Solar power output

[ruhrohraggy]

If anyone even bothers using solar anymore...I got bored and did some observations regarding solar power in ONI.

*Updates and new information gained* Big thanks to those that took the time to read / respond, I'll add your information up top here.

- Lux starts at 0 and increases linearly to peak lux of 80,000 at midday. This curve is a triangle. (Nitroturtle)

- As light is blocked on a solar panel, the curve shown below will shrink horizontally until it forms a triangle, and then shrink vertically to 0 as a triangle. (Nitroturtle)

- Bunker doors let light through on closing, but block light when opening. (Azunai333, or whatever your name originally was)

The biggest thing I wanted to look at is battery storage capacity vs. the worst case meteor strike.

I worked most of this out by observation and plotting...

The Solar Power curve is roughly linear as shown here...

Spoiler

 

I calculated 161.5 kJ total power using the data provided

Spoiler

.

To gather the data, I hooked a solar panel up to a smart battery, and watched it the day after (once the smart battery was fully charged). At the end of the cycle, the reports tab showed 156 kJ wasted, after taking into account battery runoff.

I also kept track of the incremental time using the in game reports feature, and plotted wattage observed in the tool-tip during each tool-tip update. (Which was roughly every 2.5s at normal speed)

If you look close, you can see that the plotted data is not perfectly linear, as I drew in a straight line, and you can see some space under the curve. I had a feeling this might have been the case, as the first time I worked out the math without putting in any raw data, my math was different from reported in-game power.

My guess is rounding errors or something, as it's pretty darn linear and I can't see the devs bothering to put in an nth root equation to model their solar power curve. (And I'm not looking into the code to find out)

Having worked out the % difference, it comes out to around 4% error, or around 5-10 kJ over the course of 1 cycle's worth of light.

We learn a few things from this.

It takes 100s from the start of a new day to reach 100% power output, 325 seconds of full power, and then another 100s to reach zero again. (roughly).

 

Using this information, we can then calculate how much power we lose during any interval of the day. So that's neat.

This could be a situation where meteors strike before sundown for 195 seconds. (Best case would be meteors hitting right at nightfall for 100s)

Spoiler

 

And then we have worst case...Meteors hitting right smack in the center of our day cycle for full duration. It's really not great at all.

After having done all this, I actually forgot to tack on the extra *edit* 38.2 seconds the bunker doors would spend *opening* where the panels wouldn't be getting light. Add to that any extra time they closed early due to an inefficient scanner network and...I can't be bothered to go back...safe to say, if meteors plan on hitting you like this, you're not really getting any solar power for the cycle.

Spoiler

 

 

So now we know a little more about solar panel power output.

It's basically linear.

Solar panels generate somewhere between 140 and160 kJ per cycle of full received daylight.

It would take around 9 smart batteries, accounting for runoff, to store the entire power gained from a solar panel in 1 cycle.(I let 1 solar panel power 9 batteries for 1 cycle, here's the result, 15.7 kJ * 9 = 141.3 kJ stored + 9 * 0.4 kJ runoff = 144.9 kJ, so we see some more of that error.... )

Spoiler

The worst case meteor strike basically means our solar is down for the entire day...

So if you are powering something that runs constantly with solar panels, you will need at worst, 9x smart batteries per solar panel to avoid it shutting off.

Well, that was fun. Hope some of this information turns out useful to anyone tinkering with solar panels...

 

 

 

[Neotuck]

why not use large batteries instead of smart ones?

[Neotuck]

I understand most players don't like large batteries since they are a little "leaky" but when it comes to solar panels you lose a lot more power when batteries reach full capacity 

[Saturnus]

1 hour ago, Neotuck said:

why not use large batteries instead of smart ones?

Smart batteries are superior in every way... except energy density but since area is hardly an issue for any player even that doesn't matter.
 

[Neotuck]

2 minutes ago, Saturnus said:

Smart batteries are superior in every way... except energy density but since area is hardly an issue for any player even that doesn't matter.

would you say this build I did in the last update would improve If I replaced every large battery with smart batteries?

[Saturnus]

Just now, Neotuck said:

would you say this build I did in the last update would improve If I replaced every large battery with smart batteries?

Yes. Very much so.

