APR: your source for nuclear news and analysis since April 16, 2010

Tuesday, March 20, 2012

Steam Generators: Design and details

A recent NRC blog post (see here) brings up a matter that has come into focus of late at San Onofre Nuclear Generating Station in California. The problem, briefly, is this; leaks have been detected in the steam generators at Unit 3 at the site, and Unit 2, with identical steam generators and which has been shut down for work anyway, was also inspected and the operator, Southern California Edison, has made the decision to restart neither plant until the cause of the leaks at No. 3 plant and assurance it won't happen at No. 2 plant are both secured. While No. 2 plant did have some steam generator tubes plugged as a part of normal work, and there appears to be no commonality between the few leaking tubes at No. 2 and the leaks at No. 3, SCE has decided to be cautious. (For those interested, No. 1 unit has been shut down and dismantled for some time.)

Now, what makes this story a bit more frustrating for all involved is the fact that all of the steam generators we're talking about here are brand new. These steam generators are Mitsubishi built replacements for the originals, built by Combustion Engineering, which are no longer available. This author has made some requests in the industry to find out some specific details on the replacement generators but so far these have not come back. Click here for an SCE release on MarketWatch.

Now, the NRC blog post gave perhaps the briefest possible explanation of what a steam generator is. We can do better than that here at APR - and we will.

Let's first, right off the bat, mention that there have historically been three manufacturers in this country of large, commercial, pressurized water reactor plants; Westinghouse was first, followed by Combustion Engineering and by Babcock & Wilcox. Each had its own design theories on everything; comparing two of the plants is like comparing a 1970 Camaro to a 1970 Mustang. Let's take a look, then, at some general characteristics of steam generators - we'll look at operative features and the differences between the original equipment built and installed by the vendors.


For this portion, we'll use Westinghouse steam generators. Below, our first illustration which is very basic and which depicts major parts of a Westinghouse steam generator.

The function of the steam generator is to transfer heat from the primary system (which is water that goes through the reactor core) into water in the secondary system (water that is turned into steam, used to perform work, condensed, and returned to the steam generator, never mixing with reactor water.) So then we need a piping system inside the steam generator to carry the primary water - which is at very high pressure and temperature - and efficiently transfer heat to the secondary water. This is best done with many thousands of small tubes, made of a special material (Inconel 600, very often) that will not corrode in this type of environment. As we see in the drawing above, thousands of tubes form the "tube bundle" so clearly visible taking up most of the volume of the lower 2/3 of the inside of the steam generator. These tubes are all mounted into a thick, heavy tube sheet (sometimes called a tube plate) and, as we can see, the area below these is divided in half. Hot primary water enters the inlet plenum, then goes through the tubes transferring energy to the secondary water outside the tubes; it then exits the tubes into the outlet plenum, and then goes on to the cold leg of the primary system. Since the primary tubes are shaped kind of like an upside down letter "U" they are widely just called "U tubes."

Now, before we get too far, it's important to note that the temperature difference between the water going in and the water coming out of the steam generator on the primary side isn't as high as most people think. Generally, pressurized water reactors have a differential temperature, at full power, between the water going into the steam generator and the water coming out well under 100 degrees. For example, the "hot leg" temperature at Shippingport Atomic Power Station, the first commercial US pressurized water plant, was 537F while the "cold leg" temperature was 509F, for a difference of just 28 degrees. However, the plant was moving about 7,350,000 lbs of water per hour through each of three operating loops; so, we can begin to understand that while the temperature change of the water was not that great, the total energy transferred WAS because of the massive amount of water being moved.

In later plants of course the temperature difference gets wider and the amount of water being moved becomes truly massive. By the mid 1980's the Westinghouse Model 412 design (4 loop, 12 foot core height, 3411 MWt core power) was being advertised as having a hot leg temperature of about 619F, cold leg temperature of about 558F for a difference of 61 degrees. This plant design had four loops, each of which had a seven thousand horsepower electric motor-driven pump moving about 34,600,000 lbs of water per hour.

