Upgrading a FIT/MCS PV+ST system to PV only + battery

[wookey]

We installed solar thermal in 2008, and 2.9kW of grid-tied PV in 2009.

Since then the house has has IWI, EWI, 3G windows, an extension, MVHR, loft insulation and dramatically better airtightness so is most of the way to EnerPHit. (Useful talk I gave on that here: http://wookware.org/house/retrofit/ (youtube link there if you prefer to be tracked as you watch :-) ))

The solar thermal is now a bit of a liability as it makes the house too hot in the summer. We could fit bigger heat dumps and a better hot water tank, but lovely as free hot water from the sky is, heat you can't turn off in summer is less than ideal, especially at this latitude (54 degrees). Some hand-wavy sums suggest that just using the roof space for PV instead with heat-pump-heated hot water is just as efficient, much more flexible (electricity is more useful than heat), makes more money (the heat is worth maybe £80/yr, the extra electricity £400-odd). Carbon sums are harder, especially with the battery, but I think it still makes sense, especially in the context of heat pump and EV which should arrive over the next 2 years.

So, as we currently have scaffolding up it seemed like a good time to have a redesign. We want to add a battery anyway now that they are cost-effective even for low users. The FIT has 9 years still to run and being a really early one the index-linked rate is now ridiculously good (79p/kWh generated) so keeping that is important, even though it makes the system design trickier. A few threads on here about upgrading FIT systems were very helpful.

So after a couple of weeks research I have come up with the design described below, which is a bit unconventional (Victron kit with the new panels DC-coupled). I was expecting to find a few AC-coupled hybrid-inverter solutions to compare with, but given my requirement of zero dependence on external servers, I couldn't actually find _any_ other options, which was surprising. So asking here if I have missed something before I spend a load of money (and a few other queries about other details I'm not quite sure of yet).

Existing system:
Simple MCS install of 2.88kW (16 Sharp NU-180E1) with a 2.65kW Fronius IG30 inverter, from 2009 (right at the start of
FIT). Fitted in a U-shape around pre-existing solar thermal (40 vacuum tubes) install):
http://wookware.org/house/retrofit/pics ... .jpeg.html
Roof faces 158 degrees. That has an export meter and we export about 85% of generation (under SEG not FITs now).

Considerations are:

• * preserve existing FIT (79p/kWh generated) for remaining 9 years
  * maximise self-consumption/PV generation
  * DNO has applied 5kW G100 export limit
  * avoid dependence on proprietary software or remote servers, whilst maximising control (I am keen to avoid the risk of a supplier folding and some of my equipment becoming either paperwights or significantly degraded).
  * annual consumption is currently low: 1100kWh electrical, (and zero gas), but 3-5kW heat pump and EV likely in next 2 years which will put it up by about 1000kWh for the heat pump and another 2000kWh for an EV
  * roof is 8.94m x 4.35m
  * smartmeter installed, so can utilise smart tariffs

My proposed plan is:

1) Replace existing 16x180W 14% Sharp panels with 7x410W Longi Hi-MO 5m 410WP ( LR5-54HPB-410M) (
These are same capacity within 10Wp (2870Wp instead of 2880Wp)
Keep Fronius IG30
That should comply with https://www.ofgem.gov.uk/sites/default/ ... _FIT_Gener\
ators_V18.pdf chapter 3, and that part will operate as before through its own existing generation meter.

2) Add:

• 7*435W Longi Hi-MO 6 (LR5-54HTH-435M) (Same physical size as the 410W HiMo5 so fits nicely). Array will be 7x2 portrait.
  Fogstar Seplos Mason 15kWh battery (EVE MB31 cells) These are amazing value at £1800 for 15kWh. The BMS will pretend to be anything you like: (22 protocols over CAN or RS485)
  Victron Multiplus II 48/5000

The panels could be DC or AC coupled.
DC would be Victron MPPT into batts
AC would be AC inverter into mains

Control with VenusOS on Pi (or CerboGX)

So the basic idea is that I can fit twice as much PV on the roof as I currently have, but one half has to match the existing FIT array so needs to be 2.88kW. The remainder would be 7 435W panels (i.e 3.05kWp)



This leads to various questions/options

1) DC-coupling vs AC-coupling efficiency/cost/convenience?

Sources generally say DC-coupling is more efficient, but I'm not sure that's true if you export most of the output. We currently export 85%. I'm struggling a bit to work out how that changes by adding a 15kWh battery, given average usage of 3kWh/day + hot water of ~2kWh/day.

