9 replies to this topic

[B-2 Spirit]

Hello experts,

 

I am currently doing a debottlenecking study and focusing on Flare system adequacy.

The flare type is pipe flare.

 

The design specs of the flare (existing design) is

Flow Rate:324,073 kg/hr (Power Failure)

Temp:25 degC

Mwt:30.7

Pressure Drop across tip: 1. 5 pisg

Mach No. 0.41

 

the unit has been debottlenecked twice already and the total power failure which I gather from the existing system (already debotlenecked) is

in excess of 500,000 t/h

temp:98 deg C

Mwt: 34

Pres. Drop across tip: 3.44 psi

Mach No. 0.57

 

I have limited expereince in flare systems and as per my understanding the flare is not adequate, but the unit is operating. Looks like the issue was highlighted but no action is being taken on this.  

 

Can anyone share their experience on flares and whether the above scenario is acceptable.... ?? what are the various factors efecting the decision. Is it because of a very low probability of total power failure sceanrio that they might have decided to operate the unit. What about incomplete combustion in such a scenario? might lead to vapor cloud formation etc....

 

 

 

-best regards,

B2Spirit.

 

[shan]

1.  Mach No. should be less than 0.5 per 6.4.2.1.2 API 521 (2007).

2.  Your flare scrubber may be unable to eliminate all the 600 um diameter or larger liquid droplets when the  flow rate is higher and the liquid/vapor density differential is less.

[Bobby Strain]

You know that you can add devices to mitigate the load during power failure. I wouldn't be conserned about the flare tip. Worry about the back-pressure on relief valves during this high relief. And, like Shan says, the KO drum. Many drums are undersized even for the design flow, and the API sizing procedure is grossly in error. Those who know don't use it.

 

Bobby

[oscarsender]

Hello,

 

It may help to check if they have done some flare reduction studies like dynamic simulation or HIPS. The flare load that exceeds 500,000 kg/hr may be calculated using conventional method in which it just add the relief load of all afected equipment, but in reality they may not all relieve at the same time depending on some credible instances. Dynamic simulation helps estimate the reduced load based on these. Additionally they may have an existing safeguards in the plant (HIPS) which provides some instrumentation to reduce or stop some very large reliefs (i.e, they have an existing trip that stop the flow of steam to the reboiler of the column that has the largest relief during power failure).

[Bobby Strain]

And it is going to be a lot hotter in the vicinity with this higher heat release. You should check radiation levels at the higher load.

 

Bobby

[gegio1960]

Which origins have the single loads contributing to the GPF load?

Which kind of actions have been taken in the 2 previous stages of debottlenecking?

I think only a serious dynamic study could improve the understanding of this situation (if other steps of improvement, like HIPS implementation, have already been made in the previous flare load reduction steps).

[B-2 Spirit]

Dear All,

thanks for your valuable inputs.

 

-B2Spirit. 

[Linda1978]

Hi,

 

I think you have to:

 

1- re-check size of flare headers. make sure you are not exceeding the mach no. of 0.5 in main header and mach no. of 0.7 in branches.

2- re check the flare network using a professional flare network software (read more about flarenet here)

3- check the heat intensity and isopleths. Make sure the stack height and radiation intensity are adequate. if you have multiple flare stacks use a professional software like flaresim. if you have one flare stack use available simple applications (e.g. see here) or do hand calculations (there is a simple method in GPSA).

4- check the knock out drum and seal drum and re-check the size (bobby strain has good software for sizing)

 

good luck

Edited by Linda1978, 04 June 2015 - 12:55 AM.

[Bobby Strain]

Linda,

     The software is good for design, not just preliminary sizing. Using API procedure can get the designer in trouble. But it's not available always unless you are a subscriber. At the moment, the sample has the vapor flow locked. I unlock it from time to time, but I unlock it at random.

 

Bobby

[Linda1978]

Sorry Bobby,

 

I know  your software is good. I have corrected my sentence. Please accept my apologies.