Power Distribution Woes

  • Thread starter gerald beaudoin
  • Start date

Thread Starter

gerald beaudoin

A couple of weeks ago we had a fuse blow in a mainswitch which was supplying a 3 phase 575volt distribution trough. This occurred during the weekend when there was no one around to notice the effects. All the VSD's (Variable speed drives rated at 575v) connected to this trough blew 2 of their protective input fuses which were very conservatively rated as compared to the manufacturers specs. The drives were however, all short circuited on their input ! I tried simulating the condition (without the drives connected) by removing a fuse and taking some voltage measurements with a scope. I could not see anything out of what I would have expected to see....reduced voltages and one phase missing. There is a 3phase delta transformer connected to the same trough so I tried the test with and without it connected but the results were the same. Back EMF from the transformer??? Closer examination of the situation revealed that the original blown fuse was carring an excessive load...95A on a 100A fuse. We reduced the load to the trough so the fuse will no longer be a problem.....and I have isolated the transformer loads in case of phase loss, but the question still remains.....why would a missing phase blow out the drives????? at reduced voltage????? This one cost us $6000 worth of low hpdrives.....I'm sure glad it wasnt the 75hp units we have elsewhere in the plant. Any comments would be appreciated Gerald Beaudoin
When a VSD looses one input leg, the current in remaining legs will increase in an attempt to deliver same power required by the load. If the internal fuses are not sized correctly, the corresponding diodes can blow. However, the diodes in the leg that was lost should be ok. Another possibility is the electrolytic capacitor(s)failing due to higher ripple current with single-phasing condition. This would then take out the diodes, except the lost leg. Another possibility, that could take out all the diodes, could be a voltage spike/surge caused "somehow" by the lost source leg. But, most VSDs also have MOV's on their input legs. Interesting. I suggest you have the VSD manufacturer do a root cause analysis based on what exactly failed/and how in the input circuits.
Input fuses on the drive will stress when a line voltage unbalance exists. You do not need to lose one phase merely have an unbalance. The high lines will try to supply most of the current needed to power the motor. The diodes in the high phase lines will also be stressed but usually not to their limit. When the stressed fuse opens the remaining lines will draw excessive current and could damage the remaining fuses. Although a line loss will result in the remaining line to try to support the total current load, do not overlook input voltage imbalance as that can also lead to fuse failure.

Alfonso Padilla

As said in other comments, one missing phase will make the drive to demand more current on the remaining ones, thus incresing current on the other two phases fuses (and diodes). I once experienced a problem like yours. We found the fuses protecting the drives though being the right current rating, had a slow response curve, We changed to semiconductor fuses, which have a simitar I2t characteristic as input diodes or SCRs. We also found some trouble with the MOVs, so we changed them to the next available voltage level and higher Joule rating. No problems since then. Alfonso Padilla QPS Control & Automation [email protected]

Colvin, Chuck

Most VSD will shutdown on an undervoltage or phase loss. I would not count on the 95A load being the problem. Sounds more like a surge or incoming spike that happened so fast the fuses did not blow soon enough to protect the drives. Or possible you could have some harmonic currents. Carlton Colvin Electrical Project Engineer E-mail [email protected] Klockner Pentaplast of America 3585 Klockner Road Gordonsville, VA 22942 1(540)832-3600 Ext 636 1(540)832-1583 Fax 1(804)242-9780 Cell

Anthony Kerstens

1. How do you know it was a short and not an overload? You could open the fuse, look for the link, and find it in two pieces if it was an overload. If it was a short, you would find very little of it, if anything. 2. If you take one fuse out, that leaves you with a single phase powering the drive. And one power transistor (or SCR) doing the job of three. I would be asking why the drive controller didn't do it's job in detecting the single-phase and fault out before being damaged. Although given multiple failures here, I might suspect your drives don't detect faults, or are configured not to do anything on the fault. In either case, you should be opening the fuses to determine whether it was a short or an overload, and checking into your drives' fault detection capabilities. Anthony Kerstens P.Eng.
I worked for a VSD etc. manufacturer for ten years until 99. I never saw low-med HP VSDs with phase loss protection. Also, a phase loss will not cause and undervoltage. The 3 phase rectifier/capacitor section will simply become single-phase condition. The cap bus voltage will be the same but with higher ripple. In my previous e-mail, I also noted the current will increase in an attempt to deliver the same power.

