Appendix 4 Primary Steam Generators

In the pool vs. loop discussion it was remarked that the primary steam generator was seen by some as the “holy grail” of LMFBR design. Conventional wisdom suggests that elimination of the intermediate loops would be desirable from at least two points of view. First, the capital cost of the plant would potentially decrease simply by elimination of all the IHTS components and real estate needed to house them – the IHXs, the IHTS pumps, and the associated piping. Second, the steam conditions would be better. If reactor outlet temperature is 1000°F and there is no IHX, steam temperature can be increased by whatever LMTD exists across the IHX – in the case of the approach proposed, 40°F. Raising the temperature of 2500 psig steam from 900°F to 950°F raises its enthalpy by about 35 BTU/lb, which, if returned to the steam plant would translate into about 3% better thermodynamic efficiency. A 3% improvement in thermodynamic efficiency would increase electric power output more than 7%, so a 1200 MWe nominal plant would become a 1300 MWe nominal plant. This is a sizable payoff. The payoff is increased further by the elimination of the power required to drive the intermediate system pumps.

However, there is a cost involved. First, with the departure of the IHXs the steam generators become a part of the primary system and would need to be enclosed within the containment boundary. Bringing the steam generators inside containment raises the specter of a steam or hot feedwater leak inside containment likely compromising the position taken earlier that the containment need not be designed for high pressures. Second, from the section 7 discussion of the heat transport system, primary steam generators may very well translate into four primary loops as opposed to the two in the design approach advocated to this point. The result is the containment boundary grows. Moreover, the isolation valves in the IHTS system now become part of the primary system as does the sodium water reaction products system (SWRPS) and its associated tanks. The SWRPS flare stack introduces another problem since it would need to be contained, which would be no trivial matter as the combustion gases following flaring would be very hot. It is probably reasonable to conclude that if a SWRPS is needed, a primary steam generator is simply not a tractable idea.

There was one sodium cooled reactor design that probably had primary steam generators and that was the Sea Wolf submarine’s first reactor plant and its associated prototype. (The prototype was designated “S1G” and the plant on the Sea Wolf “S2G”). The word “probably” is used because not much is known about that plant design except to the people who were involved in its design, construction, and operation since it remains classified. But from what little that has been disclosed, it appears there was an evaporator, steam drum, and superheater all fabricated of type 347 stainless steel. The choice of stainless steel was a mistake and the superheater wound up being out of service for the life of the S2G Sea Wolf plant probably because of caustic stress corrosion, but absent the superheater, the plant was still capable of making 80% of full power. The individual units in the steam generating system probably had double walled tubes and a double tube sheet with NaK between the tubes and the tube sheets. NaK would have had to have been used as the intermediate fluid to allow the steam generators to have been placed into something approximating wet layup, assuming that the sodium side of the steam generators could have been drained somehow. It is unlikely that Sea Wolf had any system for accommodating sodium water reactions. Other than the problems with the superheater, the only significant problem experienced on S2G was a primary sodium leak.

A double walled steam generator design was developed as a part of the CRBRP project by Westinghouse and was tested to well in excess of 15,000 hours at ETEC without incident. The CRBRP project had no intention of using the Westinghouse design as anything other than a backup steam generator in case difficulties developed with the reference hockey stick design. There was no serious thought given to its use as a primary steam generator. The unit had provisions for monitoring the interspaces between the tubes but there was no double tube sheet and there was no NaK between the tubes. Although the unit did not show any evidence of leakage or degradation of any kind from either the water or the sodium sides, 15,000 hours of testing is a far cry from the 60 or more years that should represent a typical nuclear plant lifetime, so the ETEC testing is considered inconclusive, at least as far as application of the design for use as a primary steam generator is concerned. Moreover, there was no post test evaluation of the unit. A much smaller unit (about 3 MWth) furnished by the Japan Atomic Power Company (JAPC) was tested in parallel with the Westinghouse unit at ETEC. After 10,000 hours of testing, it was returned to Japan for post test evaluation. JAPC did not make the results of their post test evaluation public, but presumably they are available.

