How does sensorless field oriented control algorithms (induction motor) regulate slip? 3

[Electrical_Engineer]

This is a long question involving closed loop motor control and I couldn't find any resources on this. ANY help is appreciated.

Is it possible that the VFD ends up operating at the high slip region when running DIRECT sensorless rotor flux oriented control? Direct sensorless rotor flux control, as the name implies, estimates the rotor flux DIRECTLY using the math model. Which means as long as there is rotating rotor flux, it can be estimated. After estimating rotor flux, it sets the direct and quadrature current based on the setpoint in the rotor frame.

So now imagine this situation: The rotor is forced to spin at 10Hz through external load. A VFD starts open loop V/F at 60Hz. So, now, a rotor flux is generated with synchronous speed 60Hz. The sensorless direct rotor flux control is switched on. The algorithm align the frame to the rotor flux, which is spinning at 60Hz, and sets the Id and Iq. Which is a problem, right? there is nothing telling the algorithm to slow down and it just keeps following the rotor flux at 60Hz, even though the rotor is only at 10Hz.

So essentially, do field oriented control algorithms regulate slip just by their design or is there a chance they could end up in the high slip region?

 

[waross]

Not all VFDs are capable of catching a motor "On the fly".

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[Electrical_Engineer]

That's a good point. But it could also happen during normal FOC operation.

For instance, the motor is running at 50Hz and stable FOC operation. Now there is a sudden load change and the rotor speed drops to 10Hz. Now will the FOC maintain a 50Hz input or does the FOC algorithm automatically end up running at a lower frequency? Because the FOC algorithm does not look like it differentiates high slip or low slip and FOC is also theoretical possible at high slip.

 

[itsmoked]

There is no "sudden" to the FOC process. FOC algorithms measure the rotor position every single pass of each electrical phase. It essentially sees all motion. It can't even provide the next output update without knowing the fractional position of the rotor. The current too. The actual high level control will get the message things have gone haywire and either fault to protect all the hardware or tool the drive output down to something that won't toast the electronics.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[FacEngrPE]

The exact answer is likely vendor and model specific. It could also depend on the motor.

https://www.lnelectric.com/2020/10/what-is-a-high-slip-motor/[/URL]]
Standard NEMA Design A, B, or C motors operate at one to five percent slip. A NEMA D motor has a high slip of five to thirteen percent.

The larger the design slip angle the lower the rated speed.

I would expect that a load change that resulted in a motor dropping speed from 50 Hz to 10 Hz would result in a trip on max amps, unless the drive was set to constant torque, or torque control. Unusual, but some drives can do this.

 

[LionelHutz]

As pointed out, the algorithm operates much, much, much faster than the rotor speed could ever change so you'd never change the rotor speed without it knowing. There is also a self tuning function that must happen before the algorithm can be activated, and that lets it learn the motor parameters enough it would easily know the rotor is slipping.

Also, an induction motor at high slip draws a lot more than rated current and the VFD output transistors could never handle that current level unless it is very oversized.

 

[Electrical_Engineer]

@LionelHutz @FacEngrPe @itsmoked Thanks for your inputs.

I agree the FOC algorithm runs much faster than the mechanical process so there is no really "sudden change". But the transient aspect can be ignored for my question.

Another way to ask the same question: Suppose I run the controller in torque mode and set a torque or (Iq) reference, and the load reaches a steady state speed (which is not measured or estimated). The controller is generating a 30Hz (on average) sine wave at steady state. Is there a way to estimate what the load speed actually is without explicitly calculating slip? And if I don't care about the load speed, can I assume the FOC is doing its thing and the controller is operating efficiently?

The motivation for this question is this: Is slip estimation only necessary in case of high performance speed control and NOT needed for torque control? I remember reading about algorithms that use slip information ONLY in the speed control loop and not in the torque control loop. Since torque control loop ignores the slip/rotor-speed, is there a risk of running the motor inefficiently in high slip regions if I use just torque control?

