass/epoxy
using modulated temperature differential scanning calorimetry”,
Journal of Composite Materials, 38: 163-175.
Kaul, P. B., Day, K. A., & Abramson, A. R., 2007, “Application of the three
omega method for the thermal conductivity measurement of
polyaniline”, Journal of Applied Physics, 101.
Kestin, J., Paul, R., Clifford, A. A., & Wakeham, W. A., 1980, “Absolute
determination of the thermal conductivity of the noble gases at room
temperature up to 35 MPa”, Physica A: Statistical Mechanics and its
Applications, 100: 349-369.
Kestin, J., & Wakeham, W. A., 1978, “A contribution to the theory of the
transient hot-wire technique for thermal conductivity
measurements”, Physica A: Statistical Mechanics and its
Applications, 92: 102-116.
Kubičár, L. u., & Boháč, V., 2002, “Intercomparison measurements by pulse
transient and stepwise transient methods on Perspex and stainless
steel A310”, High Temperatures - High Pressures, 34: 135-146.
208 References
Kubicar, L., Bohac, V., Vretanar, V., Barta, Ε., Neuer, G., & Brandt, R.,
2005a, “Thermophysical properties of heterogeneous structures
measured by pulse transient method”, International Journal of
Kubicar, L., Vretenar, V., Bohac, V., & Tiano, P., 2006, “Thermophysical
analysis of sandstone by pulse transient method”, International
Kubicar, L., Vretenar, V., & Hammerschmidt, U., 2005b, “Thermophysical
parameters of optical glass BK 7 measured by the pulse transient
method”, International Journal of Thermophysics, 26: 507-518.
L
Lee, S.-M., 2009, “Thermal conductivity measurement of fluids using the 3
omega method”, Review of Scientific Instruments, 80: 024901-024907.
Leslie, J., 1804, “An Experimental Inquiry into the Nature and Propagation
of Heat”, (1rst ed., pp. 552-553), London: Printed for J. Mawman.
Lewis, R. W., Nithiarasu, P., & Seetharamu, K. N., 2004, “Fundamentals of
the Finite Element Method for Heat and Fluid Flow”, (1st ed.),
Chichester: John Wiley & Sons, INC.
Lima E Silva, S. M. M., Duarte, M. A. V., & Guimaraes, G., 1998, “A
correlation function for thermal properties estimation applied to a
large thickness sample with a single surface sensor”, Review of
Lin, L. Y., Lee, J. H., Hong, C. E., Yoo, G. H., & Advani, S. G., 2006,
“Preparation and characterization of layered silicate/glass fiber/epoxy
hybrid nanocomposites via vacuum-assisted resin transfer molding
(VARTM)”, Composites Science and Technology, 66: 2116-2125.
Lockmuller, N., Redgrove, J., & Kubicar, L., 2003, “Measurement of thermal
conductivity with the needle probe”, High Temperatures - High
Pressures, 35-36: 127-138.
Log, T., & Gustafsson, S. E., 1995, “Transient plane source (TPS) technique
for measuring thermal transport properties of building materials”,
Fire and Materials, 19: 43-49.
Log, T., & Metallinou, M. M., 1992, “Thermal conductivity measurements
using a short transient hot-strip method”, Review of Scientific
Longo, G. A., 2008, “A steady-state apparatus to measure the thermal
conductivity of solids”, International Journal of Thermophysics, 29:
664-677.
M
Maeda, Y., Sagara, H., Tye, R. P., Masuda, M., Ohta, H., & Waseda, Y.,
1996, “A high-temperature system based on the laser flash method to
References 209
measure the thermal diffusivity of melts”, International Journal of

