Parallel Pumps Versus Serial Pumps

Almost everything there sounds about right, with some allowance for specific piping details. I don't see how the 30 psi pump discharge increases to 35 psi, though, unless there is a drop in pipe elevation.

1 foot of elevation = .433 PSIG loss.

I agree with Tornado. How do you get that 5 PSIG increase?

That is what we refer to in the trades as a BF (brain fart). Sorry. It was a bit stinky.

right, alright, start with 30 psig.

I had one of the maintenance technicians arguing that pumps in series do not increase pressure. Pimps in series do increase pressure, so why not pumps?

Scenario 1: rather less than three times the flow. What is being ignored is the piping system pressure/flow curve. The operating point is where the system curve and the three pump curve intersect.

Scenario 2: rather less than three times the pressure for the same reasons as scenario 1.

The description in the original posting would be more closely followed with positive displacement pumps than with centrifugals.

The unlabelled scenario is unfathomable, as no information is available about the piping system characteristic curve, as suggested in the response to the other two scenarios.

Has this plant had a HazOp?

If in doubt, consult a qualified Piping Designer.

We were instructed by the inquiring James to remain blissfully ignorant of those pesky details about length of pipe, and 90° turns etc. in the unlabeled scenario.

So I shall remain, as always, ignorant and blissful.

I am not sure what you meant by HazOp. We have all manner of safety training and safety engineers. Would be nice if plant retrofits were actually reviewed by an actual engineer of mechanical systems.

I agree with what you stated about scenario I and II, of course the pumps are not unitary sources (just like electrical sources have internal resistance), and the flow circuit is not a unitary circuit (so there are impedance elements not named or pointed out, such as how many turns the pipe makes, how many fittings are present, etc.

It would seem obvious to me that if the line pressure is 30 psig in the basement, then while under considerable flow or even no flow, the pressure must decrease as one approaches ground level (30 feet up say arbitrarily), and another 5 feet to get on the oil cooler skid.

HazOp is short for Hazard and Operability Study, which is routine in well-managed industry. There is a wealth of information about the topic on the internet.

The forum has no information about flow rates. Pressure drop and flow rate are interdependent owing to fluid friction, which is a function of a range of variables that are not known.

If in doubt, consult a qualified Piping Designer.

True, there is no data on flow rate, because there is no flipping flow meter (or pressure gauge) on the pipe at grade level (referring to the 4" pipe).

I do know this: if the system is supposed to have 160-175 gpm flow rate in this cooler, it better darn well have it, otherwise it is scale city, poor heat removal, etc.

A 4" pipe might deliver that flow as an open pipe when the source is 150 feet away, 35 feet below the application, but there are about 6-8 90 degree turns, and also a multi-port "Christmas Tree" valve to select which cooler A or B or both. Operators make matters worse by trying to get more cooling by selecting both coolers, which really just plugs both coolers with scale sooner. I am pretty sure a 4" pipe that long with that many turns, and a honking big complicated valve with a 35 ft rise needs more than an initial 30 psig to begin with.

Solutions may be found in any textbook on piping design.

I have already done my analysis of the situation, utilizing finite losses, pipe resistance to flow, and the like, and also the thing with elevation change which does reduce dynamic pressure quite a lot in this case. Thank you for confirming I went about this the correct way.

You do agree at least that the pressure cannot be as high above ground on this system as below?

The forum knows nothing of the fluids in use. As this is bigger than a Mechanical Engineering issue, consult a local Utilities Engineer.

With three pumps in parallel, flow rate will be much less than 300 gpm but significantly more than 100 gpm. The pressure will be more than 20 psig. This holds for essentially replacing the single pump with three. If you change the whole system to accommodate additional flow, then pressure and flow will depend heavily on the changes.

.

With three pumps in series, once again attached to otherwise the same system, flow will be significantly greater than 100 gpm since pressure fed into the system will have almost tripled. Pressure will be less than 60 psig because flow rate will have increase and for pressure to have tripled, each pump would still need to be increasing pressure by 20 psig, i.e. the pumps would need to be pumping against a system that limited flow to 100 gpm.

.

I'm uncertain what you are asking in the third. Is there negative gravity in this hypothetical?

Did I forget to remind you that on Scenarios I and II, you are allowed to change the system (piping, etc) as needed?

Then these scenarios would be correct for a downstream installation that consisted of no fittings other than the discharge connection of each pump, in the first instance, and of the third pump in the second.

