PSA on using VFDs and motor Service Factor

[Jraef]

Just a cautionary tale.

I have a situation right now where we put in some VFDs on pumps, even AFTER double checking the motor nameplate FLAs, and are now being told that the pump OEM was intending on running the motors into the Service Factor continuously, so they are saying we sized the VFDs incorrectly (we were not told that).

Just so everyone knows, when you run a motor with a 1.15 Service Factor from a VFD, the motor NO LONGER HAS A SERVICE FACTOR! Most "inverter duty" motors state that right on the motor nameplate (they did in this case) by showing that if run from an inverter, the Service Factor is 1.0. Non inverter duty motors might not state that on the nameplate since the motor mfr will tell you they are not meant to be run on an inverter drive, but if you ask, they will tell you, no service factor.

Here's an example (not from this project, but another one I am working on right now where we had a similar issue in that the Engineer insisted on sizing the VFD to the Service Factor Amps).


The issue is that "Service Factor" is a mainly thermal issue in that the motor is designed to handle 115% more heat than a full rated load would create in the motor, so if you overload it, it will be able to handle that. Years ago NEMA specs used to say "temporarily", but they took that out at some point and all it says now is some weasel words to the effect of "running a motor into the Service factor may result in decreased motor performance and life". But running a motor on a VFD (inverter drive) creates additional heating in the motor too, so the 1.15SF motor can handle EITHER an overload of 115%, OR running on an inverter, but not both. Apparently some OEMs and MEs are unaware of this fact and it can bite you if you are the one supplying the VFD.

 

[wwhitney]

So why does that name plate say 681A (1.115 SF) in the top section, but 715.1A (1.0 SF) in the "inverter duty" section?

715.1A is 1.05 * 681A, which sounds like the Inverter Duty is using up 0.10 of the SF, leaving a 1.05 SF. But the manufacturer has chosen to label it at 715.1A /1.0 SF rather than 681A / 1.05 SF.

Cheers, Wayne

P.S. If I understand correctly, raw VFD output is a very non-smooth voltage curve, and the VFD relies on the inductance (?) of the motor to smooth the output. Would putting some smoothing equipment between the VFD and the motor (load reactor?) reduce the inverter-related heating in the motor windings?

 

[synchro]

So why does that name plate say 681A (1.115 SF) in the top section, but 715.1A (1.0 SF) in the "inverter duty" section?

715.1A is 1.05 * 681A, which sounds like the Inverter Duty is using up 0.10 of the SF, leaving a 1.05 SF. But the manufacturer has chosen to label it at 715.1A /1.0 SF rather than 681A / 1.05 SF.

Cheers, Wayne

P.S. If I understand correctly, raw VFD output is a very non-smooth voltage curve, and the VFD relies on the inductance (?) of the motor to smooth the output. Would putting some smoothing equipment between the VFD and the motor (load reactor?) reduce the inverter-related heating in the motor windings?

The PWM output voltage waveform of the IGBT switching devices has to maintain fast edges or there would be significant power dissipation when the IGBTs are conducting. And so if you add a filter at the VFD output it would need to have inductors at its input with a high enough inductance to allow this swing. I suspect that the core and windings of this inductance might have comparable heating to the inverter-related heating of the motor. And so you might reduce heating in the motor but it would appear in the filter instead, and the filter adds extra costs. I don't have quantitative comparisons but these are inputs based on some technical judgement and experience.

 

[wwhitney]

I suspect that the core and windings of this inductance might have comparable heating to the inverter-related heating of the motor. And so you might reduce heating in the motor but it would appear in the filter instead, and the filter adds extra costs.

Right, for a new installation you'd just size the motor properly given the use. But for an existing installation of motor and VFD, if you ran into the SF issue described in the OP, I was wondering if you could add a properly sized VFD output filter to reduce the motor winding heating, and to what extent. Could you get all the SF back, or maybe just part of it?

Cheers, Wayne

 

[synchro]

Right, for a new installation you'd just size the motor properly given the use. But for an existing installation of motor and VFD, if you ran into the SF issue described in the OP, I was wondering if you could add a properly sized VFD output filter to reduce the motor winding heating, and to what extent. Could you get all the SF back, or maybe just part of it?

