Power Factor Correction on Highly Variable Loads
Power Factor Correction on Highly Variable Loads
(OP)
Prologue
Our facility uses a 10 MVA, 69kV:12470V transformer to feed a mixture of 12470:480 transformers (mostly delta-wye, but with one 3750 KVA delta-delta, and a 2500 KVA wye-wye). The utility doesn't charge for low power factor, but we're planning several process line additions, and must either significantly improve power factor (nominally 78 to 80%) to free up capacity, or do extensive work to increase KVA.
One complicating factor is, while we have a number of high power across-the-line, and delta-wye started motors a significant portion of our load are DC motors ranging from 250 HP to 700 HP driven from S-6 bridges. Newer machines - about 1200 HP worth per nameplate - are using AC motors driven with ABB ACS800 drives, but, so far, none were built using the -ULH ultra low harmonic option. Our newer AC and DC drive installations are equipped with 3% line reactors or isolation transformers, but those on older installations are connected directly to the bus.
One proposed solution is to install controlled capacitor banks in the 12470 transformer yards to operate at nearly unity. While this would free up KVA and eliminate the need to re-work the 10 MVA transformer and lines to it I'm feeling incredibly queasy at the prospect of potential harmonics and resonance issues.
I'm thinking a better scheme is to install properly sized capacitor banks at larger (30 HP and up) across-the-line and delta-wye started motors, install line reactors on DC drives that aren't already so equipped (eventually we'll be phasing these out in favor of AC drives), and to specify ABB ACS800 drives with the -ULH option for new installations.
I've read previous Eng-Tips threads, but haven't yet run across one on this specific question ...
In several cases - process pumps, cooling tower fans, and air compressors (where we use a single VSD compressor for trim control, and our sequenced fixed capacity compressor motors run fully loaded most of the time)KVAR sizing isn't very difficult.
However, each of our process lines use a 75 HP or 100 HP reclaim granulator, and on these the load profile changes dramatically. During normal process operations they run lightly loaded, but must be sized to allow for periods of heavy scrap generation.
It seems to me using a controlled PFC bank for these would be problematic (in addition to costly) due to the very 'peaky' load profile. If I size to correct PF at nominal loading then it'll over-correct under heavier loads.
I'm leaning towards KVAR sizing to achieve 95% PF from whatever the power factor is at peak recorded loads (in this case, from 85% to 95%), and am interested in any opinions pro or con, and/or if there are rules-of-thumb when sizing KVAR caps for highly variable loads.
Trending for a typical installation is attached.

Our facility uses a 10 MVA, 69kV:12470V transformer to feed a mixture of 12470:480 transformers (mostly delta-wye, but with one 3750 KVA delta-delta, and a 2500 KVA wye-wye). The utility doesn't charge for low power factor, but we're planning several process line additions, and must either significantly improve power factor (nominally 78 to 80%) to free up capacity, or do extensive work to increase KVA.
One complicating factor is, while we have a number of high power across-the-line, and delta-wye started motors a significant portion of our load are DC motors ranging from 250 HP to 700 HP driven from S-6 bridges. Newer machines - about 1200 HP worth per nameplate - are using AC motors driven with ABB ACS800 drives, but, so far, none were built using the -ULH ultra low harmonic option. Our newer AC and DC drive installations are equipped with 3% line reactors or isolation transformers, but those on older installations are connected directly to the bus.
One proposed solution is to install controlled capacitor banks in the 12470 transformer yards to operate at nearly unity. While this would free up KVA and eliminate the need to re-work the 10 MVA transformer and lines to it I'm feeling incredibly queasy at the prospect of potential harmonics and resonance issues.
I'm thinking a better scheme is to install properly sized capacitor banks at larger (30 HP and up) across-the-line and delta-wye started motors, install line reactors on DC drives that aren't already so equipped (eventually we'll be phasing these out in favor of AC drives), and to specify ABB ACS800 drives with the -ULH option for new installations.
I've read previous Eng-Tips threads, but haven't yet run across one on this specific question ...
In several cases - process pumps, cooling tower fans, and air compressors (where we use a single VSD compressor for trim control, and our sequenced fixed capacity compressor motors run fully loaded most of the time)KVAR sizing isn't very difficult.
However, each of our process lines use a 75 HP or 100 HP reclaim granulator, and on these the load profile changes dramatically. During normal process operations they run lightly loaded, but must be sized to allow for periods of heavy scrap generation.
It seems to me using a controlled PFC bank for these would be problematic (in addition to costly) due to the very 'peaky' load profile. If I size to correct PF at nominal loading then it'll over-correct under heavier loads.
I'm leaning towards KVAR sizing to achieve 95% PF from whatever the power factor is at peak recorded loads (in this case, from 85% to 95%), and am interested in any opinions pro or con, and/or if there are rules-of-thumb when sizing KVAR caps for highly variable loads.
