Natalie: I’m going to go ahead and get started now. Again, thank you all for joining us. We’re extremely excited to have you all here today. Good morning to those of you in North America and, good afternoon. I see we have a lot of European friends on the line as well.Today we’re going to be talking about prognostics for warranty and full service agreements, and I’ve brought on Edward Wagner who is our Chief Digital Officer and he is responsible for building our new markets and product direction plan, and he has also built six software companies and IPO’d two of them. He is located in Boston, Massachusetts and I will display his contact information here on this screen for you guys, so please feel free to reach out to him at any point in time. I’ve also brought on Gerald Curtin, who is our director of our asset answers here at Sentient. He’s responsible for program and product management, and he focuses solely on satisfying our customers and working on our various Digital Clone products with them. Gerald comes to us from GE, where he was the product line leader, and he focused on full service agreements, so he’s really an expert in this field today. He also worked on pulse points and aftermarket wind parts.
For those of you who don’t know me, my name is Natalie Hills. I’m the manager of revenue marketing here at Sentient and I’m really the first person who introduces Sentient to our prospects and our customers. I run our global webinar series. Feel free to reach out to me at any point in time. I have many resources that I would be happy to share with all of you.
Just before we get started, I’d like to tell you a little bit about our webinar series. We do do this once a month, and three per quarter. We always start with a deep technology discussion, and then we move to life extension applications where we talk about how we can apply our Prognostic Digital Clone technology to different industries such as aerospace, wind, or even automotive. Then we wrap it all up with a higher value business conversation on the industrial internet, so really how do you apply that prognostic technology to live machines in the field to make them work more efficiently.
We have a great library of resources on our website. It’s under the video library tab on the top right-hand corner there. We’ve been doing this for almost two years now, so there’s quite some topics on there, and please feel free to reach out to me if there’s one you think you might be interested in, I’d be happy to send you over a direct link for that.
I also want to encourage the audience today to insert questions. On the right-hand side of your screen, there’s a GoToWebinar control panel, and at any point in time, please input them in there from the start of the conversation to the end. I did flat out some time to review these with Gerald and Edward later on today at the end of the presentation. At this point in time, I’m going to pass it over to Edward Wagner, our chief digital officer, and he’ll take you through a little bit of the history of Sentient and who we are, and then he’ll pass it over to Gerald for really the meat of the FSA discussion today. Thank you, Edward and please take it over.
Edward: Natalie, thanks, and again, hello everybody. Those of you who have been following us know that we are having this webinar specifically for our European customers and partners. As the business has grown, we’re getting a lot of requests from Europe, and so we are starting a webinar series earlier in the morning to be able to help our European partners see these live.
Just a little bit of background about Sentient. Those of you who don’t know us, we were funded to develop new technology to be able to understand how we could service rotating equipment more effectively. We were funded by some of the world’s largest operators to be able to do this, and if you’re on the line today, I want to thank you. We were supported by them because there was a need to figure out how to service and understand serviceability much earlier in the product life cycle, as I’ll cover in these slides today, and then I’ll transition over to Gerald.
Next slide please. Specifically, our core technology was developed to understand when cracks begin to initiate in rotating components much earlier than you would have known in diagnostics or visual or almost any other technique, which is far right in the graph at the bottom, where by that time you have long crack problems that’d be picked up by vibrations or oil sensors. Our simulation technology picks up the crack initiation, nucleation, and growth very early in the life of the machine, and what we’re able to do is to notify the customer when we see initial damage occurring, and the customer can make changes as needed. This technique we call prognostics. People have been studying about prognostics for years. We’re the first commercial product to reach the market that’s prognostic versus diagnostic based.
Our claim to fame is we are founded on a material sciences model, not a data model, and that’s our deep technical expertise inside of Sentient. We calculate the remaining useful life based upon that crack initiation in the loading, and give that information to the customer as soon as we can, and then the customers can perform what-if scenarios. What if they do an up tower component replacement, what if they change the lubrication, what if they make certain operational changes? How will that optimize performance and extend life?
That enabled us to get into the asset and fleet management markets so that we could look not only over a fleet but we could look at individual assets and give remaining useful life calculations back to our operators. This reduces, in general, significantly the cost and the time of testing, and specifically when we’re faced with the limitations in physical design testing. We can do a lot more computational testing at the same time and at much lower cost. I’m not going to spend a lot of time with this. Our discussion today is on FSAs, but as Natalie said, there’s a number of webinars that are online that go into this in some detail.
