Why are off grid solar setups only 12, 24, 48 VDC?
60VDC is the cut-off for Safety Extra Low Voltage, or SELV, as spelled out in UL 60950-1. Besides being lower voltage, SELV circuits are also isolated from the mains by reiniforced insulation, which has specific spacing and materials requirements.
In general terms, SELV voltages are ‘touch safe’, meaning that they don’t present a shock hazard with direct contact. 48V falls below this SELV threshold with some margin. It’s also conveniently four ‘12V’ lead-acid batteries connected in series (really up to about 58.8V at full float charge.)
Voltages above the SELV level are considered in the same class as line voltage, and typically require an electrician to install. Reason? Electricians are familiar with codes and techniques to protect against inadvertent contact with potentially lethal voltages, including use of proper materials, fusing, fault protection, enclosures and cable routing.
More here: https://www.edn.com/electronics-blogs/power-supply-notes/4414411/What-does-SELV-mean-for-power-supplies
Field-assembled battery hardware can't go outside of SELV without help
The primary limitation on the DC bus voltage of most off-grid systems is indeed due to touch safety limits (60VDC/42.4VAC SELV limit), but that's not due to who's installing it. Instead, the issue is parts availability: lead-acid single cells and monoblocs of the sizes used in off-grid systems generally are not available with touch-safe terminals due to manufacturability and application diversity issues. This can be somewhat overcome, depending on the environment, by using a battery cabinet or battery room as the touch safety boundary, or by using a factory-assembled and listed energy storage system, but that leads us to our next issue.
DC switchgear is hard
Light-duty LV AC mains switchgear (MCBs and associated mounting/bussing systems, as well as fusible switches/disconnectors, garden-variety mains fuses, and so forth) is, of course, readily available for typical utilization voltages. However, only a limited subset of this gear is rated for DC service at all, and if so, its ratings will be limited to around 48-60VDC. Furthermore, disconnectors and such intended for solar service, while rated for high DC voltages, have very low short-circuit/interrupting ratings in the grand scheme of things. This is because solar panels are inherently current limited sources: no matter how long your string is, it won't put out much more than its rated Isc no matter what you do to it, and solar panel Isc values are on the order of amps, not kiloamps.
This means that you need much heavier-duty switchgear for DC service at mains-voltage-equivalent DC voltages, as battery strings are capable of kiloamp-class fault currents (easily equivalent to a mains source in this regard), and DC arcs are indefinitely self-sustaining once struck vs. AC arcs which will self-quench at zero crossings. Furthermore, even heavier-duty gear such as DC rated industrial-type MCCBs and heavy-duty fusible switches is often limited to 125VDC for single pole breakers and 250VDC for all-pole switching in multiple pole devices. While there are a few fusible switches out there that have ratings up to 600VDC, these are also limited by the inability to get fuses rated for mains OCPD service at voltages above 300VDC with only a few exceptions.
Interrupting ratings are another issue; achieving DC system short-circuit ratings over 10kA requires careful component selection, and even the most robust breaker and fuse designs available only achieve 100kA or less of DC interrupting rating. (For comparison purposes, most MCBs are available with either a 10/22kAIC series rating or a 10, 14, or 22 kAIC straight rating for AC service, depending on product line, voltage and where you are on the planet, and industrial MCCBs often have ratings up to 200 or 300kAIC. AC mains fuses are similar to MCCBs in this regard, as well, with modern current-limiting HRC fusing achieving 200 to 300kAIC, even in its consumer-accessible incarnations.)
TLDR: Industries use industrial gear, and homes don't have 50kw demand. Really.
This cannot be emphasized enough:
DC is one nasty customer.
It is easy to get complacent after spending a youth and a career working with docile, harmless 5-24 volts DC, or well-behaved 100 - 240 V AC voltages because of its frequent zero crossings.
DC in that same range is a mean drunk. You may have been in very old houses and felt switches that had a definitive SNAP when switched on or off. Those are throwbacks to when house power was DC, and they snap the contacts quite wide, to assure an arc is snuffed. Above that, you need magnetic or pneumatic "blowouts" designed to pull the arc up into an arc chute to blow it out. ,
Because if DC arcing gets underway, it will burn through almost anything. (note in this tram-fire video how the arc takes awhile to start, and then tears the tram apart. Keep watching, it relights a couple of times.)
Look at the DC ratings for contactors and relays. You will see very different voltage ratings for DC than AC.
As a result, the various regulations treat higher voltage DC differently from low voltage, and allowable maximums are typically in the 30-50 volt range.
Two thresholds of particular interest to solar panel installers (who work on roofs) are the 2017 Rapid Shutdown rules. You must now leave "aisles" between groups of solar panels for firemen to access the roof. This means there are groups of panels that are still adjacent.
