It is currently Wed Feb 21, 2018 6:17 pm



Post new topic Reply to topic
Author Message
ekstra   ara
 Post subject: Inherient current limited devices like UPS.
PostPosted: Thu Aug 23, 2012 8:42 am 
Offline
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 483
Location: New England
How are people treating devices like VFD's and UPS's which are inheriently current limited by design - not by protection. For example I have a 40KVA 208 3Ph UPS, with 150A output fuses. By nature of the double conversion process (converting AC to DC, and then DC back to AC) the unit is inheriently current limiting. I would estimate dead bolt fault current at approximately 300A. There are internal CT's that would trip the unit into fault before the fuses actually blew, but they are not overcurrent devices.

How would you approach this for analysis?

1) Use the 2 second rule.
2) Generalize that the 'under 125KVA' supply would apply

What would change if the UPS was 480 3ph @ 40KVA? Could no longer apply the 125KVA limit exception because of the voltage.

Could we even sustain an arc at this low current level?


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Thu Aug 23, 2012 9:45 am 
Offline
Arc Level

Joined: Mon Jan 18, 2010 11:35 am
Posts: 445
Location: Wisconsin
i have seen UPS's with solid state bypasses that operate when the output circuitry senses are short circuit. The intent of the bypass is to provide enough fault current so that protective devices will function. Check the specifications.

Also, there are often 'maintenance' bypasses for UPS and VFD, you need to include the possibility that these may be operated.


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Thu Aug 23, 2012 6:34 pm 
Offline
Plasma Level
User avatar

Joined: Tue Oct 26, 2010 9:08 am
Posts: 1878
Location: North Carolina
Two potential cases.

First, the drive does current limit and has a limited peak output which is part of the drive specifications. Generally this does not exceed twice the 100% continuous loaded case. Peak output is published for UPS's. The overload capacity of the drive/UPS comes into play and generally there is some documentation (though frequently scant) which describes what happens in this case and how long before the drive detects the overload condition and shuts down.

Second, assume the internal link capacitor fails. It sometimes takes some effort but you can find the size which gives you the energy which you can then apply as an arcing fault by converting to cal's and then directly calculating the area of a sphere at the appropriate distance (cal/cm^2 at a point). This corresponds roughly to the case of a dead short of an IGBT.


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Fri Aug 24, 2012 8:45 am 
Offline
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 483
Location: New England
I agree with all above, but let me clarify further.
1) In most cases a fault current rating of 2 times the FLA is probably a good guess.
2) VFD's typically do not have bypasses.
3) UPS may occassional have bypasses, so both conditions would have to be check, UPS or Bypass. But in bypass the fault currents are going to be much higher because you are basically calculating fault current on a conductor to a point in the circuit. So its more like any panel or load fed from a conventional source.

The premise of my question remains how to deal with the current limited devices. In above example, 150A fuses with 222A (2xFLA) is going to take several minutes to open, may even ten minutes depending on fuse type. Just plugging in the numbers into the IC equation is going to calculate out to some cal/cm2 value, probably a very large one because of the clearing time, but is it representative of the real world. If you could generate an arc at all, I seriously doubt it would be sustained for ten minutes with only 222A availabe as fault current. So on these kinds of devices some alternate approach, and/or exemption looks to be needed. So what do we do for an approach? Use two second rule? The IEEE original 1584 said the had difficulty sustaining an arc on a transformer under 125KVA and 208V. Assuming impedance of say 5%, the fault current would be something like 7000A. So if a UPS is only 222A fault at 208V can we apply the same exemption?
What if it was 480V, and fault current was then 100A. Could we even get an arc with only 100A fault current available?
Applying the math here just doesn't seem to make sense.

Comments.


