diy solar

diy solar

Mitigation of demand from utility with solar

Like @Ampster if you find one let us all know -- i would be interested in at least looking at how it works ... I mean by its very nature it would need a transfer switch of some sort ... but the guy looking over my shoulder as I type says he has heard of these without transfer switches but their costs are in the 10's ... and uses software and blah blah blah - (which in my mind is till a transfer switch of some sort)

I put two simplified one-line diagrams here:

First diagram is a most common grid-tied system like I currently have with SRP. For simplicity I put jut one inverter with 3 MPT channels and skipped combiners, fuse boxes, disconnects and other equipment not important for concept understanding

Second diagram contains additional equipment shown with red. I believe it can achieve "Meter Zero" mode during peak time without transfer switch. It's DC-coupled with existing Grid-teed inverter via vacant MPT channel "C"

This is how it's all going to work:
  • Current transformers on solar and utility lines duplicate existing SRP meters here. They measure solar energy production and energy flow to/from grid at any particular moment providing data to computer. Based on those parameters I can calculate grid demand at any particular second
  • If time 8pm-2pm, it's non-peak time and electricity is free from solar or cheap from grid. Computer turns on relay to charge battery bank via BMS
  • If time is 2-8pm (peak time), battery charging is off. Electricity is expensive
  • If time is 2-8pm and Solar power >= power in grid line, there is no grid demand as it's 100% covered with solar. No action required
  • If time is 2-8pm and solar power < power in grid line, this is where I'm consuming expensive peak-time electricity from grid and demand charges apply. Following sequence of events will happen:
  • Computer turns on additional "off-grid" inverter to get 240VAC from battery bank
  • Double-voltage rectifier gets 480VDC from 240VAC
  • Computer-controlled PWM provides 90-480VDC to vacant MPT channel "C" in existing grid-tied inverter
  • MPT inverter believes sun is shining again and feeds that power to grid/loads via main distribution panel like it usually does with regular solar energy
  • Software in computer monitors power in grid line and sends command to PWM to increase or decrease power output from battery bank to MPT "C" in grid inverter. Its characteristics should simulate technical parameters of regular 480VDC solar array (e.g. 12 of 40V/300W panels)
  • Having feedback from grid line to battery power output, software maintains "Meter Zero" mode on grid line as long as battery lasts
  • If battery capacity is not sufficient to cover 100% of demand during on-peak time, software can be adjusted for some demand allowance. E.g. up to 3kW or so. Since demand charges in utility payment plan have three tiers (please see previous messages), first 3kW of demand will cause only ~$25 demand charge in monthly bill. It's pretty bearable and can be consumed
  • Bigger battery bank and more powerful second inverter can achieve 100% demand mitigation and true "Zero Metering"/TOU mode

Typical grid-tied system
1593448135037.png


DIY Meter Zero system:

1593448201973.png
 

Attachments

  • Common grid-tied solar system.pdf
    42.5 KB · Views: 3
  • DIY Meter Zero.pdf
    67.7 KB · Views: 10
Last edited:
Don't be too quick to dismiss that rate plan. Have you done the math about going on that rate and buying a smaller battery system that would allow you to self consume your solar and not buy anything at that peak rate?
You can also add solar panels a lot less expensively than inverter and battery capacity. However there is an optimum system design for that rate plan, just as there is one for your existing rate plan.

I compared E-13 and E-27. With my energy consumption trends I need to double my system from 12.96kW to ~25kW in order to achieve Zero utility payment. Even not considering investment funds, I physically don't have room on my roof for more panels

Yes. I did detailed calculation. With current situation E-27 (demand plan) is more beneficial for me even with demand charges than E-13 (non demand/peak rip-off) plan

Demand mitigation with batteries looks more achievable than non-demand/rip-off. I already cover with solar ~100% of my monthly electricity consumption. It's just needs to be shifted to off-peak time eliminating demand fees
 
I compared E-13 and E-27
I forgot that your solar production already covers your consumption. Is that true every month? With that low price for solar generation it would be tough to leverage TOU rates and use grid as a battery.
 
I forgot that your solar production already covers your consumption. Is that true every month? With that low price for solar generation it would be tough to leverage TOU rates and use grid as a battery.

