93rd General Meeting Presentation
Emerging Issues in the Power Generation Industry
Dr. John Siefert
The following remarks were delivered at the General Session of the 93rd General Meeting on May 12, 2025. It has been edited for content and phrasing.
INTRODUCTION: Dr. John Siefert is the Area Lead for materials research in the Generation Sector for EPRI. His experiences in welding research are diverse, and he is the primary or contributing author to more than 150 manuscripts.
His slide presentation can be found here.
DR. SIEFERT: Good afternoon. I was listening to Tim (Kennedy, keynote for the Opening Session) this morning. I thought, “How am I going to open up this discussion?” And the one thing you must be thinking is, “What the hell is a Ph.D. going to teach me about power generation?”
We're going to discuss a couple of stories and walk through some emerging issues, but this has many layers to it. And you've already seen some of the themes come out in a couple of presentations.
First, I'd like to address: What is EPRI? EPRI was formerly known as Electric Power Research Institute. We dropped that and are now called EPRI, because of the acknowledgment that energy is starting to drive the world. We want to be a part of that, and not just in the electric sector.
We're very fortunate to have approximately $500 million in funding each year, spread across several different disciplines, programs, and sectors. However, you can divide EPRI in various ways. If I'm fortunate enough to receive funding, I usually receive between $5 million and $10 million. So, about 1%-2% goes into materials dedicated research, of which some is for repairs, which we'll discuss later, and part of that has made its way into the National Board Inspection Code (NBIC). My participation in Part 3, Repairs and Alterations, is one of the most rewarding experiences I've had in my career.
I'm not going to read through each of these boxes, but if we start at the bottom, what could possibly go wrong? That's the first thing to consider in this discussion, but we're really at a crossroads. To give you a military analogy, because I grew up in a military family, there are a few features and systems in a power boiler that require more than just the ASME Code. They require more than just the frontline infantry.
Some things require special operations, even on the design side, to be done correctly. What are we doing to train those people and update the codes and standards? Well, not enough, because some of the things we're trying to do today genuinely aren't just state-of-the-art. They're well beyond what we've done historically, whether it's the materials we're using, the systems and temperatures we're designing to, I've heard a lot of discussion about some of the pressures that you see in some of the more common boilers. Some of the piping systems we're talking about today operate at 4,500 psi every day. There are hugely different systems in some cases, but let's discuss what goes wrong and what we're doing to try to fix a portion of that.
EPRI issued an industry alert, and this was the first-of-a-kind document. EPRI never envisioned having to do this, quite frankly. It had to do this because who is doing this anymore? The original equipment manufacturers (OEMs)? I returned to EPRI from Babcock & Wilcox 14 years ago. Why? Because they closed the research facility.
Who are the OEMs? There are dozens of them. Today, they're manufacturing heat recovery steam generators and, to a lesser extent, some of the large power boilers seen around the world. However, they're no longer issuing industry alerts like they once did, so we had to take this step. We did this because it wasn't just one failure. This is one of the major misconceptions, and I often hear it when I come to codes and standards. They say, “It's nice, John, to bring an issue, but is this a one-off?” No, it's at least a two off. Two is fine. Two is in play. Is it a three off? Is it a 10 off? We're talking about hundreds of leaks. Fortunately, we're not aware of any failures that have ruptured today, but we're discussing multiple materials, the mainstay alloys used in these high-temperature systems, and we're talking about failures occurring as soon as 35,000 hours. Does anyone know how many hours are in a year?
MEMBER: 8,760.
DR. SIEFERT: Excellent. So what is that? That's about five years of operation. I'm not sure about you, but 35,000 hours doesn't align with the intent of the code. Fill in the blank for any code.
This is a system where the expectation is that these last 300,000 to 400,000 hours, because that is the age of the U.S. coal-fired fleet. Let's discuss a modern combined cycle, and let’s try to put the tee issue into context. I wasn't sure how familiar the crowd would be with the power boiler, my cycle plant, but here we have a plant that has four gas turbines, four heat recovery steam generators, and two steam turbines. In total, a set like this generates about a gigawatt of power, equivalent to the size of a nuclear reactor, so it's a pretty big cycle.