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Sat Aug 25, 2012 7:32 am 
Offline
Plasma Level
User avatar

Joined: Tue Oct 26, 2010 9:08 am
Posts: 1878
Location: North Carolina
First off, you're getting your information off the internet. So as usual, caveat emptor. As to my qualifications...I'm not a drives engineer at least by trade. I know several of them and can vouch for their abilities. However, I currently have on my plate an "upgrade" to a drive system that has an 8 digit price tag in American dollars. I've spent close to that in the last 3 years on many other projects that involve drives but this one is by far the biggest so far. Whether it fails or succeeds, my name will probably show up in the mining news. I've got one shot at this to get it right or find a new line of work so if it fails, I'll also be putting out my resume.

To give you some idea of what can go wrong with drives and to borrow from my background and current project, 6 sets of "helper drives" (AC motors configured to load share with DC motors in a large servo system) were installed in Australia. The drive engineers claimed that the drives were so fast that they did not need semiconductor fuse protection and could shut themselves down before any damage occurred. A major application design error was made. The servo system in question spends a lot of time in stall or moving very slowly. The drives were sized according to RMS rating which is for typical applications but unfortunately in this case, it was the continuous current (DC) rating of the IGBT's that mattered most. This means that the surface area of the silicon IGBT's was about half the size it should have been. The Australian mining companies that own these drives were very unimpressed with the speed of the drives shutting themselves down around 18 months later when the IGBT's began exploding and sending shrapnel through the drive housings. Apparently the drive was not fast enough to shut down the power electronics. This is a simple application of bolted faults and a clear demonstration of why high speed semiconductor protection devices are mandatory.

Now you would think that everyone would have learned from this experience. However, I was recently approached by the largest mining equipment manufacturer in the world and they had the same drives with the same lack of protection and made the same arguments for why I didn't need it despite trying to sell me a system where the cost of installing the protection would have amounted to about a 0.1% cost increase to manufacture it. I was not able to determine at that point whether the amount of silicon had increased or not but fortunately it didn't matter as they priced themselves out of the market relative to their competitors.

I'm going to go out on a limb here and assume you are talking about static drives, no matter what the application is. In dynamic drive systems such as synchronous condensers, Ward-Leonard loop drives, and MG set frequency converters, you would analyze it just like any other motor/generator system. Your biggest hassle is going to be determining the decay rate of the asymmetrical current. The largest drives that I currently have to deal with actually fall into this class with some of them able to put out up to 8000 A DC (though at low voltage). The key in this case is to determine and do everything at the peak power point where the amount of energy is at a maximum. If you are running up against these describe your situation and we can probably walk through the analysis.

As for static drives...UPS's and drives are not magical high tech devices that cannot be understood by mere mortals. One one side of the device you have what is generally called a "converter". It can be a very simple, passive device such as a diode rectifier. Or it can be multiple voltage controlled devices (IGBT's or FET's), or current controlled devices (GTO's or GCT's) which all perform the same function: a semiconductor switch used to convert AC into DC or (depending on the timing) convert DC back into AC in "regenerative" mode if the drive/UPS is capable of this. In between we have the DC circuitry. This may consist of up to 3 components. There is always a DC link capacitor these days though it can also be an inductor. Generally capacitors are used because they are physically smaller and less expensive. There is also sometimes a brake chopper in a drive which is yet another semiconductor switch and a resistor. There is also a battery charger/DC-DC converter in a UPS which does voltage conversion down to a lower voltage via a charge pump (more switches and capacitors) as well as battery control/monitoring. Finally on the output side we have more semiconductor switches forming the inverter stage. In addition to all of this especially older drive designs may use one or two isolation transformers, which may have multiple taps though the tapping is unimportant for safety analysis purposes. It's only purpose is to step down the voltage in stages to allow series strings of IGBT's to operate at lower voltages (12, 18, 24, or 36 pulse drives). The other exception is that in the case of a cycloconverter there is no DC link at all because it converts directly from AC to AC. The analysis is somewhat simpler and so is the drive since it only needs 9 IGBT's for full regenerative conversion instead of the usual minimum of 12 devices.

