We have a small reactor with a filter element in the bottom. The reactor is working at atmospheric pressure. The reaction mixture is a suspension. During reaction the liquid part of the suspension is pumped by a gear pump from the filter (reactor) outlet back to the top of the reactor. The flow at which the liquid is passing the filter is dependent of the pressure difference over the filter. Now I want to speed up the recirculation flow as high as possible without cavitating the pump. To increase the pump flow, the flow through the filter has to be increased, which can be achieved by increasing the pump flow. Viscious circle? If the pump flow is much higher than the flow through the filter, the pump will cavitate. At the one hand the pumpflow is needed to create a depression under the filter and to create a pressure difference resulting in a higher flow through the filterelement. However if the pumpflow is too high, the pump will cavitate. How do I select the best pump to increase the recirculation flow to a maximum without cavitating the pump? Which specification is the most important? The pump suction capacity? Other? And how do I best control such a system? Is a gear pump the best type of pump for such an application? Who can help me?
Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Pump selection problem 5
- Thread starter gilden
- Start date
-
1
- #2
i think you will need to wait for others to help you, but instinctively i'd say your most important specification would be NPSH(required). the second your desired throughput.
i think the only problem you will encounter is the speed with which the process will adapt to changes here. you need to increasine pump throughput just to the point where cavitation starts, then let the system adjust, then once again increase, and so on. it is not per se a vicious circle, as you put it.
but, as said in the beginning, this is just intuition, not knowledge, so wait for others to comment.
hth,
chris
i think the only problem you will encounter is the speed with which the process will adapt to changes here. you need to increasine pump throughput just to the point where cavitation starts, then let the system adjust, then once again increase, and so on. it is not per se a vicious circle, as you put it.
but, as said in the beginning, this is just intuition, not knowledge, so wait for others to comment.
hth,
chris
-
1
- #3
For a moment forget the pump. Pumps do not create a "depression" under the filter.
The flow to the pump is determined by the static head available in the reactor less any resistances, such as the filter's, on the way to the pump.
Since you are speaking of an atmospheric vessel, estimate the friction drop on the way to the pump for the new higher expected flow rates, deduct it from the available static head, and you'll obtain the available head at the working temperatures. If this head is lower than the vapour pressure of the liquid, you'll have cavitation in any pump.
The question now resides on what pump could handle a bit of cavitation without harm.
And all this is assuming no dissolved or entrained air, or other non-condensables, are apt to be released in the pump.![[pipe]](/data/assets/smilies/pipe.gif "[pipe] [pipe]")
The flow to the pump is determined by the static head available in the reactor less any resistances, such as the filter's, on the way to the pump.
Since you are speaking of an atmospheric vessel, estimate the friction drop on the way to the pump for the new higher expected flow rates, deduct it from the available static head, and you'll obtain the available head at the working temperatures. If this head is lower than the vapour pressure of the liquid, you'll have cavitation in any pump.
The question now resides on what pump could handle a bit of cavitation without harm.
And all this is assuming no dissolved or entrained air, or other non-condensables, are apt to be released in the pump.
-
1
- #4
Two persons who can help you better than anybody else are pump manufacturer and filter manufacturer, in my view.
First get the data of flow rate vs pressure drop across the filter from the filter manufacturer. The minimum flow rate will be at maximum pressure drop(when filter gets loaded). Then check NPSHa as suggested by 25362.
Secondly, get a pump which safe minmum flow is below the minimum flow rate across the filter. Otherwise, pump may get heated up and will be damaged as you are recirculating the fluid again and again.
In my opinion you can try for a screw pump.
Regards,
First get the data of flow rate vs pressure drop across the filter from the filter manufacturer. The minimum flow rate will be at maximum pressure drop(when filter gets loaded). Then check NPSHa as suggested by 25362.
Secondly, get a pump which safe minmum flow is below the minimum flow rate across the filter. Otherwise, pump may get heated up and will be damaged as you are recirculating the fluid again and again.
In my opinion you can try for a screw pump.
Regards,
-
1
- #5
[/u]bchoate[/u]
Permit me to reiterate the situation:
Small, atmospheric reactor containing a suspension. Filter element in bottom of reactor. Gear pump circulation pump.
The need is to increase circulation flow.
The NPSHR (required) for the pump in service is fixed. The problem is the NPSHA (available). NPSHA is related to the static height of the liquid over the pump suction, the pressure over the liquid in the vapor space, head losses across the filter, and head losses in the piping. When recirculation rate is increased the head losses across the filter and the piping are increased reducing NPSHA below NPSHR with the expected result.
