How to start backup BFW pump without deadheading?

[KernOily]

Good afternoon guys. A (semi) dumb question for you from a pumping system newbie.

I have a large steam generation system consisting of three 900 hp split-case 3600 rpm synchronous speed pumps feeding ten steam generators. They are on auto-transformer start because the power system won't take A-T-L starts. Minimum disch flow for these pumps is 250 gpm. Full load feedwater demand is 1226 gpm into 3810' TDH. Feedwater temp is 200 deg. F. Three pumps, two on-line, one 50% hot standby. This is a constant-head application, i.e. as steam generators are brought on and off-line, I still need 3810' TDH. There is not much friction loss at all across this system; the system curve is VERY flat. Control is by PLC. The only automatic controls present are the inlet feedwater rate control valves at the generators (Fisher E body) and the min bypass disch recycle valves on the pumps (CCI Drag valves). All other valves are 1500# manual gate valves (I know, overkill... don't get me started...).

These are oilfield steam generators. Feedwater source is an API 650 tank. There is no reheat cycle, no economizer, no deaerator, no steam drum, no condensate recovery, no blowdown, no nuthin'. Feedwater in, steam out, into the ground, and gone forever.

My question is thus. I am designing the pump switchover procedure. Suppose pumps A and B are online and I want to start C and shut down B, for whatever reason. The switchover procedurer is manual.

The Question: How do I start C without deadheading it and ramming it backwards on its curve?

As I see it, the procedure would be to:
(1) hit start button for C pump
(2) C pump completes its pre-start warm-up cycle
(3) block valve in C pump disch lateral is closed
(3) Motor starts and min bypass disch flow control valve opens to 100%
(4) pump is bypassing 250 gpm back to the feedwater tank

Now I'm stuck. If the operator starts to open the block valve on C pump, the system won't take the water so the pump backs up on its curve - yes? The min flow bypass is still open so 250 gpm is still going out through the bypass but no rate is going out the discharge becasue the system won't take it.

I might have just answered my own question. Am I missing something here? Will I back up C on its curve before B goes off-line? I am anticipating the switchover procedure to take about 10 minutes or so.

Thanks guys!
Pete

Thanks!
Pete

 

[BigInch]

The system will try to take the water, unless you have a flow control valve downstream of a common discharge point. Without that control valve, more water will tend to be forced into the pipe leading to the generators. That will have one of two possible effects,

1.) the pumps will fall back on their curve, ie lower their flow while they increse their discharge head towards the shut-off (dead head) output as they attempt to force more flow to the generators,

2.) if there is little resistance to accepting more flow at the generators, their inlet pressure will vary somewhere between what was required for the old flowrate and what additional resistance (if any) is offered to the new flowrate.

If the minimum flow bypass is still open, that flow will be the difference between what the pumps want to flow and what the generator's will accept. It won't necessarily remain at 250, as probably some additional flow will be accepted by the generators and the 250 in the bypass will decrease accordingly.

The changeover can be accomplished by opening the bypass on B to draw off from B, whatever amount the generators are refusing to take. Once B's bypass is open, but not flowing 250 yet, the flow in C's bypass will tend to increase by the same amount, so the net effect is to reduce the flow to the generators by the same amount as well, getting that feed back to 750 gpm. Continue the process each time, increasing the bypass flow of B, reducing the overflow to the generators, getting them back to almost 750, while you bring up the flow in C's bypass to almost 250, with the remaining bit, sent to the generators.

If you do the above in smooth increments, you will wind up with B offline and C online, with only small changes to flow going to the generators, which eventually return to 750 when C's bypass is full closed and B's bypass is full open.

Then hit the stop button on B.


BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[JJPellin]

I seem to have a slightly different picture in mind for the configuration of your system and controls. Starting with your chronology, the C pump is blocked in from the system and pumping 250 gpm through its individual minimum flow bypass. The other two pumps are each contributing about 600 gpm to the total flow. Since the C pump is running at a lower flow than the other two, it has a higher discharge pressure than they do. As you start to open the block valve on the C pump discharge, it pushes flow into the system. As the total flow tries to increase, the pressure controllers at the generators start to pinch back, holding the total flow and pressure constant. As the block valve at C pump reaches fully open, all three pumps must be at the same pressure since they are tied together off the discharges. You still have the same total flow going to the generators. The spill-back at C pump is confusing. If there are no control valves between the discharges of the three pumps, you really can't say that the flow through that spill-back is coming only from C pump. If the spill-back is set up as I picture, the flow from C pump will approach 400 gpm to match the flow from the other three. That is a total of 400 gpm; 250 gpm spill-back and 150 gpm outgoing flow. But, as soon as the total flow exceeds 250 gpm, the spill-back will start to pinch down. At that point, the system should be self stabilizing. As the spill-back on C pump pinches down, more flow from C will enter the system backing the other two pumps down to 400 gpm each. It should be simple matter of tuning the spill-back controllers and the pressure controllers at the boilers. All you should have to do is slowly open the valve at C. Once it is fully open, the spill-back on C should be closed. Then you slowly close the block valve at B. As the flow from B drops, the spill-back at B will slowly open and the downstream pressure controllers will hold the total flow and pressure to the generators constant. Then you shut down B. It is really that simple.

If the spill back is not automated as I describe, it still works the same. As C pump valve is fully open the total flow will be about 1450 gpm; 1200 gpm to the generators and 250 gpm back through the spill-back. But this total flow will be equally shared by all three pumps. Each will be pumping about 485 gpm. Then you gradually close the spill-back at C and each pump will drop back to a flow of about 400 gpm. Then you gradually open the spill-back at B and the flow from each pump goes back up to 485 gpm. Then you gradually close the block valve at B and the flow from the other two pumps increases to 600 gpm each.

On new projects I tend to see a lot of confusion around individual spill-backs in systems like this. The spill-back on C is reading the total flow from C. But, ultimately, once the individual block valves at each pump are open, that spill-back is diverting flow from the common discharge combined from all three pumps. It really isn't an individual spill-back at that point. With all valves open, all three pumps have the same discharge pressure since there is no imposed pressure drop between them. If they are at the same pressure, then they must be at the same flow (assuming all are in equal condition and identical). The system will tend to balance itself out quite nicely. If there is any instability in this process, then retune the control valves to make the transition smoother.

 

[BigInch]

I thought 50% standby of the capactiy of two pumps is 1 pump, but managed to get fixated on the 250 flowrate, forgetting its a minimum flowrate, and wound up with a total flowrate of 750 for the 3, which looks dead wrong now. I think if you substitute 600 x 2 = 1200 (something), for the 750 gpm I have written above, there should be no other major difference in the switchover process. I also agree with JJ's description, the bypass flow will be any excess flow from all three pumps. The three pumps must act together, given that there are no individual control valves between each pump and the common header.

It should work out OK, if the bypass control valve can handle flows from 250, up to say 400, or up to 600 gpm would be ideal.


BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[JJPellin]

Reading back over my earlier comments, they seem a little confusing even to me. I also missed in your original description, the fact that all spill-back valves are fully automated and controlled to make sure each pump is above the minimum flow rate of 250 gpm.

As described, this system should be extremely stable and easy to start, stop or switch pumps. The three pumps will balance out against each other and none of them can be backed up below their minimum flow because each has a spill-back sized to divert the total required minimum.

The only way that any instability or upset should occur would be if a pump were started up or shut down without being blocked in from the system. This would cause a sudden change in pressure and flow that could result in a transient upset while the controllers at the generators tried to respond. In most of our boilers, this sort of a transient upset would not be any concern at all. In fact, the standby pump would normally be set up to auto-start. In the event of a false auto-start, there would be a momentary spike in flow and pressure while the control valve responded. The pumps would share the flow equally and no harm would occur.

We have several boilers set up this way. The procedure to switch pumps should take just one or two minutes. I would step it out as follows:

1. Start up C pump blocked in from the system.
2. Gradually open the C pump block valve.
3. Gradually close the block valve at B pump.
4. Shut down B pump.

 

[BigInch]

Then the only question is exactly how any flow differential (temporary imbalance between bypass flows and excess generator flow, if any) will proportion itself between the generators and bypass, as the transition progresses, which we can't determine without knowing the system, bypass Cv and pump curves.

