Professional Documents
Culture Documents
com
ScienceDirect
Most process intensification research and its applications have We posit that intensification can also be achieved by
focused on process and equipment design modifications. In making operational and control changes to a process
this paper, we present an overview of existing developments (whether conventional or intensified by design). In this
and opportunities in dynamic process intensification (DI), which context, we define dynamic process intensification (DI) as
comprises changes to dynamics, operational and control of a ‘any change to the dynamics, operation strategy and/or control of
chemical process, that result in substantial efficiency a conventional or intensified system, that leads to a substantially
improvements. After reviewing a series of examples, we identify more efficient processing path.’ [5].
several fundamental principles underlying DI, including
equipment design and operational approaches. We conclude The purpose of this article is to systematically review
with a number of challenges and opportunities related to DI existing DI approaches, define future directions and
system design, operation and control, and fault detection and research targets, as well as potential challenges and road-
isolation. blocks to achieve them. A set of structural/operational
process features and elements of DI will be identified,
Addresses which will then serve to construct a taxonomy of pro-
1
McKetta Dept. of Chemical Engineering, The University of Texas at cesses and technologies that fall within the realm of DI.
Austin, Austin, TX 78712, United States
2
Energy Institute, The University of Texas at Austin, Austin, TX 78712, Motivating examples
United States
3
Institute for Computational Engineering and Sciences, The University of
The focus of this paper is on continuous (rather than
Texas at Austin, Austin, TX 78712, United States batch) processing, implying that the product(s) must be
made in a continuous stream. Thus, the dynamically
Corresponding author: Baldea, Michael (mbaldea@che.utexas.edu) intensified processes considered here operate in a peri-
odic/cyclical fashion. They reach a periodic steady state
Current Opinion in Chemical Engineering 2018, 22:48–53 which, on the average, corresponds to the desired produc-
tion rate(s) and/or product quality. To achieve this, a
This review comes from a themed issue on Process systems engi-
neering: advanced manufacturing storage ‘buffer’ may be required to attenuate any fluctua-
tions in these performance indicators that are associated
Edited by Karen Fletcher, Jim Bielenberg, Mahmoud El-Hawlwagi
and Ka Ming Ng
with the periodic operations. Further, it is expected that
some figure of merit (e.g. energy use per unit product) is,
For a complete overview see the Issue
all other things being equal, better for the dynamically
Available online 5th October 2018 intensified process than for a conventional process of
https://doi.org/10.1016/j.coche.2018.08.003 equal capacity, if the latter exists. We begin with review-
2211-3398/ã 2018 Elsevier Ltd. All rights reserved. ing existing dynamically intensified systems and subse-
quently use this information to extract some generic
features of DI systems.
While it was surmised that this mode of operation exploits Reverse-flow reaction systems
the nonlinear influence of temperature on the rates of Reverse-flow reactors (RFRs) have been proposed as an
competing reactions, initial reporting of these improve- intensified design for carrying out endothermic reactions
ments was largely based on empirical observations. Fur- [20]. The system comprises a tubular reactor whose
ther work concentrated on identifying first, the generic operation consists of two separate, discrete steps: a heat-
conditions that a system must fulfil in order for periodic ing step, whereby a fuel is (catalytically) combusted to
operation to be beneficial and second, determining the heat the catalyst bed, and a reaction step, where the
optimal parameters of the cyclic forcing function (i.e. process stream (typically fed to the reactor at he opposite
amplitude and period) for maximizing this benefit [7–11]. end) passes over the hot catalyst and undergoes the
desired reactions. Pressurization/depressurization, purg-
ing, or other steps can be inserted in the operating cycle as
Cyclic distillation needed, and multiple reactors can be operated in parallel
Research carried out in the 1960s [12–15] also revealed to ensure a (relatively) constant flow of product. The
the possibility of operating distillation systems in a cyclic reaction system reaches a cyclic/periodic steady state,
fashion. Cyclic distillation in this case involved segregat- rather than having a single steady-state operating point
ing the fluid traffic in the distillation column, with fast (i. as is the case with most continuous processes. Several
e. in the order of seconds) switching between (downward) applications of reverse flow reactors have been reported,
liquid flow and (upward) vapor flow. Segregation reduces including the production of syngas [21], SO2 oxidation
unnecessary mixing between the vapor and liquid phase, [22], and the mitigation of volatile organic compound
and between the liquid material on adjacent trays. Lit- emissions [23]. Operational complexity notwithstanding,
erature reports claim increases in stage efficiency and it is argued that reverse-flow systems have capital expen-
throughput compared to conventional columns [13–15], diture advantages compared to a conventional system of
as well as improvements in energy use [16]. Columns equivalent capacity. The latter would require, for exam-
intended for cyclic operation use customized/dedicated ple, a separate heating furnace and a feed-effluent heat
internal components (including special trays that have no exchanger to maintain the required thermal regime.