So...the view that we have from 30,000 feet is that we have some switching devices that are converting AC or DC power into AC or DC. These are called "converters" or "inverters" and have lots of other names depending on what they do. We may or may not have a variety of energy storage devices as well. From an arc flash hazard point of view, we are done here. The simplest failure model that adequately explains all the components will result in a conservative estimate of arc flash hazard potential.

haze10 wrote:
I agree with all above, but let me clarify further.
1) In most cases a fault current rating of 2 times the FLA is probably a good guess.


This is not a guess. The exact value would be 1.87 times the rated maximum current which is the current (voltage) that can occur at least until the IGBT explodes (semiconductors are very thermally limited). Generally it occurs if you have an IGBT that fails to open properly at which point you have a direct phase-to-phase fault. Though the arcing fault current is likely to be less than the bolted fault current it nonetheless provides the data required to model the system using IEEE 1584. We then need only look at the protection systems in the drive to find the opening time.

All other issues with the inverter itself result in the capacitor bank dumping it's internal charge out quickly. Although impedance does play a part, the quickest way to the solution is to calculate the DC link capacitor bank dumping it's charge out all at once. Again, failure mechanism is unimportant. In this case we convert the energy in the capacitor to calories, then divide by the area of the sphere in question (in cm^2) to calculate calories/cm^2 in a worst case scenario.

With batteries, nominally you need to know the series resistance. This is often hard to get and requires a lot of time calling manufacturers. 10 milliohms is the lowest number I've ever seen and gives you a conservative estimate if you know the battery voltage.

You should run the line-to-line short circuit fault leading up to the drive and obviously this could happen internally as well but it cannot extend "downstream" of the drive/UPS due to the electrical isolation that occurs in the case of AC drives though it could happen in offline or line interactive drives which don't have isolation. This is the one point where even an extremely oversimplified schematic really comes in handy.

As to the bypass switch...you are right. You can have a failure in that thing which would correspond to bypassing the UPS. At the load this is not really a problem since the load may or may not just see some very strange harmonics. At the UPS/bypass switch is the issue. However, even assuming the UPS itself is in some sort of worst case condition of being 180 degrees out of phase with the line source, the result is a short circuit within the UPS and we are again back to the IGBT's acting a resistors (for a short period of time before they explode) and 1.87 times the maximum current. Downstream again it would be more of a short circuit scenario.
So...although it is helpful to know what is going on inside a drive to be able to understand how to analyze it, in the end, you don't actually have to know very many details to perform an arc flash hazard calculation.

This is the reason why I recommended the "x2" factor initially. You are welcome to do a deeper analysis yourself of your particular drives and UPS's. What I've found though in general is that the capacitors are generally too small to provide much more than a concern, the batteries despite making a big mess exploding from internal pressures make a big mess, and that the "x2" dead short condition is the worst case condition limited only by the drive internal/external overcurrent protection devices. It also nicely re-iterates the reason that you need to follow the section of the manufacturer's manuals describing the required external protection if the drive doesn't have it.


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Wed Aug 29, 2012 12:44 pm 
Offline
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 483
Location: New England
PaulEngr wrote:
First off, you're getting your information off the internet. So as usual, caveat emptor. As to my qualifications...I'm not a drives engineer at least by trade. I know several of them and can vouch for their abilities. However, I currently have on my plate an "upgrade" to a drive system that has an 8 digit price tag in American dollars. I've spent close to that in the last 3 years on many other projects that involve drives but this one is by far the biggest so far. Whether it fails or succeeds, my name will probably show up in the mining news. I've got one shot at this to get it right or find a new line of work so if it fails, I'll also be putting out my resume.

To give you some idea of what can go wrong with drives and to borrow from my background and current project, 6 sets of "helper drives" (AC motors configured to load share with DC motors in a large servo system) were installed in Australia. The drive engineers claimed that the drives were so fast that they did not need semiconductor fuse protection and could shut themselves down before any damage occurred. A major application design error was made. The servo system in question spends a lot of time in stall or moving very slowly. The drives were sized according to RMS rating which is for typical applications but unfortunately in this case, it was the continuous current (DC) rating of the IGBT's that mattered most. This means that the surface area of the silicon IGBT's was about half the size it should have been. The Australian mining companies that own these drives were very unimpressed with the speed of the drives shutting themselves down around 18 months later when the IGBT's began exploding and sending shrapnel through the drive housings. Apparently the drive was not fast enough to shut down the power electronics. This is a simple application of bolted faults and a clear demonstration of why high speed semiconductor protection devices are mandatory.