Next slide please. Our technical approach is based on these four steps. We predict the remaining years for the life of a component or components under loading conditions, then we look to confirm those predictions using SCADA or HUMS or various other ways to see if we get the predictions correct. Then we make that information available to the customer to game, meaning to do what-if studies and understand cost trade-offs, and finally, we make recommendations on life extension that the customer can now follow. In some markets, we can drive those recommendations directly to the machines, and some markets, we have to make recommendations through structured processes, but in general we predict first, we confirm second when we game, and provide a number of applications to be able to play what-if studies, and then we provide optimized information for life extension back to the customer.
Next slide please. And as I said, this is in contrast to a traditional diagnostic approach where you are using sensors, absorbing that information, creating databases, looking for baselines and changes to those baselines. We know what that costs you. The prognostic approach is different. The cost is much much lower. The initial value proposition is we get these predictions done within 60 to 90 days, which is extraordinary. We do this by building a multi-physics model, a digital clone of the component or components or systems that we’re looking at, and when we run these sensitivity studies under a lot of high-performance computing. Once we have a prediction, we only need to alert the customer if there’s a change in that prediction, so we manage by exception. It is a software product. It’s delivered as a Software as a Service, because the back-end high-performance computing is quite significant. We do that at the University of Buffalo, and then we make the results and asset answers available to you through a browser.
Next slide please. Ultimately, what we’re trying to do is to get information to the operators as early as possible when we start to see problems on bearings and gears and systems so that they can take whatever operational decisions that they want to extend life, and this is working for us. If you wait to the time where the vibration indicators begin to alert you, you’re very late in that life cycle, too late to extend life. What we’re really doing is extending the remaining useful life of machinery through the use of prognostics. Hopefully, that makes sense.
Specifically, next slide please, when we get into a customer engagement, there are a series of reports that we’re providing to our operators and the OEMs. This is an example of those reports., a fleet ranking of the assets, in this case gear boxes, from worse to best. Make sure that we’re picking up problems in gear boxes that are not on an existing watch list, and budgeting can be done. What gear box will fail, where and when, and then plans can be made for supply and inventory management systems.
Number two, if there is a chance to do up-tower component replacements, we make sure the customers know that, because that’s the optimal thing to do, and so we are not only looking at the systems level. We’re looking at the prognostics on the sub-component level, so we can give a list of assets to a customer that we have reported on that can be serviced by an up-tower component replacement instead of full gear box replacement. We’ll take a look, number three, at all gear boxes that can be extended by derating, and what’s the impact on life, and same for uprating, for those operators who are thinking of uprating their machines. What’s the impact on the life of the gear box?
Number five, what is the as-is state of the asset, the bill of materials, and then once we have simulated what will be the to-be state of that bill of material so we can get that information back into the ERP systems, and finally, take all of this information and create five-year budgeting plans based on this data to estimate the spending per site. It’s a lot of information coming out of the systems, but ultimately we’ve done a pretty good job of providing a new level of planning information on a site-by-site basis.
One more slide. Slide eleven, please, Natalie. Just to give you an idea, we are now at about 14,000 assets that we’re managing globally. We’re very proud of that. Our customers in general are falling into two buckets, customers who really want the most tested products in their markets. There’s not enough time and money to do more hardware testing, so they want to leverage our software computational testing environments to create the world’s most tested products, and then products with the lowest cost of operation where we can extend the remaining useful life of those products and get the costs down, especially in the O&M areas.
I just wanted to go through a couple of slides to introduce Sentient. We’re very proud of what we’ve done. The company is growing rapidly. Now I’m going to transition over to Gerald to talk specifically about FSAs, and if you have any other questions regarding Sentient, there is an enormous amount of materials online, and of course, as Natalie said, you have my contact information. Thank you. Let me transition over to Gerald, and we’ll get started.
Gerald: Thank you Ed, and good morning to my friends in the United States, and good afternoon to my friends in Europe. Are you an owner with units coming out of warranty or FSA, and looking to decide what to do next, or are you an original equipment manufacturer or an owner evaluating an Install Base FSA extension? The good news is the same underlying questions are there. How do I quantify my risk, and then how do I manage my risk?
