This two hour program is for anyone wanting a rapid introduction to solar ahead of a project. It is primarily focused on residential design considerations.
If you would like to take this for continuing education, please register here.
If you would like to go back to the slideshow, click here.
Welcome to Two Hour Solar Power.
My name is John Cromer and I am a NABCEP PV installer, as well as a mechanical engineer from the University of Pennsylvania, and master electrician. I’ve worked in solar for twelve years, largely in residential and commercial markets. I’ve taught solar throughout this time as well.
Today I will share with you ten important considerations to know when approaching a solar power project.
- Performance Estimation
Let’s perform a PVWatts example to determine how much energy a solar array will produce. We’ll use PVWatts to do this, a free and online resource published by the Department of Energy. The same data is used by commercial design software for energy modeling, and so PVWatts will tell you much environmental data about the project site under close look, despite its basic appearance.
PVWatts makes it easy to get a good idea of how an unshaded array will produce its electricity for every hour of the day, every day of the year, based on local weather data. Input your address and system size to get started.
I recommend performing a PVWatts example using a 1 kilowatt solar array size, to keep things simple. We can see that a 1 kilowatt solar array in Houston, TX, facing due south at a 20 degree roof tilt angle, will produce 1400 kilowatt hours of energy per year.
Summer days will produce more than winter days, because the days are longer.
Moving a step back to more closely examine the input data, the roof tilt information button states that a common 5:12 roof tilt is at a 22 degree tilt. The solar array is facing 180 degrees azimuth, which is another way to say due south. A 90 degree angle would face due east. A 270 degree azimuth faces due west. All this can be modeled in PVWatts.
Now we know enough to run an experiment. Does a due west facing array perform any differently from due south?
At a 5:12 roof pitch, the performance difference between south and west facing is 9%. Similarly, the performance difference between south and east is 9% . If the array is at a steeper tilt, this difference increases. For example at a 12:12 roof pitch, which is a 45 degree tilt angle, the performance difference between due south and due east is 13%, still not much. Moving north, up to Indianapolis, the performance difference between at a 12:12 roof pitch for due south vs. due east is 23%.
In other words, the orientation of the solar array, or even the tilt angle, does impact array performance, but perhaps not as much as commonly thought. You can find solar arrays bolted 90 degrees down the sides of buildings, such as in New York City where electricity is expensive and rooftops are cramped. Or for off-grid power in Canada to avoid snow. Similarly, at a shallow tilt angle, the north side of the roof isn’t off the table. So while east and west facing solar arrays are fine, and even north-facing solar arrays in some circumstances, most solar arrays in the northern hemisphere face mostly south. So far as roof tilt goes, the array will follow the roof. Tilting up off the roof rarely makes sense, especially with structural considerations. Steep angles will increase production, but also increase installation and operating cost, as the roof will no longer be walkable.
If you memorize your local PVWatts solar performance ratio for a due south rooftop array, at a standard roof tilt, you will be able to perform off-the-cuff solar performance estimates to impress potential clients, friends, and coworkers.
If I know for my area, 1kW of rooftop solar will produce roughly 1400 kilowatt hours per year, then 1 watt of solar will produce 1.4 kwh per year. That figure might be discounted 10-20% if the tilt angle is atypical, but for the most part these calculations can be performed mentally in conversation to be later confirmed with a computer model. 1 kilowatt array produces 1400 kilowatt hours per year. So a 10 kilowatt array produces 14000 kilowatt hours per year. A 5 kilowatt array produces 7000 kilowatt hours per year. A 5 kilowatt east-facing array produces around 6000 kilowatt hours per year. A home that uses 11,000 kwh of electricity a year will need a 8kW solar array, give or take a watts.
Before leaving PVWatts, let’s take a closer look. At the very end of the PVWatts calculation, the option is presented to download an hourly performance estimate.
Additional site information is revealed, such as the amount of sunlight in the air and on the solar panel itself, as well as rooftop temperature, ambient temperature, wind speed, and more.
All of these factors are being used, along with conservative assumptions about the solar panel itself, to produce an estimate that accounts for both sunny and cloudy days, the changing position of the sun in the sky throughout the day, performance losses due to heat, voltage drop, and system availability.
The two major items PVWatts does not account for is shade or snow. So far as shade goes, I recommend using commercial software to build a 3D model and perform shade analysis based on that. For snow, one simply needs to discount the winter season, perhaps entirely, if it can be expected that snow will sit atop the array for an extended period of time each year.
The short of it is, by extracting this hourly data into a spreadsheet, PVWatts data can be used as a foundation for more detailed design, such as estimating off-grid battery bank capacity, quantifying the economics of selling back electricity to the utility, or evaluating alternate billing rate structures from the electric provider.
- Payback Calculation
Calculating solar payback is another essential industry skill. A great starting point is to calculate simple payback, which indicates how many years it will take for project expenses to be recovered. Let’s assume you want a ten year simple payback on your solar project. What should your budget be to achieve this goal?
Let’s assume the solar array has an effective generation rate of $0.10 per kilowatt hour. This number largely depends on how much the utility is required to purchase back solar power during the day, particularly if the customer does not have batteries. But for now, let’s assume the customer earns $0.10 per kilowatt hour for the solar production, regardless of whether it is sold back to the grid or used on site.
From the PVWatts section, the local production figure comes back into the play. A common ratio for much of the United States is that one watt of solar will produce 1.4 kilowatt hours per year.
Multiplying 1.4 kwh/year by $0.10/kwh lends a new payback metric, $0.14 per watt per year. If the project needs a ten year simple payback, then it should have a budget of $1.40 per watt.
The upfront tax credit is commonly included in simple payback calculations. Currently, the tax credit is 30%, set to drop to 26% in 2020-2021. After taking the 30% tax credit into account, the project budget should be no greater than $2 per watt to achieve a 10 year payback.
According to recent solar installation pricing provided by EnergySage, average solar installation pricing is closer to $3 per watt. Applying a 30% tax credit, 1.4 kwh/year/W production rate, and $0.10/kwh electric rate, the simple payback would be 15 years.
In other words, simple payback is primarily a function of production rate and electricity value. PVWatts can be used to determine how solar production value varies somewhat throughout the country. In Reno NV, 1 watt of solar may produce 1.6 kilowatt hours per year. In Buffalo New York, the same watt of solar produces closer to 1.2 kilowatt hours per year, a 25% difference.
But the price of electricity, as well as utility buyback rates for solar power, vary much more widely than solar production. For this reason, it is safe to assume that solar power markets are driven primarily by the price of electricity, with the amount of sunlight in the air playing second fiddle. Because of the price of electricity, solar power has a faster payback in New York City than in Houston, TX, despite Houston having cheaper installation cost and better solar production.
The simple fact is that New York City electricity costs 2-3 times as much as in Houston TX, allowing New York’s project economics to fair better despite its higher installation prices and lower system production.
Now let’s take a look at a typical residential project budget.
Solar has dropped in price over the years but cost reductions in the most recent years have been offset by increases in import tariffs. So based on EnergySage pricing, even 2015-2016 budgets are still realistic of what you can get today with the import tariffs included.
In dark blue, the cost of the solar panel itself hovers between 40 cents per watt up to 60 cents per watt including the price of the import tariffs. The panel hard cost is about the same whether or not it’s a residential, commercial, or utility-scale panel. The mark-up on the solar panels that a distributor charges, which is more to residential markets than commercial, is modeled up here in grey which we’ll get to shortly.
