Virtual Reality Saves Money!

For a long time the resources and effort required to create complex simulations of manufacturing and process operations was avoided for all but the toughest problems, and when they were constructed, it was often a last resort. "Simulations? We can't afford that fancy stuff. That's just for the 'big boys,'" was heard for years throughout operational units around the world, but not any more.

By Dave Harrold, CONTROL ENGINEERING October 1, 2000

KEYWORDS

Simulation

Design/analysis

Education/training

Quality assurance

Safety

Testing

Sidebars: Simulation complexities Justifying simulator benefits

For a long time the resources and effort required to create complex simulations of manufacturing and process operations was avoided for all but the toughest problems, and when they were constructed, it was often a last resort.

“Simulations? We can’t afford that fancy stuff. That’s just for the ‘big boys,'” was heard for years throughout operational units around the world, but not any more. More and more companies are learning, often the hard way, that investments in simulations can save time and money.

Simulations are really imitations that take on the appearance, form, or sound of something that is, or could be real. Often the word simulation is combined with the word models to form “simulation models” and is meant to imply a high level of accuracy, but isn’t well quantified leaving the buyer/user in somewhat of a doubt as to what to expect.

Other terms frequently used in conjunction with simulation include high fidelity, dynamic, and steady state. Alpha Sim Technology’s (Houston, Tex.) definition for high fidelity, dynamic simulation is: “A computer-based, mathematical model of the unit, process, and/or entire plant. It is developed according to engineering first principles. It is model software, continuously solving the differential equations of state, running in real-time. The occurrence [deviation from target] is 0.5% under steady- state conditions and 2% under dynamic or upset conditions; and is a virtual process reality.”

It’s worthy to restate that hard values validate the completeness, accuracy, and performance of simulations, increasingly used for design, debug, analysis, and training.

Design simulations

Use of simulations to assist in the design of “stuff” has blossomed since the proliferation of desktop personal computers (PCs), windows user environment, and modular software development techniques.

An area often overlooked as a simulation candidate is business planning. Once business strategies for becoming more flexible and agile move from the handwaving and overhead projection arena into the “how are we going to do that?” arena, entire business processes will need to be designed, analyzed, and reengineered. Business process simulation software, such as Micrografx’s (Allen, Tex.) iGrafx Process, allows conducting “what-if” scenarios on business processes without actually disrupting day-to-day operations.

A few years ago, no one in his or her right mind would have attempted to simulate an entire production facility before it even existed. But that was yesterday, and this is now. Using simulation software tools from Tecnomatix (Herzliya, Israel), Siemens AG’s (Frankfurt, Germany) project manager Reinhold Ufrecht didn’t hesitate to create a virtual shop floor and then simulate and analyze the operational usability prior to actual construction.

Commencing with a conceptual design of each processing area, and adding production workflow simulation, Mr. Ufrecht was able to test, analyze, and refine the design until he was completely satisfied machines were properly spaced and production workflow was optimized. “We are under heavy pressure regarding time and cost. Leveraging the advantages of factory simulations, we are certain that our production concepts are right from the beginning,” says Mr. Ufrecht.

The use of design simulation is not limited to Germany; it’s being used all over the world. For example, Honeywell’s (Phoenix, Ariz.) Hi-Spec Solutions group regularly uses simulations to validate the operational design for Australian based Laminaria. One simulation involved the design of a debutanizer column scheduled to be located in a remote area of Indonesia. Thanks to the accuracy of the simulation, the control system software was well debugged when the startup engineer arrived for what was planned to be the first of three trips. Fourteen days later, the entire unit was operational and the startup engineer went home, for good.

In the United States, design simulation software from MSC.Software (Los Angeles, Calif.) is recognized by the Federal Aviation Agency as an accepted standard for design and analysis of stress, vibration, heat-transfer, acoustics, and aeroelasticity for airframe manufacturers seeking design certification.

And in Canada, Magnola Metallurgy (Danville, Quebec) uses simulation software to determine chemical reaction kinetics, physical property relationships, and other operational data required to design, build, and operate a serpentine mineral transformation process to produce magnesium. Once the design is finalized, the simulation adopts a new role when it is connected to the Honeywell control system where the simulation assists with control system software debug and operator training.

