19 replies to this topic

[ayan_dg]

General thumb rule says that Control Valve should be sized in such a way so that the "Normal flow" through the valve occurs at 70 % of the valve opening.

For pump minimum flow line the normal flow & maximum flow are same so in this case should we size the valve in such a way so that pump minimum flow takes place at 70% of valve opening or at some higher value say 90 %

[katmar]

Is it true to say that the maximum and normal flows are the same? I presume you are referring to a "spill back" line that ensures that there is always flow through the pump. The cheap and simple way to do this would be to install a restriction orifice that gives a constant flow of (say) 10% back to the supply tank. In this case it would be true to say that the maximum, normal and minimum flows are all the same.

But if you are going to the trouble and expensive of installing a control valve in the spill back line, then presumably the valve will open and shut according to the flow to the process stream. If more than 10% of the pump's capacity is flowing into the process then you can have zero flow in the spill back line, and I would regard this as the "normal" flow. The spill back flow rate is not too critical, so I would happily size the control valve at more then 70% capacity. In fact I would be tempted to size it at 100% and use the simplest cheapest valve available.

[ankur2061]

Ayan,

I am providing a schematic for centrifugal pump spillback or minimum safe continuous flow loop using a control valve. The notes along with the schematic give a brief insight into the operation and sizing of the spillback loop along with the control valve.

Will appreciate comments from knowledgable forum readers.

Regards,
Ankur.

Attached Files

•  Centrifugal_Pump_Min_Flow.xls   29.5KB   2103 downloads

[djack77494]

Ankur,
Your schematic with accompanying notes clearly convey the concept. Please excuse my bluntness, but I think your schematic can be improved upon. First and foremost, and especially if you have a single pump as shown, I would branch off with the minimum flow line close to the pump. My preference would be inside the discharge isolation valve even though this may result in the need for an additional isolation valve in the minimum flow line. The main advantage to this approach is that the pump would be protected even if the discharge isolation valve were inadvertently closed. This configuration would also work for a multiple pump installation. Secondly, I have always preferred to measure the TOTAL flow being discharged from the pump, and to use this process variable for controlling the minimum flow control valve opening. I really like the very direct approach of using the variable that you really want to control, if at all possible. Face it, you don't really care about the discharge header's pressure. You do care about the total flow through the pump. Why not measure and control that variable? Also, your pressure sensing point is too far downstream. You could sense a low pressure if the isolation valve is closed or the check valve is sticking, even though the pump flow is small. I'd locate the pressure sensing point as close to the pump's discharge flange as possible. Same for the local PG, but be sure to use two separate taps. I do like the fact that you ran the minimum flow line back to the suction vessel and not just into the suction line. That is important. Also, I agree that a pressure control loop would probably save you a bit of capital, but I think the added expense of a flow control loop is well worth it.
Regards,
Doug

[ankur2061]

Dear Doug,

I appreciate the fact that you have done a critical review of the sketch provided by me.

Please note that I have not shown all the valves that are required to be provided. For example, I haven't shown the isolation valves to the control valve and neither have I shown a by-pass for the control valves which would definitely be required if you are providing an engineering P&ID.

A flow control loop as suggested by you in the forward flow line with the flow measurement element in the main discharge line and the control valve on the recirculation line is also a viable option.

I do prefer pressure control loops because pressure as a variable has the most dynamic response compared to other variable such as flow, level etc.

In fact I would appreciate if readers of this post would describe schematically the various logical minimum safe flow configurations that they have seen or implemented.

However, the underlying principle for a minimum safe flow is that the pump should not run at shut-off leading to heat generation and destruction of the pump sealing system and in this context I disagree with you that the discharge pressure is not important.

Doug, maybe if possible you can send some typical configurations of minimum safe flow as a schematic as adopted by some engineering/process licensor copies.

Meanwhile I will also try to develop some other schematicss involving minimum safe flow with differences than what I have shown in the sketch provided.

Regards,
Ankur.

[JoeWong]

Doug / Ankur,
Interesting discussion...Just can not stop myself to step in...


Doug,

...First and foremost, and especially if you have a single pump as shown, I would branch off with the minimum flow line close to the pump. My preference would be inside the discharge isolation valve even though this may result in the need for an additional isolation valve in the minimum flow line. The main advantage to this approach is that the pump would be protected even if the discharge isolation valve were inadvertently closed. This configuration would also work for a multiple pump installation.

Agree.

Secondly, I have always preferred to measure the TOTAL flow being discharged from the pump, and to use this process variable for controlling the minimum flow control valve opening. I really like the very direct approach of using the variable that you really want to control, if at all possible........ You do care about the total flow through the pump. Why not measure and control that variable?

Agree.

Also, your pressure sensing point is too far downstream. You could sense a low pressure if the isolation valve is closed or the check valve is sticking, even though the pump flow is small. I'd locate the pressure sensing point as close to the pump's discharge flange as possible. Same for the local PG, but be sure to use two separate taps.

Agree.

I do like the fact that you ran the minimum flow line back to the suction vessel and not just into the suction line. That is important.