Here’s the electrical balance of system material budget. I don’t recommend cheaping out on electrical balance-of-system material as there’s some nice things you can add to a solar array to improve its quality, beyond the panel or inverter. A generator bypass switch at the top of the service panel can allow the home-owner backup power capability for the whole house. Many potential solar owners are surprised to learn battery-less solar does not have the ability to provide power during the blackout, and even battery-based systems typically do not supply the whole house. Because battery prices are still dropping, the most cost-effective solution for a solar owner who wants back-up power capability can be to use a backup gas generator for now and wait for the price of batteries to drop further. Owners want their systems looking clean and polished, which can be achieved with a little bit larger balance-of-system material budget for internal conduit runs, breakers and subpanels, instead of fuses and clunky disconnect switches. Modest upgrades will not increase project cost exorbitantly but help the project obtain a nice, polished look.
So how would one calculate the payback on a system installed at this budget?
Again, PVWatts is used to determine how many kilowatt hours of electricity a solar array will produce per year. Use a 1kW array as an example, and divide by 1000. Most likely, one watt of solar will produce between 1.3-1.5 kilowatt hours of solar production per year.
Next, identify the local net-metering policy by visiting www.dsireusa.org. DSIRE is a website of green energy incentives and policies at the local, state, and federal level. While visiting DSIRE, check for any state sales tax exemptions. But also use it as a starting point to determine the utility buyback regulations.
A strong consumer net-metering policy results in a one-for-one retail-priced exchange for the electricity the system will outflow onto the grid. In other words, every kilowatt hour the system outflows onto the grid results in a credit of equal value to kilowatt hours delivered by the electric provider.
The most solar-friendly net-metering policy results in electricity consumed on-site versus electricity sold back to the power company being valued the same.At the other end of the spectrum, without batteries, it is safe to assume that 2/3rds of system production will outflow onto the grid. After all, society uses electricity 24/7 whereas solar only produces between morning and evening, with most of its production in the middle of the day. So if the goal is an array that offsets 100% of a facility’s energy use, roughly 2/3rds of that production will either need to be stored in a battery or sold back to the utility. With so much production needing to be bought back, if the array lacks batteries, it is a good idea to contact the utility to confirm the net-metering policy before proceeding any further with a project. Likewise, it is necessary to examine the electric bill to back out any fixed charges from the bill before determining the effective generation rate of the solar array.
In any event, the effective generation rate of the solar array is either 100% of the utility energy, assuming a solar-friendly net-metering policy, or on the other end of the spectrum, up to 2/3rds of the array could be discounted as much as 80%, such as if the utility buys back for the federally mandated minimum – an amount known as avoided cost. Once the local buyback policies are determined, the effective energy rate of the solar array can be quantified in a $/kilowatt hour unit.
Now payback can be calculated. The project budget divided by the project size to get a dollar per watt figure. Solar production is measured in kilowatt hours per watt per year, and the effective generation rate is billed in dollars per kilowatt hour. Combining these figures together to get the simple payback measured in years.
Lastly, check the value of any tax credits or incentives. The federal tax credit for solar, which includes batteries, can vary between 10-30% depending on governmental changewinds. Sales tax exemptions can take a year or more off of system payback.
Assuming a $2.51/W installation budget, a state sales tax exemption worth $0.16/W, a federal tax credit of 30%, a generation rate of 1.3 kilowatt hours per watt per year, and an effective generation rate of $0.10/kwh, the simple payback of this solar array would be 13 years.
- Array Layout
At a very high-level, solar array layouts are an exercise of how many rectangles can fit into the larger rectangle, which isn’t too daunting of a do-it-yourself task. Shade on a solar array should be avoided, but a small amount of shade can be managed. A good rule of thumb is to completely eliminate any solar production from the PVWatts model for the time that a solar array is shaded. I perform shade analysis using commercial solar design software, because the market leading software companies integrate LIDAR data into their services, making shade analysis accurate and easy. This is the same data as what Google Earth uses to make its 3D models of trees and buildings. However, solar installers may also use specialized field survey tools to calculate shade loss percentages, or even basic trigonometry to get a rough idea of shade conditions to be avoided.
It should be obvious that a good solar jobsite is one with a large open surface area with access to sunlight. While partial shading can be designed around, the whole point of solar is to not be in the shade.
In the PVWatts section we learned an east facing solar array will produce 15-30% less than a south-facing solar array, depending upon the tilt angle of the roof. Let’s expand that philosophy to also consider the north side of the roof. A due north facing array in Indianapolis, IN at a 45 degree tilt angle produces 55% less than it’s south-facing counterpart. Combine the two roof surfaces together for a total unshaded yield of 145%. An East and West facing array would produce 77% less than their south-facing counterpart, for a total unshaded yield of 154%. In other words, if one covered the entire roof with solar panels, roughly the same amount of energy would be harvested from the roof, regardless of how the building is oriented.
Southern roof surfaces are certainly more desirable than east or west facing roof surfaces, and north-facing roof surfaces should be considered last, and only if there is project budget remaining. But consider economies-of-scale. Upgrading a residential project from a 20 solar panels to 40 solar panels will not double the design, labor, or balance of system material costs. Nor would it double the permit fees, engineering costs, or sales commissions. This is called economies-of-scale. The larger the solar project gets, the cheaper it gets per watt. In a region with an average installation price of $3 per watt, a small residential project might cost $5/W and a large residential project could cost $2.50/W. By increasing the project size, the economies-of-scale can drop the project cost more quickly than the performance loss resulting from moving onto less-than-ideal roof surfaces.
The complexity of the roof is as important as roof tilt and orientation. A walkable roof maxes out at a 5:12 roof pitch. The steeper the roof, as well as the height of the roof, as well as the type of roof can substantially increase project complexity and cost. As a solar installer, I want to select a couple large roof surfaces with access to sunlight. The orientation of the roof is only one consideration. In fact, in a southern climate with a shallow roof tilt, I might cover the entire roof if the budget and electric bill allows.
Solar is an outdoor rated electrical device designed to withstand the elements. In a catastrophic hail storm, a solar array is more likely to save the insurer money by avoiding a roof replacement, rather than cost them money with a solar array replacement. But that assumes the solar array protects the entire roof. Perhaps the greatest conflict in solar design is that between installers and architects. Solar wants a large and boring roof, whereas an architect wants complex roofs broken into little sections.
Rather than list the nuances of every solar design and optimization theory, let’s keep it simply by suggesting a good residential solar project consists of at least one pallet of solar panels, covering the best two or three roof surfaces available. Then if additional budget allows, more panels and less optimal roof surfaces can be considered.
Likewise, try to avoid placing a couple of solar panels off by themselves on a small roof surface, regardless of how ideal the orientation is, simply because of installation complexity. A pallet of solar is almost always between 20-30 solar panels roughly 3 and 1/3rd feet by 5 and 1/3rd feet. So good roof surfaces that are rather large and boring, regardless of orientation. Similarly, an east-west roof can be as viable as a north-south roof in terms of project siting.
I don’t even mind systems which produce more electricity than the building consumes. You can always find a use for more electricity, such as reduced reliance on gas or wood heat in the winter.
One of the more important decisions of solar design is considering rooftop accessibility, as slanted roofs are not designed to be accessible places. But with a solar array located on the roof, that roof surface will inevitably need someone back up on it for array servicing at some point in the near or distant future.
It is possible to cover an entire slanted roof from edge-to-edge with solar panels and maintain compliance with international building and residential building codes. It is also common for local authorities, such as fire departments, to add additional stipulations, such as a common “ three feet offset from the sides and top of the roof” requirement.
The main concern is that the building need an exhaust plan to ventilate smoke from attic spaces and lofted ceilings in the event of a fire. So if a sleek edge-to-edge look of a modern residential solar home is desired, know that it is possible to obtain exemptions from the 3’ offset rule on a slanted roof even where mandated, if other kinds of ventilation systems are planned in those spaces.