Robustness of the underlying tools is an important requirement of any software claiming it saves time. That’s especially true of simulation software used as part of the design process. In addition to ease of use, and of course price, buyer/users should pay special attention to the availability, appropriateness, and completeness of application-specific library modules.

For example, consider an assignment to design an anti-lock braking system. The software tools should provide advanced algorithms for designing and analyzing multivariable controls that contain uncertainty caused by unmodeled, nonlinear dynamics or modeling errors. Robust Control Toolbox is one of several “off-the-shelf” modular libraries available from MathWorks (Natick, Mass) designed to help solve specific design and analysis problems such as anti-lock braking systems.

If the design assignment is in the chemical arena, seek simulation tools with complete and robust thermodynamic, transport, and physical property databases as found in products from Aspen Technology (Cambridge, Mass.), Simulation Sciences (Brea, Calif.), Fluent (Lebanon, N.H.), AEA Technology (Calgary, Alberta, Canada), and Simons Technology (Atlanta, Ga.).

Simulation models used during the design process are also proving beneficial when used to test and debug the final product or solution.

Simulation aids debugging

Often, simulation investments are justified as training aids, but as Chevron’s Tim Wilmarth says, “The real value of simulation proved to be our ability to identify and fix design flaws. We estimate our ability to catch design problems early saved us one week in a three-month startup schedule on a single project.”

Mr. Wilmarth didn’t put a dollar value on a week’s worth of production, but if you’re producing 10,000 gallons, pounds, or widgets each day for seven days, and they sell for $1 per unit, it’s easy to document a $70,000 addition to the bottom line, thanks to a pre-startup debug using simulations.

Bill Mackin, senior project engineer at Armstrong World Industries (South Gate, Calif.) is another evangelist of the benefits simulation provide in debugging control and automation system software. Using SST’s (Waterloo, Ontario, Canada) PICS Simulation software to mimic a control systems I/O subsystem, Mr. Mackin routinely debugs batch control systems throughout Armstrong plants. “Over an 18-month period we replaced relay logic with seven PLCs [programmable logic controllers]. Prior to using PICS, we wired toggle switches, pilot lights, thumbwheel switches, potentiometers, and summation boxes representing our load cells to the PLC I/O system. It was like musical chairs, with engineers toggling switches and turning pots. With PICS we could concentrate on debugging the control software,” says Mr. Mackin. (See Application Update, in this issue.)

Like Chevron and Armstrong, Southwestern Public Service (SPC, Amarillo, Tex.) successfully used time-lagged I/O tieback simulations to debug portions of control systems in the past, but they also recognized the limitations tieback simulations had when testing complex interlocks and control strategy interactions.

Facing an unacceptable 16-week outage to retrofit existing boilers with new controls, SPC’s management searched for time saving alternatives. Part of the timesaving came in the form of a high fidelity, dynamic simulation from Esscor (Solana Beach, Calif.). Once the Esscor simulation was connected to the Foxboro Intelligent Automation control system, all aspects of the control logic and graphics could be tested without “faking-out” or forcing inputs and outputs. With the help of the Esscor simulation, SPC shaved eight weeks off the outage schedule.

Simulation software has been developed to meet a variety of debugging needs. For example, where time-lagged I/O tiebacks and low- to medium-fidelity simulations are suitable, Cape Software’s (Houston, Tex.) VPLink, SST’s PICS, and Munger’s (St. Louis, Mo.) Mimic software may fill the bill. (See Simulation complexities sidebar.) High-fidelity simulations may best be developed using products such as AEA Technology’s Hysys or Hysim; Simons Technologies Ideas; Simulation Sciences Dynsim, Hextran, or Netopt; Aspen Technology’s Aspen Plus or Aspen Dynamics; or Fluent’s Computational Fluid Dynamics.

If simulations, especially high-fidelity ones, can return high benefits such as experienced by Southwestern Public Service’s eight-week outage reduction, imagine what those same simulations might produce if used daily. That’s exactly what most simulation software packages offer today—the ability to become high-fidelity, dynamic simulations that run along-side the real process, often using real process inputs to predict where key dynamic performance indicators (such as quality, throughput, and cost) will be 15, 30, or 60 minutes from now.

Simulations for analysis

Simulations have been around for over 40 years, mostly as training aids. In the past few years, the benefits provided by simulated “what-if” analyses have begun to receive deserved attention and respect.

