Multiple VFD Pumps to Common Header 2

[KevinAtBurnsRoe]

I have an application where I have 4 variable speed pumps that can supply feedwater to a common header that in turn feeds up to 4 industrial boilers. It takes up to 3 of these pumps operating in parallel to handle the load of all 4 boilers. There is also a smaller, constant speed pump used for very low loads. The pumps are not identical, 2 are motor driven and 2 are turbine driven. Our goal is to maintain a constant header pressure. I would like suggestions on how to control the speed on the pumps over the load range, including cutting in and out pumps as required. Pump starting and stopping will be manual by operators based on some alert.

 

[rzrbk]

This will be highly dependant on the way the different curves match up. I would suggest by controlling the header at a constant pressure, similar to some firewater systems. Provided that there is a pressure that meeds your process needs and falls in the operating range of all the pumps.

As to logic on starting/stopping pumps I don't know

 

[ScottyUK]

What are the relative sizes of the pumps? This will be significant in how the control scheme can be implemented.

Given similar size pumps I would be tempted to make the electric pumps the main modulating element for both ease of control and efficiency. Turbines usually have optimum efficiency at one speed, and power output drops away very quickly with speed decrease - not a great match for a variable speed pump duty. If the turbine runs at fixed speed then control will probably require a modulating valve on the pump discharge, which is lossy. A nested loop of pressure controller and flow controller would allow the turbine driven pump(s) to hog the load with the electric pump(s) making up the shortfall. Ideally the electric pumps would have a slightly greater maximum output than the turbine driven pump to allow reasonably bumpless control as the fixed speed pumps start up and drop out.



----------------------------------

Sometimes I only open my mouth to swap feet...

 

[BigInch]

You said 2 are motor driven and 2 are turbine driven, but didn't mention if are there any other differences in the pumps themselves. Are the 4 pumps equal in other respects? I suspect they may not be, probably running at different speeds. Do they have the same BEP H and Q ratings?

If there are significant differences in the pumps, optimizing their operating configurations and starting or stopping sequences could be made much easier with a transient flow analysis that includes detailed control equiment modeling capability.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[ccfowler]

The previous comments should be most helpful, but I would want to also consider the type of turbines used and the reasons for using them. Is there some process use for the exhaust steam from the turbines (presuming that these are steam turbines)? It seems possible that the use of the steam pumps could alter the operation and economics of the connected system. Do the boilers serve a common header, are they all the same size, capacity, etc.?

If these turbines are of the size range that I am assuming (relatively small compared to multi-thousand horsepower units), their heat rate probably implies substantial relative energy costs compared to the electrically powered pumps.

Another question that comes to mind is the load profile for these pumps. If most of the time the load is great enough for the electrically powered pumps to operate at or near full speed, then the adjustable speed drives' losses simply represent a needless parasitic loss, and constant speed drive for these could make more sense for both initial and operating costs.

From the information provided, it seems likely that the turbine driven pumps would probably be best kept in reserve for emergencies and high load periods, and the electrically powered pumps probably do not need to have adjustable speed drives. This would allow simplification (and substantial savings) in the control system both initially and long-term. (Boiler feed pump duty is not usually one of the applications showing the greatest energy savings from the application of adjustable speed drives. Some energy savings may be realized from adjustable speed drives, but they are usually quite modest and economic justification is frequently marginal at best.)

 

[BigInch]

Since boilers usually represent a relatively high constant head, a small turndown in pump speed by a VSD control is followed by an unacceptable reduction in pump's discharge head.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[Valvecrazy]

What is the temperature of the water at the pumps? What do you mean by turbine driven pumps?

 

[ScottyUK]

Turbine driven pump: a pump driven by a turbine, probably using steam in this instance.

I'm curious: what difference would water temperature make to the control strategy unless the water underwent a state change?


----------------------------------

Sometimes I only open my mouth to swap feet...

 

[Valvecrazy]

I just wanted to know if what was powering the turbine pumps could be easily or automatically turned on and off. Also, the temperature of the water determines if a pressure reducing type valve could be used or not. Pumps moving cool water can be greatly restricted with a valve without damaging anything. Pumps moving warm or hot water cannot be restricted very much. I do not think that control valves are a lossy way to control pumps. They may actually be the best way in this application.

 

[EdStainless]

Will the target header pressure ever change, or is it a fixed value?
You need to lay these pump curves out next to each other and give them a good look. You aren't going to be able to use VFD, since you need (near) full pressure.
It sounds like you will use the turbine pumps for base load and then turn the electrics on and off to follow load shifts.
And inspect the check valves carefully.

= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection

http://www.trent-tube.com/contact/Tech_Assist.cfm

 

[BigInch]

valvecrazy,

Do you say, "Pumps moving warm or hot water cannot be restricted very much.", because the pumps will overheat faster, since they are already running at the hot water's temperature? If not that, what other reasons are you thinking of?

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[Valvecrazy]

Hot water has no lubricity or cooling ability. When pumping hot water, a fairly large minimum flow, or recirculating flow, must be maintained. Cool water is a very good lubricant and has great cooling abilities. When pumping cool water, a pump can be restricted to a very low minimum flow without hurting anything.

A control valve on the discharge of each pump can be set to maintain a constant pressure on the header. A very small bypass amount can ensure that each pump maintains proper cooling and lubrication. These control valves don't care if the pumps are driven by electric motors, steam turbines, or diesel engines. Pumps can be turned off and on manually, or automatically with a simple pressure.

With each pump having it's own control valve, each pump is only bucking it's own back pressure and not that of the other three pumps. Without a control valve, when pumps are run in parallel, the lowest head pump can be deadheaded at low flow rates by not being able to buck the back pressure of the other pumps.

Also, power consumption drops off as flow is restricted with a valve, almost exactly the same as it does when the RPM of the pump is slowed using a Drive. I think Constant Pressure Valves would be a good way to control these pumps and maintain a constant pressure on the header.

 

[BigInch]

Granted that the temperature of cool water is lower as is the viscosity, so obviously longer running at reduced flow is possible, and lubricity must also be greater.

Temperature is not the only concern. Low flowrates in most pumps do hurt, by causing increased bearing wear due to unbalanced hydraulic forces. Flows less than 60% of BEP are not recommended by API, although with a VSD, the pump would not produce such proportionally high unbalanced forces, since discharge pressure is reduced.

Each pump having its own control valve does not ease the situation. The pressure of the header (assumed to equal the highest discharge pressure of any connected pump) must be reached by any other pump in order for any other pump to discharge into that header. A pump producing less pressure than the header pressure must have its own discharge check valve in order to avoid being spun backwards by fluid attempting to enter from the higher pressure header. Placing a control valve betwen a pump discharging at a pressure lower than the header will still not improve that situation. You could only close it in an attempt to avoid backspin. Flow from the higher pressure header would enter the even lower pressure at the control valve's outlet and still attempt to backflow into the pump.

Additionally, when a centrifugal pump is deadheaded its discharge pressure is deadheaded at shutoff pressure, which is the same pressure, if it has a discharge control valve or not. Deadhead pressure with a control valve is not reduced and is in fact usually the maximum pressure that a centrifugal pump can produce. And, power is not being expended on the fluid, except for minor internal recirculation flows and all the rest of the power used by the pump to overcome internal fluid, bearing and stuffing box friction is being converted to heat. There is no possibility for improvement by deadheading a pump at a lower pressure or any other pressure other than its shutoff pressure,... without a VSD.

On the other hand, a VSD driven pump, especially with recirculation, could be used to lower the shutoff head corresponding to a lesser rpm, and reducing the heat load. Even a VSD driven pump without recirculation would produce less pressure and less heat. Total power consumption does drop off with lesser flow at reduced discharge head, and since head drops with the rpm^2, and flow drop is linear, a VSD w/ recirculation can improve the heat load.

True there is less power consumption at lower flows, with or without a control valve on a non VSD equipped pump, but most all power delivered is converted to heat at the lesser efficiency, so a lesser power consumption is paid for with increased resulting temperature load to the pump.

It also don't think that power consumption with a VSD drops, as you say, almost the same as if flow is restricted with a valve (on a pump without a VSD). Power consumption drops with the cube root of the pump speed on a VSD equipped pump, but power consumption on a pump without a VSD looks like that on any typical centrifugal pump curve with reducing flows, which is an inverted curve that depends mostly on the pump's hydraulic efficiency at flows away from BEP Q. Power consumption with a VSD dropping with the cube root of rpm and with only a very slight change in efficiency at different rpms and the same change in efficiency with flowrate as a non-VSD equipped pump, would appear as a cubic curve and power consumption drops very fast. I think those curves are very different, however the power consumption of a VSD equipped pump with a control valve, or a VSD equipped pump without a control valve would indeed be equivalent. The control valve has no effect on power consumption of the pump, as it mearly increases or decreases resisting head and consequently only changes the power required to flow into the system at any given control valve setting.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[KevinAtBurnsRoe]

Thank you everyone for your input so far. There have been several questions that I'd like to answer now to help clarify the application.

The characteristics of the 4 main pumps are as follows:
P1 - Turb - 388 GPM
P2 - VFD - 194 GPM
P3 - VFD - 194 GPM
P4 - Turb - 300 GPM
All the pumps are designed for 620 ft TDH with water at 227 DegF. Disregard the small pump (P5) in the auto pump control scheme.