The footprint of this particular unit has drawn comparisons to a football field. However, suppose we didn't start to delve into the heat recovery stream generator and the steam transported from that system to the steam turbine generator. In that case, we can look at a simple isometric diagram, as shown here. This is for a main steam system. We have a second heating system that would look very familiar, but in this particular configuration, there are four tees. Some systems might have about a dozen tees, some might have a couple. It ranges quite significantly, but for this particular site, because it's a two times two on one, they would have had a total of about 20 tees.
What does a tee look like? Now, a bit of a funny story. My dad is not an engineer; he's a pilot. My dad was reading an article I wrote not too long ago. He said, “Man, you talk about girth and crotches a lot. What happened to you?” I said, “Dad, if you think the engineering terms are bad, you should see what the welders call their defects.” So, these are very real terms. If you haven't seen a tee, there are two crotches. Three girth welds are needed to join that tee into the system. We've seen leaks in the crotches, and we've also observed leaks in the girth welds aligned with the crotch. We've seen leaks in those welds, whether they were shop or field welds. There’s a misconception in the industry that field welds are worse than shop welds. It's not always the case.
There have always been leaks on the tee side of the pipe side. And there have been leaks in the branch, or what we call the main welds. So effectively, you can get leaks or cracks almost anywhere in these welds or crotch positions. Before we delve into what we're seeing and why that presents a significant challenge, let's address the issue at hand. I say we almost as EPRI, because we don't attend widely to the nonnuclear codes and standards. The nuclear guys may. They sit and argue, and they coordinate. There are about 25 guys from EPRI who go to ASME Code to help Section XI. We send two and sometimes three, because I can attend, but in most cases it's two.
When we discuss tees, we address the necessary updates to the codes and standards. The piping system is designed to Section I or B31-1, depending on where the split occurs between boiler external piping and non-boiler external piping. The tee itself, interestingly enough, has a stamp A-234. Has anyone looked into an A-213? Has anybody read one of these product form specs? You're not missing much. They're a bit lacking in guidance, but they often refer to other requirements. In this case, A-234 heavily references some of the requirements in A-960. And then there's yet another code or standard to which the tee is designed to.
That's ASME B16.9, and here lies part of the problem. I think I know the answer to this question. Has anyone looked into ASME B16.9? For the most part, ASME B16.9 consists of tables. And for tees, it governs the C and M dimensions. That's all it is. It governs the envelope in which the tee must fit. Anyone who has done the calculation, or perhaps had some experience with the designer or insurer, can tell us who designs the thickness of the tee? It's not one of those five codes and standards on the screen.
Therein lies one of the first issues we have. No one has actually checked in what the thickness in the crotch position should be, because look, I'm not a mechanical engineer; I'm a Ph.D. in materials science. But I'm told that it’s not a good thing if we put a hole in a pipe. If we do that, we have to reinforce it, and we have to reinforce it if we're operating in the creep range for hundreds of thousands of hours.
What else have we seen? These tees are now in service. They're very under-designed, but if you look at this slide on your left, we have examples of tees that were stamped as one material and found to be another. I'm sure the paperwork for this team was perfect. How does this entity, which has stamped this tee, still have a code stamp? What if that was a carbon steel tee and not some high-temperature material that did some reasonable time in service?
That's a problem. If you look at the bottom of the slide, you see those vees? Those vees should not be in a high-temperature tee. Those are what a mechanical engineer will tell you are stress concentrations. The vees are there because these tees can be difficult to form and reintroduce defects that have to be removed. However, because there is no thickness calculation, the vee can be almost any depth, as they may or may not be checking the thickness in the crotch position.
Then we get to another manufacturer, and we see in the top right-hand corner that they removed a lot of material when they did their vee. What did they go back and do? They repaired it. This material that we're talking about, by and large, they see significant reductions in performance when welds are introduced, because the heat-affected zone is grossly under-maxing in strength to the base material unless we do a full normalization, intemperant heat treatment.
