Applications of Dynamic Simulation

Abstract

Applications of dynamic simulation today span R&D, Design, Operations, and Training. Proper implementation of dynamic simulation
can deliver substantial benefits including tighter design, reduced project cycle time, smoother start-ups and optimized production.
Because many of the existing dynamic simulation tools are time-consuming and difficult to use, these benefits are often not fully
achieved. In this paper we identify the major areas of application for dynamic simulation, outline the important issues and potential
benefits within each area, and provide a profile of a dynamic simulation tool that removes the barriers-to-use, and allows the benefits
of dynamic simulation to be more fully realized.

Introduction

Applications for dynamic simulation can be found in virtually all areas of process engineering. The tangible benefits of using dynamic
simulation can be seen in tighter design, reduced project cycle times, smoother start-ups, and optimized production. These benefits
are typically derived from improved process understanding. With all the advantages of dynamic simulation, it is sometimes surprising
to discover resistance to its use. This resistance is understandable, however, when we examine the background and nature of many
of the existing dynamic simulation tools.

Due to the unique combination of computer programming, numerical integration, modeling, and sophisticated chemical and
thermodynamics skills required, dynamic simulation has remained largely in the hands of "experts" over the years. Additionally,
because of the extremely large number of calculations that dynamic simulations require, these simulations have been reserved for
very large and powerful main-frame or mini-computers. As such, only the most complex and challenging of process designs seemed
to warrant the use of dynamic simulation. Today however, thanks to advancements in computer hardware and software technology,
such as Object-Oriented Programming [7], and the development of new ways of packaging dynamic simulation [2,3], this technology
is rapidly evolving into a tool for everyday use by many engineers.

With the barriers-to-use disappearing, we believe that dynamic simulation will begin to grow in popularity and use throughout the
industry. In the following sections of this paper, we discuss the major applications of dynamic simulation, outline the important issues
and potential benefits within each area, and provide a profile of what we believe to be the ideal dynamic simulation tool.

Applications

Applications for dynamic simulation today are diverse, spanning R&D, design, operations, and training. The following highlights some
of the more significant areas:

● Conceptual Design
● Process Design
● Control Strategy Development
● Fault Tree and Safety Analysis & Evaluation
● Plant Control System Check-Out
● Operator/Engineer Training
● Process Improvement

Conceptual Design

Conceptual design is an area which to date has seen little use of dynamic simulation. It is not because dynamic issues are
unimportant at the conceptualization stage. On the contrary, operability is an important selection criteria for any new process. Instead,
the reason for not applying dynamic simulation is that there is currently no software tool on the market fast and flexible enough to
permit rapid screening of process alternatives.

Ignoring for a moment that suitable dynamic simulation tools are not available, what would we like to learn from the simulations at the
conceptualization stage? Two issues come to mind:

● Reactor operability including heat management

● Equipment overdesign requirement for control

Reactor Operability. When considering a new reaction process, there are many factors influencing the choice of reactor type. Among
those factors, safety and operability are preeminent. Reactor control is often a matter of heat management, and different types of
reactors have different characteristics when it comes to heating and cooling. For example, jacketed CSTRs have the surface area/
volume aspect to consider, whereas tubular reactors are prone to developing hot-spots. Being able to simulate the proposed reactor
system and a preliminary heat management control system would be invaluable in selecting among various reactor types for a given
process.

Overdesign for Control. Overdesign sounds like something we should avoid. However, when considering a new process, we can not
assume that it will handle external disturbances and still remain at its design rate. A control system needs some room to maneuver
flows and heat duties as it tries to correct for incoming disturbances. This extra room is the required "overdesign" (see [1]). For some
processes the overdesign needed for control can be quite expensive thus making those processes less attractive candidates.
However, the required overdesign information can only be practically obtained from dynamic simulations, with realistic control
systems and anticipated disturbances.