The turbine drives are steam turbines with 150 psig inlet and 5 psig outlet.

The intent is to control the discharge header at a constant pressure, each boiler that is fed from the header has its own feedwater control valve. Each pump has an automatic recirc valve for low flow protection. There is a preference to operate the turbine driven pumps when possible as their exhaust steam is used for feedwater deaeration.

I hope this answers most questions. I welcome your further input.

 

[Valvecrazy]

As long as the water is cool, very low flow rates are possible without hurting the pump. I have found that at low flow, unbalanced hydraulic forces or radial deflection, is usually much less harmful to the pump than vibration or resonance caused by the VSD at certain critical speeds. As long as you have a good radial bearing and the shaft slenderness ratio has not been cut to the bare minimum, radial deflection causes very little problem with pumps up to about 200 HP.

The pressure of the header ( I WOULD NOT assume to equal the highest discharge pressure of any connected pump). Normally the BEP is quite a bit less than dead head pressure.

When delivering constant pressure or head, a VSD can only slow a pump down very little, because you loose head by the square of the speed. In these applications the power curve of a full speed pump being throttled is very close to the power curve of the same pump operating on a VSD. See this article.

http://www.cyclestopvalves.com/comparisons_13.html

However, since the temperature is 227 degrees, a throttling valve is not a good fit for this application. A VSD with a transducer on the PID loop may be the best way to hold a constant pressure on the header.

 

[BigInch]

ValveCrazy,

No I didn't say (or mean) at the shutoff pressure. The header would be at the pressure of the pump(s) with the highest discharge pressure corresponding to the actual discharge head for the actual flow at operating conditions. That is not going to be at shutoff flow, unless the header had no flow too and backed the pumps back to their shutoff discharge head. Of course also provided that the header was not overly sized permitting a large drop in pressues from one pump outlet to another to occur within the header.

In order to run 4 pumps into the same header, you must have very similar curves that produce the same approximate discharge head for any given flowrate you require to go through any individual pump at the same time. If one pump produces a discharge pressure too low in relation to the other pumps, backflow into that pump's discharge from the header will begin, or if prevented by a check valve, the pump will start backing up towards shutoff pressure until it reaches header pressure at a lower flowrate.

Failing to have pump curves that yield approximately equal discharge heads/pressures at each pump's required flowrate at any given time, you must install controls that will result in the discharge head or pressure seen at the header's inlets from each pump to be roughly equal.

In any case, no matter how you control it, this is not easy to accomplish with pumps that have widely differing curve characteristics and an unstable system is often the result where one or several pumps wind up contantly fighting the others and no or widely oscillating flows and pressures are a possibility depending on the steepness of the pump curves and their intersections with the system curves. It will be more complex, if the inlet pressures and flowrates to each boiler are to be varied as well.

I would sincerely recommend that you do or have done an accurate operating simulaton of the pumps and boiler feedwater system to determine if stable operations in the reqired range of operating flowrates and pressures can actually be expected.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[ozmosis]

My experience of using VSD's on the cold water feed pumps has shown that the actual speed/flow range is rather limited due to the safety tolerances required of the steam boilers. You must take this into consideration and in conjunction with the person responsible for the boiler control.
There is a pretty good software package provided by Danfoss called MUSEC (

http://www.danfoss.com/BusinessAreas/DrivesSolutions/Softwaredownload/DDPC+Software+Program.htm

) that, as BigInch points out, will provide some sort of 'simulation' to give you an idea of what could be achieved. It will not give you the final answer so should be taken in this way (i.e. don't rely on it!)

 

[BigInch]

Hi sed2developer,

What is sed2, BTW?

Lots of stuff on that page. I was thinking of doing a simulation incorporating accurate pump and system hydraulics coupled with the control response. Will any of the listed programs help? If so, which one(s)?

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[ozmosis]

BigInch
the program I was referring to is 2nd from the bottom "Multiple Unit Staging Efficiency Calculator". It's obviously designed with their VFD's in mind but offers a lot of experience of hydraulics and drives. Not 100% guaranteed but points you in the right direction.
Just to prove that I don't work for Danfoss, sed2 is the Siemens VFD that I was responsible for the Product Management when it was in development. I'm afraid we don't have a program like the MUSEC that Danfoss have!

 

[BigInch]

Thanks for the answer. My brother-in-law was involved with a project of a similar name, but I suppose that was "SEED", so no connection.

OK, looks like it could be interesting, but unless it has hydraulic modeling capability, it won't hold my attention.

Thanks for the pointer to it.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com