24 replies to this topic

[K J]

Dear all,

       Can any one suggest how to reduce power consumption in motor driven fixed speed multistage centrifugal compressor by operational changes. Actually we have three stage N2 compressor. For fixed discharge pressure of 23.5bar if by available margine(operation 2.7bar incresable up to 3.2bar) we increase suction pressure its suction guide vane close so is there any way for power saving by operational changes.

[PingPong]

Whether we like it or not, to compress a certain amount of N₂ from 3.2 to 23.5 bar requires a certain amount of power, depending on the efficiency of the compressor and the inlet temperatures of each compressor stage.

 

Apart from increasing suction pressure and/or decreasing discharge pressure, I don't see how you can lower power consumption by operational changes only.

 

Improving the efficiency by modifying the compressor (check with vendor) will be costly, if possible at all.

 

Lowering the inlet temperatures of all stages will require larger coolers and/or colder cooling medium.

[GSaikrishn]

see if the compressor spillback valve is open. closure of the spillback valve is direct reduction of the compressor power.

[thorium90]

I would definitely advise against randomly closing the spillback valve as suggested above. The valve is for controlling the pressure not for controlling compressor efficiency... gas compressors have certain inherent differences from liquid pumps.

Edited by thorium90, 28 October 2013 - 09:11 AM.

[K J]

This is our DCS graphic page in which PIC1623 has good margine but if compresor suction pressure is increased its guide vane will close. looking at this is there any way.

Attached Files

•  N2 compressor.xlsx   366.91KB   66 downloads

[PingPong]

I don't see what PIC1623 has to do with the inlet guide vanes of the compressor. Those seem to be regulated by PIC1650RS but it is not clear where the pressure sensor of PIC1650RS is located.

 

According to PI1001 the N2 from the ASU has a pressure of 4.7 BarG but the pressure at the compressor suction is only 2.7 BarG, meaning that PCV1623 has a pressure drop of 2 bar, which seems extremely high in relation to the absolute pressure.

 

You best contact the company that designed this unit and ask them what the hell they were thinking when they designed this system.

[thorium90]

This is what I gather from your screenshot.

 

It would appear you have an ammonia plant with a reformer for making hydrogen and an ASU for generating nitrogen for this purpose.

HIC1004 is some input parameter to control FIC1002 N2 Compressor Discharge flow to a certain amount wrt to the hydrogen flow? and in this case it is at 75.7% open and 23.08barg.

 

PIC1650RS is actually the remote setpoint calculated by the Performance Controller. The pressure PV is taken from PI1631.

The Performance Controller controls the compressor by controlling the IGV which is currently at 22.7% (23% is the commanded value from the controller), and by controlling the recycle valve FV1602 which is currently at -0.2% (0.0% is the commanded value from the controller).

 

Evidently your setup is one where capacity control is by a combination of IGV and recycle control whereby both discharge flow and pressure are controlled by the controller

 

When suction pressure is increased, the discharge pressure and flow will also increase. Since the recycle is already fully closed, the IGV is then closed to reduce the flow and pressure back to the setpoint.

 

I too feel the dp across PIC1623 might be a little too high. PIS1623A/B/C are all showing low pressure alarm. Perhaps the discussion would benefit from knowing the design pressures for the compressor inlet and outlet as well as the flow and more info on this valve, otherwise it could be hard to make an informed judgement.

 

Is it possible to modify the setpoint for PIC1650RS? There is still some leeway on FIC1002 as it is only 75.7% open, assuming this screenshot is of the plant running at full load. In fact, the discharge pressure PI1002 appears to already be giving a high pressure alarm (HI)!!! If you reduce the discharge pressure setpoint, FIC1002 will open more to achieve the required flow and overall you will definitely get some improvement.

Edited by thorium90, 30 October 2013 - 10:10 AM.

[PingPong]

Please note that the "recycle" or "kickback" via the normally closed FV1602 is merely the antisurge control valve. It is not for nitrogen flow control to the ammonia plant, but to prevent a too low flow through the compressor if the forward flow through FCV1002  is below the surge limit of the compressor.

