flumoxed by flux vector control

[cryomark]

I am uncertain about my understanding of open loop vector control with a VFD. My take on it is that the drive makes a motor speed calculations/adjustments based on recorded motor parameters and measured I, V and their phase relationships.
Is this close?
We have an application which is giving us some trouble.
The motor speed readback from our open loop vector application is very noisy. At first we investigated instrumentation; then considered noise; It is in fact real variations in the VFD output.
The load is a 2 cylinder, reciprocating gas liquifier. Pistons are connected to a flywheel, which is coupled to the 7.5HP AC motor by 1:6 pulley/v-belt arrangement.
Because of the 300 psia input pressure, the motor functions as a brake. This means the motor is always being over driven. We are using a regenerative braking device but at low speeds(Heavy load) there is almost no regeneration. The apparent speed variation, as reported by the VFD, is ~10% of the nominal speed(300RPM min-1500RPM max)
Is it possible that the over-driven condition of the motor is confusing the VFD's vector calculations?
Am I all wet?
Am I waiting for some thoughtful input?
You betcha.

 

[rhatcher]

First, you need to distinguish what information provided by your drive is real and what is calculated or estimated. Clearly, "speed feedback" with an open loop vector drive is an estimated value based on calculations made using frequency output and current. In most cases the only real values available from the drive will be voltage, current, and frequency. Everything else is calculated. Be careful though as I most recently worked on a system with a 400 hp Toshiba drive which only measured output current on two phases and estimated the third (!). The drive was tripping out on overcurrent upon start up and was registering negative speed and torque among other things (load was an arbor saw). This had led the poor engineer at this facility to assume the drive was at fault and only after replacing the drive (component by component) and not fixing the problem were we called to help. This drive could calculate every parameter you could think of except what your wife was cooking for dinner that night, all based on three phase voltages and two phase currents as input. He had looked at the two measured phase current values just long enough to note that neither was in overcurrent when the drive tripped. However, the wealth of erroneous data somewhat obscured for him the fact that the two measured phase currents were imbalanced with the higher phase current having a sagging phase voltage associated with it. I am still not sure why the drive was tripping on overcurrent since I didn't bother looking at the third phase (calculated) current. We went to straight to the motor and (with a surge tester) found that the phase in question had shorted turns. In short, all of the calculated values were erroneous once that one phase shorted since the calculation model (obviously) assumed balanced current. A poor design if you ask me. So, you are not all wet yet....

Back to the issue at hand, I am wondering if your problem may be a tuning issue. Are the speed variations periodic? In other words, is there a repeating pattern in the speed variations?
I am not directly familiar with your application, but I imagine that it would be very similar to a 2 cylinder air compressor except that it is a regenerative load(?).
If so, your motor will see a load cycle of 2 periods of high load separated by periods of somewhat light load for every six revolutions. Although a flywheel is present, it's dampening affect will not be complete (to do so would require an impractically large flywheel compared to the actual load). In addition, the drive belts will have some elasticity based on the type used and the amount of tension applied, making the system's response somewhat soft or spongy. If the VFD is trying to speed regulate itself with this type of load it may be overshooting the mark, causing a periodic (repeating) fluctation in speed. If this is the case you may reduce the fluctations by tuning the drive PID loop.

I hope this helps...

 

[cryomark]

Thanks, rhatcher!
The varying load from gas expansion was the cause of over regulation of the VFD output. Fluctuations in the high pressure gas supply and the output sink pressure were complicating matters but when these were factored in, the correlation of "noise" to piston stroke was obvious. We managed to massage the PID loop to the point where regulation, response and readback are all acceptable.

 

[jbartos]

Suggestions:
1. Simulation by suitable software could be used to verify your PID fine-tuning

http://www.speedcad.com/

2. A flywheel could be added
3. Caution could be exercised since the fine-tuned PID does not account for potential future irregularity in the system functioning. There are no guarantees.