[Neotuck]

Just now, Saturnus said:

Yes. Very much so.

elaborate

I know about the leaking problem but when I did a test in debug mode the overview showed more power wasted with smart batteries then the large ones due to them reaching max capacity sooner

[Nitroturtle]

This is interesting information, but I want to point out that only 10 of the 14 tiles in a solar panel are needed for max power generation.  It would be interesting to see how this effects the calculations.  It should make the chart look more like a triangle, but I'm not sure what the duration of the flat portion at the top would look like.

In my setup, I layer the panels so that most of them only have 10 tiles exposed.  This allows me to fit 36 solar panels, while leaving 16 bunker doors worth of space for rocket silos.  Only 6 of the panels have all 14 tiles exposed, while the other 30 panels only expose 10 tiles.  I have around 290 smart batteries on the circuit to store all of the power, and even then I'm over-producing by quite a bit.

Spoiler

Here's the save if you want to play with it:

Planet3.sav

[Saturnus]

15 minutes ago, Neotuck said:

elaborate

I know about the leaking problem but when I did a test in debug mode the overview showed more power wasted with smart batteries then the large ones due to them reaching max capacity sooner

As you can see in the OP. You need 9 smart batteries for every solar panel to have 100% coverage for the worst possible meteor shower. Or you need 5 jumbo batteries to account for the added loss in jumbo batteries.

9 smart batteries is 1800kg of refined metal which is unlimited as it rains down from space with refined iron, or you have metal volcanoes.
5 jumbo batteries is 2000kg of raw metal which is a limited resource.

9 smart batteries generates 5400 dtu of heat requiring 0.450 wheezewort to cool
5 jumbo batteries generates 6250 dtu of heat requiring 0.521 wheezewort to cool

[SackMaggie]

The day has come SOLAR PANEL RATIO in ONI.

[ruhrohraggy]

4 hours ago, Nitroturtle said:

This is interesting information, but I want to point out that only 10 of the 14 tiles in a solar panel are needed for max power generation.  It would be interesting to see how this effects the calculations.  It should make the chart look more like a triangle, but I'm not sure what the duration of the flat portion at the top would look like.

Wat? I'm not sure what your thought is there...so...respectfully, I am not making assumptions as to what you mean, and would require further explanation.

[Nitroturtle]

4 minutes ago, ruhrohraggy said:

Wat? I'm not sure what your thought is there...so...respectfully, I am not making assumptions as to what you mean, and would require further explanation.

The solar power curve you have above has a rising portion, then a flat part on top, then a falling portion.  In my setup, where 10 of the 14 tiles on the panels are exposed to light, I would expect the rising and falling sections to be larger and the flat part on top to be shorter.  Because in your graph, you are using 14 tiles exposed, which will reach the amount needed for max output earlier than panels with 10 tiles exposed.  Hopefully that makes sense, if not I can try to explain it better.

[Azunai333]

9 hours ago, ruhrohraggy said:

After having done all this, I actually forgot to tack on the extra 76.4 seconds the bunker doors would spend opening and closing where the panels wouldn't be getting light.

This is only half right.

The bunker doors will still let light through while closing it:

While opening:

[ruhrohraggy]

20 minutes ago, Nitroturtle said:

The solar power curve you have above has a rising portion, then a flat part on top, then a falling portion.  In my setup, where 10 of the 14 tiles on the panels are exposed to light, I would expect the rising and falling sections to be larger and the flat part on top to be shorter.  Because in your graph, you are using 14 tiles exposed, which will reach the amount needed for max output earlier than panels with 10 tiles exposed.  Hopefully that makes sense, if not I can try to explain it better.

 

Nvm, I get what you're saying. I'll need to look into exactly where light peaks out. If it turns out that received light doesn't peak out at when a single solar panel maxes out, then yes.

I'll look into it.

 

17 minutes ago, Azunai333 said:

This is only half right.

The bunker doors will still let light through while closing it:

While opening:

Good information to add to the thread. Thank you. So this means only 38.2 seconds total gets wasted during a full, powered bunker door cycle.

Do you know if the bunker door lets in 100% light while it is closing?

 

[Azunai333]

7 minutes ago, ruhrohraggy said:

Do you know if the bunker door lets in 100% light while it is closing?

Yes, they do. The blocks of the door are nonexistent, so the speak.

Sadly this is true for incoming meteroids, too...

