Disimilar capacity pumps connected to common header

It depends upon the difference in pressure between the pumps' inlets and the pressure in the discharge header, which is approximately the same for all. Take the pumps' pressure/flow performance curves, compare with the above data, and go from there.

Do you mean both suction and delivery are on common headers? If it's only the suction and it's properly sized, no problem, provided each pump is correctly spec'd for the block it's feeding.

Cheers........Codey

suction header and discharge header are common to all the pumps.Only common factor is all are designed for 20 Meter water column head.

Any pump can operate for any block.On the day of my visit they were operating 2 nos 22 KW and one 37 KW , the water temperature was 10 Degree C and when it was changed to two numbers of 37 TR the water temperature dropped to 9 degree C.

I think we need a diagram here.

"Any pump can operate for any block." What is your meaning of a 'block'? If any pump can operate for any block, there must be a bunch of valves to route the water to the different blocks...

Enervac -- A group of pumps like that, randomly selected without consideration of good engineering practice is liable to perform in a manner that results in some operating efficiently and others performing inefficiently (with respect to energy efficiency and noting that centrifugal pumps operating far from their maximum efficiency region are also liable to mechanical or electrical overloads). Some careful analysis of the performance curves of individual pumps and the various possible system curves resulting from different combinations or the available pumps might explain troubling system operation events.

However, given the size of the motors in your example it may be reasonable to assume that some good engineering went into the total system design. The way to select pumps for such a system is to look to a family of hydraulically similar pumps from one manufacturer's offering where the relationship of head to capacity is the same in percentage terms; but the capacity at maximum efficiency point increases by some reasonable multiple. (this will necessarily result in increasing specific speed as the size of the pump goes up.) That appears to be the case in the 7 pump example you cited.

The reason for this is that the system is likely designed to operate at a pressure that is consistent with most economical construction. However in the case of systems such as HVAC cooling tower systems the weather conditions as well as other variations in the demand from the facilities served may demand widely varying capacities. Note here that this particular combination of pump sizes will give system operating engineers considerable flexibility in selecting the right combination of pumps operating in their best efficiency regimes to service the demand.

The whole idea here is to get pumps to pump against a constant selected discharge head that matches their best affiance pressure well. Generally such a family of pumps will have similar impeller diameters and a general increase in vane width as the capacities go up. The trick in engineering this combination will be making the right pump selections given the important consideration that the general shape of the head capacity curve changes as specific speed goes up rather than staying the same (in percentage terms as I mentioned before) which is the ideal.

Remember, the key to good performance here is matching of the head capacity curves. System problems can arise if pump wear and the resultant performance loss, unusual system changes, unauthorized system modifications or poor operator performance/ failure of automated control systems occur. One thing to watch out for is operation of the smaller high specific speed pumps in a capacity regime well below the maximum efficiency point such that wear ring deterioration takes place. This will accelerate as the failing pump operates closer to shutoff head eventually resulting in "churning" and showing noticeable heating in the casing due to energy from the motor not leaving with the now greatly reduced water flow.

The danger in such a system if poorly managed could occur when one of the smallest pumps operates under continuous duty under a primary automated control and operator inattention to secondary system tuning is absent. As the pump wears it continues too try to meet system requirements but since it cannot it operates in a regime of ever increasing radial load that accelerates wearing ring deterioration. At this point it should be taken off line for repair. However, the system operating engineer may not be aware of the decreasing flow rate either because flow meters are not installed or working properly or training is lacking.

The system design engineer, rather than opting for the least expensive version of the lowest capacity pump (perhaps the 11KW pump in this example) will do well to specify a version with a "heavy duty" shaft/bearing frame or larger frame motor if close coupled that will reduce shaft deflection at near shutoff operating conditions and perhaps eliminate the kind of pump failure I've described above.

Note hear we are dealing with a well engineered system example. A poorly designed system cobbled together with the cheapest available pumps may well have one or more pumps that do not have the head capability to match the other pumps and will simply sit there and run at reduced or no capacity ("churn").

Another second order consideration is that it is important to remember that the total delivered head of a pump (what the rating curves are based on) is the sum of suction head and discharge head. Undersized suction piping may well push unmatched pumps with higher NPSH requirements into a cavitation regime where they lose capacity and can rapidly become damaged when the entire system is operating near maximum capacity.

Bottom line here is that pumps operating in a parallel system require some serious attention to matching.

Hope this all helps, Ed Weldon

Theoratically, there should not be any problem since all pumps are operating and designed for 20 metres head.

In practice, how ever, you will need to check for performance of each pump. Does flow from all pumps in operation add up to the sum of flow each pump is supposed to deliver?

