FIRE TESTING OF A FULL SCALE WOOD FILLED FLASK SHOCK ABSORBER

Year
2001
Author(s)
P. Purcell - British Nuclear Fuels plc
File Attachment
33535.PDF54.33 KB
Abstract
In many flask shock absorber designs, wood is used as the prime energy absorbing medium. Usually the wood sections are closely packed within a sealed steel casing with the grain orientation arranged to optimise impact performance. In large flasks over 1500kg of wood may be fitted. Under IAEA impacts from 9m and the subsequent 1m punch test the sealed casing may split or puncture causing the wood to become exposed during the subsequent thermal test. BNFL operate a number of flasks using wood filled shock absorbers and as part of a package safety development programme decided to conduct a thermal test on a typical example. An NTL 11 flask lid shock absorber was manufactured to original specifications then ‘modified’ to represent impact damage. This was bolted to a heavy steel slab representative of the flask lid. This was fitted with thermocouples at salient positions and then the whole unit was erected in a fire test facility, where it was subjected to an all engulfing hydrocarbon fire lasting 30 minutes. This test had two objectives: 1. To demonstrate the thermal protection given to vulnerable lid components during the thermal test 2. To assess the consequences of extended combustion of the wood material. The test demonstrated that the lid shock absorber offered nearly total thermal protection to the lid during the fire, the temperature of the dummy lid rising by a few degrees. This was due to the wood remaining largely intact during the fire, a situation not normally assumed in the current analysis. This has demonstrated the inherent pessimism in the existing safety cases for BNFL packages which assume all the wood burnt away during the 30 minute fire. The exposed wood in the shock absorber did continue to char for some time after the end of the thermal test but this was conclusively shown to have no adverse effect on the flask safety case. BNFL consider this test was effective in demonstrating that boundary conditions applied during the thermal analysis gave pessimistic results.