Filter Sizing

[MacMcMacmac]

Good Day Folks,

Just wanting a little guidance on filter sizing for an upcoming installation for a combustion test cell.

Some background:

About a year ago I was asked to procure a vortex oil/water separator for a pair of reciprocating compressors (Worthington BDC). To size the separator, I simply took the stated ICFM of both compressors (~9500cfm combined) and ordered according to that figure. The discharge pressure of both recips is 300psi max. I obtained a 10,000cfm unit rated at 300psi. Was my methodology correct, or should I have taken something else into account? It seems to be working well, except for some float drains that are going to be replaced with pneumatically operated full-port ball valve style autodrains to correct a plugging issue with the current small-orifice (5/32") drains.

My current issue is that I have been asked to size an appropriate coalescing filter for a new installation in another test cell. One stream of air is coming from the two recips, while a second stream will be coming from a pair of Atlas Copco HM7 centrifugals. The recips flow through a dedicated filter rated for 10,000cfm in the basement, so that part is taken care of. The problem lies with the centrifugals. They are rated at 7pps each. I need to convert this to a CFM figure to properly size out a new filter. Here's where I always get confused.

I take the weight of air at .0807 lbs/cuft. Each compressor is good for 7pps, so 7 lbs/sec / .0807lbs/cuft = 86 cuft/sec = 5204cfm X 2 compressors = 10,408cfm. I realize the air will not be "standard" weight, as of course it will have moisture in it, but it works well enough for my needs.

Considering one of these centrifugals @ 2500hp is already equal to both recips at 1250hp apiece, it seems like I either grossly oversized the vortex separator, or I am underestimating the flow from 5000hp worth of centrifugals. Where am I making my mistake?

Thank you for any guidance you can offer.

 

[LittleInch]

The big thing missing is delivery pressure, both max rated and actual. Also temperature and intercooling is missing. What matters in a cyclone unint is actual velocity which is nearly proportional to pressure. Hence 10000 cfm at 100 psi is twice as fast as it is at 200 psi ( well within 10% or so). If you don't state operating pressure ranges you won't get the correct size equipment.

What joint pressure are you going to be running these things at as the atlas copco may be pressure limited on the low end pressures. Centrifugals tend to want to work in a defined band of pressure.

The PD compressors fair enough - apparently rated for 300 psi, but the centrifugal compressor?
If you could post the data sheet or technical details of the atlas copco as I did a quick google of it and no real details came up.

Compressing air to that sort of pressure needs a multistage compressor and possibly intercoolers and coolers on the outlet??

My guess is that the AC unit is a high volume, high pressure machine and motor size is for some higher pressure than your recips. They look like multistage also to get to 300 psig?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[MacMcMacmac]

Thank you for responding.

The Copcos are rated to 300psi as well. They are four stage units, with intercooling and aftercooling. All compressors will be running at 300 psig.

The Copcos have discharge blow off valves, so we can tailor the pressure to whatever we want it to be, not the most efficient setup, but we need very precise flow control for some of our tests. This test will need 20 lbs/sec, which dovetails nicely with what the recips and Copcos can put out at full power.

Assuming the centrifugals are operating at max pressure and flow (300psig, 15 lbs/sec),fully aftercooled to approximately 80F-90F, what CFM rating should my filter, or cyclonic separator be?

I am leaning towards another cyclonic separator, as the ones we installed seem to be working quite well. Combined with filtration, we should be able to get the air quality we need. Water and oil carryover have negative effects when combustor testing as it lowers temperatures and messes with emissions measurements.

If all else fails, we can enlist our 8000hp, 32lbs/sec, 300psig Cooper/Cameron to carry the entire load, but as it is an expensive machine to operate, this is a fallback option. It is fully intercooled, but not aftercooled. This is not a big issue, as the air will be sent through a process heater which will raise its temperature to as high as 1200F anyway.

 

[LittleInch]

On that basis another one of the same looks right to me - just need to specify the CFM, actual operating pressure and away you go... CFM looks to be your calculated 10 to 11,000 but clearly is a bit sensitive to inlet pressure variance due to altitude etc

Why your centrifugals appear to be over powered I'm not sure, hence a look at the data sheet would be interesting.

The centrifugals should accept a control valve varying flow up to some maximum pressure? Hence much easier to control flow than the recips.

1200C inlet air temp? What are designing - a blast furnace?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[MacMcMacmac]

Thank you very much for your input. Based on this, it would seem I got a little aggressive in sizing the separator for our two recips, but velocity through the system tends to be high anyway. The researchers always ask for more!

The temperature, flow and pressure needs to simulate the conditions inside the combustion chamber of a power turbine, so the air needs to be quite hot. It must be very precisely controlled, down to tens of grams of flow per second and within a few degrees. Flaming out the nozzles results in a lot of wasted time resetting conditions, since back pressure within the system varies quite a bit once we are fired up and running through the heaters. It's quite a performance keeping everything happy.

If the recips weren't so cheap to run I'd rather just stick with the centrifugals, but they are legacy machines (1950s) and have paid for themselves many years ago.

 

[LittleInch]

That sounds like a very hard job to control to that level of flow demand. controlling anything to 1% or less of max flow is a big ask. Instrument float and variance will be more than that.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.