Centrifugal Compressor Stonewall Operation 2

[LaithMarwin]

Hello I am currently at a plant startup and have a question. We have a centrifugal compressor here and the current conditions have it operating way off the right side (stonewall) of the compressor curves.

Essentially, the compressor is designed for the following conditions:
Suction Pressure: 320 psig
Discharge Pressure: 1000 psig

It is feeding in to a pipeline with a current pressure of about 500 psig which is way out of the operating envelop of the compressor.

When we run the compressor it operates at maximum RPM (turbine engine driver) but only delivers less than half the design flow rate. I guess it makes sense as the discharge resistance is so low that it can only operate in a choked flow or stonewall condition.

My question is what is the best way to get maximum output from the compressor. If my understanding of the compressor curves is correct the maximum possible flowrate for a stonewall condition should be where the stonewall line is (i.e. where the stonewall line intersects the appropriate compressor speed line) but that is not what we are seeing.

My thoughts are:

1. Operate the compressor at a lower speed and try to get on a curve in the compressor envelope (i.e. lower speed where the discharge pressure is a lot closer to our conditions).
2. Keep operating the compressor at full speed in stonewall and try to build up the pipeline pressure.

The item which has me the most confused is that the compressor is designed for suction pressure control but the compressor anti-surge valves cannot be out into AUTO until we are in the operating range of the compressor. That is at least what the compressor vendor is saying which seems very confusing. When we operate the compressor controlling the anti surge valve in manual we seem to enter surge conditions as the discharge system resistance changes. Is the compressor vendor correct? I always believed that one could put the compressor anti surge valves in AUTO at any given moment in time and the compressor would take care of it's self as it knows how much flow is required based on the operating conditions.

 

[zdas04]

People keep wanting to pretend that dynamic machines will operate like a PD machine. They won't. Your multi-million dollar compressor was designed to do 3 compression ratios with a suction of something around 335 psia and discharge around 1015 psia. That is where your curves are valid. That is where performance is predictable and makes sense. To do 3 ratios probably takes a 2 stage machine (something like 1.75 ratios per stage) which wants first stage discharge to be around 580 psia. You are actually doing 1.5 ratios across the whole skid. If anything is working I'd be impressed. If you think the 500 psig discharge is more or less permanent, then you can look at single-staging the compressor. If it is just a phase, then I'd look at a discharge backpressure controller and jack it up to design conditions.

David

 

[LaithMarwin]

Hi David thanks for the quick response and information. I agree that centrifugal ompressors are not like PD compressor where all you need to worry about is the differential pressure.

The line pack pressure is slowly being built up and I wanted to just a discharge restriction in order to get the compressor into the proper operating range however, this does not seem to be possible as we only have motor operated valves and everyone seems hesitant (this some merit) of modifying a motor operated ball valve to act as a throttle to build up line pack pressure. My thought was simply to try and move the compressor into the proper operating range while we use it to try and build up line pack pressure. Attached is some performance curves (Suction Pressure 317 psia @ 80 F)

Right now in a choked flow condition (machine at maximum speed) we are only able to flow about 40 MMSCFD (Suction Conditions: 350 psig @ 80 F Discharge Conditions: 600 psig @ 214 F) where as if you look at the curves we could move a lot more mass if we ran the machine at a lower speed and suction pressure and temperature (to be inline with the 80% speed curve).

As per the attached curve I believe that if we ran the machine at 80% speed and lowered the suction pressure to 302 psig (317 psia) and the speed we would be in the operating envelope for the machine and move a lot more mass flow (60 to 70 MMSCFD I calculated the mass flow to be around 161,488 lb/hr). I know that the curves are dependent on a specific set of conditions and need to be re-run based on new conditions.

I am not a compressor expert and before I start telling the "experts" how to run their machine I wanted to get some feedback. They seem to feel that running the machine at maximum speed in a choked flow condition is better and will build the line pack much faster and that is the maximum flow that the machine can do.

Any thoughts you have will be much appreciated.

One last question is that the compressor does not seem to go into AUTO on the surge controller and I find that to be very puzzling. I figured that one could use suction pressure control (even at these low flows) and have the compressor hum along with the anti surge in AUTO but the surge control valves is completely a manual operation.

Thanks for any feedback you have.

