Exhauster Performance Estimation for Gas Turbine Altitude Testing

[MacMcMacmac]

Good Day. Here is a rather long-winded explanation of my issue, and a request for assistance if you have a spare few minutes.

We have had a failure on our 2MW exhauster set. Last year, we spent quite a lot of time and money getting one of our 1350hp motors on this unit to within acceptable vibration levels. Unfortunately, during our annual maintenance shutdown, it has been discovered that the twin to it has a 50% resistance imbalance on one of the wound rotor windings. I can hear the crane loading the motor as I type this. What is even more unfortunate, its that we are smack dab in the middle of a very important test campaign, and we were relying on this exhauster set to allow us to reach the altitude points requested by the client.

There is a second machine in our facility which will allow us to reach a majority of the test points. It is a 5MW compressor/exhauster. Right now, it is set up the same as the 2MW machine: a low pressure stage exhausts into a pipe, which passes the air through an intercooler into a "high" pressure stage which discharges to atmosphere.

As it is configured right now, the 5MW machine is rated for 24,000cfm and 100 psia. The first stage on the 2MW machine is rated at 1250m3/min, or roughly 44000cfm. Obviously, the 5MW will flow nowhere near the 2MW first stage, so it is natural it cannot achieve the same altitude settings once flow through the test rig exceeds its capacity.

There is an option open however. The 5MW machine as delivered and installed, had the capability of being configured to flow in a "parallel mode" of operation whereby both stages of the compressor/exhauster could be exposed to the same inlet pipe, and both discharge to the same stack. A second inlet valve to the second stage and a second discharge valve from the first stage were opened, and the path through the intercooler was isolated. In this mode, we essentially give up pressure capability for a much greater flow capability.

What I am wondering is, would two LP stages acting in parallel have the capability of producing the same vacuum levels as the 2MW in a LP/HP staged configuration. I know from my years of working with compressors that higher pressures necessitate staging, and since the 5MW would essentially be turned into a single stage, high flow, low pressure compressor, how much vacuum potential would I be giving up?

The 2MW LP stage is rated at 1250m3/min (44000cfm), discharging into a HP stage rated at 403 m3/min (14000cfm). 1st stage discharge rating is 4.4psia, 2nd stage is 15psia, if I am reading the 1930's era German language tags correctly. We have virtually no literature on this machine, but curiously, I have the performance curves from the 1930s.

The 5MW LP stage is rated at 24,000cfm, the HP stage is rated at 11,300cfm when pulling from atmosphere in parallel mode. The LP stage is rated at 39psia discharge from 14.5psia inlet, the HP stage rated at 45psia discharge from 14.5psia inlet. Together, they still do not quite equal the flow rate of the 2MW first stage, but it is still considerably better than the 5MW configured in a 2 stage arrangement.

Since the pressure ratios of the 5MW compressor even in parallel mode are much higher than the 2MW stages, I am thinking we are not losing much vacuum potential, if any, since the 2MW stages are rated much lower in the pressure domain. This makes sense, since the size of the 2MW air ends are quite large compared to their motors, and easily over-amp the motors if inlet valve settings are too aggressive. The 5MW on the other hand, is a largely set-and-forget machine which rarely exceeds a 1MW draw no matter what condition we are at.

Unfortunately, the parallel mode was nullified a few years ago when the second discharge valve from the 1st stage was removed due to corrosion and the fact that it had never ever been used in that mode.

The restoration of this option would require some minimal piping mods and the installation of a new valve. I am hoping that I can make the case, and if it turns out to meet our requirements going forward, it would give us 100% redundancy in case of another 2MW unit failure, or we might even consider retiring the 2MW unit completely, which I am fully in favour of since the failure rate of this machine seems unacceptable. I think we are on the back end of the bathtub curve with this machine. I have compiled a list of multi-stage exhauster manufactures and brochures on the network as a major hint.

The only thing that concerns me, is that the 5MW will surge at around 39,000ft of simulated altitude, whereas the 2MW will proceed up past 60,000ft before entering the surge zone. The current test seems to require test points just outside the current flow capacity of the 5MW as configured.

Thanks for reading this far. If you need clarification on anything let me know. As always, I appreciate any guidance.

 

[georgeverghese]

Reading from your assessment of the 5MW machine when operating in parallel mode,

"Together, they still do not quite equal the flow rate of the 2MW first stage, but it is still considerably better than the 5MW configured in a 2 stage arrangement."..

it sounds like you already know what flows this machine will achieve at the desired simulated altitude ?

