Hector Torres’ question
Process control migration best path forward – duplicate what was there or design for the future? How do you achieve both when schedule and cost are constraints or better yet roadblocks?
Answer from Greg McMillan
I think you should do an opportunity assessment (OA) for process control improvements before the project definition. A typical OA takes one to two days. Operations, maintenance, process, and automation engineers and technician should all attend. The ISA book, Advanced Control Unleashed: Plant Performance Management for Optimum Benefit offers the OA approach in each chapter and additional OA questions in an appendix. The OA is a great time to get the additional measurements you need. Even if you don’t have time or info to justify the purchase of the field instrumentation, if you can provide the process connections, you can enable a lot of future opportunities. I would look for opportunities for flow, pressure, temperature, and analyzer sample connections. I will send some ideas on portable measurements and analyzers that could be temporarily connected to explore and document improvements and benefits. Again Hunter Vegas can offer practical guidance.
Hunter Vegas, who has just graciously agreed to be part of our Mentor program, is an excellent resource for these and other project questions. I did a three-part series of Control Talk columns: Successful Retrofit and Automation Projects.
Answers from Hunter Vegas
Hector Torres has asked yet another “million dollar” hard to answer question but I will try to provide some practical suggestions for addressing his issue.
I actually gave a talk at the last ISA Automation Week conference on a topic that was related to this. I’ve attached the “ISA version” of my slide deck Managing Successful Automation Retrofit Projects. The actual version I presented had much more entertaining slides to keep everyone awake but this version provides the “meat” of the talk using those painful ISA bullet point/small font slides that we all have come to know and love. The gist of the talk was that MOST people start their control system migration journey by picking the control system and then trying to make it work. This talk suggested that you do the reverse and consider exactly WHAT the system needs to do and what the constraints are for cut over and THEN pick a system that will satisfy those constraints. This dramatically improves your opportunity for success. The talk also covers a wide range of “gotchas” that can burn you badly if you don’t consider them very carefully during the design process.
There are three main types of control system migration projects… they are:
An in-kind replacement
In this type of project is usually a migration of the control system to a later revision. Most of the conversion will be automated and there will be no (or very little) migration of the wiring. This situation is usually straightforward so I won’t address it now.
Start from scratch
You either have NO process control system or the process is changing so completely that you are essentially starting over. However you asked about control system migrations so we’ll pass on this topic right now as well.
Mostly the same but different
This is the project you are asking about. There is no “easy” migration path forward for the current control system so you are going to have to replace it. Since management is spending the money they want “improvements” (usually very poorly defined). You probably have very tight budget and cut over time constraints, and the operators want it to “work exactly the same except they want it to work better” – (which will also be very poorly defined and will change depending on which shift or operator you ask).
Greg’s advice for an OA is a good one – as you may be able to define opportunities for improvement that will justify some additional capital spending and generate management support for the project. You can also point to inability to get spare parts, limited capabilities of the current system, etc. etc. as a means to get the capital going…but that is a whole OTHER topic that I will try to avoid as well.
So let’s get back to your question – Once you have money and have decided on a control system, how do you migrate over the software? Well that depends a bit on the process you have….
If you have a continuous process then you probably WILL be converting the bulk of it directly into the new system with little change. However you can look to add several things to improve the plant and justify the project. You might consider:
- The control improvements Greg discussed – (Better control around the columns, feed forward control, advanced multivariable control, etc.) All of this can usually be “laid over” the existing controllers even as the system is running.
- Automated start-up/shut down/product change sequences – Many plants have taken to automating these sequences to shorten transition time and improve safety. Again this is something that can usually be “laid over” the existing controllers as the system is running. It will usually require a lot of off-line simulator testing to make sure everything is right but most newer systems have that capability so testing is easy to do.
If your current control system is NOT programmed in an ISA88 fashion then I would say, start over. The advantages of a well-designed ISA88 configuration are so extensive that the cost of the additional engineers is easily recouped in a short time. Note that I did say “WELL-DESIGNED ISA88 configuration.” A poorly designed ISA88 system can create more problems than it solves. Ironically I gave an ISA talk on tips and tricks for applying ISA88 in chemical plants. It provides a lot of suggestions on things you might consider if you are putting in chemical plant batch system (or really any batch control system). (Editor’s Note: “S88” in the slide deck is a shorthand reference to the ISA batch control standard, ISA88.)
Regardless of what process you have, you will always want to keep an eye on the future. If you are putting in anything include spare capacity (network switches, field junction boxes, marshalling panels, I/O cards, sizing power supplies, whatever.) You don’t have to install the spare capacity, just design for it so that you can add extra controllers, I/O Card racks, operator stations, etc in the future. If you do this, you’ll likely save yourself a tremendous amount of money (and effort) down the road and will be able to add future improvements relatively cheaply.