Overload protection downstream of inverter on off grid solar pv system

This also might depend on whether the RCD is voltage-dependent, and also how quickly the inverter output decays on a real fault.

However, not sure whether you are asking about ADS for protection against electric shock, or 'overload protection' because these are two very different things.

gkenyon said:

This also might depend on whether the RCD is voltage-dependent, and also how quickly the inverter output decays on a real fault.

Indeed - the inverter output may well collapse in the face of a low impedance fault, but (as I understand it) the voltage-dependent RCD behaviour is co-ordinated to shock protection requirements - i.e. if the L-N voltage at the RCD remains above 50V the device should still operate normally. If it drops below that the RCD may well not open, but the voltage available for a shock should also be below 50V - hence not a hazard (at least not in the conventional electric shock sense). Non voltage dependent RCD should trip if the residual current exceeds its threshold regardless of the voltage available (as I found out once when trying to verify the N contact of an RCBO with a multimeter - even on the bench with no mains connections at all the damn thing trips before you can get a reading!)

   - Andy.

AJJewsbury said:

Indeed - the inverter output may well collapse in the face of a low impedance fault, but (as I understand it) the voltage-dependent RCD behaviour is co-ordinated to shock protection requirements - i.e. if the L-N voltage at the RCD remains above 50V the device should still operate normally. If it drops below that the RCD may well not open, but the voltage available for a shock should also be below 50V - hence not a hazard (at least not in the conventional electric shock sense).

Does this not depend on which version of the product standard is being considered, and which classification is accorded by the manufacturer?

To clarify, all final circuits are fed from individual RCBO's and we do have a N to PE link and an external earth, so I'm concerned only with overload protection. All the conductors are selected with ratings greater than the associated nominal trip currents of the associated protective thermal magnetic trips and I can prove using the adiabatic equation that they are above the minimum size inferred.to deal with the extended trip times. Also well aware that the inverter has its own more sophisticated protection mechanisms and this is my dilemma. I'm trying to design a properly engineered solution where there is discrimination between protective devices up and down the supply and that is proving difficult to achieve.

Talking to a colleague who has the experience with grid-connected systems I now understand that this is a problem unique to the off-grid situation where there is no connected grid to provide the fault current and thus the predictable trip performance or to put it another way, a predictable supply impedance rather than the unpredictable "virtual" variable impedance of the inverter.

All responses greatly appreciated. I've revised my original design and I think I now have an acceptable compromise.

I wouldn't worry abut trying to achieve discrimination between the inverter's internal protection and downstream MCBs/RCBOs as likely you'll never achieve it reliably. That's not unique though - you'll be pushed to show any discrimination between as B32 MCB and a 13A fuse (whether in a plug or feeding a number of sockets on a fused spur) or between 30mA RCDs where the regs require one following another (e.g. caravan inlets and pitch sockets), or where a submain feeding a CU is protected by any size of MCB. That's just life I'm afraid. The regs don't require discrimination (sorry, selectivity nowadays) in all situations - just tell you how to do it when it is required.

Also be a bit wary of the adiabatic with longer disconnection times - although it always errs on the side of safety, the err can grow to unreasonable proportions much above 5s and you can start getting absurd results like needing Iz >> In.

Although the wording isn't as clear as it used to be (at least to me), the 2nd paragraph of 435.1 is your friend here - in general an overload protective device, with a suitable breaking capacity, may be assumed to provide fault protection to the downstream conductors. (Not true for every type of device, but should be OK for common or garden ones as far as I know).

   - Andy.

gkenyon said:

Does this not depend on which version of the product standard is being considered, and which classification is accorded by the manufacturer?

Ah, Graham (as usual) you know much more of the various standards than I do - please do enlighten us!

   - Andy.

If you've selected B tripping curve RCBOs and sized them as tightly as reasonable to the loads then you've probably done all you can from a general perspective.  There is no guarantee that any rating RCBO will trip magnetically as that depends on the inverter's response to overload - it is possible that the inverter will shut down before the RCBO trips magnetically in the event of an overload, particularly where the overload causes the output voltage to collapse (e.g. in the near short circuit case).