None of those five codes or standards requires them. You have tees in some cases with vees; some tees that aren't stamped properly; some tees with repairs that are undocumented, not properly heat-treated, causing failure in 35,000 hours. In some cases, even when the tee has good paperwork, the right material, and is properly designed, we still see failures in the girth. It’s because they're placed way too close to that crotch or that zone of reinforcement, which is supposed not to be allowed. However, that's not one of the design criteria governed by B16.9, A-234, A-960, or Section I of B31.1, which allows those codes and standards to be used for their piping systems.
So this is the management slide, right? We tell them all the time it's about safety, but I'm not sure if that falls on deaf ears sometimes, but they do respond to this occasionally. Now, the cost is incredible. I can give you many different cases, but the thing that I'll never understand for a large utility is that – or small one for that matter – if it cost me five to 10 times on the back end to do it right, that is replacement cost, why wouldn't I save that money and do it right at the front end?
And that's because we're back on a fundamental problem. Those outside of this room likely do not appreciate codes and standards for what they are. They are a minimum set of requirements. You will hear them use language like a gold standard. Is ASME a gold standard? Is the NBIC a gold standard? Absolutely not. It should never be used in that language. It is explicitly stated in the forward of these documents that they constitute a minimum set of criteria.
We estimate that just for tees, we will revisit this point because there are a couple of elephants in the room. But just for tees, worldwide, we're talking about an exposure of more than $10 billion. And that's being conservative. If you held a gun to my head, I think I could say it's more like $15 billion or $20 billion.
Before we get to the elephant in the room, what’s EPRI doing? Because the codes and standards space is so entangled and so difficult to decouple and deconvolute, at least for one organization alone, the one thing we can do is write a spec. We're writing spec right now, but one of the things we recognize is that there have to be different options. We can't go back in the best case; for instance, consider machine forging. That's the metal option. Lots of people are trying to source machine forgings. The way this is done is that a large block is procured, and then 80% of that is machined and removed by weight to produce the machine forging. How many entities in the world are equipped to do that or want to do that in a few weeks? None, not unless you have a CNC machine in your garage you can run 24/7, right?
Now, machine forging, depending on who you're talking to – and there are a couple of manufacturers – those guys are not equipped to do the calculation for you. They're not equipped to do a final analysis if you want to use B-16.9 upfront language to better design the part for you. They have to rely on simple tables. We estimate that just to update the tables for B16.9 for tees would cost us $2 million. And that is because we have to assess performance not only for simple things like overload, tensile behavior, or even fatigue, which is simpler than creep, but also for long-term performance using more sophisticated means, adopting an allowed special forces approach, because there is no other way.
So the spec will come, but we have an end user right now. They are building a combined cycle plant in the Midwest. Do you know what it says in their spec? No B16.9 tees in my plant. I got a call, a desperate call saying, “John, will you help?” “What do you need?” “I'm building this plant. I have B16.9 tees on the ground. The EPC chose to completely ignore the language in their spec.” And what did the rest of the people around the EPRI room say to end users? They said, “Yeah, that's pretty typical.”
Now, we're two of the elephants in the room. If you look at A-234, it's not just tees, it's fittings. What other types of fittings are there?
MEMBER: Elbows.
DR. SIEFERT: Excellent. There are end caps, torispherical end caps, and reducers. If you asked me if I was worried about those other fittings, I would say yes. Whoever said elbows first, I said there were maybe 10 tees in a system or two systems. There are probably three times as many others. If we think there's an issue in elbows, and we find one, that $10 million is going to pale in comparison.
The other elephant in the room? Seam welds are a huge challenge. And to give you an idea of the challenge with seam welds – I promise you this is my only nerdy plug – but if you look on the right – and you don't have to be a Ph.D. materials scientist or a mechanical engineer. But if my expectation for this test condition was that the material should fail at about 13,000 hours, we were given an elbow with four seam welds: one at the intrados, one at the extrados, and a seam weld in each of the pieces of pipe that were welded to that opening.
