While I was out in Utah, I also took a chance to look over some large solar installs in process (or completed some while ago). Of particular interest to me was some of the tracker technology for a large farm under construction - because it’s often hard to find details on how this sort of stuff works, unless you’re actually in the bid process for a large (5+MW) solar plant.

This plant, of course, was a bit more than 5MW…

Why Do I Care?

I’ve been curious about how “big commercial solar farms” are done for some long while, because their cost per watt numbers, even with labor, are exceptionally low. Most of the newer farms seem to be sub-$1/W installed. A revision on my A-frames recently got down into that price range, but that was a DIY install, so no labor costs on it (and homeowner installed).

I keep wanting to figure out how to do some additional solar on other chunks of our hill - we have access to quite a bit of “generally useless rocky south facing hillside” if it gets zoned properly, and if I were to put solar on it, I’d want to put a few MW of single axis trackers up. The A-frames are nice for residential installs, but if you’re really aiming to optimize every aspect, the single axis trackers seem worth the costs - you make better use of panels, and can do some neat tricks with morning/evening optimization depending on what sort of panels you install.

If anyone has a few spare million laying around they’d be interested in investing in a small solar farm, I’m exceedingly well sited for such a thing. I’m about half a mile from a good sized substation that has more interconnection going in soon, and I’ve got 19kV of three phase running down the street near me with plenty of load in both directions, so I don’t think I’d need much of anything in the way of grid upgrades or transmission upgrades for a small (few MW) plant. It’d be a good source of skilled and unskilled local labor use as well, punching holes in the ground and such.

But, what’s a big plant look like up close before it’s been made operational? Mostly, an awful lot of I-beams in the ground with pipe attached. And a lot of bare dirt and dust.

Empty Sections and Panel Pipe

The plant, as I found it, was still under construction. Almost everything looked done in this section, with the exception of the panels - and I’m unclear as to why. Perhaps the panels weren’t available in the needed quantities, or perhaps they were finishing out other sections entirely before putting panels on.

The plant is… just this, over and over and over. There’s a single main support pipe for the panels, hanging in pivots that allow it to rotate back and forth to track the sun. This is all supported on small I-beams either concreted into the ground, or perhaps just driven in with some variety of large impact driver. There’s a decent amount of adjustment on them, as I’ll show later.

Take a look at the very end - there’s a pair of shock absorber type devices that seems intended to prevent vibration and oscillation of the panels at the end of a long, clearly somewhat elastic bit of pipe.

Looking left, you can also see that the pipe follows the terrain. The ground is mostly flat, but this system seems entirely capable of following dips in the terrain - which should reduce overall costs. I don’t know how much of a slope the system can work on, though.

This is the end pivot. Important things to notice here: All the parts simply bolt together. There is a lot of adjustment in how things can bolt, which means that the installation of the I-beams can have loose tolerances. Look at the width of the slot in the lower mount. Good for keeping costs down. Also, the yoke that the shock absorbers connect to is a separate device, simply U-bolted over the pipe. It’s not a different model of the main pivot. And, bolted to the pipe with easy-to-install U-bolts, the mounts for the panel frames, in whatever form those take. As far as I can tell, the shock absorbers are only at the end, with the main motor drive installed in the center. I’m not certain how stiff they are, but to prevent any sort of nasty vortex shedding oscillation in the rows, I would assume “very stiff.”

In conversations with people about DIY single axis trackers, there’s been a lot of talk about bearings of various sorts (mostly of the “large” and “ball or roller” variety). How would you mount and align them, how do you adjust them for minimum friction, etc. Clearly, we’ve been approaching the problem very wrong. That’s not what commercial trackers use. This is a plain, not-even-greased-as-far-as-I-can-tell bearing consisting of “a rod driven through some stamped metal supports.” The core tube simply pivots back and forth below the support, not aligned with the axis of rotation. You’ll see how that works later.

The stamped metal main supports are, as far as I can tell, identical - just reversed to each other, and then riveted together. “Minimum Viable Support” sort of work here - which, of course, is the art of engineering. As I’ve heard it described, “Anyone can build a bridge that will support a truck. It takes a good engineer to build a bridge that can only barely support a truck.” I certainly build my stuff hell-for-stout because I’m not a mechanical engineer…

I genuinely don’t like that empty hole there. It’s probably for the mounting of the shock yoke, but that’s also a nasty little space for endless wasps to build homes inside the pipe.

