The following discussion is part of an ongoing column called Ask the Automation Pros, authored by Greg McMillan, 2010 ISA Life Achievement Award recipient, industry consultant and author of numerous process control books and articles. Submit your questions to [email protected] with “Ask the Automation Pros” in the subject line. Browse previous questions and answers here. Past Q&A videos are also available on the ISA YouTube channel; you can view the playlist here.
Erik Cornelsen’s question
What is the best way to simulate process control to ensure productive automation projects? How much effort should realistically be allocated to simulation, testing, and training? What tools are well suited for doing this effectively? How do the effort and methods differ between existing plants (e.g., a migration project) and new plants?
Julie Smith’s response
Great topic! One of my favorites.
In my experience, the simulation effort depends on two things: the process complexity and the simulation goal. Whether it’s a new or existing plant is less important.
As with any project, it’s important to define the scope early and clearly. What are you trying to simulate and why? If the purpose is to validate code, that will have a very different scope than, say, operator training or sizing the cooling system for a reactor. It will also help define the model’s fidelity. A simple tieback can be used for code validation of discrete processes, but not for an exothermic batch reactor. Your highly hazardous unit operations, like reactors and distillation columns, are going to take more effort to simulate properly, but the rewards are also greater. Match the fidelity to the process and add complexity where needed. Most chemical plants will have areas of both high and low complexity within their scope.
The fidelity will also naturally feed into tool selection. Many commercial control systems have a built-in simulation tool that is adequate for some jobs. Again, it depends on the goal. If you are trying to assess unit coordination and shared resources for a batch process, the built-in tool is probably fine. In that scenario, you are less concerned with what happens inside the vessels than with the relationships between them. But if you are trying to assess interlock adequacy or make quality predictions, then you need a higher-fidelity tool that can simulate behavior on a first-principles level within the unit operation. This will often require a third-party tool.
Make sure you allow plenty of time during the factory acceptance testing (FAT) to exercise the simulation thoroughly. Try to see how many ways you can break it. You won’t think of everything, but you will have a huge leg up on commissioning. Your operators will thank you.
Building on these points, simulation should not be viewed as a single model or a one-time project deliverable, but rather as a combination of models applied across the plant based on control objectives.
Edin Rakovic’s response
Simulation should not be viewed as a single model or a one-time project deliverable, but rather as a combination of models applied across the plant to support different control objectives. In practice, some areas of the plant can be adequately represented using simple tieback models, such as first-order plus dead-time approximations for basic control logic validation. In contrast, other areas require higher-fidelity dynamic models to effectively test more complex strategies such as cascade, ratio, and constraint control. This layered approach ensures that the modeling effort is aligned with the area where it provides the most value.
From a project execution standpoint, simulation development for virtual factory acceptance testing that is focused on control strategy validation is typically on the order of 10% to 15% of the control strategy implementation effort. The greater benefit is realized when simulation is treated not as a phase within an automation project, but as a program with its own lifecycle.
As discussed in the Prosera Perspective article, “From Project to Program: What Actually Works in Simulator Lifecycle Management,” simulation delivers the most value when it is positioned as a long-term asset that spans design, commissioning, training, and ongoing operations. This aligns with the concept of a multipurpose dynamic simulator, where the model initially supports automation design and testing, then evolves into a platform for operator training, maintenance validation, and continuous improvement.
For both migration and new plant projects, the overall strategy should be consistent. Simulation requirements should be defined early through workshops, and model fidelity should be aligned with intended use. Migration projects benefit from an established process that can be replicated with high confidence, whereas new plant projects benefit from early implementation, allowing simulation to support procedure development and management of change throughout the plant lifecycle.
Greg McMillan’s response
While a tieback simulation is often cited as sufficient for migration projects, first-principle dynamic process simulations offer so many more possibilities for better operator training, greater process understanding, and improvements in proportional-integral-derivative (PID) controller tuning and applications.
Rarely are all controllers tuned for best performance. Also, there are often advances in PID algorithm features, upgrades in instrumentation and final control elements, or changes in measurement scales or final control element capacities that will benefit from PID tuning, preferably by automated tuning software. If a PID uses derivative action and changes from original series to newer standard form or in the extreme case where a PID working in engineering units (in some programmable logic controllers [PLCs]) versus percentage signals (in distributed control systems [DCSs]) or vice versa, you must retune the PID controllers.
However, you need to include all the possible sources of dead time, such as mixing delays, transportation delays, and instrumentation lags and delays. You also need valve installed flow characteristics, and instrumentation resolution and rangeability. To test procedure automation for
startups and batch automation, your simulation must be able to handle empty or low liquid volumes.
The opportunities opened by dynamic first-principle simulations justify the investment discussed in my 2026 Control article “Digital Twins Key to Prosperous Process Control.”
If you are studying a few loops, you can use software to identify the open-loop dynamics and put them into a focused simulation consisting of a total loop dead time and an open-loop time constant and self-regulating or integrating gain that includes the effects of measurement instruments and final control elements.