We tested all four. We have facilities in Charlotte for that. When we tested this X service elbow, the best-performing long seam weld failed at about 1,600 hours. That's an eight times reduction in the expectation and performance. That’s not doing too well, but maybe if I know the problem, I could manage around it. We then tested the elbow seam welds, and they saw a further 13 times reduction in the extreme.
The expectation was 13,000 hours, and the worst-performing elbow was about 200 hours.
Someone is going to say I don't run the plant at 1,157 degrees and 12.3 ksi. I can see that. But if this variable holds true in the long term, it means that some of these components could fail in 10,000 hours or a million hours. Does anyone know how long a million hours is? I guarantee it outlasts everyone in this room. And it's going to outlast my kids.
That’s the sobering story, but I'm here to bring a message of hope. The message of hope is that the market is evolving rapidly, and it will force us to change. I’ll mention a few things at the end that may be a bit contentious, as that's my nature. But if we're going to change, we have to rethink how we do business radically. How we're thinking about our codes and standards that we love so dearly, because if I had this discussion two years ago, I could have told you the market is flat, there is no load growth in the U.S.
This was a projection given at a meeting two months ago by one of the largest end users in the Southeast U.S., and these are their load growth projections, which are just phenomenal. In some cases, they're projecting more than 10% growth per year, so from zero or negative growth to 10%.
What is this causing others to do? You see it right there, right? Those of us who like coal, it also means we're going to install a bunch more plants, and they aren’t going to be a bunch of solar panels and wind turbines. It needs to be part of it, but it's going to be a lot of combined cycle plants, too. If we think we have an issue now, we're going to have more of them if we're not too careful.
What is EPRI doing to address some of these concerns, because some of the thermal assets, especially aging thermal assets, are more of a likely engine before the bread and butter that EPRI is well-versed in? We can think about it in the two different thermal fleets. If you look at the distribution of coal-fired units in the U.S., the old ones stop at more than 300,000 hours of operation. You just saw that Southern Company said they're not shutting theirs down; they're going to get a lot more runtime. The modern supercritical fleet, which includes a couple of dozen modern plants, is actually approaching 100,000 hours, so they're no spring chickens.
These combined cycle plants, which they envisioned building in the year 2000, were intended to be peakers. We're going to build them as throwaways. We'll run them 20 years, dump them, and we won't use them anymore. Well, those things are now 22 years old on average. They're being run, in some cases, based on more than 150,000 hours, which is quite interesting for what was once a throwaway item.
Now, the background, the U.S. fleet. The combined cycle plant has been responsible for 50% of the generation over the last couple of years. What does that mean for me, because I'm also a welding engineer by education? It means we need a lot more repairs. One of the things that EPRI has done, working closely with Part 3 since approximately 2013, is to bring more industry research to help not only improve some of the language but also introduce new items into Part 3, particularly alternative weld repairs.
That investment from industry represents more than $7 million in research funding to make it happen. It's not enough for us to introduce things into the code; we have to ensure that the words we put in are still relevant. And much of the research we're doing today is tracking the performance of these repairs so that we can be absolutely sure and competent when we stand in a room like this and when someone asks, “Do you really have faith that you can repair these materials without heat treatment?” I can say to you unequivocally, yes.
The research for us and repairs are not a dime. The backbone of that information is substantial. More than 800,000 hours of creep testing – most on feature creep tests. Those tests cost $2 to $10 an hour.
I mentioned we're tracking repairs. We can point to more than 10,000 repairs and Grade 91 steel across 15 end users who have shared their information with the longest inservice repairs now more than 70,000 hours of service. Considering that Welding Method 6 was introduced in 2015, this is a phenomenal achievement for the industry.
The other reality, at least for the post-construction world, is that plants will have to operate with cracks. This makes some people nervous. In instances, we can clearly show that running with the crack is less risky than trying to repair it. Some of these components, like this valve you see here, are not the way we should design valves, but this end user got this kind of valve. The valve in those corners is 16 inches in thickness. If you made them repair those cracks, and I said to you we didn't need to do that, they wouldn't be very happy. I'm not sure they could actually repair it anyway and get access. Has anyone tried to do a 16-inch-thick weld before? I mean, I did like 3 or 4 inches, and those were pretty big; 16 is a whole different ball game.