DC-DC MPPT controllers are 98-99% efficient.
The Multiplus II is >95% from 400W to 2kW when inverting (outputting) 93% at 3kW, 90% at 4kW
when inputting (charging from AC) it is >92% efficient from 600W to 2kW.
Battery round-trip should be about 95% coulombic efficiency.
AC PV inverter is ~93% typical efficiency? That's true for my Fronius. I don't have a power curve to see how it changes from 0-2.65kW

So:
Charging from grid @2.4kW, use locally: 0.91x0.95*0.95=82%
Use/export AC PV as generated: 93%
Use/export DC PV as generated: 0.99*0.95=94%
Use/export AC PV via battery: 0.93*0.95*0.95=84%
Use/export DC PV via battery: 0.99*0.95*0.95=89%

Which does suggest that The DC-coupled PV is no worse and probably slightly better than the AC coupled.

AC coupling with a cheap inverter would cost less in kit than the two DC MPPTs I need for these 7 panels.

Has anyone worked out how to use https://github.com/TrystanLea/solarmodel ? I think that can do the modelling, but I'm not sure how it's supposed to be run - through a webserver presumably? Which file is the top-level page? Maybe it only works from within EMONCMS?

2) Limiting export to 5kW

If I have multiplus II with a CT meter on the outgoing line, can it control other AC PV inverters on the line so that all of them together will stay under 5kW? Victron docs talk about feature of Fronius inverters 'zero-export' which makes this reliable. If there is only the one existing (fronius IG30) 2.65kW PV inverter then the Multiplus should be able to throttle-back to ensure that it never sends out more than is currently being exported. But if I had two (AC-coupled PV inverters) that could go over 5kW without the multiplus exporting anything. That sounds like it isn't going to work without active control of those inverters, and couple be an argument for DC-coupling the remainder of the panels.

The multiplus II can control the grid frequency to make other inverters throttle back, but presumably it can only do that when disconnected from the grid, as it has to stay frequency-synced to the grid when it's connected. Right?

3) MPPT voltages 3x2? 3+4 3x2+1? Get an RS?

Victron DC coupled MPPTs are very efficient (98-99%) but also quite low-voltage in comparison to typical PV inverters. So attaching the new section of the array is annoyingly tricky: there are plenty of amps but not enough volts. 7 435Wpanels has peak voltage (-10C) of 302V which is too much for the 150/8 and 250/* MPPTs. The MPPT RS can do it (450V/100A) but that's a £960 machine so adds a lot to system cost.

One £300 150/50 can do 2 sets of 3 panels in parallel which is good, except that we have room for 7 so it would be nice to use 7. I could use a 150/50 and a 250/50 but that's £300+£360 so is getting expensive again. One extra panel is worth about 390kWh/yr so at least 55 quid on current SEG, more if it saves importing. So the extra MPPT+panel pays for itself in 5-8 years, so I think that is worth doing.

I did think about wiring up 7 panels as 1 string of 3 and another of 4 and parallel them up? Or having 2 parallel strings of 3 plus 1 more in series? In both cases I'm pretty sure you get the same power out as 6 panels so it's pointless.

Are there any other cost-effective ways to solve this? Should I look at getting a Multi RS solar 48/6000/100-450/100 instead? Does that do all the same things as the Multiplus II as well having a pair of high-voltage MPPT? Cost is slightly higher but then it's 20% higher power.

Are there other 410W panels of about the same size but with significantly lower open circuit voltages so 7 will go in <250V?

4) If AC which inverter?

Connecting the panels over AC avoids all the faff with voltages and MPPTs above and costs £400-£950 depending which inverter one picks. But I couldn't find any PV or hybrid inverter which gave good local control. Is there one? Everything seems to be phone apps, remote servers run by the suppliers, or pushing buttons on the front of the machine.

Necessary functionality includes

• * complying with the G100 5kW export requirement.
  * compatible with Fogstar rather than own expensive proprietary battery
  * timed charging from grid when import rate is low/negative
  * Export of all surplus PV once battery and hot water are full.
  * Export to grid when price is high (and it will be sunny tomorrow)
  * Interfacing with OpenHab or Home Assistant

Are there really _no_ AC inverters which allow fully local control?
There are loads of inverter suppliers. Have none of them been reverse-engineered yet?

5) Grid-Parallel or in-line Multiplus?

The Multiplus II can be installed either T-ed off the incoming AC line 'Grid parallel', or with all AC going through it. Does anyone have the pros and cons of this (for places that have reliable grid) straight in their heads?

