You are on page 1of 7

2010, 12th International Conference on Optimization of Electrical and Electronic Equipment, OPTIM 2010

High-Voltage Ceramic Station Post Insulators


- Least-Effort Determination of Bending Strength
under Short-Circuit -
Norbert Stein*, Amir M. Miri**

* FGH Forschungsgemeinschaft für elektrische Anlagen und Stromwirtschaft


Hallenweg 40, D-68219 Mannheim, Germany. email: norbert.stein@fgh-ma.de

** Universität Karlsruhe, IEH, D-76128 Karlsruhe, Germany


email: miri@ieh.etec.uni-karlsruhe.de

Abstract - Stress concentrations due to the rather abrupt computational effort and in the interest of the simplicity
variation of the cross-section at the transitions from core to of the calculation method, the structural analysis of a
shed have an important influence on the elastic material complex large-scale system as an open-air substation
stresses and the allowable loading of ceramic insulators. requires restrictions in the detailedness of modelling of
For the user the effect is of immediate relevance only for the
the structural components sufficient to achieve so-called
non-quasi-static dynamic bending case, which is the
prevailing short-circuit loading of the post insulators of a global identity of the dynamic behaviour, i.e. displace -
substation. It can be described in terms of local stress- ments, of the complete mechanical structure – the system
concentration factors. of conductors and their supporting elements and sub-
structures.
The paper reports on a special Finite-Element calculation
method allowing to determine the required sets of In the case of a post insulator, that, on short-circuit, is
concentration factors of insulator types with least subjected to bending, this results in its reduction to a
discretisation- and computing effort possible. It thus cylindrical or conical, even, that is shed-less, beam of
becomes feasible to determine the static local stress maxima identical lateral oscillatory behaviour. This procedure
between sheds along the insulator and to derive from their
fully satisfies the immediate purposes.
maximum value together with the rated value of Minimum
Breaking Load MBL the value of allowable material stress
inherently assigned to the insulator by the manufacturer. It Problems may arise in the subsequent consideration of
may be named Minimum Breaking Stress MBS. Since [1] strength, i.e. the withstand to the calculated short-circuit
rates dynamic material strength as not lower than static, a loadings, where the ceramic post insulators represent a
general dynamic strength analysis can be performed. particularly sensitive component. Ceramic material,
Exemplary results from static and dynamic studies on a unlike that of the metallic structural components, does not
standard type 110 kV post insulator are presented, that are possess their ductile phase strength reserve, so that
supported by static test results. overloading will immediately result in failure.
The studies show that the dynamic cros-sectional stress
distribution differs from the static case and results in higher The general insulator strength consideration requires
stress-concentration factors than the respective static. Yet, the knowledge of
test experience [2, 3, 4] have shown that the slow dynamic - the absolute maximum value of material constraints
response of the insulator may compensate fast loading of the insulator over the local and time domains and
effects as regards its strength. Yet, this particular influence - the respective maximum allowable stress value.
requires more and closer study to come to a simplified
determination of the Equivalent Static Loads (ESL) for the
dynamic short-circuit case. A. Maximum Material Stress

Tests and calculations [4, 5, 6], and in particular this


I. INTRODUCTION
paper, show that the sheds of insulators, due the rather
abrupt variation of the cross section at the transitions
The state of the art of assessing the mechanical effects
between core and shed, result in massive stress
of short-circuit currents in high-voltage substations is col-
concentrations of a size that, on cantilever loading, has an
lected in [3] prepared by the ESCC-TF of CIGRE 23-11.
important impact on the local cross-sectional stress
For reasons of practical feasibility in preparatory and
maxima. The same effect, while existing to a lesser

978-1-4244-7020-4/10/$26.00 '2010 IEEE 118


degree for other types of loading, is of immediate good example for it, while the so-called pinch force of bundle
practical importance only in the considered case of conductors is certainly not.
dynamic bending.

