People and Processes

Brian Nessen:  “You refer to a champion in the plant. Who would be the best person or title suited for this role? And, can you expand on the importance of that role with regards to your presentation?”

Bill Alkema:  “It’s important to have someone that is intimately familiar. At Desert Basin, we used an operations supervisor because that person has the intimate knowledge of the on-goings in the plant. He has intimate knowledge of all the systems in the plant, and he’s got a vested interest, obviously, to make sure that all these tools a are being used effectively. Also, he stays current with all the round-the-clock issues in the plant.”

Bill Nieman:  “When you’re interacting with SmartSignal, how do you typically interact back-and-forth with them?”

Moh Saleh:  “We either trade emails or voice calls, because no plant remains static. And so when things change, SmartSignal detects the change. Over the weekend, the operator might notice it. During the business day, the SmartSignal EPI*Center specialist [WatchList Analyst] may detect it. …If it’s a low priority, an email would be traded. Other times, someone would call. And, once a week we hold a conference call with the SmartSignal specialist, and we  go over all the issues that have been identified and look at resolutions for previous ones.”

Bill Alkema:  “I’ll add to that. We also encourage our operators to log in and look at the WatchList on SmartSignal. When they look at the WatchList, they can then do some pre-investigation if it’s during the night and they need something to do.  It’s a good time for them to go in and start investigating why a certain parameter is abnormal.  They might go back and look into the logs and see that it was replaced yesterday during the daytime. So, there are a lot of things that come up on the WatchList that we can resolve quite quickly. You’re looking for those ones that suddenly don’t get resolved and those operators actually can write work-order forms and make sure they [work orders] get taken care of first thing in the morning.”

 Bill Niemen:  “Operators are typically used to responding  to alerts on the control panel and responding to things instantaneously. It’s kind of unusual for plants to put the SmartSignal information right in front of their operators. Why did you guys decide to do that?”

Bill Alkema:  “Well…they HAVE time to do it. It gives them an interest in it. At night, even if we’re running through the night and things are stable, we’re usually going into the evening, we’re at base load, things are pretty quiet, and they have the time to do it. It just keeps them interested in it. Granted, some operators are better at it than others or more interested.”

Moh Saleh: “As I mentioned during the presentation, we had a perfect case.  It was Sunday, nice quiet day, units just plugging away, and the operator was just looking through the screens and noticed… ’Wait a second, how come I’m getting an alarm? I don’t remember seeing it earlier.’ And he starts drilling down and he realizes ‘oh, something really IS wrong somewhere.’”
Bill Alkema:  “He was showing no NOX.”

Moh Saleh: “Right! I mean from the control system’s perspective, it was happy. You know, zero NOX is better than 3 PPM! So deviations are detected much earlier than the control system alarm levels.”

Bill Nieman:  “I often use the term ‘Multiple Converging Lines of Evidence.’  As I listen to you talk, what I hear is that you’ve got a staff that is pretty competent to be able to use this full suite of tools effectively.  Could you say anything about the kind of training you give them? And how long it takes to ‘come up to speed’ in these systems?”

Moh Saleh:  “I think more than just staff, it’s the interest of the staff infused in these tools. The majority of these tools, unfortunately, what happens is – basic human nature. If it’s very complicated, you don’t use it. I mean, it’s very hard to keep people utilizing it, and all the tools we have at Desert Basin are, I can say, very easy to utilize. I don’t use the DCS historian because the biggest reason is it’s complicated for me. The other tools were fairly intuitive.  And so I use them…”

Moh Saleh:  “There is a question out there about ‘Can you access these systems from offsite stations such as from home?’”

Moh Saleh:  “Unfortunately, the equipment doesn’t fail when you want it to fail, so a lot of these issues come up during weekends, middle of the night, and we ‘map in’ to our work computer, and we use all these systems from home.”

Bill Alkema:  “But we go through our desktops.”

Moh Saleh:  “We go through our desktops. That is correct. For all of them.”

 Return to categories

Financial Returns

Bill Nieman:  “…You mentioned early on that it’s important to build value and establish value for the champion and the management. How is it that you track the ROI of the various systems and roll that up in a way that your management can appreciate the value you’re getting from online monitoring?”

