Step-Up to Step-Down Transformer Winding Configurations

[Milo902]

Hi,

Long-time lurker, first-time poster here with a configuration I would like to get the hive opinion on.

I'm designing a 480kW, 480/277V PV system w/ 60kW Solectria string inverters. Although it's interconnecting into a 480/277V switchgear, the interconnect is located approximately 2,000' away from the array. I wanted to avoid stepping the voltage up, but it's just not practical to run 480V for this distance. I'm planning to step up to either 4,160V or 12,470V (most likely 4,160 but it will be a simple economic analysis) to transmit the power to the interconnection before stepping back down to 480V.

My proposed configuration is to step from 480/277V G-Wye to 4.16/2.4kV G-Wye at the array, then from 4.16/2.4kV G-Wye to 480V delta at the interconnect. The 480V side of each transformer would be protected by a 3-pole breaker, the 4.16kV side of each transformer would be protected by built-in expulsion and current limiting fuses.

Here's the thought process/justification:
On the utility side, it is prudent to specify an un-grounded configuration to avoid excessive neutral currents in the event of voltage imbalances. A delta configuration allows zero-sequence current to circulate, preventing zero-sequence harmonics generated by the PV inverters from passing through and providing an effective ground for the 4.16kV neutral. The 4.16kV system would be G-wye to allow the fuses to operate in the event of a ground fault on the MV system.

The inverters require a 4-wire system, so this constrains the primary (inverter) side of the step-up transformer to a Wye-G config. Given that the utility is isolated at the step-down transformer and there won't be any voltage/current monitoring on the MV system, the G-Wye 4.16kV configuration would minimize any chances that a lost-phase on the MV system would go un-detected by the inverters. Although a Delta winding would work, I believe this introduces the possibility of the inverters continuing to produce power even if a phase is lost on the MV system due to the single phase fusing used for protection.

If there is a phase lost on the interconnection side than this may not be detected by the inverters due to the delta winding. Unlike the MV feeder, there will be a revenue meter on the interconnection side so the voltages can be monitored and addressed if there is an issue. Also the facility would notice this phase loss as well.

This appears to be a relatively straight-forward design but I like to get alternate opinions, any thoughts?

 

[electrofelon]

Perhaps this is relevant, just found it in an inverter manual and thought of your post:

If the neutral on the Utility Side is grounded, the transformer core
structure must be 4 or 5 limbs to detect an open phase condition on
the Inverter Side of the transformer. Special detection configurations
are required to implement Inverter shut-down on loss of utility phase
if the transformer core is a 3-limb construction.

 

[kwired]

60kW @ 480 volts is about 72 amps.

I've run many 100 HP irrigation well motors with supply conductors in the 1500 foot range - and a time or two maybe closer to 1900 feet. Those typically are 118 FLA on the nameplate so even more kVA then you have. I've increased conductor size for voltage drop, I don't see using MV cable, associated gear plus two transformers costing less then increasing size of conductors, especially aluminum conductor like we usually are using for those irrigation feeds. 250 kcmil aluminum is what I always have run to 100 hp at 1300-1500 length, longer then that I do some calculating to see what I have, Significantly shorter I might do some calculating also.

 

[jaggedben]

60kW @ 480 volts is about 72 amps.

...

Yes but evidently there are eight inverters (or perhaps six or seven), and I'm sure he would be combining the output before any step-up. Would you still make the same comments at 500 amps?

 

[electrofelon]

The best deal without stepping up would be to send the dc strings back, then you would be working with ~700 volts. PV wire is available in Aluminum. Combining 1 inverters worth of strings, if one used 350MCM, that would be in the ballpark. Now would be the time to reread that article in solarpro from a while back about the economics of not going crazy to keep voltage drop low Last I checked 350 AL THHN was 1.05 per foot, but we would need PV wire. WAG that it is $1.50/ft, that would be $6000 per inverter, with no EGC. Multiply that times 8. A 500KVA transformer is around 13K IIRC. 15KV primary cable is 2 something a foot. $2.50? Need 6000 feet = 15k. Wow its actually kinda close. Might be worth a closer look. One thing you can usually expect with MV is you will pay a lot for it if you cant do it in house. Dont forget your transformer losses. .5% each 24/7 plus load losses.

Disclaimer: super quick and dirty analysis, take with a grain of salt.

Edit: Maybe send the 480 back. Could skip the expensive PV wire, but would need three conductors (dont need a neutral for the solectria inverters, see my other post) works out about the same and it would be less of a bastard system and avoid the combiners and hassle of changing up and down to 350 PV wire.

 

[kwired]

Yes but evidently there are eight inverters (or perhaps six or seven), and I'm sure he would be combining the output before any step-up. Would you still make the same comments at 500 amps?