Robert R. Stephens Pennzoil Products

<< 2. If you take one fuse out, that leaves you with a single phase powering the drive. And one power transistor (or SCR) doing the job of three. I would be asking why the drive controller didn't do it's job in detecting the single-phase and fault out before being damaged. Although given multiple failures here, I might suspect your drives don't detect faults, or are configured not to do anything on the fault. >> At this level of fault the SCR or transistors are faster than the fuses, generally speaking. That's why they will get damaged before the fuse opens. That is also why solid state fuses were developed but the truth is their is a race between the SCR and the solid state fuse as to which will sense the fault first. My experience has been that it's about 50/50.
The typical input circuitry on an AC VF drive is a *diode* bridge. I have never found one of these drives to have input phase loss detection. In fact, some designs do a line / bus compare to figure out when to fire the regen circuit and when operated on the "wrong" two phases, the regen circuit will cook the resistor bank. As far as protecting the input bridge . . . I have rarely found a fuse that is cost effective for use across a large number of drives given the probability of input circuitry failure due to large fault currents. Suitable high speed fuses are expensive, they are comparatively large, and they sometimes negate the UL rating of the drive. Ken Brown Applied Motion Systems, Inc. http://www.kinemation.com
Responding to G. Beaudin's plight: You experienced the "classic" case of single-phase operation on a bank of three-phase electrical loads, all supplied from the same feeder. Each drive (as seen from the protective fuses downstream from the trough) is considered as an "equivalent" three-phase load. Then, upon loss of supply to one trough phase the remaining two phases will "draw" 1.732 times the prefault load current. The increased current then destroys the semiconductor device junction. It is my conclusion that the fuses then operated to "clear" the resultant short-circuit of the semiconductor. Note that the current increase will be the same magnitude regardless of the equivalent "electrical" configuration of the load, i.e., delta-connected or wye-connected. As "food for thought" following are several points to consider: 1) Proof of this phenomena will be revealed by the fact that the same two phases are involved in each of the drives. 2) Even if equipped with single-phase protection circuits, I doubt that such devices could detect (5 to 8 ms) and then operate the contactor (an additional 10 to 15 ms) quickly enough to protect against semiconductor damage (probably less than one-half cycle). 3) The primary purpose of the main upstream trough protective fuse is to protect the trough conductors. Its secondary function is to serve as backup if the individual drive's protective fuse fails to clear an electrical fault. While it will protect a conductor it will never provide adequate protection for the semiconductor, simply because its nominal capacity is several times that of the individual drive's fuse. 4) Make sure that the drive's fuses are sized based on the thermal withstand capability (I^2xt) of the semiconductor devices, and not on their rated "full load amperes". Also consider current-limiting type. 5) To insure against future single-phasing, substitute a circuit breaker for the main trough fuses. 6) While possible, it is highly improbable, that a power surge is responsible for what has been described. 7) The scenario described above ignores the voltage restoring effects of large motors that may be connected to the same 575 Volt supply. Contact me if additional information or detail is required. Regards, Phil Corso, PE Trip-A-Larm Corp (Deerfield Beach, FL)
Further to my earlier response on the subject: 8) It is unlikely that operation of both intact-phase fuses was due to overload. Fuse clearing-time for an overload of the magnitude noted earlier (1.73 x ) is in the order of seconds. Furthermore, the "band" related to clearing time would preclude coincident "fuse" operation. Regards, Phil Corso, PE Trip-A-Larm Corp (Deerfield Beach, FL)
Some additional tips/comments regarding seemingly "unrelated" or "head scratching" types of electrical failures: The third most costly (and most common) type of electrical failure involves premature operation of low-voltage fuses. This can occur in either the distribution feeder (G. Beaudin's case), or the motor-starter. Have you ever experienced motor failure following loss of one fuse in its starter? Even though the starter was equipped with "overload" detection? The second most costly (and more rare) type of electrical failure involves premature operation of fuses in the primary of distribution transformers. This can also lead to whole-sale destruction of low voltage equipment. Now, for the most costly (and not as rare as is thought) type of electrical failure. It is the "arcing ground-fault". This phenomenon often leads to massive melt-down and fire in low-voltage (even 208 V) distribution systems. Utility service equipment, bus-duct, distribution panels, and very large MCC's can be obliterated in seconds... yes, seconds. Analogous to colon cancer, arcing ground-fault failures are preventable. The usual cause: inadequate attention paid to ground-fault protection strategy. Interested in more? Regards, Phil Corso, PE Trip-A-Larm Corp (Deerfield Beach, FL)
