A build-it-yourself, adjustable ground-mounted rack and DIY installation provided deep savings on a PV system for this Wisconsin family.
In 2006, I refurbished a classic Jacobs Wind Electric wind turbine and installed it on our family’s property in Merton, Wisconsin. The “Jake” produces an average of 110 kWh per month, offsetting about one-third of our household’s total electricity use on a yearly basis. However, in the summer months, when winds average 8 mph, the Jake was only generating about 40 kWh per month. That’s when I started thinking about installing a PV system to boost our renewable energy production.
In 2011, the prices of PV modules were continuing to drop. I began researching system components, incentives, and permitting requirements. I set an aggressive goal of designing and self-installing an adjustable, ground-mounted PV system for $1 per watt (after incentives), and began my PV project in earnest.
A rooftop array was ruled out for several reasons. First, there was the shade caused by a 70-foot-tall tree near the west side of the house, which severely limited the available solar window. But we were loath to cut it down—the summer shade it provides helps reduce our air-conditioning loads. The largest roof sections face east and west, and while not a deal-breaker, orienting the array to meet Wisconsin’s Focus on Energy (FoE) program rebate requirements, which states that modules must be installed within 45° of due south, would have meant mounting them in an aesthetically unpleasing way.
Fortunately, although trees border our property, there was still plenty of room on our 1.3 acres to find a suitable solar window for a ground-mounted system and meet the town’s minimum 20-foot setback requirements (see “Finding True South” sidebar). The window wasn’t “perfect”—my analysis showed that some shading of the array would occur in December and January from maple trees that lie just outside of our south property line—but that energy loss could be offset by adjusting the array’s tilt monthly (see “Optimizing the System”).
Once we’d determined that there was a good solar window, we examined our electricity loads (see “Electrical Loads” table) to arrive at a reasonable PV system size. According to our utility bills, our household consumes an average of 321.5 kWh of electricity per month, or 3,858 kWh per year, which costs us about $630. (We use natural gas for our forced-air furnace, water heater, oven, and cooktop. We also have two wood heaters, both of which are used often in the winter.) But besides meeting our existing electrical loads, we also wanted the system to generate enough electricity to recharge a plug-in hybrid vehicle or electric vehicle in the future.
Wisconsin’s FoE program offered a cash-back rebate (up to $2,400) for grid-tied PV systems, and we could also take the 30% federal tax credit. Balancing these economic and energy goals, I arrived at a 6.1 kW system, which would max out the FoE rebate and provide extra electricity for an EV. This system size would also maximize the inverter’s energy-conversion efficiency and stay within the weight parameters of the wooden rack structure I had in mind.
On January 2, 2013, I applied for Wisconsin’s FoE Renewable Rewards program. To be eligible for the program, I had to retain the services of a PV installer to oversee and sign off on the installation. I hired Trang Donovan, who provided me with design reviews, recommendations, and a final system inspection for compliance to the National Electrical Code.
Solar energy systems are eligible for a federal tax credit of 30% of the system cost, with no upper limit, until December 31, 2016. I also applied for solar renewable energy credits (sRECs) through Ethos Renewable Power, which would pay $13 per megawatt-hour (MWh) generated.
After the Rewards program application was approved in February 2013, I began sourcing PV system components. The $1-per-watt goal was an aggressive cost target set to achieve a return on investment (ROI) of at least 18%.
I implemented several strategies to make the most of every solar-generated electron at my site.
Adjustable, ground-mounted rack. Although mounting the PV array at a fixed tilt is the most straightforward strategy (and the least amount of work, since you don’t have to adjust the rack), it incurs energy penalties. The sun changes its angle from the horizon, being higher in summer than winter, so annual energy output can be increased by seasonally adjusting module tilt. From the sun-path chart I generated (bit.ly/UOsuncharts), I determined an array tilt to maximize the system’s energy output for each month. I’d need to be able to adjust the array tilt from a minimum of 19.6° in June to a maximum of 66.5° in December. Over the course of a year, I calculated that the monthly tilt adjustment would yield about 4.6% more energy compared to a fixed array (see “Estimated Production” table).
In snowy regions like Wisconsin, a steeply tilted array sheds snow more quickly, which is important for maximizing the system’s overall production. Following a 4-inch snowstorm, for example, our steeply tilted array is snow-free the following sunny day, while I’ve observed that it can take more than a week for the snow to melt from a fixed-tilt, ground-mounted PV array that’s on a church property 1.5 miles away.
In the summer, a ground-mounted system has better airflow around the array, which means lower module-cell temperatures and higher energy production compared to a similarly sized roof-mounted system. Since PV modules have a negative temperature coefficient, keeping them cooler increases their power output. For example, my modules have a temperature coefficient of -0.41% per degree Celsius (i.e., for every degree above 25°C, the PV modules’ power output is derated by 0.41%). Roof-mounted PV modules typically run 25°C or more above ambient temperature, while ground-mounted modules typically run slightly cooler (i.e., my estimate is 22°C above ambient). The annual average high temperature for my region is 13.5°C. For a roof-mounted system, the production hit due to heat would be 338 W [6,100 W × -0.41%/°C × (13.5°C + 25°C) - 25°C]. With a ground-mounted array, the average power decrease would be 263 W [6,100 W × -0.41%/°C × (13.5°C + 22°C) - 25°C]. So in my case, a ground-mounted PV system results in 75 W (338 W – 263 W) more power output per peak sun-hour of operation. This translates into an additional 124 kWh per year (4.52 sun-hours/day × 0.075 kW × 365 days/yr.).