For real-world solutions, consult a qualified Piping Designer.

I think the real answer is simply to remove this one cooler from present system, and maybe its associated generator oil cooler, and put them on their own dedicated air-cooled closed loops, best answer, virtually nil corrosion, and a good closed loop maintains this by not losing any water, or virtually no water. No scale forms, since the water keep just going around and around with no losses by evaporation.

Nonsense. Hire someone who has experience of these things. It will be economical in the long run.

Why is putting oil coolers that have very high service (and surface) temperatures on soft water or even just a closed loop with hard water (at the start) a bad idea or nonsense. I already know the water and present treatment with the supplied flow velocity is woefully inadequate for the chemistry to work properly ( I have been experienced with this system for time since 1999 until the present, and have watched this item closely during every outage. I complained about this to management even during the re-power commissioning of this unit (new GT, old ST). They didn't listen then either. The Production Superintendent at that time and the next one were degreed mechanical engineers.

So now, do you still say "nonsense"? If you do, I call B.S.

As this thread has wandered far from <...Parallel Pumps Versus Serial Pumps...> into areas more suited to consultancy over the personal messaging system, at which point a schedule of rates can be negotiated, it is now time to unsubscribe from the thread.

<unsubscribes>

Change system as needed? As needed for what?

If you can alter the system as needed, you can get whatever results you need/want. Meaningless.

I mistook this for a real world scenario question.

The first two scenarios are just thought experiments, and I believed I made that clear from the start.

The third scenario (about the oil coolers) is real world, and is a problem that is worthy of solving. I have tried, and management either blows it off ("we have gotten by so far"), but with up 67% blockage of small tubing with scale material that resembles concrete in make up and physical properties (or worse), I would say we are playing with old dynamite.

no there is no negative gravity: circulating pump is actually located 30 ft. below grade in the basement. The output pressure (flowing to all points, including the 4" pipe and the 20" pipe. is 30 psig. When I climb back up out of the basement and go over to the oil cooler skid for the scavenge oil, that inlet is average of an additional 5 ft up (sort of, since there is a redundant set of generator oil coolers higher than 5 ft that source from this supposed to be 85 gpm (but it is not), and the lower scavenge oil coolers about 3 ft above grade at 150-160 gpm approximately by design, but in practice are very doubtful to be that high a flow, probably more like 70 gpm or even less. This lower oil cooler has very small diameter cooling tubes (1/4" stainless steel, with SS tube sheet, about 400 tubes within the A and the B side combined.)) Sorry about all the (). I have contended for many years now with management that this part of system should be on a separate water supply, closed loop cooling with 3 degree hard(soft) water (German degrees).

We have had reasonably good heat transfer, even with some deposits that are less than 1 mm thickness on tubes in condenser, other steam turbine oil coolers, and hydrogen coolers. The deposits were there and could not be removed by my predecessor (they are a material whose composition resembles concrete) by normal acid cleanings. Almost no calcium carbonate present. Concrete remover/tool cleaner will remove the deposits but management will not cooperate with me about attempting to clean very old tubes, as they will not risk much.

The newer stainless steel oil cooler for scavenge oil off the GT bearings is running hotter than any other cooler in the plant, has insufficient flow velocity, and is sourced directly off the cooling tower circulating water pumps. I have told "them" over and over we must do something before we trash the bearings in the LM6000, but no one is listening, or caring apparently. Too busy with "managing".

I have learned it really does not matter how loudly one repeats this scenario to management. They are stone deaf.

At any rate, these latter tubes have been as bad as 67% complete blockage during one outage about 2 years ago. We had to get a Good-way drill and DRILL them open. This was after the concrete remover could only open about half the blocked tubes and get the remaining material gone back down to bare metal. So, OMG, WTF, why won't they F---G listen?

Listening is not the issue at that facility. The issue is one of ignorance.

Enlightenment might arrive with the design calculations for the installation in question.

Since I am not allowed to just go out and hire an engineer, and they will not believe a chemist, what course of action do you recommend?

We have no cliffs to jump off of. I can do the calculations based on GE's original specs for the water supply to the cooler, and what I expect to find with present supply, and have already done it, and submitted. One issue is the revolving door of Production Super's in recent years.

Have you pointed out the amount of down time required to do an overhaul and the cost of tech time and travel and parts and loss of revenue that downtime will cost?