I upgraded my system to 12.96kW in the beginning of April this year. So far it covers 100% of my monthly consumption and even sells some energy to grid. However I cannot tell yet about July-August (typical months with HIGHEST consumption around the year) due to HVAC energy hogging. Also daily daily generation dropped now from 105-110kwHr to 92-95kWHr due to triple digit ambient temperatures and inverters throttling to protect themselves from overheating

E-27 (demand-based) plan has fair prices from utility for buying and selling electricity. "Off-peak" time it's ~4 cent/kwhr and "On peak" time it's ~6 cent/kwhr. Their prices are identical for buying and selling energy. They don't make profit from it perfectly allowing grid to be used as a "storage battery"

Their main profit is from demand charges when sun is low or no sun and demand is highest and unavoidable. Unless customer has some kind of TOU management system or so dedicated to savings that sits in hot house from 5pm to 8pm with all lights and appliances off
 
A gas dryer would probably be the lowest cost and would be a benefit year round. If you could display the usage so that people are aware that using the stove or oven will put you over the 10 kw mark they may decide to wait. If the baked potato is going to cost $180.00 more at 7:30 than at 8:01 they may wait.
 
Now, when Summer is over, here is more detailed statistics of energy generation and usage in my household. Please see a screenshot from utility bill

"On Peak kW Max" - that's infamous monthly peak demand explained on 6/27/20 in this topic which I'm trying to mitigate

As I mentioned earlier, my solar equipment was upgraded in March from 7.8 to 12.96kW by installing more panels and second inverter. So counting overall it covers more than 100% of total energy consumption for March-October. And monthly bill for electricity continues to be 80-90% consisting of demand fees

Recently I installed part of equipment shown on diagram #2 from 6/29/20 above with PZEM-16 electronic meters and computer interface. Just for monitoring now. Will share some results soon if anybody interested to look at them and discuss further ideas

1604794654016.png
 
I turned on my DIY "Smart Demand" system on 10/18/20 with preset demand limit 5kW in PowerShell code running it. Diagram below shows how effective is demand mitigation with manipulation of just A/C compressor based on momentary and average demand levels during peak time

1604873559728.png

This is how it's working:
All measurements are done via PZEM-16 modules described above and transferred via MODBUS interface to Server-2016 computer with PowerShell code running 24/7. Sure, RaspBerry-PI or Arduino would be much cheaper. But I already have powerful server running at my home for different purposes unrelated to this topic. So I just decided to piggyback it for SmartDemand as well
  • Weekends and off-peak periods 8pm-2pm Monday-Friday when demand is not countable, system is just monitoring power flow every 3 seconds not interfering with any loads. Logs are saved to CSV-file for further analysis
  • On-peak time from 2 to approximately 5pm Monday-Friday if solar production is sufficient to cover demand, system is not interfering. However it becomes alert to intercept demand peaks
  • All fun starts between 5-8pm when demand is higher than solar production and it's causing demand charges from grid. System keeps measuring current demand from the beginning of each 30 minutes interval (starting at 00 minutes or at 30 minutes) using same algorithm as used by utility. If demand reaches preset limit (5kW as I configured it), then server gives a command to relay for turning off A/C compressor. That relay is installed between thermostat (Nest) and AC itself
  • Power consumption instantly drops by 5kW (load from A/C compressor) and demand value (as average from the beginning of cycle) goes down. That process is not instant and takes time (typically 5-15 minutes) depending of other loads running at house during that time.
  • When calculated demand value reaches 4kW (Demand limit - 1kW), program calculates how many minutes passed since moment of forceful A/C compressor powering off. If it's more than 10 minutes, relay restores circuit between thermostat (Nest) and A/C compressor allowing Nest to make decision whether A/C should run again or not.
  • If less than 10 minutes passed since last running of A/C compressor, system waits up to 10 minutes to protect it from overloads due to short-cycling
Certainly, October-November is not the best season for testing such system as ambient temperatures went down after Summer. But on those days when outdoor temperatures still reached 90-100F (please see diagram above), system managed to keep demand below 5kW (preset limit). Indoor temperature didn't went out of control for more than 1F as A/C was still running almost as long as needed, but in shorter cycles making sure not exceeding preset demand limit within each 30-minutes interval