Good question. It would take some analysis with models of the filter and motor to know for sure. The added filter would have an impedance that might limit the peak current and therefore torque available from the motor, but whether that would be enough to make any significant difference would have to be analyzed as well. Winnie (Jon) would know more about this than I do.

 

[Jraef]

Right, for a new installation you'd just size the motor properly given the use. But for an existing installation of motor and VFD, if you ran into the SF issue described in the OP, I was wondering if you could add a properly sized VFD output filter to reduce the motor winding heating, and to what extent. Could you get all the SF back, or maybe just part of it?

Cheers, Wayne

From an engineering standpoint, you may be correct. People who make DV/DT filters tout how they help the motor run cooler but yes, you just move the losses into the filter, plus it can lower the output capability of the VFD by lowering the voltage getting to the motor. That's minor though, DV/DT filters are a good thing when you need them sometimes. I rarely let that get in the way.

But from the motor mfr's standpoint, unless you built the filter into the motor, you cannot make a claim on the motor's capacity based on what the user may or may not do or add to the circuit.

In the case of the one in the image, the SF still shows as 1.0, so you cannot use any overloading. I assumed the added current was in losses.

 

[paulengr]

Just a cautionary tale.

I have a situation right now where we put in some VFDs on pumps, even AFTER double checking the motor nameplate FLAs, and are now being told that the pump OEM was intending on running the motors into the Service Factor continuously, so they are saying we sized the VFDs incorrectly (we were not told that).

Just so everyone knows, when you run a motor with a 1.15 Service Factor from a VFD, the motor NO LONGER HAS A SERVICE FACTOR! Most "inverter duty" motors state that right on the motor nameplate (they did in this case) by showing that if run from an inverter, the Service Factor is 1.0. Non inverter duty motors might not state that on the nameplate since the motor mfr will tell you they are not meant to be run on an inverter drive, but if you ask, they will tell you, no service factor.

Here's an example (not from this project, but another one I am working on right now where we had a similar issue in that the Engineer insisted on sizing the VFD to the Service Factor Amps).
View attachment 2554135

The issue is that "Service Factor" is a mainly thermal issue in that the motor is designed to handle 115% more heat than a full rated load would create in the motor, so if you overload it, it will be able to handle that. Years ago NEMA specs used to say "temporarily", but they took that out at some point and all it says now is some weasel words to the effect of "running a motor into the Service factor may result in decreased motor performance and life". But running a motor on a VFD (inverter drive) creates additional heating in the motor too, so the 1.15SF motor can handle EITHER an overload of 115%, OR running on an inverter, but not both. Apparently some OEMs and MEs are unaware of this fact and it can bite you if you are the one supplying the VFD.

The VFD is required as if a few years ago to provide the overload relay function. You provide the FLA and service factor. In sizing charts for overloads in starters the same thing is .done. If you want a bigger motor, buy it. A compressor manufacturer is looking at duty cycles (load/unload) and they frequently provide goofy nonstandard service factors on their equipment and run “into the service factor” during load cycles. They also load test the system so performance is designed in. A pump manufacturer on the other hand has absolutely no business doing this. Pump curves are not that precise and there are too many variables such as process temperature that cannot tolerate such a tight margin. Fire them. This is their screw up and you can get enough data from the VFD to prove it.

If you stop looking for excuses and read the name plate more closely you will also see it has an allowable turn down ratio. NEMA MG-1 provides charts showing a Part 30 (non-inverter duty) motor with 1.15 service factor can be turned down to around 50% of design frequency as a minimum in a constant torque condition. On a PD pump this applies but in a centrifugal pump it can be turned down to typically 10% before the lack of integral fan cooling and the inherent fixed flux of the motor become a thermal issue.

As to “magic waves” from a VFD it is a rectangular waveform that is designed to operate a motor with the same performance characteristics as specified by the motor manufacturer. The fact that the name plate shows a 1.0 power factor included an increase in FLA (hence more upper end torque margin) and a turndown ratio. The motor manufacturer is allowing for a greater turndown than just 2:1 as per Part 30.

