As I said earlier, you can't expect 50% PWM to be equal to 50% max speed.
You can measure speed and/or flow and control motor speed indidectly. e.g. 0 to max, not 0 - say 1000 RPM proportionately.

PID control (basically matches setpoint to measured value) is often done in software. I personally have not implemented auto-tuning, but I have done PID.

 

Semi old thread.
1. you could put it through a variable resistor adjusted by a servo that changes the 'tapped' position on the variable resistor. Power the servo through PWM based on the valve positions.

2. or through different rated resistors. You can parallel the output from the resistors to the motor and have separate contractors driven by outputs on the board to allow power through the desired resistor rating.

Not really efficient but cheaper than a VSD.
Interested to know if you solved the problem in your project. Let us know how if you did, I'm interested

 

Thank you Cory. Not resolved quite yet, but looking at using Arduino as a controller. As for KISS's comment.. the good thing about this is that I don't need a variable flow so much as a stepped flow. I plan to map out the different flows needed based on the state of the system and see what power levels are needed to change the motor output. If X, then Y PWM. If the physical functionality of the motor changes/degrades, it won't be accurate any longer, so I see your point about closing the loop, but again, cost is an issue. Besides, if functionality did become an issue and a closed loop controller was in place, I think it would have to keep increasing power to the failing motor and just hasten its death. Maybe that isn't the case?