I recently had a similar system, but with two pump stations each pumping into a 1500 meter (mostly) static head to reach a 2800 meter sumit. The problem was the capacity of the bypass control valve when the operators wanted to start and run on standby for a long time at near 100% pump capacity, while the upstream pipeline started and eventually gave them enough NPSHR to open up to the pipelines. They generally got too anxious, started too early at excessive rpm and overheated the pumps and cut out the CV with a sustained high dP (close to shutoff head) flashing flow (due to the small maximum flow that the CV was capable of handling), while waiting to get enough upstream pipeline suction pressure.

Pump changeovers were easier, at least until they pulled suction pressure down below the trip setting.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[JJPellin]

BigInch,

I really do need to learn how to post a diagram or picture. I think we are describing slightly different configurations. Based on his original description, (and a few assumptions) I am picturing the following:

1. Each pump has a flow measurement that measures the total flow from that pump (outgoing flow plus spill-back).
2. Each pump has a minimum flow spill-back that comes off before its block valve.
3. The spill-back valves are programmed to spill-back as much as needed to keep the flow from that pump above 250 gpm (proportional control)
4. After the block valves the three pump discharge lines all tie together in a common header.
5. The common header branches out on the other end with control valves set up to maintain constant head pressure to the generators.

In this configuration, each pump is protected since it has a spill-back sized to pass the required minimum flow (250 gpm). Once the block valves are open, the three pumps (or two if only two are running) will balance out to equal flow since there is no pressure drop between them. The generators will always receive the correct head pressure based on their controllers. The only way they could get into trouble would be if the generators demanded more total flow than the running pumps could produce. The pumps would then run off then ends of their curves and the pressure would drop. This could happen if they were running two pumps and one tripped off at a time when the generators were demanding more flow than one pump could produce. But during a switch, with three pumps running, there should be no way to have a problem like that. Even if the demanded total flow was less than three times the minimum flow for one pump, the pumps would still be protected by their individual minimum flow spill-back lines.

The difference in our interpretation probably has to do with my assumption about the function of the spill-back control valve. I am assuming it is a proportional control that will open only as much as needed to keep the total flow from that pump above 250 gpm. We have a few pumps that have spill-back valves that are not proportional and open fully if the total flow drops below the set-point. This valve acts like an orifice. In this set-up, there must be a dead-band in the controller on the spill-back so that it does not get into a hunt (open, closed, open, closed). If you are picturing this type of spill-back control set-up, you are absolutely correct that the response during switching cannot be predicted without more information.

But if the spill-back controllers are proportional and the system is as simple as I imagine, it should be very stable and switching should be easy.

 

[BigInch]

J,

Well, I'll tell you there isn't too much sense in our discussing this alone w/o additional input from the OP, and I don't really see much, if any, difference in our understanding of how it should work, the end result appears to be the same, but what the heck... it could be interesting.

Your configuration sounds perfect w/o any diagram needed, whether the valves are PID, or O/C doesn't make a whole lot of difference in the end, only how it gets there, bit by bit travel or by a one shot full open.... which really just makes the transients larger in the process of getting to the smooth steady state end result.

I think we may have a slightly different concept in how the CV's work, being that they either work together as an equivalent valve or they work independently of each other. I don't think you would necessarily need flow measurement from each pump, if the bypass valves happened to be sized nicely, they would function pretty well either way. As you say, even an orifice would work (pressure control). And hydraulically, a CV is just an orifice that can vary with actuator position.

I tend to think, for example if 2 bypass valves (BYV) were 100% open, the flow in each BYP would tend to be equal, since the pressures at the "T"s at both the suction and discharge header must be equal for all pumps, and assuming all pump curves are all equal, each pump with an open BYV will tend to be at the same point on their individual curves. The third pump, even though its BYV remained closed, would also tend to reduce it's flow and move to the same op point of the other 2, after the other 2 pump's BYV were opened, since the suction pressure at the common point in the suction header would tend to go up, and likewise the common discharge header pressure would tend to go down. So what I believe is the flow through any open BYVs is the net spill back from all parallel pumps as each pump tends to equalize its operating point with the others. Any unequal operating point on the curve is only a transient point in the move towards total equalization.

