Butterfly Valve Water Flow Test - Torque Coeff Const???

[AUtigerUSCtrojan]

Hello All,

I am writing a test procedure for water flow testing a 10 inch butterfly valve. The application is 10000 GPM liquid hydrogen at 15 psig at the inlet. In addition to pressure drop, I am trying to model the dynamic load (torque) to the blade/shaft assembly experienced at the application flow condition with a water flow test. In order to have dynamic similitude, the Reynolds Number should be constant, but that results in a test flow rate 5 times that of the application. Instead, I will be keeping the dynamic pressure constant to match pressure drop. Will the torque applied to the blade/shaft assembly during the water flow test equal that of the application flow condition? I remember reading in the literature somewhere that the coefficient of torque w/r/t dynamic pressure (Ctd = Torque / (1/2 Density * Velocity ^2) is function of Reynolds number and viscosity, therefore the torque should be different in these two cases. Is there a way to match torque between these two flow conditions?

Thanks,

Chris

 

[bao2ye]

no, I do not think you should keep the pressure drop constant. You need to consider variance pressure drop since the disc is moving. Also the GPM changes.

 

[AUtigerUSCtrojan]

I was not clear. I am keeping dynamic pressure constant between the application and test conditions, not constant during the test. This is just to define the initial flow condition. As the blade rotates the flow/pressure is what it is and is not being controlled. I am trying to come up with what the intial flow condition should be to match forces on the blade/shaft assembly. Similarity in torque is of concern, but maybe pressure drop similarity will suffice. That is my question.

 

[bcd]

Hello,

Unfortunately you are heading to a dead-end. Your best bet to approach this problem using actual media properties is CFD. The accuracy of CFD has its limits and it is difficult to corroborate the results.

Dynamic torque is created by a non-uniform pressure distribution across the valve disc. The shape and thus the resultant magnitude of the force on the disc and center of pressure from this pressure ditribution in relation to the shaft centerline will vary with the media density, viscosity, temperature, pressures, and flow rate. It is an unsolvable problem when approached this way.

So how is it done? Flow tests are done with water and the torque measured at the valve shaft is measured in both while opening and closing the valve. Mathmatecally subtrascting the two results eliminates the effects of friction. Knowing the valve Cv, flow rate, and differential pressure at each measured position will allow you to calculate and factor known as the dynamic torque coefficient (Cd). To predict the torque induced by flow around a butterfly valve disc, you multiply Cd by the valve port diameter^3 times the differential pressure. This yields the dynamic torque. To this you need to add back in friction depending upon the desired direction of rotation. Note that the equation for dynamic torque does not include any media properties, just a dimensionless factor and differential pressure. The media properies are included when calculating the differential pressure at each angle of opening.

If you are using a commercially available product, most manufacturers can give you the curve of Cd versus position for their product.

bcd

 

[adrenalin]

hi everyone,
I have a question related with this topic.
In fact much more simplier then that
Is the dynamic torque applied on the disc by the flow, towards the opening direction?
I mean should I get help from the flow while opening a butterfly valve?
thanks a lot...

 

[AUtigerUSCtrojan]

bcd,

Unfortunately we are the manufacturer. It is my understanding that the torque coefficient (Ct) method is useless unless the Reynolds number of the test case and that of the actual application are similar. Ct differs greatly from Re ~ 10^5 (water flow test case) to Re ~ 10^7 (actual application) and won't produce correct torques if calculated from one condition and applied to the other. I think I will have to choose a water flow test condition and get data, model the test condition in our CFD tool (CFDesign) and compare the results to the test data to validate the CFD tool, and then run the CFD tool at the application condition and assume the results are correct. Any other suggestions?

 

[bcd]

Hello,

The method you proposed is about as good as you can do. You are theoretically correct that Reynolds number effects the coefficient, but lacking other methods, the coefficient is generally good for the normal conditions to which butterfly valves are exposed. Validate the CFD results then apply the analysis method to your actual conditions. It's a lot of work but is the best practical option for accuracy.

bcd