downcomers). The concept of cyclic operation was also
applied to extraction columns [17]. In general, adding Oscillatory baffled reactors (OBRs)
such technology to a chemical plant is likely acceptable OBRs are tubular reactors that diverge from the conven-
for new investments, but retrofits of existing distillation or tional plug-flow design in two important ways. From a
extraction towers may prove costly and problematic. design perspective, the flow encounters resistances (baf-
fles) placed at equally spaced intervals along the flow
path. From an operational perspective, an oscillatory
Periodic distillation based on output multiplicity motion is imposed over the net flow by means of a
Early research [18,19] demonstrated that (binary) distil- secondary pump/actuator. Combined with the baffled
lation columns exhibit steady-state multiplicity. Of par- structure, the oscillatory fluid motion simulates the oper-
ticular interest is output multiplicity, where a column ation of a cascade of CSTRs, effectively achieving a
can produce distillate of two different purities with the desirable plug flow profile. OBRs provide an intensifica-
same reboiler duty. A recent development in the tion opportunity for reactions that require long residence
dynamic intensification of distillation processes relies times and where a conventional plug-flow reactor would
on exploiting this nonlinearity [5] to lower column have an impractical length. We note that the idea of
energy use (defined as the sum of reboiler and condenser oscillatory mixing has been used for the intensification
duties) for a distillate stream of desired purity. This of other processes; see Ni et al. [24] for a review.
dynamic intensification approach relies on the observa-
tion that the target product can be obtained as a blend of Cyclic adsorption processes
two auxiliary products (one of higher purity, the other of Pressure swing adsorption (PSA) is a widespread tech-
lower purity than desired), both of which have lower nique for separating gas mixtures, and has been applied
specific energy consumption. Thus, the operating con- to, for example, air separation (to generate oxygen or
ditions (column pressure, boilup rate) are periodically nitrogen) and the separation of hydrogen from syngas.
switched between the operating points corresponding to Fundamentally, PSA relies on the difference in adsorp-
the two auxiliary products, and the distillate stream is tion rates between the components of the mixture on a
collected in a buffer tank whose average composition given solid adsorbent, as well as on the influence of
equals that of the desired product. The control signals pressure on the amount of adsorbed component. A PSA
may be continuous or step-wise, with periods in the system comprises of two (or more) adsorbent-packed
order of magnitude of the time constant of the column beds, which are operated in a cyclic fashion, in a carefully
itself (minutes to hours). Importantly, this concept relies choreographed sequence of discrete steps that are imple-
on a conventional distillation column and does not mented in practice by changing the positions of several
require any dedicated hardware. automatic valves that control the flow and feed/extraction
capital expenditure. For such equipment, periodic Challenges and opportunities for future
operation can be regarded as a sufficient condition for research
(dynamic) intensification. Conceptual system design
In many circumstances (as is the case for cyclic distil- A key challenge for process intensification in general, and
lation, reverse-flow reaction systems and SMB chroma- for DI in particular, is finding a systematic methodology
tography), the equipment used in a dynamically inten- for identifying candidate systems [27,29,30]. While the
sified process is customized/dedicated and cannot operate underlying principles listed above provide some guid-
in a conventional continuous fashion. Such inherently ance, there is — to our knowledge— no rigorous means
dynamically intensified systems must be designed for functionalization and for systematic generation of
simultaneously with their operating strategy, which periodic intensified device designs. This task becomes
requires significant more effort than in the case of even more arduous given that the design of the physical
standard equipment meant for steady-state operation. system must be coupled with the design of a dynamic
In this case, periodic operation is a necessary condition operating policy. While outside the scope of the present
for intensification. For such systems, an ‘equivalent’ paper, we note that a further complication can arise when
continuous process (using ‘standard’ equipment) capa- semicontinuous periodic processes are considered (e.g.