Maybe you could do the bearings yourself, maybe. I'd want a factory tech to do it if it was mine.

I guess that won't matter to the bean counters until it takes a dump.

I have pointed that out, that wiping the bearings in an LM6000 gas turbine could be as expensive as a $40,000,000 mistake.

This thread is growing like Topsy. This has got stuff-all to do with pumps in series and parallel.

There may be a cast iron bath in the flow loop now for all the forum can tell.

I suspect you have already done this, but it doesn't hurt to repeat:

make sure you CYA.... recollections of who said what can change rapidly when the SHTF. Documenting communication in writing through the proper channels of your concerns and the potential downside is prudent. Heck, maybe someone will even recognize it as a CYA letter and be spooked into initiating meaningful review.

How about years worth of well documented inspection report with photographs?

How about 17 years trail of emails?

I have also told every single one of my direct supervisors and anyone else I get to sit still for 5 minutes.

^{Sounds like you have it covered. Just wanted to make sure.}

I am totally sad, that I could not explain the layout just with words. My bad.

There seems to be a process fluids and piping incompatibility issue at the facility, which would be best addressed by consulting a qualified Process Engineer.

How in hell can the pipes and the process fluid be incompatible? Water and mild steel.

This is not seawater, but brackish or near brackish. It is sourced from well fields that have only about 244 mg/L total hardness as CaCO3, moderate alkalinity, and 20-50 mg/L silica as SiO2. Yes, it will make a scale that you cannot easily remove.

We get away with this on condensers, etc., because the temperature extremes are not as high as they are with scavenge oil from gas turbine bearings.

The oil coolers for the gas turbines would do well to be on a softened water closed loop.

Sounds like the heat exchanger is a local high point in the system. Is there a connection for venting at the high point?

What NPSH is available to the pump 30ft down when running? Is the water degassed?

Seems like flow restrictions in some tube in your heat exchanger placed at a local high point could lead entrainment gas to partially fill some of the more restricted tubes....especially with the suction inlet some 35 feet lower.

The pump suction inlet operates in gravity flood of 24" header pipe from cooling tower basin (that is discharging at approximately ground level minus 15 ft.). This is open-cycle cooling system, so the water is air-saturated certainly. The water is somewhat warm at the return point, about 75-88 °F, so there is little likelihood of it warming up enough anywhere that matters to gas lock some tubes, except these oil coolers that are the only coolers in our system that exist physically above ground, and are subjected to the highest differential temperatures.

The pumps are not starved but they do date back to 1957, when the Westinghouse Steam Turbines were first installed.

There is a provision for venting the oil coolers at the oil skid nearby the LM6000 gas turbine.

So far, we have not scorched the oil, or wiped the bearings on GT, but I would not rule this out in a possible scenario where the system creeps up on 100% blockage of the cooler tubes.

Okay, so the pumps have roughly 15 feet of water for NPSH. The heat exchanger is 35 feet above the pumps, so in static conditions filled with water should have 20 feet of water pulling pressure below atmospheric.

This would suggest that, if the return path is not extremely convoluted and no other major restrictions are in the return path, that there is a good chance absolute pressure is below atmospheric through some portion of the heat exchanges (assuming the heat exchanger represents a significant portion of the head loss for that loop).

If the water is saturated with air, reducing the pressure will bring gas out of solution. The tubes don't have to experience a complete loss of flow for a significant reduction in performance. Even if gas coming out of solution is flushed through completely, the areas where the gas contacts the tube will see reduced heat transfer.

the pressure at the filters is still 20ft above atmospheric, given that the pressure measured a the tap is gauge pressure, and is given at 30 psi x 2.31 = 69.3 ft. Not to imply that under high velocity conditions there is not an opportunity to cavitate in the HX tubes, but the relatively low temperature and fairly high absolute pressure (69.3-35) + 14.7 psi = 29.5 psia, would require velocity that should make all kinds of noise easily detected.

(NPSH requirements for pumps provided by the manufacturers are usually given in psi/ft absolute)

Doesn't work like that. The pressure will not be 29.whatever at the top unless there is a closed shutoff valve stopping all flow after that point (with pumps still energized).

In a real system with flow, you can't just add pressure as if it were a static column. To prove this point, just add up the pressure going from the bottom of the return instead.

Pressure at the return at water level in this open system is atmospheric, right? How many feet above that is the heatexchanger? Right, so that is the feet of water that would be drawing pressure lower than atmospheric.