Flat roofs, such as commercial roofs, have their own rules and regulations, which are commonly staying 6’ from the edge of a roof. That is simply good practice to discourage maintenance workers from getting to close to the edge of the building, although it can make solar on top of tall skinny buildings more difficult.
At any rate, there are valid reasons to stay off the edge of the roof with the solar array.
Wind speed is greatest at the corners and the edges of the roof, such that the solar array will result in more force on the roof truss during extreme wind if installed on the edges of the roof. The roof itself may overhang the structural wall, and so the attachments on the interior of the roof will be stronger. The roof cap or ridge may need space for servicing. A design change in the field may be more easily overcome if there is some empty space remaining on the roof. Lastly, as an installer, there is a big difference between climbing onto a roof to access a troublesome solar panel, as opposed to disassembling the solar array to clear a pathway to where the trouble is.
I love the solar bling bling look that takes a normal boring roof and turns it into something special. But the more accessible the roof, the better. In fact, if the solar array is being installed on a rooftop that is wholly inaccessible, that is a good time to spend additional money on quality, top-shelf components (and labor) to better ensure the trips back up on the roof are infrequent.
A pre-built roof is not intended to be a construction site. Clearances around the skylights or other roof-mounted equipment become walkways for solar construction workers. When leaks occur, it’s usually in those areas rather than caused by the solar array itself.
Shingles in particular do not take much wear and tear. But a well-built solar array should extend the life of the roof. Shingles are primarily degraded by exposure to sunlight, so shading a shingle will increase its life. But plan the project with the intent to reduce the total amount of time spent on a roof as much as possible. Pre-assemble as much as practical on the ground to reduce the time spent on the roof itself.
Let’s end the array layout section with an interesting energy metric. The Department of Energy pegs commercial average electrical usage being 14 kwh per square foot, and let’s assume its that for residential.
How many kilowatt hours per square foot does a solar panel produce in a year? If a solar panel is 3 1/3 feet by 5 1/3 feet, and also assume it is a typical 300 watt solar panel. Let’s assume one watt of solar produces about 1.3 kilowatt-hours per year, so multiply 300 watts by 1.3 w/kwh/year to get about 400 kilowatt hours per year. Dividing by 18 square feet results in approximately 21 kilowatt hours per square foot per year from a solar panel.
Of course every building’s energy use will be different and site conditions can impact solar production. But the point is that solar panels can produce more energy per square foot than what an average building uses per square foot, by a healthy margin. This implies that a solar array can produce all of a building’s energy needs by harvesting energy from the roof, even without covering the entire roof to do so, at least for a one-story building. On a two-story building, it is still possible to generate a 100% energy offset when all sides of the roof are used, which is possible if the roof has a walkable tilt angle. Of course the taller and skinnier the building the harder it will be to take that building completely off-grid.
At this point, a design algorithm emerges for selecting the ideal roof services for planning a solar array on a roof. Start by identifying large, open sunny areas on the roof. South-facing roof surfaces are best but anywhere between due east and due west is just fine. If the roof has a 5:12 pitch or less, consider the north side of the roof as well.
If the project skips some of the design phase, such as not being reviewed by a structural engineer, nor having a third-party inspector get into the attic to verify the lag bolts hitting the rafters, nor checking with the local jurisdiction to see if there are local roof offset requirements, then it is safe to maintain a 3-foot offset from the roof ridge and rakes (i.e. the top and sides of the roof). Even a slightly larger offset has benefits. A 4 foot offset gives the installation crew even more flexibility if encountering any unforeseen roof obstacles or structural issues. The further away from the roof edge the array is mounted, the easier the loading is on the roof. That extra space is useful for maintenance. So there are many reasons to not put the solar array all the way to the top or sides of the roof.
The bottom edge of the roof is a different case. I don’t want workers to close to any edge of the roof, but filling bottom region up with solar is best for snow and rain run-off. So a good solar array will start the solar panels a couple inches up from the bottom edge, such that torrential rain will not overshoot the gutters, and then run continuously up to about 3-4 feet from the top of the roof where the budget and design allows.
Personally I would only bother with roof surfaces that can fit 8 solar panels or more, such that the installation crew can work efficiently and a good installation price can be obtained. A pallet of solar panels is between 20 and 30 panels, and a good-sized residential project can be two pallets – which is about 15 kilowatts. Of course more difficult installations are possible and occur regularly. But solar panels do require some maintenance access. Failure in the safety electronics could take a panel offline until someone can put hands on it. A long period of time without rain could result in dust and pollen accumulation on the array surface. Proximity to a dirt road can do the same. Roof access is useful simply for cleaning the array.
So in planning an array layout, identify large, open, accessible roof surfaces, and fill them up with solar panels in a maintenance friendly layout, before moving onto less ideal surfaces so long as there is budget and an electric bill remaining.
- Inverter Design
How do you determine how many solar modules go on a particular circuit which are then wired into a particular inverter?
Inverter manufacturers provide online software to assist this design process. They take the specific information from the module specification sheet, such as the voltage, amperage, and temperature coefficients and use that information to suggest configurations which are compatible with their products.
Close examination of PVWatts data can tell us the maximum inverter size required without overspending on inverter capacity. Typically, inverters are 10-20% undersized compared to solar array nameplate capacity, although oversizing as well as more significant undersizing is not unusual.
Solar cells are prepackaged into solar modules (which is the industry term for solar panels), and circuits of solar modules are labelled by the industry jargon as strings. So there’s an inverter called a string inverter that at the residential level is typically one inverter for the entire system, with the inputs being multiple strings of solar panels.
The opposite approach is to use microinverters where there’s one inverter is installed behind every single module on the roof. Microinverters are particularly interesting because the produce AC instead of DC output up on the roof, eliminating the DC home run cable requirement to be in metal. Which I think is nothing more than DC discrimination, because the metal is only there for phyiscal protection, so there’s no reason do differentiate between AC source circuits and DC source circuits. But that’s a digression.
The third inverter approach is the most popular for rooftop solar, an approach called DC optimizers that is somewhat of a hybrid between string and micro inverters. DC optimizers put a box behind every solar panel on the roof that only regulates a DC-to-DC voltage output rather than doing a full conversion to AC on the roof. The DC to AC inversion instead happens at a string inverter down on the ground. This string inverter has less stuff in it, because the voltage has already been converted by the optimizer on the roof. But this architecture is slightly less expensive than micro-inverters.
Both micro-inverters and DC optimizer systems are more expensive than string inverter systems. But either of these systems are more accommodating of string inverters for shade, and so should be selected whenever there is shade on the roof. Of course, shade and solar power do not mix, but partial shade, such as a single tree shading a portion of the array in the morning or evening, can be substantially mitigated through the use of these products instead of a string inverter by itself.
Some installers do not like DC optimizers or micro-inverters, because these platforms put more electronics on the roof than string inverters by themselves. And those electronics can fail. If a single micro-inverter or DC optimizer fails on a roof, the most cost-effective answer may be to leave it alone until there is another good reason to get up on the roof, rather than incur the cost of replacing the failed unit.
In the absence of shade, which system is bestis a matter of opinion. Without getting too nuanced, I’d recommend micro-inverters for do-it-yourselfers, DC optimizers for experienced installers, and string inverters for ground mounts.