The maximum material constraints of the genuine F. Dynamic strength


insulator with sheds are therefore relevantly higher than
the respective determined for the shed-less insulator The designer needs a practicable, i.e. relatively
model. uncomplicated procedure for judging the dynamic
strength of a post insulator. This could be achieved by
furnishing a set of k-factors for the respective insulator
B. Stress-Concentration Factor type. Then, it becomes possible to determine the relevant
genuine stress values on the insulator from the respective
The actual maximum bending constraint at a particular values of the global approach and, on the other hand, the
coordinate transition area on the genuine insulator can be required value of maximum allowable stress MBS
expressed as the respective value on the same-diameter- inherently given by the rated MBL. The subsequent
core model insulator multiplied by the local stress- analysis must consider the domains of coordinates plus
concentration factor. The factor relates to the insulator time.
type, the considered transition area on the insulator and
the type of loading - bending in our case. If cross- The following reports on the application, testing and
sectional stress distribution varies between static and a on exemplary results of a special Finite Element
dynamic case, the respective factor also relates to the technique, that represents the least-effort method - in
loading time function. terms of least preparatory and computational effort
achievable – to determine the detailed stresses on a
The factor must also relate to the strength hypothesis to genuine insulator with sheds for any form and material of
be applied for the considered loading case and material. insulator and any loading time function.

C. Strength Hypothesis
II. CALCULATION METHOD
For the case of brittle material, as ceramics, were The insulator subjected to bending represents an axi-
failing occurs without a marked plastic strain, the strength symmetric structure under non-axisymmetric loading, i.e.
hypothesis of maximum principal stress shall be applied. not a completely axisymmetric problem that could be
Stress-concentration factors k must therefore relate onto treated as plane. In principle the actual problem can be
σ1, hence k1 . solved applying 3-dimensional FEM elements.

Yet, a special FE type - the so-called Harmonic


D. Maximum allowable stress
Element – gives the possibility to treat our non-
axisymmetric problem in a quasi-axisymmetric manner
As the sole mechanical rated value as regards bending
[8-15], thus allowing to make use of the really enormous
loadability of the post insulator the manufacturer assigns
effort reducing advantages of the plane FE model. Only,
the Minimum Breaking Load MBL , the permissible static
the application of that method brings the problem
bending force at the insulator head and equivalent to
solution into an affordable range for normal computers.
permissible bending moments. Although the MBL
inherently contains a static value of allowable material
The harmonic element is called special in-so-far, as
stress MBS, this value is not known.
only a small number of commercial FE-programs offer
this type of analysis.
The inherent value of MBS could be easily determined
if the stress-concentration factors k1 of the insulator for
The special feature of the harmonic elements is, that
static bending were known.
the circumferential load need not be evenly distributed as
for a „plane“ problem and thus permits a non-
After [1] and [4] the static value of porcelain material
axisymmetric loading. For these elements a distributed
strength may be also applied for the dynamic short-circuit
load F‘(θ) is developed into a Fourier series of harmonic
case.
load functions, where θ is the circumferential coordinate:

F‘(θ) = A0+A1cosθ +B1sinθ + ........ + Ancos(nθ) +


E. Post-Insulator Strength
Bnsin(nθ) (1)
The withstand of a post insulator can so far only
In principle any load distribution can thus be modelled
judged for the cases of static to quasi- static bending. The
using the appropiate number of terms, where each term
latter case may be assumed for quite a number of open-air
substation layouts. The fall-of-span of stranded conductors is a will require a separate analysis and the effects have to be

119
superimposed. If only the term A0 is specified, we have
the case of ideal axisymmetric loading. III. EXEMPLARY CALCULATIONS

The non-axisymmetric loading of the insulator with a The first HE-studies on insulators, static as well as
single bending force may be treated with just one term of dynamic, were performed on the 17 shed, 110 kV, 800 kp
the first harmonic, using either one or two components, MBL post insulator Z110N-800 DIN 48120 that had been
see Fig.1, thus only requiring a single analysis per time the object of [4] and [5] and of which, therefore, more
step. Version b) that defines a resultant uniform lateral data than usual were known. On a remaining specimen
load in the global x-direction was used in the from [4] strain-gauge measurements of a higher precision
calculations. than in [5] were also done.

For comparison a second insulator of identical general


geometries (height, core diameters), fittings and MBL but
with the higher number of 23 sheds, alternating large and
smaller sheds, was then also analyzed with the intent to
increase experience, extend the number of sets of data
and to demonstrate trends.

