Appendix 6 The hot leg vs. cold leg primary pump controversy

The decision of whether to place the primary system pump in the hot leg or the cold leg is one that occupied considerable attention on CRBRP. The cold leg would generally be preferred as it operates 250°F lower in temperature than the hot leg and is exposed to a much less severe transient environment. Even for a simple reactor scram, the hot leg will experience a down temperature transient on the order of 200°F in the space of less than a minute as a result of the prolonged coastdown of the centrifugal primary system pumps. Since stainless steels having both low thermal conductivity and a high thermal expansion coefficient, are widely used in the primary circuit, thermal shock should be avoided to the extent possible. It is worthy of mention that this thermal shock problem was explicitly and successfully dealt with on the SRE (which was operated at full power with a reactor delta T of 460°F.) and the Hallam reactors with a device that was called an eddy current brake. 1 This “eddy current brake” was essentially an EM pump working backwards, that was activated on the occasion of reactor trips and it effectively eliminated the hot leg temperature transients. The precedent for placing the pump in the hot leg was set by the SRE and followed by Hallam. The eddy current brake was located in the cold leg.

Lower temperature in the cold leg means higher density and lower volume rate of flow for a given weight rate of flow, simplifying pump design. In a tightly packed reactor core such as is found on breeder reactors, the core pressure drop traditionally has been on the order of 100 psi. If the pump is in the hot leg, the primary side of the IHX will be the highest pressure point in the primary system next to the pump discharge itself. Since the secondary side of the IHX must be maintained at a higher pressure than the primary side, putting the primary pump in the hot leg raises the secondary side pressure by 100 psi above what would have been the case with a cold leg primary pump. Increasing secondary side pressure complicates steam generator leak detection and imposes greater challenges on the design of the sodium water reaction products system. It also increases the consequences of any secondary system leak.

This controversy is closely related to the discussion in section 7 and is the result of the interplay between two perceived requirements.  First is the desire for as much of the primary piping system as possible to be inspectable, which means that much of the primary piping system will be unguarded.  Second is the need to accommodate the postulated double ended pipe break in the design.  The reactor and pump cover gas systems are connected and both operate slightly above atmospheric pressure so as to prevent the reactor level from dropping below the outlet nozzles in the event of a double ended pipe break somewhere in the PHTS.  The pump is of a sump suction design so as to eliminate the sodium seals in the pump that proved so troublesome on the SRE plant.  In a sump suction pump, the shaft seal is actually sealing the cover gas.  The sodium level in the pump is determined by the head losses between the reactor outlet nozzle and the pump itself.  With the pump in the hot leg, the pump level drawdown is about 12 feet, which rises to over thirty feet for the cold leg pump location.  While 12 feet is considered manageable, a 30 ft. drawdown would result in a much longer shaft and was generally considered on the CRBRP project to be an unrealistic option.  For these reasons, on both CRBRP and FFTF, the pump was located in the hot leg. 

Earlier sodium cooled reactors avoided this problem in various ways.  For both SRE and Hallam, mechanical pump seals were used that were directly exposed to sodium.  In the case of the SRE, the mechanical seals directly led to a flow blockage in the core when the oil (tetralin) that was used to lubricate and cool the seals found its way into the primary sodium system and blocked flow to core assemblies.  The fuel failures that occurred in that reactor are celebrated by anti-nuclear activists in the San Fernando Valley to this day.  Since the SRE experience, there hasn’t been much interest in deploying pumps with sodium seals.  This is somewhat academic anyway since the PHTS pumps were located in the hot leg on both the SRE and Hallam.

The Fermi-1 designers solved the drawdown problem by installing double-walled piping throughout the primary circuit.  With double walled piping, there is no incentive to maintain reactor cover gas pressure at atmospheric pressure or to equalize the pump and reactor cover gas systems.  The reactor cover gas pressure was allowed to increase as the primary pump speed was increased; essentially eliminating pump drawdown to levels even less than the hot leg pump plants.  The Fermi-1 approach had the additional advantage of allowing vertical expansion loops in the primary circuit resulting in a significantly more compact containment than was achieved on FFTF and CRBRP.  The Fermi-1 containment was just 72 ft. in diameter.  In contrast, the FFTF containment had a diameter of 135 ft.  The designed thermal power of Fermi-1 was 300 MW which compares well with FFTF’s 400 MW.  If one assumes that containment footprint should be proportional to thermal power level, then its diameter should increase with the square root of power level and FFTF’s containment diameter should have been about 83 ft.  At 83 ft. diameter, FFTF’s containment footprint would have been about 62% smaller that it turned out to be.  While there may have been other reasons why the FFTF designers did not adopt the PHTS layout approach used at Fermi-1, the inspectability of the primary system welds played a role as was discussed earlier in section 7.  In addition to the FFTF, there had been a precedent for a hot leg pump.  The Karlsruhe reactor, KNK-1 which was modified to KNK-2 had a hot leg pump.  The German follow-on plant, SNR-300, also had a hot leg pump as well as the then planned German commercial sized plant.