My obsevation in one case where only two pumps were operated in parallel was that the flow was about only 20% more than the individual pump capacity. We had to replace the pumps with one single pump for better efficiency.

rsalaskar -- Your example of only 20% improvement with a second pump was probably much related to the system head being mostly friction head and rather than static head such as a tall building with cooling towers on the roof and generous sized piping with minimal friction losses.

Your example was probably much related to the system head being mostly friction head and rather than static head such as a tall building with cooling towers on the roof and generous sized piping with minimal friction losses.

Remember for water flow in pipes the friction head loss is proportional to the square of the flow rate (capacity). Double the flow rate and the head loss increases 4X. Since a well selected centrifugal pump, i.e. one whose operating point is in the range of maximum efficiency has only around 20-30% increase in pressure at shutoff. Put two of these operating in parallel against straight friction loss in a piping system and there will be only head increase somewhat less than shutoff head in each pump available to push water through the system. Look at the head-capacity curve for the pumps and you will likely see that the capacity for each at that operating condition is only somewhat above half the capacity of the single pump. This explains the unsatisfactory performing example in your experience. To the extent that the system head includes a component of static head (which loses have no first order effect from capacity changes) the improvement in performance from parallel pumps will be greater.

Ed Weldon

Spot on, Ed. The key is static head vs. friction head. If the system is such that it is 100% static head, and each pump is sized for this as TDH, then each pump will produce its' duty flow (according to the capacity/head curve) no matter what combination of pumps is running. It only gets tricky when friction head is introduced, as more/larger pumps operating will increase flowrate in the system thus increasing friction head and TDH, and each pump will operate differently according to the nature of its' curve, including the possibility of 'churning' if system head exceeds max pump head on a particular pump.

Ed, you are right, it was case of frictional head. The pumps were used for cooling tower with water delivered to many small HVAC units at different distances and with a common cooling tower.

If all the pumps were specced correctly for the same head, only their delivery volumes will change to the respective blocks. If then also you have enough supply for all the pumps, they can and will run together and do their respective duties without cavitation. You may also find that the higher you go you will probably run the bigger pumps where the smaller pumps will only be used when the duty required is less. This is also done to save enrgy costs. (Why run 22kW when 11 will do)

For better efficiency, and to save energy costs, would it not be better to use a single pump with inverter drive (VFD)? This way, you could also keep the duty point closer to the best efficiency point.

Not always Holzfeller, Remember that a motor typicall dont like to run below 50 % of its rating( as in frequency inverters), and thus, if the duty calls for a 37kw 1450 rpm pump you can,t get the 50% required at 700 rpm from a centrifugal pump. Maybe if you install a 3000 rpm pump and throttle that to 1500 you will get the 5o% (maybe), but, when will you recover the capital costs?

These are centrifugal pumps, so flowrate is not proportional to speed (unless you ignore pressure). Don't forget affinity laws; if the speed is doubled, the flowrate doubles but the pressure is squared, and vice versa. Since we have a fixed differential pressure (static head) and the pump will always be operating at the 20metre point on the curve, a small change in speed will lead to a larger change in flowrate, so the overall speed range would not be that great, easily within the 25Hz-60Hz range.

I do agree, but i'm trying to show that additional control and expenditure may not be financially viable.

Yes, I see your point. In fact it is impossible for us to evaluate without knowing a lot more about the installation and its' purpose. Still, whether the control is manual or automatic I would have thought it would be much simpler and cheaper to have one pump rather than seven. Inverters are much cheaper, more versatile and more user-friendly than they used to be. Although, there could be many very good reasons for using seven pumps (Ebay bargain, plenty of standbys in case of failure, etc.)

Replying directly:

From your different posts, it seems that the pumps are all in parallel, suction to discharge.

The kW differences seem to suggest that there is a need for different flow volumes at different times or at different load scenarios. The 20m head being constant cannot be possible in a closed circuit without trottling the pumps outputs individually.

The Logics seems that who ever designed the system, must have some kind of control that will start different pumps On/OFF to maintain the same pressure differential accross the chiller's heat exchange. Therefore, if more load is required, more output flow is needed, and more circuits are opened to the flow... then more pumping capacity is required...etc.

This system is complicated and you need a correct piping circuit diagram with all the control system operation description to adivise or at least for you to sort out and try to redesign. Otherwise, if all is working fine, leave it as is...

{Normally, chiller cooling circuits are designed with a constant flow accross the evaporator, needing the same pumps to run all the time. The Loading and Offloading of fan coils is done with 3 way bypass valves so as to keep the flow constant and avoid a lot of complications ... otherwise, you need to program a PLC or something like that to keep all the components and the users happy...}

Your comment regarding cooling circuits USED TO be correct, with constant flow through the chillers and a primary-secondary loop with variable flow through the end-user distribution system. Recent developments (and better chiller controls) have led to current "high-end" CHW designs using variable flow through the chillers AND the end-user distribution system.