Magnus, G., 1861a, “The conduction of heat by gases”, Journal of the
Franklin Institute, 72: 130-132.
Magnus, G., 1861b, “XIV. On the propagation of heat in gases”,
Maitland, G. C., Mustafa, M., Ross, M., Trengove, R. D., Wakeham, W. A., &
Zalaf, M., 1986, “Transient hot-wire measurements of the thermal
conductivity of gases at elevated temperatures”, International
Malinaric, S., 2004, “Parameter estimation in dynamic plane source
method”, Measurement Science and Technology, 15: 807-813.
Malinaric, S., 2007, “Uncertainty analysis of thermophysical property
measurements of solids using dynamic methods”, International
Malinarič, S., & Dieška, P., 2009, “Improvements in the dynamic plane
source method”, International Journal of Thermophysics, 30: 608-618.
Matsumoto, T., & Ono, A., 1991, “Radiative heat exchange method for
thermal conductivity measurement applying a perpendicular heat
flow to a thin-plate sample: Principle and apparatus”, International
Matsumoto, T., & Ono, A., 1992, “Thermal conductivity measurements of
Pyroceram standard reference material based on radiative heat
exchange (II)”, Proc. The Thirteenth Japan Symposium on
Thermophysical Properties, Tokyo, 129-132.
Mani, N., 1971, Precise Determination of the Thermal Conductivitites of
Fluids Using Absolute Transient Hot-Wire Technique, Ph.D. Thesis,
University of Calgary, Alberta, Canada
Martin, T. O. K., Salmon, D. R., & Taylor, D. R., 2009, “Investigation by
NPL of Transient versus Steady State Thermal Conductivity
Measurement Methods”, Proc. 30th International Thermal
Conductivity Conference - 18th International Thermal Expansion
Symposium, Pittsburgh, Pennsylvania, U.S.A.
Maxwell, J. C., 1860a, “II. Illustrations of the dynamical theory of gases”,
Maxwell, J. C., 1860b, “V. Illustrations of the dynamical theory of gases. -
Part I. On the motions and collisions of perfectly elastic spheres”,
McIvor, S. D., Darby, M. I., Wostenholm, G. H., Yates, B., Banfield, L., King,
R., et al., 1990, “Thermal conductivity measurements of some glass
fibre- and carbon fibre-reinforced plastics”, Journal of Materials
Science, 25: 3127-3132.
McLaughlin, E., & Pittman, J. F. T., 1971a, “Determination of Thermal
Conductivity of Toluene - Proposed Data Standard - from 180 to 400K
210 References
under Saturation Pressure by Transient Hot-Wire Method. 1. The

Theory of the Technique”, Philosophical Transactions of the Royal
Society of London Series a-Mathematical and Physical Sciences, 270:
557-578.
McLaughlin, E., & Pittman, J. F. T., 1971b, “Determination of Thermal
Conductivity of Toluene - Proposed Data Standard - from 180 to 400K
under Saturation Pressure by Transient Hot-Wire Method. 2. New
Measurements and a Discussion of Existing Data.”, Philosophical
Transactions of the Royal Society of London Series a-Mathematical
Menashe, J., & Wakeham, W. A., 1981, “Absolute Measurements of the
Thermal Conductivity of Liquids at Pressures up to 500 MPa”,
Berichte der Bunsengesellschaft/Physical Chemistry Chemical
Physics, 85: 340-347.
Metaxa, I. N., 2006, Thermal Conductivity of Nanofluids, Phd. Thesis,
Aristotle University of Thessaloniki, Thessaloniki (in Greek).
Min, S., Blumm, J., & Lindemann, A., 2007, “A new laser flash system for
measurement of the thermophysical properties”, Thermochimica
Acta, 455: 46-49.
Milano, G., Scarpa, F., Righini, F., & Bussolino, G. C., 2001, “Ten years of
parameter estimation applied to dynamic thermophysical property
measurements”, International Journal of Thermophysics, 22: 1227-
1240.
Miller, M. S., & Kotlar, A. J., 1993, “Technique for measuring thermal
diffusivity/conductivity of small thermal-insulator specimens”,
Review of Scientific Instruments, 64: 2954-2960.
Mittenbühler, A., 1964, “Apparatur zur Messung der Wärmeleitfähigkeit
von feuerfesten Steinen, Körnungen und Pulvern”, Berichte der
Deutschen Keramischen Gesellschaft 41: 15-20.
Moisala, A., Li, Q., Kinloch, I. A., & Windle, A. H., 2006, “Thermal and
electrical conductivity of single- and multi-walled carbon nanotube-
epoxy composites”, Composites Science and Technology, 66: 1285-
1288.
N
Nagasaka, Y., & Nagashima, A., 1981, “Absolute measurement of the
thermal conductivity of electrically conducting liquids by the
transient hot-wire method”, Journal of Physics E: Scientific
National Institute of Standards and Technology, 2000, Definitions of terms
and modes used at NIST for value-assignment of reference materials
for chemical measurements, NIST SP 260-136. Washington, U.S.A.
Nieto de Castro, C. A., Calado, J. C. G., Wakeham, W. A., & Dix, M., 1976,
References 211
“An apparatus to measure the thermal conductivity of liquids”,

