I am looking at an arc flash calculation for a battery bank. It will be done using Annex D of NFPA 70E. I have read that, read Chapter 12 in Jim Phillips book and reviewed the manufacturer of the battery's material.

In order to begin I need to determine the bolted fault current at the battery terminals. In order to do this, I need the internal resistance of the battery bank. Lacking any testing, how does one determine that?

Thanks in advance.

I don't have any short cuts or rules of thumb for that one. It seems it would be like assuming the fault current unless you had good info. Perhaps others with more battery experience might have a rule of thumb for you.

Thanks Jim. I received a response from the battery manufacturer giving me the values at the time of manufacture for the specific cells I have.

well I guess this was a much harder question than at first I perceived it would be based on the overwhelming response. Had the same result on eng-tips website, so I must be breaking new ground.
Basically I took the manufacturer's values, neglected the intercell connectors to arrive at worst case scenario at the battery terminals. I did this absent battery maintenance test results. Then I multiplied by 3 based on Annex D to arrive at worst case for the dc system. This was about 5-7 cal/cm^2 and used that for the whole system.
Hope that helps others in this situation.

You cannot ignore intercell connections and bus bars as they are on the same order as the cells themselves. The other challenge is that you can't assume full voltage at the terminals. You get a massive voltage drop under short circuit conditions. The big challenge is deciding what the external series resistance can be. When bolted connections are commonly 20-50 milliohms, and battery series resistance is less than that, it becomes difficult to honestly get below 100 milliohms externally at best. Then using battery series resistance, intercell resistances under dead shorts, and you will probably find you are off by an order of magnitude for typical substation batteries (125 vdc) and even more for most other cases except large 48 vdc telecom arrays if those still exist.

PaulEngr is right about the bus connections. Also by there are more losses by the time you connect battery cables to termation inside switchgear with more drops across fuse holders, dc breakers, etc.

I have had a similar question regarding DC. We have 2 UPS units for our data center that have the batteries set up for 480V. I wondered what the arc flash rating of that cabinet would be.

Generally whenever you look at drives, UPS's, etc., you have to look at 4 potential arcing fault sources:
1. Faults that are generally in the input (line side) sections. May have to consider downstream contributions from the drive/UPS itself here but in practice it is usually not done. This calculation is not much different from any other device.
2. Any bypass contactor arrangements in which case you have to look at both the case of UPS/drive present and UPS/drive not present at the load.
3. Faults on the output side, without a bypass contactor. This is generally limited to around 200% of output capability of the UPS itself as the assumption here would be something severe such as a shorted IGBT.
4. Internal faults such as shorting out the batteries or capacitors. For capacitors in particular since all the energy can be released nearly instantaneously the best approach is to look at energy stored in the capacitor and then calculate distance exponents. For batteries, internal series resistance is critical.

I'll bet your UPS battery is not 480 V. In fact it is likely to be MUCH less because the inverter is a charge pump and can produce nearly arbitrary output voltages using charge pump methods.. You are lucky to find 125 VDC power supplies much less keeping a UPS up at >120 VAC, although I've seen them once in a while.

By ignoring the intercell resistances and other resistances I arrive at worst case as I am using a 2 sec time for the arcing time. Playing around with adding in other resistances lowers the bolted fault current and therefore the arcing current and incident energy.
To be conservative, i ignored those and arrived at worst case and then multiplied by 3. this resulted in a 5.28 cal/cm^2.

I have seen UPS for 3-phase 480VAC. This type of inverter is used for grid stabilization applications.

Was the BATTERY configured such that the current available at the output terminals was 480 V?

Since it is a charge pump, the input (batteries in this case) and output (line/load voltages) don't have to necessarily have anything to do with each other.

When going directly from AC to AC (cycloconverter), you can only get to about 95% of the supply voltage. This creates a 9 IGBT bridge rather than the conventional 12 IGBT case since there is no DC bus, but this design is relatively rare.

When going from AC to DC (converter) or DC to AC, GENERALLY you are limited by two factors. First, you are limited to the peak voltage on the input. Second, thyristor ratings have to be twice the peak/trough input voltage. iwth a standard low voltage IGBT the swing is 1400 V. So for a standard 480/600 V class drive (and lower), the maximum possible input voltage (AC) is just shy of 700 V. For traction and STATCOM's they typically use nonstandard transformers and operate at 690 V to push power to the device limits (nonstandard transformer is cheaper than more IGBT's and more copper).

On the DC side of things, we also have DC to DC converters. These work by alternately closing and opening the thyristor switches (transistors) between a DC storage device (capacitor or inductor) and the two DC buses. If a capacitor is used, the output voltage is based on the impedance that the capacitor sees so we can achieve extremely high output voltages for a short period of time. Thus DC-DC converters are either boost, buck-boost, or buck types depending on whether they boost or buck (lower) input voltage to the output.

We can't do this in AC. Our AC choice is a transformer, or mixing AC and DC converters. Since transformers are very cheap (relatively speaking) this is usually the option of choice for design.

OK, now with batteries, each cell in a battery is usually capable of only somewhere around 1-3 volts depending on the chemistry. They are wired together in sries to get higher voltages (such as 6 or 12, 24, 48, or 125 volts). The wiring can be external or internal.

The limitation is that eventually you can get uneven distribution of storage if you have a weak cell or two, and then this can in turn cause increased voltage stress on the remaining cells during charging because the weak ones will act as a short (or even reverse polarity). So with batteries there is a safety threshold beyond which you need to put them in parallel strings and boost the voltage externally.

Finally in a UPS applicable since all power semiconductors are effectively variations of diodes and switches, the output is never a "true" sine wave. It is always switched. To clean this up and make "clean" power, an LC filter (mix of capacitors and inductors) is used. A transformer can easily be part of the filtering design. If it is also being used to bump the voltage up to avoid having high voltages on the DC bus (and batteries), so much the better.

SO...480 V (and higher) UPS's definitely exist. I have a couple. But 480 V battery strings would be very rare. Not that you can't ignore manufacturer guidelines and build one anyways, but it's generally not a good idea especially for something that is supposed to be a device to improve reliability in the first place.

Perhaps I should have said batteries. The battery modules for the 1MW UPS/grid stabilization application take up 3 tractor trailors and the inverters take up another one. The DC voltage is on the order of 1000V.