What prognostics does is puts you in the driver’s seat in quantifying and managing risks, and today I’m going to go through that with a particular emphasis on quantifying, because that’s the first step. You got to quantify before you figure out how to manage.
Let me start out with going through some of the common contract tools for managing unplanned wind turbine risks. For those of you who haven’t heard the term unplanned, you’ve got a planned maintenance when you change out batteries, when you’re changing out oil, and then all other events that are unplanned, so a pitch card failing, a gear failing, et cetera. We’re going to talk about the latter. If you have units under warranty, you’re going to fit into one of three products. One is an in-and-out warranty, which is essentially bumper to bumper, so crane, labor, parts, or parts and labor, or parts only. From an unplanned maintenance coverage, the first one in and out has full coverage, and then the other two have a certain degree of coverage, or I should say a certain cap based on the percent of the turbine price.
Covering unavailability. Unavailability when your turbine’s down while the grid is up. Your in-and-out warranty, you’ve got an availability guarantee. For the other two products, you don’t. And then greater than 30 days unavailability, which we call business interruption, the warranties don’t cover that, but your insurance products do. We’ll get into that in a moment.
And pricing and term. Typical warranty, two years in the States, five years in Europe and other markets. The price, typically you don’t know what the price is. It’s built in to the price of the turbine, so it’s capitalized, it gets depreciated. It’s great. Those are the products for warranty.
Now, if your units are outside of warranty, you really have three options. Your first one is your balance sheet. That’s where all the risk is on your balance sheet. You’re paying for it. Second one is insurance. Property and plant insurance, you also have machinery break-down in some cases. Essentially from an unplanned maintenance cost coverage, the property and plant insurance will cover, I should say has a deductible. Your deductible is typically 100,000 to 250,000. The range you choose impacts your premium that you pay, and then there is a cap to the value of the turbine, typically.
Your coverage, however, especially when it comes to capital component failures that are over a $!00,000, tends to be limited, so your first gear box that fails or your first blade that fails will be covered in full, typically. The second one, 75% of the value, third one, 50% of the value, fourth one, 25% and then the fifth one, you’re on your own. Unavailability insurance typically doesn’t cover that. Thirty-day business interruption, it does cover that as long as it’s caused by something that’s sudden and accidental. Those words are extremely important in the insurance business. The pricing term, well, if you’re trying to manage your risk for long term, insurance is not the way to go, because you’re going to get a one- to two-year fixed price, and for those of you who ever got into a car accident, if you’ve hypothetically run or backed into somebody in your driveway, you know that when you go and claim your premium increases. I know that.
Your second option, or I should say third option here besides your balance sheet, is what we call a Full Service Agreement, or what our European friends call an LTCSA, Long Term Contractual Service Agreement. Again, it’s very similar to a warranty in concept. It’s full coverage. It may not cover the full turbine price. Typically, it’s some multiple of the service agreement price. The unavailability is covered just like a warranty, a full in and out. There’s no business interruption.
The real benefit with FSA is that it’s covering anywhere to 5 years to 15 years, and there’s actually longer ones out there, but typically 5 years to 15 years fixed price or stepped price. All in all, I want to point out the reason the FSA exists is because of the insurance, and the insurance for those who are trying to cover risk, insurance really doesn’t do it. It covers the big ticket hits. It doesn’t cover all the small ones. It even does have limited coverage, as I mentioned earlier, on the big ticket hits if they keep happening.
Natalie why don’t we run over to our poll? Let me get that up. We’ll take about 30 seconds here, and who’s listening in? Are you someone who has turbine warranties ending in the next two years? Are you someone who has an FSA or LTCA ending in the next two years? Are you evaluating an Install Base FSA or are you just a Sentient science groupie? We know who you are. If you wouldn’t mind filling that in over the next 20 seconds we’ll take a look and move on.
All right, Natalie. Lots of groupies, as I figured [laughs]. Well fantastic, we have a little bit of everybody on. Let’s go into the primary question of, how do I quantify my future risk? For the reliability folks, this curve should look familiar. If it doesn’t, it’s called the bathtub curve, because it looks like the shape of a bathtub.
