Thermal Hazards Laboratory

Recent Fire Testing and BLEVE Research at Queen's University

Department of Mechanical Engineering
Queen's University
Kingston, Ontario CANADA

key words: boiling liquid expanding vapour explosions ( BLEVE ), fireball, blast, jet fire, projectiles, pressure relief valve, pool fire, torch fire

The latest test program for Dr. Birk and his team to work on is the Pressure Relief Valve (PRV) Field Trial. This field trial is a continuation of the BLEVE Testing and PRV Testprograms conducted by Dr. Birk. These test programs showed that commerical PRVs can have highly variable behavior and computer modelling suggests that this variable behavior can have an impact on the outcome of a fire engulfment accident. In this case the PRV characteristics of interest are the blowdown and the cycling behavior. The computer modelling suggests that different blowdown and cycling behavior can affect the tank stress environment and the energy content of the lading.

The PRV Field Trial tests involve 500 gallon ASME code propane tanks heated by an array of liquid propane fueled burners. The burners are used (rather than pool fires) to ensure test-to-test repeatability of fire conditions. The object of the test is to show if different pressure relief valve operating characteristics have any signficant impact on the fire event outcome (i.e. BLEVE or not BLEVE).

This work is funded by the Dangerous Goods Directorate of Transport Canada through theTransporation Development Centre. The tests are carried out at the Munitions Engineering Test Centre of the Defense Research Establishment ValCartier in Quebec Canada.

The following Figure shows the tank ready for the test. You can see the burners (2 on vapour space, 12 in liquid space). This burner systems provides approximately the same total fire input as a fully engulfing fire with an effective fire temperature of 816 deg C.

The next figure shows the same tank but also inludes the bunker that protects the computer controlled PRV. In these tests the normal tank mounted pop action PRV is replaced by a computer controlled fast opening ball valve. This was done so that PRV action could be carefully controlled to give the desired action. In these tests we were interested in how the outcome would change if the PRV had a small blowdown or a large blowdown, for example.

The next figure shows the tank after the test was complete. As can be seen this tank ruptured but did not BLEVE. The failure opening is quite large and this resulted in a massive 2-phase propane jet and large transient jet fire which propelled the tank off of the tank stand.

Three propane tests were conducted in the summer of 2000 and six tests were conducted during the spring of 2001. In the 2000 tests all of the tanks ruptured but none of them BLEVEd. In the 2001 tests all of the tanks BLEVEd. The BLEVEs in 2001 were caused by the addition of a single vapour space burner (recall that the 2000 tests had two vapour space burners) which increased the size of the severely heated zone. This outcome clearly showed the signficance of the size of the severely heated zone.

The following sequence of figures shows a tank BLEVE. This BLEVE is an example of a 2-step BLEVE where the failure starts as a small pin hole and it grows to a length of maybe 5-10 cm. The crack growth then stops for a second or so. During this time a vertical jet release is seen. The tank internal pressure transient during this few seconds involves the following:

  • jet release causes pressure drop in vapour space
  • pressure drop causes a liquid boiling (flashing) response
  • vapour generated by liquid repressurizes tank
  • increase in pressure causes crack to start growing again
  • tank fails catastropically to release BLEVE

frame 1: tank initial failure (rupture crack stops with a length of 5-10 cm) and small opening produces vertical jet fire (duration 1-2 seconds)

frame 2: crack grows rapidly during tank pressure transient to cause total-loss-of-containment and BLEVE.

frame 3: fireball grows as propane cloud is consumed by fire

An important finding of this work was that tanks with large blowdown PRVs failed later than tanks with PRVs with small blowdowns. The delayed failure means the tanks have lower fill levels at failure and this means reduced energy and hazards at failure. For example, the fireballs were much smaller for the tanks with large blowdown PRVs.

The results so far show that variations in PRV operating characteristics do impact the outcome of a fire impingement event. Dr. Birk and his team will be returning to METC in the spring/summer of 2002 (June 10 to July 5) to carry out four more tests in an attempt to firm up conclusions from the previous nine tests.

For more information the reader should contact A. M. Birk at