To increase circulation flow several factors must change, alone or in concert.
1) the filter area needs to increase to reduce head losses. This probably isn't possible, since, as I understand it, the filter is an integral part of the reactor.
2) larger piping is needed to reduce head losses. This is particularly important in the suction line since you must maintain a flooded suction for the gear pump to work right. Larger discharge piping may also help.
3) liquid return to reactor - is it subsurface or does it discharge into the vapor space. Subsurface re-entry increases the pump's required discharge pressure to produce the desired flow.
4) Do you have the best pump for this application? The gear pump is probably not the best option. Consider this alternative: circulate liquid and suspended solids (if that does not adversely affect the desired reaction). A laminar flow pump, such as the disc flow pump, can pump slurries up to 80% solids. The NPSHR is 1/3 to 1/2 of an equivalent centrifual pump. If circulation of the total stream is possible, then the filter in the reactor could be eliminated and placed somewhere else if product filtration is needed.
El Cajon, Ca. 619-596-3181
I believe the disc flo pump can potentially improve this reactor operation.
Bill C.
Permit me to reiterate the situation:
Small, atmospheric reactor containing a suspension. Filter element in bottom of reactor. Gear pump circulation pump.
The need is to increase circulation flow.
The NPSHR (required) for the pump in service is fixed. The problem is the NPSHA (available). NPSHA is related to the static height of the liquid over the pump suction, the pressure over the liquid in the vapor space, head losses across the filter, and head losses in the piping. When recirculation rate is increased the head losses across the filter and the piping are increased reducing NPSHA below NPSHR with the expected result.
To increase circulation flow several factors must change, alone or in concert.
1) the filter area needs to increase to reduce head losses. This probably isn't possible, since, as I understand it, the filter is an integral part of the reactor.
2) larger piping is needed to reduce head losses. This is particularly important in the suction line since you must maintain a flooded suction for the gear pump to work right. Larger discharge piping may also help.
3) liquid return to reactor - is it subsurface or does it discharge into the vapor space. Subsurface re-entry increases the pump's required discharge pressure to produce the desired flow.
4) Do you have the best pump for this application? The gear pump is probably not the best option. Consider this alternative: circulate liquid and suspended solids (if that does not adversely affect the desired reaction). A laminar flow pump, such as the disc flow pump, can pump slurries up to 80% solids. The NPSHR is 1/3 to 1/2 of an equivalent centrifual pump. If circulation of the total stream is possible, then the filter in the reactor could be eliminated and placed somewhere else if product filtration is needed.
El Cajon, Ca. 619-596-3181
I believe the disc flo pump can potentially improve this reactor operation.
Bill C.
-
1
- #6
I think you have some possibilities, you may:
a) Increase the filtration surface of the filter. More surface - > less dP - > more NSPHa
b) Use a pump that has a low NSPHr for example: a liquid vacuum pump.
c) Relocate the filter after the pump. I do not know if the process can tolerate this.
in each case you may need to control the discharge of the pump to avoid getting cavitation.
RGS
a) Increase the filtration surface of the filter. More surface - > less dP - > more NSPHa
b) Use a pump that has a low NSPHr for example: a liquid vacuum pump.
c) Relocate the filter after the pump. I do not know if the process can tolerate this.
in each case you may need to control the discharge of the pump to avoid getting cavitation.
RGS
Eppur si muove. You wouldn't be able to circulate the suspension at any rate higher than the gravitational flow obtained by opening the valve at the bottom of the reactor. Having measured or estimated that flow rate with a constant head (liquid level), you then can proceed to select the pump. ![[smile]](/data/assets/smilies/smile.gif "[smile] [smile]")
- Thread starter
- #8
I would like to reply on the comment of 25362. Without recirculating back to the reactor, I have drained the reactor, once by opening the valve at the bottom of the reactor, and once by draining it through the pump. The draining by the pump was 3 to 4 times faster. If 25362's comment is right, how can this be explained?
- Thread starter
- #10
Reply to 25362's question: everything was equal in both filtrations. If you want, I can give you the filtration times:
without pump:
100 ml in 32"
200 ml in 75"
300 ml in 134"
with pump:
100 ml in 10"
200 ml in 21"
300 ml in 33"
without pump:
100 ml in 32"
200 ml in 75"
300 ml in 134"
with pump:
100 ml in 10"
200 ml in 21"
300 ml in 33"
To gilden,
Have you repeated the tests or are these the results of single tests ?