Trending for a typical installation is attached.






RE: Power Factor Correction on Highly Variable Loads
as your question refers to a quite complex situation, it will be very difficult to disucss it in such a forum. Nevertheless, let me try to provide at least an initial analysis.
- As you operate all systems of drives in you system you are faced with all kind of non-active power
- DOL induction motors: fundamental reactive power, varying with load
- DC-drives: fundamental reactive power, varying with speed and load
- AC-drives: harmonics, varying with load
It is important to note, that methods that work well for systmes suffering from only one of these issues will not work well in your system:
- Connecting capacitor banks to motors as it has been a neat approach in the past, will not work as these capacitors will suck in harmonics from DC- and AC Drives nearby
- Also variable capacitor banks added somewhere in the system will need series inductors to protect them from harmonics
The suggestion to add a variable capacitor/filter bank at the MV level therefore sounds reasonable to me. A properly designed system will consist of several branches consisting of capacitors and inductors (and some resistors) to provide a high pass-filter function. Be aware that there must be always a trap for 5., 7. 11., 13. harmonics connected in the system to avoid resonances.
However, I'm aware that this approach is quite costly, as several MV breakes are needed. Unless the upstream system has also harmonics or PF limitations, upgrading the transformer and the cables might be a more attractive solution (as this will also provide energy savings and headroom for further extensions).
RE: Power Factor Correction on Highly Variable Loads
Do not try to analyse the system by using some "integral" numbers like PF or THD. You need to assess the topics by the value of the fundamental reactive current and the amplitudes of the individual harmonics.
RE: Power Factor Correction on Highly Variable Loads
I get the impression that your load mixture is well above the 15 - 20 percent non-linear load that one usually thinks is safe to avoid resonance between leakage inductance and PF capacitors. And, given that the load seems to be all over the place most of the day, it will be very difficult to avoid resonance kicking in occasionally.
SVC is what one looks to in cases like these. You may be in the low range for SVC, but I think that there are SVC Light versions that could be worth while looking at.
Gunnar Englund
www.gke.org
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
RE: Power Factor Correction on Highly Variable Loads
In the old days before drives became ubiquitous and power factor correction was at times more of an art than a science there were a few tricks that we used. Depending on the distribution layout of your plant and the location of your harmonic mitigation equipment you may find this helpful.
(This discussion applies only to your conventional induction motors. Drives, both AC and DC are a separate issue.)
Consider a large motor on the same feeder as a small motor.
The large motor has local capacitors to correct the power factor to unity. The small motor is uncorrected.
The power factor of the feeder will be based on the combined kW of both motors and the KVARs of the smaller motor. The KVARs of the smaller motor will have little effect on the feeder PF and the feeder PF will stay close to unity.
When practicing the "Art" of PF correction, we would correct the larger motors to unity and then evaluate the effect on the plant overall PF. These correction capacitors would be connected directly to the individual motors so as to be switched by the motor starting contactor.
If we needed more KVARHrs to bring the plant PF up. We would look at motor ratios. Specifically the ratio of large, PF corrected motors to smaller under corrected motors. We would look at the total KVARs required by the uncorrected motors and consider over correcting the larger motors to supply these KVARs. Once our correction on a motor exceeded the amount required for correction to unity we connected the capacitors directly to the contactor, bypassing the overload relays. The limit that we used was not more than twice the capacity required to correct the motor to unity.
Which motors to correct? The most effective as defined by:
"KVAR correction times running hours per month". For example in a lumber mill, dry kiln fans are a prime target. There are larger motors but the dry kiln typically operates 24/7 while the rest of the mill may only operate one or two shifts.
A large motor with a number of smaller motors that are related to the operation is a good candidate for 200% correction.
A motor which is somewhat independent or other motors such as a compressor may be over corrected but the over correction may be more conservative.
What happens when a large, over corrected motor is run for maintenance when no other motors are running? Even when a plant is idle, there may be a lot of equipment connected which will use KVARs. The transformers are a good example.
Even if the plant PF goes leading it may not be by much and most utilities can use a few extra KVARs capacitive, particularly a utility which doesn't charge PF penalties.
What happens when the plant is operating and a large corrected motor goes down?
The point of this exercise is to free up service capacity. When a large motor goes down, that will free up more capacity than you will lose by the loss of PF correction.
When practicing the art of PF correction, you must often step back. look at the big picture and ask yourself;
"What if...?" If the art of PF correction saves you the expense of a PF controller it may be worth considering.
Keep in mind that KVARs are voltage dependent.
Bill
--------------------
"Why not the best?"
