Tripping of Compressor While Starting Another Compressor in a Different Building

Could be due to mechanical resonance!

Hi electricalexpert65, can you elaborate, how it could be due to mechanical resonance?

The only thing mechanically connecting to these two sets of compressors 'X' and 'P' (at two different buildings) are two interconnecting piping between their common discharge headers. While starting P2, we kept vibration analyser at one of the interconnecting pipe near 'X' common discharge header, it was indifferent.

One more information is transformers supplying 11 kV power to a particular group of equipments are as follows:

Transformer-1, Section-1 : Z4,Z5,X1 & P1

Transformer-2, Section-2A : X2 & P2

Transformer-2, Section-2B : Z1,Z2,X4 & P3

Transformer-3, Section-3 : Z3,Z6,X3 & P4

It could be seen that the problem faced are mainly with compressors P2, X2 & X4 which are connected with transformer-2. Vise-versa is also true, when X2 is started, effect is observed on P2. Vibration is abnormally increasing on compressor shafts, but increase in motor vibration is not significant. However, increase in motor current observed during that time.

For similar effect many other combinations of compressors also observed, but on lower scale.

Hope, I provided enough information for experts to react. Still more can be provided if required to zero-in towards root cause. We have conducted several experimental start-ups to find root cause but still straying.

Your list shows 3 transformers feeding the groups of compressors. Are the transformers all identical? What is their MVA rating? What conductor size is being used for the feeds to the compressor motors? Transformer 2 has more total load. Section 2B has as much load as each of Transformers 1 and 3, so Section 2A is added on top of that. Could there be a significant voltage drop in Transformer 2's feeders when you start up the offending compressors? Have you put a power analyzer on the circuits to see what happens to the voltages and currents? Big motor start = big inrush current, maybe voltage drop below acceptable levels? Does this happen when the Section 2B loads are not all running (i.e., reduced load on Transformer 2)? Maybe you need a larger transformer for the 2A & 2B loads.

I'm not sure how that translates into mechanical vibration of the compressor shafts, but if the motor is having trouble getting adequate starting power, there must be some effect.

Thanks PeterT for your comment.

Transformers are all identical. Conductor size for compressor 'X' is 3x300 sq.mm and for compressor 'P' is 2x300 sq.mm. Actually I have mentioned 11 kV loads only. These transformers have huge capacity and are connected with several other loads, for example compressors Y1 to Y6, which are 6.6 kV induction motors. Transformers have almost equal load distribution. Voltage drop while starting rarely go below acceptable level.

like this phenomenon

Since they have the #2 transformer in common has all the wiring and connections been checked? Lose connection may cause those type of symptoms. Maybe problem is in the core of the transformer. You don't give the power of the transformers. If the wiring not the problem and you have two transformer of the power rating that #2 is. Swap them out and see if the problem follows the transformer.

Our electrical counterparts have gone through these issues. Ones connections for some loads were swapped, but no encouraging result. I shall try to get more information on it.

If they are all synchronous, I would suspect a problem with your excitation supply. It may be that the added load is causing your other compressor motors to slip out of synch, called slipping a pole. That can cause vibration. Do you have synchronous protection modules on all of there controllers?

Compressors groups 'X', 'Z' and 'P' are synchronous, 'Y' are induction. We have not suspected problem with excitation supply. Can you explain how starting a compressor can cause "Slip out of synchronous" to running compressors in other building? I understand that speed of synchronous motors already running are governed by grid frequency.

Yes, we do have synchronous protection modules on all of the controllers.

This problem could be due to the surge of the running compressor when starting another compressor connected on the common discharge piping. Is there check valves installed in the discharge piping ? Surging of centrifugal compressor is a momentary back flow of compressed gas in the turbine, typically occuring when head pressure become too high. Have you tried to slightly reduce the load of the running compressor while starting the other one ?