For roof installations, there must be a switch accessible to firemen that will do two things. a) Reduce voltage within a group of panels to 80 volts DC or less. b) Reduce voltage between groups to 30 volts are less.
If system voltage is less than 30 volts, then no special provisions are needed.
This is a direct reflection of the hazards of higher voltages.
For a non-roof installation this is not an issue; food for thought.
Tall battery stacks are less reliable
Batteries are a series string of cells of 1.2 to 3 volts per cell. As such, a high voltage stack has a lot of cells. The more cells, the bigger the risk of a cell failure limiting or downing the entire stack. EVs have largely conquered this with exotica batteries, but you won't have the same luck with plain old lead-acids.
Battery voltage needn't match solar voltage
There is nothing wrong with high voltage on the solar panels (noting it is inherently current limited) and low voltage on the battery pack (which has literally no current limit, and will cheerfully explode). The solar charge controller would need to buck-convert, but it's doing that anyway.
Do you really need 50 kW?
Yes, if you were doing this commercially, e.g. to run a server farm or grow lights, then yes - you would subdivide the project into smaller chunks that fit the available inverter hardware. Keep in mind multiple inverters can't have their outputs paralleled because they can't sync. If they fall out of phase with each other, that amounts to a dead short between them. If you are expecting inverters to automagically sync with each other or with the grid, reset that expectation.
If this is a commercial venture, e.g. running ventilators and pumps at a remote mine, then you ante up into industrial tier equipment, and pay the nosebleed prices for same. The very fact that you're contemplating cheap Cheese seems to cast this as a homeowner issue.
Homes don't do this
One of the great fallacies in the "eco" movement is unit substitution so you can keep your processes exactly the same. Remove nitrate fertilizer and insert bat guano, remove Roundup and replace with brand X, but still farm in exactly the same way. Raise a building with brand X parts instead of brand Y and claim LEED certification (because the parts are LEED), while completely ignoring earth-sheltering, passive solar design and other real winners in the field. That approach is wrong-headed. Tragically common, but wrong-headed.
And honestly we get this more from naysayers: looking at their 200A main panel breaker, multiplying by 240 and declaring they need a 48KW solar system, and the naysayer immediately tosses out some numbers to prove a 48kw solar system is totally impractical. No poop Sherlock. In this case the naysayer is maliciously swerving into unit substitution specifically to assassinate the idea of off-grid living even being practical.
Of course, now, a 50kw solar system actually is imaginable, but it's still just as ridiculous as when the naysayer "proposed" it.
A random house with no conservation measures draws 1000 watts on average. That's according to the power companies, who rate power plants in terms of number of houses served. That's a fairly big system (100KWH of battery to ride through 100 hours of storm), so you think about loads and how to minimize them, particularly vampire loads which are 24x7.
The biggest vampire load in your whole house is the inverter. Even with no load, it burns 1% of its rating simply by being "spun up". So at 50kw, that's 500 watts of power -- remember what an average unoptimized house takes? So you're already there - how wasteful is that!? So keeping inverters sanely sized is important.
And you do that by thinking loads through carefully, and not having one giant customer that controls the inverter sizing.
Careful load design
Here's what you don't do: Avoid a $600 new fridge by keeping your old one, causing you to need to provision an extra $3000 of PV capacity to power the inefficient thing.
So for instance, electric baseboard heat is Right Out. Electric oven, no; use gas. Building heat should be passive solar design then an active solar-thermal system. Hot water, solar thermal with storage tanks.
There's a fair exception: Heat pumping is perfectly reasonable, because the 20ish % efficiency of the solar panel x the 300-1000% efficiency of the heat pump > what you could possibly get from solar-thermal.
The upshot is the only really large electric load you'd expect from an off-grid home is the air conditioner. The inverter would be sized accordingly.
As such, 48V is plenty of battery.
On-demand hot water
I've been racking my brain on what load exactly requires 50kw for a short time, and my conclusion is "on-demand hot water heat". I love 'em, but yeah. No... Just no.
That violates two tenets: making heat with PV, and running an inverter for a load that doesn't care about AC.
At the very least, you'd hotshot it off DC directly, perhaps using simple boost converters per channel that only spin up when the heater is calling for heat on that channel. But since you'd be doing 3-4 channels anyway on a large heater, you'd just do 3 smaller heaters - located right at the spigot, so you zero out that long, very expensive wait time for hot water to reach the spigot (with the now-filled pipe's hot water being abandoned).
The right way to do this is to use solar-thermal with a large storage tank, possibly augmented with a heat pump. The storage tank gets warmed as hot as you can possibly get it (90C if possible), then it's either heat-exchanged to make domestic hot water, or used as the heat sink for a heat pump (tanked) water heater. Heat pumps are more efficient when pumping "downhill", so that big solar-thermal tank of hot water saves a lot of PV energy even if it's not used directly.