I missed that there were multiple units.

Now that changes my thoughts to if you are going to have medium voltage anyhow to look into either having POCO run a service to the PV site or subscribe to MV as your service voltage.

 

[Besoeker]

60kW @ 480 volts is about 72 amps.

Per the OP, it is a 480kW system.
Still don't think I would go for the step up/step down option. Too much extra complication and additional points of potential failure. Not to mention the additional skills/qualifications/authorisation required.

 

[electrofelon]

Per the OP, it is a 480kW system.
Still don't think I would go for the step up/step down option. Too much extra complication and additional points of potential failure. Not to mention the additional skills/qualifications/authorisation required.

Quick and dirty comparison of "trading" transformer no load losses for more voltage drop. I get you could have 4% MORE VD without tranny than with to break even. And that is just no load losses.

 

[kwired]

Per the OP, it is a 480kW system.
Still don't think I would go for the step up/step down option. Too much extra complication and additional points of potential failure. Not to mention the additional skills/qualifications/authorisation required.

Certainly something to consider, I initially had blinders on and only saw 60 kW - definitely don't think the step up/step down would be worth the investment for that kW level.

I know a fair amount of things associated with over 600-1000 volts applications - but would never install such equipment without additional training first. I'd want more $$ for such skill set if I had it also, not to mention I would likely have some equipment that goes along with doing such work.

 

[Milo902]

Thanks for the input.

The system consists of (8) 60kW inverters, which at 480V is approximately 584 amps. (6) 500 Cu conductors/phase gives me approximately a 2.3% voltage drop at full load. This would be in addition to the voltage drop from the inverter to the AC combiner box (one run is up to 400'). The interconnect is to a customer switchgear, so I can't go directly into the POCO's 12.27kV system.

Regarding AC vs DC, moving the inverters to the interconnect would require DC combiner boxes at the arrays (the Solectrias have direct fused inputs) to the tune of $800 or so each. On the conductor side, it would need 500 Cu as well to keep voltage drop around 2%. The advantage would be only (16) conductors are required instead of (18), but it would be more expensive PV wire. In my experience 480V AC vs 1000V DC tends to be pretty similar in terms of line losses/cost and this is no exception.

Good point on the neutral, on the Solectria I can leave off the neutral and just use a jumper from the EGC to the neutral terminal. I will have some small single phase loads (under 1kVA) at the array for a DAS system. I could use a 480V L-L xfrmer and then there will not be any L-N loads.

I think the next step is to look at my hourly plant output over the year and compare losses and equipment costs. The MV transformers will run 12-19k each, plus concrete pads and additional installation costs. If it's looking pretty close, there are certainly advantages to avoiding the transformers and simplifying the design and install. Liquid fill transformers also have significant lead times. From a design perspective I'm comfortable w/ MV equipment, but there's a lot to consider with respect to winding configuration (hence the questions).

I'll report back with some numbers after I get lost in excel for a few hours.

 

[Besoeker]

I'll report back with some numbers after I get lost in excel for a few hours.

Good idea.
One other thing to consider, and you probably have, is how the insolation level varies throughout the day and to zero after sunset. How would you deal with transformer light load/no load losses under those conditions?

 

[Besoeker]

I missed that there were multiple units.

Maybe you, I and Jaeref need to go to the same optician...........

Sorry mods - couldn't resist.

 

[electrofelon]

Good idea.
One other thing to consider, and you probably have, is how the insolation level varies throughout the day and to zero after sunset. How would you deal with transformer light load/no load losses under those conditions?

Right I am not exactly sure how to calculate that. I Assume there is some software that can model those loading conditions. Basically though, using the max inverter output current for your voltage drop calculations will give unrealistically high energy loss numbers. Also, as I mentioned in an earlier post, there was an article about VD and solar systems in one of the trade publications a while back and they concluded that most systems are designed with excessively low VD numbers to the point that it is not the best economically overall.

Edit: "voltage rise" is also an issue though and can result in the voltage rising above the inverters voltage window, so that needs to be taken into account.

 

[ggunn]

Right I am not exactly sure how to calculate that. I Assume there is some software that can model those loading conditions. Basically though, using the max inverter output current for your voltage drop calculations will give unrealistically high energy loss numbers. Also, as I mentioned in an earlier post, there was an article about VD and solar systems in one of the trade publications a while back and they concluded that most systems are designed with excessively low VD numbers to the point that it is not the best economically overall.

Edit: "voltage rise" is also an issue though and can result in the voltage rising above the inverters voltage window, so that needs to be taken into account.