Now that we've had an idea of what's going on inside the primary side of the steam generator, let's take a look at the really complicated portion - which is the secondary side. This area of the steam generator has to admit feedwater from the feed and condensate system, allow for heatup and mixing of the incoming feedwater (so as not to thermally shock the U-tubes), provide for an area to allow water to rise up through and boil in the tubes, separate out the water droplets from the steam (and return the water to the steam generator while allowing only steam to exit), provide for blowdown (letting out steam or water for cleaning), allow for measurement of water level in the steam generator, allow for sampling of water (to control chemistry to prevent corrosion), and some other things. It's a massive, complicated piece of equipment! Let's look at a really detailed illustration which comes from a visitor's center brochure covering Connecticut Yankee - an old Westinghouse PWR plant. As with all illustrations here, click to enlarge.

On the right side of this illustration we see the feedwater inlet; this supplies a ring that runs around the steam generator, spraying the water downward into the downcomer region that is outside of the tube bundle, but surrounds it completely. As the feedwater goes down, it both mixes with the water raining from the steam separators (more on those in a moment) and heats up. At the bottom, it makes the turn and moves up into the tube bundle area where it is heated to the boiling point; it blasts upward toward the top of the tube bundle. (Follow the arrows.) The steam, laden with water droplets carried along, enters a set of cyclonic or swirl type moisture separators that force the mixture to swirl rapidly; the steam can make the turns, but the water cannot and is thrown off and rains back down into the downcomer area. After this, a second set of moisture separators is encountered in this design which is labeled as the "mist extractor assembly" at the very center top of the steam generator. Again, steam is allowed to continue on to the outlet to the steam system, while water returns to the downcomer region. All this water returning isn't just a gentle rain - it's a rainforest gully-washer of Biblical proportions. Well, anyway, it's a lot of water. Many other design features are also shown.

Let's now take a brief look at some of the basic differences between the Westinghouse, Combustion Engineering, and Babcock & Wilcox steam generator designs and plant layouts. Again -- this is a basic introduction.

WESTINGHOUSE: In general, Westinghouse plants have either two, three or four steam generators depending on how powerful the plant is. Each steam generator has its own specific hot leg, cold leg and reactor coolant pump. Below is a general basic view of a Westinghouse four-loop plant (as was described second above when we were talking about temperature differences and flow rates.) The steam generators, internally, are quite like those already shown.
Below, a typical Westinghouse steam generator as depicted in WASH-1082.

COMBUSTION ENGINEERING: CE plants always have two steam generators, no matter how powerful the plant is. Here is a view from WASH-1082 of a typical Combustion Engineering reactor plant. Note that while there are only two steam generators, and two hot legs, there are two cold legs and two pumps for each steam generator.

Below, from a Combustion Engineering - Consumers Power specification brochure covering Palisades is a view of one of that plant's steam generators.

Below, from WASH-1082 is a slightly more detailed illustration of a typical CE steam generator.

BABCOCK & WILCOX: These are the most unusual steam generators. These steam generators do not use U-tubes; instead, they use straight tubes - and the primary coolant passes vertically down through the steam generator in these tubes, from the top to the bottom. These steam generators are very tall - and the run of primary coolant pipe that forms the hot legs is enormous. Below, a general perspective of a Babcock & Wilcox reactor plant.
Below, the "Raised Loop" configuration of the very late Babcock & Wilcox plant design.

Below, from WASH-1082, we see a partial cutaway of a Babcock & Wilcox steam generator. These are generally known as "single pass" steam generators, or else as "once through" steam generators. Note also that on Babcock & Wilcox plants there are always only two steam generators, with one hot leg and two cold legs (and thus two pumps.)