My understanding:
Grid-parallel:
* Makes siting the Multiplus II much easier because it doesn't need to be near the incoming line
* separate CT meter needed to see power from AC-coupled inverter
* Essential loads max power is ~4.4kW

In series:
* Is max power on AC limited to 4.4kW? or is that only essential loads and grid-out can do 12kW (50A?)
* need to worry about 'factor 1.0' rule https://www.victronenergy.com/live/ac_c ... or_10_rule (no more than 5kW PV on 5000VA machine. (I will have 5.9, but could couple the old AC system direct to grid to deal with this, or get the Multi RS solar 48/6000/100-450/100)
* No separate CT

As we have reliable grid here, maybe there is no advantage in the series configuration - it's only useful for off-grid people?

OK. Apologies for this very long post. I hope at least that it is clear.

[Joeboy]

Quick question to chase down. Do you have to match your existing fit array? I thought you could oversize and claim the fit percentage only?

I'd ask around here on the Fogstar value for money. Not sure it's quite as good as all that. A few diy builds here with different needs have beat that price. All depends on how diy you want to go? Current prices excl bms, cabling, shipping is 51 us dollar per lf280k unit. I have 15 per stack but piggybacking on an existing bms system for control.

Not sure on local access hybrid inverters, they do seem.to.be online access via smartphone. Maybe a cheap tablet hooked into the hybrid units own WiFi would suffice?

I'll be selling a Victron.150/60.solar controller in a few weeks. Might be of use to you and help keep prices down? I kept to strings of three panels and Voltage matched as I went adding more strings.

Can't go bigger in panels to match a 2x3 set up to a 7's output?

A great read and quite an upgrade, I'll be back later. Do you have a budget you are hoping to come in at?

City plumbing are very competitive for PV if you've not looked there.

Lastly, your generation from fit, is the 79p the combined figure for generation & deemed export? A decent number of us have dropped the deemed export side of fit and went for actual. Might not work for you though?

[MrPablo]

Depending on the hybrid inverter, you may be able to handle logging and control over modbus or the like.
I have a Solis hybrid EH1 that I don't think I would recommend for a power user, but this is now controlled via a self made control stick using ESPhome.
I haven't actually looked at the inverter screen in months.

One benefit of AC coupling is the ability to have higher peak output - that could be useful to cover local consumption if needed.
Of course, you then have the idle power consumption of a hybrid inverter to consider as well.

[Stinsy]

1) Your plan is a good one and will work. If not for FiT you’d swap your existing inverter for a bigger hybrid, but you need to maintain the AC from your old system flowing through your generation meter to keep the free money flowing. Therefore AC-coupling is the best (and only) solution.
2) Solar thermal has been obsolete for a long time. I’d be happy to bore you with the long list of drawbacks but seeing as you’re already a convert I’ll hold off (unless you really want me to!). Unfortunately it has stuck in official documentation as the best thing since sliced bread! Even today solar thermal makes a massive (positive) difference to EPC scores whereas solar PV and a diverter gets you marked down for “electric DHW”! PV is ridiculously, stupidly cheap nowadays. The panels are so cheap as to be almost free, inverters and batteries too have come down a lot (unfortunately solar installers have filled the gap with unjustified rates). So filling any available space is worthwhile.
3) You don’t want “dependence” on cloud servers. I get that. However few inverters are genuinely dependent on an internet connection and the server in China still being up. Every one I’ve knowledge of can function just by pushing buttons on the front, or by connecting a computer to it locally. Having internet connection and servers functioning just adds a layer of convenience. There was a guy writing his own software for the LuxPower inverter (that I use) to keep internet based monitoring/maintenance should the servers go down, I’ve not kept up to date with that project…
3) You’ve designed a really nice system with premo Victron components. You could achieve something very similar with a made-in-China hybrid inverter from Sunsync or Solax or whoever and a set of LF280k cells. If you’re not bothered about cost and RoI then no worries, but you should at least understand how much extra you’re paying.

[Gareth J]

Do you need to move/replace the old panels? What is the gain? Space? Location? They'll be a bit more efficient for sure but that seems like it will be hard to recoup.

If you're going the export limitation route with DNO anyway, you may as well tessalate the maximum kWp (within reason!) you can on the roof and use an inverter that meets export limitation requirements (g99?) to modulate to the 5kW requirement. For simplicity, on the new inverter, may as well make it a hybrid?

It will be worth doing a back of envelope calculation to decide if, in summer, it's more cost effective to heat water via a heat pump or resistive loads + diverters. With SEG, or a smart outgoing tariff, it'll probably make sense to heatpump it. But wear on a relatively expensive heat pump may be an overlooked factor too.