Jimmy Carter
RE: Power Factor Correction on Highly Variable Loads
I also have the experience of installing and handling individual caps at motors in industrial plants. Both have pros and cons. I am not trying to design it remotely. However, If the primary objective is to free up capacity of 10 MVA xfmr, my initial hunch would be to go with PFCC at the transformer. Easy to control, monitor and maintain. harmonics are not difficult to mitigate. In fact that gets more difficult with distributed ones. Peaky load profile at individuals lines may not be that peaky at the service level. Even a combination of the two approaches would work.
You may want to do a thorough study and then decide.
Rafiq Bulsara
http://www.srengineersct.com
RE: Power Factor Correction on Highly Variable Loads
2 - You didn't allow some margin to prevent over-excitation from residual voltage at lowering frequency (increasing Xc) following switching off of motor/cap? (That would seem contrary to traditional recommendations.)
=====================================
(2B)+(2B)' ?
RE: Power Factor Correction on Highly Variable Loads
=====================================
(2B)+(2B)' ?
RE: Power Factor Correction on Highly Variable Loads
I have several times taken the PF of a motor from x% lagging to the same % leading and never had a problem. Just lucky I guess. Some of the motors usually were unloaded when they were normally stopped, some were loaded (propeller fans).
Bill
--------------------
"Why not the best?"
Jimmy Carter
RE: Power Factor Correction on Highly Variable Loads
thread237-262325: No load current on a motor
From everything I have read, the standard approach should provide a margin against resonant condition for switched-capacitors during coastdown. To do that, we should undercorrect based on a conservative low estimate of vars (i.e. the no-load vars)
http://www
At the above link is Siemens' recommendation:
This is the first I have heard anyone suggest to correct based on full-load vars, ignore the 90%, and perhaps even overcorrect (that's what I understood you to say).
Perhaps you can comment whether you agree you are using a non-standard approach and whether there are certain application factors that would be relevant for success of this strategy (I would think it is particularly risky for a high-inertia, low torque load that takes a long time to coast down, and for medium voltage vs low voltage motors).
=====================================
(2B)+(2B)' ?
RE: Power Factor Correction on Highly Variable Loads
=====================================
(2B)+(2B)' ?
RE: Power Factor Correction on Highly Variable Loads
After reviewing our utility's analysis and recommendation see I've incorrectly assumed MV capacitors to provide an additional 9.6 MVAR would be located in our transformer yards. Our layout is currently a short underground run from their 10 MVA yard to a getaway pole, the next pole fitted with 1200 KVAR of switched caps, and final two terminal poles have isolation switches for underground runs to each of our two transformer yards. This run is done with 336AL. Their proposal is to put up another 8 poles each fitted with 1200 KVAR of switched capacitors to yield 10.8 MVAR total.
The existing 1.2 MVAR is switched from a security-tagged enclosure near the bottom of the pole, but I don't know the details of the control scheme employed. This pole is fitted with 6 capacitors and 3 oil switches connected to the line through a set of fused isolating switches, and are shunted to ground with what appear to be MOVs, but I don't see any series reactors.
SVC Light technology sounds excellent in that it would address our immediate concern of PF correction to free up capacity as well as providing harmonics mitigation, and I'll put it up for discussion. The smallest one I saw in ABB's literature was rated 0 to 30 MVAR, and would be surprised if installing such a system would cost less than the alternative plan of installing fans on the 10 MVA transformer, replacing the 560A OCR with a higher-rated one, and running a parallel pole line (since the 336AL wire run would otherwise be overloaded).
It just so happens we recently installed 100 KVAR of capacitors on one of our 300 HP air compressors to reduce line current. Uncorrected, one phase was consistently into the 395 to 400 amp range, and taking out the 400A fuse on that pole. I set up a power logger (at the main disconnect, so some of this load is the cooling fans) to check before and after performance. Prior to installation power factor was 84.5% at 398 line amps when the air end was loaded, and 51% PF and 182 amps unloaded, and with capacitors connected loaded PF was 98.5% at 343 line amps, and 90.6% at 110 line amps.
This compressor is fed from the wye-wye 2500 KVA transformer, and it's bus serves fewer non-linear loads than the others - primarily, a 200 HP VFD via a 3% reactor (plugged into the bus nearly adjacent to this compressor), and one of the older 400 HP DC drives installed without a line reactor, and approximately 150 feet away.
I'm thinking about connecting a logger to record compressor capacitor current, and harmonics signature, then slowly ramping the VSD-driven air compressor through it's speed range. Would my thinking be correct in expecting to see unusually high capacitor currents and harmonics at certain VSD speeds if this particular load has a tendency towards creating resonant conditions?
Another question is prompted by electricpete's Siemens TechTip #20 link cautioning against capacitors connected across the motor when open transition reduced voltage starting is employed. The aforementioned air compressor is open-transition, there isn't room in the enclosure to add resistors to make it closed transition, and I'd just as soon not have the caps always across the line.