The world of engineering keeps changing- hopefully (usually) for the better.

Thanks for the info on development...

If you have any site(s) on the topic I would like to visit them and gratefull.

On the Question put by the OP: I thought that he needed to be made aware that what he saw might be OK since he was asking ..

Thanks

energygod -- Of course variable frequency drives and modern electronic controls are state of the art. But note that the OP refers to an existing system and it is reasonable to assume it is already installed. Also note that the OP's location is India and possibly an area where there may not yet be a well developed technological base to support such equipment.

Beyond that there may be cost factors and even cultural biases that work against the installation of such sophisticated equipment from foreign sources. When I worked in the pump industry 40 years ago variable speed motors were just starting to make their appearance in pump applications and as one might expect these first applications were in the building industry for high rise office building HVAC and house water pumps. At the time there was a lot of heartache over early failures in some applications.

So an important consideration here may be that similar applications in India or other emerging nations whether or not the systems are designed and manufactured locally may be in a troublesome stage of evolution.

Another cost consideration has to do with the cost of labor. Technical decisions based on cost can be deeply swayed by labor costs. In this case the cost of trained operators standing watch over meters and manual controls and occasionally doing some simple arithmetic or reading of charts may be much less than the cost of sophisticated imported equipment as well as the cost of repairing it in the case of failure.

Ed Weldon

I value all of your input immensely, Ed. But I have to take issue with this. In terms of many technologies, especially pumps, you should not refer to India as an emerging nation. They have experience (much from the UK) that goes way back, and many ignore the fact that they have been doing things for decades that our oriental friends are only just starting. Many European pump companies (eg KSB) have been manufacturing pumps in India for decades, and any cultural differences tend more strongly towards the British than anything else (many of us are proud of this fact!). Many installations in the Industrial, Civil, Petrochemical engineering, etc industries are based on UK training and support. I don't know where most inverters are made these days (you might find as many, or more, made in India as in China), but it is wrong to assume that our friends from the Indian subcontinent are not fully aware of the potential benefits, or culturally or economically (or in any way) averse to their use. They have been using sophisticated, imported equipment much longer than my friends and neighbours in the US.

As far as I see, there are only two possibilities:

(a) The system has been designed as such for a very good technical reason, of which we are probably not fully aware without good drawings, sketches, specs, etc..

(b) The sytem is designed by an idiot or beaurocrat (Synonyms?).

(b)

Holzfeller -- I'm not trying to put down India. But in terms of technology they are still behind other world leaders in many areas. I admit I cannot fairly judge the level of pump technology in India as it is today. Likely I have digested too many anecdotes and formed unfair predjudices. Like it or not India is often described in the western media as an emerging economy. But all of that is immaterial here.

What we should be doing is trying to understand the situation on the ground given an OP who for whatever reasons has not told us a lot about the details behind his question. When I look at a problem I try to look objectively at all the underlying factors. Just because some non technical factor is not complimentary or politically correct does not mean it should be ignored.

I know little about the state of the art in liquid system control technology, only what I would ask of some expert about what they could or could not do. What I do know is that variable frequency AC motors are still fairly expensive and thus would demand a fairly high return on the investment in terms of energy savings.

Seems to me that there is a practical maximum speed dictated by NPSH requirements of the production pump design and a practical minimum speed below which efficiency dictates that a lower capacity pump running at maximum efficiency provides more efficient power usage.

Now a problem of determining best cost profile (assuming a highly variable HVAC system) of a small number of VFD pumps versus a larger number of single speed pumps may be easy if it is just one VFD pump versus two or three single speed pumps. But when the number of each type gets larger the solutions become the realm of a serious computer analysis involving many non linear variables. With 7 pumps in three different capacities the number of practical available net capacites is probably in the range of 20 or 25, maybe more and still there is room for redundancy to support increased system relaibility. Would this level of technological expertise be available to the local engineering firm or the local in country pump manufacturer? Or would it be necessary to go to the international manufacturer and pay for his high gross margin requirements even if the manufacturing costs are similar by virtue of both manufacturing plants being in the same labor market?

Visualize yourself as the engineering firm in India that has a contract to do the mechanical engineering of systems for a new office complex. Whose system do you go with and write into the specs? Lots of costs involved. The local pump manufacturer may be as well versed in the hydraulic design of pumps as KSB. But does that level of knowledge extend to newer electical, electronic and control technologies and possibly minor tweaks to impeller designs and the ability to put them together into a system that will return maximum energy efficiency?