For example, Washington Group International (Boise, Id., formerly Philadelphia based Raytheon Engineers & Constructors), found steady-state thermal oxidizer simulations didn’t accurately demonstrate the affects of changing compositions and waste flow rates. Therefore, control strategies were unable to respond quickly enough to these changes and the thermal oxidizer had to be bypassed, causing temporary regulatory compliance violations.

By converting the steady-state simulation to a dynamic simulation, Washington Group engineers were able to observe vent gases changing from zero to maximum flow (and back to zero) in less than one minute. Seeing and analyzing this dynamic behavior helped in modifying field instrumentation and quickening control strategy response. Those, in turn, keep the thermal oxidizer in regulatory compliance throughout the operating range. Though Washington Group isn’t saying, avoiding paperwork and possible fines associated with regulatory compliance violations must have paid for simulation development, analysis, and all necessary changes.

International Paper (Purchase, N.Y.) is also a believer in using simulations to seek improvements and claims it has identified capital effectiveness improvements from its simulation system in the order of $100 million annually.

A distillation column at Sunoco’s Sarnia (Ontario, Canada) facility experienced unexpected pressure spikes that resulted in unstable operations for several hours following the spike. Working with Honeywell Hi-Spec Solutions, a dynamic simulation of the column, overhead accumulator, and associated piping was created. Analysis of the dynamic simulation revealed that as liquids accumulated in a piping trap the pressure gradually rose. Eventually liquid collection blocked the vapor flow, creating a pressure spike that dropped rapidly once the liquid accumulation was displaced. As a result, equipment modifications were made to eliminate the liquid trap in the piping.

Facing changes to operational philosophy’s in the deregulated power industry, a large commercial power producer needed to economically swing a boiler’s loading from 600 MW per hour to 100 MW per hour or less, still meet all EPA requirements, and not damage the boiler. At a minimum, the producer understood new control system tuning and alarm parameters would be required to maintain boiler control at the lesser firing rate, but what other changes and considerations would be required for things like burners, pumps, fans, control valves, etc.?

Using a high-fidelity, dynamic simulation model from Esscor—with capability to adjust burners, resize fans, pumps, valves, etc. on-the-fly—the producer identified the required physical and operational changes to quickly change boiler load. Once the necessary physical and operational requirements were understood, the real boiler was retrofitted and the control system modified to incorporate a variety of operational states. In fact, thanks to simulation modeling and what-if analysis, that 600 MW per hour boiler can now operate as low as 55 MW per hour efficiently and economically.

Simulation for training

In 1995 South Carolina Electric & Gas’s (SCE&G, Cope, S.C.) management faced bringing a new generating station on-line, with boiler design (and control strategies) new to SCE&G, using new operators.

According to SCE&G simulation manager Glenn Westberry, “Management’s decision to install a high fidelity, dynamic operator simulation unit early in the project’s life has paid countless dividends. Using the simulator, we identified several logic problems in the control configuration that would have caused unexpected unit trips during startup. Since startup we regularly use the simulator to conduct normal and abnormal operational training exercises for operators, and we have found it to be the best place for engineers, technicians, and operators to work out ‘what-if’ improvement scenarios without jeopardizing on-line operations. Our operators and technicians are always asking when they can get more simulator time.”

When Phillips 66 (Borger, Tex.) decided to build its first Methyl Mercaptan (the “stink” added to natural gas so you can smell a leak) production facility, a training simulator was included as part of the project. The investment paid big dividends. A startup, predicted to take 200 to 350 hours, was complete in just 66 hours.

Surely not all that savings can be attributed to the process-training simulator, can it?

According to Phillips employee David French, maybe not 100% of the savings could be directly contributed to the high fidelity, dynamic training simulation, but much of it was. Because of the accuracy of the simulation, extended operator simulator time, and many successful simulator startups, the operators learned which areas of the process required special attention as startup progressed. Without all those simulator hours and the confidence and competence they create, Mr. French is confident the startup would have taken much longer and have been more difficult.

There’s the tangible benefit high-fidelity simulation training produces, the confidence and competence operators obtain in their own ability to routinely handle normal and abnormal situations. When operators can do that, the number of unplanned shutdowns and subsequent results will reduce, often dramatically.