Whether the BYV are PID controlled or O/C, or if 1, 2 or 3 were open the net flows will all tend to equalize, depending of course on the time after each one opened. If there were 3 Control (pressure or flow), the flow through each would have to be proportional to their individual Cv settings, using the common and equal pressure drop between suction and discharge T's, which must always be equal for the 3 valves and pumps too.

If you would like to talk more about this, why not get my e-mail from my web page and we talk about this 1-2-1, rather than fill up these pages with our various (& probably erroneous?) assumptions, which really won't get any better without having the benefit of further input from the OP anyway. We can do just as well on our own.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[KernOily]

Guys thanks for the AWESOME replies. I owe you lunch. Next time you're in central California, I'll buy. We have great Mexican food here

I do want to provide a couple more bits of information to hopefully nail this:

- The system is brand new, still on paper. We are currently in P&ID development. The pumps are out for RFQ.
- Each generator has a flowrate control valve (Fisher E-body) and a flowmeter (orifice plate) on its inlet. The normal flowrate setpoint per generator is 4200 bpd, or ±123 gpm, hence the 1226 gpm full load for all 10 generators.
- As I said above, there is no metering or controls in the individual pump discharge laterals, in the discharge bypass laterals, or the combined pump discharge header. The flowrates from the generator meters are summed in the PLC and that summation is then used as the process variable for the control of the minimum bypass valves.
- The system must be able to provide feedwater for one generator online up to all ten online. For one generator, the rate is 123 gpm, so that means one pump online at 250 gpm min flow: 123 gpm going to the load and 122 going to min flow bypass. One pump will serve up to four generators; beyond that, the second pump will have to be started.
- The current plan is for the PLC to control the min flow bypass valves (PID) by looking at the sum of the rates coming in from the generators.

Argh. I just realized this won’t work without a flowmeter in each pump’s bypass lateral. Otherwise I have no signal with which to provide an input process variable to control the position of the bypass valves.

For example, if I have one generator online at 123 gpm, then the pump has to run at 250 gpm because that’s the min flow. So that means 123 gpm going to the load and 122 gpm going to bypass. But the bypass valve must also be able to pass the entire 250 gpm min flow at some % open, say 70%. To throttle the bypass down to 122 gpm I need some input signal to do that. I do have constant backpressure on the bypass header, so I guess I could do it by valve position, since the Cv is known vs. position, but that might get me into trouble as the valve trim wears...




Thanks!
Pete

 

[BigInch]

Right you need a few more FTs in there somewhere. If your pump curve is flat, running off of a PI might be tougher since there may not be a big signal change with dQ. Valve position is not highly accurate, and if there are any major over-swivels before actually returning to the control point... will the PID overcontrol and then undershoot? FT might be more stable.

Whatever...

Then you might ramp up something like this, unloading the gens with B while loading with C.

ramp 1 2 3 4 5 6

PB 615 615 492 369 246 123
gen 615 492 369 246 123 0
byp 0 123 246 246 123 0

PC 246 246 369 369 492 615
gen 0 123 246 369 492 615
byp 246 123 123 0 0 0



BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[ccfowler]

74Elsinore,

In addition to the above discussion, which is excellent and well founded, I would concern myself with two additional issues. (I am presuming that these will be identical pumps.)

1- Where does the minimum by-pass flow go? If it returns to the source tank so that it can mix tolerably well with the bulk flow, then all is OK, but if it is piped to short-circuit back to the same pump's suction piping, then you may have a problem. The recirculating flow can warm fairly quickly, and then the generated head (due to reduced fluid density) may not be enough to provide enough pressure to deliver flow into the header supplying the boilers.

2- Do the pumps actually have a head vs. flow curve that rises continuously to shut-off? If not, does the curve rise sufficiently below the minimum flow rate to assure enough head (and pressure) to overcome the pressure in the discharge header?

As a practical matter, your synchronous motor drive will actually be somewhat less forgiving than the more common induction motor drives. Please bear with me in this explanation.

If these were to be induction motor driven pumps, the operating pumps would, presumably, be carrying greater flow rates and would load with their motors to a greater slip rate (roughly proportional to their load). The starting pump which would be operating at a lesser flow rate (250 gpm) and drawing less power thereby operating with a proportionally lesser slip rate. Thus, the starting pump would be able to develop a slightly greater head due to its greater shaft speed (head being proportional to the square of shaft speed).