Agree.

Doug... we have many common understanding and agreement...

Face it, you don't really care about the discharge header's pressure.

Agree with comments.
If the pump having drooping curve and the suction pressure can vary much, you may have interest to maintain pump head. You will concern about the suction and discharge pressure.

For many cases, i agree with you the discharge has less / no concern from pump minimum flow protection perspective.

Also, I agree that a pressure control loop would probably save you a bit of capital, but I think the added expense of a flow control loop is well worth it.

Emmm....I struggle quite a while to figure out how it save a bit of capital... isn't the flowmeter will have same range when you locate it closed to user side (like the sketch presented by Ankur) or locate it at pump discharge ??



Ankur,

I do prefer pressure control loops because pressure as a variable has the most dynamic response compared to other variable such as flow, level etc.

For a centrifugal pump control, i feel that your don't need a very sensitive & quick response. Response of a flowmeter is sufficient. If pressure control in reactor, probably your argument stands...

However, the underlying principle for a minimum safe flow is that the pump should not run at shut-off leading to heat generation and destruction of the pump sealing system and in this context I disagree with you that the discharge pressure is not important.

For a centrifugal pump, the pump head and couple with flow. If the flow is managed, so do the pump head.

If the suction pressure is vary much, infact the pressure control would lead to unnecessary recycle...


Hope above does not make uneasy feeling...

[Zauberberg]

Keeping the pump in a flow condition which is always above the minimum flow, is most effectively done by providing flow measurement in both discharge and recirculation line - especially if pump belongs to critical service where fluid overheating and subsequent seal failure could cause fire hazard (i.e. pumping fluids above their autoignition point - HGO, long residue, FCC slurry etc). In such cases it is more like mandatory practice, rather than engineering judgement.

Configuration presented by Ankur is very common one, but mostly used for other purpose - when the pump is feeding downstream process unit and excess feed is continuously being sent to a storage tank (or another unit, but this is not a frequent case). I completely agree that pressure controllers are by far the fastest loops in the plant, but properly tuned flow controller can do the job without any problems. That is why level controllers are cascaded with flow controllers (from a suction vessel or column bottoms) - because FC offers faster response and higher flexibility in maintaining set point of a control loop, compared to single LC/LV configuration.

[katmar]

There are two points that I question

I do prefer pressure control loops because pressure as a variable has the most dynamic response compared to other variable such as flow, level etc.

At the point where you need the spillback valve to start opening the pump curve is most likely to be very flat and it could be difficult to get stable control because relatively large changes in the spillback valve position will have a minimal effect on the pump's discharge pressure.

Also, I agree that a pressure control loop would probably save you a bit of capital

How so? The flow element already exists in the sketch and the DCS could control the spillback valve from this signal, saving the cost of the pressure transmitter. The cost of the valve would be the same, whether it was linked to the pressure or the flow.

[djack77494]

Wow, this topic has generated a lot of commentary in a short period of time. Ankur, I will reply to your response before addressing any of the many other comments presented.

I appreciate the fact that you have done a critical review of the sketch provided by me.

You are quite welcome and I hope that my comments are taken in the constructive manner in which they were intended.

Please note that I have not shown all the valves that are required to be provided. For example, I haven't shown the isolation valves to the control valve and neither have I shown a by-pass for the control valves which would definitely be required if you are providing an engineering P&ID.

Understood. We are in agreement here and in most of your points that follow.

I do prefer pressure control loops because pressure as a variable has the most dynamic response compared to other variable such as flow, level etc.

I am not appreciating the point you are trying to make here. In my mind, pressure and flow loops are both very fast. Compare either to level or temperature and I'd be in total agreement with you. As soon as the flowrate through the pump changes, the flow measurement element and transmitter should immediately sense this and send the signal back to the controller. Similarly, movement of the control valve would effect an instantaneous change to the controlled flowrate. So, I have no concerns over the dynamic response of a flow control loop.

In fact I would appreciate if readers of this post would describe schematically the various logical minimum safe flow configurations that they have seen or implemented.

However, the underlying principle for a minimum safe flow is that the pump should not run at shut-off leading to heat generation and destruction of the pump sealing system and in this context I disagree with you that the discharge pressure is not important.

Doug, maybe if possible you can send some typical configurations of minimum safe flow as a schematic as adopted by some engineering/process licensor copies.

Meanwhile I will also try to develop some other schematicss involving minimum safe flow with differences than what I have shown in the sketch provided.

Extra diagrams of suggested systems would be nice. Hopefully we'll see some. As for me, I have to figure out how to do that.

[djack77494]

At the point where you need the spillback valve to start opening the pump curve is most likely to be very flat and it could be difficult to get stable control because relatively large changes in the spillback valve position will have a minimal effect on the pump's discharge pressure.

Harvey makes a good point here, and I'm very inclined to agree that it leads me to flow controlled spillback as being the better (but certainly not only) solution for many pump applications.