A double walled unit fabricated of 304 SS was installed in the SRE with mercury as the intermediate fluid. Hallam also had a double walled unit. Neither SRE nor Hallam was intended to be a primary steam generator. The Russian SVBR lead bismuth reactor was designed with duplex tubes. EBR-I had stainless steel double walled tubes with a copper layer between the tubes. The tubes were connected to a header rather than a tube sheet. The design led to a physically large unit in comparison to the amount of steam produced and it was not economically practical for application to large units. The Dounreay Fast Reactor in the U.K. had parallel tubes in a copper heat transfer block. Although effective in preventing sodium water reactions, the design would not scale up economically. EBR-II had 2¼ Cr 1 Mo double walled tubes in a recirculating configuration with eight evaporators, two superheaters of equal sizes and a single a steam drum. The units had double tube sheets with front faced tube to tubesheet welds1. Four of the evaporators used mechanically bonded tubes while the other four were metallurgically bonded. Because of the bonding, there was no realistic way to monitor the inner space between tubes for leakage. Unit performance was generally satisfactory2 but the design features (metallurgical bonding, mechanical bonding, front faced tube to tubesheet welds, double tube sheets, lack of rapid detection of a leak to the inner-space) have generally fallen from favor. The choice of the material of fabrication, equally sized superheaters and evaporators, and the recirculating configuration influenced the CRBRP. Otherwise, there is no known double wall tube experience worldwide that is directly applicable to LMFBRs.

It is instructive to consider a double walled tube concept along the lines supposed to have been used on S1G/S2G with a double tube sheet and NaK as the intermediate fluid. Presumably, a modern version of such a concept would be fabricated from some kind of high chrome ferritic steel as opposed to the 347 SS used on S1G/S2G. It would probably be a straight tube unit, which would cause it to be longer than its helical coil counterpart. It would be necessary for the NaK to be a fully contained system on each steam generator probably connected to some sort of surge tank covered above by an inert gas. The pressure in the surge tank would probably be maintained somewhere near atmospheric. A system for detection of leaks between the NaK and the surrounding air would need to be provided, which might argue for maintaining the NaK pressure slightly below atmospheric, however air in-leakage into the NaK system could be difficult to detect (remember Superphénix). More likely the system would be maintained slightly above atmospheric to prevent air contamination of the NaK. The cover gas system would probably not require monitoring except for the control of its pressure.

A leak from either the water side or the sodium side would be immediately detectable. If the leak were from the sodium side, the pressure would increase to the IHTS pressure at the site of the leak which would be no less than the minimum IHTS pressure, probably at least 20 psig. If there were a leak from the water side, there would be a strong chemical reaction between the water and the NaK. The NaK surge tank would need to be provided with some sort of relief protection in order to be able to accommodate the reaction products from such an event. Because of the relatively limited quantity of NaK in each of the steam generators, flaring of the generated hydrogen would probably not be required, but the surge tank relief would probably discharge to some kind of reaction products tank. The steam generator would be immediately isolated on both the water and sodium sides once any leak had been detected, regardless whether from the sodium or the water side.

Recovery from any leak would involve a protracted outage and could require replacement of the steam generator, which would be complicated by the fact that it is located inside containment. The plant operator would be obliged to wait the ten days necessary for the Na24 activity to decrease low enough to permit entry into the cell. The primary sodium side would need to be drained. The failed tube(s) would need to be located somehow. Plugging tubes, particularly those connected to the inner tube-sheet would not be a simple task, particularly since the tube-sheets would have been designed to be in close proximity to one another to minimize unit length and NaK inventory. Once the damaged tubes had been repaired, it would be necessary to restore the NaK system, which would have become heavily contaminated as a result of the leak, particularly one from the water side. This would require bringing in makeup NaK, NaK purification equipment, and some sort of NaK pumping capability along with connecting piping, valves, required safety provisions, and so on. NaK has a bad reputation of being many times more difficult to handle than sodium and is not the sort of thing that would be welcomed inside containment.3

It is from all these considerations that the primary steam generator is not a very appealing idea, at least with the technology that exists at the current time. As has been stated earlier, one of the most important advantages of sodium is that it is benign to the materials that contain it. The same cannot be said of water. So long as sodium is the only working fluid inside the piping systems within containment, there is reasonable confidence that the interior of the containment will be a relatively uneventful place. Once water is introduced into the containment the picture changes.


1 To accomplish a “front faced weld” the tube is passed through a hole in the tubesheet and a fillet weld is made between the outside of the tube and the tubesheet.  The main problem with this procedure is the crevice between the tube and the tubesheet where contaminants such as halides can hide out and later cause stress corrosion cracking of the tubes.  “Back faced welds” are made by machining tube stubs onto the inside surface of the tubesheet then butt welding the tubes to these stubs.  The resulting welds are more inspectable, have better integrity, and the crevice is eliminated.

2 Buschman, H. W.; Longua, K. J.; Penney, W. H.; Operating Experience of the EBR-II Intermediate Heat Exchanger and Steam Generating System; ASME/IEEE Joint Power Generation Conference; September 25-29, 1983.

3 NaK forms a superoxide, KO2, which is potentially explosive.