Bill Alkema:  “Bill, I can’t claim that we track it formally in terms of the ROI. We know what we’re paying for the systems, and we know when we make a “catch.” You know-- when a full load trip costs you a quarter-million dollars, and you’ve actually avoided one. With experience, we’ve known that some of these problems we’ve found might have led to a trip if we hadn’t caught them. If we had let them go until the alarm system/control system caught them that might have been too late.  I can’t say it’s a formal method of ROI evaluation, but we know we’ve gotten payback, and management understands this.”

Moh Saleh: “From the engineering perspective, I’ve gotten involved many times in the investigation of trips or abnormal behaviors. Quite often, without the tools, the unit would NOT be started, you know, after a trip, because we don’t know. As an example, we had a trip in which our exhaust pressure went high, and we dumped the unit. It was mid-winter and now, had you gotten the trip and you didn’t have any ability to go back and trend reliably and look at a pattern that went over a year, when I ran the analysis I found that over the course of the last year you can find that the pressure went up during winter, down during summer. And that day, it was a very cold day, high humidity, and I trended the humidity and found, yup, it does make an impact, and we just hit the trip level. So, we made the call right there to restart the unit immediately, versus, had we not had all these diagnostic tools, it would have been ‘okay, now we have to open up. We have to cool down the unit for a couple of days, open up the exhaust, go in there to take a look to see…did something collapse in the first ?’ so…they DO recover the cost.”

Bill Nieman:  “it sounds to me like you’re saying that you’re collecting a lot of data all the time and storing it into your historian.  And you’re using these various systems to allow you to get multiple converging lines of evidence about what a problem is, where the problem might be, and how long you have to respond to that problem before it causes you some kind of unplanned outage.  Is that a fair summary?”

Moh Saleh:  “That is an excellent summary.”

Bill Nieman:  “Could you say anything about the time-to-value of some of these systems? Once you’ve installed them, does it take weeks, months, years, or longer before you start getting the kind of value you expected from the systems?”

Moh Saleh:  “I guess I’ll give you an analogy. Suppose you go off fishin’; when will you catch a fish? You know, it could be within the first 5 minutes, or it could be many months. It’s very hard to say when that would happen.  It really is because your plant could run perfectly fine for an extended duration of time without any problems. On the other hand, the very next day you could have it. I know that in one of our monitoring systems, we were installing it, and …we could not start the turbine. We were going nuts on it, trying to figure it out. And we said ‘hey, did you catch the last day or two of data?’ And they said ‘yes,’ and we said, ‘okay, can you see what changed over the last 24 hours?’…and we found the problem.  It’s just that it was a digital input, it did not go into the logic, and we did not have an alarm in the control system. So right there, in our minds, because we had to start the plant, it paid for itself.”

“Now, that’s a perfect catch; within minutes, we were able to recover. It could have been six months before something like that happened.”

Bill Nieman:  “One thing I get from this anecdote is that it doesn’t take long to set up profiles for these systems, that you can use a relatively small amount of data. Can you say anything about the set-up time of these systems?”

Moh Saleh:  “The set-up time for our particular plant…it went very quickly because we had historical data. So, based on the historical data, the profiles were built fairly quickly. Now, PI [OSIsoft] made it much, much easier; but reality is that you could have built the data, from my understanding--somebody from SmartSignal can expand on this--you could have built the data from the normal historical data from other historians.”
 

Bill Nieman:  “There are a few questions here that ask about specific problems that you might have had. There’s one that asks if you are experiencing exhaust cracks due to cycling. There is another that asks about bleed valve issues. Could you talk about some of the issues that you’re having as you cycle your plant up and down and the kinds of surprises you thing you are avoiding by using some of these systems?

Moh Saleh:  “There are certain parameters that there is no instrumentation to monitor.  As an example, the exhaust cracking: there is no instrumentation to monitor that and, unfortunately, that’s one of those things you have to put up with in life. But there ARE other parameters that we detect that we CAN make a judgment call on. For an example, here:  Boiler feed pumps, right? We’ve had issues where the bearing temperatures start to deviate, and we know we’re starting to have a problem on them. Be it that the cooling fan died on the bearing, and all these indications show us that we’re starting to have a bearing problem.  All these indications are minor and the control system WILL NOT pick these up. By the time it [the control system] picks up on that alarm level, you’ve got something very, very serious going on and your response time is fairly limited. What all these tools do for us is allow us to detect these [problems] very early in the game, develop a plan for monitoring it [each of them],  and, if it’s serious enough, then make the judgment call to create a forced outage and find what’s taken place.”