Presume these machines are centrifugal compressors, and that you have drawn these conclusions from calculations done with data on polytropic head vs flow curves for the LP and HP stages of 5MW machine.

 

[MacMcMacmac]

Hello George, Thanks for replying to my tome!

Here is all of the information I have on the 5MW compressor performance:

I find it odd that the flow through the second stage only goes up by about 200cfm when coupled to the first, but it is what it is.

I am banking on the much better pressure ratio of these stage compared to the 2MW exhauster will allow a decent altitude performance despite the fact they are both pulling from the same inlet and not acting in series.

I guess what I really am trying to figure out is why the 2MW can pull higher altitude with less power. The only thing I can figure is that the first stage 2MW air end is so much bigger it can flow more at lower pressures without overloading the motors, whereas the 5MW is biased toward pressure production at the expense of flow. By opening up both stages to the same inlet source, I am harnessing the flow capability of the second stage, while decreasing the pressure potential of the entire set. This is irrelevant, since pressure production is moot when in use as an exhauster to atmosphere.

The 39psi and 45psi capability of the standalone 6MW stages are much higher than the pressure potential of the 2MW, Combined, they should flow about 35,000cfm together. My expereince whith exhausters is much less than my experience with compressors, but I figure they are just low pressure compressors in the final analysis.

Thanks for any input. Let me know if my reasoning is faulty anywhere.

 

[georgeverghese]

When exhaust pressure is 14-15psia, in theory, LP and HP casings operating in parallel ought to give you much more than 35000cfm, but to pin this flow down (at least approximately), we need the performance curves for these 2 casings ( flow vs polytropic head and flow vs polytropic eff). It is also not clear which of these machines (the 2MW or the 5MW) are centrifugal or axial flow compressors. From the GPSA, it looks like these are centrifugal machines, since inlet flow is < 80 000 acfm.

There is also the risk of poor extrapolation of the current polytropic eff curves (assuming you can find them) to operating envelopes which are far removed from the design case operating envelope, which is the case here. But I guess a stab in the dark is better than none. Ideally, the compressor vendor ought to give you new performance curves for these intended new operating conditions.

By the way, that intercooler could do a lot better than 194degF if it was a bigger unit with more heat exchange surface area, in the current series flow arrangement. A bigger unit could have brought exit gas temp down to 100degF, which would then free up compression flow and reduce power on the downstream HP casing.

The process controls for this machine may also need to be carefully studied and modified for these new operating conditions.

 

[MacMcMacmac]

Thank you George.

Some further snooping turned up the following curves:

The line of interest would be the short fellow at lower right. Flow looks promising. Outlet pressure is above ambient, so it looks like it might do the job. No indication of how performance might drop off under altitude conditions. It's hard, maybe impossible to figure this out without actual running. I found some literature on the network that said 14.5kpa in suction mode when in a series configuration, which corresponds to about 49,000ft of altitude. Delivery at that inlet pressure is 1.93kg/sec, or about 4lbs/sec. I'm not sure if temperature is being taken into account.

As you suspected, these are single shaft, multi stage centrifugal compressors.

The oem, Brown Boveri, is long out of the compressor business I'm afraid.

Yes, the intercooler is a minimum effort affair.

Thank you.

 

[pierreick]

Hi,
You should contact ABB.

https://global.abb/group/en/about/history/heritage-brands/brown-boveri-cie-bbc

BTW you got a decent document!
Pierre

 

[georgeverghese]

Presume the note below that composite curve for parallel operation which reads VZ503p + VZ503p is a typo error - should be VZ503p + VZ504p.

If one were to make some assumptions on polytropic eff, there most likely is info here on the individual curves for VZ503p and VZ504p to enable calcs to extrapolate performance (approximately) at any desired suction pressure. These are laborious calcs which are made less painful with the aid of a process simulator (HYSIS or Simsci/Pro-II or similar) or if you have means to hire an engineer familiar with such calcs, who may have a polytropic analysis spreadsheet to help. If the results look promising, you could then use these derived observations as justification to request approval of capex allocation for the intended mods. You can also use these calcs to verify accuracy by predicting performance in parallel mode operation to see if calcs match the composite Brown Boveri curve. None of this work comes free of charge, as there is a fair amount of specialist work involved.

 

[MacMcMacmac]

Yes, it's time to make a case and do some work.