National Electric Code requires solar conductors to be de-energized on command, and a strict interpretation of the latest National Electric Code would mandate the use of DC optimizers or micro-inverters when solar is installed on roof. There is no off switch on a solar panel, so module-level panel electronics are the only means to completely isolate the solar array during an emergency and bring the system voltage down to a safe voltage. So on a roof, it is wise to use module-level panel electronics, despite the added failure points and slight cost increase. On the ground, it is less important.
Regardless, the circuit sizing process is useful to know.
Because micro-inverters are module-level, micro-inverter manufacturers typically provide a compatibility list or calculator, which essentially tells the user which micro-inverters are compatible with which solar panels. The micro-inverter specification sheet will then reveal how many micro-inverters can fit on a given circuit. Micro-inverter design is flexible and any regular electrician with AC experience will feel comfortable installing them.
In this example, a micro-inverter can have up to 16 micro-inverters per 20A branch circuit. If a pallet of solar panels is 25 panels, the system would require two branch circuits of up to 16 modules each. One circuit can be a different size than the other, it doesn’t really matter which direction the modules face, even within a circuit, because the micro-inverters allow each panel to operate independently of its neighbors.
Experienced solar installers will often opt for DC optimizers with string inverters, which bring the code-compliant and shade tolerant advantages of micro-inverters, along with some cost savings. Microinverters cost more than DC optimizer solutions. The compatibility of a solar panel to a particular DC optimizer is similar to the compatibility of a micro-inverter – it is a matter of solar panel voltage compatibility and can be determined by using the manufacturer provided sizing resources. But DC optimizers have an additional design step, which is to also specify a string inverter. So I will demonstrate string sizing first, and then illustrate how it applies to DC optimizers.
Fronius makes a popular string inverter and their sizing process is similar to other manufacturer circuit sizing software. The solar module is selected, along with the inverter. Typically, string inverters are 10-20% undersized compared to the nameplate size of the solar array. This is because the size of the array is the input whereas the inverter capacity is the output. There are losses between the input and the output, as much as 20% in the summer. Again, these losses are modeled in PVWatts and so the exact relationship between DC input and AC output can be closely examined. But the short of it is that the AC inverter is commonly undersized compared to the array size although it doesn’t have to be. So for one 8kW pallet of solar modules, you might look at inverters sized between 6.5kW to 8kW.
At any point, the solar panel and inverter are selected, along with local record cold and average hot temperatures, and then the manufacturer sizing software will reveal all the acceptable wiring configurations of the components.
Most string inverters today have multiple circuits inputs capable of operating independently of the other circuits. It’s possible to have two circuits of five to ten solar panels, or one circuit of seven panels with one circuit of 10 panels, just as an example. Higher end inverters have multiple maximum power point tracking circuit inputs (MPPT), such that shade or mismatched orientation of one string would only impact that string and not pull down the entire system. This is an advancement from string inverters of years past.
For inverters with multiple MPPT, circuit sizes can differ on one circuit compared to another. Similarly, a DC optimizer or micro-inverter system can be thought of to have an MPPT for each solar panel on the roof, such that each solar panel is its own unique system. So circuit design with module-level panel electronics have greater flexibility than with string inverters.
Within a single MPPT, whether it be a circuit or a single module, the cells all have to face the same direction. Say a pallet of solar has 25 panels. This string inverter could accommodate two circuits of eight solar panels facing southeast and one circuit of nine panels facing southwest, for a twenty five panel system working on a single string inverter.
DC optimizer sizing is similar to both string sizing and micro-inverter sizing. Here is a 132 panel layout using some odd-ball thin-film solar panels with atypical voltage and amperage ranges.
The string sizing software suggests that for this 132 panel, 20 kW array, two 10kW string inverters should be used, distributed between six total circuits using the P405 model DC optimizer. This is further confirmed by the specification sheet which reveals up to 25 modules can be added to a given circuit with this particular optimizer. DC optimizers work by increasing the panel voltage and decreasing the panel amperage. This allows longer circuits than stand-alone string-inverters or micro-inverters to be installed, as more power can be pushed through the same size conductor. As an installer, I like DC optimizer installs because of their long circuit lengths, which can mean simpler wire runs back from the roof to the ground.
Now that we know the number of circuits the array will have, the next design step is to identify where the start and stopping part of each circuit. This information is useful in determining where the home run circuits will land on the array up on the roof. Special care should be taken during installation to make sure these circuits are plugged up correctly. Having these locations planned in advance reduces mistakes in the field.
There are two approaches to circuit layout.
I think it best to go for a simple, logical circuit layout even if it means using more circuits than necessary, if those circuits can start and stop at logical places. For example, starting at one side of the array perimeter and ending on the opposite side, moving across the array in a straight line.
The alternate approach is “snakes in a basket” where the circuits randomly start and stop as a winding circuit path is drawn throughout the array. So the first circuit starts here and ends here, and the second circuit starts here and it ends here, resulting in a compact design where the start and stopping points of the circuits are randomly distributed throughout the array layout. But if the circuit map is not available, servicing such an array can be hellacious. Clear, logical circuit layouts keep it simple enough for a solar installer to figure out through visual inspection.
Interconnection options are available either as a “load side” or “supply side” connection. The supply side connection is located between the customer meter and the main service panel. The supply side connection can equal but not exceed the rating of the electrical service. That means if the site has 200 amp service, then a 200 amp energy system can be connected to it. After all, the conductors feeding the building are rated for 200A. Regardless of whether the power to the building comes from the solar array or the utility, it is throttled by the 200A main breaker at the electric service panel.
Alternately, a load-side connection can be selected. The load-side amperage allowance is less than the supply-side allowance, and so is generally selected for systems which will power less than 100% of a building’s energy.
As a starting point, NEC dictates a 200A electric service cannot be fed with more than 200A of power, whether it comes from the utility service through the main breaker or from the solar array through a load-side breaker, or any combination thereof. In other words, if a 200A service panel has a 200A main breaker, then it has zero amps available for a solar load-side connection.
Obviously this is not the case for solar, and NEC allows an exception when the solar breaker is located at the opposite end of the busbar from the main breaker, rather than inserted somewhere in the middle.
So if the main breaker is at the top of the panel, then the load-side solar breaker is located at the bottom of the panel, an additional 20% amperage allowance can supply the panel. In other words, a 200A service panel can be fed with 240A of power divided between the utility main breaker and the solar array. This could be 200A of utility power and 40A of solar or 180A of utility power and 60A of solar or any other combination.
This code provision is necessary, because it is common to have more than 200A of load breakers on a service panel. Count the breakers on any electric service panel and its likely to total 300A or more. An assumption is being made that not all the loads are used at the same time. If 300A of load were pulled through a 200A panel, the panel would heat up and start a fire. The 200A main breaker is there to prevent that from happening. But now solar is also feeding the panel, from the other side of the busbar, and so the panel can potentially draw more than 200A of power, which would be a fire hazard.
Code allows the extra 20% capacity when solar feeds the very bottom of the busbar, not because it’s being nice, but because when solar is landed at the very bottom of the busbar, the actual cross-sectional area of the busbar will still not exceed 200A, because the electricity will not combine with the main power supply before flowing into the load. If the solar array were landed anywhere else other than the bottom of the service panel, potentially it could combine with the utility load and exceed the busbar rating.
So the takeaway is either the solar array will land between the utility meter and the main breaker as a supply-side connection, such as for a larger solar array, or multiple solar arrays alongside battery inverters. For this reason, I recommend installing a 200A disconnect switch, as well as a dedicated electric service panel, which will adequately prepare the site for future expansion such as the ability to combine solar with batteries to run the home off-grid during a power outage. But if simply offsetting the electric bill is the goal, easier project logistics can result from simply landing the solar array or battery inverter or both at the bottom of the busbar. A supply-side connection requires a power shutdown from the utility. A load-side connection does not.
