Hacker News Reader

Comment by ctdonath | Hacker News Reader

I run my “office” on portable solar much of the summer. Two takeaways:

Solar averages ~10W/m^2 long term (days/weeks) with wild fluctuations. Getting easy reliable output retirees a lot of surface & batteries.

When those storage batteries deplete, it’s a hard stop. Recharging takes considerable time, competing for scheduled direct use. Plan on at least doubling panel surface just to recover from outages, and plan on more storage than you expect because long uncooperative weather happens more often, with more consequences, than you expect.

Unless you are willing to shut down completely at times, or spend enormous sums, you must have a reliable on-grid source ... which then may as well be your primary.

View on HN

A reliable, secondary source of power to account for consecutive days of bad weather doesn't need to be "on-grid."

Most robust off grid setups that I've seen involve the use of a backup generator.

For anyone considering an off grid mobile office in a van or other vehicle. When the sun isn't cooperative for too long and I find my storage battery being depleted too low, I simply start my engine and start charging the battery using my 12V DC cigarette lighter port.

I have 200W of solar feeding a 100Ah 12V battery with an MPPT charge controller and I thought I had calculated enough redundancy to handle up to a week of bad weather, but end up using my alternator to charge when the sun is unreliable more often than I thought I would.

PS: Are you using a PWM controller or a MPPT controller? The latter can increase efficiency by up to 30%.

> Solar averages ~10W/m^2 long term (days/weeks)

That figure surprises me. What I’ve heard before is that full sun is roughly 1kW/m², and that over time you should expect to get 4–5 useful hours (depending on latitude and weather conditions of where you are—some will be outside this range on either side), so if you take 21% efficient solar panels, the total would be 1000 (watts) × 4 (hours) / 24 (hours) × 0.21 (efficiency) = 35W.

10W is a lot less than 35W. I presume I have some incorrect assumption or am overlooking something.

(I’m interested in this because I’m planning to build a velomobile and live out of it while going round Australia for a year. Power systems are something I’ve been putting a fair bit of theoretical research into, but I have no practical experience whatsoever. I’ve been planning to build solar panelling into the body.)

35 is pretty close to 10 when you start at 1300 and prune obstructions.

Now address clouds, angles, dust, snow, opacity, malfunctions, storage inefficiencies, panel aging, and chronic optimism. 10W.

Most of that is what the 4–5 useful hours per day figure is supposed to take care of (that’s a year-round thing, too, averaging summer, which may be as much as double that, and winter, which may be half).

Hmm… storage inefficiencies? Are you including battery storage and perhaps an inverter in your calculations? I guess that’d account for some loss.

I’m guessing also from “snow” that you may be in a place that gets comparatively little sunlight?

I wonder if "4-5 useful hours per day" is per sunny day, so you can cut your number in half again for cloudy days.

No, this is from averaging it across the entire year, based mostly on locations in the south of Australia (Gippsland, Melbourne, Adelaide, Ballarat, those sorts of areas), but I looked briefly at other parts of the world also. The idea is then that a panel rated at 250W can produce an average of about 1kWh per day when pointed in the optimal direction and never shaded (but without tracking the sun); the average winter day may get you only 500W, and the average summer day over 1500W, with plenty of variation within those seasons day by day also.

42,339 MW of PV installations produced 39,401 GWh of energy in 2017 in Germany[1], so about 38 days of peak generation, or roughly 10.5% of peak. A square meter of PV panels produces around 150W. So 15W average for a panel in a good location in Germany. 10W/m^2 seems like a good estimate if you want your stuff to work reliably year round. Better locations than Germany would do better of course, and restricting yourself to just summer probably helps quite a bit too (winters are dark and cloudy in Germany).

[1] https://en.wikipedia.org/wiki/Solar_power_in_Germany#Statist...

Germany is north of Maine, it’s a terrible location for solar power and heavily subsidized. At best they hit a capacity factor of 11.6% in 2018. By comparison the US had a capacity factor of 28% for utility-sized PV plants in 2015, but it’s dropped as less efficient locations have been added.

A square meter of PV panels produces around 150W, that’s some low efficiency panels, you can easily hit 200+W/M2 using a high efficiency panel like SunPower Maxeon 3.

Solar isn’t a great fit everywhere. So, people really can see wildly different outputs from the same setup based on location. At 28% capacity factor averaging 60+W/M is very possible.

Thanks for the reference to the 200W panel!

The most shocking thing I ever learned about solar panels is that they behave as a rectifying diode when they are not receiving sufficient photons.

That means you can screw up the current from an entire bank of solar panels by having just one of them in deep shade. So if you want maximum generation, you either have to site them in a very clear area, or wire them all in parallel instead of serial, which means much fatter wires to reduce transmission losses, and then larger conversion losses?

