Pump and Control Valve, Chicken or the Egg?

[Patassa]

I've worked myself into a circular reasoning of death here. Help.

I'm starting in a phase 2 of sorts of someone else's design and project. The gist is this, the former principle mechanical engineer ran a really rough 1st draft of hydraulic calculations to spec a new pump and a new flow control valve for estimate purposes. He assumed 20 psi dp at max flow rate (water, 1600 gpm) across the control valve. He then used this pressure drop in addition to the line losses and static heads to size the pumps.

Fine

Now insert me into this project..

To do the next round of estimates, I have to "tighten" up the calculations to get more accurate quotes on equipment.
I recalculate the line losses using the 3D models of piping with accurate dimensions, so far so good, the head loss is slightly different as expected. But now I turn to the control valves and I'm thinking, man, I really need to get an accurate dp across these so that I can more accurately size my pumps.

Here is the chicken and egg problem. As far as I can tell, you size a control valve using a specified dp across it (the difference between your pump outlet pressure and your final destination pressure). But, I don't have this in reality because I'm basing my pump head off of bogus control valve dps in the first place. (I'm referencing the Emerson/Fischer literature on the standard method of specing a control valve. In other words, I calculated my pump dp based on a control valve dp (that came out of someone's imagination), so I can't very well use my pump dp to calculate my control valve dp without circular reasoning.

How do I get out of this conundrum?
*Known information is max and min flow requirements as well as piping diameters.

 

[Latexman]

I don't see the conundrum. Send the preferred vendors a spec to quote on, and start using the performance characteristics of a real control valve. Then, the Stage 1 assumptions can be forgotten. No more circular reasoning (whatever that is). Then, touch up the pump requirements (spec) with 3D model quality piping losses and real CV pressure drops. And, you are ready to get a quote on the pump.

Good luck,
Latexman

Technically, the glass is always full - 1/2 air and 1/2 water.

 

[Patassa]

Which specs would that be? Again, the Fischer literature calculations require a dp across the control valve, this is based on pump head. I have neither in reality.

 

[Patassa]

Circular reasoning = using one assumption to base another assumption then checking the first assumption using the second assumption. That's what it is.

 

[Latexman]

"Send the preferred vendors a spec to quote on, and start using the performance characteristics of a real control valve". That's which spec it is. What's wrong with using the 20 psid? That's a real pressure drop. Does it result in a reasonable control valve that gives good control? If so, you are in business. What is the total frictional pressure drop in the run including the CV?

As long as the assumptions are reasonable, and you buy a pump and CV that reasonable meet those assumptions and subsequent calculations, you are good to go!

Good luck,
Latexman

Technically, the glass is always full - 1/2 air and 1/2 water.

 

[zdas04]

Isn't this just an iteration? Pick a pump and pick a real control valve. See how they play together in the model. If it doesn't work pick the next size control valve (either bigger or smaller depending on the problem) and re-run the model. Repeat until the system curve (including the control valve) meets the pump curve at your desired flow rate. I've solved this kind of problem with slide rules and got results that worked. I'm sure with a 3D model you can do at least as well as we did back in the day.

David Simpson, PE
MuleShoe Engineering

Law is the common force organized to act as an obstacle of injustice Frédéric Bastiat

 

[LittleInch]

What's intriguing me here is what this control valve is doing. What is it's control input? Flow, pressure, temp?

Normally you provide a valve vendor with min, max and normal conditions. If your aim is to minimise dp at max flow to reduce your wasted energy, you can provide min and normal flow and dp and then give max flow and say min do, vendor to advise. Only you know what your system is trying to do so I'm double guessing here, but you don't needed to design the valve yourself, give vendors the key info and they will tell you what the dp is at your max flow whilst still working within the 20 to 70% open which is a normal range for most control valves

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[JJPellin]

I don't see the dilemma. The control valve selection is not as constrained to the system as the pumps is. I would go to the control valve experts and tell them the fluid, range of flow and line size. They would recommend a valve in order to get the best value and reliability. I would not even need to tell them the DP. They would come back with an optimized valve, and then I would select a pump to run within the model of the system using that valve.

Am I missing something here? If the pumps selection required me to have a valve with a 100 psi DP, I would not select that valve. I would rerate the pump or change the pump selection to reduce the valve DP. I know that dropping 100 psi across the valve represents a waste of energy and is likely to result in poor valve reliability. If my valve expert told me that the most economic and reliable valve for my flow range and fluid would have a DP of 10 psi, I would design for that.


Johnny Pellin

 

[Patassa]

Sorry, I can see how to some of you the solution here is obvious, you clearly have experience doing this before, but I don't. I wasn't aware that this was supposed to be an iterative process when trying to solve it yourself, but I certainly see how that makes sense and how the dilemma goes away.