Process Design. One of the most important contributions that dynamic simulation can provide is in the area of process design. In
particular, dynamic simulation is well suited to address the ever increasing need to design plants for operability. The goal here is to
find not only an effective control structure, but to design a process that is easy to control. This must begin with a thorough
understanding of the process; how it operates, how disturbances are propagated, what features dominate the dynamic response,
what variables are most important to control, etc. To achieve this level of understanding, dynamics must be considered from the
beginning of the design/development process.

Figure 1a - Two-Column Ternary Separation

Understood and accounted for early in the design phase, dynamic controllability issues are normally easy to accommodate. Left as an
afterthought, however, they often appear as unexpected operating problems, and are usually difficult to handle. Adding a few extra
stages to a column, resizing a heat exchanger, or installing extra temperature sensor nozzles while still on paper is trivial. Making
 such changes once installation has begun, or -- as is too often the case -- once operation has begun, is difficult and often quite
expensive.

Figure 1b - Single Column

The potential benefits of considering operability at the design stage are considerable. In [5] it is shown how a simple, financial project
model can quantify what is achievable. For example, a 5% increase in capital productivity from process simplification has a net worth
of about $4 million on a $100 million fixed capital project. While such improvement opportunities not uncommon, they are rarely
implemented because experience shows that tight design often leads to poor operability. Only when the improved design can be
tested for operability before a decision is made, can the benefits be realized.

Figures 1a and 1b illustrate an example of process simplification at the design stage. Two distillation columns are normally used to
separate a ternary mixture. However, the mixture can also be separated in a single column with a side-draw. While the single column
is less expensive from a fixed cost standpoint, it is also harder to control and therefore carries the risk of becoming an operational
problem, or worse yet, a production bottle-neck. If the side-draw column later has to be replaced by two new columns, we face a very
expensive proposition. On the other hand, if the side-draw column can be properly designed to account for dynamic operability and
control issues, then the single column constitutes a real improvement in capital productivity.
While capital productivity can be achieved through tight design, it can also be realized from optimized production. A plant which is
designed with operability in mind and which is run by well trained operators, can produce considerably more than a plant where
operability issues were never explicitly addressed. Dynamic simulation allows for optimized production by revealing trouble spots,
operability problems, and bottlenecks in the design before they become operating liabilities. In [5] it is shown that a 5% improvement
in process throughput has a net worth of $5 million for a $100 million fixed capital project.

Based on the benefits of capital productivity it would seem as if we should always strive to reduce capital and optimize production.
Unfortunately, this is not always possible. The reason is that one often encounters a trade-off between steady-state cost-saving
opportunities and dynamic operability. For example, from a steady-state viewpoint it is always beneficial to minimize holdup in
processing equipment and surge vessels. The reason is that holdup means large equipment which translates into fixed capital. From
an operational standpoint, however, holdup can make the difference between a smooth working plant and an inoperable one. In
operations, holdup means isolation from disturbances, and may mean being able to run one section of the plant when another is
down. However, too much holdup is not good either, even from an operational standpoint. Extra holdup means long process cycle-
times which implies that corrections to quality problems take longer. The extra holdup has now become a liability rather than an asset.
The steady-state/dynamic trade-off can only be resolved through dynamic simulations.

The value of the results from a dynamic analysis can be quite impressive, especially when encountering a design/control trade-off.
For example, when we are faced with the decision of increasing capital expenditure for the sake of improving operability, we must be
certain that the additional costs will not exceed the benefits. An example of this dilemma is given in [6]. Here the heat management
control system for an adiabatic, fixed bed reactor is considered. From a steady-state investment viewpoint, a reactor feed/effluent
heat exchange system with a bypass, as illustrated in Figure 2a, would be the most economical. However, as is shown in [6], such a
low cost design can lead to severe operational difficulties, mostly due to the feedback loop formed by the heat exchangers. When the
costlier design shown in Figure 2b is adopted, the operational problems go away since the process heater effectively decouples the
reactor feed from the reactor effluent.