But even for small interconnections, supply-side connections can be made. Sometimes the insides of the building electrical system can be a mess. A supply side connection, made outside the building where the electricity leaves the meter and enters the building, can keep the system inspection outside the building.
Here is a small commercial electrical room. How are we going to interconnect into this system? The room is surprisingly code-compliant, other than the fact that this space in front of the electrical equipment should not be used for storage. But the room is messy. The last thing I want to do is bring the inspector inside this room.
Back on site, we look at the outside of the building and see where the electric cables leave the meter and enter the building. The supply-side connection will intercept and tap into these conductors at this point. Currently there is an electrical box called an LB where the cables enter the building, feeding the two separate 200A panels coming out of the 400A meter base. Ahead of the array installation, the electrician had the utility power down the building. The building service entrance conductors were then pulled out of the service panel, out through the LB, and back to the meter connection. The LB then swapped with a larger junction box and the conductors are routed back through. This process was scheduled when the building was not in use and took less than an hour.
So when the solar array was ready to install, all that was needed to be done was to open up the junction box and tap onto the conductors, a process that took only a few minutes. Minimal business activity was interrupted as a result of the supply-side connection.
Tapped conductors still need to be protected by overcurrent protection, so installation of a service panel with breakers is necessary, or something more simple like a fused disconnect switch. The conductors are tapped using tap connectors in the junction box.
Here is the cable going into the service panel on this commercial project example. Piercing insulated tap connectors were used instead of standard tap connectors, because they install faster. Standard tap connectors require cutting the conductors, stripping back the insulating jacket, bending the conductor wire, and other small tasks which require time and hand strength. The piercing insulated tap connectors simply crank down onto the conductor. But if the building can sit without power for a couple hours, most electricians will prefer to use regular tap connectors where possible. Regular tap connectors are less costly and are easier to reconfigure at a later date than their piercing-insulated tap connector counterparts.
- Racking Design
For any given roof surface, there is almost always a racking attachment made specifically for that kind of roof. The standard way to attach to a shingle roof is to use a flashed L-foot mounting bracket, with a lag screw stand-off that is integrated into flashing which slides underneath the shingle.
It is not a perfect process. In new construction, it is much better to plan the project and mount the flashed attachments as the roof is being shingled, rather than after the fact. Otherwise you have to get up underneath the shingle with a pry bar and pry out roofing nails.
There are above the shingle mounting systems, as well as tile replacements, clips for standing seam or corrugated metal roofing, or even sealed standoffs designed for penetrating through metal roofs.
Some racking has better integrated cable management than others, but the whole system is pretty much the same. A rail is attached to the L-foot, and then the solar panels are clamped down to the rail. Mid-clips that space and clamp down the solar panel onto the rail between two panels, and end clips that go on the very end. Additional components help with grounding as well as splicing rail together.
Pro installers will stagger the L-foot attachments to hit every rather going across a roof, with consideration given to maximum rail span between attachments, as well as specific wind and load considerations that might require using less than the maximum span.
Like inverter string sizing software, racking manufacturers provide an online design tools to assist in racking bill of material development and sometimes structural load evaluation. In this design software example, the array layout is modeled and some building details inputted.
In another example, the number of rows and columns of array subsections are inputted into the sizing software separately.
The racking design software will ask about the environmental conditions: what’s the local wind speed? what’s the local snow load? what’s the exposure category? Are there objects around that are gonna break up the wind? These questions will determine the spacing between roof attachments as well as the gauge of the rail.
At the end, with the array layout and building information inputted, the result is a racking material list ready for procurement. In this example, three different kinds of solar rail are recommended with various strengths. For these three rails, maximum spans are given for different roofing zones. Zone 3 is the corner of the roof, with the greatest wind speed and the least amount of span. The interior of the roof is zone 1, the roof cap and eaves are zone 2.
Is the longest span the most desirable? The longer the span, the fewer attachments needed to be installed. It is logical to think the fewer penetrations made in the roof, the better. But the fewer penetrations made, the greater the force on the attachment. It is important to evenly distribute the load over the roof truss, meaning the penetrations should hit each rafter as the array moves across the roof. So a solar array might not get much use out of the thickest rail gauges which could allow for longer spans. A strong and cost-effective may be to go with a cheaper rail and conservative attachment spacing.
Moving along in the software, the array layout, location, and attachment spacings are defined. The numbers are crunched to provide the downward, uplift, and tangential forces on each attachment. In this example, there is a down force of 170 lbs and an uplift force of 100 lbs. The corners of the roof have forces three times that amount. This reinforces what we already know, which is that the corners of roof and the corners of the array will experience stronger wind load than the interior of the array. Moving the array in a few feet from the interior will reduce the load on the array, but even so the corners of the array will experience greater wind load than the interior.
Finally, the rail is selected and the racking material list is produced. The report provides how many sticks of rail are needed, as well as the associated clamping hardware such as mid clamps, flashing, lag screws, rail-to-rail splices, grounding straps, and other odds and ends. Combined with our inverter list and the solar panels themselves, the speciality material is ready to order from the distributor.
In fact, taking the time to develop an array layout and material list based on the equipment a distributor carries opens many doors. By signing up for an online distributors email lists, the installer-level pricing and product available can be determined. The solar supply chain is very open. Even non-installers, if presenting themselves, can access installer-level pricing from many online distributors.
- Aesthetic Considerations
I tend to work in lower-priced energy markets and find myself having to strike a balance between price and cost-effectiveness. I don’t care about much about solar panel efficiency or brand name manufacturers. I am fine with generic solar panels and more picky about inverter quality and racking with integrated cable management. But the one solar panel upgrade that I always recommend to residential customers is “all-black” solar panels, which have a black frame instead of a silver frame, and a black plastic backsheet instead of a white backsheet. Lower end all-black panels will still have grid-lines on the front of the solar panel, although the lines fade with distance. Top shelf all-black panels can look like a rectangular piece of pitch black glass, but they cost a pretty penny too. So I tend to go for more mid-range all-black panels to produce a good aesthetic, because as they say in fashion, black never goes out of style.
So in my design aesthetic, I am going for all-black modules, internal conduit runs, and symmetrical array layouts that fill up most of the roof surface where the budget and design allows.
The only error that this installer made in this picture is that there’s a plumbing vent right here on the middle of the array. The design obviously called for a continuous row of solar panels, but the designer missed the painted roof vent that is sticking up from the roof, and so the installer simply stuck the solar panel out at the end of the array instead. Maybe the client says, “Well, it’s on the back of the house so nobody will see it.” But rooftops are visible the further away one gets. In fact, I commonyl will ask clients to walk away from their home and then turn around and snap a photograph, if I am designing an array offsite, simply to reveal additional roof detail.
It is possible to simply replumb the vent on top of the roof. Roofing vents are required to be a certain height off the roof to assist with gas dissipation. An installers might want to give the vent stack a little haircut with a bandsaw, which may be electric code compliant but could violate other building code. To maintain the plumbing vent height, it is a simple matter of rerouting the pipe under the array and out the top, accomplished with two 90° bends of plumbing pipe.
I’m very interested in is how to take a rooftop and build the roof out of solar panels rather than to use solar shingles for that reason. But solar shingles are not readily available, and the ones that do exist are not widely considered to be cost-effective, despite the marketing hype. My opinion is that the solar shingles that are on the market do not look better than having an all-black array with a layout that makes aesthetics a priority.
To achieve a “hover-like” effect with the solar array, with all the cable and racking tucked underneath the array rather than sticking out the edges, the racking cantilever span from the racking design software is important. The last L-foot attachment on the roof is placed within the array perimeter, as well as within the cantilever span of the rack,, such that the last bit of solar rail will be cantilevered over the L-foot.