Bill Alkema:  “One thing you mention, Bill [Nieman], in your question, was bleed valves. Right now, and Moh might correct me if I’m wrong, I don’t think we necessarily, other than what the turbine control systems can pick up, we might not pick up something like that.  But, as Moh mentioned, we also are putting in another monitoring system in place, and that’s our heat-rate monitoring system [EtaPro]. And that will definitely pick up if there is a bleed valve issue--where it is leaking by, or something like that. That system will give us one more angle to come at these problems or to see if we’ve got issues.”

Moh Saleh:  “Now, having said that, I believe that if the compressor bleed valves open, compressor discharge pressure may get impacted.  And so, because you’ve learned the profile, then that small difference in pressure will be detected because at that load level - with all the parameters in which the turbine is running at that particular time - it’s a deviation. It’s not enough for the control system to alarm at, but it’s a different signature, and I think these tools would be able to detect that deviation.”

Bill Nieman:  “It sounds like you’re saying that your SmartSignal system and your EtaPro have pretty sensitive abilities to track the ‘normal’ conditions and then point out deviations in those normal conditions even across this wide range of loads that you’ve told us your plant typically goes through. Is that right?”

Moh Saleh:  “That is correct.”

Bill Nieman:  “There are actually some questions about how do we determine normal behavior? What kind of turbine systems can SmartSignal be used on?”

Bill Nieman:  “Let me just say that, typically, SmartSignal would take about a year’s worth of historical data from a PI system.  We have a staff of experts here who are knowledgeable about the way turbines operate, and they sift through that data to identify the normal conditions of operation for that turbine--whether a base-loaded machine or one that cycles, the way yours does. Having established all the normal operating conditions across a range of ambient temperatures from summer to winter, and the variations in load demand that would be put on that turbine, then the software understands, really, all the normal variations that would be occurring in the system. And, in real-time, when we’re monitoring, it’s very simple to identify abnormal problems using the very sensitive precision that you mentioned earlier.  Anything that really ‘pops up’ is clearly abnormal, either due to some abnormal operating process that you started using or due to some kind of damage to your equipment.

This methodology can be applied to all kinds of turbines--not just the Siemens Westinghouse 501SF at your site--but it’s been applied to the range of GE frame machines, from frame 5, 6, 7, 9 to Siemens V machines to Alstom and Solar machines. So, the same methodology can be applied across all your turbine equipment and across all your balance-of-plant equipment – which we really haven’t spoken much about today.”

Bill Nieman:  “How about problems related to your fuel valves, fuel contamination, valve varnishing/sticking – do you guys have issues like that at your plant?”

Moh Saleh:  “We haven’t had varnishing yet, but those kinds of issues/deviations would be detected. Fuel contaminants you’d be able to detect.”

Bill Alkema:  “We have had fuel contaminants. The Siemens units have ‘witches hats.’  These systems, especially our dynamic monitoring systems, will pick up those fluctuations on a very real-time basis. It’s because of their quick sampling rate.”

Bill Nieman:  “So, of course, the SmartSignal models are built around trying to detect the sort of ‘failure modes’ that are characteristic of the equipment,  and two of the models that typically get built are one,  a fuel model  that looks at things like fuel flows, fuel pressures...And then you also have the very sensitive monitoring of your combustion system through, in the 51F the blade path temperatures and the exhaust temperatures, as well as the flashback temperatures…all of those combine together to give you a really comprehensive picture of how your combustion system is responding to the fuel that is being fed into the machine. “

Bill Nieman:  “Also, here is one of the experts from SmartSignal, Mike Reed. Do you want to say anything about detecting fuel problems, Mike?”

Mike Reed:  “Well, we can typically find some of these in Delta P’s; we can find them in spreads on the exhaust temperatures. There are a myriad of little different items that will be deviations that set up a flag that leads to looking at other sensors to figure out where the possible ideas are of where it’s going wrong.”

Would the system be able to detect an inlet bleed heat valve sticking problem?

