Orientation vs self consumption

Z

Zee

Senior Member
Location
CA
Maximizing self consumption is the new normal around here.
(And will be soon wherever you are as the grid becomes saturated with solar, and solar loses its value mid-day.)

Gone are the days of blasting the maximum amount of solar power to the grid mid-day for very high credits, then using grid power at cheap rates at night.

So does it make sense to start designing arrays around split orientations (east and west) to spread the solar production throught the day and allow the house to self consume more solar than send it to the grid for pennies?

Or face them all South and put in a ton of batteries?
 
My intuition is that batteries are going to be necessary regardless in most cases, but that pointing arrays in multiple directions will be more cost effective as far as it goes. That is, if you have good software to compare scenarios with arrays pointing different directions to scenarios with all pointing south, I think you'd find that some percentage of the time the software recommends one fewer battery unit for the former scenario.

You need good modeling software that you can trust, which I think is a whole other issue right now, but there you go.
 
It depends on the energy usage pattern of the home and the TOU rate structure. If the energy usage can soak up the PV energy during the day then you can even the production out with an E-W orientation. If the usage peak is 5-8pm then no amount of module orientation is going to help you, it's battery time. Usually day rates are lower than evening rates, sometimes significantly, so a system without a battery is stuck producing lower value energy. But the cost of a battery system can quickly eat into any money saved from the annual energy offset. Good system modeling that includes PV system production and financial modeling of the energy usage is really the only way to know what will get you the most bang for the buck. The days of simply matching annual PV production to annual energy usage are fast disappearing with the replacement of NEM with other less advantageous compensation systems.
 
  • Like
Reactions: Zee
Thanks. I mis-spoke a bit. I do understand that batteries will be necessary regardless of panel orientation.
I was wondering:
-given batteries
-maximizing self consumption
-a typical "home-in-the-afternoon/evening" household scenario
...... would placing the panels deliberately East-West vs all West be worth it? Splitting the panels E-W adds labor cost and a 5-15% hit or so in output for the Eastern array (morning fog vs PM sun).
My initial Q ......of choosing between an E-W split vs all S........was silly because usually the roofs either face E-W OR S-N. So choosing between E-W vs S is not an option 99% of the time.

This whole MSC (maximizing self consumption) scenario has really gotten me thinking about how best to use up your own solar.
A couple conclusions about optimal design and payback pop out:
1. The more a customer uses power mid-day, the fewer batteries you need, the better your payback.
2. If they use most of their power in winter, they can't buy enough batteries to make it work. No battery will store summer sunshine till winter. So the heat pump crowd has a misaligned power usage pattern for solar/battery systems. The batteries just cannot ever recharge in winter.
3. If they have an EV, then having an integrated smart EV charger (enphase) that uses solar power when its peaking (mid day) would seem like a huge boon.
4. Heavy A/C users would seem like good prospects. They have mostly summer time and day time use. This is closely aligned with solar output.
5. Overall, tweaking your usage to earlier in teh day than ideal would seem worth it. I.e. many people start the A/C later in the day, and the heat when they get home, but it may be worth setting these to kick on early to capture sun.

Just thinking out loud here and wondering what conclusions others have reached?
 
Zee said:
Thanks. I mis-spoke a bit. I do understand that batteries will be necessary regardless of panel orientation.
I was wondering:
-given batteries
-maximizing self consumption
-a typical "home-in-the-afternoon/evening" household scenario
...... would placing the panels deliberately East-West vs all West be worth it?
Click to expand...
If there are batteries, what difference would it make what time of day the energy is harvested?
 
Assuming usage throughout the day that is morning afternoon and evening.....wouldn't you need fewer batteries if you had panels facing East and West?
 
wwhitney said:
For pure self-consumption, battery is sized based on the integral over the day of (PV production - houes load). So reducing that integral reduces the necessary battery size.

Cheers, Wayne
Click to expand...
Batteries come in discrete sizes; it might not make enough difference to jump from one size or count to another. One thing I know about batteries: buy more than you think you will need.
 
ggunn said:
Batteries come in discrete sizes; it might not make enough difference to jump from one size or count to another.
Click to expand...

Which is what I alluded to in post #2 where I said some percentage of the time a good analysis softwareqouls recommend one less unit. It's both a function of the size of the difference as well as whether the difference happens to fall across the size of a multiple of units.

One thing I know about batteries: buy more than you think you will need.
Click to expand...
For self-consumption in a grid-tied scenario I don't think really true.
 
Zee said:
Thanks. I mis-spoke a bit. I do understand that batteries will be necessary regardless of panel orientation.
I was wondering:
-given batteries
-maximizing self consumption
-a typical "home-in-the-afternoon/evening" household scenario
...... would placing the panels deliberately East-West vs all West be worth it? Splitting the panels E-W adds labor cost and a 5-15% hit or so in output for the Eastern array (morning fog vs PM sun).
My initial Q ......of choosing between an E-W split vs all S........was silly because usually the roofs either face E-W OR S-N. So choosing between E-W vs S is not an option 99% of the time.

...
Click to expand...

