a. Coolant contamination. Coolant contamination persisted. It was eventually cleared by installing an external cold trap of much larger capacity, by adding plugging meters to detect impurity, and by additional pipework for balancing of cover gas volumes. Some of the now unused purity lines still remain blocked.

b. Design changes. Because of design changes, the original zero power physics calculations made on the ZEUS pile were invalidated. Therefore the reactor had to operate at 1 MW until it provided its own physics data to provide confidence for its high power operation.

c. Reduced stability. At high burn-up at 45 MW operation, less stability margin was apparent. This was apparently due to fuel expansion within cracks at the higher burn-ups, thus reducing the axial expansion feedbacks that were a negative contribution. This was no problem, but it pointed to the need for reactivity feedback measurements with adequate perturbations. The DFR oscillator had а І0 perturbation that was far too small for this investigation of feedback reactivity.

d. Radial blanket overheating. Due to maldistributed flow in the radial breeder at low flows, the blanket was severely damaged by stationary bubbles caught in buoyant down-flow conditions. Six hundred elements were removed, 200 only after cutting swollen sections. In fact, insufficient attention had been paid to the hydraulic design of the blanket sections, as these sections had low power ratings.

e. Core fuel pin damage. As entrained gas bubbles during start-up may adhere to unwetted fuel pins and then, due to this adherence enhanced by their buoyancy, stay long enough to blanket the pin as the power is raised, some fuel pins have been subjected to failure. This failure mechanism is troublesome in reactor operation but has been valuable in the sense that in no cases has there been any evidence that pin failure has spread or propa­gated (see Section 4.4.2) (26).

f. Primary circuit leak. A small leak in the primary circuit caused a one — year operating delay in order to find and repair the leak. The location of the leak was very difficult to determine, because the reactor is a 24 loop system any one of which could have caused leaking sodium to appear at the point where it was detected in the leak jacket.

First the cover gas was pressurized without affecting the leak. Then 10 gm of gold tracer were introduced into the primary and the tracer’s presence in the leak jacket subsequently confirmed the leak presence. Successive isolation of the leak location was made possible by adding helium to the cover gas and lowering the sodium level slowly to try to determine when the leakage stopped. This confirmed that it was in one of 24 bottom outlet pipes. Then, with a pressurized leak jacket and manual movement of the heat exchangers attached to each outlet pipe in turn, inward bubbling from the leak jacket was detected.

The failure was finally determined to be a fatigue failure of a bad and misaligned weld combined with thermal stressing due to cold NaK entering the hot outlet stream from a subsidiary purity line just upstream of the weld. Neither weld nor thermal stressing was sufficient to cause the failure alone. The purity line was removed and the weld remade to solve the problem.