No batteries, a high-efficiency inverter, and high-wattage modules. I chose a batteryless grid-tied system for several reasons:
There are many batteryless grid-tied inverter choices, but since I was already satisfied with the Power-One 4.2 kW Aurora transformerless inverter used with my Jake, I decided on an Aurora (Power-One) 6 kW inverter for our PV system. The 6 kW inverter can process two PV strings with independent maximum power point tracking, and has a CEC-rated efficiency of 96.5%.
I chose Helios Solar Works 7T2 (305 W) modules for several reasons. First, I wanted to keep my hard-earned dollars in the United States, and these modules were manufactured in Milwaukee—close to home. I could rent a trailer for $20 and drive to the Helios Solar Works factory (a mere 25 minutes away) to pick up the 20 PV modules, saving significantly on shipping costs. Second, putting 10 of the PV modules in series hit the Aurora 6 kW inverter’s voltage “sweet spot.” And, third, dimensionally, these modules allowed for a reasonable rack wooden header length that would minimize module deflections caused by wind and snow loading.
Maximizing PV output. The voltage of PV modules in series is additive, but the string’s current is limited by the lowest module current (Imp)—so it is best to place all the modules with the highest Imp in the same string. Using the manufacturer-provided data measured for each module, I methodically went through the list to select modules to string together. This allowed each series array to perform within 0.3% to 0.4% of their theoretical maximum power output.
Minimizing electrical losses. Properly sizing PV electrical conductors can decrease power loss from voltage drop. Generally, the total DC voltage drop should be less than 2%. I used #10 copper PV wire in the PV output circuit, which results in only a 0.7% loss. A tolerable AC voltage drop (inverter output circuit) is between 1% and 1.5% for grid-tied systems. For the inverter’s maximum output of 25 amps, the AC voltage drop was 0.7% using #2 AWG aluminum wire.
I knew I could economize and improve my system’s return on investment by building a custom rack structure using locally available, relatively inexpensive lumber and readily available metal components. Had I purchased a manufactured ground-mount rack, I would not have had any chance of achieving my $1-per-watt goal. The most similar manufactured option was a metal rack (about $4,892, with shipping). Besides being more than twice the cost of a home-built rack, this rack would have been much bigger (46 lineal feet versus 33), subjecting the array to more shade because of its length, and was not as adjustable, as it only went to 50°—I wanted my rack to tilt to 66.5°.
The most complex part of the custom-rack design process was calculating loads (see “Design Loads” sidebar). I also needed to create a simple yet durable tilt method that would accommodate some design imperfections, since nothing would be dimensionally “perfect.” Pillow blocks—bearings used to provide support while allowing rotation—used with 2-inch-diameter, schedule-40 pipes, would allow each of the roughly 1,500-pound rack assemblies to be adjusted. The additional cost for this method was $360; it will likely be recovered in about seven years when assuming the additional 4.6% output that results from being able to adjust the array monthly. During the array’s expected 25-year lifetime, the tilting feature has an ROI of 14.3%—well worth the extra design effort and expense.
The racks that hold the PV modules each consist of two 2-by-12 headers mounted onto a 2-inch-diameter pivot pipe. Six 2-by-12 joists sit on joist hangers between the headers, and were glued and screwed together. Eight 1-by-4 stiffeners were glued and screwed to the joists. Joints were sealed and the counter-bored lag screws were topped with silicone caulk before the entire rack was coated with deck stain to protect it from degradation.
The support structure had to be robust to support the two wooden racks. The horizontal force of the wind on the center I-beam support was calculated to be 3,470 pounds. A 4-by-8 inch I-beam made from ASTM A992/A572 (grade 50) steel provided a two-to-one safety factor. The 18-inch-wide, 4-foot-deep concrete footings were also designed to handle this wind loading with sufficient margin, and were excavated below the 4-foot frost line to ensure that the structure would not shift from freezing and thawing. Rebar in the bored holes increased the loading capability.
Both wooden racks were pre-assembled to ensure proper fit, then taken apart for assembly on the tilt structure. Before mounting the PV modules, a strip of aluminum-foil tape was bonded to the top of each 1-by-4 stiffener. Each module was mounted to the wood frame using 5/16 stainless steel hardware through four of the specified module mounting holes. Modules were spaced 1/4 inch apart. A #6 bare, stranded copper wire serves as an equipment-grounding conductor (EGC); each module was connected to the EGC with a WEEB ground lug. The 10 modules in each string were connected in series using their MC4 quick-connect plugs.
The inverter was installed in the existing wind turbine inverter shed. The shed already had 240 VAC and a subpanel with a spare circuit breaker location. Mounting the inverter in the shed gave a short PV output circuit cable run (approx. 58 ft.) and protection from the weather, animals, and vandalism.
My original target for the PV system was $6,100 (or $1 per watt), after incentives. The final cost came in at $7,661 (or $1.25/watt)—not too far off-target. The tilting feature has been working flawlessly. Adjusting the two arrays takes less than five minutes; I actually look forward to it, as I know tilting increases our PV array’s monthly electrical output. Our grid-tied system is net-metered with a monthly true-up. Rather than paying electric bills, we are now receiving monthly checks from the utility. In its first full 12 months, our grid-tied PV system generated 8,618 kWh—within 1% of that predicted by PVWatts. In that time, the system has earned $1,286 [($0.136207/kWh × 8,618 kWh) + $13/MWh × 8.618 MWh)] worth of electricity, resulting in a system payback of about six years. Under our grandfathered net-metering agreement with the utility, we are paid at the residential retail electricity rate for any surplus electricity that our PV system produces each month. If the earned credit for our PV-produced electricity is less than $25, it is carried over to the next bill. So far, though, we’ve been receiving monthly checks from the utility.