The following general results of a more principial


nature relate to the 17 shed DIN insulator. The results of
the studies in terms of actual values, tendencies and
general conclusions are to be found in chapter IV. below.
Fig. 1 Versions of load distribution using the A1-term of Eq. (1)

In the present context, it is important to note that the


Harmonic-Element technique can also be applied for
modal analysis, the eigenfrequencies, that are, together
with the displacements, the essential and available control
data for checking the quality of any model. A further
effort reduction advantage is, that only half of the
longtudinal cross-section, see Fig.2, need be discretized.
The FE-analysis computer program used in the study was
ANSYS [14, 15] of Swanson Analysis Systems, Inc. .

A. Test versus analytically accessible data

A cylindrical beam, one end fixed, the other end free


can be assessed with precision by the analytical methods
of elastic theory. Therefore, the quality of the method
was checked for a not too finely discretized harmonic
element model of a beam versus the usual analysis of the
same beam, both with a perpendicular bending force at
the free end. Global geometries and material data were
those of the 110 kV insulators used in [4].

The relative deviations of the characteristic values


determined with the HE-model against the analytic
calculation are convincingly small:

- Maximum displacement 0.85 %


- Maximum bending stress 0.43 %
- First eigenfrequency 0.27 %
- Second eigenfrequency 3.3 %
- Third eigenfrequency 7.3 %

In a further stage of the study, method and calculation Fig.2 Harmonic-element model of Z110N-800 DIN 48120,
were also checked against strain-gauge measurements on lower part with sheds S1 to S5 shown
one of the exemplary insulator types; see III.C below.

120
A. Harmonic-element insulator model Fig. 5 , showing iso-stress lines for the region of the
lowest sheds S1 and S2, demonstrates the stress
The DIN-insulator was discretized into a harmonic- concentrations at the transitions from core to shed, giving
element model of ca. 3500 elements and 4500 nodes, also the respective values of k1 .
requiring a CPU time of ca. 20 s for a static run on a
2.5 GHz PC. Fig.2 shows the lower part of the
insulating body.

B. Static calculation

The results of a static run for a bending load of 1,75


kN in terms of the principal stresses σ1 along the surface
of the insulating body is shown in Fig. 3 . Coordinate h
is the coordinate Y of Fig.2 .

Fig. 5 Calculated iso-stress lines for region of lowest sheds S1


and S2 and values k1 at the transition points

C. Test of calculated against measured values

The calculation could be checked against measure-


ments from bending tests performed on the same
insulator. By means of strain gauges the specific material
elongation on the porcelain surface - strain - can be
Fig.3 Calculated principal stress σ1 on the insulator surface and measured.
reference value σ1r for the shed-less insulator of same
core diameter Between the lower sheds S1 and S2, where calculation
had shown the highest material contraints, the
Fig. 4 shows the resulting stress-concentration factors outer fibre strains in the principal bending plane were
k1 = σ /σ1r for the values of Fig. 3 . The absolute measured with applied chains of strain gauges. The
maximum directly above shed S1 is 1.6 , while the chains consisted each of 10 individual strain gauges with
maxima upwards from this point are constant with values an active length of 1.5 mm that were spaced at 2 mm. The
of 1.5 . parameters used for the tests corresponded to those of the
calculation.

The tangential strain of the FE analysis ε is to be


determined with

εyy = 1/E {σyy - ν (σxx +σzz)} (2)

Due to the modelling technique the tensor component


σyy of the calculation is always tangential to the surface
and σyy and εyy are unidirectional. E was taken from [4]
with 89000 N/mm2 and Poisson‘s ratio ν was applied with
0.3 .

Fig.6 a) shows the calculated values of surface strain


against the strain-gauge measurements from 4 separate
Fig.4 Calculated stress-concentration factors k1 tests in 6 b). The agreement satisfies in all respects and
in particular for the absolute maximum value near shed
It should be noted that even the cylindrical parts S1. Small differences near shed S2 can be explained by
between sheds show an increase of bending stress of the slight geometrical differences between the DIN
nearly 10% against the shedless insulator. The stress- insulator used for calculation and the actual insulator
concentration factors to be applied in actual strength used for tests.
considerations are the regional maxima.

121
IV. DYNAMIC CALCULATION

The performance of the harmonic FE-model in the


static and modal analysis shows its suitability also to
simulate the dynamic. Beside their value as criterion for
compatibility, the eigenfrequencies give indications as to
requirements as to integration time steps.

