"Ask the Experts" is moderated by Béla Lipták, process control consultant and editor of the Instrument Engineer's Handbook (IEH). The 4th edition of Volume 3, Process Software and Networks, is in progress. If you are qualified to contribute to this volume, or if you are qualified to answer questions in this column or want to ask a question, write to firstname.lastname@example.org.
Q: In the past, you've written a lot about the need for a larger role and the added recognition that our process control profession deserves. You also talked about applying the principles of process control to other processes like the economy, the control of rivers, or the global energy future. You wrote about the role universities, publishers, technical societies and conferences could play in gaining that recognition, but to my knowledge you never talked about the steps we could/should take to get there. Why not?
A: The answer is simple George—I did not, because I don't know. It is much easier to identify problems than to give a recipe for fixing them! It is easier to say that process control can increase safety by monitoring the methane concentration in mines, designing safe software for acceleration control in cars, designing deep-sea oil wellhead shutoff valves (the same as we design all critical safety shutoff valves) by partially stroking them periodically to check out the functioning of the loop, or guarding the cooling systems of nuclear power plants, which could be the next logical targets of the terrorists.
Yes, these problems are easy to solve because, as presetned so far, they all involve only engineering. What's much more difficult is to spread the news that there is a new profession that can solve these problems and do more. Accepting and understanding this takes time. It also takes time to convert our engineering societies, conferences and publications from being adverising forums to becoming the wells of new knowledge, or to take back process control from the "dot com types" and return it to people who understand the processes they control. It might take a lot more accidents at our 40-year-old refineries, aging nuclear power plants and even older mines, before people realize that it is our profession that can not only fix them, but can also meet the challanges of the coming post-industrial age.
Q: I would like to ask about the preventive maintenance (PM) of SIS valves in our plant. Our shutdown valves have no bypass and therefore, scheduled maintenance requires a plant shutdown. Yet we have no plans to install bypass systems due to their high cost.
As a result, the operation and maintenance departments asked to move these SIS valves from the preventive maintenance category, and move them into the shutdown maintenance category to have 100% compliance in PM rather than deferring PM of SIS valves to shutdown.
I would like to have your opinion on this issue. Is it acceptable if these SIS valves will receive their scheduled maintenance during shutdown? Am I right that safety measures will not be affected, since they should be done in shutdown only anyway?
Fahad S Al-Howimil
AThese questions came up during the evolution of ISA SP84 standard, starting in 1982. As you know, this standard, co-adopted with the IEC 61511, is a qualitative approach to availability numbers for SIF. That means, the transmitter, the logic solver and the final control element (S/D valve) will function when called on to shut down the process unit safely. The availability calculation for the Function-SIF takes into account components, design, completeness of testing and the frequency to meet SIL level requirements. There is a Technical Information (TI) as part of ISA standard SP-84-2005 with details on how to test and maintain these S/D valves per standard. In general, these valves account for 45% of the availability number. In other words, your deficiency in design and testing frequency will increase the risk level manyfold. You have three choices:
- Declare that these systems are not safety instrumented systems or functions. This will have non-compliant regulatory or insurance-related consequences
- Talk to the valve manufacturer about how to add test features, so that you can test them periodically (bump test for 8% to 10% movement) to ensure its functioning. You do not need bypass valves to do this. There are several prepackaged test kits, such as ASCO and others, on the market to accomplish this.
- Set up your DCS or asset management system to do partial-stroke testing online to validate the valve function. This is commonly done in systems similar to yours where valves are considered critical, but not available for testing on a periodic basis. This will require some additional components to be added to the existing valve and software setup to collect data.
If it is any comfort, the ISA SP 91 committee is reviewing a draft addressing the situation where the valves are not SIF, but are classified as critical S/D valves. In that case, testing, training and documentation shall be followed similar to SIF, but not rigorously quantitatively. This will enable operating units to meet basic safety needs to own and operate, which will satisfy regulatory and insurance needs.
Please check with ISA web page (www.isa.org) for some papers on this topic. I recall an engineer in your company who gave a paper on this very topic eight years ago in one of the ISA conferences. Check your company archives, too.
A: The short answer to your question is yes, you can move the SIS valve testing to the shutdown maintenance list, but at a cost.
SIL availability calculations are performed to determine how frequently a shutdown device will be called on to act, but fail to do so. The higher the SIL rating, the lower the likelihood of failure on demand. These availability calculations are based on the dangerous undetected failure rate of the device and the frequency of testing.
You can lengthen the testing frequency (as you suggest), but the likelihood of the valves failing to close on demand will be increased. The question you have to answer is this: Will these valves still satisfy their SIL requirement under the new testing schedule?
People have devised ways to get around this issue with some success. One option is partial-stroke testing. This operation is performed with the use of solenoids and limit switches, which allow you to partially close the valve while in operation. This proves out the SIL control system, the wiring to the valve and the solenoid, but it doesn't really guarantee that the valve will fully close and seal upon demand. Still, it does check the functioning of a large part of the loop, and that may be enough to satisfy the SIL availability requirement.
A: The interval between testing valves depends on the performance required. You did not state the original and proposed test intervals, and so your question can't be answered qualitatively (other than saying that the "safety measures" definitely will be affected by such a change).
Was a calculation done to justify the change? What recommendations (if any) might the valve manufacturer have for test intervals? Automated partial stroking is often an effective means to extend the required manual full-test intervals. There are over a dozen manufacturers of the systems required to implement such solutions. Note that failure rates can change depending on frequency of valve movement and testing. The standards, technical reports and many books describe and show how to perform such calculations; it is nothing more than simple algebra.
Paul Gruhn, PE, CFSE,
Q: I am working in a power plant in Kosovo. As there is no possibility here of finding good literature on the control of processes I am writing to you. Can you provide me with a mathematical model description for water treatment plants—especially for the part of a water treatment plant were I will try to implement predictive control.
A: In process control, two general types of models are used—first-principle models and black-box models developed from the operational data. For linear, model-predictive control, black-box, parametric (ARX, ARMA, state space) or non-parametric models (step responses) are applied. The typical procedure for building a model is applying pseudo- random excitations, collecting data and generating and validating the model
Most MPC product providers have tools for building models. You also can use Matlab Process Identification Toolbox.
You also can read more on developing models at www.easydeltav.com/BOL/10.3/index.html
Also, a practical general overview on building models and model application for tuning and MPC control can be found in Instrument Engineers' Handbook (IEH), Vol. 2, Chapters 2.13 to 2.17. You can also read in the same volume in Chapter 8.39 about wastewater treatment controls.