These single-pass steam generators do not have anywhere near the amount of water on the secondary side at all times that Westinghouse or Combustion Engineering steam generators do; according to "The Second Nuclear Era," Combustion Engineering steam generators have a secondary water inventory five times that of the B&W steam generators. This makes the pressurized water plants with U-tube type steam generators less rapidly responsive to secondary side transients .. such as loss of feed, or overfeed after scram .. since there's more secondary water to absorb transients caused by rapidly changing temperature of feedwater.

On the other hand, though, the single-pass steam generators are much more efficient and actually allow a tiny degree of steam superheating -- there is no superheat to the output steam at all on U-tube type steam generators. Superheats of between 30 and 60 degrees are achievable with vertical, single pass steam generators; while this does not allow the use of "dry steam" turbines without multiple stages of separation and reheat, it does drive up overall plant efficiency.


As you have probably already guessed from looking at the illustrations which include dimensions, steam generators are exceedingly large and heavy components. They are also difficult, time consuming and exceedingly expensive to manufacture.

The three major reactor vendors who offered pressurized water reactor plants all manufactured their own steam generators (except for the earliest Westinghouse plants.) A generic description of a 250 MWe steam generator in "The Nuclear Industry, 1969" (USAEC) gives rough dimensions of 68 feet in height, 14 feet in diameter and a weight of 330 tons. Westinghouse, in the middle 1980's, specified its Type F steam generator as weighing 346 tons dry, 422 tons normal operating weight and 560 tons flooded full. This steam generator could produce 3,813,000 lbs/hr of steam at between 920 psia and 1000 psia.

Because of the complexity and mass of this piece of equipment, it had to be one of the earliest ordered during the construction process for a plant. According to WASH-1174-71 ("The Nuclear Industry, 1971") the time that steam generators had to be ordered from the manufacturing facility was slightly over four and a half years before the expected criticality date. (Only the reactor vessel, turbine generator, and condenser & auxiliaries had to be ordered earlier.) This same volume mentions that in 1971 the three companies manufacturing steam generators (Westinghouse, with facilities at Tampa, Florida and Lester, Pennsylvania; Combustion Engineering, with plant at Chattanooga, Tennessee, and Babcock & Wilcox, with plant at Barberton, Ohio) could together supply about 60 steam generators total among them per year which at that time was enough to supply the expected build rate for the next ten years.


It suffices to say that the worst thing that can happen in a steam generator is for the U-tubes to leak or rupture, allowing primary coolant to get into the secondary side water and steam. This is called a primary to secondary leak, and is the reason we're talking about San Onofre at all. I should mention that very tiny leaks are detectable - they're then found by hydrostatic testing when the plant is shut down, and the normal practice is to plug the tubes. Steam generators are always built with so many tubes that even plugging a hundred of them won't affect the steam generator's ability to provide the rated amount of power.

There are a few ways to get U-tubes to leak. Poor chemistry control on the secondary side can lead to corrosion, and leakage. Poor design can cause crushing of the tubes under flow impingement. Poor fabrication and manufacturing can lead to leaks in anything man ever made - and this could be true for steam generators as well. I could go on, but it's important to note here that SCE does not yet know why there are tube leaks on the San Onofre generators and we'll just have to wait for their report or something more from the NRC before we can make any kind of judgement call.

In Depth: Dan Yurman at IDAHO SAMIZDAT wraps up the entire situation at San Onofre.


The first commercial pressurized water reactor plant that had unitized, vertical, U-tube steam generators was Yankee Rowe (Yankee Atomic Electric Company.) That plant went on line in 1960. Prior to that, steam generators in commercial plants had separate heat exchangers and steam drums. Let's take a look at some of these.

First, we'll take a look at two different designs that were used on the Shippingport Atomic Power Station. Two of that plant's four loops used Foster-Wheeler steam generators, while the other two loops used Babcock & Wilcox steam generators. (Note: As first built, Shippingport only required three loops to operate at full power and these are the figures quoted earlier.)

