Recip Compressor Selection

[zdas04]

Like most Oil & Gas Facilities Engineers I know, I've always left frame and cylinder selection to the packagers (within reason). I'm currently working of a feasibility study and didn't want to ask too much of packagers on a project that may never turn into a sale so I did the preliminary sizing myself.

The application was 9.8 MMCF/d of 98% CO2 2% CH4 going from 93 psia to 861 psia (elevation is 7,300 ft, suction temp 90F) which works out to about 1,380 hp.

It looked to me like the JGK/4 was a good fit for the task (Ariel Performance calculated 312 hp/throw in the first stage, 504 hp in the second, and 242 hp in the third running at 900 rpm).

I asked a packager for lead time on a JGK/4 matched with a CAT 3608 GLD and he came back saying that the application would be a lot better with a JGC/4 (without any clear definition of "better"). To match the machine to the flow, I either needed to reduce the first stage cylinders from 12.5 to 10.5 (which increased the hp requirement to 1586) or slow the driver to 725 rpm.

The JGK is 5.5 inch stroke rated at 635 hp/throw (max for a 4 throw is 2,540 hp). The JGC is 6.5 inch stroke rated at 1,035 hp/throw (max for 4 throw is 4,140 hp).

Do any of you have a feeling of why the packager would prefer the JGC/4 in this application (especially when they have packaged more JGK/4 frames than anyone else, yeah, it is that packager)?

David

 

[pmover]

David,

a longer stroke means lower piston speeds, which the JGC model does possess. other than the differences in rods loads, it does appear the JGC is more $.

like yourself, i'd be more inquisitive about the matter.

good luck!
-pmover

 

[ivymike]

David,

I'm told that the k-frame is more typically matched with a 1200rpm engine, and the c- and d-frames are more typically matched with 1000rpm engines, such as the 3608 and 3612.

 

[dcasto]

The JGK frame is a long time frame item these days. JGT is avaialable more readily.

The JDC WILL NOT WORK WELL. They have normally match the C frame to the 3608 due to an RPM rateing of engine.

pmover, longer stroke is faster piston speed, because the engine speed is fixed. The K frame is also required due to the gas molecules being bigger, CO2 is a big molecule and valve losses and volumetric efficencies are a problem.

Why not a 3606, it'll do 10.1 MMSCFD at 1000 RPM (site rated 1680) If you want a tad more, a lot of people use electric drive coolers which frees up another 90 HP so you can get 10.1 MMSCFD at 900 RPM

 

[zdas04]

With the altitude derate and the speed hit, I get the 3608 GLD (with engine-driven cooler) has about 1,667 hp at 795 rpm. The JGC/4 will move 9.8 MMCF/d at this speed (the compressor run called for 1,408 hp). I don't have a problem with the driver running at 84% of site hp. 120% with a 3606 GLD would give me pause (derated hp at 1,000 rpm is 1,731 hp, slow it down to 795 rpm and the site hp is 1369 hp, the compression load with auxiliaries is 1,614 hp). I know that I can probably come up with a mix of cylinders that will move 9.8 MMCF/d at 1,000 rpm, but the hp (and fuel) would increase markedly, and the "pseudo-q" value would be out of range (that is where I started).

I contacted Ariel about the application and they said that they like to limit the pseudo-q to under 13 for CO2 and they like to run the compressor as slow as possible. This was before I had the packager suggest the JGC/4 instead of the JGK/4.

David

 

[dcasto]

why slow it down, use a JGT frame

Cylinder Data: Throw 1 Throw 3 Throw 2 Throw 4
Cyl Model 14-1/8T 14-1/8T 12-1/2T 9-5/8T
Cyl Bore, in 14.125 14.125 12.500 9.125

9.85 MMSCFD @ 950 RPM at 1399 HP, it'll fit a 3606

 

[zdas04]

Makes sense, all the numbers work, slightly more hp than the JGK/4, but pretty good. Thanks.

But, the question was - why would someone pick a 1035 hp/throw machine when a cheeper 635/650 hp/throw machine would work much better?

David

David Simpson, PE
MuleShoe Engineering

http://www.muleshoe-eng.com

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

The harder I work, the luckier I seem

 

[ivymike]

my response above was based on a conversation with a former packager who was at my desk for another reason when I remembered your thread. To quote him more accurately, he said "that's a rookie 101 question... the k-frame is matched with a 1200rpm engine, and the c- and d-frames are matched with 1000rpm engines."

He didn't spend half a second thinking about it before he gave that answer... so perhaps your packager didn't either?

He also told me who THAT packager probably was... 'cause I didn't have a clue.

 

[zdas04]

Dang, over 20 years of doing this stuff and I ain't yet out of Rookie 101. It's been a while since someone tossed one of my questions into that particular bin. Thanks for the answer.