There might be just enough room to shoe-horn in another IEC form factor 150A contactor, and would like to know if there were any downsides to sequencing on the capacitors after the transition to delta is complete? This would also make it easy (by using a TDR) to enforce the caps stay off-line for enough time to assure the motor is spun down to zero speed before attempting a restart.
RE: Power Factor Correction on Highly Variable Loads
But, in view of your information concerning leakage reactance I will be cautious the next time I do PF corrections.
Thanks for the heads up.
Bill
--------------------
"Why not the best?"
Jimmy Carter
RE: Power Factor Correction on Highly Variable Loads
In my experience, a VFD will always produce the same harmonics with the harmonic levels changing due to the load on the VFD. I'd expect the most likely time you'd see an issue is with the VFD running at full-load and full-speed. It would be interesting to see what currents the capacitors are experiencing.
You don't want to rapidly switch a motor with capacitors connected on and off. However, in this case you can connect the capacitors to the main starting contactor as long as it remains energized during the transition.
RE: Power Factor Correction on Highly Variable Loads
Perhaps I'm not thinking about this correctly, but isn't this precisely what occurs during open transition?
When first turned on the motor is wired for wye operation when the 1M and 3M (wye shorting) contactor are closed, then, some short time after the motor is at speed, the 3M contactor drops out as 2M turns on to reconfigure wiring to delta.
It's only a few milliseconds duration, but during that transition from wye to delta the motor is open circuit.
RE: Power Factor Correction on Highly Variable Loads
RE: Power Factor Correction on Highly Variable Loads
RE: Power Factor Correction on Highly Variable Loads
During the time it takes to drop out 3M wye contactor and seal 2M to complete the delta running configuration the motor acts as a generator. Closing 2M slaps it across the line again with little chance of generated voltages being in sync with the line, and provides an opportunity for high transient current (and torque) until sync is restored.
Seems to me adding KVAR caps would tend to exacerbate this problem ...
RE: Power Factor Correction on Highly Variable Loads
The issues arise when the capacitors are connected to the motor in such a way that when one contactor opens, there is a circuit through the capacitors and the motor.
Bill
--------------------
"Why not the best?"
Jimmy Carter
RE: Power Factor Correction on Highly Variable Loads
RE: Power Factor Correction on Highly Variable Loads
I can see why one wouldn't want to connect KVAR caps in an open transition arrangement (or closed transition, for that matter) at the 2M contactor load side since they'd be effectively shorted while the wye contactor is closed, and start charging through the motor windings until 2M contactor seals.
Another hazard with this wiring would be if the motor were shut off, then restarted before the caps discharged ... I'd imagine a big bang would ensue when the 3M wye contactor closed during the restart with a good chance of welding it's contact sets together, and possibly destroying the caps.
That leaves open two questions:
1). Why do Siemens (and - I've been looking around - Square D, ABB, Cutler-Hammer, and several others) caution against using KVAR capacitors on open transition starter configurations?
They recommend placing the caps at the contactor line side, and I don't see any difference (except for not requiring the MTOL be set at a lower trip setpoint, and higher line current through the contact sets) between placing them there versus at the 1M load side.
2). Why do these sources specifically mention open transition as being a problem, but not closed transition starting?
A motor equipped with closed transition starting uses a set of high power, low ohm resistors and yet another contactor to maintain motor connection throughout the transition.
I'm wondering why these different manufacturers don't mention closed transition starting as causing difficulties?
It's advantage over open transition is to provide enough motor current via these resistors to keep the line supply and rotating motor field in sync. When KVAR caps are added to the mix, and connected to 1M they would be supplying magnetizing current throughout the wye-to-delta process.
RE: Power Factor Correction on Highly Variable Loads
I suspect they mean a setup where the caps would effectively replace the closed transition resistors. In that case, the open motor leads would be shifted some amount from the line phasing. I suspect it would be close to 90 degrees out.
Also, your last paragraph doesn't tell the whole story. When the shorting contactor opens those motor leads are not in phase with the line power. The resistors pull them in phase before they are connected at full-voltage.
RE: Power Factor Correction on Highly Variable Loads
Rbulsara is correct in utilizing a PFCC at the transformer. But at lower subs, IE 480V might be another option. At least one of the manufacturers has active switching on their LV PFCC's. IE, below a setpoint, it activates a 75 or 150KVAR stage to correct the pf. If this is not enough, it activates another stage, etc. Once pf raises above another setpoint, for a predetermined amount of time, it will begin dropping out stages and monitoring the pf. This would aleviate your over-correction concerns.
RE: Power Factor Correction on Highly Variable Loads
A lps for your 16 Jan 11 11:23 excellent post.
Muthu
www.edison.co.in