These two illustrations are from the complete Shippingport Atomic Power Station press package in the APRA collection. Note the many spidery pipes that are used to pipe steam/water mixture up to the drum or separator section, and the many downcomer pipes used to bring returned saturated water down from the drum to the heat exchanger.

The Foster-Wheeler design used straight, once-through or single pass design on the primary side with straight tubes and straight shell. The Babcock & Wilcox design was also single pass, but the whole heat exchanger portion was bent into a "U" so that both the tubes and the shell were "U" shaped. The different designs were fabricated this way to allow investigation of the relative merits of the two designs.

Below, one of the lower heat exchanger sections being removed from Foster-Wheeler Corporation's plant at Mountaintop, Pennsylvania.

Babcock & Wilcox used the same general design that they supplied for Shippingport in their own Indian Point 1 and NS Savannah plant designs before changing over to the once through design. Below is an illustration of one of the Savannah's two steam generators.

The illustration below is from the US AEC's photographic essay for the 1958 Geneva conferences entitled "Atoms for Peace / USA 1958" and shows what is described as a heat exchanger for Indian Point being drilled. This kind of operation, on a larger scale, had to be performed to make the tube sheets (or, "tube plates") of every steam generator. The large number of penetrations through the plate weakens it; the plates are made extremely thick to compensate.

We have mentioned Foster-Wheeler; this company did not stick around in the fabrication of large primary plant components, but did manage to get an order to build the secondary steam generators for the dual-cycle Dresden-1 which was constructed by General Electric. These were vertical U-tube type steam generators, and were in service prior to those at Yankee.. so that these might be considered the first vertical U-tube steam generators in any commercial plant, while those at Yankee Rowe were the first at any commercial PWR plant. Below is one of the Foster-Wheeler steam generators built for Dresden-1.

The advantages of having the heat exchanger and the steam/water separation performed inside a single, more compact unit seem obvious, so that there is no need to explain the transition to vertical U-tube steam generators as soon as they were developed.

I hope this has been a good, basic introduction to PWR steam generators; now, people everywhere can get a handle on just what is going on inside one of these things, how they're built, and perhaps begin to understand what's being described about San Onofre's primary-secondary leaks.

6:00 PM Eastern Tuesday March 20, 2012


  1. Will, just out of curiosity how do they handle the plugging of the tubes? Depressurize, climb in and do it? Or remotely somehow?

    What do they do with the primary water during this plugging? Is the primary water radioactive and if so to what degree?

  2. Yes, you'll do it with the loop isolated, and drained on the primary side. There are manholes to get access to the ends of the tubes in the inlet / outlet plenums. The radiation consideration isn't really the primary water, but rather the radiation from irradiated plant metals and from any corrosion product deposition - so sometimes if "stay time" is very short inside the plenum plugging can be done remotely. Welding the tubes is one way to seal them, while Babcock & Wilcox had an explosive slug method to plug tubes.

  3. I should have said drained, not necessarily "isolated."

  4. And now a further update on conditions that are not getting much better at San Onofre, courtesy of Dan Yurman at Idaho Samizdat:


  5. Hmmm, this is very worrying indeed. Excessive wear on more than a hundred tubes. Wear from what? Foreign debrie (left behind after assembly?), incorrect assembly of support plates?

    Does anyone know?

    MHI (the Japanese) takes QA work seriously so I am not just a little surprised about this...

    1. > 129 tubes in ... Unit 2 at San Onofre.
      > Previously ... Unit 3. Both ... by
      > Mitsubishi Heavy Industries ... 2009.

      Any idea how many other plants use the same products? With almost all Japan's fission plants shut down I'd guess they're all getting a close look over there as well.

    2. The March 20 Idaho Samizdat article has numbers:

      "9,700 tubes .... several tubes show unusual wear with up to a third of the tube wall worn away. Another 69 tubes showed 20% wear, and there was 10% wear in another 800 tubes."