Figure 2a - Feed/Effluent Heat Exchange with Bypass

(Click to enlarge)
 Figure 2b - Feed/Effluent Heat Exchange with Heater
(Click to enlarge)

It should be obvious from the preceding discussion that dynamic simulations can play a crucial role in process design, and that the
financial benefits of making the right decisions are substantial. But how much can we afford to invest in dynamic simulation? When
the stake is $4-5 million we can easily justify having four to five engineers on the task and still get a healthy payback. However, with
yesterday's dynamic simulation tools, the number of engineers is not the issue; it is the time it takes to complete a dynamic simulation
study. The following quotes from Vogel [8] summarize the problem: "The large amount of time required to develop a simulation
starting from (scratch) made it prohibitive for us except in critical cases.... The design schedules allow little time for dynamic
simulation".

While we can report the same experience, it might still seem puzzling to many why the project does not allow for dynamic simulation
studies if they are so beneficial. The answer is simple; time is money. In [5] it is shown that one extra year in the project schedule,
even without additional costs, represents a net present loss of $3-5 million on a $100 million fixed capital project. In addition, the cost
of a delay is certain, while the savings that we aspire to achieve with the delay are not. The magnitude of a somewhat uncertain
payback must be quite large to compensate for the cost of a project delay. Consequently in the past, only a few of the most critical
cases were studied. Only when the dynamic simulation study carries a zero time penalty can we afford to look at dynamic operability
in the entire process, which in the end, will deliver large paybacks. The availability of fast, modular, interactive and highly integrated
dynamic simulation systems like HYSYS from Hyprotech, is making this feasible. Dynamic analysis can now be completed
simultaneously with steady-state flow sheet design adding no extra time to the project while reaping all the benefits.

Process Control Strategy Development

The use of dynamic simulation for control strategy development is natural. Dynamic simulation provides insight into the process by
allowing the control engineer to:

● Determine how disturbances propagate through the system

● Understand the relative sensitivity of process variables to process upsets
● Investigate process and control-loop interactions
 ● Determine the effect of equipment sizing or arrangement changes on disturbance rejection and overall operability
● Determine the effects of ambient conditions on process operation
● Compare the dynamic performance of alternative control strategies
● Perform control-loop tuning
● Investigate start-up, shut-down, low-, mid-, and max-throughput operation

The advantages of such studies are many, owing to a deeper understanding of the process. When involved during the design phase,
these studies result in an improved plant design that is intrinsically easy to start-up, control and operate. Easy start-up alone can be
quite valuable as demonstrated in [5]. In one example, two months shorter start-up of a $100 million fixed capital process was worth
$3 million.

Also, when the design considers the dynamic operability of the plant, a much simpler control strategy usually results. This means that
operating personnel are more likely to understand the strategy and will be better able to maintain the system. Despite the wide-spread
use of process automation, many plants run with controls set in "manual" because the control system is too complicated, too difficult
to maintain, or is ineffectual. On the other hand, a simple, well designed and understood control system is usually left in automatic,
thus making the best use of the capital investment in terms of plant throughput and product quality. Again, the primary benefit of
dynamic simulation is improved process understanding. And it is this understanding that allows for the simplicity, flexibility, and
reliability demanded of today's most competitive process plants.

Fault-Tree and Safety Analysis & Evaluation

Safety is an area that has always been an important factor for most process plants. Stricter regulations regarding hazardous spills and
waste generation, and the ever-increasing fines that result from infractions, have placed an even greater emphasis on fault-tree and
plant safety design.

Using dynamic simulation, realistic chains-of-events may be modeled, tested, and evaluated against industry standards and
guidelines. In this way, problem areas in the plant can be identified in the design phase, and can either be redesigned, or fitted with
ancillary safety equipment. Additionally, by being able to model the sequence and duration of events leading to a fault or safety
incident, plant personnel are able to define more effective safety action plans and shutdown/evacuation strategies.

Again, using this information early in the design stage helps to ensure that the process design is intrinsically safe, and that personnel,
property, and equipment are protected.