 

[pmover]

your situation is not unique . . . rotating machinery operates at the intersection of the system resistance line and, in this case, the compressor operating speed line. since there is little restriction (i.e. low compression ratio), operating a centrifugal compressor while in stonewall is thermally inefficient (just burning fuel). as far as damage, well, that question is best answered by the compressor mfg. extended operating periods in stonewall may/may not impact the compressor. since the compressor is discharging into a pipeline, it will take time to develop sufficient resistance (i.e. differential pressure across the compressor), which is dependent upon pipeline diameter/length, downstream pipeline outflows, and compressor capacity. so, at start-up of pipeline compressors, initial operation at stonewall (if no other units are in parallel operation) will occur. hopefully, this was a matter reviewed/discussed during design stages with the compressor mfg to ensure no operational/mechanical problems will occur.

using a ball valve to add system resistance is not a recommended practice as the ball valve seals can be damaged (high gas velocities) and it is extremely difficult - impossible to control. ball valves are not designed for control purposes (mainly open/close operation). do not attempt.

regarding the surge control, vague information is provided as to why the controller does not operate in AUTO. investigate the design, instruments, and cable connections to ensure proper installation. not sure what mfg controller is being used.

compressor outlet conditions can be predicted if inlet conditions are known (P, T, Z, Q, etc.) and compressor rpm. lowering inlet pressures will decrease mass flow. of importance is volumetric flow.

hey, the compressor curves are not attached . . .

hopes this helps and good luck!
-pmover

 

[LaithMarwin]

Hi pmover, thanks for the insight and help. I uploaded the curves to the Eng Tips server but for some reason it did not take. I will try again here.

I have a couple more questions for you.

1. In terms of the surge valve I am totally confused. Every facility I have been at the compressor is designed so that one can manually close the ASV until it calculates that a possible surge could occur and then it opens automatically. Here there is a surge map (Pd/Ps versus dP1/Ps)which is defined as:
dP1: orifice meter pressure drop
Pd: Discharge Pressure
Ps: Suction Pressure

It is a simple curve and according to the vendor the ASV's cannot be put into AUTO until such time as the "cross hairs" on the screen go to the control line. Does that simple mean that since the compressor is so far out of it's operating envelope it does not know if it is surging? This seems puzzling as in a typical curve there is a minimum mass flow (based on pressure and temperature conditions) that the compressor needs to be out of and it would know when it was in surge well before the "surge noises" start. Is this your experience? This morning we were almost able to get the compressor to the "control line" but as we were almost fully closed on the ASV valve the compressor started to surge and the cross hairs moved all over the map.

(Note: We have checked all instrumentation for the ASV system and everything seems to be in check.)

2. As this compressor is normally designed to send gas to a pipeline what will happen if the pipeline pressure starts to change (i.e. resistance changes)?

3. In order to get into the best operating envelope I figure that it makes the most sense to run the compressor at a lower speed and get into an operating envelope correct?

Here is the link to the compressor curves:
Your file's link is:

http://files.engineering.com/getfil...7b0aa0097a0&file=Export_Compressor_Curves.pdf

 

[LaithMarwin]

Hi pmover, thanks for the insight and help. I uploaded the curves to the Eng Tips server but for some reason it did not take. I will try again here.

I have a couple more questions for you.

1. In terms of the surge valve I am totally confused. Every facility I have been at the compressor is designed so that one can manually close the ASV until it calculates that a possible surge could occur and then it opens automatically. Here there is a surge map (Pd/Ps versus dP1/Ps)which is defined as:
dP1: orifice meter pressure drop
Pd: Discharge Pressure
Ps: Suction Pressure

It is a simple curve and according to the vendor the ASV's cannot be put into AUTO until such time as the "cross hairs" on the screen go to the control line. Does that simple mean that since the compressor is so far out of it's operating envelope it does not know if it is surging? This seems puzzling as in a typical curve there is a minimum mass flow (based on pressure and temperature conditions) that the compressor needs to be out of and it would know when it was in surge well before the "surge noises" start. Is this your experience? This morning we were almost able to get the compressor to the "control line" but as we were almost fully closed on the ASV valve the compressor started to surge and the cross hairs moved all over the map.

(Note: We have checked all instrumentation for the ASV system and everything seems to be in check.)

2. As this compressor is normally designed to send gas to a pipeline what will happen if the pipeline pressure starts to change (i.e. resistance changes)?

3. In order to get into the best operating envelope I figure that it makes the most sense to run the compressor at a lower speed and get into an operating envelope correct?

