HPU Sizing from time history of forces

[n1210933]

Hi,

I am designing a test rig which will apply forces to a rotating shaft via hydraulic cylinders. We have a spreadsheet giving the size of these forces against time, so my question is this:

If we know the deflection to the shaft caused by a given force, can I use the gradient of the force/time graph to determine the oil flow required? Ie. I know the deflection at time 1 and I know it at time 2, so I know how far the cylinder has to travel between these times, and therefore its speed. This means I can determine the required oil flow from the steepest gradient of the graph.

Is this a reasonable approach?

Thanks for your help,

Pete

 

[IceStationZebra]

Sounds perfectly logical to me. Just remember when sizing the HPU to take pump efficiency and line losses into account.

If you have a dedicated load case and really want to optimize the system you could use accumulators to augment the flow during peak flow demand. But you would need to calculate this carefully as you will not have any margin for error and you will limit changes to the load case. You may need an accumulator or two if using a variable displacement pump anyway, depending on the rate of change on your specimen, as the pump can only react so fast to changes in flow demand.

ISZ

 

[n1210933]

Thanks, just needed to check if my theory was correct!

 

[kcj]

I doubt the pump concept is fast enough.

Is this a signficant distance/deflection, or more of a change in force applied to roating shaft? How fast does it need to operate.

Sound more like you are building something like an MTS test rig.

Then you will need for sure a variable pump (but I suspect it will be quite small for the average flow over an entire cycle), an fairly large accumulator (pump cannot respond fast enough), servo valve mounted on actuator for high natural frequency, temposonics type of feedback, and probably pressure transducers/load cells to measure force. All of this needs a fast motion controller to sort it all out.

Peter may be around here somewhere.

kcj

 

[IceStationZebra]

Since the output is force against a rotating shaft I assumed the output would be on the small side. I was also thinking the same thing about MTS, but it all really depends on the response time and accuracy required.

If the force changes relatively slowly one could use a simple pressure transducer and PLC. A really "dumb" system could simply toggle between 2 or 3 preset pressure regulators (which I have done with pneumatics before). A variable orifice could be used to tweak the rise/fall rates.

Pete, if you need design help there are plenty here willing to throw in their $0.02. I just commissioned a 200hp MTS HPU, and as KCJ said there are others with extensive controller knowledge.

ISZ

 

[n1210933]

Hi,

Thanks for your input, its very helpful. We have two MTS HPUs in our facility already and we'd like to use them again so that we can borrow capacity from them when not in use. I'm in touch with them already but we are doing some preliminary design so that we can do some costings.

 

[PNachtwey]

I thought kcj explained what is required very well.

If we know the deflection to the shaft caused by a given force,

Do you know that? Isn't that what your are testing?

can I use the gradient of the force/time graph to determine the oil flow required?

Yes but why not put a load cell on the system to measure force or put pressure transducer on either side of the piston. The one can measure force by:
Force=CapArea*CapPressure-RodArea*RodPressure.
This is not quite as accurate as using a load cell because there isn't a good way of measuring seal and other frictions. A good controller can then generate any force wave form you want within reason and the controller will attempt to follow the it. You don't need to calculate flows ahead of time. You then record the position or deflection as a function of time. The force and position/deflection data can be recorded as a function of time an uploaded into to a PC where it can be converted to Position/force or force/position in something as easy as Excel. You just need to set up the graph to plot the x and y data the way you want. BTW, you should be able to measure the positions and forces at 1 millisecond intervals or faster.

Ie. I know the deflection at time 1 and I know it at time 2, so I know how far the cylinder has to travel between these times, and therefore its speed. This means I can determine the required oil flow from the steepest gradient of the graph.

One can also generate a motion profile and record the pressures/forces and then do the same as mentioned above.

What hasn't been explained yet is how the linear actuators are twisting the shaft. If the shaft that has a arm that sticks out that the actuator pushes on then there is a need for some trigonometry to calculate the true displacement or applied torque. If the actuator is pushing the rack for a rack and pinion then no trig is required.

Let us know what the quote is. There are probably some cheaper alternatives.