David

 

[dcasto]

the Rookie 101 answer is correct, EXCEPT, this isn't Rookie 101 gas. Big ol molecules like CO2 and Propane require very slow piston speeds, stroke times RPM should be less than 5000, 4000 is better.

The vendor doesn't know that, he just knows the rate frame RPM and the driver RPM, so the C frame goes with 3600 seies CAT's.

You need a 300 RPM, 14" stroke machine, alas, the closest thing is an AJAX.

 

[zdas04]

Dcasto,
You can probably guess who my client is (biggest producer into your company's processing plants), and I'm simply not going to encourage them by even mentioning an AJAX (they love the things).

btw, I changed my report to use your JGT/4 Cat 3606 LE and the projected cost of the project went down over $1MM.

David

David Simpson, PE
MuleShoe Engineering

http://www.muleshoe-eng.com

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

The harder I work, the luckier I seem

 

[dcasto]

I think I know the exact project. Are you using a screw package up front? Are you going to condense and pump after this unit?

 

[zdas04]

Yeah, I can get the stream to the compressors at about 4.3 psig, so to get to 90 psig is about 1,400 hp of flooded screws. I'm spec'ing a pair of Ariel 357 mm screws since that is the only 800 hp screw in general production (Howden has one, but getting information about Howden's is like pulling teeth, Kobelco may have one but they told me that I don't have a good enough reason to get a copy of their program, and I can't even find anyone to ask at MyCom).

I looked at compressing part way and then chilling to liquid and pumping to pressure. Funny thing is that with pure CO2, you get adequate sub-cooling at -20F and 250 psig. With 98% CO2/2% methane you have to go to -30F and 350 psig to get the same sub cooling. The shape of the saturation curve makes a J-T valve (constant enthalpy) and Turbo-Expander (constant entropy) way less useful than I had expected. Pumping from 350 psig to 900 psig only takes 60 hp or so, but chilling from 24F to -30F takes something like 2,800 hp. This option cost 73% more than just compressing it.

If you ever need to deal with Thermodynamic properties of mixtures I highly recommend the NIST REFPROP database. It cost $200 and probably saved me a month's work. I was able to generate p-h diagrams for the mixture and then overlay the pressure/enthalpy traverse on the diagram to get a really quick view to see when I was approaching a phase change. I got the suction/discharge conditions for each stage of compression from Ariel Performance and then went to NIST and got enthalpy. It was really cool to see a five stage compressor follow the constant entropy line in the compressor cylinders and the constant density line in the coolers--this stuff really works. I calculated compressor hp using the NIST enthalpy's and got values within 2-5% of Ariel Performance. Thermodynamics can be really fun with high quality data.

I need to stop bubbling and go back to studying my Rookie 101 course.

David

 

[ivymike]

yeah, get busy.

 

[insult2injury]

The other thing you can do with the NIST software is use the dlls for excel plug-ins, programming projects, etc. which are independent from their GUI. All you need to know is a little Visual Basic, Fortran, or C++.

I2I

 

[dcasto]

I have NIST REFPROP, but I use my process simulator for complex jobs.

 

[zdas04]

Dcasto
This is the first job I've done in 10 years that 2% give or take made any measurable difference in a design. I mostly work in blood-guts-and-feathers upstream operations where you are pretty much expecting to get it wrong since no one has ever figured out a reliable way to predict well response to changing line pressure at low pressures. You over-design everything to the point that a 30% error won't be a catastrophe. I haven't run a process simulator since I got trained in an early version of HYSYS (I think) in 1984 on Amoco's mainframe. REFPROP was a revelation for me.

I2I,
As much fun as this project has been, I'm unlikely to need REFPROP again for a few years. Doing the Visual Basic stuff a month ago would have saved me some time, but the cut-and-paste capability in the program was amazingly powerful (and quite intuitive).
David

 

[dcasto]

Here answers I get from my CAT/Ariel representatives


·Yes, the JGK frame is still made.

· You can’t run a 3608LE at 750 rpm. It stumbles along at that speed. The 3606LE is a better choice at 900 rpm.

· There are plenty of 3606LE engines available. 3608LEs are scarce.

· The shorter the stroke the better for CO2 with lower piston speeds.

· The JGT frame would be good and if you still have VE problems then add a stage to your pressure range.

 

[zdas04]

I GIVE. I turned in the report last week with a recommendation of a Cat 3606LE and an Ariel JGT/4.

Thanks for your help.

David

David Simpson, PE
MuleShoe Engineering

http://www.muleshoe-eng.com

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

The harder I work, the luckier I seem

 

[dcasto]

Beer at Steamworks, I'll buy.