Trying to find a solution to the enigma, here are some thoughts that may help to understand what happened:
Suspensions of particles flown through a pump have always the potential for size reduction because of collision between the particles themselves, and between the particles and the metal walls of the equipment, mainly by the pumping effect. Thus, if one of the tests was carried out after a pump run, the particles may have been reduced in size creating a different friction drop through the filter than they would before a pump run.
To see whether the suspensions in both tests were about equal, transfer the samples to a graduated cylinder with a stopper and measure the hindered settling rates of the slurries.
The settling rate, L/h is proportional to the square of the average equivalent particle diameter assuming all other factors, such as densities of liquid and solid, as well as viscosity of the liquid are equal on both slurries, and that Stokes' law applies.
For example, if the hindered settling rate doubles, it means that the average particle diameter has increased by about 40%.
Duplicate tests can be carried out by reslurrying the cylinder contents by inverting it several times. It is recommended to use same size cylinders for the sake of repeatability and reproducibility, so wall effects will cancel out.
Would you care to comment ?
Have you repeated the tests or are these the results of single tests ?
Trying to find a solution to the enigma, here are some thoughts that may help to understand what happened:
Suspensions of particles flown through a pump have always the potential for size reduction because of collision between the particles themselves, and between the particles and the metal walls of the equipment, mainly by the pumping effect. Thus, if one of the tests was carried out after a pump run, the particles may have been reduced in size creating a different friction drop through the filter than they would before a pump run.
To see whether the suspensions in both tests were about equal, transfer the samples to a graduated cylinder with a stopper and measure the hindered settling rates of the slurries.
The settling rate, L/h is proportional to the square of the average equivalent particle diameter assuming all other factors, such as densities of liquid and solid, as well as viscosity of the liquid are equal on both slurries, and that Stokes' law applies.
For example, if the hindered settling rate doubles, it means that the average particle diameter has increased by about 40%.
Duplicate tests can be carried out by reslurrying the cylinder contents by inverting it several times. It is recommended to use same size cylinders for the sake of repeatability and reproducibility, so wall effects will cancel out.
Would you care to comment ?
- Thread starter
- #13
Answer to both 25362 and phex:
Dear 25362,
Both filtration tests were performed starting from a settled suspension. And I repeat: the filtration with the pump was done without recicrculating the liquid back to the reactor. The partcle size diameter was the same in both filtrations. So I drained the filter (reactor) once simply by opening the bottom valve of the filter, once by opening the bottom valve that was connected to the pump. In the first case, the liquid was directly collected into a drum, in the 2nd case the pump pumped the liquid into the drum. Draining with the pump was much faster. How come? Isn't this caused by the suction lift of the pump (1.5mWc), creating a "depression" under the filter?
Dear 25362,
Both filtration tests were performed starting from a settled suspension. And I repeat: the filtration with the pump was done without recicrculating the liquid back to the reactor. The partcle size diameter was the same in both filtrations. So I drained the filter (reactor) once simply by opening the bottom valve of the filter, once by opening the bottom valve that was connected to the pump. In the first case, the liquid was directly collected into a drum, in the 2nd case the pump pumped the liquid into the drum. Draining with the pump was much faster. How come? Isn't this caused by the suction lift of the pump (1.5mWc), creating a "depression" under the filter?
To phex
You'd be right if it were a closed loop system, but, as indicated by gilden, it is an open circuit. The flow to the pump is under a constant-level and gravitational (the reactor is under atmospheric pressure), so it can increase only if the downstream pressure is lowered by a suction lift, which gear (e.g., positive displacement, not centrifugal) pumps are capable of producing. The pushing pressure towards the pump is always the same, and in this case it is not the pump's discharge.
No doubt positive displacement pumps can create a "depression" but the flow depends on the acting differential pressure up the point of dissolved air release and/or NPSH limitations. See gilden's tests.
To gilden:
1. Only to be sure: was the sequence of testing first "no pump", then "with pump" ? The first test was done with a liquid that had a certain time for settling the suspension. Could the second test, "with pump", have been done with a "clearer" liquid that was probably credited with some more minutes of settling, and was more depleted of fine particles than the first ?
2. Since when opening a valve to the atmosphere the flow was pushed by a differential pressure depending only on the liquid level, and when operating the pump its lift tripled or quadrupled the flow rate, it means that the pressure beyond the filter was brought down to about 5 psi. Is it possible to measure the pressure at the pump's suction nozzle ?
Could you please comment ?
3. In short, gilden is right, indeed a gear pump may produce a vacuum (aka lift, depression). A pump doesn't actually suck, but can reduce the absolute pressure, and in this particular case the "pushing pressure" is always atmospheric. Positive displacement pumps are sometimes used to prime centrifugal pumps.