Journal of Physics E: Scientific Instruments, 9: 1073-1080.
Ning, Q. G., & Chou, T. W., 1998, “A general analytical model for predicting
the transverse effective thermal conductivities of woven fabric
composites”, Composites Part A: Applied Science and Manufacturing,
29: 315-322.
O
Ohta, H., Tsukida, T., Tomota, Y., Shibata, H., Ogura, G., & Waseda, Y.,
1991, “Measurement of thermal diffusivity of pyroceram 9606 by the
laser flash method”, Journal of the Ceramic Society of Japan.
International ed., 99: 471-476.
Oram, R. J., & Wolff, E. G., 1998, “Simple Guarded Hot Plate Technique for
Making Thermal Conductivity Measurements of Composite Materials
at Cryogenic Temperatures”, Advances in Cryogenic Engineering
Materials, 44: 349-358.
P
Pantano, A., & Averill, R. C., 2002, “An improved thermal lamination model
for analysis of heat transfer in composite structures”, Journal of
Composite Materials, 36: 701-719.
Paradise, M., & Goswami, T., 2007, “Carbon nanotubes - Production and
industrial applications”, Materials and Design, 28: 1477-1489.
Parker, W. J., Jenkins, R. J., Butler, C. P., & Abbott, G. L., 1961, “Flash
method of determining thermal diffusivity, heat capacity, and
thermal conductivity”, Journal of Applied Physics, 32: 1679-1684.
Peralta-Martinez, M. V., 2000, Thermal Conductivity of Molten Metals, Phd.
Thesis, Imperial College, London
Peralta-Martinez, M. V., Assael, M. J., Dix, M. J., Karagiannidis, L., &
Wakeham, W. A., 2006a, “A novel instrument for the measurement of
the thermal conductivity of molten metals. Part I: Instrument's
description”, International Journal of Thermophysics, 27: 353-375.
Peralta-Martinez, M. V., Assael, M. J., Dix, M. J., Karagiannidis, L., &
Wakeham, W. A., 2006b, “A novel instrument for the measurement of
the thermal conductivity of molten metals. Part II: Measurements”,
Perovic, N. L., & Maglic, K. D., 1991, “Thermal diffusivity of pyroceram”,
High Temperatures - High Pressures, 23: 437-443.
Pfriem, H., 1938, “Nichtstationäres Verfahren zur Messung der Wärme -
und Temperatur - leitfähigkeit von Flüssigkeiten”, Zeitschrift Des
Vereines Deutscher Ingenieure, 82: 71-72.
212 References
Pillai, C. G. S., & George, A. M., 1991, “An improved comparative thermal
conductivity apparatus for measurements at high temperatures”,
Powell, R. W., Liley, P. E., & Ho, C. Y., 1966, “Thermal conductivity of
selected materials ”, National Bureau of Standards, NSRDS-NBS 8,
Washington: U.S. Dept. of Commerce, National Bureau of Standards.
Powell, R. W., & Schofield, F. H., 1939, “The thermal and electrical
conductivities of carbon and graphite to high temperatures”,
Proceedings of the Physical Society, 51: 153-172.
Preston-Thomas, H., 1990, “The International Temperature Scale of 1990
(ITS-90)”, Metrologia, 27: 3-10.
Priestly, J., 1781, “Experiments and Observations relating to Various
branches of natural Philosophy; a Continuation of the Observations
on air ”, (1rst ed., Vol. 2, pp. 375-378), London: Pearson and Rollason.
R
Ramani, K., & Vaidyanathan, A., 1995, “Finite element analysis of effective
thermal conductivity of filled polymeric composites”, Journal of
Composite Materials, 29: 1725-1740.
Ramires, M. L. V., Nieto De Castro, C. A., Perkins, R. A., Nagasaka, Y.,
Nagashima, A., Assael, M. J., et al., 2000, “Reference data for the
thermal conductivity of saturated liquid toluene over a wide range of
temperatures”, Journal of Physical and Chemical Reference Data, 29:
133-139.
Reddy, J. N., & Gartling, D. K., 2001, “The Finite Element Method in Heat
Transfer and Fluid Dynamics”, (2nd ed.), Boca Raton, Florida: CRC
Press.
Rides, M., Morikawa, J., Halldahl, L., Hay, B., Lobo, H., Dawson, A., et al.,
2009, “Intercomparison of thermal conductivity and thermal
diffusivity methods for plastics”, Polymer Testing, 28: 480-489.
Robinson, H. E., & Flynn, D. R., National Bureau of Standards, Washigton
(Private correspondence of Powell et al., 1966).
Roder, H. M., 1981, “A Transient Hot Wire Thermal Conductivity Apparatus
for Fluids”, Journal of Research of the National Bureau of Standards
(United States), 86: 457-493.
Rudtsch, S., & Hammerschmidt, U., 2004, “Intercomparison of
measurements of the thermophysical properties of polymethyl
References 213
methacrylate”, International Journal of Thermophysics, 25: 1475-

