load reactor temperature

[bafco]

I have a customer with a Powerflex 40 VFD, 10 HP, driving 5 1.5HP motors. We employed a single MTE load reactor between the VFD and the motors. As long as the system runs on a 4kHz carrier, we don't have a problem. Unfortunately, this customer has found the audible motor noise undesireable, and has increased the carrier to 12kHz. With the given load, that is the maximum carrier allowed according to AB's derating curve. Since this increase in carrier, this customer has smoked 2 reactors. I simulated his conditions in the lab today, and found that at 4kHz, the reactor surface temp was around 180 F. At 12kHz, rose rather quickly to almost 300 F. This customer also almost always runs his motors at 25~30 Hz.

The reactor has the following ratings:
Fundamental Amps - 18
Max Amps - 27
0.8 mH.

Is there a way to predict the required load reactor, while maintaining an equivalent impedance and level of protection as the model currently being used. I think all I need is a reactor that can handle the thermal load and being slammed with a 12kHz carrier, but, I have no idea how to size it. The reactor currently in use has a published carrier limit of 20kHz, but the melted terminal block begs to differ.

 

[Skogsgurra]

Can you say if core or winding is hot? If the core is hotter than the winding, you probably have ringing that heats the core.

I have had several cases with motor reactors and rather long motor cables where each switching has caused ringing between reactor and motor cable. You can see it if you put a "measure loop" around the core and view the induced voltage on a scope.

If you see heavy ringing with higher than the switching frequency - you got yourself an "induction oven" that heats the iron.

The core heats proportionally to ringing frequency to some exponent (eddy current losses mostly) and if you switch at 12 kHz instead of 4 kHz, you also get three times more heat from the ringing.

If you can see ringing, try to damp it out using resistors parallel to the reactors. That is what ABB does in their du/dt filters. You will probably need something around 10 ohms and 25 watts. But it is better to experiment with different ohm numbers until you find a good value.

 

[GordS]

How about the following method?

Heat generation inside the reactor should be mainly I*I*R loss and eddy current loss. You can measure I at 4 kHz and you can measure R ( remember to account for the temperature dependence of R). So at 4 kHz, you can calculate I*I*R. Knowing the surface temperature ( you measured it), you can calculate the amount of heat convected and radiated away ( search for websites that give equations for convection and radiation heat loss from surfaces). So knowing the total heat generated ( which equals the total heat convected and radiated), and knowing the I*I*R component, you can calculate the eddy current component at 4 kHz. To calculate your maximum frequency, choose some higher frequency and increase the eddy current component by the square of the frequencies. This will give you a new value for the heat generated. Go back to the heat convection and radiation equations to calculate the new surface temperature. ( Do 2 iterations every step with the second iteration adjusting the valuer of "R" based on the reactor surface temperature calculated in the first iteration. Allow about a 10C hot spot allowance)

This could likely conveniently be done with a spreadsheet .

 

[bafco]

Thanks guys. I'll give both of these a shot. It's better to try these than to purchase a Sine wave filter. The price makes me shiver a bit.

 

[DickDV]

Why not contact the reactor manufacturer. These reactors have a maximum rated carrier frequency spec and you are likely over the spec.

They may very likely be able to give you derating figures for the reactor you have. You may, for example, be able to run 4 of the 5 motors on one reactor and purchase an inexpensive small reactor for the remaining motor to acheive the derate.

Just one other thought--did you attempt to run the five motors without any reactor. Giving the sizing info you supplied, it seems to me that you just might be ok without one. In addition, I've never been able to sense any additional motor quieting from 8khz to 12khz. Especially with the reactor smoothing, I'd be surprised if the motors are noisy at all at 8khz.

12khz just might be overkill.

 

[bafco]

Thanks Dick.

Unfortunately, our multi-motor applications also rely on the reactor to reduce not only voltage overshoots at the motor, but to reduce cable charging currents as well. When we run the 460V units without the reactors, our overload relays begin to trip out due to charging currents. I've scoped a 1HP application with and without a reactor, and seen peak to peak current reductions in the neighborhood of 10 amps, so the reactor kind of does the job all the way around for us. (I wish I could post my scope screens on here.) I will also try to get away with reducing the carrier frequency a bit.