I agree with @Peter Nachtwey: as I said in my first response, the problem is not in the PID.
- What is the pressure of the hydraulic fluid at the discharge of the pump when the press measurement is hovering at 68PSI?
- What is the pressure of the hydraulic fluid at the press when the press measurement is hovering at 68PSI?
- Is the valve open a little bit, or perhaps leaking a bit, when the press measurement is hovering at 68PSI?
Also, I think I may understand the process now; let me know if the following is correct.
There are three, maybe four, relevant elements in the system. In order from highest hydraulic fluid pressure (HFP) to lowest HFP, the are the following:
- The discharge of the hydraulic pump has the highest HFP in the system
- Next is the control valve, across which there is a drop in HFP
- The magnitude of the drop in HFP here is a function of valve position, flow rate, and possibly downstream (down-pressure) elements
- Next is the volume of hydraulic fluid above the press itself, that effects the pressing of the press onto the product
- Finally a bleed(?) valve or orifice or other restriction, across which there is a large drop in HFP, and through which hydraulic fluid flows back to the pump inlet or pump intake reservoir.
There are obviously hydraulic lines in between those elements, across which there is some (I assume minimal) pressure drop.
So a simple, first-order, steady-state model of the system is
HFPpress = (HFPdischarge*Rbleed + HFPintake*Rcv) / (Rcv+Rbleed)
where
- HFPpress is the HFP above the press
- HFPdischarge is the HFP at the pump discharge
- HFPintake is the HFP at the pump intake
- Rbleed is the resistance of the bleed device, the reciprocal of its flow coefficient
- Rcv is the resistance of the control valve, the reciprocal of its flow coefficient.
Am I close? Am I missing anything significant?
Caveats
- The actual system is obviously far more complex i.e. non-linear, and not always at steady-state. But if we are using a PID, then we are treating it as more or less linear in the region of operation, and the PID's job is to get the system to a target steady state.
- Again I am ignoring pressure drops through the hydraulic lines, or at least lumping them into the Rbleed and Rcv model parameters.
- This model ignores compressibility (half a percent per kPSI?), so there is an In-Out=Accumulation part of the model that I am ignoring, where Accumulation may be linear, or at least monotonic, with HFPpress, but that would only affect the dynamics because dAccumulation/dt is zero at steady state.
- A fifth element would be the intake (reservoir) of the pump.
- The last element, the bleed, may not be present, or maybe it is "leakage" to a sump
- I understand fluid statics and dynamics fairly well but am willing to learn more, and I don't know a lot about the details of hydraulic systems.