Extremely stringent discharge volume recommendation in API 617 9th Ed 1

[JustSomeRoark]

Hello,

Annex E "Guidelines for Anti-surge Control Systems" in the latest revision of API 617 9th Ed includes some recommendations (i.e. the annex is marked as "informative) on the design of anti-surge control systems for axial and centrifugal compressors. Section E.2.5 "Trapped Volumes" states that:

"The pipe volume between compressor discharge nozzle, anti-surge valve, and check valve should be minimized. It should not exceed the volume flow which the compressor discharge handles in 0.5 seconds for axial and 1 second for centrifugal compressors."

In my experience with dynamic simulation of compression systems I have never, ever come across a system that complies with this requirement. I find it extremely impractical and unrealistic due to the following:

1. A significant percentage of the systems I've deal with include a water/air cooler at the discharge of the compresor, upstream of the antisurge line take-off. In these cases (a large majority) it will be absolutely impossible to achieve such a small volume.
2. Most of the times, operating companies' standards require that the antisurge valve is located at a high point, with free draining requirements to either side. Therefore the volume of the inlet line of the ASV (i.e. included in the discharge volume of the compressor Sect E.2.5 makes reference to) will be significant.
3. The take-off of the PSVs at the discharge of the compressor (in case they are required) must be located upstream of the discharge check valve to protect the compressor in case of such check valve failing in closed position, therefore the volume of the inlet line must be included also in the discharge volume. Since the PSVs are often located at an elevated position the added volume will also be significant.

I cannot fathom how such recommendation was included in API 617 9th Ed. and fear that some OEMs may use it to question the results of dynamic simulations carried out by third parties.

Your comments on this would be really appreciated. Do you find this recommendation helpful or simply absolutely unrealistic and unachievable?

Regards.

 

[georgeverghese]

At the outset, this requirement does seem startling. And I agree with 1,2,3. Agree this will rattle many a process engineer and the piping engineering team. And the 3rd party dyn sim engineers too.
So I ran a worst case scenario with the smallest throughput compressor I believe to be approx 100acfm at the inlet (from the GPSA) on my compressor performance spreadsheet. Suction press = 10barabs, discharge = 27barabs, inlet acfm = 116, feedrate = 1500kg/hr with a mol wt = 17, t = 25degC. Resulting discharge volume = 100am3/hr at a polytropic eff = 76%, discharge t = 90degC(approx). This requires a DN50 discharge line, and with a 1second discharge (actual) volumetric throughput trapped volume (which amounts to 0.028m3), this amounts to some 14m of DN50 line. Assuming the air cooler has some 100m of 1/2 inch tubing, the trapped vol for the cooler is 0.011m3. This reduces permissible exit line length to 9m of DN50 pipe.
Can this be done, with ASV recycle inlet piping included, and the discharge check valve just d/s of the cooler ? - I wouldnt say its impossible, but it would require some effort. It will be difficult to get any feedback from the authors of this annex, I think.
Presume you have done a quantitative check on an actual compression system.

 

[bicycledisciple]

Standard practice - take exception or add $$'s for the extra R&D time to achieve this.

RG

Ron
-------------------------------

http://www.georgeron.com

 

[pmover]

the greater the pipe volume between asv and compressor discharge flange, the longer it takes to reduce system resistance across the compressor. the system resistance is greatest at compressor inlet/outlet flanges. the volume quantity is 1 consideration, among others, in determining surge margin.

like you wrote, the API text is information and not mandatory.