The last solar panel on the array lands on the rail, and then the rail will be cut to length with a cordless bandsaw. This effectively hides the racking underneath the array perimeter.
With experience, the rail can be pre-cut down on the ground.
Most racking manufacturers have a plastic cap to cover the nub of cut rail, hiding any rough edges. Some top shelf racking systems will go a step farther, hiding the end clips of the array underneath the module frame, although it complicates to achieve the very best high end look.
The other part to making the rooftop array aesthetically pleasing, which is important for resale value, is to use internal cable runs through the attic. Internal cable runs are no more complicated than external cable runs. They make the array look better and keep the conduit and cable off of the hot roof. The less wire for birds and squirrels to chew on, the better (and there are accessories for critter management). But basically, keeping all of your cable and racking underneath the array allows it to achieve a very pleasant, almost magic hovering effect on the roof.
Here is the end result, using internal cable runs to avoid having conduit on the outside of the rooftop. Twenty years from now, the modules will not appear terribly obsolete so long as they are still generating electricity.
- Planning the Conduit Run
In the northern hemisphere, it is common for solar to be on the south side of the roof. But exterior electronics, such as an outdoor solar inverter, should be kept out of direct exposure to sun, and so are commonly mounted on the north, east, or west side of the building, typically along side the existing electric meter. My preferred method to achieve an internal cable run involves a high material cost, but because it takes a direct route through the attic, it really is no more expensive than routing the conduit along the exterior of the building. Attics are uncomfortable to work in, but so is a hot roof. Speed is the issue here. The question becomes how do you actually land the the cables coming off the array in order to go into the attic?
High-quality solar installers are comfortable drilling a hole through the roof. The transition is made at the last solar panel in an accessible corner of the array, such that it can be identified later via visual inspection. But code allows the transition box to be tucked underneath a solar panel, to protect it from weather as well as improve the aesthetic look of the installation. This is a specialized solar transition box called a Soladeck, which is small enough for the job has integrated flashing to get up underneath the shingles. The cables come out from the attic to both land on this terminal block, meeting up with the solar cables from the roof which enter through a cable gland. Of course, this is a specialized box which costs about $100 just for the empty shell.
There are inexpensive, code-compliant ways to make the transition into the roof in a workmanlike fashion. For example, you could get a flashed pipe boot can be found at the local hardware store. Electrical conduit could then be stubbed up through the pipe boot for the cable to transition between the roof to the attic. One caveat on trying to make the rooftop transition work elegantly with generic, off-the-shelf components is that there is only about 4” of clearance between the roof deck and the top of solar panel itself. Often junction boxes found locally are 6” deep. So if the goal is to hide the transition box underneath the array, by the time one considers the height of the pipe boot and the height of the box, it is easy to be in conflict with the array itself.
When using generic off-the-shelf components, I will commonly skip the box on the roof, simply by transitioning the cables through the conduit via a cable gland and then landing a box in the attic, accessible and just underneath the array. In short, you can achieve a quality installation with off-the-shelf generic parts, but this too requires knowledge and planning rather than last minute scrambling.
DC conductors when inside the building, are required by code to be protected by metal conduit. It’s confusing as metal conduit is commonly associated with a ground path, but in this case it is not for grounding, but instead for physical protection. A rodent is less likely to chew up a wire if contained in metal conduit. A nail or screw is less likely to puncture a power cable if the cable is contained in metal conduit. Some installers will select micro-inverters to avoid this requirement, allowing the home run to be run in AC-rated romex. Running metal conduit as a retrofit through an attic can be a difficult task.
I prefer DC optimizer systems with long circuits, and I will spend more money on balance of system material if it improves installation quality or speed. To enclose my interior DC home run cables in metal from the roof to the inverter on the side of the building, I use a bundled cable product called MC cable, which stands for metal clad cable. The conductors are already bundled together, in a metal wrapping that encloses them fully. This is expensive stuff – one DC circuit of MC cable will contain two full-sized cables plus a ground, and costs just under $3/foot. But costs are kept in check with DC optimizers allowing fewer circuits than other kinds of inverter systems. My favorite MC cable has four full-sized cables plus an undersized ground at about $3.50/ft, which would give me two circuits total with two positives and two negatives plus a ground. It is expensive, but can be quickly routed through an attic while meeting the DC metal requirements, making the code-compliant installation go very quickly. The MC cable can be landed on the Soladeck box and then run through and out the soffit on the underside of the roof eave, on the north-side of the building where the inverter is landed.
MC cable is only rated for damp rather than wet conditions – which makes sense as the metal wrapping isn’t nearly as weather-resistant as a complete metal tube such as EMT conduit. Yet the outside of a building is considered a wet condition, unless the outdoor area is sheltered, such as a covered parking area, an awning, or a porch.
So the MC cable run transition to the inverter can be accomplished in two ways. It’s easiest to pull the MC Cable through out from the attic through the soffit, and then land on a junction box. I will then strip off the MC cable wrapping and transition the cables into EMT conduit through the box before landing on the inverter. Alternately, the transition can be made inside the attic if the permit office opposes any MC cable outside the building due to its damp, rather than wet, environmental rating.
For this homerun, most installers will size the cable to be #8 or #10 AWG. I usually go for #6AWG MC cable, because the MC Cable comes with an undersized ground. The minimum ground wire to ground the solar rail up on the roof is #8, so I select a #6 MC cable to take advantage of the #8 ground wire included in the cable bundle. Otherwise I would have to run the ground separately. The ground will land on the rooftop transition box, before landing on a ground lug on the solar rail, completing the grounding run from the inverter on the side of the building up to the solar array on the roof. The solar inverter is then tied into the building ground.
Therefore, the easiest code-compliant way to bring two solar circuits from the rooftop down to the inverter is to use “#6/4 plus undersized ground” MC cable or #6/2+g where there’s only one solar circuit. The two conductor pair will be used for the positive and negative end of the solar circuit. In other words, a “#6/4 + g” conductor MC cable will cover two solar circuits and a “#6/2 + g” will manage one solar circuit (or serve as an in-attic jumper between subarrays on the roof), in addition to providing the ground cable. MC cable is expensive but it makes the array look real nice and installs quickly.
- Battery Sizing
Batteries are a fascinating aspect of the solar industry, and could easily overshadow in the solar industry in the years to come. But the real applications of batteries are a little bit different than what you might initially expect.
Many homeowners expect batteries to be an economic solution to resolve the mismatch between solar production, energy use, and grid policy. If a utility does not provide a good net-metering policy, storing the energy onsite to use later can lower the electric bill. Even without solar, if the utility has a time-of-day variable electric rate structure, it might be appealing to charge the battery during off-peak times and use the battery during peak times, to lower the electric bill.
But batteries do not have infinite life, and there is a real cost to installing and running the battery. Let’s assume the batteries of the Tesla PowerWall 2 comprise half of its $6,600 material cost, resulting in a $250/kwh price. The Tesla Powerwall 2 is rated for 37,800 kwh of energy output, leading to a $0.09/kwh hard cost which does not include installation, sales tax, design fees, or any other project expenses aside from the hard cost of the batteries themselves.
Utilities who are keenly aware of battery lifecycle costs can simply adjust their policies to render residential battery ownership as a luxury rather than cost-effective expense. A time-of-day rate difference of $0.09/kwh would result in the customer simply trading one bill for another, or even losing money if all project expenses were considered. An economic rate differential would have to be twice that amount, closer to $0.20/kwh, in order for the battery to be run in a cost-effective manner. Only a few customers will be able to find an advantageous time-of-use rate structure for their home. But others will value the utility of having some backup electrical storage capacity on site. But the point is that it is very hard to make batteries economic under most residential electric rates.