On the left-hand side, you have your wear-in period, where the further up you go in the y-axis, the higher the failure rate, and then the further you go on time on the x-axis, the longer the asset life. Stage one, wear-in, that’s where you have quality defects and manufacturing defects. This is the typical life cycle of a product. You can pretty much fit any product in there.
Category two is the random failure stage. That’s where you’re seeing impacts of operating stresses, maintenance issues, accidents, and then you get to the third stage at some point in time which, is the wear-out. That’s when you start experiencing back pains and knee pain, or if you’re a wind turbine, you see fatigue issues, et cetera. So, the wear stage. Every product you can fit on this curve at some point in time. Some are better than others in the beginning, some are better than others on the third stage and the second stage, but how do you quantify your risk at where you are?
The first approach, you can use past history. It’s always valuable to learn from the past. Weibulls, which is failures over time. Another approach is your current condition. You look at borescope reports, you look at condition monitoring system, you look at visual inspection, all the valuable tools for understanding where you are and what your current risk is.
The third option is using physics-based modeling of wear on your gear box, on your main bearing, on rotating components that’s fatigue. The challenge is, the first two, the risk of you depends on where you’re measuring on the bathtub curve. If your asset is in the wear-in stage, and you’re creating a Weibull, and your Weibull shows that your failure rate is decreasing essentially to zero, well, great. That’d be fantastic, but is that real?
If you’re measuring in the random phase, then you have a slightly increasing failure rate over time in this particular case. Again, is that real? Now, the good news is with physics-based modeling, or the breakthrough really, is that you’re looking at stress and wear portion of the curve so it doesn’t matter where you’re measuring on the curve, because you’re predicting through material science the actual wear stage and operating stress stage. I just want to point out that is the primary differences between using past history and current condition. It’s looking in the past, and then the physics-based modeling is really looking in the future, so it doesn’t really matter where you’re looking from. You’re going to get your accurate curve.
Another way of looking at this. Today, the general rule of thumb on say one-and-a-half, two-megawatt turbines, unplanned maintenance costs are seen as 40 to 60% of total maintenance cost. There’s a lot of variables, environment, turbine type, that fall into that but again, as a general rule of thumb, 60% of that is from the drive train, 20% pitch and converter subsystems, 5% from blades, pitch bearings, and then the rest is 15%.
When you’re going to plan, the typical approach is look at your Weibulls, take this into account and that’s your go-forward plan, but the question is, does this really represent your turbine’s future performance and future risk? For me, in my past history, I’ve seen two of the same exact turbines in different locations, one in a 35% capacity factor environment, the other one in a 50% capacity factor environment and the reliability rates are vastly different. To me, this is a very generic tool, but if you want to get accurate predictions, this is not a one size fits all.
Let’s talk about looking ahead. As Ed mentioned earlier, there’s a couple of prognostics approaches out there. They’ve been around for a while, but we’ve come with something new. There’s standard bearing lifing, which is ISO 281, there’s standard gear lifing equations, ISO 6336, and then there’s Digital Clone. What we’re really adding is we’re looking at the factors that impact life. Load, speed, lubricant properties, surface finish, component microgeometries, and material quality.
What we do is we take those factors, we simulate and we identify. Again, this is a repeat slide. We simulate crack nucleation through crack initiation so that we’re getting you ahead of the curve so you can make decisions to extend the life of your asset before it’s too late, before components start getting damaged from other components, micro pitting, macro pitting, gears, shearing, et cetera.
For the folks on the line who have upcoming FSA and warranties ending, let’s talk about how to use prognostics. In general, you fit in a bucket where you don’t necessarily have a solid grasp of your failure events and your cost. Let’s face it, you’ve seen the OEM climb up your turbines. They go and work on it. You’ve seen them come in for a gear box replacement. They’ll tell you, but what you don’t know is what component it was replaced with, what the failure rates of that component that they replaced it with, and what the costs were for the event.
So, how do you use prognostics? The first is to quantify major component risk, damage risk, by site, by asset. In this particular case, we’ll say your FSA ends after 10 years. You have a used Digital Clone. You get a site-level gearbox damage estimate over time, and use Digital Clone on the main bearing, and you get fatigue damage. Sorry, you get a fatigue damage Weibull over time for main bearing.
