Positive displacement pumps in series

[PUMPITUP7]

Hello all,

There is not a lot of information on this subject so i've decided to try this route to see if any one might have some experience with this. I'm in the frozen entree manufacturing business and the specific application is to pump sauces through heat exchangers to the point of meal assembly in a continuous fashion that matches production speed. We already have a system for this and those of you that are familiar with the sanitary pump industry will know that the most common types of pumps are the circumferential piston (known as lobe pumps), progressive cavity, AODD (air operated double diaphragm) and Sine pumps.

In normal circumstance two PD pumps in series is rarely done because inevitably it is impossible to balance the two and they end up fighting each other and causing cavitation or pressure spikes. Even if the second pump was placed exactly in the physical middle of the system, the fact that we are cooling the sauce means viscosity will be changing and dynamic head due to friction will not be the same in each half of the system.

The reason we wish to have two pumps in series is that while a single pump has enough flow and pressure capacity to handle the conditions of the system, the pressure is detrimental to the product, reducing quality and identity of ingredients. With two pumps in series, the second placed approximately midway in the system the hope is that each pump will see approximately half of the TDH (total dynamic head) as a single single pump would. For example if the discharge pressure of a single pump in this system is 100psi then theoretically with two pumps in series each pump would have approximately 50psi.

In order to avoid the issue of balancing flow I am proposing to place an open bypass around the second pump. This is where I run it to problems trying to predict what will happen. I believe the bypass would allow any excess flow from the first pump to bypass the second in downstream direction. If the fist pump falls short of supplying the needs of the second, the flow from the discharge of the second would recirculate back to its suction upstream over the same bypass. In this set up, both PD pumps would be same model, size and horsepower. Speed for each would be controlled from one common level transducer (4-20mA loop) so that both pumps run the same speed. Each pump would have its own VFD so we could do some slight scaling if we wanted speeds to be slightly different (a one time compensation adjustment). As mentioned above, even if everything was theoretically perfectly balanced there will always be slight differences due to cooling and viscosity changes. We run a different kind of sauce every day so we would not be able to track and compensate these differences.

So back to the bypass around the second pump, my question is would the first pump be exposed to the full system TDH or only half? Would the pressure in each system self balance or would it randomly wander (System TDH would never change (ie100psi) but pump 1 may see 70psi while pump 2 sees only 30 and minutes later 60/40 then 30/70 the other way.

Anyway, I don't want to lead into it too much more. What do you all think? Anyone have a similar setup that is actually functioning?

Oh I should also mention that I am aware that having the 1st pump feed a buffer tank and then the second pump draw off that tank would solve this but that would require twice the instrumentation and also complicates the CIP cleaning of the system. Now I have a tank to clean and essentially two separate piping systems which would require two separate CIP pumps. This is what I want to avoid.

Kind regards,

 

[LittleInch]

An open bypass will not work as the second pump would do very little. The open bypass would simply equalize pressure across the second pump and not allow the second pump to actually generate a differential head unless your bypass was really small, then you will get a lot of shear in your product.

If you had a single fluid pumping at a single rate then this would be difficult. For a liquid system a very small difference in fluid volume being pumped( i.e. flow rate) makes a massive difference in pressure.

The only way I see this being done is to have a master slave relationship between the first pump and the second.

The first pump you set for flow control on your VFD (master).

The second pump (slave) VFD controller looks at its suction pressure and controls on that to keep it at a certain set point ( say 10 psi or whatever your pump inlet is into no 1 pump). If you're in the hydraulic middle, but with slightly different viscosities, then the difference in pump power shouldn't be too different if you have the same or similar inlet pressures as you will have the same mass flow.

If you have a fairly gentle start / stop system then this is your best chance but even then starting and stopping this system won't be easy. I guess a bypass around the second pump with an NRV might work. Start the first pump at low flow ( not more than your 50 psi) then start the second pump, let the system stabilize then slowly increase speed on the master pump until you get to your desired flow rate.

The tricky bit on any PID controller is getting the speed of reaction right. Too fast and it never stabilizes, too slow and doesn't react fast enough to changes in the master pump.



Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[PUMPITUP7]

@littleinch

Thank you for your reply. I understand the master slave scenario. I agree the PID control could be very difficult. Start up would be difficult. I don't have dedicated operators so looking for something dummy friendly. The system is cleaned every night and starts empty the next day.

I'm curious about your comment on the bypass. If the second pump was to generate no pressure differential then only two things could be happening (to my knowledge). Either all of the flow the pump generates would move back to its suction or the pump is fully cavitating continuously.