If the specification of the inverter, particularly its response to overload, is known then you can potentially be a little more optimistic.  To take an example, one manufacturer of inverters (Victron) specifies an overload capability of 2xInom for either 2 minutes (where the output voltage can be maintained) or 30 cycles (where the output voltage cannot be maintained).  For a 10kVA inverter, (Inom=43A), that should operate the magnetic trip on 6A RCBOs and depending on what load the inverter is under, maybe 10A ones too, assuming a B tripping curve.  If you're really lucky, and the inverter isn't under much other load, you might even get a 16A RCBO to trip magnetically.  Once those overload limits are exceeded the inverter shuts down and retries every 30 seconds, doing that three times before requiring manual reset.  Note that those characteristics are specific to that manufacturer - they can't be taken as indicative of the behaviour of inverters in general.

If selectivity is an important requirement then you will need to be involved in specifying the inverter so as to achieve the behaviour you need (although if you want selectivity for faults on large loads then be prepared to make yourself unpopular by saying the inverter should be two or three times the size!).  It should also be noted that an inverter's performance under overload conditions is heavily dependant on the DC source - if the battery system is not able to deliver the powers involved (e.g. if the DC supply is insufficiently stiff due to issues such as battery under sizing, battery degradation, skimping on the size of the DC cabling, etc.), then the inverter won't deliver its specified characteristics.

AJJewsbury said:

Ah, Graham (as usual) you know much more of the various standards than I do - please do enlighten us!

In the versions of BS EN 61008 and 61009 amended to 2017, there are different classifications of voltage-dependent RCD and RCBO.

I understand the technical performance requirements have been revisited in the latest versions of the standards although there will be a cutover period before these are fully applied to all products on sale.

More information here: https://www.beama.org.uk/static/dda8f294-e43c-484f-8061e92b1f1e64e0/Technical-Bulletin-Guide-to-Residual-Current-Device-RCD-E-Marking.pdf 

Thanks for that Graham - lots I didn't know there! I'd not come across the "E" markings before.

So, if I've understood it correctly, we need only worry about devices marked "E2" or "E3" (as the other types are OK for "indirect contact" protection (i.e in the case of L-PE faults)) and those seem to be limited by note b to "Only devices integrated in one unit with a socket-outlet or designed exclusively for being associated locally with a socket outlet in a same mounting box." - so that makes me think of what we know in the UK as BS 7288 socket RCDs (or perhaps those industrial enclosures that have a modular DIN rail enclosure above a BS 4343 (as was) socket outlet).

If that's the case, it's maybe not a worry for devices intended to be used in CUs/DBs?

   - Andy.

AJJewsbury said:

So, if I've understood it correctly, we need only worry about devices marked "E2" or "E3" (as the other types are OK for "indirect contact" protection (i.e in the case of L-PE faults)) and those seem to be limited by note b to "Only devices integrated in one unit with a socket-outlet or designed exclusively for being associated locally with a socket outlet in a same mounting box." - so that makes me think of what we know in the UK as BS 7288 socket RCDs (or perhaps those industrial enclosures that have a modular DIN rail enclosure above a BS 4343 (as was) socket outlet).

The ambiguous area ... which is also drawn out in the BEAMA Bulletin if I remember correctly, is that some devices intended for use in CUs and DBs are marked E3. The BEAMA Bulletin explains why this might happen.

So, yes, I guess E1 or no marking are apparently OK for ADS without further enquiry, but E2 or E3 should not be selected to provide ADS, unless further due diligence is undertaken with the manufacturer.

The Bulletin also tells us what's happening in the future with the newer standards, to avoid this uncertainty.

AJJewsbury said:

So, if I've understood it correctly, we need only worry about devices marked "E2" or "E3" (as the other types are OK for "indirect contact" protection (i.e in the case of L-PE faults)) and those seem to be limited by note b to "Only devices integrated in one unit with a socket-outlet or designed exclusively for being associated locally with a socket outlet in a same mounting box." - so that makes me think of what we know in the UK as BS 7288 socket RCDs (or perhaps those industrial enclosures that have a modular DIN rail enclosure above a BS 4343 (as was) socket outlet).

The ambiguous area ... which is also drawn out in the BEAMA Bulletin if I remember correctly, is that some devices intended for use in CUs and DBs are marked E3. The BEAMA Bulletin explains why this might happen.

So, yes, I guess E1 or no marking are apparently OK for ADS without further enquiry, but E2 or E3 should not be selected to provide ADS, unless further due diligence is undertaken with the manufacturer.

The Bulletin also tells us what's happening in the future with the newer standards, to avoid this uncertainty.