So what you do is now that you have that you can actually evaluate how your risk of those major components compares to the FSA extension quotes that you’re receiving, because you’re evaluating, right? Your FSA is ending, your warranty’s ending. What do I do next? Do I do self-perform? Do I go after Full Service Agreement? Do I go with a third-party, just O&M, and take all the balance sheet risk for the unplanned maintenance?
Chances are if you had an FSA that you signed eight years ago, you got a pretty good deal, so that’s the gray bars. The OEM or the third-party is coming to you now with significant increased FSA price. You don’t know what the cost were spent, so you don’t know if you’re getting a good deal or not, so you make a couple of calls. You find out what a gearbox replacement event costs for your turbine, you find what a main bearing replacement cost event is for your turbine. You can actually use these estimates to estimate your cost.
In this particular case, there’s quite a large gap between what the FSA cost is overall and what your major component risk, which is supposed to be again a sizable portion, 60% of your unplanned risk. So you can ask yourself, “Are you getting a good deal?” Now, on the flip side, this could say quite the opposite. This could say that you have a lot of gearbox risk or main bearing risk coming up because of the environment, because it’s the lubrication you’re using, whatever. For a variety of reasons.
So the FSA may look like a fantastic deal. Now, in that particular case I would caution you, you need to pay very close attention to the contract and understand your caps, because that determines your level of coverage. You may have a lot of failures, but then your caps run out, so you’re not covered after a certain period of time. It’s not a good situation. That’s how you can use prognostics if your FSA or warranty is ending.
For the other folks who are evaluating an Install Base FSA, you’re in a bit of a different camp. You have historical data, you have knowledge of operating the turbines, or at least have the costs estimates of the operation, but how does that translate into future risk? That’s the fundamental question, because you’re looking at 5-year, 10-year FSA, Install Base FSA. Do I go with that, or do I do this myself at lower cost?
So, in this particular case, the first thing you can do is compare your particular site or asset failure history to your expected fatigue risk, and in this particular case you see a gearbox. I have a graph here, and in 2014 this particular site, there were a lot of gear box failures. So you create your standard Weibull, and it shows that in 2014 your [inaudible 00:31:11] failure rate was about 10%. In 2020, it’s predicted to be around 50% cumulative, okay?
To start, you could do some planning to that. However, is that real for your site? You want a Digital Clone model, and Digital Clone says there’s going to be wear, risk at a much more rapid rate. The good news is you have a similar total failure here roughly, about 5, 10% difference at year 2020, but the significant concern or highlight here is that you actually have quite a different ramp rate of failure, so if you’re making a decision on FSA or you’re making a decision on owning, how you manage this and how you cover this is the difference between making or breaking your profit.
Also, because you have a history, because you know what components are in there, you can actually go into component- or supplier-specific risks. In this particular case, this is actual high speed bearing comparison of a gearbox. We have two different suppliers, we have several critical life factors shown here, and the moral of the story is from a hardness rating of the metal, supplier A is better, from a microstructure inclusions standpoint, supplier A is better, and from a microstructure quality or material quality standpoint, supplier A is better.
Great. So what does that mean? Well, you put these factors in through Digital Clone, and one of our outputs is what we call a stress-life curve. So on the y-axis, we have increasing contact pressure increasing stress, on the x-axis, increasing cycles. Ideally, you want to be as far up and as far to the right as possible. So here we have supplier A, as you can see, is way further up and way further to right, so it can handle much higher stress and much higher cycles, but that’s not the only story here. So, you have to take into account cost, and also where that component is located, and how it’s loaded in the unit.
So if that supplier A is actually much more costly, it may make sense if you’re operating here up in the contact stressor. However, if you look down here, Supplier B, if your turbine’s operating in this particular component, and the gearbox is operating here, Supplier B may be just fine. So again, this is stress-life curve, and depends on what component is where.
At the end of the day, what you’re doing in both of these scenarios is using prognostics to understand risk and premium. Can it predict this car crash? No. Can it predict gear and bearing fatiguing? It can, and it does. So to wrap up, if you’re an owner, you want to use prognostics on units coming out of warranty or FSA to quantify your major component risk to compare risk versus your FSA premium, and also insure your FSA contract coverage is adequate. Are you going to have the right number of major components events covered?