A. Time-function of the loading

For the dynamic analysis the time function of the


dynamic tests [4] was adopted. It is based on the two-
phase short-circuit time function of force for a maximum
force value of 8000 N, which is the MBL of the
considered insulator – see Fig. 7 . The rate-of-rise dF/dt
of the bending force represents one of the possible
extrema of the short-circuit dynamic range. Pinch forces
of bundled conductors could form another, higher one.

Fig. 6 Outer fibre strain ε between lowest sheds S1 and S2


a) Calculation with Harmonic Elements
Fig. 7 Dyn. bending force on the post insulator of 8 kN MBL
b) Measurement with strain-gauge chains in 4 separate 1 2-phase short-circuit force
test series 2 Time function of the applied bending force

D. Modal Analysis B. Damping


Besides the displacements, the modal analysis of the After test calculations without damping and structural
HE-model of the genuine insulator against the global damping ratios of 3% and 7% , the following calculations
technique reference model, represents the most important used a realistic value of 3%, also proposed in [7] .
feature for checking the conformity in their oscillatory
properties and behaviour and, hence, for the practical
applicability of the stress-concentration factors. C. Dynamic Stresses

For all specimen of the post insulators used in [4] the Fig. 8 gives the results of a dynamic run in terms of the
first eigenfrequencies were measured and also on some principal stresses σ1 for a number of significant time
the second and the third. steps.

The following assembles the values from the modal The parameter of the curves is time. It can be clearly
analysis performed on the HE-model versus the distinguished, how the point of highest stress moves into
measurements of [4]. The good agreement is obvious. and through the structure starting from the top.

Calculated Measured At t = 10 ms a „static“ distribution is reached. This


1. Eigenfrequency/Hz 44 41-47; mean 44 time coincides with the moment, where the maximum of
2. Eigenfrequency/Hz 287 264-278 the dynamic stress-concentration factor k1(t) curve is
3. Eigenfrequency/Hz 788 769-773 reached – see D below .

122
D. Dynamic stress-concentration factor In principle, it seems reasonable to apply these regional
dynamic maximum values k1(t and h), that could be
For the determination of dynamic k-factors also the named k1max , for the practical assessment of dynamic
reference model has to be subjected to the same time maximum stress. They have the rate-of-rise of the
functions. To be applicable in practice, the reference loading time function as upper limiting parameter.
insulator of identical core diameters had to be modelled
with the usual beam elements to identical oscillatory As shown in exemplary form in Fig. 10 a) for the DIN
behaviour – displacements, eigenfrequencies, damping. insulator and in Fig. 10 b) for the 26 shed insulator of
same height mentioned 2nd paragraph of chapter III. ,
Beside their known variations with the coordinate of analysed under completely identical parameters, the
height, the dynamic factors also vary with time, as it is regional dynamic maximum values k1max are distinctly
above the respective static values. The values 1.6 and 1.7
for the 17 shed and the 23 shed insulator are the absolute
static maxima. The respective ratios dynamic/static are
quite considerable 1.18 and 1.23 .

The fact that the dynamic cross-sectional stress


distributions are different from the static makes strength
considerations more complicated in an unexpected and so
far undefined extent.

Fig. 8 Calculated principal stress σ1 on the insulator surface at


dynamic loading . Parameter: time step

shown in Fig. 9 for the node of highest stress just above


shed S1. The maximum is reached at about 10 ms –
compare the loading time function.

Fig. 10 Comparison of static and dynamic k1 factors


a) 17 shed DIN post insulator
b) 23 shed post insulator

V. SUMMARY, REQUIREMENT FOR FURTHER


WORK, FINAL REMARKS

The studies were designed to find the principle of a


practicable general access to the assessment of the
dynamic short-circuit strength for high-volltage post
insulators, i.e. the easier access to the determination of
the so-called Equivalent Static Load ESL defining the
Fig. 9 Variation of the local stress-concentration factor k1 with ultimate allowable dynamic loading
time

