Lightning Protection For Off-Grid Long Distance Solar Array

[JohnMcNutt]

Folks,

I have been tasked to develop some kind of a solution for a problem of an off grid solar/inverter system sustaining repeated lightning damage. Probably to the charge controllers although that is not 100% known to me.

This is a standard off grid system, by standard I mean (have not visited the site personally, but this is how it was described to me) ~300 watt panels maybe 10-30 of them on a ground mounted metal frame out on some rocky point, hooked in series up to a voltage of about 250 VDC and then run a couple hundred feet underground to a room where there are Midnite/Outback charge controllers, some ~30 kWh of lead-acid batteries and an Outback inverter setup.

They probably have some Midnite SPDs which are basically some relatively small MOVs in a box at either end.

I have been tasked with this because I am also an RF guy and I see that RF sites with standard Polyphasers can take direct hits and be undamaged over many years, but they all operate at DC ground.

The problem as I see it here is that we don't have the luxury of anything being DC grounded since it is supplying DC power. So lightning strikes on/near the array, raises the potential of the metal frame of the array to several kV, this overcomes the insulation resistance of the solar modules, and current flows back on the positive lead and again overcomes insulation resistance on a charge controller and damages something along the way.

I wonder if the MOV Midnite SPDs are too little, too late.

I can think of three solutions:

1. Increase PE conductor size, possibly even ABOVE power conductor size, so the voltage of the array, voltage of the metal objects back in the inverter shed, batteries, etc. all raise in potential in lock step and lightning current does not flow on any normally power-carrying parts, especially semiconductors.

2. Source and install MOVs or some other type of arrestor that reacts faster, shunts more current, or both.

3. Install some sort of inductance on the positive power lead. Unknown effect on the charge controllers if they quickly try to stop incoming current flow as part of normal operation. Perhaps a hefty Zener diode and/or capacitor beyond the inductor would be prudent. This is a stretch but I know that Polyphaser is big on LC networks these days instead of gas discharge so I wonder if it could be scaled up for this application. A bit of a stretch I am sure.

Thoughts?

 

[itsmoked]

Any chance a storm's approach could disconnect the array entirely for the duration. Little energy will be captured in the passing overcast anyway.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[JohnMcNutt]

I like the storm disconnect idea.

Certainly the value of energy captured during a storm would be much less than the potential value of damage caused.

What kind of device could be sourced to disconnect the array? Is there such a thing as a lightning discharge detector?

 

[itsmoked]

There are dozens of lightning detectors on the market. I'd hack one to be continually powered. I'd have it trigger a timer once a strike was detected within a few miles. The relay would be a re-triggerable delay type so as long as lightning was still being detected nearby the delay would continuously restart.

It would operate like this:
A close detection would trigger the re-triggerable time delay relay that would go active and drop a contactor that disconnects the solar array from the ground equipment for 15 minutes. As long as strikes keep happening the 15 minute delay keeps resetting. Once no strikes have been seen for 15 minutes the array would be automatically re-connected.

This system would work fine if lightning storms tend to sweep past your site. If your site is uniquely located at a point where storms actually form locally on top of the site then it would be less effective as described above. In that case you would need to step up the game considerably to an E-field detector that can detect the fields forming in a storm that will directly lead to lightning.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Sparweb]

I understand everything that Keith's proposed, and it sounds clever and effective, but allow me to cast out an old piece of cowboy wisdom: "The pioneers get the arrows, the settlers get the land".
There may be some characteristics of this solar array's installation or its site that makes it an ideal place for a FIRST lightning strike in the area. The history of lightning strikes would indicate that's possible. The lightning detection & disconnect idea doesn't work if these solar panels themselves are the first site of the strike.

1. Make sure to verify all conductor sizes...
2. Why haven't the "relatively small MOVs" been replaced with much bigger units? Have the existing arresters been checked for adequate grounding?
3. I don't know much about charge controllers, but I don't think they will enjoy added inductance. IIRC most models actively regulate the current in the panels with PWM programs switching at several kHz.

Some people have done creative things with spark-gaps.

STF

 

[itsmoked]

Spar; You're certainly not wrong on that. Could be a savvy lightning protection guy would point John to a specific error in his layout that if corrected would solve the entire problem. I've seen some crazy good and detailed threads in here based on lightning protection schemes described by folks. I don't recall which forum(s) they were in.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[JohnMcNutt]

Gents, I have not seen the site. I am trying to gather more theoretical info on how it ought to be done.

The relatively small MOVs are an off the shelf surge protection device made by Midnite for this application. I just know they are there because I have been told. And they are industry standard. But seat of the pants tells me that these little things are not really adequate for serious surge currents.

In RF there are some pretty good off-the-shelf products that come with pretty good warranties and even equipment guarantees. But it is quite the opposite for DC. As well as the protection schemes required also being quite opposite.

 

[racookpe1978]

Check also your assumptions about the DC ground, and the lightening ground for the frames, power line towers, and invertor building frame. Seems that the typical DC power system has DC+, DC -, and a ground required (often to the frame if a vehicle, to the house ground of a building installation.) I know my invertors require they be grounded - separately from the DC negative.

 

[VEBill]

Once upon a time, lightning hit a tree on the corner of our property, about 150 feet from our house. Blew the tree to smithereens. Meanwhile, inside the house, there was a ONE FOOT long spark arcing between some wiring. It just so happened to be in direct line of sight where I happened to be looking.

Point being, the quarter-inch gap of the contactors will only be wide enough for the tertiary surges. Secondary surges induced in nearby wiring may easily jump the contacts' gap. Primary strokes will vaporize the contactor, and then the current will use the resulting plasma to complete the circuit to ground.

 

[Sparweb]

John, I commend your "think first then troubleshoot" approach.

A few more thoughts. "Small" is a relative term. The arrestors on my WT inverter are the size of my fist, and they aren't actually "MOV's" they are "silicon oxide varistors" if that matters ("Delta 302DC TVSS" if you want to see one; compare to the Midnite units).

I did not rely solely on resistance measurements when I placed ground rods around my WT tower. I also hooked up a 12V battery positive to the tower rod and the negative to the rod driven beside the inverter. Note that this won't work unless you include a power resistor in series to keep the test current under control. Then I used the resulting current through the loop and accounting for the resistor and long test wire resistance to estimate the total earth resistance. The earth was dry at the surface and I got a few milliohms, meaning a ground current of several kiloAmperes will cause a ground voltage differential of only 10's of volts between the rods. The value of the test is not really for the calculated result, which is suspect at best, but to confirm continuity to the earth of the ground rods. This is my "back of the envelope" analysis. A proper analysis includes important factors like the rise time and "action integral".

STF

 

[OperaHouse]

Any added inductance at the would help significantly at the charge controller with MOV array before that. A CORCOM like EMI filter would be suitable. Panel buss is generally not grounded. Midnight does not make any PWM controllers. It is being fed into a capacitor like any switch mode power supply

 

[waross]

Don't increase the size of the power conductors. The resistance of the conductors helps to dissippate the energy of the strike when the MOVs conduct.

Bill
--------------------
"Why not the best?"
Jimmy Carter