1482.
Rudtsch, S., Stosch, R., & Hammerschmidt, U., 2002, “Intercomparison of
measurements of the thermophysical properties of Crown Borosilicate
Glass BK7”, Proc. 16th European Conference on Thermophysical
Properties, London.
S
Sabuga, W., & Hammerschmidt, U., 1995, “A new method for the evaluation
of thermal conductivity and thermal diffusivity from transient hot
strip measurements”, International Journal of Thermophysics, 16:
557-565.
Salmon, D. R., 2001, “Thermal conductivity of insulations using guarded hot
plates including recent developments and sources of reference
materials”, Measurement Science and Technology, 12.
Salmon, D. R., 2009, “A critical analysis of European standards for thermal
measurements at high temperatures: I. History and technical
background”, Measurement Science and Technology, 20: 015101.
Salmon, D. R., Baxendale, S., Groboth, G., Hay, B., Hammerschmidt, U.,
Brandt, R., et al., 2003, The certification of thermal conductivity and
diffusivity properties of Pyroceram 9606 as a reference material up to
1000 oC - Final Technical Report HTCRM. (Report No. SMT4-CT98-
2211).
Salmon, D. R., Roebben, G., Lamberty, A., & Brandt, R., 2007, Certification
of thermal conductivity and thermal diffusivity up to 1025 K of a
glass-ceramic reference material BCR-724. (Report No. EUR 21764
EN). Geel, Belgium: European Commission, Institute for Reference
Materials and Measurements.
Salmon, D. R., & Tye, R. P., 2000, “An intercomparison of heat flow meter
apparatus within the United Kingdom and Eire”, High Temperatures
- High Pressures, 32: 19-28.
Salmon, D. R., Tye, R. P., & Lockmuller, N., 2009, “A critical analysis of
European standards for thermal measurements at high
temperatures: II. Recommendations for inclusion in a new standard”,
Measurement Science and Technology, 20: 015102.
Schleiermacher, A., 1888, “Ueber die Wärmeleitung der Gase”, Annalen der
Physik und Chemie, 270: 623-646.
Sheffield, G. S., & Schorr, J. R., 1991, “Comparison of Thermal Diffusivity
and Thermal Conductivity Methods”, American Ceramic Society
Bulletin, 70: 102-106.
Song, Y. S., & Youn, J. R., 2005, “Influence of dispersion states of carbon
nanotubes on physical properties of epoxy nanocomposites”, Carbon,
214 References
43: 1378-1385.
Stâlhane, B., & Pyk, S., 1931, “NY metod för bestämning av
värmelednings-koefficienter”, Teknisk Tidskrift, 61: 389 - 398.
Stefan, J., 1872, “Sitzungsberichte der kaiserlichen Akademie der
Wissenschaften, Mathematische-Naturwissenschaftliche Classe,
Abteilung II”, (Vol. 65, pp. 45-69), Wien: Aus der K.K. Hof und
Staatdruckerei.
Stosch, R., & Hammerschmidt, U., 2003, “A new statistical approach to
thermal conductivity evaluation of transient hot-wire and transient
hot-strip signals”, High Temperatures - High Pressures, 35-36: 55-66.
Suliyanti, M. M., Baba, T., & Ono, A., 1995, “Thermal Diffusivity
Measurement of Pyroceram 9606 by the laser flash method”, Bulletin
of NRLM, 44: 301-309.
Sweeting, R. D., & Liu, X. L., 2004, “Measurement of thermal conductivity
for fibre-reinforced composites”, Composites Part A: Applied Science
and Manufacturing, 35: 933-938.
T
Thompson, B., 1786, “New Experiments upon Heat. By Colonel Sir
Benjamin Thompson, Knt. F. R. S. In a Letter to Sir Joseph Banks,
Bart. P. R. S”, Philosophical Transactions of the Royal Society of
London, 76: 273-304.
Thompson, B., 1799, “Essays, Political, Economical, and Philosophical”, (1rst
Amer. ed., Vol. 2, pp. 457), Boston: David West.
Thompson, W., 1889, “Popular Lectures and Addresses”, (Vol. 1, pp. 73),
London: London Macmillan.
Thostenson, E. T., & Chou, T. W., 2006, “Processing-structure-multi-
functional property relationship in carbon nanotube/epoxy
composites”, Carbon, 44: 3022-3029.
Tleoubaev, A., & Brzezinski, A., 2005, Combined Guarded-Hot-Plate and
Heat Flow Meter Method for absolute thermal conductivity tests
excluding thermal contact resistance. In H. Wang, W. D. Porter & G.
Worley (Eds.), Thermal Conductivity 27: Thermal Expansion 15 (Vol.
27, pp. 502-510). Lancaster: Destech Publications, Inc.
Tsederberg, N. V., 1965, “Thermal Conductivity of Gases and Liquids”, (pp.
2-12), Cambridge, Massachusetts: M.I.T. Press.
Turnbull, A. G., 1961, “The Thermal Conductivity of Molten Salts”,
Australian Journal of Applied Science, 12: 325 - 329.
Tye, R. P., & Salmon, D. R., 2001, “Thermal conductivity Certified Reference
Materials: Pyrex 7740 and Polymethylmethacrylate”, Proc. 26th
International Thermal Conductivity Conference - 14th International
Thermal Expansion Symposium, Cambridge, MA.
Tye, R. P., & Salmon, D. R., 2002, “NPL Initiatives in the Development of
Thermal Properties Reference Materials”, International Journal of
References 215
Tyndall, J., 1863, “Heat Considered as a Mode of Motion”, (1rst ed., pp. 26,
238-242), London: Longman, Green, Longman, Roberts & Green.
V
van der Held, E. F. M., & van Drunen, F. G., 1949, “A method of measuring
the thermal conductivity of liquids”, Physica, 15: 865-881.
Vozar, L., & Hohenauer, W., 2003, “Flash method of measuring the thermal
diffusivity a review”, High Temperatures - High Pressures, 35-36:
253-264.
W
Wagner, P., 1969, “On the technique of measuring the high-temperature
thermal conductivity of a rod”, Journal of Physics D: Applied Physics,
2: 293-300.
Wakeham, W. A., 2009, “The Transient Hot-Wire Technique: Not Quite the
Answer to a Maiden's Prayer”, Proc. 30th International Thermal
Conductivity Conference - 18th International Thermal Expansion
Symposium, Pittsburgh, Pennsylvania, U.S.A.
Wakeham, W. A., & Assael, M. J., 2000, Thermal conductivity
measurement. In J. G. Webster (Ed.), The Measurement,
Instrumentation and Sensors Handbook: CRC Press.
Wakeham, W. A., Nagashima, A., & Sengers, J. V. (Eds.), 1991,
“Measurement of the Transport Properties of Fluids”, Experimental
Thermodynamics, (1rst ed., Vol. III). Oxford: Blackwell Scientific
Publications.
Wakeham, W. A., & Zalaf, M., 1985, “Thermal Conductivity of Electrically-
Conducting Liquids at high Pressures”, Physica B: Physics of
Condensed Matter & C: Atomic, Molecular and Plasma Physics,
Optics, 139-140: 105-108.
Wang, S., & Qiu, J., 2010, “Enhancing thermal conductivity of glass
fiber/polymer composites through carbon nanotubes incorporation”,
Composites Part B: Engineering, 41: 533-536.
Weber, S., 1917a, “Experimentelle Untersuchungen über die
Wärmeleitfähigkeit der Gase. I”, Annalen der Physik, 359: 325-356.
Weber, S., 1917b, “Untersuchungen über die Wärmeleitfähigkeit der Gase.
II”, Annalen der Physik, 359: 437-462.
216 References
Wichmann, M. H. G., Sumfleth, J., Gojny, F. H., Quaresimin, M., Fiedler, B.,
& Schulte, K., 2006, “Glass-fibre-reinforced composites with enhanced
mechanical and electrical properties - Benefits and limitations of a
nanoparticle modified matrix”, Engineering Fracture Mechanics, 73:
2346-2359.
Williams, I., & Shawyer, R. E., 1991, Certification report for a pyrex glass
reference material for thermal conductivity between -75 oC and 195
oC - CRM 039. (Report No. EUR 13358 EN). Brussels: European
Commission, Institute for Reference Materials and Measurements.