So the head loss in a real flowing system makes up the difference between the 29 psig you calculated from the discharge side and the feet of vacuum from the suction side.

If the bulk of the head loss occurs at or even before the hx, then the pressure can be less than atmospheric there...and not just in special geometries.

The friction loss at the estimated flow rate is much smaller than the static head components, ignored in this case when citing below atmospheric pressures.

The discharge line returns to the pump suction, so is essentially same static head up & down, except for density of water, warmer on the drop leg. When using a pressure gauge, static head is dominant in this system.

It is highly unlikely in this case that the friction pressure drop in this case could exceed 14.7 psi, at 100 to 200 gpm in a 4" pipe, and if the pressure drop occurred at the heat exchanger entrance, possible, you still would have pressure above atmospheric where your pressure gauge is. Beyond the pressure gauge, who knows. Except that there would be considerable noise, which, in a practical situation, would be commented on, we might assume. Also, with so little flow, we might think that the fluid would not pick enough heat to be noticeable on the exit piping, also not remarked upon.

Friction pressure drop should be "astronomical" across the HX when it could be up to 67% blockage as observed during last year's outage. I wonder if conditions did not allow a vacuum to be produced on the downstream side of the HX? I will have to look into that when we are online, and see if.

You don't contest that the pipe discharge to atmospheric conditions will be at atmospheric pressure, right?

I hope you can further agree that since flowrate being less than the speed of sound in water, pressure will be transmitted, even upstream.

If those things are not disputed, it doesn't matter if you think the total amount of head loss is a large value or not. The head loss will be the difference between the pressure entering and exiting the system. The exit is at atmospheric. The head loss is distributed across the system.

Low flow does not make pressures below atmospheric less likely, just the opposite.

In static conditions, i.e. the pump is turned off, the HX in question (scavenge oil cooler) is just below the point some 350' away where the cooling tower basin is showing a normal water level. That tells me it is flooded with about 1.5 ft H2O head positive under stasis.

You earlier said the return was 15 or 20 feet below ground.

Oh, come on! Are you telling me this scenario is also under the rule of 'system subject to change to make whatever answer you decide correct'?

The return to the cooler tower for the main header (after the condenser) is about 35 ft in the air. The return from the coolers back to the pump suction is still about 35 ft below grade, just like the take off from the pump discharge.

From an earlier post of yours...

'...The pump suction inlet operates in gravity flood of 24" header pipe from cooling tower basin (that is discharging at approximately ground level minus 15 ft.). ...'

Ground level minus 15 feet, discharging as an open system. The HX is supposed to be 5ft above ground. Oh wait, don't tell me, I know: we can adjust ground level as necessary to get the pressure you want?

Please refer to comment #87, wherein I include a crude sketch of this system. I don't have proper drafting tools, so please bear with me the excel spreadsheet objects.

Water. ok. Previously it was either oil or a gas turbine.

Please describe the process in terms of fluids and flowrates. Please state the pH and the piping materials of construction. Please describe the water treatment regime in the water circuit and the blowdown criterion.

The raw (city potable water supply) is evaporated to the point in an open cycle cooling system (cooling tower) until the cycles of concentration is in the 3-4 range. This is really pretty low compared to what some system achieve by way of water conservation.

Incoming:

Alkalinity: 180 mg/L as CaCO3

Total Hardness: 220-270 mg/L as CaCO3; Ca/Mg ratio is about 1.2-1.4

Chloride: 200-300 mg/L as Cl

Sulfate: 100-300 mg/L as SO4

Silica: 20-50 mg/L as SiO2

Cooling tower water: ~3-4 cycles of concentration on all species, conserved by measurement. Sulfuric acid is added to control pH to 8.2-8.35, with good results with respect to calcium carbonate. Chlorine/bromine is used as oxidizing biocide to control microbe growth/attach, or algal blooms.

MgSiO3 - Not reaching saturation value

CaCO3 - all saturation indices indicate slightly corrosive, not scaling

Corrosion inhibitor - very effective agent for steel, aluminum, yellow metals that is yielding 1-2 mpy general corrosion rate, and pitting less than 0.5 mpy. (My formulation with anionic scale inhibitors based on consults with a leader in the industry)

No formula exists for continuous treatment that will inhibit the concrete-like scale formed by this water in contact with metals. No calcium carbonate scale is ever found, on the other hand.