Wouldn’t it be a more pleasant working environment if, like SCE&G’s Mr. Westberry, you too had operators and technicians requesting more training opportunities? It’s possible! When training is meaningful, interesting, timely, and when management gets involved in the outcome, everyone wins, and simulated training can play a key role in making training meaningful, interesting, and timely. (See CE , Aug. ’00, p. 63).

As one trainee said, “There were times when the simulator was so close to the real plant, I forgot I was running a simulation.”

High fidelity, dynamic simulations are routinely used by military and commercial airline pilots to ensure pilot competence and confidence in handling normal and abnormal situations.

We put men on the moon, and brought them safely home using simulations for design, debugging, analysis, and training.

Visibility may not be as high as that in our world, but simulations still can play a major role in safe, efficient automation and control.

Simulation complexities

Simulations are developed to meet a variety of needs and range from the simple and straightforward to highly complex models. But without agreement among all parties, expectations of capability can be missed with disappointing results.

Alpha Sim Technologies’ (Houston, Tex.) definition for a high-fidelity, dynamic simulation is: “A computer-based, mathematical model of the unit, process, and/or entire plant developed according to engineering first principles. It is model software, continuously solving the differential equations of state, running in real-time. The occurrence [deviation from target] is 0.5% under steady state conditions and 2% under dynamic or upset conditions; and is a virtual process reality.” That defines what to expect in performance, but what about simulation software capabilities? What should be expected, what is required to meet specific requirements?

No standard definition exists, but personal experience, conversation with folks in the business, and reading literature from several simulation software suppliers, seems to make the following guidelines appropriate for identifying low-, medium-, and high-fidelity simulation software capabilities.

Low-fidelity simulation software provides time-lagged input and output tiebacks.

Medium-fidelity simulation software includes functions such as process deadtime, first-order lag, accumulator, full function calculator, preconfigured algorithms for heat exchangers, jacketed tanks, and easy connectivity to popular spreadsheet and high-fidelity simulation software packages.

High-fidelity simulation software includes libraries of physical plant equipment; databases of thermodynamic, transport, and physical property of chemicals and organics; and advanced algorithms capable of determining complex liquid and vapor mixtures.

To avoid disappointment later, define the purpose and required fidelity of the simulation before contracting to have a simulation constructed. One way to ensure you get what you want is to have a basic understanding of the capabilities of the software that will be used.

Justifying simulator benefits

Justifying a simulators’ benefits for training is a difficult sell without hard evidence and quantified numbers. To gain information, the power industry commissioned the Electric Power Research Institute (Palo Alto, Calif.) to oversee a study related to simulator use. The 1993 report concluded that 20% to 30% of forced plant outages were the direct result of operator or maintenance errors and that installing and using a simulator to train operators could be associated with an average yearly savings of $4,532 per installed megawatt. Thus a 1,000 MW production facility could expect a whopping $4,532,000 of value each year. Assuming a simulator cost $1 million, the payback period is less than three months.

Few industries have the means or clout to commission such reports, so users are left to figure it out for themselves. In conducting your own study, evaluation should consider at least the past year’s performance with an examination of each unplanned shutdown, or near miss, quantifying the effects of:

Lost time injuries or deaths;

Lost production (units per day, selling price per unit, number of lost days);

Damaged equipment (cost to repair with spare or new parts plus downtime);

Startup and restart costs;

Wasted feedstock costs;

Wasted energy costs (steam, electricity, chilled water, etc.);

Regulatory compliance costs (paperwork, reporting, fines, public relations); and

Contractual cost due to missed delivery deadlines.

Some of the events are difficult to quantify with hard numbers, but in this case, best guess will get the job done. Once you have the numbers apply them to the formula [ PBP = (S – d) / C ] where PBP = Pay Back Period, S = Simulator price, d = Damaged equipment cost, and C = Cost of lost production.

As an example, suppose the simulator cost (S) is $500,000, damaged equipment cost (d) is $200,000, and lost production (C) cost $50,000 per day. Then, PBP = ($500,000 – $200,000) / $50,000 = 6 days to recover the simulator cost. (Note the repair or replacement cost is subtracted directly from the simulator price under the assumption that simulator training would have eliminated the need to repair or replace the equipment.)

But no one really wants to wait until they have an unplanned shutdown with $200,000 in damage to gather hard numbers to justify a training simulator.

Perhaps an alternative justification method is to take to heart the success stories of those who have gone before.


Related Resources