123
The applied Finite Element technique of Harmonic [2] Mathejczyk, M.; Stein, N.: Kurzschlußseilzüge
Elements has been successfully tested and found engebündelter Doppelseile in Schaltanlagen. ETZ-A, 97,
particularly suited to tackle the problem with least effort 1976, pp.493-498
in terms of preparation and solution of the structural
analysis problem, particularly on smaller computers.
[3] CIGRE Technical Brochure 105: The mechanical
The achieved calculation results have been shown to effects of short-circuit currents in open-air substations,
be extremely accurate and have been checked against test Paris, 1996
results.
[4] Stein, N.; Bauer, E.; Brandt, E.; Dannheim, H.;
The results of the voluminous study have brought the Lehmann, W.; Meyer, W.; Pietsch, K. – Dynamic
intended aim considerably nearer, yet, have also detected behaviour and strength of high-voltage substation post-
unforseen new problems – the dependency of stress-con- insulators under short-circuit loads. CIGRE-Report 23-
centration factors on the loading time function - that need 12, 1984
to be object of further studies. In so far, only a narrower
spectrum than hoped for, but covering the applications [5] Stein, N. – Strain-/stress-concentrations in insulators
with rigid conductors, was achieved. due to sheds. CIGRE Proceedings 1984, Vol.1, Gr.23,
pp.45-46
Studies on other insulator types, faster (pinch force)
time functions, as well as a closer look on the transitions [6] Miri, A.M.; Stein, N.: Kurzschlußbeanspruchung und
of the insulating body to the metal fittings, in particular at –Festigkeit von Hochspannungs-Stützisolatoren. Etz
the bottom end are needed. The, at once, safe and Archiv, Vol. 10 (1988), H.3, pp. 89-96
economic selection of post - insulators is of particular
importance and the complicating influence of part of the [7] Fraikin, R.; Lilien, J.L.: Dynamic behaviour of. h.v.
studies needs to be discussed among experts. post insulators; Belgian tests, model proposition and
future work. CIGRE SC23-81(WG02)-80-IWD
Test experience [2, 3] has shown that the slow dyna-
mic response of the insulator will usually compensate [8] Bathe, K.F.: Finite-Elemente-Methoden, Berlin,
fast loading effects as regards its strength, which may Springer, 1986
have the consequence of a lower value of Equivalent
Static Load than the actual maximum dynamic load. Yet, [9] Zienkiewicz, O. C. – Methode der finiten Elemente.
also this particular influence requires more and closer München: Hanser, 1984
study. Since the most effective method for further
studies is now available, tested and found most accurate, [10] Clough, R.W..; Rashid, Y. – Finite element analysis
these studies will not require a too high effort. of axisymmetric solids. J. of the Engng. Mech. Div.,
Proc. of the Amer. Soc. of Civ. Eng. (ASCE) 91 (1965)
A consecutive study on the adaptive optimization of H. EM 1, pp.71-85
the form of the problematic transitions, while keeping
creeping distances and core diameters, has been [11] Wilson, E. L. – Structural analysis of axisymmetric
performed after this study and has resulted, for instance, solids. J. of the Amer. Inst. of Aeronaut. a. Astronaut. 3
in a reduction of the static maximum concentration factor (1965) H. 12, pp. 2269-2274
of the DIN- insulator by 22 % from 1.6 down to 1.25 .
[12] Buck, K. E. – Zur Berechnung der Verschiebungen
Composite insulators, that are in growing use for new und Spannungen in rotationssymmetrischen Körpern
substations [16], do not face the same problem as ceramic unter beliebiger Belastung. Diss. Univ. Stuttgart, 1970
insulators due to their silicone surface and sheds. The
elasticity of these parts is considerably higher than that of [13] Bathe, K.J.; Wilson, E.; Peterson, F.; Wunderlich,
their core and thus stress concentration is no problem at W. – SAP IV-B, Beschreibung und Benutzerhandbuch.
all. Additionally composite insulators possess a so-called Ein Programmsystem zur linearen statischen und
safe failure mode, which is an advantage in the case of dynamischen Berechnung von Tragwerken. Ruhr-Univ.
overloading. It is envisaged to investigate in a future Bochum, 1979
study their dynamic properties under short-circuit. The
present study will keep its relevance for refurbishment, [14] Ansys user’s manual rev. 4.2, vol. I & II. Houston:
uprating and special cases. Swanson Analysis Syst.

[15] Ansys theoretical manual. Houston: Swanson


REFERENCES Analysis Syst.

[1] Logunov, L.P.: On the impact strength of porcelain [16] Papailiou, K.O.: Composite Insulators are gaining
support pin insulators. Soviet Power Engineering No. 12, ground. IEEE Transmission and Distribution Conference,
1974, pp. 729-734 New Orleans 11-16 April 1999. Vol.2., pp. 827-833

124

You might also like