[Nitroturtle]

 

35 minutes ago, ruhrohraggy said:

 

Nvm, I get what you're saying. I'll need to look into exactly where light peaks out. If it turns out that received light doesn't peak out at when a single solar panel maxes out, then yes.

I'll look into it.

Cool, yeah that's exactly what I'm saying.  Solar panels with all tiles exposed are wasting possible light as they max out at a much lower level than they end up getting during midday.  Panels with all tiles exposed will produce max power once light reaches 51,213 lux (716,981 total), but peak light output reaches 80,000 lux.  So with 10 tiles exposed, max power is produced with 71,698 lux, but obviously for a shorter portion of the day.

Setting up staggered panels seems to provide significantly more power as it's able to use more of the available light for the given area, though I've never done the math to figure out the numbers.  That's why I was curious how it would fit into what you've shown above.

What I'd really like to know, is what the optimal panel arrangement would look like.  I've only ever gone with 4 levels, as this fits nicely with the door sizes and allows 12 panels per 16 doors.

[ruhrohraggy]

1 hour ago, Nitroturtle said:

Yeah, that's what I'm talking about, and I think you are missing something.  Solar panels that are partially covered can reach full output during midday.  Only 10 tiles (5 columns of light) are needed at midday to reach peak output (I think it's more like 9 actually, but that's not possible obviously).  So with my setup, the panels that only expose 10 tiles have a shorter period where they are producing max power, but they do still reach max power.

Yep, I got it, took me a second to think about it, I just woke up. You are right.

Having done a quick in-game check...The plotted curve for solar lux would be exactly a triangle since it peaks at 80,000 lux at midday for a brief moment.

So in the case of decreasing tiles of solar light to a solar panel the graph would shrink horizontally down into a triangle, and then vertically down to 0.

Your flat top begins whenever the panel reaches peak wattage. So it'd be something like this, I imagine.

 

The rate of change for decreasing tiles of exposed light is probably linear as well, so you could work out total generated each day by a solar panel, with 1 thru 6 tiles blocked.

1 hour ago, Nitroturtle said:

Setting up staggered panels seems to provide significantly more power as it's able to use more of the available light for the given area, though I've never done the math to figure out the numbers.  That's why I was curious how it would fit into what you've shown above.

What I'd really like to know, is what the optimal panel arrangement would look like.  I've only ever gone with 4 levels, as this fits nicely with the door sizes and allows 12 panels per 16 doors.

And this, I find to be quite an interesting question.

I saw your parabolic lookin' solar panel arrangement in the file you provided for the liquid hydrogen discussion, and I had thought the same thing...Is that more ideal than the small staggered arrangement I used in my pulse scanner demo?

Will have to look into this for sure. I think the best way to go about this would be to do it experimentally, since the hard math seems to carry some error with it.

You would also need to decide what optimal means. Optimal per horizontal tile, which would give the most total power output for the length of the map? Optimal per total space occupied, which would use the least amount of resources? Optimal in terms of practicality?

[ruhrohraggy]

So I ran some fun tests, and I'll post my numbers, conclusion, the save file for anyone else who wants to tinker with it, and then some pictures.

I ran 5 tests.

1) A tight staggered, compact set of 3 panels

2) A spaced staggered, set of 3 panels

3) Panels side-by-side

4) Nitroturtle's parabolic configuration

5) An extremely stacked parabolic configuration.

I setup a big line of automated batteries and let each configuration of panels charge them up sequentially, to see how many batteries would get charged. Granted, there's a tiny amount of power used in the power shutoffs and in runoff, but i don't care, the system is equal for all configurations.

I then quantified the data into 5 categories. As shown below.

Here are the results :

Of the 3-panel configurations, we can see that the Compact Stack has the best space efficiency, but offers the worst solar panel efficiency. (But not by much)

The Spaced Stack has the best of both worlds, but excels in neither. You get a reasonable bump in solar panel efficiency, for not much space efficiency cost. This is probably overall, the most efficient of the 3 panel configurations.

Standard layout obviously gives you the best solar panel efficiency, but has the worst space efficiency.

I exactly copied a section of the parabolic panel layout that nitroturtle used in the save file regarding liquid nitrogen. I'm surprised the results were a little different than the Spaced stack, since it's basically the same layout. I'm going to attribute this to inconsistencies and rounding errors. It's pretty darn close though.