JPL.Thierion-Cesard

Thanks JPL-Thierion-Cesard for your concern but it is not at all the case of surging, because if compressor surges, its flow rate and discharge pressure also fluctuate. Here affect is only on vibration. I want to further clarify here that compressor groups 'X' and 'P' are meant for totally different consumers. Only its common discharge headers are interconnected for any eventual use because both compressor groups handle air. These interconnecting lines are not charged during observations.

My first thought was a back pressure and loading issue but I now understand that the headers are not interconnected under normal configuration, which is the condition for this observed phenomenon, also verified by no discharge pressure fluctuations.[p/]

The observed motor current increase could indicate a load increase, a voltage drop or pf change.

Is the pf correction device working correctly in auto?

In manual what can you actually change manually or is the pf correction static when in manual?

During the startup with vibration episodes have you ever measured the rpm of the motors that appear to be interacting and manifesting as vibration? What is the vibration frequency/period especially at what appears to be a resonance from 8-13 seconds after turn on? What is the rpm delta when this occurs?

Intriguing.......

GA Wal for trying to understand the problem. Here I shall try to clarify.

Is the pf correction device working correctly in auto?

Yes, it is. Normally pf is 0.9 leading. In auto, excitation current is regulated automatically to maintain pf.

In manual what can you actually change manually or is the pf correction static when in manual?

The pf correction device can also be put in manual. In manual the excitation current is set as constant. So, while starting-up other compressor, pf of running compressor may drop to some extent, say up to 0.85 and again come back to normal. We experimented keeping alternatively one of the compressor in manual mode, but no any advantage.

The observed motor current increase could indicate a load increase, a voltage drop or pf change.

Load (flow rate and discharge pressure) in running compressors were unchanged. The observed motor current increase in running compressors could be due to drop in voltage or pf.

We do not have facility to measure motor rpm of running compressors or which is started. We can try to take coast-down plot (BODE) while stopping.

Normally pf is 0.9 leading.

I am hoping that we have a difference in perspective on leading (current vs voltage) here. I would figure pf should be 0.9 lagging (with vars lagging vs vars leading). As I read your comment, I assumed (based on leading pf) that you to meant that you are using lower excitation which can easily lead to slip as mentioned in an earlier post. If the excitation is high enough, I would assume that you meant 0.9 lagging. How is your pf set or monitored on your system? Is there a switch that changes from leading to lagging, or is there a sensor which gives an indication of leading or lagging?

If you are actually at 0.9 pf leading, I would expect slip of synchronous motors to be your problem.

Thanks RG2 for your comment. I discussed about it with our electrical colleagues. It seems the possibility you have described can occurs in generators and not on synchronous motors.

How long has this phenomenon been endured? Is it getting worse/staying the same?

Just out of curiosity, what is the compressed air (it is air, right?) being used for? Irrelevant, but that seems like a big compressor to me.

We noticed this since 2007. Compressor sets 'P' have separate operation group, so some earlier incidents might have gone unnoticed. It is varying.

'X' are main air compressors for feeding a large air separation plant. 'P' are for general plant air supply.

Paralleling machinery whether electrical or mechanical, is not without the risk of unintended interaction between machines, loads and/or control systems. I found this article at

http://www.ccijapan.co.jp/products/pdfs/cci/274.pdf

"..Regardless of design or application, centrifugal compressors have one common characteristic. At low capacities, orderly gas flow through the compressor is impossible since gas velocities are too low to be converted to the required pressure-energy level in the discharge line. When discharge pressure exceeds impeller discharge pressure, backflow occurs until the discharge pressure is less than impeller outlet pressure. At this point, flow reversal again occurs, and compressor operation becomes unstable, resulting in excessive vibration and probable damage to the compressor. This phenomenon is called stall and produces a surge in the system. As a function of its design. every centrifugal compressor has a stall-flow surge point at any given operating speed. The gas flow corresponding to this surge point is fixed for a constant speed compressor, but changes at each operating speed for a variable-speed compressor.

In either case, good industry practice dictates that compressor gas flow should never drop below a margin of 105 percent of the surge point. It is for this purpose that
compressor recycle valves are used to appropriately recycle gas flow from compressor discharge to suction to prevent operation below this critical flow..."