While this may seem to be a minor point, I have encountered cases where induction motors operating at greater than expect slip rates actually explained seemingly odd (and troublesome) performance issues.

 

[KernOily]

Guys thanks for the replies. I do appreciate your time, effort, and help thus far.

BigInch - thanks for the verification on the meters. I added the meters to the individual bypasses on the P&ID. I was definitely not comfortable using the control valve position to control the flowrate.

ccfowler - To answer your questions: (1) The minimum flow bypass indeed goes back to the source tank, which is a 5000 bbl atmospheric tank (but which is not open to atmosphere) about 1/4 mile away. Short-circuiting back to a suction line esp. on a high-energy pump is a BIG no-no, something I have to educate the folks around here on, all the time.

(2) The pump curve has a continuous rise but not per API 610. The preliminary curve rises from 3800' at BEP to 4350' at min flow. In the RFQ package, we specified that it was OK to deviate from the API 610 required-percent-head-rise-to-shutoff because we are trying to minimize the amount of head we have to burn up across the generator control valves. I am trying to get as flat a curve as possible so the real curve will likely be less rise than this one. But it will have a continuous rise, albeit small (hopefully).

Thanks guys! Pete

Thanks!
Pete

 

[JJPellin]

I now see that many of my assumptions above were incorrect. Without individual flow measurements at each pump (or in each spill-back line) the control becomes slightly more complicated. But the basic principals still hold. I think you can use a calculated valve position for the spill-back valves. But I would recommend a little more conservative approach because of the possible error (as you noted, when the valve wears, etc.) If the manufacturer's recommended minimum flow is 250 gpm, I would suggest using a higher number for the setting in order to leave some margin for error. It should not be too much of a burden to set the minimum flow for 300 gpm, each. That way if your calculated flow through the spill-back line was off by as much as 20%, you would still be protected. I am also concerned about a few other potential problems.

Since your flow measurement is not located close to the pumps, there are some potential problems. You need to have a good procedure for starting up this system initially. In order to pack the lines from the pumps to the generators, you will have to pinch back on the pump to keep it from running off the end of the curve. Since the flow measurement is at the other end, the PLC will think you have zero flow and will open all valves wide open (spill-back and generators flow control). Running off the end of the curve can do severe damage to the pump. We had this problem in a desulfurizer unit where the control valve was perhaps 150 yards away from the pump. While they were packing the lines, the pump would flow out at too high a rate and cavitate severely. Regular pump wrecks were the result. Initially we implemented a procedure so that they pinched on a block valve and held the pump to constant discharge pressure while the line packed. Eventually, we added an instrumented spill-back and automated the whole process. This unit has to start up and shut down several times per year.

As you were describing, the start-up with one generator and one pump seems manageable. But once you have two, then three then four generators running, it gets more complicated. Will the start-up and shut-down of the pumps be automated too? Coming up in rates does not tend to be the problem. We see more problems as rates come down. If they were running on six generators and needed a total flow of ~740 gpm, they would need to have two pumps running at about 380 gpm each. But if a couple of generators came down (shut down or tripped) and the required flow dropped to ~615 gpm, the two pumps would be getting very close to their minimum flow. The PLC would be satisfied since the total flow was above 600 gpm. But this far back on the curve, any differences between the pumps could be significant. The PLC would assume that the pumps were flowing 307 and 307, but they could easily be pumping 400 and 215. Back on the flat part of the curve, these differences can be amplified. Check out the shape of the curve in that region. Note the amount of head difference between 250 gpm and 400 gpm. You may be surprised that the head difference back there is a pretty small percentage of the full head produced at 250 gpm.