Wulf, R., Barth, G., & Gross, U., 2007, “Intercomparison of insulation
thermal conductivities measured by various methods”, International
X
Xie, H., Gu, H., Fujii, M., & Zhang, X., 2006, “Short hot wire technique for
measuring thermal conductivity and thermal diffusivity of various
materials”, Measurement Science and Technology, 17: 208-214.
Y
Yang, G., Migone, A. D., & Johnson, K. W., 1992, “Heat capacity and
thermal diffusivity of a glass sample”, Physical Review B, 45: 157-
160.
Yang, S. Y., Ma, C. C. M., Teng, C. C., Huang, Y. W., Liao, S. H., Huang, Y.
L., et al., 2010, “Effect of functionalized carbon nanotubes on the
thermal conductivity of epoxy composites”, Carbon, 48: 592-603.
Yen, C. L., Tseng, H. C., Wang, Y. Z., & Hsieh, K. H., 1991, “Thermal
conductivity of glass fiber reinforced polypropylene under high
pressure”, Journal of Applied Polymer Science, 42: 1179-1184.
Z
Zawilski, B. M., Littleton Iv, R. T., & Tritt, T. M., 2001, “Description of the
parallel thermal conductance technique for the measurement of the
thermal conductivity of small diameter samples”, Review of Scientific
Zawilski, B. M., & Tritt, T. M., 2001, “Dynamic measurement access, a new
technique for fast thermal conductivity measurement”, Review of
Zhou, T., Wang, X., Liu, X., & Xiong, D., 2010, “Improved thermal
conductivity of epoxy composites using a hybrid multi-walled carbon
nanotube/micro-SiC filler”, Carbon, 48: 1171-1176.
Zhou, Y. X., Wu, P. X., Cheng, Z. Y., Ingram, J., & Jeelani, S., 2008,
“Improvement in electrical, thermal and mechanical properties of
References 217
epoxy by filling carbon nanotube”, Express Polymer Letters, 2: 40-48.

