3 replies to this topic

[dozerboy]

We have several relief valves with high pressure natural gas relieving to atmosphere. The set pressure is 700 psig and the inlet temp will be ambient and will vary from 20F to 90F depending on the time of year. The discharge piping is vertical and is above any work platforms by at least 10ft. During a recent PHA the question of gas dispersion came up if one of these valves were to relieve. Not knowing exactly how to appropriately estimate the exit temperature, we did an adiabatic expansion calculation, which indicated the natural gas will actually be more dense than the surrounding air and now we are concerned that the gas will create a significant hazard as it will tend to move back towards the ground rather than readily disperse. However, there are natural gas relief valves all over the industry which relieve to atmosphere, so we are doubting that this is the appropriate method to evaluate the scenario. Anyone ever work on this issue or something similar?

[Lowflo]

Flammable vapor releases from PSVs rarely ever pose a hazard, as long as they are directed away from personnel and other equipment. You need to consider the possible radiation exposure if the relief stream ignites. However, the risk of VCE (vapor cloud explosion) is extremely small, even for gases that are heavier than air. That's because of the velocity of the stream exiting the tailpipe. Typically, the exit velocity from a pop-acting PSV is 300-500 ft/sec, and often higher. Rarely is it less than 300 ft/sec unless you greatly expand the tailpipe. At these velocities the jet stream naturally induces high amounts of air due to the low pressure zone created by the jet exiting the tailpipe. The dispersion rate is very high, yielding a surprisingly narrow flammable envelope. As long as the vapor remains a vapor (doesn't condense into droplets) and has sufficient velocity, then the flammable envelope will rarely ever drop below the point of discharge.

You can convince yourself of this by running some test cases using a gas dispersion tool such as PHAST.

This is also explained in API 521 (5th ed, 6.3.2.2). In API 521, the criteria for acceptance is based on a Reynolds number calculation rather than velocity. However, the effect is the same.

[kkala]

We have several relief valves with high pressure natural gas relieving to atmosphere. The set pressure is 700 psig and the inlet temp will be ambient and will vary from 20F to 90F depending on the time of year. The discharge piping is vertical and is above any work platforms by at least 10ft. During a recent PHA the question of gas dispersion came up if one of these valves were to relieve. Not knowing exactly how to appropriately estimate the exit temperature, we did an adiabatic expansion calculation, which indicated the natural gas will actually be more dense than the surrounding air and now we are concerned that the gas will create a significant hazard as it will tend to move back towards the ground rather than readily disperse. However, there are natural gas relief valves all over the industry which relieve to atmosphere, so we are doubting that this is the appropriate method to evaluate the scenario. Anyone ever work on this issue or something similar?

In a Safety study (2006) we had used Aloha dispersion model (free from EPA), which noted that light gases (e.g. NH3, CH4, etc) could behave as gases heavier than air (i.e. falling onto the ground) at low temperatures. Then we used Aloha to estimate dispersions of "heavy" hydrocarbons (MW>29) in the air, but dispersion of "light" gases of low temperature is expected to be similar. There may be specific dispersion software for them though.
Using a dispersion model, one can estimate the areas where concentration is between low and high flammability limits (having risk for explosion). For instance, one can estimate the ground area (if any), where ground level concentration of CH4 is lower than C1 (corresponding to high flammability limit) AND higher than C2 (corresponding to low flammability limit) to assess the risk. This has to be done for several atmospheric stability classes. Dispersion models do not give precise results, on the other hand they indicate extent of dilution of CH4 with atmospheric air (mentioned by Lowflo).

Edited by kkala, 10 May 2010 - 01:54 AM.

[kkala]

The risk of jet fire should be also assessed, as mentioned by Lowflo. At least for equipment ruptures, statistical possibility for jet fire or vapor cloud explosion does not seem to differ much, say 50% - 50% probability if the gas is ignited (others report different probability, e.g. 60% - 40%, etc). Time of gas ignition from the moment of release can be (say) 2 min or more, depending on the probability of meeting an ignition source in the area (road, refinery, etc). All these guidelines can be found in books or texts about Safety Studies; e.g. Institude TNO of Netherlands has published 5 books on the subject apart from software. Of course hypotheses of a Safety Study can be considered arbitrary, e.g. actually gas explosion can occur immediately or hours after a release; at any case they provide assessment of risk under conditions defined by the "guideline" (this is not yet code).
Considering gas dispersion models, I asked a colleague knowing about them; he does not know a model specifically for dispersion of gas getting heavy because of its low temperature. At any case googling "gas dispersion model" will result in much information. Be aware of promising models, they may require a lot of meteorological data to give a result.