Main coolant flow is 10,000 gpm by one circulating pump, although there is a double redundant pumping system, so that 20,000 gpm is available at full steam load on steam turbine (20 MW output). We usually operate at less than 50% full load output on the steam turbine, but near full load on the gas turbine.

In scenic 1, you will gain the sum of volumetric of the pumps.

In scenario 2 pumps in series you will gain the sum of the all pump heads

you may be right...My fault, 4" line at 10,000 GPM

anyways since this is hypothetical... lets just talk potential... and ignore the pipe capabilities.

Even without rupture...

For scenarios in this universe....

Pumps in parallel replacing one pump hooked to the same system, with just the necessary piping to make the connection of the parallel addition(s)....

Flow rate will increase but will be less than the amount proportional to the change in number of pumps. Also, pressure will increase. So, 1 to 3 above. For conditions to move from 1 above to 2, a duplicate of the entire system could be attached to each pump, and if you wanted them in parallel, just add the connection....not that cross connections will have meaningful flow.

.

For pumps added in series pressure will increase, but less than what would be proportional with the increase in number of pumps. Flow rate increases because additional head is provided.

The increase in flow is realized in every pump, meaning each pump is experiencing less resistance and so develops less head on an individual basis. So, 1 to 3 not 1 to 2 above.

Ordinarily that would be relevant, but the OP explicitly stated that the piping systems could be assumed adjustable.

That is just showing the starting point from which to make adjustments. In my adjustments I opt to change things around and then bring it back to exactly as it was....ya know, so there is something meaningful for comparison.

I know you can change things around to get whatever you want, but taking that option kind of throws the OP under the bus by pointing out that formulating a question where you can shift the variables around to arrive at any answer leaves a lot to be desired.

<...formulating a question where...shift the variables around to arrive at any answer leaves a lot to be desired...>

Quite. The answer is a lemon:

I hope you will share that piping arrangement. I know theoretically it should be possible...but have no clue about the specific path.

ROFLMAO! This is getting more humorous as we go along.

Idea being that for pumps in series (assuming they could take the abuse of that situation) - to hold Q constant requires essentially tripling the output head.

Idea being that for pumps in parallel (assumption in this case is that constant head pressure at output nozzle is required, thus triple the flow, approximately. That requires each pump to be operating at the same NPSH each as one single pump would (each sucking out of a pond for example).

Lif can be interesting. Dealing with what you find unexpected with humor is not a bad option.

How extraordinary.

Quite.

It blows my mind that you guys cannot read a f* scenario and provide a straight answer. The d* boundary conditions were handed to you on a silver platter.

"4" line branching off the main header (20" diameter)"

<...ignore the pipe capabilities...>

The original post does just that.

NO NO NO. The damn 4" line is at 30 psig (at the pump discharge), and it is a mere side arm off of a 20" pipe that is under 30 psig, and flowing 10,000 gpm. Come on man, use the noodle a little bit.

All the 4" pipe is supposed to deliver is this: 160-175 gpm to scavenge oil cooler, and about 80-100 gpm to the generator oil cooler (situated above the scavenge oil cooler).

The problem is the change of elevation reduces feed pressure to the oil coolers (real scenario).

No that isn't it... .

in all seriousness, Normally when sizing pumps, I ignore the line size and if available, I use the systems curve.

But I admit, I didn't fully read your OP.

If you don't know the line sizes, how do you determine the system curve?

The systems curve is already determined.

The next is sizing the pump. When I have the systems curve why would I waste my time on piping... again.

Oh, so somebody else already did it.

No, I actually would do, but for sizing, if you already entered the sizing in to calculate the systems curve, then you have the required head and flow, why reiterate line sizes.

Nor do you really need them for a pump curve for selecting a pump. even though some pumps curve will list the inlet/outlet sizing for that pump.

That's not what you said in post 53. So which is it: do you ignore line sizes, or do you look them?

I think you can't see the forest through the trees.

I have the systems curve, why would I have to deal with lines size, that been done.

Here I'll help you point it out that its a typo that so difficult for you.

"in all seriousness, Normally when sizing pumps, I ignore the line size and if when available, I use the systems curve."

Better?,

And thank you for being so dam ANAL^ytical.