You would be much better off if you could get a flow measurement in each pump discharge line. Measurement of the flow in each spill-back line would not be as helpful. As I noted above, when the valves are open, these spill-back lines are not really individual any more. They come off the common combined discharge manifold. If you can't get individual flow measurements, I would suggest a few other possibilities:
1. If all drivers are motors, you can use motor amps to understand if a pump is getting into trouble. Either alarm on imbalance in amps, or trip the pumps based on low amps. When the pumps are test run at the factory, ask that they run your pumps with your motors at the exact voltage that you have in your system. Ask them to provide motor amps versus flow. Once the pumps are installed, you can do a field test to verify the amp draw when running at minimum flow and set up a low amp trip to protect the pumps.
2. Cycle the start-up and shut-down of the pumps. If this function is PLC controlled, I would not suggest always running A for single pump operation and A/B for two pump operation. As pumps are started and stopped, have them cycle through all three so they get equal run time. If this is not PLC controlled, then have a set switching program that cycles through all three to keep them all with similar run hours.
3. Change your monitoring strategy when running in the dangerous ranges. At five generators running, one pump is running fat and happy. At 10 generators running, two pumps are running fat and happy. But at 1, 2, 6 or 7 you could be in trouble, potentially running at too low a flow for one or two pumps. When you are running in those ranges, have your operators keep a closer eye on the pumps. And when-ever possible increase or decrease rates to get back to safer numbers.
4. Consider adding a high temperature trip on the pumps. We do this on some crude boosters (four pumps in parallel) that are in an unmanned area of the plant. We had a history of one pump (the only turbine driver) getting backed up on the curve and burning up. We installed a high temperature trip that has saved us from several wrecks. And this was much less expensive than individual flow transmitters.
5. Set up a program of testing pump performance. Once or twice per year, take a rate cut for a couple of hours and run each pump individually long enough to get a full set of performance data. You may get some push back if this costs production. But make sure you are balancing the cost of the test with the possible savings. If you catch a weak pump early it may cost $50,000 and take a week to do a planned overhaul. If you wait for it to wreck and do case damage, it could cost $150,000 and be out of service for a month or more. If the lost production costs $10,000 per year, that might be cheap insurance to save $100,000 net on the repair. Not to mention the risk of running with no spare while it is out for overhaul.

You have a bigger challenge ahead of you than I first believed. Good luck.

 

[JJPellin]

Oh, I suppose I should go back to your original question. When switching pumps you are going to have a problem. The PLC will register the total flow but will have no idea that you are opening a manual block valve. The PLC will have no way to know which spill-back valve to open. At that point, it becomes a totally manual process. The process still seems simple but requires that you switch some spill-back valves from automatic to manual operation. How will the PLC adjust its response when one spill-back valve is taken out of PLC control? The process is still basically the same:

Block in C pump.
Put the C pump spill-back in manual and open it 70% (based on your comments above)
Start C pump.
Gradually open the block valve at C pump.
Gradually close the spill-back at C pump.
Put the B pump spill-back in manual and gradually open it 70%.
Gradually close the block valve at B pump.
Shut off B pump.

Depending on how many generators are on line, the risk is that something happens while you are switching. You will have to be very careful about how you program the PLC to respond to a loss of flow. Which spill-back valve will it open and how much? What will it do if one is in manual and blocked open or shut? I would not recommend deliberately switching pumps if you have 6 generators running. The more I think about this (and remind myself about the size of these pumps) the more I think you should push for individual flow measurements at each pump discharge.

 

[KernOily]

JJ - Thanks for your thoughts. It is late here (6:30) so I will print these and chew on this tonight. Reply tomorrow. Thanks again - Pete

Thanks!
Pete

 

[ccfowler]

KernOily (name change from 74Elsinore?),

I would be very careful about wanting very flat curves on any pumps operating in parallel. This is a very easy way to get stability problems with the load shifting erratically between the pumps. What may work tolerably well with fresh, new pumps working perfectly can get very ugly as normal wear and tear take their toll. The relatively flat curves can, also, make starting additional pumps against the operating system much less pleasant.

I've seen parallel pump systems where the flow shifted rapidly (almost instantly) back and forth between pumps. These fluctuations could be much faster than your control system could possibly handle. Again, your synchronous drive system will likely be less forgiving than induction motor drives which could at least attenuate the effects of such fluctuations. The shaft speed of the unloaded (or severely under-loaded) pump in your system will not rise in compensation.

The operation mode that I would fear for your proposed system would be, in sequence:

1- Suddenly stopped (or nearly stopped) flow at one pump.

2- Suddenly reduced discharge header pressure due to net starved supply flow.