Hope this helps you understand what may be happening.

Dear RAMConsult, Please see my comment at #12. it is not at all a case of surging or stalling. As I am handling large centrifugal machines since over 25 years, I understand this phenomenon. I request you to look for other possible reasons and help me out.

Pritam, are you discounting the possibility of surge phenomena based on nil observable fluctuation of discharge pressure and flow rate only? Or, do you have another reason for your very firm belief?

We are at a point where we need to review all our observations and start to collect data, check what is working, reconcile the collected data and possibly recommission some subsystems in order to determine exactly if everything is ok.

If you don't know what is broken find out what is working

You do know that you have put up with it for 4 years now?....

When you call in consultants to help, they need to be able to confirm observations and review the history in order to properly understand and help (or give up)

Thanks Wal for your suggestions. I'm open to all suggestions, definitely not for giving-up. In late eighties and early nineties I was personally involved in commissioning and operation of these machines and later in mechanical overhauls. I have encountered and felt several surging situations and based on those observations and experience I could say that it's not the case of surging.

But still as you suggest, we shall re-look the data, conduct more experiments and come back.

No, not from the initial commissioning of the installation. As I replied earlier, we noticed it from 2007. Lot of piping modifications were done particularly at compressor building where 'P' are installed. Installation and commissioning were between 1987 to 1990. That time there was no any interconnection between these two sets of compressors. 350 NB interconnection was made in 1991 and 450 NB interconnection in 2001. There were several other small piping modifications at location 'P' during 2006 and 2007. But we are unable to check its effect. Could you suggest something?

Earlier we have checked the tightness of the shutoff valve(s) between the headers, they were good. We shall check again.

Is there some kind of vibration isolators installed in the interconnecting piping ? If not I would suggest to look if they can be easily found and installed on site on 350 and 450 NB pipes. Is it also feasible on site to temporarely disconnect the interconnecting pipes and check if the problem of vibration goes away or not ? Rgds

Thanks JPL, we have already planned for the experiment with disconnection, may be in another two weeks. But it is intriguing, how these pipes with several bends and welded supports transmit vibration to only two of the four compressors in other building? Can you throw some more light on it?

We do not have vibration isolators in these pipes.

A good point worth examining, but can it develop many years after installation? Motors are soft start, compressors' load control is by pneumatically operated inlet guide vanes. On other points I will come back later.

How far from the impeller discharge is the discharge airflow and pressure monitoring apparatus?

Can you monitor the pressure inside the impeller housing?

What type of sensors are being used here?

You mentioned that the voltage drop on motor start is within specs. Does that mean that a voltage drop is observed? Does it coincide or peak in concert with the current increase during vibration episodes? Any figures?

You did say that you couldn't measure RPM. That's too bad. Can you not borrow a hand held non-contact tacho from a mate? Do you have or have access to a variable frequency strobe light?

PT's are installed at around 1 m distance from the impeller discharge. Flow monitoring is at the down stream of after cooler (much away from compressor proper). We can't monitor pressure inside the housing.

Yes, voltage drop is very common while starting huge HT machines. Compressor 'P' take about 17-18 seconds to synchronise (1500 rpm) after starting. Voltage drop is observed up to first 6-7 seconds then comes to normal. It does not coincide with the increase in vibration episodes. Increase in vibration is observed at the compressors X2 and X4 in other building while starting P2. While stopping P2, in captured BODE plot we observed vibration peaks between 900 to 1050 rpm (may be critical speed range). Interestingly it corresponds with the time 8 to 13 seconds during which the vibrations in other compressors at 150m away are rising abnormally.

Actually, our compressor group 'X' does not have key phaser which is required for capturing speed. We have several offline vibration analysers besides on-line monitoring. Any way we are trying to do something for it.

Hey! I also have a problem with my air compressor. When it runs it fills up with air once and trips the breaker. What should I do?