Plant Control System Check-Out

Most modern process plants utilize some degree of automation. Many employ an array of single-loop controls, while others have
integrated the entire operation into a Distributed Control System, or DCS; the latter becoming increasingly popular. For any new
installation, and when significant changes are being made to an existing operation, instrument engineers and technicians must
normally "check-out" the functionality of the new control system. This often involves creating mock signals in the field to simulate
levels, flows, pressures, etc., so that the controllers' inputs can be confirmed. Similarly, the controllers are exercised to confirm that
their outputs do indeed activate the appropriate valve. While this procedure verifies the I/O integrity, it does little else to assist in the
overall check-out of the installed control structure. Ideally, the hardware would be tested on the plant itself, using real transmitter
signals, and influencing real process variables. Unfortunately, running the plant with an un-tested control installation is usually too
risky for most operations managers to approve.
Dynamic simulation, however, provides a very elegant and often quite efficient means of addressing these issues. By substituting a
dynamic process model for the real process, the plant's control system may safely be evaluated without having to put expensive raw
materials or plant equipment at risk.

Since many DCS's keep their control configurations separate from their I/O configurations, it is easy to check out the plant's actual
control configuration while interfaced to a computer model, then simply switch over to the field I/O configuration for plant start-up.

Training

On-Line Operator Training. A natural extension of the plant check-out functionality is on-line operator training. Once the control
configuration has been validated and is working as expected, the integrated DCS/Dynamic model makes a perfect operator training
station. Operators are able to practice start-up, shut-down, and emergency action sequences all on the actual control-room hardware.

Many plants recognize the importance of operator training and purchase additional control hardware dedicated to this purpose. When
one considers that the model used for training can be the same model that was used for the process design, and the control strategy
development, the incremental cost for the additional equipment is minimal.

Off-Line Training/Education. Sometimes it is impossible or inconvenient to train on the plant's actual control hardware. In those
cases interactive, graphical dynamic simulations running off-line are still extremely valuable teaching and training tools.

Off-line, engineers are free to test or illustrate any number of control concepts. For plant operators and students, fundamental process
and control concepts can be demonstrated and clarified. "What-if" sequences can be run, and the process and its control strategy
changed altogether to test alternative concepts.

As a training tool, interactive dynamic simulation has found applications from the classroom, to the training room, and to the plant floor.

Platforms. It is instructive to note the four different types of training platforms. These include:

● Tie-Back
● Emulation
● Enhanced Emulation, and
● Stimulation.
 (Figure 3 - Stimulation Setup
(Click to enlarge)

In the Tie-Back method, control signals within the DCS are tied back to the process variables through simple transfer function blocks
such as first-order lags with dead-time. This technique has the advantage of being completely self-contained within the plant control
environment, and thus eliminates any interfacing requirements. The fidelity of the process model, however, is very crude, and is
therefore not recommended for extensive use in operator training. This method can be an effective and inexpensive way of checking
out a new DCS configuration.

Emulation is a training method where the process simulation and the control operating environment are modeled together on a
computer. Since this method does not involve the actual control hardware, it cannot be used for DCS check-out. It is, however, quite
useful for off-line operator training where it is important for the look and feel of the model's controller face plates, data displays, strip
charts, and other instrumentation displays to match closely that of the real DCS equipment.

Enhanced Emulation is similar to Emulation, however the user typically has a real operator console and keyboard (or some
reasonable facsimile) to interact with the simulation. This has the advantage of providing a more realistic environment in which to
train. However, like Emulation, it cannot be used for DCS check-out purposes.

Stimulation involves linking a high-fidelity process model directly to the actual DCS hardware, as shown in Figure 3. Since the actual
control equipment is being used to control with the model, this technique is very effective for both DCS check-out and operator
training. We believe that this is the superior method for a number of reasons. First, the same model that was used for process design,
and control strategy development, may be used for this implementation with only minor modifications to the model. Thus the
duplication of effort in making a special operator training simulation or DCS check-out model is avoided. Second, by linking the actual
DCS hardware with a high-fidelity model, operators receive the most realistic and useful training. Third, this is the only technique
which is applicable to both operator training and DCS check-out.

