A series of fire tests were conducted on nine, 1.8 m3 (500 USgal) propane tanks to study the significance of PRV behaviour on tank survivability to fire impingement. In these tests three tanks ruptured (i.e. finite failure) and six BLEVEd (total loss of containment). The difference between the BLEVE and non-BLEVE failures was due to a difference in the fire condition. It is believed that these tests show some insight into the BLEVE process. In all tests the fire consisted of an array of nominal 590 kW (2 MBTU/hr) liquid propane burners. A pool fire was not used because of the uncontrolled nature of open pool fires. It was believed that very repeatable fire conditions could be achieved by using a series of burners. In the tests where the outcome was a non-BLEVE there were two burners mounted 30 cm above the tank on the tank vapour space. These burners were used to weaken the steel and to initiate a failure. To heat the liquid, there were between four and twelve burners applied below the liquid level. When one burner was added on the vapour space, all of the remaining tanks BLEVEd. This was true over a range of fill levels (at failure) of between 10–50% by volume. It is believed this added burner was just enough to weaken the tank so that any initial rupture would grow towards a total loss of containment and BLEVE. This paper presents the details of this test series and shows how severely heated length and liquid energy affected the outcome.

This content is only available via PDF.
You do not currently have access to this content.