 

[paulengr]

So why does that name plate say 681A (1.115 SF) in the top section, but 715.1A (1.0 SF) in the "inverter duty" section?

715.1A is 1.05 * 681A, which sounds like the Inverter Duty is using up 0.10 of the SF, leaving a 1.05 SF. But the manufacturer has chosen to label it at 715.1A /1.0 SF rather than 681A / 1.05 SF.

Cheers, Wayne

P.S. If I understand correctly, raw VFD output is a very non-smooth voltage curve, and the VFD relies on the inductance (?) of the motor to smooth the output. Would putting some smoothing equipment between the VFD and the motor (load reactor?) reduce the inverter-related heating in the motor windings?

In a way. A 3% load reactor reduces the output by 3.%. So a 100 HP motor and VFD is reduced to only 97 HP. To compensate a DV/dt filter uses RC elements to trim high frequency components (smooth the edges) to reduce reflected waves but at an expense of less than 1% losses. A sinus wave (LC tuned) filter again introduces a 3-5% loss but makes the output almost a pure sign with a cutoff frequency down around 100-200 Hz with a big roll off to allow cable lengths of thousands of feet without reflected wave issues.

The VFD is designed for an impedance on the output of around 100 ohms which matches the cable system characteristic impedance. Motors vary widely based on HP from around 1000 ohms on small ones down to around 100 ohms at 200 HP. This is referring to surge impedance not DC or 50-60 Hz AC. At short cable lengths 50 feet or less the mismatch does not cause the voltage seen at the motor terminals to exceed minimum surge impedance ratings (Part 30). As line lengths increase beyond that point the impedance mismatch becomes an issue and we get reflections or “ringing”. It is blocked by the motor inductance (hence reflects. It is absorbed by the cable, NOT the motor. Eventually reflected waves exceed motor insulation and we get a discharge (arc) which leads to reduced life, sometimes hours. In the motor it will be turn to turn shorts in the first couple windings. This is all basic transmission line theory. The magic waves cannot enter the motor because the inductance as stated blocks it but it is not filtered but reflected.

Basic transmission line theory is a wave travels down a conductor unaffected. At an impedance change (motor terminals) some voltage passes through while some reflects backwards. The impedance ratio determines how much is reflected. The VFD must match the cable or the reflection would curtail output. So at worst it is 200% of the input wave (100% reflected) which we approach at around 200-250 feet. Thus inverter duty motors look like a 100% solution. But with each pulse the second pulse can hit before the first reflected wave has died out giving rise to 300%+ reflected wave peaks at still longer lengths. Which is why a filter is the best solution. An RC matching network can also be used at the motor end. GE patented it but their patent ran out last year so now it’s patent free.

 

[paulengr]

From an engineering standpoint, you may be correct. People who make DV/DT filters tout how they help the motor run cooler but yes, you just move the losses into the filter, plus it can lower the output capability of the VFD by lowering the voltage getting to the motor. That's minor though, DV/DT filters are a good thing when you need them sometimes. I rarely let that get in the way.

A dv/dt filter is an LC filter with a cutoff frequency located above the VFD switching frequency. It has no appreciable impact on HVF. This claim is puffer speech. While technically true the infinitesimal improvement is not measurable. Plus the whole point of dv/dt is keeping losses minor (under 1%) compared to 3-5% for load reactors and sinus filters. You can’t achieve under 1% losses yet somehow gain it all back at the motor. That would be some perpetual energy scheme.

But from the motor mfr's standpoint, unless you built the filter into the motor, you cannot make a claim on the motor's capacity based on what the user may or may not do or add to the circuit.

Almost true. We can manipulate things like frequency, rated voltage, turndown ratio for VFDs, and service factor on the name plate for the same motor to produce different name plate ratings. Same motor but different ratings depending on the input service conditions. Motor shops provide “treating” services for unusual operating conditions such as operating a European rated motor on North American standard conditions. I get calls on this once in a while from a pump vendor that is always coming up with various junk yard MV motors.