Lead acid is cheaper, but generally speaking, lead acid is only cost-effective when storing multiple days worth of electricity – an expensive proposition for any customer. It is an expense off-grid customers pay for because they have to, not because they want to, in order to avoid the even greater cost of grid expansion.
Sizing a battery for off-grid operation is a simple matter. Take twelve months of electric bills, and size the solar array to not run out of power at the end of any month, paying special attention to summer and winter. This will be larger than a 100% net-metered solar array.
Then take the monthly data and divide by days in each month to get an average power consumption per day.
Remember PVWatts from earlier? PVWatts data can be exported into a spreadsheet, revealing solar production figures that are not only monthly, but also for each hour of the day.
PVWatts models typical weather for a given month, and so will reveal how many cloudy days in a row a customer should expect. Beware that PVWatts will not model an unusual hurricane or blizzard, but that is what a backup gas generator is for.
In my off-grid models, I start by guessing the battery size and seeing how it impacts the model. Then “recharge” the battery capacity with hourly solar production figures from PVwatts and subtract out the hourly load consumption from the monthly electric bills. I then model a few different battery and array sizes, as well as generator capacity and run time, to ensure the battery bank does not run out of power at any point in the year. For off-grid planning, I think its best to design the system not to need a gas generator, such that the generator is only there for backup. That said, I work in the southern half of the USA. In the extreme north, it is easier to assume a few days of generator use in the winter.
In any event, it is useful to plot solar production, building consumption, generator use, and battery capacity on a graph throughout the year.
At some point, the solar array will be large enough to where increasing it further won’t yield any more advantage. While solar will produce power on overcast days, it largely depends on how thick the clouds are.
The last modification to the spreadsheet is to add in generator run time to avoid the very few days the solar battery cannot generate enough power on its own.
If off-grid living is not desired, a well-sized residential battery might be sized to eliminate solar outflow onto the grid. As a rule of thumb, batteries that store ~2/3rds of a solar array’s daily summer production will be large enough to stop an array from outflowing onto the grid. And again, for anything smaller than multiple days of storage, I’d recommend lithium ion, even if cost forces the customer into second-life batteries from the electric vehicle market rather than buying new. Without getting into the details, lead acid batteries cannot be discharged as quickly as lithium ion without significant efficiency losses, so a lead acid battery which stores anything less than 2-3 days worth of power will not lend good results for the user.
Also be aware that most homes, particularly all-electric homes, will use more instantaneous power than what a single residential battery inverter will output. But battery inverters are expensive. Solar inverters are “one way” in that they take all the solar production and push it out onto either the load or the electric grid. Whereas a battery inverter is “two ways” in that it both charges and discharges the battery. This makes battery inverters cost about twice as much as their batteryless counterparts. Which makes the project all the more expensive if whole-house backup power is required.
Today, common solar battery inverters will only backup critical load panels, rather than output their power to the whole house. And because new residential lithium ion battery costs can exceed $300/kwh just for the battery itself, most solar customers are only purchasing enough storage to power a house for hours, rather than days. All that said, this is an emerging market that is rapidly evolving. Some companies are focusing on load controls, to manage a building’s electrical load during battery operation in order to better allow for whole house power during a grid outage, by making sure not all the appliances in a home turn on at the same time.
At any rate, storing 2/3rds of the daily summertime production of a solar array strikes a good balance of keeping most of the solar electricity onsite, while providing enough power for emergencies, whether it be a critical load or power for the entire house.
Commercial customers have a more optimistic picture for batteries.
Larger commercial customers are billed differently for electricity than residential customers. Maybe half of a typical commercial electric bill is the kwh energy rate, the area under this curve. The other half of a commercial bill is measured in kW demand charges, the maximum height of this curve sustained for 15 minutes over the entire month.
Let’s assume that half of a commercial electric bill results from the maximum 15 minute power draw of a building. The obvious solution is to use the battery during that period of maximum demand only. If the building does not experience its peak demand during the day, then the solar array component might not even be needed. And battery that only offsets 5% of a building’s energy use might reduce its electric by 25% or more, a 1:5 ratio. Compare that against a batteryless, net-metered residential solar array which produces 100% of a building’s energy and offsets the bill by 100%, a 1:1 ratio, to understand that commercial batteries, with or without solar, have the potential for better economics than residential solar, with or without batteries.
If a commercial solar battery were designed to eliminate 100% of the electric bill, it would only save at that 1:1 ratio. But ironically, if the project were smaller and only targeted peak demand, the project becomes more cost-effective.
This requires rather substantial modeling, based on building interval data that can be obtained either directly from the utility or through logging into the building’s electrical account and downloading it directly there.
I then model my “solar battery peaker” plant in a spreadsheet similar to my off-grid designs. But on a real project, I’ll use commercial design software, such as EnergyToolBase, to hone in on the optimal configuration.
The building load profiles is the literal shape of the building’s power use throughout the day. The load profile gets spikier towards the top, and levels out as the building transitions from peak load to base load. The most cost-effective commercial solar battery only targets the spikiest of the peak loads.
Almost every commercial facility on a demand charge can install a cost-effective battery for demand management, with or without solar.
- Energy Automation
Today’s smart home television commercials reveal how little attention is being paid to energy management. Energy management can generate an economic payback faster than solar or batteries on its own, while providing useful information for site planning.
Automated load control improves the efficiency of off-grid living in multiple ways. But any well-sized off-grid system generates an overabundance of energy during the day.
To be worth anything, this energy must either be sold back to the grid, stored in a battery, consumed by an appliance, or it is otherwise wasted. Load control can take advantage of stranded electricity on site, the cheapest electricity available, and put it to use.
For example, any deep freeze can be turned into a battery that reduces grid outflow, by only turning on during times of solar excess. Weather data can then be fed into a controller to recognize cloudy days, telling the freezer to then operate only during off-peak times as a fallback measure.
If the freezer lacks enough thermal mass inside to stay cold, the problem is easily solved with purchasing two freezers and filling each halfway up with gallons of water. The total load of the building might increase through this strategy, but the electricity being used to power the freezers can be turned on at times when electricity is the cheapest.
Solar monitoring is improving. For some time, the use of DC optimizers or micro-inverters gives the solar owner the ability to see the production of every single solar panel on the roof. But even with this overabundance of data, with a solar array on the roof, the homeowner can be left with no idea of how the energy from their solar array is actually being used.
The other piece to this puzzle is monitoring the electrical consumption of the building itself. Knowing the electrical consumption data of the building to an exact degree will only improve any design process of adding any solar arrays or batteries to the building. Net-metering allows solar projects to be built somewhat blindly, assuming that the production of the solar array and consumption of the building will even out at the end of the process. But even having monthly electric bills does not reveal the minute-by-minute level of detail of exactly how energy use of a building fluxuates.
Consumption monitoring reveals how much electricity the building is using at any given time, often creating an energy log for later reference. Many companies sell products which install current transducers inside the electric service panel to relay consumption data to a local or cloud network. Regardless, it becomes clear that residential electric loads are very spiky, with sporadic peak times where loads all turn on at once, combined with off peak times when there is not much electric load at all.
This fluctuation can be problematic. For example, if the utility buyback rate is substantially lower than what the customer is charged for electricity, the end result can be 2/3rds of a residential solar array’s production is discounted 80% when bought back from the utility. But storing that energy in a battery isn’t necessarily more cost-effective, because of the cost to cycle the battery. Batteries have other value to customers, such as back-up power, but rarely make a grid-tied residential solar array more economic.