In the first condition, how would the the flow get back to the suction? Wouldn't the flow from the first pump would be acting as check valve in the bypass line? The first pump is still providing positive flow so the second can't just recirculate that over and over. It has to go downstream. In the second condition, i can't see how the second pump would completely cavitate when there is a fixed supply feeding it. The flow from first pump would either choose to go downstream to the second half of the system over the bypass or it would choose to go to the suction side of the second pump. Path of least resistance would seem to be governing the "choice" and the suction of the second pump would be providing the the lower pressure option rather than the head represented by the second half of the system over the bypass.

Perhaps another way to describe this, In order for the first pump to see the TDH of the entire system (call it 100psi)the second would have to have zero flow (like it was off). Once it is running i would expect that it would pick up at least some portion of the total system TDH. It can't bypass back to suction if there is no place for it to go because that is already occupied by flow from the first pump. (I'm assuming perfect balance here which we know it won't be. I'm actually expecting some flow to bypass in either direction but only the unbalanced amount).

Lets say the system is empty and at start up and we follow things along. Pressure shown on the discharge gauge near the first pump would be zero (call it gauge 1). As it fills the first half of the system the pressure would rise to about 50psi. The pressure as indicated by a gauge at the end of the first half of the system would be zero (call it gauge 2). The system discharges to atmospheric pressure. As the sauce moves into the second half of the system it has two directions it can go, over the bypass or into the suction of the second pump. At this point since the pressure for either direction would be zero so it will go both ways. As the sauce enters the second pump and is discharged it can go two ways, back to its suction through the bypass or further downstream into the second half of the system. Since the pressure for either path at this instant is zero it could go either way but seconds later the pressure at the bypass will rise because that space is already occupied, pressure will be slightly higher and thus sauce will go downstream instead. As it moves further downstream the pressure
at gauge 2 will rise from zero an amount proportional to how far the sauce has traveled through the resistance of the piping of the second half of the system. For the purpose of this example lets say it has only traveled 10ft downstream and gauge 2 reads 5psi and gauge 1 reads 55psi. Gauge 3 would be near the discharge of second pump and would also rise to represent the head against it which should be the same as gauge 2. So sauce leaving the second pump has two paths to go and both of which offer a 5psi resistance. It can't go back over the bypass because that space is occupied so it goes downstream. Now as the sauce moves to the end of the second half of the system the pressure at gauge 2 & 3 will rise to 50psi (representing TDH of the second half of the system). Oh-oh, i think i just answered my own question here. If gauge 2 reads 50psi that means gauge 1 is going to read 100psi and thus I am not successful in lowering the pressure the product is exposed to.

Running through a couple bypass NRV scenarios in my head, i don't think that is going to help me either. I think i need a surge tank in the middle. Maybe i can figure out a way to make it all one system for CIP.

Please feel free to poke holes in my assumptions. I've just managed to prove myself wrong so would love to hear someone tell me i'm wrong about being wrong - LOL.

 

[PUMPITUP7]

Hi LittleInch,

Ya that's pretty much exactly what you said in the first place without so many words 50psi into second pump suction and 50psi out means zero differential and zero flow. All TDH would be transferred to the first pump.

 

[georgeverghese]

How about a single low shear Moyno sanitary food grade pump (progressive cavity type)?

 

[MikeHalloran]

I'm thinking two pumps and two heat exchangers, each running at half the current flow.
No fancy valves or controllers, just less pressure on the product.
Of course, that doubles your CIP requirements.



Mike Halloran
Pembroke Pines, FL, USA

 

[PUMPITUP7]

@George

Currently we are using a single pump. It's a Sine pump, next gen. Certa 400 series. Sine pumps are also supposed to be ultra low shear. The pump can provide up to 200psi and 3 times our highest required flow. It was oversized so that the nominal flow would require a low pump RPM (about 100RPM). Tests we've done thus far indicate the degradation of product is caused by pressure not by the action of the pump itself. A single mono PC would still be exposed to the same system TDH / pressure as our sine pump (100psi) and likely won't be any different.

 

[PUMPITUP7]

@ Mike,

Yes that would work but unfortunately it would have other ramifications for our heat exchangers. The heat exchangers need to be in series to achieve the desired cooling AND desired delta T of the coolant. Our coolant is city water and we reclaim the heat and water into our domestic water and boiler feed systems. I suppose the product side could be parallel and the coolant side in series.

Interesting supposition but still hangs on the doubled CIP requirements.