I think the guys that did this to me (and everyone else here) pretty much ignored the system curve and the specs, and the line size and said something to the effect: "Hmm...let me see, here is point A, and over there is point B, I guess 4" pipe will carry that much flow, what the heck, tell Joe to get the welder out and go run that string of 4" pipe from A to B".

In that case it does make one wonder why #14↑ has been labelled off-topic.

I agree that textbooks do cover this topic completely and thoroughly. You gents have basically proved that I was not entirely hallucinatory over this issue in my plant.

I certainly am not the culprit who marked your post off-topic. I did, however, place one GA on top of that.

We have different levels of insanity here where I work. Some of it involves doing things the same way, with higher expectations for results. Some of it involves coming up with elaborate ways to continually or repetitively do corrective maintenance on what should be engineered out of the picture.

My problem is not separating this into two completely different posts, that, and not giving a complete snapshot of the flow system, my bad. I thoroughly apologize for the fox chase on this one. It seems some of you at least got the scent of the beast.

My problem is not separating this into two completely different posts

You really can't do that, because that you'd get the 'need more info'.

The problem that I had, and considering from the responses, no one really read the whole post through, or understood it.

Unfortunately, I'm not taking anything too serious today. it -4F with a -20F wind chill, I just want the week to end.

At least by the elevation difference, plus friction & velocity head losses. If you have enough head to initially fill the discharge pipe, then you regain static head lost.

It is likely the system designer took this into account.

If your HX discharge pipe is fully flooded, then you regain part of your lost static head at the vented level of the pump suction supply. For instance, if the drop leg stays full for a 20ft drop, then your pressure available to create flow though the heat exchanger = 69.3ft at branch -30ft up minus -20ft drop = 59.3 Delta P at HX, or 25.7 psi. You need to add friction losses and velocity head loss, once you have the interacting flow & velocity dependent head loss values in equilibrium. You need to correct your pressure gauge readings by the vertical offset from the pipe, and according to the temperature of the water, these feet elevation values (2.31ft = 1 psi) are for 60F water density...

Thanks. That water is almost never 60 F, 80-90 F is more like it, and quite a bit hotter on the downstream side of the HX.

Due to the high service temperature of the HX, it needs to be on better water than present source - the open circulating cooling system (cooling tower water).

The head difference is not changed much at those temperatures. For the most part, the pressures calculated by elevation difference will be off 1/2 percent at the 90F.

So the static head gain for the drop leg will be a little less than the elevation, if the water is less dense. The lower viscosity of the water at higher temperature will help offset the difference.

Since the discharge returns directly to the main pump suction (missed that earlier), then the static head will be essentially and practically zero, leaving only friction & velocity head losses to measure. Your pressure at the HX inlet could be used to estimate flow, if the piping wasn't so large for your flow estimate. A 4" line at 200 gpm will drop about 1 psi per 100ft of length. Add up about another 10ft. of pipe length for every fitting, valve, etc., and add 20ft for each end of the 4" pipe.

If you have assumed say 200 gpm of flow, 30psi on the gauge at the header branch(69.3 ft) less (30ft x 0.995) for elevation, less 6 fittings, 2 velocity heads and 50ft of length=150ft equiv x 1psi/100 = (39.45ft or 17.08 psi) - 1.5 psi = 15.6 psi on your (30ft physical offset) gauge, that is about as close as you can get to prove 200 gpm of flow.

If the pressure measured is higher, then you have less flow, assuming you have accounted for fittings and height correctly, and have some good gauges (usually calibrated to 60F water). So to a point, low pressure at the heat exchanger can indicate good flow.

Thanks, very good assessment. It is starting to point at the real problem: the water itself is just inappropriate to the application (too high an exchange temperature (surface temperature) for water with that much hardness and silica. The thing has more or less kept the bearings on the LM6000 from burning up since being installed in 1999, but I have seen some real issues with tube blockages in recent times.

Could be the selection valves were scaled to the point of being dysfunctional? Operator error?? I know maintenance put new 6 way valve on this last outage that ended November 1.

I would love to see this HX be moved off circulating water, and put on its own system (fin-fan air-cooled loop with bump tank).

I told you all from the start that the hypotheticals would allow you to incorporate whatever piping system made you happy with the pressure and flow rate stated.

Wouldn't it be better to carry out a real-world design, then?

Yes. The thought experiments were from an argument with a maintenance operator who claimed (I) the same flow, and (II) the same pressure.