Setting demand limit lower than 5kW is technically feasible. But it won't be very efficient as even without A/C typical demand from other loads in the house (kitchen appliances, lights, computer, TV etc) is constantly around 1.5-2kw

The ultimate test this system will get next June-August. I believe it can keep demand within 5kW. However the big question is whether A/C in such operation mode still can keep indoor temperature around 77F (comfort level for our family) while outdoor is 110F? That's a BIG question

Also few questions to HVAC specialists (if there are any reading this topic): Is 10 minutes cool-down time sufficient for A/C compressor between cycles? What is the shortest time considered safe for cycles of 5ton compressor operation?

*Slightly higher demand peak on 11/3/20 shown on diagram was caused by laundry dryer turned on during on-peak time. "Smart Demand" system couldn't compensate it as A/C was already turned off during that time. I'm thinking maybe installing another relay disabling dryer during peak time or operate a heating element in dryer similar way via "Smart Demand" system as A/C compressor. The caveat of that - it would require hacking into dryer's circuitry by installing a relay before heater and connecting it to SmartDemand system
 
Last edited:
Since you have Sunny Boy, consider Sunny Boy Storage. That is SMA's "powerwall", can be used to shave peaks.

Consider a generator and transfer switch. Fire it up each day when the hours for penalty rates appear. Maybe just wired in to air conditioner. Would be more ideal to feed it into the grid (generator going to PV or battery terminal of a grid-tie inverter) but that isn't as readily available as a product.

Your roof area may be full, but what wattage panels do you have, and what dimensions? (find a data sheet).
They are probably reasonably recent and efficient, but I can increase the output of my array by 50% just by replacing existing 12 to 14% efficient panels with new 20% efficient ones.
If you deliver lots more kWh during middle of the day, will that earn you a credit to offset the demand fees?

I don't have demand fees, but I do have late afternoon rates 3x the morning rates. So delivering power morning in addition to afternoon works for me.
 
Since you have Sunny Boy, consider Sunny Boy Storage. That is SMA's "powerwall", can be used to shave peaks.
I don't claim to know hardly anything about demand priceing. However I don't understand how this isn't the most simple fix for the issue. You add battery to capture extra solar output at peak solar time (not peak grid $$) store it in batts, and use it durring peak grid $$. You dont expect that battery to last days or even a half a day, just long enough to ride out the demand period.

On non-sunny days you use off peak grid $ to charge the battery before peak $$$ hit.

I was under the impression that existing SMA products did all that ^^ (just have to add them to the system)
 
Last edited:
This product is aimed at demand-charge avoidance and zero-export applications.
Full backup during grid failure us supported with an additional box that has transfer switch and other components.



If the primary demand is cooling during evening hours, a system that made ice during off-peak and used that rather than A/C during peak hours would be a low-tech solution that might be less expensive. But it isn't a readily available product on the market like this one that can just be connected to the house wiring, wherever it can fit.
 
I'm thinking maybe installing another relay disabling dryer during peak time or operate a heating element in dryer similar way via "Smart Demand" system as A/C compressor. The caveat of that - it would require hacking into dryer's circuitry by installing a relay before heater and connecting it to SmartDemand system
A simple electric dryer has 240V across the heating element but only 120V for motor/timer.
I think a relay interrupting one hot leg is all it would take.

That relay and one in the A/C thermostat connected to a signal "< 80% SoC" are part of how I plan to make my system avoid draining battery when running off-grid.

Fancy model dryers? All bets are off.
 
I already looked at "Sunny Boy Storage" SBS6.0-US-10 battery inverter. It looks promising for my goals and it's certified by SRP (my utility company). However my biggest concern is a 360V "high voltage battery" which is pretty special, expensive and not common for the market. I'm looking something similar that can use standard 24V or 48V AGM batteries widely available everywhere

Also I couldn't figure out how that SBS-6.0 device should be wired. 1-line diagram in installation manual shows parallel connection to solar SMA inverters. But in my case solar inverters are sitting behind a dedicated solar meter according to utility requirements. And their policies don't allow hooking up anything into that circuit
 
Last edited:
Sunny Boy Storage AC couples, hook it up anywhere in the house.
There is no transfer switch if you just add SBS; it charges some of the time and supplies power some of the time. So it can charge while PV is producing around Noon, and discharge during the afternoon and evening hours. Grid is still there for starting surge.
I think battery can be as small as about 3 or 4 kWh, and I think 3 batteries x 20 kWh each may be the largest at this time.