Here is the link to the compressor curves:
Your file's link is:

http://files.engineering.com/getfil...7b0aa0097a0&file=Export_Compressor_Curves.pdf

 

[pmover]

surge control . . .

i understand and agree in that a surge control valve can be manually closed as long as the controller determines a surge condition is not present. then again, most surge control systems are in AUTO during start-up (i.e. SCV remains open until some condition exist, like compressor rpm > x, flow signal > y, or otherwise).

why is their an attempt to "get the compressor to the "control line" "? as long as the process variable (PV) is > the set point (SP), the SCV should close.

"but as we were almost fully closed on the ASV valve the compressor started to surge and the cross hairs moved all over the map" tells me that the controller SP <= PV, which indicates the controller believes the compressor is at the surge control line. So, why? how about proper zero and span for all transmitters? properly calibrated? is the controller properly configured for all input/output signals?

2) again, the compressor operates at the intersection of the compressor speed line and the system resistance line. as long as the operating point is within the "safe"/normal operating region of the compressor, why worry? typically, pipeline compressors utilize discharge pressure, suction pressure, or flow control to handle varying pipeline conditions.

3) the polytropic head and actual cfh curves are all that is needed. determine flow from the raw input data at DCS/PLC and determine poly head and plot on map. determine speed from the map at calculated flow & head and compare with actual compressor speed. if different, then there is a problem for you to analyze and solve. be sure of all input operating parameters/conditions.
as far as the lower speed, this lessens the impact of an actual surge condition if one were to occur.

is the actual gas MW the same as the design conditions? remember, that compressor wheel will deliver the poly head at the operating speed and inlet flow conditions. the outlet conditions will change if the actual gas is different from the design.

double check all data and good luck! i won't be able to follow-up til next week.
-pmover

 

[LaithMarwin]

Thanks again pmover for all your valuable insight and you are confirming my own understanding. Many things are confusing here for me and you highlighted a couple of them. All they show is mbar of flow. I said that buried in their software must be some type of actual flow (ACFM) and mass flow calculation but I am encountering resistance from the vendor to extract the data. In terms of flow I am going to crack out the old text book and do some mbar to mass flow and ACFM calculations (orifice plate calculations). We have a flow meter in the facility but really it only is valid for the outlet flow and not the flow around the compressor.

The compressor has suction pressure control and I figure that the operator just needs to enter the desired suction pressure (within the envelope) and then the compressor would use speed to maintain the suction pressure. The ASV would act automatically to ensure that there was sufficient flow based on the speed, pressure and temperature and operate on it's own. But the vendor insists that until it can reach the control line the ASV can only be operated in manual.

We did some tests on the second compressor and tried to operate it along a speed curve with the proper conditions (suction pressure, suction temp, etc). The mole weight may be slightly higher than design (20 versus 19.89) but I am not sure if that would have a major impact. What we did notice was that at 30 and 45 minute intervals there was major surging occuring we were operating at a low speed (10000 to 9800 RPM on a machine rated for 11400RPM). Here is the event that would occur:

1. The compressor would operate for 30 to 45 minutes.
2. During operation the mmbar would slowly decrease.
3. After the 30 to 45 minute interval there would be a sudden surge where the mbar reading for the flow meter would go to zero then spike up corresponding to a 10MMSCFD change in the outlet flow from the plant.

We are checking for a PSV that may be passing or some change in the downstream resistance and can only see a 2 to 3 psig change in the pipeline pressure. All the other process variables in the plant seem to be rock solid.

Thanks in advance for any thoughts you have to share.

 

[pmover]

a hasty response . . .

investigate online for the AGA 3 flow equation. download it and use it for your conditions. try:

http://www.brothersoft.com/aga-3-139459.html

and check this site for further reading.

http://www.cheresources.com/high_dp_orifice_flow.shtml

orifice plate flow measurement uses dP across the plate to determine flow, so the mbar is likely the dP across the orifice plate. mbar of flow is senseless . . . no flow units of measurement.

again, do the flow calcs yourself. i never trust "a vendor" until i verified it.

manually operating a SCV at start-up or for any mode of operation by unskilled/knowledgeable people is foolish and demonstrates a lack of understanding of the possible consequences. before any manual control of a SCV, i would be 100% certain of the I/O and controller's actions. sudden rotor thrust reversals, especially at high compressor differentials can be catastrophic to the compressor.

good luck!
-pmover (now i am gone . . .)