Looking at the pivot from the other side, you can see how the U-bolts are double nutted for the panel mounts. Interestingly, the shock absorber mount is only single-nutted (if you look back a few pictures). But it’s also bolted through some holes in the steel pipe to ensure it’s rigidly attached and can’t slip (to the left of the U-bolt). Again, there’s a lot of adjustment range in those I-beams, and this allows for aligning stuff after it’s been quickly installed. As much of this as possible seems to just be stamped metal parts. The yoke for the shocks is more substantial, but there’s only one of those at each end.

I like that the nuts for the shocks are cotter pinned in place. But I’ve no idea what the nuts/bolts for the pivot assembly are. They don’t seem to have a head you can grip to, and the nuts also seem smooth and round. Any idea what these things are? I assume some sort of “rapid assembly” system, I just don’t know what sort.

The very end of the pipe is a different style connection. This is a band-type clamp, secured in place, with a cover on the end. This should keep wasps out… but also, there are empty holes elsewhere in the pipe of a size well suited to wasps. I don’t like wasps, and do my best to build wasp-resistant things, as I’ve learned over the years just how irritating those things can be in all forms.

There is a lot of PV wire coiled on the end. Why so much? I’ve no idea. I don’t know if this will be cut to length, or just left coiled after the panels are installed, but it runs all the way back. I believe this is 10AWG, with the standard 1000/2000V insulation, though I didn’t take a particularly close look (hands off inspection only).

Tracker Motors

In the center of each string is a NEXTTracker module. These are the drivers for the single axis tracking, with solar panels and the rotation motor assembly that moves each string of panels. You’re welcome to try to find more than marketing details, but I didn’t have much luck. It seems to be an entirely self-powered module on each string - I saw no evidence of any connection to any sort of grid power to manage tracking, and the antenna on top of the motor/main bearing assembly argues that they’re wireless and quite independent of any sort of wired infrastructure. You can see the motor sticking out on this side, with what I would assume is a worm gear drive to the main pivot. You can also see how the “not aligned with the axis of rotation” problem is solved - those bent coupler segments adjust things so the motor is aligned with the pivots. Lots of flexible wiring around here.

The back side shows just how much adjustment is in this mechanism as well (where it mounts to the I-beam), and the main control module. There’s an emergency stop button, a solar panel for tracking, and some cable glands. No idea what’s in there, or if it has some batteries or capacitors to handle really dark days in the winter. I’d like to think that the “central smarts” would have units park in a sane position if fog or heavy clouds are expected (as where you point the modules doesn’t matter on those days), or perhaps that’s enough panel to run the motor slowly on really dark days. No idea! I will say that the maximum angle here should be more than sufficient to rapidly shed snow, if pointing east.

Again, TONS of adjustment range here for non-vertical I-beams. Also, is it leaking grease? That’s a grease zerk in the center of the mechanism cover.

Power Electronics

Coming off the other end of the support tubes, the PV wires just loop over to the DC disconnects. Because the area is fenced in, they can get away with wire bundles like this. If you fence in your home PV array, you could do this too - but if not, you do need to keep fingers off the wires via other methods. I like the bent-wire based wire holders back there. Again, everything here seems to be mounted on I-beams, driven into the ground. There’s a theme, and I’m fairly comfortable guessing that these aren’t concreted in place - simply driven several feet into the ground with a suitable bit of machinery. Anyone know anything about renting I-beam drivers? This could be a nice improvement over my standard “bore holes and pour concrete” style mounts for residential ground mount…

Remember, unistrut on I-beams. That’s professional grade - and, really, it is. Every aspect of a big solar project like this is cost-optimized, so it’s worth considering for your residential systems too. It works.

A row of big DC disconnects. Nice fat conduits head down into the ground, with the conduit secured into the lower unistrut rail.

Finally, the main inverter and transformer for this end - a SunGrow unit. Chinese inverters, at their finest. This is a big unit, coming in on a 20’ container frame for oceanic shipping. I’ve no idea what it weighs…

So, that’s what a big ground mount commercial solar plant looks like under construction. Hopefully some people might get a few ideas to optimize residential solar installs out of it!