Optional Automatic Backup Unit is a 200A transfer switch and 120/240 transformer. That's where the wimpy 9300W 100 ms surge capability probably won't work or you. So you could have backup for other loads, not the A/C.

I have Sunny Island, which would start your A/C off a 48V battery just fine. Possibly a pair of them, but definitely four to deliver 44 kW for 3 seconds surge. But I don't think it is programmed to perform peak shaving. It's grid input current is programmable, could be set for 10A or 112A (with paralleled inverters), but I don't think that can easily be switched based on time of day. Disconnecting grid would be the easy way.

Some of the utilities require that a battery inverter which feeds the grid can only be charged by alternative energy like PV, not from the grid. That can be accomplished/accounted for with communication or current sensors, either when charging or discharging.

AGM - for deliberate daily cycling, do the math on cost per kWh. I went with AGM for backup, but my cost per kWh is higher than my utility cost. Could be only a low-cost lithium battery will make financial sense for you.
 
Hedges, thank you for information

Below is 1-line diagram from SBS6.0-US-10 manual and my own 3-line diagram approved both by SRP (utility) and AHJ (City of Phoenix). It was inspected and officially commissioned by both those sides in April. I just erased PII from it

If I understood you correctly, I can hook up SBS6.0-US-10 directly to main load distribution panel via dedicated breaker and that's it, isn't it?

1-line diagram also suggests additional energy meter (SMA-compliant). Where is should be installed in my case? I guess it should be between load distribution panel and main house disconnect switch left from that panel on my 3-line diagram. Will that work?
 

Attachments

  • SBS6.0-US-10 1 Line diagram.pdf
    67.8 KB · Views: 12
  • 3-line diagram (no PII).pdf
    60.1 KB · Views: 8
I've never hooked up a Sunny Boy Storage, only Sunny Island and Sunny Boy, so just going by what I've seen skimming SMA's manual and brochures.

Because SBS backfeeds the main panel it would be subject to the rules on PV breakers. Main plus PV not to exceed 120% of panel busbar rating, and PV breakers to be at far end of busbar. It feeds 28A, would be at least 35A breaker, typically 40A or 50A max according to data sheet.

You have 200A panel (not 225A busbar like some Square-D, unfortunately?) 175A fuse, so 65A limit on backfeed breaker. Your PV breaker is 60A so SBS can't have separate breaker, needs to go through that one.

PV combiner panel is rated 100A so allowed 125A total. It is fed by 60A fuse, has 20A and 40A PV breakers for 120A total.
Either need to upgrade that to at least 150A panel, or add another 100A panel for SBS. Would like to connect between 60A breaker in main panel and 60A fuse, would have its own fuse (say 60A.) But fused PV disconnect switch is probably meant to disconnect all sources of consumer generated power. So SBS has to be after that switch, and both PV combiner panel and SBS each need their on fuse/breaker.

Looks to me like larger PV panel is the way to go, and SBS connects to it along with Sunny Boy.

Or larger panel added before PV panel. Or add a 60A backfed "main" breaker to PV panel (it has 6 slots) and tap off separate fused disconnect for SBS.

I think the 120% limit may be waived for a PV combiner panel which only supplies power from PV, doesn't have loads. But with SBS, there are loads too.

Yes, the energy meter would go between your 200A disconnect switch and 200A panel. It is probably two or three AC current sensors the two hot (possibly plus neutral) go through. It also connects to the three wires, I would guess at a breaker in the main panel.
 