 

The nitrogen from the ASU is regulated by PCV1623 (compressor suction), by inlet guide vanes, and by FCV1002 (compressor discharge). That is three devices to do what could have been done by only one. A nice example of doing something simple as complicated as possible.

 

This compressor could easily have been designed with only 2 stages (instead of 3) and without 2 bar lost over PCV1623. That would not only have resulted in a lower investment cost, but also in a 30 % lower power consumption, and a 30 % lower cooling water consumption. Also the IGV's are in my opinion not really required (in the past 30 years that I do process design I have never seen a multistage centrifugal compressor in process service equipped with IGV's). The FCV1002 in the compressor discharge would have been sufficient to control the desired nitrogen flow rate.

 

The N₂ from the ASU is delivered at 4.7 BarG & 21 ^oC. At the inlet of the second stage the condition is 5.4 BarG & 38 ^oC. That is only a 12 % increase in absolute pressure and only a 6 % decrease in actual volume flow, but at the expense of some 800 kW power. Ridiculous.

 

I don't know what this compressor looks like. Are all stages in one casing or not? If the first compressor stage is a separate casing then it would be interesting to investigate whether it can simply be removed, and also PCV1623 removed, and feed the N₂ from the ASU directly to the second stage at 4.7 BarG & 21 ^oC.

Edited by PingPong, 30 October 2013 - 02:03 PM.

[thorium90]

Actually, from the screenshot, the Performance controller uses the discharge pressure as one of its inputs and there are only two outputs; the IGV and the recycle valve FV1602. The discharge flow is controlled on a separate loop; FIC1002, which takes its setpoint from the reformer side. Im not sure if PIC1623 is also one of the outputs of the Performance controller as it seems odd that it would give a setpoint to the valve that would give low pressure alarms... Either there is something wrong with the alarm setpoint or someone has itchy fingers and input such a low setpoint for suction pressure.
 
I believe the reason for such a complicated scheme is because the downstream reactor needs a consistently fixed pressure of N2 at a wide range of flowrates. If the only adjustable parameters were the suction valve and recycle valve, the compressor would not be able to achieve a fixed discharge pressure over a large flow range.

Edited by thorium90, 30 October 2013 - 10:03 PM.

[PingPong]

PIC1623 seems indeed not set by the Performance controller, but simply by hand by the operator.

 

My point was: now there are 3 devices (restrictions) in series to obtain the required nitrogen flowrate to the ammonia plant, whereas only 1 would be have been required in a better design: only FCV1002 in the compressor discharge. That valve automatically corrects for the difference in pressure between the compressor discharge (upstream FCV1002) and the ammonia plant (downstream FCV1002). There is no real need for a fixed compressor discharge pressure upstream FCV1002, other than that it should obviously be somewhat higher than that required by the ammonia plant.

 

In the present design the first compressor stage does hardly more than compensate for the pressure loss caused by PCV1623. Without that valve the system could easily have been designed with a two stage compressor and 30 % less power consumption.

Edited by PingPong, 31 October 2013 - 06:54 AM.

[thorium90]

Yes I agree with you that there appears to be some unknown reason to us for maintaining a consistent pressure of ~23bar but nevertheless someone requested to buy a compressor that can maintain that and the compressor manufacturer slapped an IGV on it.
That same person also requested for a suction pressure of ~2.7bar and for some reason unknown to us too... therefore the manufacturer slapped on an extra stage.
Poor client vendor communication? Or some unknown reason, plant requirement?

Edited by thorium90, 31 October 2013 - 07:31 AM.