ble of making the same products at the same rate may [31]), where a longer-term, batch dynamic must taken
or may not exist. The former is true in the case of, for into account in addition to the (short-term) periodic
example, cyclic distillation, OBRs and RFRs, while the fluctuations.
latter applies to the PSA system. When available, the
continuous equivalent constitutes a useful baseline for Design optimization
estimating the economic benefits of dynamic Once a DI concept has been identified, the system must
intensification. be modeled, simulated and optimized. The models of DI
systems often include discontinuities, and system simu-
Operating strategy lation and optimization must deal with changes in model
As pointed out above, there is an intimate connection structure (which reflect, e.g. changes in flow direction or
between the choice of equipment for DI and the choice of system connectivity). Furthermore, system models have
periodic operating strategy, whether the latter is a neces- to accurately represent multiple phenomena occurring
sary or a sufficient condition for intensification. We iden- over spatially distributed domains, and are thus often of
tify three classes of periodic operating strategies below: considerable size. A first difficulty is thus related to
simulation and identifying the cyclic steady state, a task
Continuous, which involves using periodic control sig- that is currently largely accomplished by simulating a
nals that are continuous in time, such as sine waves. large number (hundreds or thousands) of operating cycles,
The periodic operation strategies described in Sections often in conjunction with some convergence acceleration
‘Periodic operation of continuous chemical reactors’, techniques [32]. In turn, this complicates the simulta-
‘Periodic distillation based on output multiplicity’, and neous optimization of the system design and its operating
‘Oscillatory baffled reactors (OBRs)’ fall under this policy. The latter is typically carried out using a full
category. discretization of the time and spatial domains, which
Discrete/discontinuous, whereby the control signals are results in a large-scale mixed-integer nonlinear program
discontinuous and based on a switching logic (e.g. on/ (MINLP) [33], although the use of sequential methods
off or step sequences), and their implementation is [34,35] and of surrogate, grey-box models [36] has also
associated with changes in the configuration or state of been reported. The development of efficient methods for
the physical equipment (e.g. change in flow direction, solving the simulation and optimization problems that
changes in the nature of the feed). Reverse-flow reac- pertain to DI systems remains an open topic for research
tors, PSA and SMB separations, and cyclic distillation [37,38].
columns belong to this category.
Finally, a hybrid operating strategy is possible, whereby System operation and control
the control signals are discontinuous but are applied to Periodic operation is inherently dynamic and presents
a systems whose states are continuous and whose new operational challenges in addition to the careful
structure does not change. This is the case, for exam- coordination of the switching logic. The optimal control
ple, with the periodic operation of CSTRS when the of periodic systems has received significant attention in
control signals are discontinuous. the past, with research concentrating on extending the
model predictive control (MPC) paradigm [39,40], includ-
The choice of equipment and operating strategy is then ing economic MPC [41]. The development of methods
reflected in the structure of the mathematical model of for operation fault detection for periodic systems is at a
the system, and has a significant impact on applicable relatively early stage [42]. While both model-based and
design and operational optimization techniques, a matter data-driven methods are appropriate for this class of
that will be discussed later in this article. systems, we believe that data-driven methods deserve
attention — noting, however, that they will require a 11. Sowa SW, Baldea M, Contreras LM: Optimizing metabolite
production using periodic oscillations. PLoS Comput Biol 2014,
departure from existing established methods, such as 10:e1003658.
those based on principal component analysis. 12. Cannon MR: Controlled cycling improves various processes.
Ind Eng Chem 1961, 53:629.