3- Other pump(s) run out on its (their) curve(s) attempting to make up the flow deficiency.

4- The first pump regains flow abruptly as the discharge header pressure drops sufficiently low.

This surging can continue between the operating pumps with any one taking the role of the first pump in the above description. Responses by the control system may compound the stability problem.

Perhaps it is my old-fashioned point of view, but I would not want to encounter the risks of unstable operation in this manner. The energy savings could get to look very small against the need to endure a prolonged outage to install yet another set of pumps. Those avoided costs can cover an awful lot of throttling losses. I would want to do a careful economic analysis of the potential cost of needing to replace pumps the pumps vs. the economic burden of the expected throttling losses. I would keep in mind that not all throttling losses are actually losses. The throttling does reduce the energy draw of the pumps.

 

[BigInch]

I also thought the 2-10 was not an especially beneficial configuration.

As mentioned above, system stability is important. Flatter pump curves really mean less of that and can make it difficult to keep things under positive control. Pump and system curves intersecting at small angles over the operation range generally isn't too good.

Good start-up and shut-down simulation candidate.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[BigInch]

Without a discharge control valve at each pump, (or even with them, and/or auto shut-in block valves, or checks) have you looked at what can happen if you lose power to one or both pumps?

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[ccfowler]

KernOily,

Some additional thoughts:

I wouldn't even consider operating any pumps in parallel without effective individual check valves.

With reasonably steep pump curves, the pumps should share the load fairly well, but with flatter pump curves, discharge control valves at individual pumps would be a necessity. "Identical" pumps are never really identical, and one will always hog a greater share of the load. The traditional industry standard for pumps is only +/- 10% of nominal, and with flatter pump curves, that can lead to serious load sharing issues. Even with reasonably steep pump curves, but no individual flow control valves, it is common for the total expected flow rate to be unavailable because of the need to protect the motor of the "proud" pump from overheating.

I would take a serious look at the costs (initial and long term) for oversized motors vs. individual flow control valves for this application. My guess is that bigger motors and steeper pump curves will be the simpler, more reliable, and least costly choice, but each application needs its own analysis.

Regarding stability issues, if you have ever been around parallel pumps operating erratically with sudden severe flow changes, it is an experience not soon forgotten even with relatively lower energy pumps. No control valves could ever keep up with the sudden changes, they are essentially instantaneous.

I presume that your boiler control system will have provisions to deal with sudden loss of feedwater flow conditions.

You will be using orifice plates for flow metering. These are well known devices, but based on some unpleasant experiences, I offer the following comments:

1- Over time, orifice meters almost always tend to understate the flow, and their accuracy gets pretty poor for turn down ratios greater than about 3. (Some significant improvement of accuracy for greater turn downs can be achieved with the application of much more elegant than normal computations in the metering system.) This is just part of life with orifice meters. Since your boilers are operating in an essentially ON-OFF arrangement, the traditional, simple application of orifice meters is an excellent choice.

2- As part of start-up, I would include provisions to make absolutely certain that each and every individual orifice plate's geometry is verified and documented. This includes verifying the geometry of the adjacent piping as well as the orifice plates themselves, and it also includes verifying the concentricity of the orifice opening within the piping interior.

3- Also, as part of start-up, I would include provisions that absolutely assure that each individual orifice plate is inserted in the proper direction. Simply checking the installation for which side of the extension tab is stamped is not enough. I've known orifice plates to be in place for years giving bad readings the whole time because the extension tabs were stamped on the wrong side. The fitters had simply installed them according to the instructions that they were given, and nobody with suitable knowledge of orifice meters verified their proper installation.

4- Even though they are seemingly less costly, I would avoid using flange taps at the orifice meters, if at all possible. Initial accuracy is not as good as the other choices, and long term accuracy tends to be even worse. My personal preference for an installation such as yours would be to use vena contracta taps.

My philosophy about orifice meters is that they are usually a good choice so long as custody transfer is not involved and the facility has sufficiently competent personnel to understand them. If custody transfer is involved, they would be the first choice of the buyer and the last choice of the seller.

Sorry, more ranting by an old geezer!

 

[BigInch]

High head system with no check (or a failed check) could be a heck of a blowback... and reverse flow all the way from the generators???

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com