  6. Thanks for the info Will. I assume they must have large isolating valves on the hot and cold legs between the reactor and the heat exchanger? If so they must be damn good valves.

    Also your last update. Wear within 3 years doesn't sound good.

  7. Rod, was the original idea that steam generators could be inside the containment because they would not be replaced during the lifetime of the plant? I've been wondering why the Crystal River containment had to be sawed open to replace a steam generator -- can plants be designed nowadays with replacement in mind for such large parts -- say a part of the containment that can be cut out without affecting the integrity of the rest of the containment as happened there? Or are these pieces just too big to anticipate replacing?

    (Thinking of how planning ahead, an architect can have a header be built into a wall above a span so later a hole can easily be cut out to add a door or window without reworking the structure)

  8. @Hank: Rod? Rod's not here. But I'll answer your question anyway. Steam generators not only could be inside the containment, they had to be; for starters, they contain primary coolant. Secondly, they do become high rad sources because of the deposition of irradiated and thus radioactive corrosion product deposits. Originally there wasn't serious consideration of replacement. As you can see from the article I've written, the steam generators are physically larger than the reactor vessel itself and replacement of these components is something only done if absolutely necessary to extend plant life -- that is to say, it's only done after the determination that the rest of the plant has a lot more life in it, and that even with the cost and burden of replacing the steam generators the plant is still able to be profitable.

  9. Are you available to review the findings for the public when we get the final steam generator report back from the NRC? I don't expect you to sign off on it, but it would be reassuring to have an expert like yourself to let us know if there is anything about the report that needs to be questioned further. Thx, gary@sanclementegreen.org

  10. @Gary: I intend to report on this until a conclusion is reached. I think this brings up some good and relevant questions about extending plant life by way of replacement components, and as we find that other plants require extensions of plant life or else replacement (of the whole plant, by new nuclear or something else)for their electric generating capacity we'll face this again. That's why I think it's very important to see this through and keep reporting on it.

  11. Thanks Will
    apology for the wrong name; multitasking and dropped one ...

    > Originally there wasn't serious consideration
    > of replacement.

    That makes sense. Does it now look like it's specifically the internal tubing (not the outside case and big pipes) that are more of a problem for failure over the long term?

    Has anyone looked at designing steam generators so, say, they could be uncapped remotely, and individual tubes or sections pulled out through a smaller opening in the containment to replace them -- in the same way that fuel rods are pulled out, for the same purpose?

    That won't change the existing situation -- but seems something much like the steam generators you describe has to be part of more advanced nuclear plants. Wondering if there's a lesson learnable here.

    Are they doing anything different with the newest designs that would make an unexpected high failure rate of tubes easier to cope with, should that happen again?

    I know the most modern supercritical coal plants run at temperatures much higher than current fission plants, and are pushing the design of corrosion resistant high temperature components; they don't have containment issues, but perhaps they have to plan to replace individual tubes.

    I recall Gen4 fission and "Wished4" fusion plants would run hotter than current fission plants -- and would be built using the materials tech. worked out for the supercritical coal plants. Greater thermodynamic efficiency.

    Just, y'know, looking for optimism here.

  12. Tell me if I have it wrong, but Gary looks to be from a green group opposing SONGS nuclear energy and Will is obviously pro nuclear. Yet both are in search of the truth. There should be more of this. Well done.

  13. Will, very informative website. My question is where is Areva in all of this? Do they make steam generators today? Did the other companies go out of business or stop making them? That is, because the new ones are made by Mitusbishi I guess not everything has to be the same brand. So is it possible to "mix and match" components? Also, were steam generators meant to last "forever"? Obviously they have not, but would it be possible to put a large hatch on the containment building so components could be replaced?

