a. Zero Doppler coefficient. The Doppler coefficient reduced to zero after 150 days at power due to fuel expansion within the central fuel void. It is expected that it would be available under high power conditions and would still act as a safety factor. However this supposition cannot be tested.

b. Sodium expulsion (39). Figure 4.32 illustrates the scheme by which argon was moved from tank A to tank B. During the movement valve C was inadvertently left open and argon was compressed into the reactor vessel. The vessel safety valves were not designed for the high pressure that ensued, and they had a much lower limit setting, but their design caused them to fail to open at the higher value. The pressure then forced sodium out through a dip guide tube onto the reactor face. The pressure reached 0.6 bar.

About one cubic meter of sodium was ejected and fortunately it did not

Fig. 4.32. Schematic diagram of the path of the sodium in the Rapsodie incident (37, 39).

ignite but simply glowed a rosy red, being hot. There was no smoke. A considerable clean-up job resulted.

Besides the obvious faults of the bad system of pumping argon from tank A to В while in connection with the vessel, and the bad design of the vessel safety valves, this incident was compounded by several other factors: There was no hole in the upright dip tube to equalize pressures inside and out. The cap had been left off because two people from different groups were responsible for the job and each left it to the other. Valve C was left open because the design of the control layout made it difficult to see the indicator which showed whether the valve was closed or not. In addition, administra­tive delays occurred in making a safety modification to correct the situation. These took the following sequence (57):

October 4, 1965. A fire was experienced at the reactor mock-up due to this same cause of sodium expulsion through a dip tube.

October 13, 1965. The mock-up operations group reported the incident.

November 22, 1965. A safety meeting analyzing the incident concluded that bad design was the cause and that holes should be drilled in the dip tube. A recommendation to Rapsodie to this effect was made.

February 4, 1966. Rapsodie group reported that, although inconve­nient, a hole could be made.

March, 1966. Instructions were issued to this effect to the contractors. In the same month the contractors refused and made a counter suggestion.

April, 1966. Rapsodie operators insisted on their instructions.

During the summer discussions ensued between the operators and the contractors without resolution.

October 18, 1966. The incident occurred at Rapsodie on the day prior to a visit by the Minister of Technology. Soon after all dip tubes were cut off without further delay. However…

April, 1967. The same accident occurred in another facility.

The safety lessons inherent in this sequence of events are clear.

c. Secondary circuit leak (39). During filling of a secondary circuit, a filling line became plugged due to an erroneous order of heating. The heating was insufficient to avoid freezing. However, elsewhere in the circuit other insufficient trace heating caused another plug to occur independently. Between the two plugs excessive heat caused pressures to rise to give a leak. Indication of the leak into the double containment was received but ignored as it was obtained on a panel of other known faulty instruments. Thus further leaks were caused throughout the doubly contained pipework. There was no drainage point within the double containment.

d. Fuel handling incident. Figure 4.33 illustrates diagrammatically the operation of the hold-down tube in holding and slightly spreading assem­blies around the one which is to be removed. The hold-down tube is kept in place by compressing a 0.5 ton spring to cock a latch in order to restrain a 0.3 ton upward force on the assembly to be removed.

In order to compress the 0.5 ton spring, it was convenient but bad design to use the weight of the shielding on the refueling machine that weighed 40 tons. This had the effect of overcompressing the spring. On this particular occasion, it did so and was unlucky enough to catch the edge of the hold­down tube against the marking ICZ on the top of a neighboring assembly. The horizontal edge actually caught against the upper horizontal bar of the Z. This had the effect of bending the adjacent assembly head over as shown in Fig. 4.33, so that on the removal of the hold-down tube, the hold-

Fig. 4.33. Course of incident in which the Rapsodie hold-down tube damaged the head of an adjacent assembly on which the engraved letters were ICZ (57).

down tube was able to pick up the adjacent assembly by the bent head. It did exactly the thing it was designed not to do.

The procedure that used excessive weight to compress the 0.5 ton spring was bad, but the incident points toward the need to remember that every event may not be anticipated (especially those associated with a Z engraving on an assembly!).

e. Pump intermittent operation. Initially a pump had been jammed by extraneous material that had to be cleared, but in general there was very little difficulty with the pumps. However, in one instance during operation, flow loss occurred for no apparent reason. The reactor was shut-down, and later the pump came on to full flow in 3.5 min. Then later it again stopped, this time after a few seconds. The reason was difficult to isolate.

Figure 4.34 indicates the brushes which had been sparking. This over­heated the holder which removed the brushes by differential expansion. When the pump cooled down, the brushes recontacted and the pump restarted. Visual checks did not catch this effect even during inspection. The problem was cleared by collecting the sparking on a dc collecting mechanism used as a sensing device.

f. Plugged argon lines. During operation, the argon lines quickly plugged

Fig. 4.34. Diagrammatic arrangement of brushes in the Rapsodie pump drive motor (37).

with sodium oxide crud. This was solved by pumping sodium through the argon system by an EM pump. The occurrence shows that such lines should be designed for flushing by sodium and should be used to pump sodium through from time to time.

g. Jammed rotating plug. The plug was exposed to sodium vapor that froze and gave rise to sufficient crud to jam the rotating plug. This problem was very difficult to solve and needed several attempts, because it got progressively worse during operation. Continuous rotation was first tried. Heating of the joint in order to keep the sodium from freezing failed. Shaping the plug fitting to avoid hold-up ledges, on which the crud might stick, did not work either.

The problem was finally solved by forcing helium down between the plug and the plug support to keep the argon and the sodium vapor contained in it, down in the vessel. This continuous purge also helped to cool the head.

Thus Rapsodie despite its successful operation has had a short catalog of unusual incidents from which lessons may be learned. Others may be expected before the useful life of the plant terminates (37).

These are postulated faults which are being monitored by the operators:

(a) If the flow meter associated with the siphon breaker were to need repair, its inaccessibility would make this extremely difficult.

(b) The pump cable connections are, so far, in close proximity, making a common mode failure a possibility following a cable fire. This could be remedied by separating the cabling.

(c) The serpentine concrete shielding consisted of concrete conventionally laid with organic interlayers (which contained chlorine). Despite the fact that this is in the nitrogen atmosphere, problems may arise since it is not known what the effect of irradiation on the organic interlayers might be. Has hydrochloric acid been produced? Will there be or is there already corrosion in that area close to the vessel? These questions show that the safety engineer’s task does not end with successful operation of the plant.