That's nothing. His supervisor once told me that if I sent water down a 3/8" O.D. stainless steel tube to a sample point, and had it looping back up to discharge in the sump the pump draws from (intake cooler for gas turbine), the water would not flow back up 35 feet, because of pressure loss. He really has no concept of static head, velocity head, dynamic head, or anything other than pin head.

Scenario 1. Providing each pump is the same, that is, the RPM is the same the number of stages in each pump are the same, the manufacturer's flow rate is the same for each pump, then ignoring losses for the moment and the inflow to the pumps is as required, then the output pressure and flow rate will be equal for each pump BUT not cumulative at the discharge flange.

So if you have 2 pumps with an output of 100gpm per pump, as per the pump chart at 3600 RPM (50Hz) at which the pump will be operating at its BEP of 70% to 80% efficiency, then you can expect a lot less than 200gpm, but with an increase in discharge pressure at the discharge flange.

So one pump in this pump curve gives you at point A, 470gpm, and at point B with two pumps running you get 660pgm.

Scenario 2. Place each of the same three pump is series, the effect is to increase lift or pressure while keeping the flow rate the same for ALL section connected in series.

In my job of designing submersible pumps for the oil industry, we look at the depth of the well, the fluid level of the well, static and dynamic and the THD losses.

So to design a pump to lift well fluid from 6000 feet at a rate of 400bpd or 10gpm, I wold look at the pump curve to give me the lift for a single stage

I then divide the depth by the lift/head of a single stage at the Best Efficiency Point, in the curve above that's 400 bpd, giving me a head(ft) of 16ft.
Divide 6000 by 16 = 375. As the dynamic depth = 4000ft divide the dynamic depth by lift to result in 250stages required for the pump. then the losses would be calculated.

The effect of joining the pumps in series only increases lift or pressure as the flow rate will stay the same.

As for your practical example where the PHE ONLY takes 150gpm as your pump is rated at 10,000gpm, you will either blow up the tubing, the PHE or wear out the pump as the back pressure will be so high the pump will be in constant downthrust and destroy itself. Also what is the suction rate of the circulating pump, as the PHE can only discharge at a rate equal to or less than the inflow rate!

Look at the first graph in your post. If you increase the pressure on the system curve, does the flow ever stay the same?

When you place three pumps in series, the flow does not stay the same compared to one pump on the same system. Each pump will only be experiencing a third of the total resistance/ producing a third of the total head....meaning operation is at a much lower head and higher flow rate on the same pump curve. Flow rate increases when adding pumps in series, all other things being equal, because pressure developed increases.

I. Yes

II. Yes

III. Yes, at inlet to Heat Exchangers, realistically slightly less than 15.15 psi gauge, assuming 60F water density (ignore erroneous pressure gauge Bourdan tube offset due to water hammer)

Now would be a nice time to uncork that rare bottle of wine, sit back and sip a glass.

I've never been a situation like yours, but the field techs I traveled to observe (as engineering manager, I could do that) deserve much, much praise for the miracles I have seen them perform in the field. (Once while lying in a pool of etcher solution in order to get a machine back up and running without disrupting production)

I can only sympathize with your plight, while in my present position all I do is give other people headaches like you have.

And on April 28th, 2017 I will join the ranks of old duffers putting around the golf courses, here, in Arkansas and Minnesota, depending on the season.

About spring of 2019, I will hand over all my notes, the keys to the lab, and whatever else needs to stay at this hell hole, and also pursue other items of interest less vexing.

Sorry to hear that you still have that long to suffer.

Cheers.

I am certainly feeling it today. Wifely sent me to Tractor Supply the other night to pick up a bale of compressed straw. I should have known there was a catch in this somewhere, when after I paid at the register, the young man walked up with this slung over his shoulder by the carrying strap, and swung it down and handed it off. It nearly knocked me on my butt! Today, the catch is in my "get-along".

That thing weighs about 80 pounds, or so it feels. Maybe I am getting weaker, because a box of 4 one gallon containers of titrating solution was nearly too much for me this afternoon.

I have vowed to give up drinking when I can no longer carry a case of Bud bottles out to my truck.

I let the kid at the pool store load the liquid chlorine I buy there. Then have the 17 YO unload it for me at home.

At least my back feels better today. The only place that didn't hurt yesterday was where the sun never shines. 16 L of dilute EDTA really should not have been that much of a problem for me. It is only 35.3 lb.

This is a really crude sketch of the flow path for the scavenge oil coolant.