Ungrounded Vs Grounded Inverters

Status
Not open for further replies.

pv_n00b

Senior Member
Location
CA, USA
There is a valid reason for having OCPD in both legs. I have copied SolarPro's image and placed two ground faults on it, since as we know we have to have at least two ground faults in an ungrounded system to have ground fault current flow. If one of those faults is after the point where the fused conductors are combined and the other fault is on one of the unfused conductors going into the combiner box to be combined you will have the full combiner output current, minus one string, flowing in an undersized conductor, as shown by the thick line between the fault and the combining bus. There will be no fuse to stop the fault current, the conductor will most likely overheat and possibly cause a fire.

If there were fuses in both legs in this example the fuse on the faulted negative string conductor, located between the fault and the combining bus, would open and the fault current would only be the current from one string since there would be no return path for the other string currents. You can show this same issue using any combiner box in an ungrounded system with fuses on only one leg.

While some may say this combination of faults is unlikely the fact that it would cause a great deal of damage if it happened leads me to think it's better to play it safe than to be sorry, so put in the extra fuses.
 

Attachments

  • 6_Fig_3_SP_9_3_Brooks.jpg
    6_Fig_3_SP_9_3_Brooks.jpg
    90.4 KB · Views: 5

jaggedben

Senior Member
Location
Northern California
Occupation
Solar and Energy Storage Installer
pv_n00b, I have been mulling this too, but I'm not entirely convinced that the example you've given shows a severe problem. In a transformerless inverter the GFDI will detect either of those faults if connected to a grounded AC system. The AC system will provide a partial return path. So the problem will be detected and hopefully remedied before the second fault develops.

I haven't really thought through systems that are truly ungrounded or that do have transformers and a functionally grounded conductor.
 

pv_n00b

Senior Member
Location
CA, USA
The inverter will detect the first fault and shut down the inverter. How long it then takes someone to come out and fix the system would then be a variable. I read that the average time between the first and the second fault in ungrounded systems is 2 weeks.

When it comes down to it any double fault on the output of a combiner in an ungrounded array is going to be a big problem since none of the OCP will interrupt the fault, and if we are talking about a large combiner then it's basically a DC arc welder running wild on the roof. If we are willing to accept that danger then what's a little extra risk for an unlikely fault? The OCP in one leg of the strings will protect the string conductors from fault current from other strings but they will not stop the flow of fault current of the string directly feeding the faults, so now we have a wee DC arc welder running wild on the roof. Arc fault detection and isolation at the module is the only thing that can totally stop fault current from flowing inside a single string.

A lot of array safety in ungrounded systems is dependent on the first fault being fixed before the second happens in current system design. Actually this is true for what we are calling grounded systems too since the ground bond is opened during a fault and that creates an ungrounded system. One of my concerns is that as systems get older and start to fail but don't get fixed for one reason or another we will have situations where we have faulted systems and no one knows/cares until the smoke starts. For those who don't believe that a system would be abandoned I can point out a 230kW system less than a mile from me that has been sitting on a warehouse roof for about 8 years now and is not in operation, and since the inverter is not on no one would notice if it detected a fault, which it can't because it's not on. There are not a lot of these situations, but I think they will grow as systems age. Assuming that a problem will have to be fixed if we cause the inverter to shutdown is a fools bet in my book.
 

jaggedben

Senior Member
Location
Northern California
Occupation
Solar and Energy Storage Installer
I think you raise some valid concerns. I think I would have drawn the arrows in your marked up diagram differently but I also think I follow the scenario. It is not so different from the Bakersfield fire situation. It would be better to have only one string arc-welding through ground than the entire array current flowing on that string for as long as it takes to melt down that string wiring. Fusing both sides won't address all the scenarios that fall under concerns about abandoned systems, but it could help in that scenario.

I don't think it will be that long before the code mandates module level electronics, at least on buildings (and then we can drop fusing).
 

jaggedben

Senior Member
Location
Northern California
Occupation
Solar and Energy Storage Installer
Is the risk of such a double fault current any greater with a transformerless inverter system than with a transformer-based inverter system?

No. It's rather the opposite. If the transformer-based inverter uses a fuse for GFDI, and the first fault occurs in the grounded conductor, the fuse is not very sensitive for detecting that. Of course transformer-based inverters can be designed with more sensitive GFDIs, but speaking about the models that were common a few years ago, they used a fuse.
 

pv_n00b

Senior Member
Location
CA, USA
Is the risk of such a double fault current any greater with a transformerless inverter system than with a transformer-based inverter system?

It's the same risk. Assuming the transformer based inverter has a grounded DC side, on the first fault in the ungrounded conductors the GFDI will unground the system. At that point a second fault on the newly ungrounded conductors will result in fault current flow just as in an ungrounded system.

The real problem with grounded systems is that an inadvertent ground in the grounded conductor may not be noticed until there is a fault and the GFDI opening does not stop the fault current. The Bakersfield array fire was basically the scenario I drew above for the ungrounded system. There was a connection to ground in a grounded string conductor that was not noticed. When there was a fault in a combiner output conductor the GFDI opened but then the fault current went for the faulted string conductor. Since there were no fuses in the grounded conductors there was nothing to stop the large fault current from flowing through the small string conductor, flames ensued. So while it's not a high probability scenario we can't say it has never happened.
 

SolarPro

Senior Member
Location
Austin, TX
Right, but the ground fault detection blind spot is addressed in NEC 2014. Presumably there shouldn't be any undetected first faults in systems subject to NEC 2017. Legacy systems are a different story.
 

jaggedben

Senior Member
Location
Northern California
Occupation
Solar and Energy Storage Installer
Right, but the ground fault detection blind spot is addressed in NEC 2014. Presumably there shouldn't be any undetected first faults in systems subject to NEC 2017. Legacy systems are a different story.

How does the 2014 code address the blind spot? I feel like I'mmissing something.
 

SolarPro

Senior Member
Location
Austin, TX
690.5(A). See below [emphasis added].

(A) Ground-Fault Detection and Interruption. The ground fault protection device or system shall:

(1) Be capable of detecting a ground fault in the PV array dc current-carrying conductors and components, including any intentionally grounded conductors,
(2) Interrupt the flow of fault current
(3) Provide an indication of the fault, and
(4) Be listed for providing PV ground-fault protection
 
Status
Not open for further replies.
Top