  14. @Robert: Thank you. The S/G's were really never designed with a finite life span .. at least, no one discussed that nearly as much as the life of the reactor pressure vessels. But no one initially ever planned to replace S/G's wholesale. I believe that Westinghouse can still manufacture its own S/G's in house, but the whole heavy product line of Combustion Engineering is gone. I think as of now Babcock & Wilcox also advertises the ability to make some steam generators. I would have to check. Of course, I'd also imagine that Mitsubishi underbid anyone else who did bid. Areva may never have been in the bidding for this, since they're on the wrong side of the wrong ocean. I am still trying to nail down the origin of the exact design or pattern of S/G supplied for San Onofre but right now no one is talking. Except of course for the "experts" on the anti-nuclear side who are making things up as they go along.

  15. @Keith: It does look that way, yes. I do believe in open dialogue and it's nice to have a rational discussion with someone on the other side of the issue!

  16. @Hank Roberts: The U-tubes have always been the component of pressurized water reactor steam generators most likely to fail, because of a number of factors. For example, the difference in temperature across them, and the difference in pressure. The thinness of the walls. Any susceptibility to corrosion in a high temperature steam environment (external to the tubes.) Likelihood of deposition of material at the base of the tubes, at the tube sheet. The nature of the insertion into and bonding with the tube sheet. Contraction and expansion. Flow related wear. There are more.

    The Soviets have deliberately avoided tube sheet type designs as much as possible by using horizontal steam generators with large central headers.. but these have quite their own range of problems, and later Soviet designs used vertical U-tube steam generators of yet a different design. No one has designed the foolproof pressurized water reactor steam generator yet.. but the fact that no steam generator leak has ever led to an accident should tell us something about how much of a real risk this is... or is not.

    The problem with using smaller S/G's and more of them is not only cost and complexity but efficiency. Fewer large units are better.. and with almost ten thousand tubes in each generator at San Onofre, I'd have to imagine that there's probably a fifteen percent reserve you could plug. That's a super rough guess. Don't quote me on that just yet.

    For me, right now, the lesson to be learned is that you have to replace steam generators taking into account all possible primary and secondary parameters such as the actual flow velocity achieved inside the U-tubes and make sure the generators will fit in that application in every way. One wonders if a massive test rig that could pump highly variable flow rates through a real steam generator, while taking in variable amounts of feedwater and letting out variable amounts of steam in order to match any given replacement situation could be developed. (The answer of course is "Yes, quickly" but at what cost to Mitsubishi or whoever else wants in on this replacement business?)

  17. I have been told that approximately 100 tubes in one area of one SG on Unit 3 has an unexpected "in plane" vibration in the U tube area to the point that they contacted each other. These could be plugged and the plant started but everything is on hold until the mechanism is understood and a fix for eliminating further contact is designed. Likely there will be a fix, then a run of some months, then a shutdown and inspection to verify that the fix is working. The issues on Unit 2 are different. Unexpected wear at some anti vibration bars but restart is delayed until the results of the studies at Unit 3.
    MHI also manufactured the SGs at Calhoun but the ones at San Onofre are significantly larger and the two phase flow characteristics on the secondary side will be different.

  18. @"Bob" -- that can't be your real name, can it? That's a name from a TV show.. so I assume you're unable to post inside info with your real name because your job would be threatened. Can you just let me have a hint of how you came by this information?

  19. re SONGS SGs: Are the [new] RSGs using egg-crate tube separators/dampeners while the OSGs used perforated plate design? That could distort flow,vibration,thermodynamics in difficult-to-test/predict ways.
    Great blog. Thanks to all.

  20. Nice Post About Generator parts.Thanks for the post, I will look forward to see more posts from your blog.

  21. I just wanted to say I enjoyed reading your description of the workings of steam generators. My grandfather, Don Harrod, was a metallurgist who worked on tube design for Westinghouse. I've always been interested in his work, but he never would talk about it with me. Since he passed away, I've enjoyed reading the papers of his I could dig out of the closet, many of which share the same diagrams shown here. It's nice to see his work still referenced. Thanks again for the article.