Again, my guess it would be worth it if analysis comes back with needing one less battery unit, i.e. reduces cost significantly.

But here's the thing...
It's much easier to think out which individual factors and phenomenon would increase or decrease the payback than to *quantify* them, in part or in total. To quantify it reliably, you need good software that includes a model of solar production, a model of battery losses, and accurate consumption data. (For best results, you need historical consumption data that's not likely to change much. Models here will be much less accurate than for solar production.) And you need to be able to easily run multiple scenarios for comparison. For multiple solar azimuths, that means synthesizing production totals from multiple models for each array.

It is unclear to me if we're getting solar estimating software products that do this well. I mean, for obvious reasons they don't generally let you look inside to audit how it works, because that's their IP.

And then on top of that, there's some slop for instaneous netting vs. granular 15 or 60 minute data, which if I understand correctly is never going to be in the customers favor.
 
  • Like
Reactions: Zee
jaggedben said:
Again, my guess it would be worth it if analysis comes back with needing one less battery unit, i.e. reduces cost significantly.

But here's the thing...
It's much easier to think out which individual factors and phenomenon would increase or decrease the payback than to *quantify* them, in part or in total. To quantify it reliably, you need good software that includes a model of solar production, a model of battery losses, and accurate consumption data. (For best results, you need historical consumption data that's not likely to change much. Models here will be much less accurate than for solar production.) And you need to be able to easily run multiple scenarios for comparison. For multiple solar azimuths, that means synthesizing production totals from multiple models for each array.

It is unclear to me if we're getting solar estimating software products that do this well. I mean, for obvious reasons they don't generally let you look inside to audit how it works, because that's their IP.

And then on top of that, there's some slop for instaneous netting vs. granular 15 or 60 minute data, which if I understand correctly is never going to be in the customers favor.
Click to expand...
FWIW, NREL says that PVWatts is +/- 10%. I'm pretty sure that any predictive PV software is going to be adding corrective factors to the same NREL data, so it doesn't make much sense to jump too far down the rabbit hole.
 
jaggedben said:
Which is what I alluded to in post #2 where I said some percentage of the time a good analysis softwareqouls recommend one less unit. It's both a function of the size of the difference as well as whether the difference happens to fall across the size of a multiple of units.


For self-consumption in a grid-tied scenario I don't think really true.
Click to expand...
True, I have many proposals where the payback on PV + 0 kWh of batteries is identical to PV + 10, 20, 30 kWh of batteries!
The main difference being, with PV only, it is unsatisfying for most solar customers to offset only 50% of their usage. .....vs close to 100% with the huge battery bank.

Also there are heavy winter users (heat pump) where i can keep adding batteries, but to no avail: there ain't no solar power to put in the damn things in mid winter.
 
jaggedben said:
Again, my guess it would be worth it if analysis comes back with needing one less battery unit, i.e. reduces cost significantly.

But here's the thing...
It's much easier to think out which individual factors and phenomenon would increase or decrease the payback than to *quantify* them, in part or in total. To quantify it reliably, you need good software that includes a model of solar production, a model of battery losses, and accurate consumption data. (For best results, you need historical consumption data that's not likely to change much. Models here will be much less accurate than for solar production.) And you need to be able to easily run multiple scenarios for comparison. For multiple solar azimuths, that means synthesizing production totals from multiple models for each array.

It is unclear to me if we're getting solar estimating software products that do this well. I mean, for obvious reasons they don't generally let you look inside to audit how it works, because that's their IP.

And then on top of that, there's some slop for instaneous netting vs. granular 15 or 60 minute data, which if I understand correctly is never going to be in the customers favor.
Click to expand...
Great points.

In SolarGraf (bidding software), I tested 16 panels west vs 8 west /8 east and I got the same damn results on Solargraf. So I wondered if they even made a distincition in the timing of solar output depending on orientation, but it just didnt matter in this case.......... or that they aren't that sophisticated.
Hence my post here:
In my own thought experiment, it seems E-W orientation would matter quite a bit, and allow a system to reduce the needed battery bank. As long as they used power in the morning.
-----------
Also I used to ask for exact 15-minute or hourly interval consumption data, but it never seems to upload right, so I just choose among a couple pre-provided, rough defaults for household usage patterns....... and at most I input the monthly kWh totals if available.
( That's where I found heavy wintertime users really have a hard time getting a good return on batteries.)
I figure it is not the end of the world if I undersize a battery bank, as they paid less, and saved less , and can still easily expand. Especially with the Enphase 5 kWh batteries.

(I size up a tad though with battery backup scenarioes. I don't want that DEC 24th call that the power is out.)

In no way is this superior to the proper integration of usage and production functions or accurate modeling software, but I am just trying to understand the general trends.
 
Zee said:
In no way is this superior to the proper integration of usage and production functions or accurate modeling software, but I am just trying to understand the general trends.
Click to expand...
"Accurate" is a matter of degrees and it is never 100%. What is it that stock brokers always say? Something like "prior performance is no guarantee of future results".
 
  • Like
Reactions: Zee
You must log in or register to reply here.
Top