Now you would think that everyone would have learned from this experience. However, I was recently approached by the largest mining equipment manufacturer in the world and they had the same drives with the same lack of protection and made the same arguments for why I didn't need it despite trying to sell me a system where the cost of installing the protection would have amounted to about a 0.1% cost increase to manufacture it. I was not able to determine at that point whether the amount of silicon had increased or not but fortunately it didn't matter as they priced themselves out of the market relative to their competitors.

I'm going to go out on a limb here and assume you are talking about static drives, no matter what the application is. In dynamic drive systems such as synchronous condensers, Ward-Leonard loop drives, and MG set frequency converters, you would analyze it just like any other motor/generator system. Your biggest hassle is going to be determining the decay rate of the asymmetrical current. The largest drives that I currently have to deal with actually fall into this class with some of them able to put out up to 8000 A DC (though at low voltage). The key in this case is to determine and do everything at the peak power point where the amount of energy is at a maximum. If you are running up against these describe your situation and we can probably walk through the analysis.

As for static drives...UPS's and drives are not magical high tech devices that cannot be understood by mere mortals. One one side of the device you have what is generally called a "converter". It can be a very simple, passive device such as a diode rectifier. Or it can be multiple voltage controlled devices (IGBT's or FET's), or current controlled devices (GTO's or GCT's) which all perform the same function: a semiconductor switch used to convert AC into DC or (depending on the timing) convert DC back into AC in "regenerative" mode if the drive/UPS is capable of this. In between we have the DC circuitry. This may consist of up to 3 components. There is always a DC link capacitor these days though it can also be an inductor. Generally capacitors are used because they are physically smaller and less expensive. There is also sometimes a brake chopper in a drive which is yet another semiconductor switch and a resistor. There is also a battery charger/DC-DC converter in a UPS which does voltage conversion down to a lower voltage via a charge pump (more switches and capacitors) as well as battery control/monitoring. Finally on the output side we have more semiconductor switches forming the inverter stage. In addition to all of this especially older drive designs may use one or two isolation transformers, which may have multiple taps though the tapping is unimportant for safety analysis purposes. It's only purpose is to step down the voltage in stages to allow series strings of IGBT's to operate at lower voltages (12, 18, 24, or 36 pulse drives). The other exception is that in the case of a cycloconverter there is no DC link at all because it converts directly from AC to AC. The analysis is somewhat simpler and so is the drive since it only needs 9 IGBT's for full regenerative conversion instead of the usual minimum of 12 devices.

So...the view that we have from 30,000 feet is that we have some switching devices that are converting AC or DC power into AC or DC. These are called "converters" or "inverters" and have lots of other names depending on what they do. We may or may not have a variety of energy storage devices as well. From an arc flash hazard point of view, we are done here. The simplest failure model that adequately explains all the components will result in a conservative estimate of arc flash hazard potential.



This is not a guess. The exact value would be 1.87 times the rated maximum current which is the current (voltage) that can occur at least until the IGBT explodes (semiconductors are very thermally limited). Generally it occurs if you have an IGBT that fails to open properly at which point you have a direct phase-to-phase fault. Though the arcing fault current is likely to be less than the bolted fault current it nonetheless provides the data required to model the system using IEEE 1584. We then need only look at the protection systems in the drive to find the opening time.

All other issues with the inverter itself result in the capacitor bank dumping it's internal charge out quickly. Although impedance does play a part, the quickest way to the solution is to calculate the DC link capacitor bank dumping it's charge out all at once. Again, failure mechanism is unimportant. In this case we convert the energy in the capacitor to calories, then divide by the area of the sphere in question (in cm^2) to calculate calories/cm^2 in a worst case scenario.