I did find a way to have a series system using a buffer tank at the suction of second pump (which would have to be COP cleaned). The tank would be removed and replaced with a CIP jumper allowing full 5ft/s flow velocity around the second PD pump using only one CIP pump. Only extra instrumentation is a second level transducer. Not as bad as I thought it would be.

 

[LittleInch]

Pumpitup7

I think you got it. Another way to do this is to draw it out and put different flows. If you have an open bypass, then lets say pump one is doing 50 GPM ( randon numbers).

If pump 2 is only doing 30 GPM, then 30 is going through the pump and 20 through the bypass, but the second pump is basically idling, adding no pressure difference.

If pump 2 is doing 70 then 70 is going through the pump, but 20 is now going backwards through the bypass to maintain a total of 70 and again pump 2 is doing nothing.

An analogy I came up with is this. Think of one man pulling a vehicle on a fixed incline using a long rope. He's a big strong bloke (pump 1) and can do 5 mph. His mate (pump 2) turns up and wants to help. To start with though he just walks along side holding the rope and not putting any effort in ( your open bypass or my forward only bypass). Then though he starts to pull a bit more. However if he pulls too hard the rope in front goes slack and he takes all the load which he can't do so then pulls a little less hard until both of them are pulling together. It can be done ( think tug of war teams), but isn't easy.

Same thing here. Your second pump can control on inlet pressure, but might oscillate a bit. If your control system was fast enough though then it might just work. It would certainly make it a bit easier if there was a small accumulator to take out the dips and peaks in inlet pressure into pump 2, but can understand why you don't want a second thing to clean out.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[moltenmetal]

It is possible to use a PD pump as a pressure source rather than as a flow source, but it's challenging to make it work...it involves more controls than just an open bypass.

 

[FACS]

Is the pressure problem time related?
Would it make sense to install an enclosed reservoir at the head of the second pump instead of a bypass, thus relieving the material from pressure for a bit, and then letting the second pump take it from the reservoir?

 

[Compositepro]

Given the constraints you have imposed, your only options are to:
1)lower your flow rate to reduce pressure drop
2)increase your pipe diameter to reduce your pressure drop
3)transport your product in wheeled vats

Wheeled vats or bins are very commonly used in the food industry. It is the most gently way to transport things and they are far easier to clean than pipes.

 

[electricpete]

I think LittleInch articulated well some fundamental problems with op approach.

If you (op) can figure out which pump tends to be stronger, then maybe try putting the bypass around that pump with a throttle valve in bypass line. I think you might be able to tweak the valve to get some of your sought-after advantages (such eliminating very high or low pressure in the intermediate piping between the pumps, and possibly reducing pulsations). The cost would be minor reduction in system efficiency (losses in the recirc/throttle path). Just a thought.

=====================================
(2B)+(2B)' ?

 

[1gibson]

Can you recirc the whole tank through a short heat exchanger loop until you reach the desired temperature, then switch to pump through straight pipe to get to the filling portion? I guess that only works if you have the time to cool down first, and you use the entire batch at once. But it reduces the max pressure needed, and keeps a single pump.

At that point, you might as well just drop cooling coils directly into the tank.

 

[EdStainless]

If you want simple and reliable then run a larger diameter line.
Compared to the cost of additional pump, controls, maintenance, and mistakes a new line will be inexpensive.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[georgeverghese]

Is the damage done to the food product due to pressure in the fluid or from the shearing of the product as it drops nearly 100psi through some nozzles at the food assembly area ? If it is from the latter, then using 2 low shear pd pumps in series may not help, and you would possibly be looking to get some bigger nozzles that give you the same capacity at a reduced pressure drop ?

 

[NickJ67]

A little late to the party but.....

I would (also) question what actually causes the product degradation. My own experiences suggest that it isn't pressure as such that causes the degradation, but the factors that cause the pressure. That is, mostly, velocity and shear in the line, especially at bends, heat exchangers, valves etc.

A further factor is shear / slip in a pump working against a high head, which becomes proportionately larger at low speeds.
Though a Sine pump is gentle in principle it is prone to quite large amounts of slip when working against pressure at low speed.
You might do better on that front with a generously sized, two-stage progressive cavity pump which also has very low shear characteristics but also very positive sealing.

As any pump will beat the product up to a degree, adding an additional one seems best avoided and as others have already said, I'd suggest taking a long hard look at was to reduce the losses in the system, as those losses are what are damaging the product. Bigger pipes, straighter run, fewer fittings, larger/different type of heat exchanger. Impossible to be specific without knowing your system.

Nick