The Natural History Museum of Utah Solar Install

I talked about our trip the Natural History museum a few weeks back, but on their roof is a good sized PV system as well! With some views of other PV systems!

This is a far older system (2011-2012 or so) on a “green roof” sort of roof, so it’s interesting to compare and contrast the designs. A decade ago, this sort of thing was far, far less optimized for cost.

From the roof, you can also see some rather large PV installs on the parking garages. I don’t know any details on those, but they look to be broadly similar systems - and a very, very good use of the top deck of a parking garage! Nobody wants to park up there in the sun anyway…

Scattered between panels are rooftop garden beds with embedded drip irrigation systems. Being summer when we visited, they weren’t the most impressive things I’ve ever seen. I still struggle a bit with the whole “green roof” thing - they’re popular enough, but add somewhat substantial weight to roofs. Soil, especially wet soil, isn’t light. For a small home embedded in a hillside, great. Have at it. For a big industrial type facility? They’re neat, but I question if they’re actually useful for any of the things they’re claimed to be useful at - compared, for instance, to PV panels helping shade the roof.

The panels used have square (vs the more modern rounded) silicon wafers. I know bifacial panels weren’t a thing when these were installed, but a bright white roof seems like a great spot to use bifacial panels - the backing plate is clear, so light that gets through and around the panels will bounce off the ground and add power on the way back up and through the panels. Supposedly they add some useful efficiency in the right conditions. I just want a set to build a solar pergola or something with - they’re beautiful to look at, when done right! These are 240W panels, so a good bit below modern 60 cell panels. I’ve not found any data on production degradation in the past decade, though.

Some junction boxes, with familiar looking labels, feed into a Solaron… something. I don’t know exactly what it is, but you can see here that unistrut hasn’t changed as a preferred mounting system in the last decade! This may be a rapid shutdown system - the labels are consistent with a disconnect, but I don’t see any manual actuation handles on these boxes.

I think they’re building an observatory up the hill from the museum. At least, that’s roughly what it looks like from here!

Under each row of panels are the normal manual DC disconnects I’m used to seeing. Here, operating current is 64A and operating voltage is 391V, for ~25kW per disconnect. With 240W panels, that’s about 104 panels, and I would guess that they are strings of 13 panels, with 8 strings in parallel for each disconnect here. But I’m not really sure based on this label, and inspection of the roof layout doesn’t give any useful answers either. I’ve no idea how these things are actually wired up. The installation seems to be “bolted from below” into the I-beams, which is certainly a far more labor intensive approach than the modern panel clamps - though I’m not sure there were other options when this was put into service.

I do miss living close to the mountains in New Mexico sometimes…

I will, however, point out that 1378 panels does divide evenly into 106 strings of 13 panels… and doesn’t divide into 11 or 12 panel strings evenly. It works out to about a 330kW nameplate system install, and with 457,500 kWh/yr generation, that works out to 1.38 kWh/W. This is only 5% higher than my home system’s 1.31 kWh/W number (the A-frames are not great on a per-watt production basis but make for very long solar days), so I have to assume there’s some regular shading from the mountains nearby. South facing panels should produce a good bit more. But even this is only 25% or so of the museum’s annual consumption - so they’re using ~1800 MWh/yr for that building. Whoof. And not an EV charger in sight.

A rather dated monitoring system indoors provides some other stats, some of which make no sense at all. 635kWh for the day makes sense, 54MWh for the month, OK… and 27MWh for the year so far? Uh. Some wrap around error may have occurred? The lifetime production is an impressive 4.3 GWh, which seems mostly sane for a ~10 year old system estimated at 457 MWh/yr. They’re using a high CO2 estimation value, though - 1640 lbs/MWh, which looked quite high to me. I went looking for some data, and… yeah, that’s about right. In 2018, PacifiCorp’s emissions were 0.68 MT CO2/MWh, which is about 1500 lbs CO2/MWh, and that’s been coming down over time. When you’re mostly coal and natural gas, I guess you end up there.

Nothing too interesting in the daily curve either, when we were there.

Enjoy! Hopefully you’ve found something interesting in this post! I promise this is the last Utah Tourism post for a while.