Ed Weldon

Ed, sorry for the lack of response, I've been away. And I hope I didn't sound harsh – not intended. I guess what I am trying to say is that we cannot necessarily put this system design down to cultural differences or backward technology, there has to be a good reason (probably not cost, simplicity or ease of use/maintenance) why they would choose seven pumps over one or two or three, but until we know the facts, via specs, drawings, system descriptions or whatever, we cannot know what those reasons might be and any ideas we proffer are stabs in the dark.

In many industries and applications where pumps are used (including water) the engineering contractors and system designers tend to be quite conservative, and things haven't really changed that much. I've seen this in Europe and the US. VFDs and control systems, on the other hand, have changed phenomenally in the last 20 years, but in many industries the way they are used has not. I don't think this technology in India would be unknown or out of reach. I don't know this for sure, but I would not expect this sort of system in India to be subject to the same sort of constraints that we might see in certain industries in certain Asian and African countries (where, for example, gland packing might be used in chemical process pumps simply due to lack of resources, which we would consider 20 years out-of-date and usually illegal).

You make a good point:- "Seems to me that there is a practical maximum speed dictated by NPSH requirements of the production pump design and a practical minimum speed below which efficiency dictates that a lower capacity pump running at maximum efficiency provides more efficient power usage." but we don't know if this applies until we see what the system looks like. We need to know what the system looks like, particularly upstream of the common header. If there are multiple circuits that can be switched in or out, then the total system head could be at a constant or set level at different overall flowrates, and it might make sense to have multiple pumps that are sized to this head. If you try to do this with a single pump by increasing the speed you can increase flow so far, but if the tdh remains the same you could very rapidly approach the dangerous 'end-of-curve' zone. If it is a single circuit with significant friction head, and assuming all the pumps run at fixed speed, I think it will be the pump that can generate the highest head (refer to curves) that will win, irrespective of kw rating, and the rest will mostly churn if they are effectively running at 'closed-valve'. Again, stabs in the dark?

Holzfeller -- Good to hear from you again. You and I are very much on the same page. I fully agree with your comments in your last paragraph.

When we analyse these pump questions on CR-4 there is almost always a lack of critical information to enable a good answer. So we have to make assumptions and address different possibilities. The world is full of engineering lashups of one type or another that represented the "best" solution for the situation at hand, especially where there are serious money constraints. We often look at them and utter the essentially retorical question "Why on Earth did they do it that way?".

For all we know at this point this particular 7 pump installation may have been the result of 7 stages of growth of a building complex, each step taken without any expectation of what might come next. Maybe the owner had a source of cheap used pumps and a low cost local labor source to operate and maintain them as I previously suggested. India is a land of many faces ranging from the sophistication and prosperity of Mumbai and high tech Bangalore to the dirt poor farm country of the interior. We even have situations approaching that here in the USA.

I do think that one of the problems we have with these inquiries for help from foreign sources is the difficulty that folks there have preparing a detailed description of their situation in what is for them a complex foreign language. I see lots of signs of that.

Generally mismatched pumps either in series or parallel are a bad idea to be avoided. So when someone asks what will happen without a full description of pump and system curves we have difficulty in coming up with useful answers. It's sort of like asking what will happen if you throw sand in the oil filler of a car engine. We simply don't know the exact results because it is not something that is usually done. So we are inclined to offer a trite answer like "nothing until you start the engine up".

Interesting to note that Sam Yedidiah in his wonderful Centrifugal Pump Users Guide has little to say about systems with multiple mismatched pumps.

Ed Weldon

Ed-

The only purpose for my last thread was to clear up a comment regarding the state of the art in the design of chilled water systems- not to tout the benefits of VFDs or better system controls.

So fas as I can surmise from the OPs input- the original designer of that system designed it with several pumps having the same head capacity to allow any combination of units to operate based on the system load demands.

Nothing in the OPs statements tells me whether the flow through the chillers is constant or variable- and it really does not matter.

Again- I was only advising all the players of what is happening NOW rather than what used to happen.

Do you know if there is a problem with the system that may be caused from the arraignment of different pumps, is ther evidence that one or more pumps/motors have been changed out more than others? I have installed allot of these chilled water systems and have seen pumps set up like this before. Many times you can look at how the system is wired and set up to be controlled that explains why these particular sized pumps were installed. for instance on full load the 3, 37 kw will pump, at night when less cooling is needed 2, 37kw and a 22 kw pump will run.

I have installed reverse return systems with many different pumps sized for each "zone" or area they cover. These systems usually change pipe sizes each time another air handler is tied in and could get larger or smaller according to the flow.

Also there should be some extra pumping capacity to cover the load in case a pump fails. We have one extra pump for each size in the chilled water system where I work. This system has 3 different sets of 3 pumps all tied into the same headers.

It's the hydraulic equivalent of wiring seven batteries of different sizes/capacities in parallel. One doesn't get the flowrate, or current, of all the batteries the same.