Voltage drop and voltage rise are the same thing. Due to resistance in the wire the voltage is higher at the inverter than it is at the service when the inverter is producing. Whether it is rise or drop depends on which end of the conductors you take as the starting point for your measurement. Conductors that are too small can cause the voltage at the inverter to rise above the top of its operating voltage window and shut down your system.

FWIW, the voltage at the service can be at the top of the POCO's window of operation as well, which will exacerbate the situation and may even cause inverters to shut down where you didn't think there would be a problem. In a few cases we have had to install buck-boost transformers to fix it.

 

[electrofelon]

Voltage drop and voltage rise are the same thing. Due to resistance in the wire the voltage is higher at the inverter than it is at the service when the inverter is producing. Whether it is rise or drop depends on which end of the conductors you take as the starting point for your measurement. Conductors that are too small can cause the voltage at the inverter to rise above the top of its operating voltage window and shut down your system.

FWIW, the voltage at the service can be at the top of the POCO's window of operation as well, which will exacerbate the situation and may even cause inverters to shut down where you didn't think there would be a problem. In a few cases we have had to install buck-boost transformers to fix it.

Yes I am aware that VD and VR are the same thing. I was just trying to say that VD isnt just about economics, it may need to be considered for voltage window also. One could have a set of circumstances where 6% VD is the most economical, but that would cause voltage window issues.

 

[Besoeker]

Right I am not exactly sure how to calculate that. I Assume there is some software that can model those loading conditions.

Probably there is. I have had to incorporate transformer losses in bids where I had to provide guaranteed efficiency. I divided it into Cu loss and Fe (magnetising) loss.
My assumption at the bid stage was 2% total loss of which 0.7% was Fe loss and the rest Cu. The Fe loss was assumed to be constant and the Fe varied as the square of the current.
Don't know if that helps.

 

[electrofelon]

Yes I am aware that VD and VR are the same thing. I was just trying to say that VD isnt just about economics, it may need to be considered for voltage window also. One could have a set of circumstances where 6% VD is the most economical, but that would cause voltage window issues.

I meant to say "could". Was not trying to generalize that 6% would never work.

 

[ggunn]

I meant to say "could". Was not trying to generalize that 6% would never work.

Gotcha. Conversely, if the service voltage is on the high side, even a 2-3% Vd can cause voltage window issues.

 

[electrofelon]

......(6) 500 Cu conductors/phase.......it would need 500 Cu as well to keep voltage drop around 2%. .....

If I was paying for this, I would tell you to use aluminum conductors :happyyes:

 

[Milo902]

Alright, I did some spreadsheet work to get a picture of losses over the year. My PV modeling software puts out hourly data which makes it easy to build a spreadsheet to tally losses based on actual plant loading.

As there will be a main AC combiner panel located at the array regardless of interconnection method, started my calculations at that point. For the 480V system, this is simply calculating current based on real power output (as the output does not include voltage, nominal system voltage was assumed for all calculations) and using that to estimate i^2r losses for the long run. Since this is calculated on an average hourly output, watt losses = watt hour losses, simplifying calculations. I revised my conductor to (6) 1000 kcmil AL/phase to keep prices in check.

480V system: Maximum voltage drop = 2.25%, yearly losses = 4,432kWh. This represents a 0.52% yearly loss, quite acceptable.

For the 4,160V system, the conductor losses are calculated the same way (of course) and are negligible (about 60kWh). For the transformer, I got a few bids including a DOE efficiency for the step-down (as required) and a non-DOE for the step-up (doesn't fall under DOE guidelines). The main difference is the standby losses with the DOE transformer about 245W no-load and the non-DOE about 1200W no-load. I used DOE for both as the extra cost would quickly pay for itself in losses. The transformer no-load losses are there 24/7 and thus can be multiplied out over the year (245W * 8760 hours * 2 transformers = 4,292 kWh). As we can see the standby losses for the transformers are pretty much identical to the resistive losses of the 480V run. For the load losses, the transformer was quoted with load losses of 5,200W at 500kW. As load losses are resistive, we use the i^2r formula to estimate losses at lower outputs (take the output in kW over the rating of 500kW, square the ratio, multiply it by the 5,200W load loss).

4,160V system: Maximum voltage drop = 1.33%, yearly losses = 16,150kWh. This represents a 1.9% yearly loss and is in line w/ the rule of thumb for transformers of about 1% loss/transformer yearly.

Based on this alone, it leans pretty heavily in the favor of the 480V system as rough estimates put the install cost between the two options within a few kilobucks of each other. Trenching in (6) sets of conductors for that distance isn't a huge concern, but there are a few spots we need to directional bore which introduces some headache. On the other hand, two more pieces of equipment introduce their own headaches.