Sorry for any misunderstanding. One usually tends to think of centrifugal pumps, the workhorse in so many chemical and petrochemical installations.
You'd be right if it were a closed loop system, but, as indicated by gilden, it is an open circuit. The flow to the pump is under a constant-level and gravitational (the reactor is under atmospheric pressure), so it can increase only if the downstream pressure is lowered by a suction lift, which gear (e.g., positive displacement, not centrifugal) pumps are capable of producing. The pushing pressure towards the pump is always the same, and in this case it is not the pump's discharge.
No doubt positive displacement pumps can create a "depression" but the flow depends on the acting differential pressure up the point of dissolved air release and/or NPSH limitations. See gilden's tests.
To gilden:
1. Only to be sure: was the sequence of testing first "no pump", then "with pump" ? The first test was done with a liquid that had a certain time for settling the suspension. Could the second test, "with pump", have been done with a "clearer" liquid that was probably credited with some more minutes of settling, and was more depleted of fine particles than the first ?
2. Since when opening a valve to the atmosphere the flow was pushed by a differential pressure depending only on the liquid level, and when operating the pump its lift tripled or quadrupled the flow rate, it means that the pressure beyond the filter was brought down to about 5 psi. Is it possible to measure the pressure at the pump's suction nozzle ?
Could you please comment ?
3. In short, gilden is right, indeed a gear pump may produce a vacuum (aka lift, depression). A pump doesn't actually suck, but can reduce the absolute pressure, and in this particular case the "pushing pressure" is always atmospheric. Positive displacement pumps are sometimes used to prime centrifugal pumps.
Sorry for any misunderstanding. One usually tends to think of centrifugal pumps, the workhorse in so many chemical and petrochemical installations.
It's very true that liquids can't be towed or sucked(I mean literally). But centrifugal pumps do maintain lower pressures (than the pressure acting on the free surface of the liquid) because the pressure head is converted to velocity head. The lowest pressure will be at the eye of the impeller. This pressure difference is the cause for higher flow during pumping.
The main reason why PD pumps can prime better than centrifugals is because PD pumps can handle air in the liquid. Also, it is a common practice to prime centrifugals with vacuum pumps.
Pressure at the pump suction can be checked by a compound gauge. This will be phenomenal immediately after closing the suction valve of a running pump.
Regards,
The main reason why PD pumps can prime better than centrifugals is because PD pumps can handle air in the liquid. Also, it is a common practice to prime centrifugals with vacuum pumps.
Pressure at the pump suction can be checked by a compound gauge. This will be phenomenal immediately after closing the suction valve of a running pump.
Regards,
Positive displacement pumps can take (lift) air from the suction piping and exhaust it against system pressures.
Conventional centrifugals (not self-priming) deliver a head generally expressed in ft of water. Since air is about 800 times lighter than water, they couldn't develop sufficient pressure to exhaust air into the discharge side.
Once primed they can perform smootly with suction heads down to just above the vapour pressure of the liquid or the releasing of dissolved gases.
Conventional centrifugals (not self-priming) deliver a head generally expressed in ft of water. Since air is about 800 times lighter than water, they couldn't develop sufficient pressure to exhaust air into the discharge side.
Once primed they can perform smootly with suction heads down to just above the vapour pressure of the liquid or the releasing of dissolved gases.
bchoate
Considering the problem desciption and all the responses, I still posit that the best pump for this application is a laminar flow, disc flo pump. Perhaps many of you are not familiar with this pump but you should learn about it.
In the subject application:
A. The liquid circulation rate can be increased 33 - 50% because the disc flo's NPSHR is lower than a comparable centrifugal. The pump can also run with significant amounts of entrained air meaning that it is not likely to cavitate. The amount of liquid availble for recirculation is dependent upon that filter in the reactor. It is the limiting element. Circulation cannot be increased much over the area and pressure drop contraints of the filter. The disc flo may exert some pull on the filter (increase delta P).
B. Circulating the suspension - the disc flo is very gentle on suspended solids. There is almost no reduction in particle size. We have used these pumps in crystallizer applications as have many others. The disc flo can pump grapes in water without bruising the grapes; live goldfish in water without harming the fish. Laminar flow pumps are the solution if preserving particle size is important.
Considering the problem desciption and all the responses, I still posit that the best pump for this application is a laminar flow, disc flo pump. Perhaps many of you are not familiar with this pump but you should learn about it.