Conclusions 13. McWhirter JR, Cannon MR: Controlled cycling distillation in a
Dynamic process intensification holds significant promise packed-plate column. Ind Eng Chem 1961, 53:632-634.
in terms of improving process performance. There are, 14. Gaska RA, Cannon MR: Controlled cycling distillation in sieve
however, several challenges for fully realizing this. First, a and screen plate towers. Ind Eng Chem 1961, 53:630-631.
systematic framework for identification and conceptual 15. Schrodt VN et al.: Plant-scale study of controlled cyclic
design/synthesis of dynamically intensified processes is distillation. Chem Eng Sci 1967, 22:759-767.
needed. Second, there are mathematical and numerical 16. Maleta BV et al.: AIChE J 2015, 61:2581-2591.
challenges related to modeling, simulation and optimiza- 17. Seibert AF, Humphrey JL, Fair JR: Efficiency of a controlled-
tion of DI systems, particularly related to computing and cycle extractor. Solvent Extr Ion Exchange 1986, 4:1049-1071.
optimizing the cyclic steady state. Finally, while good 18. Jacobsen EW, Skogestad S: Multiple steady states in ideal two-
progress has been made on optimal control of periodic product distillation. AIChE J 1991, 37:499-511.
processes, fault detection for such systems offers consid- 19. Koggersbøl A et al.: An output multiplicity in binary distillation:
erable opportunities. Newly developed techniques in experimental verification. Comput Chem Eng 1996, 206:S835-
S840.
these areas should be implemented in intuitive software
tools and packages that can be distributed and used 20. Matros Sh, Bunimovich GA: Reverse-flow operation in fixed bed
catalytic reactors. Catal Rev 1996, 38:1-68.
widely in both academia and industry.
21. Blanks RF, Wittrig TS, Peterson DA: Bidirectional adiabatic
synthesis gas generator. Chem Eng Sci 1990, 45(8):2407-2413.
Conflict of interest statement
Nothing declared. 22. Bunimovich GA et al.: SO2 oxidation in a reverse-flow reactor:
influence of a vanadium catalyst dynamic properties. Chem
Eng Sci 1995, 50:565-580.
Acknowledgements 23. van de Beld B et al.: Removal of volatile organic compounds
The authors gratefully acknowledge funding from AIChE RAPID, a DOE from polluted air in a reverse flow reactor: an experimental
Manufacturing Institute. MB was partially supported by the National study. Ind Eng Chem Res 1994, 33:2946-2956.
Science Foundation (NSF) through the CAREER Award 1454433 and
Award CBET-1512379. 24. Ni X et al.: Mixing through oscillations and pulsations: a guide
to achieving process enhancements in the chemical and
process industries. Trans IChemE 2003, 81(Part A).
References and recommended reading 25. Ruthven DM: Principles of Adsorption and Adsorption Processes.
Papers of particular interest, published within the period of review,
New York, NY: John Wiley & Sons; 1984.
have been highlighted as:
26. Arora A et al.: Optimal synthesis of periodic sorption enhanced
of special interest reaction processes with application to hydrogen production.
of outstanding interest Comput Chem Eng 2018, 115:89-111.
1. Stankiewicz AI, Moulijn JA: Process intensification: 27. Arora A, Iyer SS, Hasan MMF: GRAMS: A general framework
transforming chemical engineering. Chem Eng Prog 2000, describing adsorption, reaction and sorption-enhanced
96:22-34. reaction processes. Chem Eng Sci 2018.
2. Tsouris C, Porcelli JV: Process intensification-has its time 28. Juza M, Mazzotti M, Morbidelli M: Simulated moving-bed
finally come? Chem Eng Prog 2003, 99:50-55. chromatography and its application to chirotechnology.
Trends Biotechnol 2000, 18:108-118.
3. Reay DA, Ramshaw C, Harvey AP: Process Intensification:
Engineering for Efficiency, Sustainability and Flexibility. Waltham, 29. Demirel SE, Li J, Hasan MMF: Systematic process
MA: Butterworth-Heinemann; 2013. intensification using building blocks. Comput Chem Eng 2017,
105:2-38.
4. Baldea M: From process integration to process intensification. The paper discusses a novel framework for synthesis of intensified
Comput Chem Eng 2015, 81:104-114. systems, based on building blocks. The resulting process structures
can be translated to available process intensification equipment.