[Fintray]

In series:
* Is max power on AC limited to 4.4kW? or is that only essential loads and grid-out can do 12kW (50A?)
* need to worry about 'factor 1.0' rule https://www.victronenergy.com/live/ac_c ... or_10_rule (no more than 5kW PV on 5000VA machine. (I will have 5.9, but could couple the old AC system direct to grid to deal with this, or get the Multi RS solar 48/6000/100-450/100)
* No separate CT

Pass through on the Multiplus II 48/5000 is 50A but this can be increase with the power assist function which will add the output of the inverter to the incoming supply effectively giving up to 16kW output.
I have used 3 Multiplus II 48/5000 in parallel for my system as I want the inverters to be able to supply the whole house.

[AGT]

Reading with interest as in the same boat as Wookey.

[wookey]

Quick question to chase down. Do you have to match your existing fit array? I thought you could oversize and claim the fit percentage only?

Yes, I believe you can, but you have to deal with the FIT provider and get them to set some factor on your readings. But to be frank they have generally been effing useless, and we simply don't trust them to a) do that in less than about 2 years and b) not forget every time and make us go through the same rigmarole repeatedly, and maybe just have a hissy fit and declare us not in compliance.

Maybe it will be fine, but it's a lot easier to not have to tell them anything and just carry on with matched-sized panels. So that's why it's worth doing it this way.

I'd ask around here on the Fogstar value for money. Not sure it's quite as good as all that. A few diy builds here with different needs have beat that price. All depends on how diy you want to go? Current prices excl bms, cabling, shipping is 51 us dollar per lf280k unit. I have 15 per stack but piggybacking on an existing bms system for control.

Yes. I could buy bare cells and a box and a BMS and build a batt up myself (I did this for my moped back in 2007). I admit I haven't priced that up, but the Seplos Mason kits seem like a nice neat way of getting matching kit, and not being ripped off on cells that aren't what they say, for a very reasonable amount of money. And I've bought it now, so that's one thing that is decided. :-)

I'll be selling a Victron.150/60.solar controller in a few weeks. Might be of use to you and help keep prices down? I kept to strings of three panels and Voltage matched as I went adding more strings.

Yes. I'm interested. Although I've just worked out that a 150/35 is sufficient for the lower-voltage string. (and £160 rather than £306).

Can't go bigger in panels to match a 2x3 set up to a 7's output?

I don't think it'll fit on the roof, because I have to have 7 for the FIT half, so they interlock:
(separated slightly to illustrate the halves)

But maybe I can try a bit harder - there is some vertical slack. Ah yes 6 x Longi 530W (3.18kW)
fit:
So that's a bit more generation. But Voc is 48.72 @25C and -0.23%/C so 56.8V design voltage at -10C, so 3 panels won't fit in 150V. But I guess I can do all 6 in a 250/60 and the panels only cost £50 more so that is a win. Looks a bit odd, but I don't really care. I'll need to check they really will fit - it'll be a bit tight, and I'll be ignoring the MCS edge-allowance for wind loading.

Do you have a budget you are hoping to come in at?

Not particularly, but I do like a good combination of reasonable quality without paying over the odds. It looks like it'll be a bit over 4 grand, which is fine.

City plumbing are very competitive for PV if you've not looked there.

Yes, I priced some things there. I have a Midsummer energy trade account, which seems to come out about the same, and they are good people. I might have to get the 410W panels from cityplumbing as Midsummer have sold out.

Lastly, your generation from fit, is the 79p the combined figure for generation & deemed export? A decent number of us have dropped the deemed export side of fit and went for actual. Might not work for you though?

We have had an export meter from day 1 so no deeming here.

[wookey]

Depending on the hybrid inverter, you may be able to handle logging and control over modbus or the like.
I have a Solis hybrid EH1 that I don't think I would recommend for a power user, but this is now controlled via a self made control stick using ESPhome.

OK interesting. I don't have a good feel for how much control the cheapo inverters actually make available locally. Nice to know that it is possible. What about it makes you wary of recommending it?

Of course, you then have the idle power consumption of a hybrid inverter to consider as well.

My neighbour has gone this route with a SunSync 4kW inverter and both idle consumption and roundtrip efficiency seem much worse than the Mutiplus II (which seems remarkably efficient given the old-fashioned real-transformer design).

[John_S]

Can't go bigger in panels to match a 2x3 set up to a 7's output?

I don't think it'll fit on the roof, because I have to have 7 for the FIT half, so they interlock:
(separated slightly to illustrate the halves)

But maybe I can try a bit harder - there is some vertical slack. Ah yes 6 x Longi 530W (3.18kW)
fit:

Is there any reason why you can't/don't have the 7 FIT panels all in a row and the 7 new panels in another row either above or below the FIT panels? Could 7 of the Longi 530s go in the second row?