So it sounds like I have two practical options.

1. Send Emerson/Fischer my min, norm, and max flow rates, line size, and liquid information. They shoot back a valve with a Cv. I take that Cv and calculate my DP across the valve at those flows then if it's reasonable, I size my pumps with those dps in the line? Is this correct?

or

2. I go through a list of valve options, grabbing a few Cvs more or less at random and iterate until a find a valve that controls the flow the way I need, gives a reasonable pressure drop and reasonable pump size? Is this correct?

Thank you

 

[Patassa]

Oh and when I said "3D model of the piping", I just mean the output from the piping cad software, not fancy fluid flow software, the 3D model just gave accurate dimensions. I'm doing all of the hydraulic calcs on an excel spreadsheet.

 

[Latexman]

Both 1 and 2 will work. Depends on how you want to do it.

Good luck,
Latexman

Technically, the glass is always full - 1/2 air and 1/2 water.

 

[LittleInch]

The issue with 1 is that you also need to define the differential pressure at at last one of your min/ max or normal flow points as the valve vendor has no idea about your system curve. E.g. a system curve with a high static head element could vary from 0 flow to max flow between say 75% to 100% of head from your pump.

A different system curve with low static head the 0 to max flow could be from 10% to 100%. The valve vendor can only see what you tell him.

As JJPellin says, if you're dropping lots of pressure/head at normal flow, you're just wasting pump power and you really need to re-design your pump.

Work out what your system needs to do it's job (flow and pressure)and then find a pump and valve to match with the highest efficiency.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[Patassa]

I think this is where I was lost yesterday

"The issue with 1 is that you also need to define the differential pressure at at last one of your min/ max or normal flow points as the valve vendor has no idea about your system curve."


I have pump curves that came from assuming a dp of 20 psig across my control valve at max flow rate. And I think you're saying, send that 20 psig to the valve manufacturer to size a real valve for me? And they will come back with a range of Cv's for my min- regular flow values, and therefore different dp's at those flow rates, so I would then need to check those against my pump curve and make sure the pump can still give me the head I need at those other dp values and do it somewhat efficiently?

If it can't, do I change the valve or change the pump or both? Is this the iteration that has to take place?


If that is all correct, I think I understand. If not, sorry, I'm not getting it.

 

[zdas04]

The control valve sizing does not need a dP. dP is a function of upstream pressure, downstream pressure, and flow rate (it is an output, not an input, that seems to be the source of your confusion). If you have any two of these, the third falls out with valve selection. You really don't want to over constrain the problem. Your outlet pressure is your target separator pressure. You have a min, max, and design flow rate. Your valve guy will be able to recommend a valve that will do the job, and the output of that analysis will be an inlet pressure. With the inlet pressure and a flow rate, sizing the pump is trivial.

David Simpson, PE
MuleShoe Engineering

Law is the common force organized to act as an obstacle of injustice Frédéric Bastiat

 

[LittleInch]

You are looking at this from the egg point of view. The chicken in this instance is the requirement or demand.

So, the order in which this activity is done to achieve maximum efficiency and generally good design is:

Establish your system curve. What this means, and apologies if it's teaching you something you already know, is to establish the pressure and flow of your system ( we have no idea what it is, whether a one way flow like a pipeline or process or a re-circulating one like an A/C system) at a number of flow points from zero to maximum flow. THEN you work out how best to service this requirement with your pump. The duty point of your pump is then normally the max flow and pressure. To vary the flow or applied pressure, with a centrifugal pump you have two main options - either use a control valve to create a variable differential pressure across it controlled by some parameter (pressure, flow, temperature, manual set point) or install a variable speed drive which controls the speed of the pump to match the pump output to the demand.

In your case someone has decided that they are going down the control valve route. The issue with this is that if you always have a differential pressure, even at maximum flow, basically you are throwing money away as this is like driving along with your foot constantly on the brake. However in some cases it is needed, in others not. Only you or your process engineer can decide this.

So to return to the question, You need to give the control valve man at least two and hopefully more cases which match your requirement. To di this you do need to assume a certain pump curve. However control valves are, by their very nature, variable devices and hence so long as the dp in reality is within 10-15% of what you asked for, you should be ok. To size your pump, the lowest power / head you need is to be able to meet the demand at max flow / max pressure. If you a control valve in the circuit, you really want to minimise the pressure losses across it at this point so go for a valve with minimum dp at max flow and let the vendor tell you what that is, based on your requirement at min / midpoint flow.

I think you're more or less "getting it"


My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way