Greg: If I had one wish for the New Year, it would be that process control engineers use digital twin and virtual plant technology to improve process performance by using more advanced process control (APC). Nearly all the knowledge offered in articles and books written by process control legends such as Greg Shinskey and William Luyben was developed and detailed by dynamic simulations.
This is also true for my writing, as well and more recent works by Peter Morgan. We documented knowledge largely gained through simulation in ISA Technical Report ISA-TR5.9-2023, “PID Algorithms and Performance,” for which we received the 2023 ISA Standards Achievement Award.
I’ve written several articles and columns on the synergy of modeling and control. The one that stands out in my mind is the Control Talk column featuring Julie Smith, global automation and process control technology leader at DuPont, and titled “Dynamic world of modeling and control.” For my latest perspective, see the article, “Digital twins for the virtual plant” in ISA’s InTech publication. The simulation value for the Life Sciences industry is discussed in the ISA blog Q&A: “How is Simulation Leveraged for Qualification & Management of Change?”
Consider simulation to also promote the value of the automation profession, despite the challenges of work overload and project schedules. Seeing a dynamic simulation, with process metrics showing an increase in process capacity, efficiency, and reliability by APC, can get everyone on the same page, including research scientists, process and mechanical engineers, data scientists, technicians and operators. You must spend some personal time getting started. That’s what I did when I developed my first centrifugal compressor surge model, opening the door to a wider future that’s viewable in more than 400 articles, blog posts and columns, as well as in the 30 books I wrote during the past 40 years.
To further open our minds to the opportunities granted by using models to improve process performance with effective use of APC, we conclude our series on modeling and control with a conversation with José María Ferrer, who has more than 25 years of experience in dynamic simulation and control of hydrocarbon processes, and presently serves as a senior advisor at Inprocess Technology & Consulting Group.
José, how can models help in prototyping APC?
José: Both types of models (steady-state and dynamic) have value to prototype APC controllers. Steady-state models can calculate deep process gains. Dynamic models can calculate transient process responses for all independent variables, including the perturbation variables that can’t move freely in a real plant during step tests. The most important advantage is being able to calculate all process responses for all potential states of the plant, so you can analyze all non-linearities of the process. Plant tests are minimized or eliminated as a result, and a better APC can be designed to consider process non-linearities. There are multiple references for using simulation models in APC project over the past 20 years, but it’s worth mentioning one of the first APC implementations in Lysekil (Sweden), where no step tests were needed. Read about it at “Step-test free APC implementation using dynamic simulation.”
Some 15 years ago, I analyzed the myths and barriers for not having a general simulation usage between the APC engineers in a control seminar titled, “Utilizing rigorous simulation in advanced process control projects.” Now, I don´t see APC engineers as simulation model builders but as simulation model users. Building a good simulation model, if feasible, might take some time, experienced simulation engineers and the right tools. We see more use of multi-purpose dynamic simulators (MPDS), which is also called Lifecycle-OTS (operator training system), and the real-time simulator (RTS), also called digital twins. The availability of a good dynamic model is more frequent today than in the past, and APC engineers can make use of them for their projects. But remember to follow 10 rules when you request MPDS, which can be found at “10 best practices to request and exploit Lifecycle OTS v2021.”
Greg: Simulations provide details about what measurements are needed to detect equipment and piping problems. Wireless measurements can be moved as needed to provide the online data necessary to proactively deal with them using state-based control. How can models provide equipment anomaly detection?
José: All equipment manufacturers supply equipment designed to work under certain conditions (flow, temperature, density, composition, for example). Once the plant is commissioned and operating, the equipment will suffer all kind of variations to those design conditions, and their performance will deteriorate over time. RTS simulation models (models running second-by-second with real-time data) can determine the current performance of the equipment, and generate an alarm when things are not working as expected. The model is not only useful for detecting the anomaly, but it can also trigger additional simulation models to diagnose the most likely cause of the anomaly. Every piece of equipment running in a plant must comply with all the thermodynamic laws and can’t escape them. If they do, then the instrumentation is lying.
Greg: I’ve extensively used simulation to confirm strategies needed to automate startup of compressors, neutralizers and reactors, and used open-loop backups to prevent surge and Resource Conservation and Recovery Act (RCRA) violations. How can models be used to provide the best procedure automation and state-based control?
José: Some 29 years ago, I started my career as process control engineer at Dow Chemical, so I was well educated on programming DOWTRAN with the MOD5 computers. Everything was state-based. The states were called STEPs and all analog outputs, digital outputs, alarms, etc., were calculated based on the STEPs. Many manual start-up or shutdown procedures were converted to automated procedures, where parameters and code must be checked out. At that time, DOWTRAN provided the AISIM and DISIM functions to simulate all the instrumentation inputs from the plant, and test all the code before uploading it to the running plant. The simulation code had to be programmed by the control engineer, and the simulation was used to verify all state-based automated procedures. There are more details about it in this in this document, “Optimize DCS Testing and Operator Training.”
I remember another project on the north coast of Spain, where they had to do a large DCS migration project. The lead control engineer at the plant was an ex-Dow engineer and automated the startup and shutdown sequences of large distillation columns, which previously were done manually and with the taste of each operator shift. We were requested to develop dynamic process models of the distillation systems and conceptually test all new control strategies. Later, we reused those models to test all the migrated DCS code with all the new sequences and train the operators before startup. It was a clear example of how models can help verify automated sequences and DCS code, which can also be found at “Digital twin: three birds, one stone.”
Greg: Please make the personal time investment to use simulations to elevate your role in process control.