SmartSignal solutions are extremely good at detecting issues like this.  Modeling the subtle behavior across related sensors and your operating history (loads, ambient conditions, operational settings) allows them to detect very slight changes from expected behavior – much earlier than the alarm limits on the control system.  And CycleWatch, the CT startup solution, is a great way to catch these issues as the valve positions change during the startup sequence.  It models the expected envelope of your reference, “golden” starts – accounting for all the different startup conditions, and every point in the cycle – to detect such developing problems.  You can see a couple video examples of problems detected by CycleWatch on our website, including gas valves sticking and fouling of bleed air.  View the demo now.

 

Return to categories
 

Vibration Issues

Bill Nieman:  “Could you address the way you guys detect and respond to vibration issues in the plant?”

Bill Alkema:  “We have a vibration issue right now that’s come up since a recent shutdown on our steam turbine.  We have been monitoring this with our vibration monitoring system. But, we also have those going into PI, obviously, and SmartSignal.  We’ve caught them much earlier than if we had allowed them to go to alarm. We just saw it change its signature in the unit itself at load. So now, it brought our attention to it. We haven’t necessarily done anything about it as yet. We’re just on a “watch” basis and tracking it according to load--when duct burners are on, when duct burners are etc. This allows us--we have an outage planned for February--this allows us to make some plans for what we might want to do at that point in time.

Bill Nieman:  “Do you notice that you depend more on these tools to look at vibration during startup or during cycling loads throughout the day?”

Bill Alkema:  “Well, we’re only just learning how to deal with the startup on the CycleWatch. It’s relatively new; I think Moh mentioned that it’s our newest system that we put in.”

Moh Saleh:  “We do have another system there, TIGER, that also looks at the startup and shut down, and it’s got a new feature in there that trends these parameters. I use that as a tool, too. We’ve got a multitude of several different systems. It’s kind of hard sometimes to say: ‘Well, I’ve got a system here that does this type of monitoring. I’ve got another that’s slightly different.’  Can you justify cost? For us, the variation between the various systems has proven very, very useful.”

Bill Alkema:  “You mentioned earlier, Bill…about coming at problems from different angles. But, it’s just the ability-- especially when you’re managing a plant--to get confirmation. You often go into something with that slight suspicion that you might find the problem. With these tools, you have a lot more confidence if you’ve got confirmation from one tool to the next.”

Bill Nieman:  “There have actually been a few more questions about vibration that have come up recently.  How does SmartSignal understand the signatures of different kinds of startups such as a cold startup, a warm startup, or a hot startup? Can you say anything about the information you had to give SmartSignal so that they’d be able to understand the signatures for those?”

Moh Saleh:  “Historical data. And you’re absolutely right. There are going to be as you go through, let’s say, a major, there are new parameters that need to be learned.  We have to re-tune after something significant or if the temperatures vary. Then both SmartSignal and our other systems, such as TIGER, also have to be…they have to both re-learn the new parameters and the variability with temperatures. So the first time you run up with lower temperatures, you may notice that you’ll see some deviations, and you have to tune them. And that’s one of the key things with having a plant champion, in that, when you see the deviation, you’ve got to make a judgment call. You know, ‘Why did these deviate from our learned profile?’  - so, it’s an interactive process - to bring it back up to re-tune the model.”

Bill Nieman:  “So the questions about vibration keep coming in.  How your Bently Nevada system and SmartSignal complement each other in analyzing vibrations and, related to that, the analysis of vibrations in the time domain and in the frequency domain.  Could you discuss any of your analysis in the context of those two issues?

Moh Saleh:  “Whole different subjects between the two. Because the Bently Nevada is providing data that you need…you need somebody to go analyze it …. But SmartSignal is only monitoring the magnitude. It doesn’t look at the angles.”

Bill Nieman:  “So, a lot of our clients say the same thing, Moh. That, ‘SmartSignal is only looking at the RMF signal.’ We only take a snapshot, typically every 10 minutes with EPI*Center and every 10 seconds with CycleWatch. So we’re only looking at the RMF signal, basically, in the time domain; whereas, when you’re ready to dig down into the real ‘guts’ of a vibration problem and do some deep diagnosis, you need to get into that frequency domain information that you’re able to dig out of your Bently Nevada system. Does that sound right?”

Moh Saleh:  “That is correct. That’s a different tool.”