Percent off Pump Curve 2

[refinerypumps]

Is there a standard method for calculating the percent off a pump curve for pump which has degraded performance over time (wear rings, impeller wear, any other reason really)? The way I see it there are two possible methods but maybe there is additional methods I am unaware of. The two I can think of either is a constant flow or constant head method.

For example if you had pump which you measured to have a TDH of 100ft @ 100 GPM and the curve showed at a TDH of 100ft the flow should be 150GPM your percent loss based on flow would be 1- 100/150 or 33% loss. However the curve would also show at 100GPM the head should be 110ft so your loss based on head would be 1-100/110 or 9%.

Also is there a percentage off the pump curve you generally consider to be acceptable based on those two or other methods? Like is 5% considered reasonable based on measurement errors, just some losses from wear ect. What about 10% or more? Or does it just depend upon what the losses are causing process wise.

 

[itdepends]

We make money with flow- not head. So I look at flow when talking about a drop in pump performance.

However strictly speaking- head is more technically correct. The reduction in head at constant flow is a function of pump wear. The reduction in flow is impacted by the system curve.

You can have two pumps with exactly the same level of wear- and they'll give the same reduction in head but different reductions in flow if they are on separate systems with different system curves

As a chem eng/metallurgist the first part of any answer I give starts with "It Depends"

 

[JJPellin]

I am not aware of any standard for this calculation. I generally use percent of head at a constant flow. If the curve is relatively flat, this will give a much different result than you would get if you calculated percent of flow at a given head.

If you determine the loss of flow for a pump operating in a system, then certainly, the system curve will play a role. But, if you relate everything back to the pump curve as percent of flow at a given (fixed) head, then the system curves is no longer relevant. By definition, the head is constant. You are simply comparing to flow values for the pump in new condition and in the degraded condition at a fixed value for head.

I track and report on the performance of two of our largest pumps. These are our coker jet water pumps. I report on the pump performance as percent of curve at rated flow. I take the flow and head of the pumps. I convert the head to the rated flow using the affinity laws. I compare the head at the rated flow to the head value from the original test curve for the pump at that same flow rate.

(Head today at 1000 gpm) / (Head on test curve at 1000 gpm) x 100 = percent of curve at rated flow.

Johnny Pellin

 

[refinerypumps]

Thanks for the insights guys. I actually asked our pump vendor who happened to be onsite today the same question he said they generally go off the difference in head at a given flow, but there are inconsistencies so you should always state how you are calculating it.

Anyone have any thoughts on percent off curve where you can be confident pulling and rebuilding the pump will improve performance. I currently have our operations group pushing to pull and rebuild a pump because "it isn't performing as good as it used to" but I haven't seen any data that backs that claim up at all and the pump is less than 5% off the curve. To me 5% seems like it could be from anything and you could pull the pump and find nothing and just waste time and money. Pressure gauges, flow measurement, changes in specific gravity, ect.

Thanks again for your previous thoughts and any more you have.

 

[itdepends]

I'd be looking at if you see that 5% reduction under dead head before blaming the pump. Most of our "pumping" issues end up being due to restrictions in a suction or discharge line.


As a chem eng/metallurgist the first part of any answer I give starts with "It Depends"

 

[LittleInch]

The other thing to base this sort of judgement on is to work out efficiency and compare to test data. Most data should be there or easy to find. Then you have an opex cost to continued operation versus cost and disruption to removal and repair.

My motto: Learn something new every day

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

 

[refinerypumps]

Itdepends: I do not think dead heading this pump is an option from a process standpoint but I will confirm. However how would a restriction in the suction or discharge lines affect pump performance in relation to the the curve? My pressure gauges are right at the suction and discharge nozzles. Any restriction in the suction line would show up as low suction pressure. I have 30+ Nsph margin so I think I am good there. A restriction in discharge would change the system curve and push the pump back on the curve but if the loss is measured relative to the curve and not just new flow / old flow this shouldn't matter right?

Please correct if I am wrong somewhere I am still very new to this.

Little inch: good point my question though is can you actually expect to return to the labeled curve after you rebuild the pump or should you always expect to be a little under in the real world.

 

[JJPellin]

You would need to provide more information about the pump. I have large, multi-stage charge pumps that are 5% below the curve the day after we perform a full overhaul. In most cases, I would not pull a pump for a 5% loss of head. But, in some services, 5% would be enough to make the pump almost useless. What is the configuration of the pump? How many stages? Do you have historical data from previous overhauls? What damage did they find? What performance change did they see following an overhaul?

Restrictions in the upstream or downstream system will not change your evaluation of pump performance as percent of curve. But, it will change the operators perception of how bad the pump is performing. And the louder the operators howl, the harder it will be for you to resist their push to overhaul the pump.

Johnny Pellin

 

[itdepends]

Some of our pumps operate with very little NPSH margin (<2m) and I've had instances where scale caught in the eye of the impellor on liquid pumps (i.e. those that don't have inlet screens) results in a pump operating below the curve. It's a lack of NPSH/cavitation- but not one you'll pick up using an inlet gauge.

A dead head test though shows the pump performing as expected- and you can see the delta between the theoretical and actual head increasing as you test higher flows- again with my measured NPSH marging remaining "ok".

Based on your second post it wasn't clear to me that you had put gauges on the pump suction/discharge. My comments refer to operations asking for a pump change out before any pressure/flow checks are done (including a dead head).

As per JJPellins posts- if you're after more specific help you're going to need to provide more specific information.

As a chem eng/metallurgist the first part of any answer I give starts with "It Depends"

 

[Artisi]

At the end of all this, my question is why do you need it?
I would suggest that a percentage of the production lost is a more meaningful measure, irrespective of what the head or flow reduction is, ie current flow / design flow x 100 = reduction in production performance. Of course this is on the premise that the pump was operating at open discharge and not throttled in any way to reduce production output -- if operating with throttled discharge, reducing the restriction will / should return to the required output, again doesn't matter what pump reduction is in percentage as once the pump can no longer meet production requirements - either, time for maintenance or live with a reduced production output.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)