Adaptive Setting LV Circuit Breakers for IBR Microgrids

[JezNZ]

Hi All,

Been a while since I've posted here!

Currently working on the design-build of a microgrid being the first to meet regenerative certification criteria (105% generation of consumption) in the locale I am.

The microgrid will operate both grid-connected and island-mode and is IBR HEAVY (Solar + BESS). So naturally I have the problem of protection (in)sensitivity to low fault currents in island mode.

It looks like I will be headed down the path of zone-based differential protection for MV components, however for LV it is a lot trickier to find a cost justifiable solution.

Sadly the IEEE draft guide on Microgrid Protection doesn't help much other than to suggest running studies (of course??), but there are some good papers floating around. One the most promising options suggests using "adaptive setting circuit breakers" - now I assume in principle this means you have a group of settings for grid-mode and another group for island-mode - easy!

BUT - does such a product actually exist? Commercially? In my mind I am thinking of cost efficient protection like MCCBs or Power Circuit Breakers whose microprocessor trip units can be toggled between two settings groups?

Or is this just implying I need to put in high end relaying and CTs etc and have custom logic etc? Or go old school and CTs and overload relays with shunt trips and selectable relays in different modes - neither of which are paths I would like to go down.

Will take other thoughts on creative solutions people out there have come up with!

 

[proprot]

Hove you considered something like Solkor protection ?

 

[wcaseyharman]

How much selectivity do you need in island mode?
You could use standard overcurrent elements in grid tied scenarios where there is sufficient fault current, then use time coordinated under voltage relaying during islanded situations.
Alternatively, I recently used microprocessor relays to implement voltage controlled overcurrent relays on a recent MV microgrid - I needed selectivity since the generator was installed in the middle of a distribution circuit with feeds in both directions. It worked - had a fault last week and the relaying isolated the fault and kept the rest of the system in service. You could potentially implement a similar scheme with less expensive relaying and LV breaker shunt trips.
Just some ideas, perhaps gets your mind going for other practical solutions.

 

[JezNZ]

How much selectivity do you need in island mode?
You could use standard overcurrent elements in grid tied scenarios where there is sufficient fault current, then use time coordinated under voltage relaying during islanded situations.
Alternatively, I recently used microprocessor relays to implement voltage controlled overcurrent relays on a recent MV microgrid - I needed selectivity since the generator was installed in the middle of a distribution circuit with feeds in both directions. It worked - had a fault last week and the relaying isolated the fault and kept the rest of the system in service. You could potentially implement a similar scheme with less expensive relaying and LV breaker shunt trips.
Just some ideas, perhaps gets your mind going for other practical solutions.

Hi,

Yes, a reduction in selectivity is an option being floated - I'm currently in the midst of writing a memo regarding the challenges, solutions - with next steps being high-level cost analysis, so yes alternative solution are welcome.

The network I'm looking at will be both MV & LV, distributed across a 600-800 subdivision with a BESS as community level and all houses having ~8kWp solar.

One option I've come across recently is looking at using the Eaton Power Defence MCCBs at the transformer LV distribution frames - these have a remote-triggerable maintenance-mode, ostensibly for provided arc-flash but considering using this to lower the instantaneous settings during island mode and sacrificing selectivity downstream.

 

[wcaseyharman]

The major challenge I had was differentiating load current from fault current when trying to achieve selectivity. With synchronous generators the minimum fault current can be less than load current, and in order to trip for all faults the overcurrent set point had to be less than the maximum generator load current. I was able to get sufficient selectivity using voltage supervised discreet overcurrent elements with coordinated fixed time delays along with higher set unsupervised overcurrent elements for higher fault current faults (this is considering the generator decrement curve). Differential was out of the question as the microgrid is over about 30 miles of MV distribution and high speed communications are non-existent. We recently had a fault and the protection operated as designed.
Batteries would likely have a similar problem as typical fault currents are often 10% above maximum load current, at least at the MV level.
At the LV level, depending on the size of the feeder you may not need to change settings depending on the available fault current while islanded, you might need to estimate your LV faults before making that decision if you haven’t already.
You might have trouble with inrush with low set instantaneous relaying.
If your fault current varies quite a bit (solar inverters on/off during day/night) you might need to combine different protection methods - overcurrent for higher fault current scenarios, time coordinated under voltage for low fault current scenarios for example- I’d probably plan for that in your estimate to be conservative.

 

[thermionic1]

How about a smart relay (SEL, GE, etc) that operates a shunt trip coil on the LV breaker. Some relays offer unconventional instrument transformer options that may help in tight spaces.

 

[JezNZ]

How about a smart relay (SEL, GE, etc) that operates a shunt trip coil on the LV breaker. Some relays offer unconventional instrument transformer options that may help in tight spaces.

Yes - it is an option, but it seems like exorbitant cost to bring this level of protection down to the LV scale. This microgrid is a residential community, so small LV breakers.

I have done some modelling in ETAP to date, and the issues with the limited available fault current don't propagate too far down into the system into the residential properties themselves (low earth faults may be an issue - readily solvable with RCDs). The problems identified in modelling are mainly at the transformer secondary circuits and points of connection of the residential properties.

 

[JezNZ]

The major challenge I had was differentiating load current from fault current when trying to achieve selectivity. With synchronous generators the minimum fault current can be less than load current, and in order to trip for all faults the overcurrent set point had to be less than the maximum generator load current. I was able to get sufficient selectivity using voltage supervised discreet overcurrent elements with coordinated fixed time delays along with higher set unsupervised overcurrent elements for higher fault current faults (this is considering the generator decrement curve). Differential was out of the question as the microgrid is over about 30 miles of MV distribution and high speed communications are non-existent. We recently had a fault and the protection operated as designed.
Batteries would likely have a similar problem as typical fault currents are often 10% above maximum load current, at least at the MV level.
At the LV level, depending on the size of the feeder you may not need to change settings depending on the available fault current while islanded, you might need to estimate your LV faults before making that decision if you haven’t already.
You might have trouble with inrush with low set instantaneous relaying.
If your fault current varies quite a bit (solar inverters on/off during day/night) you might need to combine different protection methods - overcurrent for higher fault current scenarios, time coordinated under voltage for low fault current scenarios for example- I’d probably plan for that in your estimate to be conservative.

Yes that is exactly my problem. Regarding variations in fault current, there will be a community-level BESS (tied into the MV network) which supports the network in island mode (and also performs energy arbitrage under normal conditions). For modelling purposes, I've assumed that the solar inverters are offline under fault and the BESS is the only source of fault current.

I'm interested in how you achieved the voltage supervised discreet overcurrent elements while also achieving ride-through? Or did this not apply to your network? While I would need to do some model, I could foreseeably see an external fault false tripping such a system, with the embedded system providing overcurrent and the external fault causing voltage droop?