I'm still researching a DIY solution based on diagram #2 from my posting on 6/29/20 above. Unfortunately I don't have a vacant MPT channel in any of my two inverters. So I'm thinking about hooking up battery inverter with rectifier in parallel to one of solar strings and without PWM. It will give either 240V DC or 480V DC only when demand exceeds particular limit. E.g. 3KW and work similar way to controlling A/C compressor based on demand. E.g. If calculated demand is above 3kW, battery inverter turns off and feeds 5kW (const) to the grid and loads. Calculated demand goes down. As soon as it reaches zero or even negative 1-2kW, battery inverter turns off. Hence 30-minute demand will fluctuate around zero as long as battery lasts. So it will be used a more complex program algorithm to turn on/off A/C compressor and battery inverter to keep demand around zero

Since I would like to try hooking up rectified DC from battery inverter to MPT channel of solar inverter in parallel to solar string, will the latter be operable if sun is still shining, but output is weak during evening time?
 
So you're looking for a way to feed battery power into Sunny Boy's PV input?

One thought is rectified A/C as you suggest. I was thinking of that for a generator input. One concern I had is that if input capacitor has discharged low, when A/C is applied it drives a much higher current than PV string would have.

Some Sunny Boy have also been used with batteries e.g. recycled from Prius.

If you try to connect another source, your SB7.7-41 requires ungrounded array, must be galvanically isolated. So any inverter or generator can't have a DC path to ground. There is also a limit on capacitance to ground.

Think you're better off getting Sunny Boy Storage rather than trying to perform the same function with a Sunny Boy.

Another way could be to insert a pair of Sunny Island between disconnect switch and PV breaker panel. Connect your A/C to it as well. This gives PV/battery backup during grid failure. To coerce Sunny Island into feeding A/C load from battery, either open the disconnect switch, or just send a signal to Sunny Island's digital input which says, "You're fed by a generator now, not by the grid", and program minimal generator current. That way Sunny Island only draws a few amps from grid, provides the rest from battery.

Issue with using Sunny Island is you then have to choose between lead-acid battery (not particularly inexpensive per kWh of cycle life) and a supported lithium battery (expensive up-front purchase). Unless you do a cheap DIY battery and buy a compatible BMS.
 
My second inverter SB5000US-11 is negative grounded on DC. Actually there is a choice in it which side of DC to ground "+" or "-". I chose "-" for convenience. Also P2.1 on page #12 of its Installation Guide it Says that "Sunny Boy is suitable for use with fuel cells, small wind power plants and other DC current sources" besides of actual PV modules. So it looks like that specific model is not critical for DC quality as long as it lies within 250-600V DC. So rectified 240V A/C *1.41 = 338V DC after capacitors is perfectly within operation limit

From my observance during the day PV array hooked up to that inverter operates at voltage 360-450V DC fluctuating during the day. All PV modules have Tigo TS4-O optimizers in each module and entire string operates at level providing maximum power output while voltage is fluctuating depending of insolation

Interesting situation, what happens when const 338V DC is introduced in parallel to PV string with Tigo's? In theory PV string gives higher voltage than rectifier even with low sun. Hence that voltage applied back to diodes in rectifier might shut it down effectively disabling feed from battery system. However from other side, Tigo's designed to respect parallel connection of multiple PV strings. So I hope that Tigo's can adjust PV voltage output to 338V+few more volts still allowing feed from battery system and keeping PV feed. However this is still theory to be verified

Any comments on it?

Benefits of such wiring - no additional Sunny Boy Storage or island needed and no transfer switch. System remains AC coupled and won't exceed approved power capacity at any circumstances. Total power will be still limited by power of inverter. So it won't cause frowning from utility that I can feed to grid more than approved under some circumstances

The only remaining investment for such solution is a budged "off-grid" battery charger/inverter (something similar to FuelZero), batteries and rectifier with capacitors
 
Or maybe a square wave inverter, since you're rectifying the output anyway?
But your 338 Vpeak AC is going to be symmetric around ground unless isolated. Needs to be +338V to zero for that to work.
Do these newer Sunny Boy not have the GFCI circuit my older ones do? (a fuse to GND). Not sure how your circuit may affect it.

How about a high voltage battery, rather than inverter? (But then you'll really have to limit the capacitor charging current)

But you're right, if it can take small wind generator, then it can take an inverter.
However, that is a different algorithm to select in the Sunny Boy, think it is constant voltage. In that case the paralleled PV will be operated at that point instead of MPPT.
 
Back
Top