Zhu, J., Kim, J., Peng, H., Margrave, J. L., Khabashesku, V. N., & Barrera,
E. V., 2003, “Improving the dispersion and integration of single-
walled carbon nanotubes in epoxy composites through
functionalization”, Nano Letters, 3: 1107-1113.
List of Publications
In the following list of publications, the whole research activity

performed during the Phd is included.
PUBLICATIONS ON INTERNATIONAL JOURNALS
1. Assael, M.J., Kalyva, E.A., Antoniadis, K.D., Banish, R.M., Egry, I., Wu,
J., Kaschnitz, E. and Wakeham, W.A., “Reference Data for the Density
and Viscosity of Liquid Copper and Liquid Tin”, J. Phys. Chem. Data.
39, 033105 (2010)
2. Assael, M.J., Antoniadis, K.D., and Wakeham, W.A., “Historical
Evolution of the Transient Hot-Wire Technique”, Int. J. Thermophysics.
31: 1051-1072 (2010)
3. Stefanidou, M., Assael, M.J., Antoniadis, K.D., and Matziaroglou, G.,
“Thermal Conductivity of Building Materials Employed in the
Preservation of Traditional Structures”, Int. J. Thermophysics. 31: 844 –
851 (2010)
4. Assael M.J., Kakosimos K.E., Antoniadis K.D., Assael J-A.M., “Applying
Thermal Comfort Indices to Investigate Aspects of the Climate in
Greece”, International Review of Chemical Engineering (I.RE.CH.E.) 2:
204 – 209 (2010)
5. Assael M.J., Kalyva A., Kakosimos K.E., Antoniadis K., “Correlation and
Prediction of Dense Fluid Transport Coefficients. VIII. Mixtures of Alkyl
Benzenes with Other Hydrocarbons”, Int. J. Thermophys. 30: 1733-1747
(2009)
6. Pantzali, M.N., Kanaris, A.G., Antoniadis, K.D., Mouza, A.A., and
Paras, S.V. “Effect of Nanofluids in the Performance of a Miniature
Plate Heat Exchanger with Modulated Surface”, International Journal
of Heat and Fluid Flow 30: 691 – 699 (2009)
7. Assael, M.J., Antoniadis, K.D., Metaxa, I.N, and Tzetzis, D.,
220 List of Publications
“Measurements on the Enhancement of the Thermal Conductivity of an