[K J]

This is actually feed stock conversion plant in which old gasification based feed stock is converted to NG reforming. Existing facilities of front end ASU and back end syn loop is used. so LP N2 from ASU is compressed to 23 bar and mixed with UOP 12 bed PSA out let pure H2and fed to syn compressor to pressurise up to 71 bar and sent to old plant.

now FIC1002 is equal% valve with 31400NM3/Hr  capacity

       PIC1623 is linear valve with 31400NM3/Hr  capacity

This is cameron suplied compressor with normal suction presure is 2.72 bar and at 3.3 bar it has given trip(why trip at 3.3 bar is not mentioned in any vendor document), normal discharge pressure is 23 bar. one of my supirior was saying this compressor was actualy of high discharge pressure but by this way set for 23 bar otherwise at normal capacity only 23-24 % guidevane opening is very less.

This set up is already establised, does it have any scope of power saving by some operational parameter like lowering suction temp etc.

Edited by k.j., 05 November 2013 - 02:20 PM.

[PingPong]

Lowering suction temps of all stages should reduce the power consumption, but probably only by a few percent.

 

To determine any possible energy saving options one would need all relevant compressor data, such as original design process datasheet, compressor curves for head (or pressure ratio) and efficiency versus flow rate and position of IGV's.

 

The DCS screen data that you posted earlier are dated 12-04-2013. Is there a special reason why you posted these rather old data?

Edited by PingPong, 06 November 2013 - 05:34 AM.

[Shannoncomp]

Check the power supply phases to reduce power consumption.Mostly the power consumptions up or down then try to reduce power supply only selecting mode of power.Don't change the power mode while compressor is still working.

[curious_cat]

 

The N₂ from the ASU is delivered at 4.7 BarG & 21 ^oC. At the inlet of the second stage the condition is 5.4 BarG & 38 ^oC. That is only a 12 % increase in absolute pressure and only a 6 % decrease in actual volume flow, but at the expense of some 800 kW power. Ridiculous.

 

 

@PingPong:

 

How did you figure the 800 kW power usage of Stage 1? Just curious.

 

Sorry, I'm not too good at reading DCS outputs so trying to practice a bit interpreting this diagram.

 

Am I missing an instrument? I did notice the 167 Amps current. Did you use that to infer the power? Even so, isn't that a common motor for all three stages?

Edited by curious_cat, 16 November 2013 - 08:58 AM.

[PingPong]

Gas power of any compressor stage can simply be calculated from flowrate, inlet- and outlet-temperature, and gas specific heat.

It is simply the enthalpy increase of the gas.

 

Shaft power is a little higher than gas power due to bearing losses.

 

Electric power consumption is again a little higher than shaft power due to motor efficiency less than 100 %.

Edited by PingPong, 16 November 2013 - 09:33 AM.

[curious_cat]

Gas power of any compressor stage can simply be calculated from flowrate, inlet- and outlet-temperature, and gas specific heat.

It is simply the enthalpy increase of the gas

 

Ah ok! Thanks. That makes sense. I was mistakenly assuming you read that off the DCS display somehow. 

[thorium90]

Actually, it is possible to infer the power consumption from the screenshot. Taking a single shaft motor (that 167.3A is referring to the entire compressor as there is only one motor. The picture might draw the stages separately but its still one compressor.) . Assuming it takes 6.6kV, 3 phase with power factor 0.90,

 

6600 V * 167.3 A * 3^0.5 * 0.9 / 1000 = 1721 kW

Edited by thorium90, 16 November 2013 - 11:02 PM.

[curious_cat]

Actually, it is possible to infer the power consumption from the screenshot. Taking a single shaft motor (that 167.3A is referring to the entire compressor as there is only one motor. The picture might draw the stages separately but its still one compressor.) . Assuming it takes 6.6kV, 3 phase with power factor 0.90,

 

6600 V * 167.3 A * 3^0.5 * 0.9 / 1000 = 1721 kW

 

Yep, that's how I'd tried to do it. Is 6600 V reasonable for this sort of compressor? The high V threw me off at first.

 

I had some other semantic questions about the DCS screen:

 

(1) What are XV1051 and XV1052? Notation makes me think they are actuated control valves? But are those for shutoff duty since they have other valves for flow control?  What makes me unsure is whether it is typical to measure differential pressure across them (PDI1007 & PDI1008)? Why? To verify they are fully open?