Also, I agree that a pressure control loop would probably save you a bit of capital

How so? The flow element already exists in the sketch and the DCS could control the spillback valve from this signal, saving the cost of the pressure transmitter. The cost of the valve would be the same, whether it was linked to the pressure or the flow.

The flow measuring element shown in the schematic is downstream of the spillback branch. Thus only flow forward is measured. My description called for the pump's total flow to be measured, so I assumed an additional flow measurement of this variable would be needed. At this point, we're comparing the cost of an orifice assembly and flow transmitter against a pressure transmitter. Advantage (cost) to the pressure transmitter, though the advantage should be minimal.

[djack77494]

JoeWong / Zauberberg,

Doug... we have many common understanding and agreement...

Now it's my turn - Agree.

Gentlemen, there's nothing that either of you say here that I disagree with. Thanks to all for the interesting discussion.

[ankur2061]

Gentlemen,

Two interesting web links

http://www.chemicalp...s/2007/109.html

The above link gives some schematics for minimum continuous flow

http://www.lawrencep...vol3_i1_jan.pdf

This link gives the calculation equation for minimum continuous thermal flow.

Would request forum members to give a schematic of the minimum continuous flow encountered by them during engineering or operations.

Regards,
Ankur.

[ayan_dg]

Thanks to all for providing valuable inputs. The scheme for pressure control I feel , is best suited for application where normal flow & peak flow varies by a great extent for example take the case of Potable water supply pump in a refinery/ petro complex plant. the load will be maximum at morning , shift end time but will be very less at night. But you have to continuously operate the pump throughout the day.
For such a pump minimum flow control by means of Pressure measurement at pump discharge is a better method. Simply because of the fact one does not know exactly what wii be the resistance of the circuit at that time. If you go for flow control say the pump should always operate at 50% of its rated flow then it may happen that at certain case one does not get enough pressure at the potable water taps (especially which are at higher elevation).
So for such cases pressure control is a better option.

[djack77494]

ayan_dg,
I don't want to dissuade you from your chosen control scheme. Nor am I trying to suggest one is better than another. I can state, however, based on your latest post, that you have not properly understood the scheme I had suggested earlier.

The toal flow control approach would result in ZERO flowrate in the minimum flow line during such times as there was an adequate forward flow. Only when the downstream demand had fallen to the point where the pump's minimum flow requirements were not being met would the flow control valve open and allow "spill back".

[c.vijay]

helo,
me vijay i have some basic doubt in pump spill back.Can u pls guide for the same.
Say i have a pump pressure 5 bar pumping 5kg/hr.
my process requirement is maximum 2.5kg/hr.
considering my pump is pumping around 4.5 kg/hr
So with out spill back what will happen also if i provide spill what are the things to be considered.
vijay

[katmar]

Vijay, you will get better responses if you start a new thread, rather than tacking your question onto the end of an old one. This way you can give it a better title - one that will hopefully attract engineers who have relevant knowledge and experience. Use a title like "Control of low flow pump".

Also, the thread you added to was relevant to centrifugal pumps (although that might not have been explicitly stated) and there is almost no chance of you gettting a centrifugal to do the duty you want. The control of positive displacement pumps is entirely different from centrifugals. It may well be that a spill-back is the answer, but unless you specify what pump you are using we cannot be sure.

[sachindhopade]

hope this helps u
http://www.processcalculator.com/

[ankur2061]

hope this helps u
http://www.processcalculator.com/

Sachindhopade,

are you some kind of an agent of the above mentioned website or just a plain idiot who doesn't know what he is talking about.

[deltaChe]

Dear Sir.
It is nice to learn all of experienced master in these forum.
I feel embarrassed to put my comment in the following.

If this is Automatic bypasses, it should branch off from pump's discharge line upstream of the first valve, whether it be a check valve, a block valve, or control valve ( except for these forward flow by passes). These minimum-flow bypass can protect tow or more pumps in the same service, such as a pump and its standby spare pump, but it still have two danger:

1. If one or more of the pump that has a common bypass branched off downstream of the check valves has a variable-speed drive, such as a steam turbine, the speed setting may not always equla the other pump's running speed. When it is desirable to run both pumps
at once, the speeds may differ enough so that the higher-speed pump will develop enough pressure to hold closed the check valve of the lower-speed pump. In such a case, the lower-speed pump can then overheat and fail.

2. Far worse for common bypasses is what can happen if a running pump is tripped or stops( say, due to a power failure or operator turn off the discharge valve first then suction valve), and the other pump for any reason cannot be started. Then the common bypass automatically opens and allows the discharge system under high pressure to spill back into the suction vessel. At least such vessel exploded violently due to the wrongly placed bypass branch-off location. Even non-common individual bypass, If branched off downstream of the discharge check valve may cause such a danger.

[Mohankumar]

Dear Sirs,

Very similar to this, I am facing a problem of selecting a pump for the purpose of Flow meter testing system. I have explained my problem with a sketch in the attached file. I think whatever you have discussed so far is very much applicable to my problem. Can you please study ny problem and suggest a solution?

Mohan

Attached Files

•  Pump Requirement.doc   106KB   303 downloads