 

[itsmoked]

You're really into an area that only the specific drive makers would know how their products work. Fundamentally the drives will electrically protect themselves and the motor if the drive was setup correctly. So. If you want to answer third-order efficiency questions like you're asking it would be more effective to run the drive/motor/process in the modes of interest and measure the efficiency. That's the only way you will have an actionable answer. You could find that efficiency quickly takes a back-seat up against process stability or torque pulsation etc etc.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[jraef]

As mentioned, everyone does it slightly differently, but there are generalizations that can be made. For a lot of them, the rotor position is actually measured by looking at distortions in the stator field current made by the rotor bars passing through. You might think of this as a form of incremental encoder because the rotor position is always going to be relative to the last known position, so it loses accuracy as speed approaches zero. But a few mfrs have been able to elevate this to the level of absolute encoder, so that the drive can detect rotor position at a standstill. Suffice to say though that once the rotor position is known, all other control functions can stem from that and the proper vectors can be delivered to the motor at any point.

In a lower cost “Sensorless Vector Control” drive, that rotor position is often estimated, not calculated, so response time to a step change in load may be measured in the number of revolutions that take place before control can take effect. But in drives with “Encolderless Flux Vector Control” or “Encodeless Field Oriented Control”, the response can be measured in a few radians. The differences are in the precision of the internal current sensors and the power devoted to the math, usually in the form of a math coprocessor on the high end drives, as opposed to trying to get a low cost DSP to handle everything. There is no free lunch...




" We are all here on earth to help others; what on earth the others are here for I don't know." -- W. H. Auden

 

[Electrical_Engineer]

@jraef
I did not know about using rotor bars to estimate rotor position. That's very interesting. The textbooks did not cover that.

@itsmoked
I agree it's very hard to find the details of the algorithm in commercial drives and it's limitation without experiments. And, I realized now it's just an academic question. But I was hoping someone here knew the finer details of these control systems.

 

[FacEngrPE]

Another way to look at this question is

The motor is an object governed by physics. It's electromagnetic behavior can be described by equations. (this includes the relationships that are associated with slip angle. This is an example description "How an Induction Motor Works by Equations (and Physics)"
From this point of view the slip angle is not controlled by the drive, it is the physics result of the drives inputs to the motor.

In order for the drive to create appropriate inputs to the motor, the drive program must contain a mathematical model of the motor. For a sensorless drive, the only inputs to the math model are the parameters that can be measured at the drive.
A Volts per Hertz model, is fit to some uses, but is very simplified, and fails at many tasks.
A flux vector model is a more exact model, requires more processing power, and more accurate sensors.
If a drive needs to operate in all 4 quadrants, or needs to catch a moving load, the model complexity grows some more, and may need to detect the load inertia, and other characteristics.
I tend to consider this electrical engineering magic, and am happy it is useful.

 

[cswilson]

In the standard operating range of induction motors, torque and slip frequency have an approximately linear relationship.

In open-loop control -- DOL or V/Hz -- as the load torque varies, the speed, and so the slip frequency will change until the generated torque balances the load torque.

In closed-loop (field-oriented) control, the controller algorithm will take the torque command (which could be the output of a velocity loop) and compute the slip frequency needed to create this torque. It will add the advance angle that this frequency would create over one sample cycle to the measured/estimated mechanical advance angle that the rotor velocity creates over one sample cycle.

In full "sensored" FOC, the shaft sensor is used to detect the rotor angle and velocity. In "sensorless" (really "shaft-sensorless") FOC, the drive electrical sensors and a more sophisticated motor model (often tied to the particular motor design as others have explained) are used to compute the rotor angle and velocity.

In both cases, these calculations are updated thousands of times per second. I can't think of a situation where any decent FOC algorithm would get "caught" outside of the standard operating range for slip frequency.

Curt Wilson
Omron Delta Tau