An interesting concept proposed and patented by Westinghouse2 involves the use of a restricting barrier placed between the outlet nozzles and the sodium surface inside the reactor vessel. The sodium above the restricting barrier drained into the reactor reservoir tank which had a cover gas that was equalized with the reactor and the PHTS pump. The reservoir tank was connected to the PHTS pump suction through a control valve. How the reactor was to be refueled and where the UIS was to be placed was not described.

Another avenue that might be explored would be to eliminate the double ended pipe break from the design basis.  The precedent for establishing the double ended guillotine pipe break as a design basis event was established by the licensing of Light Water Reactors (LWRs).  It would be a straightforward matter to argue that a double-ended guillotine failure of low pressure sodium piping is mechanistically impossible and therefore not applicable to LMFBRs because of their much lower primary system pressure and lack of sufficient energy in the coolant for a pipe failure to propagate into the double-ended guillotine type failure.  The argument would be made that if a PHTS leak were to occur, it would be quickly detected, the plant would be shutdown and cooled down, the leak would be isolated by draining the effected part of the PHTS to the overflow or sodium drain tank, and repairs would be made.  Some effort was committed to this approach by at least one of the national labs during the time the CRBRP project was underway, but it was not pursued to a conclusion.  This item has been included in appendix 9. 

Any future loop-type plants are likely to incorporate the primary pump into the cold leg despite the precedent set by FFTF and KNK-2.  The motive for retaining the elevated primary system piping concept is suspect and the advantages of the hot leg location are exceeded by the compelling arguments favoring the cold leg location.  An addition to double walled piping would be to adopt an EM pump for the primary circuit. EM pumps have no shaft seal and therefore require no cover gas.  While the use of EM pumps for heat transport system applications has been avoided in most sodium cooled reactors to date because of their poor conversion efficiency of electric power input to pumping power (typically 40% at best), they are compact, have no penetrations, require no cover gas, require no lubricants, and are likely to require little if any maintenance.  Such pumps could fit nearby, directly underneath, or be incorporated into the IHXs further simplifying containment design. There is no requirement for equalizing cover gas pressure between the PHTS pump and the RV if there is no pump cover gas. A slight positive pressure in the RV may be necessary to provide adequate NPSH for the pump during plant operation. That pressure can be reduced to atmospheric when the reactor and the PHTS pumps are shut down or running at low flow.

As a final note on this subject, maintaining reactor cover gas at atmospheric pressure during operation does have a drawback.  In 1992 at Superphénix there was a cover gas leak caused by failed diaphragms in a compressor used to transport cover gas to a radiometer.  The systems at the plant did not detect the leak for three weeks.  By the time the problem was discovered, 400 Kg of sodium oxide had been formed in the primary coolant.  Restoring oxide purity required two years and required replacement of the cartridges in the cold traps, all the time with the reactor shut down.  If there were a leak in the cover gas system, it would be much better for the gas to leak out of the system than for oxygen to leak in.  Out leakage of cover gas will be readily detected by radiation air monitors whereas in leakage is much more difficult to detect, particularly at low rates.  Small quantities of oxygen readily combine with the hot sodium before it reaches detectable levels in the cover gas space and the nitrogen in the air can cause nitriding of structural components. 


1 R.E. Durand, Sodium Reactor Operating Experience, Chemical Engineering Progress, Vol. 57, No. 3, Mar 1961. See also R.J. Beeley, J.E. Balmeister, Operating Experience with the SRE and its Application to the Hallam Nuclear Facility, Atomics International, 1961.

2 U.S. Patent 3,951,738, Nuclear Reactor Coolant and Cover Gas System, George, J.R, et al, April 20, 1976