With batteries, nominally you need to know the series resistance. This is often hard to get and requires a lot of time calling manufacturers. 10 milliohms is the lowest number I've ever seen and gives you a conservative estimate if you know the battery voltage.

You should run the line-to-line short circuit fault leading up to the drive and obviously this could happen internally as well but it cannot extend "downstream" of the drive/UPS due to the electrical isolation that occurs in the case of AC drives though it could happen in offline or line interactive drives which don't have isolation. This is the one point where even an extremely oversimplified schematic really comes in handy.

As to the bypass switch...you are right. You can have a failure in that thing which would correspond to bypassing the UPS. At the load this is not really a problem since the load may or may not just see some very strange harmonics. At the UPS/bypass switch is the issue. However, even assuming the UPS itself is in some sort of worst case condition of being 180 degrees out of phase with the line source, the result is a short circuit within the UPS and we are again back to the IGBT's acting a resistors (for a short period of time before they explode) and 1.87 times the maximum current. Downstream again it would be more of a short circuit scenario.
So...although it is helpful to know what is going on inside a drive to be able to understand how to analyze it, in the end, you don't actually have to know very many details to perform an arc flash hazard calculation.

This is the reason why I recommended the "x2" factor initially. You are welcome to do a deeper analysis yourself of your particular drives and UPS's. What I've found though in general is that the capacitors are generally too small to provide much more than a concern, the batteries despite making a big mess exploding from internal pressures make a big mess, and that the "x2" dead short condition is the worst case condition limited only by the drive internal/external overcurrent protection devices. It also nicely re-iterates the reason that you need to follow the section of the manufacturer's manuals describing the required external protection if the drive doesn't have it.


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Wed Aug 29, 2012 12:48 pm 
Offline
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 483
Location: New England
I take it the energy of a capacitor is a function of its size and voltage? Can you demonstrate an example on how you do this for a VFD or a UPS? I don't quite understand how you are getting cal/cm at a distance. Does the SCRs not continue to power into a fault? So how are you opening OC devices? You are not using the IEEE equations?


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Mon Sep 03, 2012 10:36 am 
Offline
Plasma Level
User avatar

Joined: Tue Oct 26, 2010 9:08 am
Posts: 1878
Location: North Carolina
haze10 wrote:
I take it the energy of a capacitor is a function of its size and voltage? Can you demonstrate an example on how you do this for a VFD or a UPS? I don't quite understand how you are getting cal/cm at a distance. Does the SCRs not continue to power into a fault? So how are you opening OC devices? You are not using the IEEE equations?


Almost no one uses SCR's any more for any drive. They require two devices per gate (to force commutation). Only VERY old drives have them. GTO's almost entirely displaced SCR's back in the 1980's and those were displaced by GCT's in the last 10 years. On the voltage controlled frontier, IGBT's have displaced current controlled devices (GTO's and GCT's) wherever they can be used. The generic term for all of these devices is the thyristor.

OK, to recap, we have three potential sources of hazard when dealing with a drive:
1. Something shorts out on the incoming bus. This is no different than calculating incident energy due to a cable fault or on the terminals so no changes to conventional approaches to arc flash calculations. I won't speak any further to that failure mode.
2. For some strange unknown reason the drive basically turns into a rectifier and conducts at maximum current for a period of time and for some reason the drive protection electronics don't shut it down. Unusual but at least possible.
3. The internal DC link storage device has a short on it's output terminals, causing it to release all of it's energy at once. Most likely it will explode due to internal pressures (arc blast) as opposed to an arc flash (radiative heat).

The thyristors (device type doesn't matter) won't last long enough in a dead fault condition to even worry about them. They fail with I^2*t and the t part is really short. So the only way to look at the thyristors is that for some reason say a programming glitch they manage to operate at maximum output current in DC mode without the drive shutting down. The resulting current will eventually trip either the drive's internal protection or external protection (you did install semiconductor fuses as per the instructions, right?) Simply use the maximum current capability of the thyristor times 1.87 to calculate bolted fault conditions and you are done. You can get this spec easily by looking at the drive spec.