The Physics of... Power Semiconductor Protection Additional information related to the recent failure described by Gerald Beaudoin has prompted me to provide this update. The drive was not running. In other words, only the AC to DC rectifier section of the VSD was energized. Essentially no-load operation! Under this conditions the fuse provides little protection to its semiconductor. During no-load operation, current flow duration is a small part of both the positive and negative half-cycles. Thus, even though the semiconductor is subject to the destructive potential of the instantaneous value of current, the fuse isn't. It responds to only the rms value. This value is very low at no-load. Probably less than 10% of the nominal ampacity of the fuse. Thus, the fuse never "saw" the increase in current associated with phase-loss. Regards, Phil Corso, PE (Boca Raton, FL)
Tom, There is a possibility that you have a problem within your power distribution system, no information given as far as the the main feeder is it a grounded Y or ungrounded delta? Need to put hook a power distribution analyser for a period of time. Good luck. Amr Elaguizy
Phil, I have seen plants with ungrounded (open wye?) transformers on the order of 2000 kVA with a few large single motor loads (500 HP Compressors) and numerous smaller motor / drive loads. One interesting phenomenon is when one of the compressor phases shorts to ground during inrush of starting. One casualty that I have seen repeatedly during this fault condition is that newer AC Drives that comply with the latest UL rules often have an MOV network on the input with a triad of MOVs between each of the phases. The latest UL rules also require an MOV between this triad and ground. That particular MOV is always vaporized in these situations leading to catastrophic failure of the drive due to collateral damage to surrounding circuitry. I have seen entire motor rooms blow up all drives simultaneously during these types of faults. The customer opted to clip the ground MOV's out of all new replacement drives . . . no failures since that time. I would be interested in your comments on this scenario. Ken Brown
In response to Gerald Beaudoin's comment about ground fault detection (GFD) on plant's service: Tripping of the service GFD should only occur for two "ifs": a) If the fault occurs between the service and a downstream branch circuit or feeder GFD. b) If it provides backup for faults that the downstrean GFD responds to, but fails to clear. Regards, Phil Corso, PE (Boca Raton, FL)
I re-read Gerald's original message and don't see anywhere where he said not running/no-load etc. He did say no one was around when it happened. Did you get more accurate details?
If I understand this correctly, you have an ungrounded system which has a MOV connected to ground. In an ungrounded system, if you have a first ground fault nothing is suppose to happen(advantage of an ungrounded system), however, if you have a second ground fault, you have a phase to phase fault. This is what sounds like is happening. The MOV is providing a ground path combined with the compressor ground fault causes a phase to phase fault vaporizing the MOV et al. Bill Mostia ======================================================= William(Bill) L. Mostia, Jr. PE Independent I &E Consultant WLM Engineering Co. P.O. Box 1129 Kemah, TX 77565 [email protected] 281-334-3169
In response to Bill Mostia's Tue, Mar 13, 10:32am comments on the subject: Your scenario describing a double line to ground (LLG) fault is based on two assumptions: a) ground capacitance is zero; and b) the MOV''s are triggered into conduction at the phase to phase voltage value. Consider, as an example, a 480 Volt, ungrounded, distribution system. If a) is correct, then phase A, B, and C voltages are all 277 Volts with respect to ground. The system is considered as balanced. Now, if phase A becomes grounded, then phase B and C voltages will rise to, but no more than, 480 Volts with respect to ground. It is unlikely that the MOV will begin to conduct causing the LLG fault. The fact that the MOV's were destroyed is evidence that the voltage they were exposed to, was in excess of their rated trigger value. Hence, confirming the existence of an appreciable ground capacitance. BTW, not to unnecessarily frighten those of you using ungrounded distribution systems, this overvoltage phenomena is more likely to occur in medium to large LV distribution systems. Regards, Phil Corso, PE (Boca Raton, FL)
I don't claim to be an expert but the description of the failure was that when a single ground fault on an ungrounded system occurred(system operating normally before the fault), the MOV was smoked. This sounded like a condition caused by some form of phase to phase fault condition. The actual conditions I do not know and since I do not know the spec of the MOV, I cannot make a judgment of to what extent the MOV would conduct or not upon seeing phase to phase voltage. No ungrounded AC is truly isolated from ground as it is grounded through the distributed capacitance. I would think that if appreciable ground capacitance(to ground?) were the cause then either the capacitance discharge current would have to be sufficient to destroy the MOV(and voltage high enough) or the voltage caused by the capacitance on one side of the MOV caused the MOV to conduct leading to a P-P fault. I would appreciate a little more explanation of how the ground capacitance causes sufficient voltage and current to destroy the MOV(and a P-P fault not be involved). Bill Mostia ======================================================= William(Bill) L. Mostia, Jr. PE Independent I &E Consultant WLM Engineering Co. P.O. Box 1129 Kemah, TX 77565 [email protected] 281-334-3169