two things happens when the temperature drops:

The desity changes
The viscosity changes

The head is not affected by the change in density and neither is the pipeline pressureloss in velocity heads. If the viscosity drops then the pressure loss however changes - it drops. This would mean that a new "flatter" systems resistance curve could be drawn in the pump performance curve. Since its "flatter" than the cold the flow would _increase_ when the viscosity drops!

The actual pump outlet pressure could drop because the density drops. Acually the pump duty would also drop (for the same reason).

Best Regards

Morten

ksnar,

As temperature increases, viscosity decreases.  For a specified starting pressure, this will result in increased flow because the viscosity decrease will cause a velocity head drop in the system.

If you want to understand the physical reason why a less viscous fluid would experience less velocity head loss, I suggest you look at "Boundary Layer Theory" in your fluid mechanics book.

What type of pump is it? Centrifugal or positive displacement?

Steven van Els
SAvanEls@cq-link.sr

Steven,
It's a positive displacement pump.
I'm familiar with the boundary layer theory and when I understand it correct, the flowrate will increase when the temperature increases because of the lower viscosity and less pressure loss.
However our curve shows a higher flow rate at a give speed and pressure for cold oil than for hot oil.
Can this be due to an increase of leakage over the pump when the oil is less vicous? I cannot think of another explanation.  

It seems a problem of increased internal leakage, what type of pump is it, gear, screw, progessive cavity?

Best Regards

Steven van Els
SAvanEls@cq-link.sr

ksnar,
hope this posting does not come too late into the discussion...
steven's question hit it right on the head.
Positive displacement pumps (PDP) behave completely different from centrifugal pumps (CP).
The PDP have what is called a slip factor (similar to the slip in induction motors). The higher the viscosity of the fluid pumped the less is the slippage and the flow vs head curve tends to the "ideal curve": a vertical line at the design flow @ operating speed, at double the speed the new flow would be another vertical line at double the original flow (this is the definition of PDP: flow/cycle = constant)
Now, the system curve behaves differently: the higher the viscosity the higher the pressure drop for the same flow.
So we have two families of curves:
1. pump capacities at different viscosities.
2. system curves at different viscosities.
If all are plotted on the same graph where the pump curve intersects the system curve (for the same viscosity) a series of operating points are obtained at different viscosities.

Then, a similar plot is made for the NPSHR (required NPSH) and NPSHA (available NPSH), again a the different viscosities. Another curve is determined that joins the operating points at different viscosities, if the pump operates at flows higher than the flow determined by those points the fluid will reach the cavitation point.

If this sounds complicated already, it gets worse...
Now a 3rd plot is required: flow vs viscosity.
Each of the flows, viscosity points determined before (Operating points and cavitation points) are plotted obtaining two curves:
1. op flows at different viscosities
2. limit flows for cavitation (flow above this point will cause cavitation).

Where these last 2 curves intersect is the MAX VISCOSITY for operation without cavitation, in other words: this point defines the MINIMUM TEMPERATURE of the fluid for operation of the PDP without cavitation.

For centrifugal pumps the story is different, there are formulas and curves that correct for different viscosity: as a general rule the performance of the pump is deteriorated by higher viscosity fluid.

HTH
saludos.
a.

Abeltio, your post was most helpfull. Steven also helped me to pin down the problem to the increasing internal leakage.

"Assuming" you are dealing with a positive displacement pump with unvarying clearances: you will experience a certain gpm/psi slip as defined by the manufacturer. This slip factor will be based on a given viscosity. The actual slip will vary as the viscosity varies: Say the given viscosity is 100 SSU; pumping 70 SSU might result in (rule of thumb?) 100/70 x slip factor, and 500 SSU would be more like 100/500 x slip factor. Then yes, generally speaking,  you would experience less slip as your viscosity increases, which you would observe as an increase in flow.