In the case of the one in the image, the SF still shows as 1.0, so you cannot use any overloading. I assumed the added current was in losses.

It is quite unusual. It is not unusual to see reduced numbers with larger turndowns but increased current is kind of unusual. HP=torque x RPM x (conversion factors). Flux is more or less constant for a given voltage so the only thing that makes sense is the manufacturer is giving you a little more torque but doesn’t explicitly state this. That manufacturer does finite element models of their motors. There is NO extra margin.

 

[paulengr]

I take it back. The top part of the name plate gives 60 HP or 281 ft-lbs. The bottom part states (variable torque” with only a 267 ft-lb rating. It should be higher.

 

[paulengr]

I take it back. The top part of the name plate gives 60 HP or 281 ft-lbs. The bottom part states (variable torque” with only a 267 ft-lb rating. It should be higher.

Also it increased to 1200 RPM so name plate is 61 HP on an inverter. There is the current increase. The rest of the 1.0 SF is absorbed into thermal limits at low speed. Integral cooling fan air flow is proportional to the square of RPM (fan affinity law). At low speed flux (10-15% of FLA) is not changing so we run into a thermal issue at slow speeds.

Do not assume playing with SF is an inverter duty trick. It is used extensively. “World duty” motors give multiple frequency/voltage/service factor ratings. Compressor manufacturers compensate for load/unload cycles with corrected overload ratings using crazy nonstandard service factors.

 

[kwired]

What I kind of see often is either manufacturers or equipment providers are sizing things marginally to begin with. Often they get away with it working long enough they no longer need to honor any warranty once it does fail. This is a general observation and not just to do with VFD's.

Last summer I had a farmer with a unit that included a tank, small pump, and an agitator motor, plus 480 x 120 transformer to derive the 120 needed for those motors. This unit was for injecting fungicide into irrigation water. The pump wasn't really a motor but more of a solenoid design that operated a piston to do the pumping every so often (I think not really sure), but timing was adjustable. Regardless, I figured the load on this was able to be up to 7 amps, but the transformer was only 250 VA (a little over 2 amp). That transformer was burned out. I replaced it with at least 750, maybe 1000 VA, don't recall anymore. He had this for a few years - so likely no warranty had he even tried to pursue one. I think they know it will typically last long enough warranty is no longer an issue.

 

[Jraef]

Yeah, the days of over designing everything by 20% are long gone. They say it’s due to energy efficiency but there is also an element of “just enough to make it past the warranty” going on.

 

[paulengr]

What I kind of see often is either manufacturers or equipment providers are sizing things marginally to begin with. Often they get away with it working long enough they no longer need to honor any warranty once it does fail. This is a general observation and not just to do with VFD's.

Last summer I had a farmer with a unit that included a tank, small pump, and an agitator motor, plus 480 x 120 transformer to derive the 120 needed for those motors. This unit was for injecting fungicide into irrigation water. The pump wasn't really a motor but more of a solenoid design that operated a piston to do the pumping every so often (I think not really sure), but timing was adjustable. Regardless, I figured the load on this was able to be up to 7 amps, but the transformer was only 250 VA (a little over 2 amp). That transformer was burned out. I replaced it with at least 750, maybe 1000 VA, don't recall anymore. He had this for a few years - so likely no warranty had he even tried to pursue one. I think they know it will typically last long enough warranty is no longer an issue.

That’s an LMI type design. Very common.

The challenge is that in order to improve efficiency the amount of material, impedances, fan efficiency, etc. all need to be improved. Years ago we designed equipment by basic design principles then tested it to verify that it meets the minimum requirements and tweaking as needed. Now motors at least are designed entirely on a computer and tested just once. There is no margin at all.

With regards to pumps although computer design (FLUENT) is available for computer design generally 20-25% margin is recommended. At the upper end of a pump curve for instance considering power increases with the cube of speed, going up 10% in speed is a 33%!increase in power. Even small percentage increases in impeller diameter are a disaster. Hence significant margin is always recommended with centrifugal pumps.