 

(2) What is PDI1624 measuring a differential pressure across & why? 

 

(3) Where we have three similar instruments A,B, C is that a high low selector for redundancy of critical measurements or some sort of voting system? What's the  N in the diamond standing for? The "S" modifier in PIS1623 & TIS1630 makes me think these are high / low selectors depending on H / L but not sure. 

 

Apologies if the questions are too naive or nitpicking. 

Edited by curious_cat, 17 November 2013 - 04:18 AM.

[PingPong]

Gas power of all three stages is: 810 + 760 + 1040 = 2610 kW

 

Shaft power of compressor will be about 2700 kW including bearing losses.

 

To provide that shaft power with 167 Amp, power factor 0.9 and motor efficiency 96 % indicates that this motor must be operating on 11 kV.

 

 

 

(1) XV valves are on/off valves, probably related to ESD system, as well as compressor start-up sequence.

PDI over these valves are probably meant to avoid that valves can be opened while pressure difference is still too big (during start-up sequence when compressor is not yet fully pressurized).

 

(2) PDI1624 measures pressure difference over suction strainer, to detect possible fouling/plugging.

 

(3) A,B,C instruments are likely 2 out of 3 voting system.

S modifier is not Selector but probably a Switch, signalling to connected controller whether its setpoint is reached or exceeded. Or maybe it simply means Signal.

Edited by PingPong, 17 November 2013 - 06:00 AM.

[K J]

@ ping pong and cat

 

(1) This is 3.3 MW 11KV motor for drive compressor, all three stages are driven by same mator by means of bull gear.

(2) stage 1 and 2 are at 24500 RPM and 3 at 39100 RPM speed.

(3) XV1051/1052 are shut off valves.

(4) PDI1624 is suction strainer PDI indication.

(5) PIS and TIS are 2 out of 3 voting switches signaling to ESD system.

(6) N in diamond standes for start permissive and H for trip parameter.

 

Actually my supirior was telling that this compressor is modified from supplier's standard design of 35 bar discharge pressure.

[thorium90]

As much as it might seem convenient to simply blame poor efficiency on an unoptimized design caused by the manufacturer using some standard design, i think the reason could be more practical than you thought. I've seen cameron designs and i think the reason for the suction pressure to be brought down to 2.7 is to accomodate any pressure drop in the ASU to this compressor. This means the compressor will still work even if the ASU pressure goes down. Because of the use of the IGV, i believe dynamic control will suffer if the suction pressure were to vary too much. I cant prove it now, but if you are willing to share the PID settings i might be able to prove that under certain suction fluctuations, the compressor can become unstable. I always wanted to try that but never had a real compressor to test with.....
Anyway, in short i think this was done for reliability rather than efficiency. Perhaps you could also ask cameron yourself on the design methodology?
Also, 11000*167.3*3^0.5*0.9/1000=2868.7kW
Not quite that far from Pingpong calculations...

Edited by thorium90, 20 November 2013 - 09:09 AM.

[PingPong]

11000*167.3*3^0.5*0.9/1000=2868.7kW

That is electric power consumption.

 

To obtain shaft power, multiply with motor efficiency, say 96 %:

 

So shaft power = 0.96 * 2868 kW = 2753 kW

[thorium90]

Unfortunately, motor efficiency is not easily measured and is not done here either. However, the amps consumed is shown and is an easily measurable quantity. It is therefore more reasonable to calculate power consumed than shaft power. Motor efficiency is a function of stuff like power supply quality as also mentioned in post 14, harmonics, system load and other stuff like friction, vibration etc. All of which may or may not be included into a single lumped factor of 0.96.

Edited by thorium90, 06 December 2013 - 08:40 AM.

[PingPong]

I earlier estimated a shaft power of about 2700 kW from the process data.

 

Estimated from the electric data the shaft power seems to be about 2750 kW.

 

Like cos phi data, also motor efficiencies are reported by vendors. 96 % is a typical number for this size motor and voltage.

However for this motor it could very well be a percent higher or lower, but so could cos phi.