Epoxy Resin when Reinforced with Glass Fiber and Carbon Multiwalled
Nanotubes”, J. Chem. Eng. Data 54: 2365 – 2370 (2009)
8. Assael, M.J., Antoniadis, K.D., and Tzetzis, D., “The Use of the
Transient Hot-Wire Technique for Measurement of the Thermal
Conductivity of an Epoxy-Resin Reinforced with Glass Fibres and/or
Carbon Multi-Walled Nanotubes”, Composites Science and Technology
68: 3178 – 3183 (2008)
9. Assael, M.J., Antoniadis, K.D., and Wu, J., “New Measurements of the
Thermal Conductivity of PMMA, BK7, and Pyrex 7740 up to 450 K”, Int.
J. Thermophysics. 29: 1257 – 1266 (2008)
10. Assael, M.J., Antoniadis, K.D., Kakosimos, K.E., and Metaxa, I.N, “An
Improved Application of the Transient Hot-Wire for the Absolute
Accurate Measurement of the Thermal Conductivity of Pyroceram 9606
up to 420 K”, Int. J. Thermophysics. 29: 445-456 (2008)
PUBLICATIONS ON CONFERENCES PROCEEDINGS
1. Assael M.J., Kalyva A., Kakosimos K.E., Antoniadis K.D., “On the
Prediction of the Viscosity and Thermal Conductivity of Liquid
Hydrocarbon Mixtures on Considerations Based Upon the Hard Sphere
Theory”, Presented 9th IATP Meeting, Boulder, Colorado, June (2009)
2. Assael, M.J., and Antoniadis K.D., “Transient Hot-Wire Technique: From
Gases to Liquids, to Melts and Solids, and still going Strong”, Proc. of
30th ITCC and 18th ITES, Pittsburgh, PA, U.S.A. (2009)
3. Stefanidou, M., Assael, M.J., Antoniadis, K.D., and Matziaroglou, G.,
“Thermal Conductivity of Building Materials Employed in the
Preservation of Traditional Structures”, Proc. of 17th Symposium on
Thermophysical Properties, Boulder, CO, U.S.A. (2009)
4. Assael, M.J., Antoniadis, K.D., Metaxa, I.N., and Tzertzis, D., “
Measurements of the Enhancement of the Thermal Conductivity of an
Epoxy Resin Reinforced with Glass Fibres and/or Carbon Multi-Walled
Nanotubes”, Proc. of 18th ECTP, Pau, France (2008)
5. Pantzali, M.N., Kanaris, A.G., Antoniadis, K.D., Mouza, A.A., and Paras,
S.V. “Effect of Nanofluids in the Performance of a Miniature Plate Heat
Exchanger with Modulated Surface” , Proc. of CHISA 2008, 18th
International Congress of Chemical and Process Engineering, Prague,
Czechoslovakia (2008)
6. Assael, M.J., Antoniadis, K.D., and Metaxa I.N. “State of the art of
Thermophysical-Properties Reference Materials ”, Proc. of 2nd National
Conference of Metrology, Thessaloniki, Greece (2007)
List of Publications 221
7. Assael, M.J., Antoniadis, K.D., Kakosimos, K.E., and Metaxa, I.N, “An
Improved Sensor for the Measurement of the Thermal Conductivity of
Ceramic Solids”, Proc. of 8th Asian Thermophysical Properties
Conference, Fukuoka, Japan (2007)
8. Assael, M.J. and Antoniadis K.D., “On the Application of the Transient
Hot-Wire Technique to the Measurement of Thermal Conductivity of
Solids”, Presented at 7th Meeting of the International Association for
Transport Properties, Istanbul, Turkey (2007)
9. Assael, M.J., Metaxa I.N., and Antoniadis, K.D., “Accurate
Measurements of the Viscosity and Thermal Conductivity over a wide
range of Conditions”, Proc. of 1st National Conference of Metrology,
Athens, Greece (2005)
10. Antoniadis, K.D., Dandikas, D.K., Mouza, A.A., “Building novel
equipment modules for commercial process simulators”, Proc. of 5th
Panhellen. Scient. Chem. Eng. Conference, Thessaloniki, Greece (2005)
11. Assael, M.J., Antoniadis, K.D., Botsios, S.D., Chamizidis, E., Gialou,
K.E., Gkontosidou, E., Kakosimos, K.E., “Optimizing the perforation of
Greek bricks”, Proc. of 5th Panhellen. Scient. Chem. Eng. Conference,
Thessaloniki, Greece (2005)
12. Mouza, A.A., Antoniadis, K.D., and Paras, S.V., “Building novel
equipment modules for a commercial process simulator: the compact
plate heat exchanger case”, Proc. of CHISA 2004, 16th International
Congress of Chemical and Process Engineering, Prague (2004)