As an original equipment manufacturer or owner, you want to use prognostics on units for Install Base FSA to compare your historic versus your future risk, to quantify your component supplier risk, and also again to insure your FSA contract coverage is adequate. So that is how you can use prognostics today for FSA and warranty turbines and evaluation. So that wraps up my presentation. Natalie, why don’t I hand it over and go through a couple of questions?
Natalie: All right, Gerald. Thank you so much for taking us through that. That was a great presentation. I do have some questions from the audience right now, and if everyone just wants to input them, please feel free to do so while we’re going through the other questions. I can always reach out to you individually and put you in contact with our technical team. But for the first question…I think we have time for about five, so the first question is, “Can you use the same technique for components that don’t have cracks but are metal which are in extreme yield mode,” and Gerald, this is for you.
Gerald: I’m actually going to defer that to our technical team.
Natalie: Absolutely no problem. I saw that question come in, and I will put you guys in contact with …
Gerald: Please do.
Natalie: Great. Thank you. The other questions I have is, “Have any customers of Sentient used this to get a lower FSA agreement?”
Gerald: We’ve actually been used not for FSA, but for post-warranty negotiation successfully.
Natalie: Great, and I believe I have some information on those and use cases on those that I would be happy to send over, so anybody want to reach out to me directly, I would be happy to do that. The next question is, “What data is required to get the prognostics model to work, and what do you ask of your customers for the inputs?” So, Ed, I’ll pass this over to you.
Ed: Thanks Natalie. So, in general there are two buckets of information. One is we need to know the asset itself. To build the materials, we need to know what is the asset, gearbox type, and some details about the geometry. The second is, what’s the loading conditions, and that comes in through METAU [SP] or SCADA, and we use that information to serialize the specific Digital Clone model. For those of you that have 100 of the same gearboxes, we may have one DigitalClone gearbox model, but it is serialized for every specific asset based on the loading conditions of that asset. Those are the two different types of information that we collect from the customers to do the predictions.
Natalie: Thanks Ed. Gerald, this next question is for you, and it’s, “Has Sentient integrated with the current industry certification process, so some of the DNV GL or the different industry processes that are used today?”
Gerald: Interesting. We’re not integrated. We’re actually being used as an addition to today for a number of customers in the design phase, and we’re hoping at some point in time in the future to be part of the process, but we are not yet.
Natalie: Great, Gerald. Thank you. Just for the last question, I do see some more coming in, but I want to be mindful of everybody’s time. I will reach out to everybody individually. But the last question is, “Do you have industry use cases where you’ve been talking about FSA or other applications of prognostics in the wind industry?”
Ed: Sure. So yes, and many of them are up on webinars and case studies that are on the site. Obviously, we de-identify information, because we’re considered a competitive differentiation point of differentiation, but we have enough experience now in wind, in industrial, in aerospace to show you different use cases. Remember, this technology was being built over 14 years, almost 10 years of that in R&D and validating the models, so we have a lot of white papers about the models, we have a lot of now case studies of use cases, and in industry today, to drive down OEM costs and the movement of companies who are moving from a planned preventive maintenance to prognostic health management to lower their OEM costs, we have a series of case studies that are up on the website.
Natalie: Great. Thank you, Ed, and to the audience, I did say this in the beginning, but please feel free to reach out to me. There is a lot of information, so it could be overwhelming. If you’re interested in a specific industry application, we have work in helicopters, on the Apache, we have different things in the wind industry as well, so I’ll send over a specific link if you are interested.
At this point in time, I’m going to just end the question session. Again, please reach out to me if you have any specific questions that perhaps you want to be in touch with a technical team in a bit more detail, and I do want to let everybody know we will be at the London Offshore Wind O&M Forum, which is November 4th through the 5th, the first week in November, held by our friends at Wind Energy Update, and I do have Dr. Nathan Bolander, our chief technology scientist presenting there, so please feel free to join us. I can send you over more information on that as well. We’ll also be at the EWEA show for our European friends on the line. I see a lot of you here, which is the third weekend in Paris, and we’ll be walking that show, so we’d love to meet up with you guys, and I would be happy to put my team in contact with you. Again, thank you all for joining. Thank you, Gerald and Ed, for your time this morning, and I will talk to everybody next month.