Process Improvement

While the areas discussed so far have been described in light of a new plant design/installation, dynamic simulation studies may be
applied to existing processes to achieve improved operation.
One of the great benefits of dynamic simulation is the high return/cost ratio it can provide through improved process understanding.
By simply understanding the process better, small changes in the arrangement of existing plant equipment, feed point locations,
operating conditions, or control strategies, can mean tremendous improvements in yield, process variability, energy utilization, product
quality, safety, and/or waste minimization. With minimal capital expense, process understanding can bring impressive returns.

A simple example will illustrate what dynamic simulations can accomplish around process improvement. Figure 4 shows a mix tank
which provides the feed to a polymer process. In this tank, a solid intermediate from storage silos is dissolved in a solvent available
from a tank farm. The effluent of the tank is on flow control set by the down-stream polymer process. The flows of solids and solvent
to the mix tank must be adjusted to meet two objectives:

● Material balance control

● Feed composition

Figure 4 - Mix-Tank Control Example

(Click to enlarge)

A level transmitter and a density meter provide the input signals to the control system. Two controllers were configured as in Figure 4.
The problem with the system was that it constantly cycled, inducing a high degree of variability in the down-stream polymer process.
The only way to avoid the negative effects of the cycles was to reduce throughput, thus providing more attenuation from the holdups
in the polymer process. The mix tank had clearly become a process bottleneck. For many years it was believed that the reason for the
cycles was the interaction between the two control loops. It is easy to see that solids from the silo as well as solvent from the tank
both affect tank level as well as mix-tank density. It was not clear that one variable pairing would be better than the other. The
controllers were frequently re-tuned, with only marginal improvement.
A dynamic simulation study was undertaken to determine the best pairing of the loops and to find how the loops should be tuned. In
addition, we were hoping to determine ways of decoupling the loops if interactions were truly causing the problems. Configuring the
simulation on a HYSYS-like product took about 15 minutes. In less than half an hour with the interactive simulator, it was obvious that
the control system should not be a problem and that it ought to be easy to tune the controller for good performance. The simulation
study thus directed our attention elsewhere. The real problem turned out to be an oversized star feeder valve under the silo. When the
level in the tank increased above setpoint, the level controller was forced to stop the feeder altogether. Then when the level dropped
again, the feeder started but supplied too much solids, thus raising the level above setpoint, increasing the density and forcing excess
solvent into the tank. The cycle would never stop as long as the non-linearity of the star feeder was present. Ironically enough, the
cycles forced operations personnel to reduce throughput, which further aggravated the problem. At lower throughput the star feeder
valve was even more oversized. By replacing the star valve with a smaller "trim", the cycling was eliminated and plant throughput
increased. In less than one hour, a dynamic simulation had provided a simple answer that years of actual operation had not revealed.

This example illustrates the importance of having a dynamic simulation tool that presents very low barriers-to-use. In the past, a
dynamic simulation would not have been undertaken for a "trivial" mix tank. Nevertheless, this mix tank was the true bottleneck in the
whole polymer process.

Profile of the "Ideal" Dynamic Simulator

With an understanding of some typical applications of dynamic simulation, we can begin to picture what the "ideal" dynamic simulator
ought to look like.

Accurate. Clearly a solid engineering foundation is critical. Without accuracy, modeling is not only a waste of time, but may make
predictions that could lead to poor design decisions. Creating this solid foundation, however, is more than simply capturing the latest
process engineering knowledge and presenting it together in a simulation package. It requires a judicious balance of rigor and
performance that yields a tool that is at the same time useful and usable.