Gri 2011 7244 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Gri 2011 7244 PDF

Uploaded by

Copyright:

Available Formats

ARISTOTLE UNIVERSITY OF THESSALONIKI

SCHOOL OF ENGINEERING – CHEMICAL ENGINEERING DEPARTMENT

DIPL. CHEMICAL ENGINEER

Thesis submitted for the degree of

Υποβλήθηκε στο Τμήμα Χημικών Μηχανικών, Πολυτεχνική Σχολή,

Ημερομηνία Προφορικής Εξέτασης: 13/04/2011

Καθηγητής Μ.Ι. Ασσαέλ, Επιβλέπων

Καθηγητής Γ. Σακελλαρόπουλος, Εξεταστής

Μέτρηση της Θερμικής Αγωγιμότητας Σύνθετων Στερεών

«Η έγκριση της παρούσης Διδακτορικής Διατριβής από το Τμήμα Χημικών

evaluation of the methods is made by presenting and discussing three

chapter is dedicated to the analysis of the sources of uncertainty introduced

Σκοπός της παρούσας εργασίας ήταν η πραγματοποίηση μετρήσεων

αγωγιμότητας στερεών. Παρουσιάζεται ο νόμος του Fourier για τη μεταφορά

χρησιμοποιείται στις μετρήσεις. Γίνεται ιστορική ανασκόπηση των

αρχικά με ύφασμα από ίνες γυαλιού, έπειτα με πολυστρωματικούς

Acknowledgements _________________________________________________ ix

A1. The Need for Material Properties ______________________________________ 2

A2. Reference Materials __________________________________________________ 5

A3. Composite Materials __________________________________________________ 8

A4. Scope ______________________________________________________________ 10

B Methods for Measuring the Thermal Conductivity ___________________ 11

B1. Other methods ______________________________________________________ 12

B1.2.3. Transient Hot Strip Method ___________________________________________ 25

B2. Transient Hot Wire Technique ________________________________________ 38

B3. Summary ___________________________________________________________ 55

C Theoretical ______________________________________________________ 57

C1. Ideal Model _________________________________________________________ 58

C2. Finite–Element Analysis _____________________________________________ 65

C3. Summary ___________________________________________________________ 95

D Experimental Procedure __________________________________________ 97

D1. Experimental Configuration __________________________________________ 98

D1.2. Temperature Controller _________________________________________________ 106

D2. Electronic Components _____________________________________________ 107

D3. Working Equations _________________________________________________ 116

D4. Calculation of the Thermal Conductivity Value ________________________ 127

D5. Validation of the Technique _________________________________________ 132

D6. Uncertainty Analysis _______________________________________________ 136

D7. Summary __________________________________________________________ 139

E Thermal Conductivity Measurements ______________________________ 141

E1. Solid Materials _____________________________________________________ 142

E2. Composite Materials ________________________________________________ 179

E2.1.1. Samples Preparation ________________________________________________180

E3. Summary __________________________________________________________ 193

F Innovation – Future Work ________________________________________ 195

F1. Innovations ________________________________________________________ 196

F2. Future Work _______________________________________________________ 197

References ________________________________________________________ 199

List of Publications ________________________________________________ 219

The chapter constitutes a brief introduction on the thermal

A1. The Need for Material Properties

on the contrary there are numerous and varied sources.

*MSDS: Material Safety Data Sheet

It should be noted that the thermal properties of materials are among

A1.1. The Importance of Accurate Measurements

A2. Reference Materials

a) Reference Material (RM)

b) Certified Reference Material (CRM)

c) Standard Reference Material (SRM)

d) NIST Traceable Reference Material (NTRM)

The ISO Committee on Reference Materials (ISO/REMCO) has

The objective of this committee is to carry out and encourage a broad

Table A.1 ISO reference material guides