*Corrected for inability to count little squares*

And lastly we have the super-parabolic configuration. It has by far the worst solar efficiency, and gains negligible power efficiency. Though it still gains a little according to the table.

If you were to go for absolutely the highest possible power from solar on the map, you would use a parabolic configuration.

Anyway, here's the save file :Solar Panel Battery Meter.sav

Here's the photographic evidence.

Compact stack :

Spoiler

Spoiler

 

Spaced Stack :

Spoiler

Spoiler

 

Side-by-side (standard layout)

Spoiler

Spoiler

 

Parabolic layout

Spoiler

Spoiler

 

Super-Parabolic Layout :

Spoiler

Spoiler

 

So, yeah, that was fun too. I actually had the most fun watching the little battery meter do it's thing. /autism

Next I'll take a look at how blocked tiles affects a solar panel...and see if there's some ideal amount of tiles that can be blocked.

 

 

 

[Iriswaters]

It feels like the best way to store power for when solar goes down is to have backup nat gas/petrol/coal, rather than 999 smart batteries...

[Yunru]

33 minutes ago, ruhrohraggy said:

If you were to go for absolutely the highest possible power from solar on the map, you would use something like the super-bolic setup...Which isn't really practical...I think.

What? You just showed that regular parabolic charged the most batteries, thus generated the most power.

O.o

[ruhrohraggy]

3 hours ago, Yunru said:

What? You just showed that regular parabolic charged the most batteries, thus generated the most power.

O.o

Let me say this another way.

The map is only so long horizontally...

So to get the most power generated per cycle, you need to use the layout with some combination of the best space efficiency, and the best solar efficiency. Whether it leans more towards space efficiency, whether they need to be equivelant, or more space efficiency, I don't really know. My instincts say that they'd need to be somewhere in-between. That's why I said "Something like it". I'm not quite sure, and have to examine how blocked light affects solar efficiency, and then find some equilibrium.

I will setup the super-bolic setup in the same floor-space used by the regular parabolic and see how many batteries I can charge with it...brb.

Turned out it eeked out a smidge more power, as the data suggests. Not really worth it imo though.

[Nitroturtle]

My goal was to optimize use of the horizontal space since it has a hard limit.  Seems like my setup does a pretty good job in that regard. I don't care about materials used since that's never been much of an issue in my games. 

It looks like there may be some gain from adding layers, though I doubt it would be much.  As I mentioned, the four high stacks also work out to exactly 12 panels for 16 doors, which makes everything fit nicely. 

I'd bet the optimal layout would have as many 10 tile panels as possible and as few 14 tiles as possible. So probably a wedge up or down from each side with a silo in the center. Or the silo on one far side, which I can't stand due to the camera scroll limits at the edge of the map. 

Thanks for doing all that testing. Definitely very interesting stuff. 

[Mariilyn]

Wouldn't have it been simpler and more precise to look at the daily reports for the power generated at the end of one cycle for each setup though ?  Ultimately, I think the most practical piece of info would be the amount of kJ generated per horizontal tile. 

[ruhrohraggy]

1 hour ago, Mariilyn said:

Wouldn't have it been simpler and more precise to look at the daily reports for the power generated at the end of one cycle for each setup though ?  Ultimately, I think the most practical piece of info would be the amount of kJ generated per horizontal tile. 

But that involves more math...And this involved making a fun automated battery meter. Logic will not apply. It gives us basically the same result though.

Also, the kJ wasted per cycle doesn't seem all that reliable to me, it was slightly different each day when testing out power cycles for 1 solar panel. With stored energy, it's more definite, I think.

With the amount of batteries, just convert each battery into 20kJ stored energy and do the math per tile. It's technically the same information presented differently...

Something like this ye?

[Mariilyn]

Yeah! But I decided in the meanwhile to test them for myself using the method I described instead of doing the math (lol). I looked at the energy created, not the energy wasted, and it was simple. The results are much the same but slightly higher, probably due to the absence of battery runoff.

Also the SuperParabolic configuration is actually 26 tile large and not 20, so there is no meaningful difference between the two parabolic layouts.

Anyway, that was interesting to compare! For me these numbers are a bit more relevant since I don't use large battery banks. 

 

Screenshots :

Spoiler