We're finishing and starting a second phase of research on fitness for service to inform what needs to be done for power generation issues. There are significant gaps, even if some people don't want to hear it, in fitness for service one or BS 7910, which is the other international code for this, when it comes to power generation specific issues, so we're hoping to fill some of those gaps.
There is extensive documentation for this as well. I won’t belabor this point, but there is also a nice position paper. You could click on the link in the slides and get there. You can see what we said five years ago and the gaps we needed to address. We're well on our way to doing it. Not just looking at post-construction for repairs, but also addressing post-construction for running with damage in the event I cannot get replacement components or need to run in that condition.
Some of the challenges we discussed, and maybe there are a couple that we haven't emphasized so much, but we need to start thinking about. The tees that were the onset of this presentation lay bare the deficiencies in simple design rules. There is no easy path to overcome this. If I said that we need a special forces approach because we're basically sending our infantrymen to do the job of something very sophisticated, that doesn't tell the story. We're sending men and women right out of basic training to do this job.
The large OEMs I'm referring to have armies of engineers. A lot of those engineers are yours truly, right? What about the industry? They were trained by large entities with engineering excellence at the forefront. That is not the case today. It is a very different landscape. Load growth is happening. It's happening around the world. You're going to hear a lot of people talk about data centers and AI. That's driving in part load growth, but the other things driving load growth are industry movements, or regionally in the U.S., we're seeing some
industries come back. Some move from high tax locations, and we're seeing the residents follow. In the Southeast U.S., where I come from, a significant portion of the load growth is driven by residential and industrial demand, in addition to data centers.
Open-ended question: Who should be responsible for industry alerts? What if I came to you next time and list 10 other issues we need to address? Who should be doing it? Is it EPRI? I'm coming from EPRI, but it probably shouldn't be EPRI. I tell you that because how many of you have gone to EPRI.com to look for something? We don't do a good job of marketing because we're a not-for-profit organization. We're not trying to promote ourselves in those cases. I'm happy to help, but I'm not sure it's EPRI.
We also need to consider increasing needs. I'm going to bring up the dirty word that some of you are going to ridicule me for afterward. Temporary repairs. What if someone can demonstrate they can do a temporary repair and, at the next major outage, replace that component? Would that be acceptable? Think about it, because I'm also here to tell you that we talk about safety with respect to the vessel. In Texas, people freeze to death without power in the winter and die from heat stroke without power in the summer. But if we can't generate power and we don't have a 15% margin – sorry to say it, but that's reality – what are we going to do? We're going to have to consider other options.
A few people before me already said it, but what are you prepared to do to help the industry? What are we prepared to do, because the challenges are very real, not just for post-construction, where my passion is, but also new construction with a landscape, which is much more complicated.
Last, but not least, I am doing something, but it's not enough. We are trying to raise awareness, so we publish a lot. That was in my bio. I do write a lot. I do that because I don't know any other way. Maybe I'm old-fashioned in that regard. Some people share it through TikTok videos and other content, which, quite frankly, I'm not familiar with. But I did write an article for the BULLETIN. I have written some that appeared recently in the AWS Welding Journal, in the May 2025 edition.
For those of you who deal with Grade 91 steel, we released a publicly available document last year, the fourth edition, to help educate people on how to do things right and what to incorporate into specifications if they can enforce them. And then for the tee issue, there are two PVP papers. We plan to pay the open-access fee so people can download them. Those are not Ph.D.-level papers. They are very much like the review I provided to help inform and raise awareness about this very important issue, because, as much as I would like to believe otherwise, it is very much the tip of the iceberg. And once we're prepared to recognize that, then we'll be prepared to fix ourselves. Until then, fixing this problem is going to be incredibly challenging until we're prepared to do it.
Thank you all very much.