5. Yan L, Baldea M, Edgar TF: Dynamic intensification of
distillation operations. AIChE J 2018. (submitted for publication). 30. Baldea M: A pinch-like targeting framework for systematic
thermal process intensification. AIChE J 2018, 64:877-885.
6. Horn FJM, Lin RC: Periodic processes: a variational approach.
The paper discusses a novel general framework for synthesis of intensi-
Ind Eng Chem Proc Des Dev 1967, 6:21-30.
fied systems. It extends the concept of pinch analysis from streams to
7. Sterman LE, Ydstie BE: The steady-state process with periodic units, with the goal of achieving intensification via direct heat exchange
perturbations. Chem Eng Sci 1990, 45:721-736. between unit operations, possibly by carrying multiple conventional
operations in a single, intensified device.
8. Bailey JE, Horn FJM: Comparison between two sufficient
conditions for improvement of an optimal steady-state 31. Phimister JR, Seider WD: Semicontinuous, pressure-swing
process by periodic operation. J Optim Theor Appl 1971, 7:378- distillation. Ind Eng Chem Res 2000, 39:122-130.
384.
32. Smith OJ IV, Westerberg AW: Acceleration of cyclic steady
9. Bailey JE: Periodic operation of chemical reactors: a review. state convergence for pressure swing adsorption models. Ind
Chem Eng Commun 1974, 1:111-124. Eng Chem Res 1992, 31:1569-1573.
10. Bittanti S, Fronza G, Guardabassi G: Periodic control: a 33. Kawajiri Y, Biegler LT: Optimization strategies for simulated
frequency domain approach. IEEE Trans Automat Control 1973, moving bed and PowerFeed processes. AIChE J 2006, 52:1343-
18:33-38. 1350.
34. Nilchan S, Pantelides CC: On the optimisation of periodic show that this facilitates the simulation of the process, and also introduce
adsorption processes. Adsorption 1998, 4:113-147. a novel time relaxation-based optimization approach for periodic
processes.
35. Ko D, Siriwardane R, Biegler LT: Optimization of a pressure-
swing adsorption process using zeolite 13X for CO2 39. Lee JH, Natarajan S, Lee KS: A model-based predictive control
sequestration. Ind Eng Chem Res 2003, 42:339-348. approach to repetitive control of continuous processes with
periodic operations. J Process Control 2001, 11:195-207.
36. Iyer SS et al.: Integrated carbon capture and conversion to
produce syngas: novel process design, intensification, and 40. Erdem G et al.: Automatic control of simulated moving beds.
optimization. Ind Eng Chem Res 2017, 56:8622-8648. Ind Eng Chem Res 2004, 43:405-421.
37. Pattison RC, Schmal JP, Pantelides CC: Efficient computation of 41. Huang R, Harinath E, Biegler LT: Lyapunov stability of
cyclic steady states in periodic adsorption processes using economically oriented NMPC for cyclic processes. J Process
the JFNK method. In Proceedings of 2015 AIChE Annual Control 2011, 21:501-509.
Meeting; AIChE, Salt Lake City, UT: 2015.
42. Wang RC, Edgar TF, Baldea M: A geometric framework for
38. Tsay C, Pattison RC, Baldea M: A pseudo-transient optimization monitoring and fault detection for periodic processes. AIChE J
framework for periodic processes: pressure swing adsorption 2017, 63:2719-2730.
and simulated moving bed chromatography. AIChE J 2017 This is one of the few contributions available on fault detection for periodic
http://dx.doi.org/10.1002/aic.15987. processes. The authors use a geometric approach based on representing
The paper discusses a novel modeling, simulation and optimization process data on a Kiviat (spider) diagram. Fault detection proceeds in two
strategy for periodic processes. The nominally algebraic full discretization steps: inter-cycle detection, which identifies a faulty cycle, and intra-cycle
(time and space domain) of the system model is converted to a pseudo- detection, which identifies the moment of occurrence of a fault and
trasient model in differential-algebraic equation (DAE) form. The authors attempts fault isolation.