In the subject application:
A. The liquid circulation rate can be increased 33 - 50% because the disc flo's NPSHR is lower than a comparable centrifugal. The pump can also run with significant amounts of entrained air meaning that it is not likely to cavitate. The amount of liquid availble for recirculation is dependent upon that filter in the reactor. It is the limiting element. Circulation cannot be increased much over the area and pressure drop contraints of the filter. The disc flo may exert some pull on the filter (increase delta P).
B. Circulating the suspension - the disc flo is very gentle on suspended solids. There is almost no reduction in particle size. We have used these pumps in crystallizer applications as have many others. The disc flo can pump grapes in water without bruising the grapes; live goldfish in water without harming the fish. Laminar flow pumps are the solution if preserving particle size is important.
- Thread starter
- #18
To 25362's 1st question:no, both tests were exactly performed in the same conditions.
To 25632's 2nd question: it should be possible to measure pressure in suction line, but not yet done (equipment ha sto be modified a little bit). I will consider to install a pressure gauge in the suction line.
To bchoate: concerning remark b, I repeat the suspension is not recirculated, only the liquid part of the suspension, since there is a filter in the bottom of the reactor (before the pump). However, I will check in detail the pro's and con's of a disc flo pump.
To 25632's 2nd question: it should be possible to measure pressure in suction line, but not yet done (equipment ha sto be modified a little bit). I will consider to install a pressure gauge in the suction line.
To bchoate: concerning remark b, I repeat the suspension is not recirculated, only the liquid part of the suspension, since there is a filter in the bottom of the reactor (before the pump). However, I will check in detail the pro's and con's of a disc flo pump.
Compositepro
Chemical
The solution to all your problems is to put the filter after the pump. With the filter on the suction side you can never get more than 15 psi across the filter, usually much less. Place it on the dischargew and you have full pump pressure available for filtering. However, a gear pump does not handle solids well and can develop dangerously high pressures if the filter plugs and there is no pressure relief valve bypassing the filter. A centrifugal pump could work. A double diaphagm air operated pump would probably be best for this application. It can handle large solids and it operates a whatever speed which will maintain the set output pressure.
The comment that liquid cannot flow out of the reactor faster than it would drain by gravity is not correct. Any suction on the outlet will increase the flow.
Centrifugal pumps will pump some air if it is entrained as bubbles in the liquid. Self-primping centrifugal pumps contain liquid resevoirs that recirculate liquid through the pump and in the process entain air in the liquid on the suction side of the pump and allow it to separate on the discharge side.
The comment that liquid cannot flow out of the reactor faster than it would drain by gravity is not correct. Any suction on the outlet will increase the flow.
Centrifugal pumps will pump some air if it is entrained as bubbles in the liquid. Self-primping centrifugal pumps contain liquid resevoirs that recirculate liquid through the pump and in the process entain air in the liquid on the suction side of the pump and allow it to separate on the discharge side.
All compositepro comments are correct.
Just one point about the filter's purpose -and the filtered particle characteristics- which haven't yet been clarified by gilden.
1. If the presently filtered particles aren't participating in the reaction by being settled out and removed from the suspension, thus lost, then one has to question the need of the filter.
2. If the filter's only original purpose was just to protect the pump, one may not need it at all to speed up the circulation, and its removal would be a logical step.
The pump could, then, serve to enhance the suspension charateristics by prividing mixing and possibly a measure of beneficial particle attrition.
3. Relocating the filter downstream -if however needed for process reasons- may involve its re-design considering the pump's possible particle attrition effects.
4. Removing the filter from the reactor's bottoms would, nevertheless, require some knowledge about the slurry, such as particle hardness, size distribution, densities, apparent viscosity, concentration, etc., for a proper selection of the pump.
Gilden's comments are welcomed.
Just one point about the filter's purpose -and the filtered particle characteristics- which haven't yet been clarified by gilden.
1. If the presently filtered particles aren't participating in the reaction by being settled out and removed from the suspension, thus lost, then one has to question the need of the filter.
2. If the filter's only original purpose was just to protect the pump, one may not need it at all to speed up the circulation, and its removal would be a logical step.
The pump could, then, serve to enhance the suspension charateristics by prividing mixing and possibly a measure of beneficial particle attrition.
3. Relocating the filter downstream -if however needed for process reasons- may involve its re-design considering the pump's possible particle attrition effects.
4. Removing the filter from the reactor's bottoms would, nevertheless, require some knowledge about the slurry, such as particle hardness, size distribution, densities, apparent viscosity, concentration, etc., for a proper selection of the pump.
Gilden's comments are welcomed.
- Status
Similar threads
- Locked
- Question
- Locked
- Question