Fast. Experience has shown that much of the process understanding that comes from simulation occurs during the model-building
phase. Interaction with, and immediate feedback from the model are key elements to the effectiveness of any process modeling
endeavor. Submitting runs batch-style and waiting for results is not only inconvenient and inefficient, but it removes the valuable "live"
link between the engineer and the model. Further, for any dynamic simulation tool to be attractive to the R&D and Design
communities, it must be fast. Building a quick model, scouting some alternative flowsheets, or examining several control candidates,
must require only a minimal investment in time and effort in order to be effective here.

Broadly Applicable. Ideally, all of the functional requirements for all types of applications should be available in one place.
Conceptual design, process design, dynamic operability analysis, control strategy development, DCS interfacing and check-out,
operator training, and on-going process improvement for all kinds of processes should share a common environment. This
environment, however, needs to be more than a group of functional engineering tools artificially linked through file sharing or
swapping. By seamlessly integrating all of the functional capabilities into a single environment, information generated in one mode is
fully and immediately available for all others.

Easy to Learn and Use. In order to break the barriers that prevent its wide spread use, dynamic simulation must be both easy to
learn and to use. By using an intuitive, graphical user environment, and a comprehensive selection of configurable unit operation
modules, simulation tools can make tremendous improvements in this area. Literally thousands of man-years of process engineering
and modeling experience have been accumulated in the simulation industry. With creative ways of packaging this information
available, there are very few reasons why engineers should have to write custom code or compile input files or subroutine calls in
order to run simulations. By making simulation technology easy to use and available to all engineers via configurable modules, we are
placing the process understanding into the hands of those who are most able to put it to effective use.

The following table summarizes the general application areas and the particular features most important to each.

R&D Fast, versatile

Design Fast, accurate

Process improvement Easy to use

Training Easy to learn, user

friendly

Table 1 - Application/Feature Summary

Summary

It should be clear by now that the applications of dynamic simulation are indeed wide spread and diverse. It should also be clear that
dynamic simulation has a great deal to offer R&D, Design, Operations, and Training. The barriers to using dynamic simulation have
been steep in the past. The requirements of a unique combination skills and a very large main-frame computer limited the number of
practitioners. The cumbersome interface coupled with the need to program or compile code in order to run simulations made it
undesirable to use. Advances in computer technology and performance, coupled with new ways of packaging dynamic simulation are
eliminating these barriers.

In order to achieve the full potential of its benefits, dynamic simulation must emerge in the form of accurate, fast, and easy-to-use
tools with broad applicability. Hyprotech is responding to this need with HYSYS, its integrated system for process engineering and
control.

References

1. Fisher, W.R., and Douglas, J.M., "Analysis of Process Operability at the preliminary design stage", Comp. Chem. Eng.,, Vol. 9, No.
5, 1985, pp. 499-515.

2. Marquardt, W., "Dynamic Process Simulation-Recent Progress and Future Challenges" Proceedings of CPCIV (Editors: Arkun, Y.
and Ray, W.H.) AIChE, New York, 1991, pp. 131-180.

3. Longwell, E.J., "Dynamic Modeling for Process Control and Operability". Proceedings of the 1993 ISA International Conference,
ISA Research Triangle Park, NC.
 4. Luyben, W.L., Process Modeling, Simulation, and Control for Chemical Engineers, 2nd ed., McGraw Hill, New York, 1990.

5. Tyréus, B.D., "A Financial Project Model". Unpublished paper available upon request from ProSim, Inc. 1993.

6. Tyréus, B.D., "Control and design applied to a reactor-heat exchange system". Miami AIChE Meeting, November 1992.

7. Tyréus, B.D., "Object-Oriented Simulation" In Practical Distillation Control (Editor: Luyben, W.L.), Van Nostrand Reinhold, New
York, 1992.

8. Vogel, E.F., "An industrial perspective on dynamic flowsheet simulation". Proceedings of CPCIV (Editors: Arkun, Y. and Ray, W.H.)
AIChE, New York, 1991, pp. 181-208.

http://www.hyprotech.com/support/papers/dynam/default.asp (13 of 13)12/3/2003 10:32:17 AM