Fire relief load reduction for flare system 3

[GreenHawk]

I have been working on a flare system to which the PSVs’ relief load due to fire scenario is exceeding the capacity of the flare system (due to an expansion in the plant). I know that using API's methodology for relief load calculation normally gives smaller relief load as compared to NFPA's methodology.

Additionally, I have also been told that if we divide the plant area into different fire zones with some restricting boundary wall around each fire zone, this can prevent flow of flammable fluids from one zone to another. By construction of this restriction boundary wall around each fire zone, we could be able to reduce the relief load to flare due to fire scenario. I had tried to look in API-521, API-537, etc. but was not able to find anything which endorses the use of this method and which gives more details on how to implement this boundary wall concept in the plant. Can anyone share their experience of implementing a similar idea in their plant? Additionally, can you also guide me if this aspect of flare load reduction has been discussed in any API or NFPA standard? Any other associated advice is also very appreciated.

Thanks.

 

[poli60]

Greenhawk,
some brainstorming points follow, with reference to API std 521, 5th ed (2007):
- Which kind of plant? (In a complex system, the fire contingency is very rarely the controlling load for the flare)
- Are you considering the fire on all the equipment to calculate the total load to flare?
- How much is the plot area? (See para 7.1.2, fire areas of 230-460 m2 max)
- Do fire fighting and drainage facilities exist? (You could further reduce the fire loads; see 5.15.2.2.1)
- Did you consider to reduce the fire load by applying proper insulation to the biggest pieces of equipment? (See 5.15.5....)
Hope this help.
Kind regards

PS: I don't know if 5th ed is the most updated one. Could someone confirm/comment? Thanks

 

[GreenHawk]

Poli60,

Thank you for responding on my questions. The plant which has this flare system is a complex plant with a couple of different sections. I have combined the fire relieving capacity of all pressure safety valves in the plant to calculate the total load to flare. This means that I'm considering a fire which exists all over the plant which comprises of several fire circles.

Does the para 7.1.2 in API-521 state that the effect of fire could not go beyond an area of 2,500 to 5,000 ft²? What I’m asking is whether I should consider that there will not be any domino effect beyond an area of 5000 ft². The total area of this plant is above 628,000 ft².

There are fire monitors, drainage and insulation on equipments, and I'd assume that their credit had already been taken in estimating the required relief rate. Therefore, I was wondering that how can I reduce the relief load to flare (e.g., by confining the fire to a limited portion of the plant, etc.)?

I would appreciate if you can share your thoughts on this case. Thanks.

P.S. I think that 5th edition of API standard 521 is the latest edition (Source: IHS). I have an addendum of this edition with an issue date of May 2008.

 

[poli60]

Greenhawk,
fire relieves are normally considered as single relief scenarios for the flare loads (ie only the highest relief load is considered for flare sizing/verification).
This is my experience in refineries, where good firefighting measures are always applied (eg local protection + ring protection + fire brigade).
Regards

 

[don1980]

Greenhawk,

There's no prescriptive guidance for dividing a plant into fire zones. API 521 gives some very high-level guidance but it leaves the details to the user.

Identify areas of the plant that have (or don't have) liquids that can cause a pool fire. Different companies have different criteria for what causes pool fire. Get a plot plan and highlight the trenches, curbs, dike walls. Those are hard boundaries which the burning liquid pool won't cross. Check the design of the ground surface. Is the cement slab crowned to cause drainage to flow to the trenches? If so, the high point of the slab is generally be regarded as a boundary.

API 521 states that fire zones are generally confied to 2500 ft2 of plot area. Look are the areas inside the boundaries and further divide them into fire zones based on your judgment and the 2500 ft2 guidance from API.

Once that's all done, check the flare system. One at a time, assume each fire zone is on fire, and calculate the flare system load based on the PSVs in that zone (those that have a fire scenario).

Then do a similar analysis on other "global" scenarios such as loss of power and loss of cooling. Loss of power and loss of cooling are commonly the same thing - they generally have the same effect. In the great majority of cases, loss of cooling creates a worst case load for the flare system. A fire zone might set the sizing for a particular subheader, but fire is seldom a worst case scenario for the fare system as a whole.