I'm not sure on the details, but according to my solar installer newer panels (or inverters or some other component) are no longer susceptible to this.

The key term with this is bypass diodes. Not all panels will be equipped with them (which puzzles me, given the substantial benefit and what I imagine to be near-trivial BOM cost).

I learned this at the Exploratorium in SF. They had an exhibit consisting of a miniature solar panel with an output reading and a small token that could cover a single panel. Very fun trip!

I don't think this is the case with newer panels. Our boat stern mounted panels atleast had one panel shaded at anytime. The output drop was proportional to the area shaded. The bigger losses came from the angle of the sun. Output was mostly logarithmic, almost nothing at sunrise/sunset, full output at 12pm, 50% by 3pm.

It's still true of the cells in the panels. But it's typical these days to add a device between the panels and the DC circuit than can improve the situation (e.g. by cutting the panel out of the circuit if it isn't productively contributing the array's generating capacity at the moment).

So the problem should be corrected at the panel level, but still exists at the cell level?

Yes, the diode behavior is (I believe) a fundamental property of the cells.

One way around this is to utilize cheap micro-inverters on each panel. This makes it easier to add new panels as your usage goes up (or funds). Also easier to monitor their performance due to shading, dust, etc.

I remember back in 1989 or so(!), my mom used to work at our summer cabin which had no electricity using a ~1m^2 solar panel, hooked up directly to a car battery, which was hooked up to a small laptop (a Compaq, I think).

I have no idea if such a setup would be remotely feasible today.

Oh, it’s feasible. That’s close to what I’m fiddling with (just GoalZero.com products instead of bare car batteries etc). It’s quite workable, but you have to be keenly aware of usage & limitations, and fiddle with stuff often (ex.: coping with passing shadows). Most people are expecting complete reliability - that’s costly.

Panels are better, MPPT are cheaper, and the right laptop uses hardly any power... I would think it should be more feasible in every possible way aside from increased expectations.

A Surface Go averages something like 5W, and a 100W array can provide that in virtually all daytime conditions even without aiming.

Additionally, the right laptop can use its internal voltage regulator to convert the 12-16 volts from the battery into the internal voltage.

One of the secrets of the openwrt community was that those little linksys routers used a 9v power brick but the power regulator was still stable somewhere north of 12v, giving you a lot more options for repeaters or mesh networking.

12V DC 0.5A was typical, stable from 5-30V depending on the exact model.

> you must have a reliable on-grid source ... which then may as well be your primary.

Not at all.

With net metering it makes sense to be on the grid during the day when you're making more power than you need, then if and when you need it you can pull power from the grid.

In Australia my Dad got a 5kW system fully installed for $5k AUD, and he gets paid a few hundred dollars from the power company every three months now.

Problem there is, with enough customers doing the same, there’s a great deal of idle capacity waiting to fill the big gaps. That’s costly. Works fine now when solar is the anomaly (what I’m doing, plugging into grid when a prolonged storm depletes stored power), but keeping 100% of current grid capacity in unused reserve for 1% incidents (heavy clouds for >1 weeks) isn’t practical. At some tipping point, classic grid power is driven out of the market ... and then an anomaly hits, with no grid to back into.

They are still saving fuel, and with weather forecasts they are not 'anomalies', they are planned handovers. You just need to work out the financials.

You could supplement with your own generator. Do you consider fuel to be an enormous sum, relative to solar?

I’m focusing on solar-only solutions.

Once you introduce sourcing as cheap & reliable as fossil fuels back into the equations, dropping solar entirely becomes a rational solution as the frequency of solar “incidents” becomes apparent and non-FF mitigation costs soar. I find most people don’t grasp how unreliable solar is.

Where do you live ?

You are making very bold claims about solar being unreliable, but I'd completely expect reliability of solar to vary significant from one location to another.

Southeast USA. On the whole a good place for solar. I’m trying to impress on people the many fiddly little obstructions that add up to significant reductions, and that despite prolonged good periods there are poor exposure times that fast burn thru stored energy and require substantial collectors to recharge in time for the next weather event.

If you're intent on using solar anyway then I don't see how having a fallback generator would compel you to just give up on the solar. How often do you have to repair or replace panels?

It’s just plain easier & cheaper to run FFs than solar. About the only serious disadvantage is noise.

This is very misguided - anyone in serious remote area operates renewables with FF backup, because FF is expensive and delivery difficult and unreliable. If you ever been to remote lighthouse or similar, youd realise logitistics in delivering tonns of FF is not something you can take for grantes.

Those 200 KG of solar panels will save you many metric tons of oil over their lifetime

Small-scale alcohol fuel production is an option. Grow sugar beets and make moonshine. It's still solar power. (No pun intended.) The molecules of alcohol are made of air and water and power from the sun.

Reply on news.ycombinator.com