Having a capacitor drive the reaction is a little different because in this case we have an energy storage device in the system. The fault mode is a little different. Assuming the capacitor is charged to maximum voltage, the maximum charge stored in the capacitor is 1/2 * C * V^2. This is basic physics. Here's a link to an online capacitor calculator:
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/capeng.html

Plugging in some numbers let's suppose we have a 500 microfarad DC link capacitor (rather large) in a 480 V drive and we are running at 480/277 V. The rectifier bridge will be full wave so the actual voltage will be 1.87 times 277 V, or 518 V. So plugging these into the above calculator, we get 67 Joules of energy. Now we need this in calories so hopping over to this web site we can convert it:
http://www.onlineconversion.com/

Our 67 joules works out to be 16 calories. Now assuming that there is no resistance and all that energy is released at once and that it is all converted efficiently into thermal radiation (a stretch but we're being conservative here) then we can convert 18" to centimeters or 45.72 centimeters.

Now the area of a sphere is 4*pi*R^2. So plugging in 45.72 centimeters and dividing into 16 calories by this gives us a final result of 0.00061 cal/cm^2. In fact without much thought we can recognize that we need either to work with much higher voltages or 4 orders of magnitude larger capacitor banks, on the order of several hundred farads, before the amount of energy is significant.


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Tue Sep 04, 2012 11:45 am 
Offline

Joined: Mon Nov 19, 2007 5:25 am
Posts: 29
Location: Titusville, Fl.
Well done... Glad to see these devices can be worked on, w/o as much concern for arc flash, HRC 0 it is...


Top
 Profile Send private message  
Reply with quote  
 Post subject:
PostPosted: Wed Sep 05, 2012 5:45 am 
Offline
Plasma Level
User avatar

Joined: Tue Oct 26, 2010 9:08 am
Posts: 1878
Location: North Carolina
321Liftoff wrote:
Well done... Glad to see these devices can be worked on, w/o as much concern for arc flash, HRC 0 it is...


Just to be clear...I'd still check the "capacitor case" carefully using the same math and logic if you are dealing with higher voltage or larger currents such as 100 MW motors. This is not something I "made up". It is following a very similar method that has been used for calculating DC arc flash incident energy, with the same caveat that although the energy conversion is not perfect, it gives a conservative result sans actual laboratory tests.

Second, it does not abdicate the other two cases...calculating the "front end/upstream" arc flash potential using conventional methods on the line side, and calculating the downstream "maximum output as a full wave DC rectifier" on the output side. These two cases will dominate the result.

And it does not eliminate the effect of arc blast. 6 drives were built designed with Siemens traction drives for "helper drives" on six draglines in Australia. They were built to support the UDD (universal dig/dump) design sponsored through CSIRO. Siemens claimed their drives were so good that semiconductor fuse protection was not required (I recently received the same sales pitch last year) and that the drive could shut itself down before causing any harm to personnel. All six drives were designed with 80,000 hour design life. However, a fatal error was made. The drive engineers involved used the spec sheets for the thyristor packages and correctly sized the drives for RMS operation. What they did not consider is that in this particular application the drives spend a significant amount of time either inching or standing still, which means that the drives must be designed for absolute DC maximum current flow, not RMS. This reduced the design life due to Arhenius equations from 80,000 hours to around 12,000 hours. Approximately 18 months after installation, as predicted (in hind sight), the IGBT's in the drives began exploding and sometimes sending shrapnel through the cabinetry. Due to safety concerns, the drives were shut down on all 6 UDD systems and UDD itself was shelved as a "failure".


Top
 Profile Send private message  
Reply with quote  
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 10 posts ] 

All times are UTC - 7 hours


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  
© 2017 Arcflash Forum / Brainfiller, Inc. | P.O. Box 12024 | Scottsdale, AZ 85267 USA | 800-874-8883