But many power outages last minutes, not days. Furthermore, during the power outage, if a building has a solar array on top, it can keep the battery charged during the day, so the home is only at risk of losing power at night.
But again, there is the problem that a Tesla powerwall by itself only outputs 20 amps, but even an energy efficient air conditioning unit can momentarily draw 60 amps due to momentarily internal heating strips. A residential service panel, after all, is rated for 200 amps. So even if the Tesla powerwall were wired to power an entire house, only 10% of the service panel’s maximum capacity can be used at any given time, and despite any extra surge capacity of the battery unit, the whole house would lose power the minute a 60 amp load turned on for more than a minute While the entire house might only average 10 amps of instantaneous load over the course of a year, the minute-to-minute loads can surge if all the home loads turn on at the same time. This forces the owner to buy numerous Tesla Powerwalls if the goal is to backup the entire house.
Imagine only having the budget for a single Tesla Powerwall, such that only a critical load panel can be backed up during a blackout. This forces the customer to choose which electrical loads should be saved during the power outage, and air conditioning is not eligible for the list.But I grew up in Houston Texas. I would give up my lights, internet, and the food in the fridge before giving up my air conditioning in Houston’s 10 month long summer.
Load management is useful in an off-grid setting. Sunny winter days will still produce excess electricity. Electric heat is generally frowned upon in off-grid design, but during times of energy surplus, it can reduce load on a heat pump, wood burning stove, or gas heater. An increasing number of both solar and battery inverters have relay signals which can trigger a relay breaker to turn loads on and off, as a function of solar production or battery state of charge.
All of the home food freezing and hot water heating can be load controlled to only occur during times when the building’s electricity is least costly. The freezer method was discussed above, but as another example, a hot water tank could be programmed to only turn on when the building energy load is low. This forces the hot water tank to only operate during off-peak times. Alternatively, the hot water tank could be programmed to only turn on when solar production is above one kilowatt. This forces the solar hot water tank to only turn on when the solar array is on to help reduce outflow issues.
Plus battery inverters are expensive. Load control reduces the amount of inverter capacity required to power multiuple devices at the same time. So a home may be fully backed up with two Tesla Powerwalls instead of four (although the duration of the storage for two powerwalls would be half that of four powerwalls). A fully off-grid house might be run with as 15kW battery inverter system instead of a 30kW inverter system.
Load controls can increase the longevity and efficiency of batteries. While Lithium ion batteries do perform better than lead-acid batteries, but they aren’t perfect. The faster they discharge, the less efficient they get.
But even without solar, or without batteries, even in a non-emergency setting, energy load controls can result in substantial cost savings.
Any hot water tank can be put on an intelligent control circuit to take advantage of a utility time-of-use rate structure. This could be triggered via the inverter relay to an automatic switch. This kind of setup is similar to how a whole house automatic transfer switch is triggered to start a generator and switch between grid and backup power. Except it applies to a single circuit,
Circuit-level mechanical relays are great to manage a couple heavy load circuits, can only go so far. But there’s no intelligence for optimizing them for anything more than basic controls which come pre-programmed from a controller. Such options are not always flexible to reprogram for a particular electric rate structure or grid policy,
Digital controls would provide greater customization with less required wiring.
With digital controls, preferably automated, non-critical but high use energy devices can be put to use. Digital controls can allow energy devices different modes of operation, based on site considerations. The same controls which throttle a building electric use when operating off a battery can also take advantage of time-of-use rate structures during normal operation. In this sense, digital controls allow the home or business owner to optimize their electric use for how they are being billed for their electricity.
Commercial customers have the greatest opportunities for savings through load control. Once commercial building energy exceeds 200 Amps – the maximum energy use of a standard electric service panel, they begin to incur demand charges. Demand charges are calculated at the highest 15 minute period of peak electricity use for the entire month. Even with a modest demand charge of $10 per kilowatt, during that one 15 minute period, electricity is costing the building as much vas $40 per kilowatt-hour. Automatically reducing a commercial facility’s energy use during this time period, by turning of refrigerators, hot water tanks, and other non-critical devices for a short period of time can generate substantial cost savings, with 1-2 year simple paybacks without any subsidy.
Consider a modest hotel chain, the kind of hotels where each room has an air-conditioner built into the wall. Everyone checks into the hotel in a rush, get to their rooms, and crank that AC or heat, resulting in a huge spike in electric use. Staggering the air conditioner run times would result in the same amount of electricity being used as before, but the peak electrical demand of the hotel is reduced.
Turning the air conditioners on and off from the front desk, as a function of check-in and check-out status, would result in further energy savings. Telling all the air conditioning to turn off for 15 minutes in all but a select few rooms would result in a cash windfall for the hotelier. An optimal system can save the hotel money without noticeable difference in comfort to the guest.
How would one implement a system like that?
You could install a large box of electrical relays, which would install alongside the electric service panel. Every circuit on the panel would be relayed controlled, and then programmed by an external controller. Such systems cost 3-4x as much as a traditional electric service panel.
Even so, circuit level controls only go so far. Device-level control can provide even greater control, for similar cost. A controller which can manage device level controls can potentially communicate with any smart home device in the house – including light switches, 120V wall plugs, and even heavy 240V loads.
So far as the controller goes, I use Home Assistant, one of the ten most popular open-source software projects in the world.
There are other home automation platforms, which usually cost more, have limited functionality, but can be much easier to use right out of the box. Home assistant requires some programming knowledge, and it helps to have a dedicated computer to run it. There is an entire community dedicated to running Home Assistant on a $35 computer called Raspberry Pi.
I use Home Assistant because other building automation controllers do not seem to prioritize energy optimization features. the commercial smart home controller I’ve found do not make energy controls a priority, and so I’ve had to learn how to program within Home Assistant to enable the control methods discussed above. So I thought I would introduce what I’ve learned about smart homes in the context of building energy management.
First is that energy load controls can communicate wirelessly without connecting to the building internet. This reduces wiring and internet security concerns. It is a best practice to put wifi-enabled devices on their own wifi network, a feature higher end internet routers can manage through software controls rather than literally installing a second wifi network.
But some smart home devices are more like installing a second physical wifi network, but instead of using a wifi antennae for communication, it uses its own antennae on an alternate frequency. This keeps the smart home device communication physically isolated from the building’s internet. The home assistant control hub will benefit from an internet connection, but it is not needed for Home Assistant to operate.
In my case, I used Zwave devices, which use a frequency similar to old school wireless telephones back before the cell phone era. But other platforms include Zigbee, 433 Hz RF frequencies, high frequency infrared, as well as 2.5 GHz and 5 GHz wifi. Without getting too far into which frequencies are best for which kind of communication, I’d say that absent a hard-wired connection, wifi devices works the best, but the consumption meter I use is Zwave so I went with that.
A consumption meter is installed in the electric service panel. A 40 amp wireless control switch has been installed on an electric tank water heater (or for the hotel mentioned earlier, each in-room air conditioning unit). Plug-in wall plugs can be installed freezers, refrigerators, or UPS power systems. Behind-the-switch control boxes can control ceiling fans.
In a similar manner to using an Amazon echo to ask Alexa to turn an Amazon light switch on or off , home assistant is programmed to give a wide range of “internet of things” devices a common vocabulary to enable communication between them. Home Assistant is open source, and can be accessed and controlled over local internet provided the user has the encrypted username and password.
One the devices are registered with Home Assistant, which is a tedious but not too difficult process, the automation menu allows the user to select which devices to automate based on data that the devices are collecting, as well as additional programming considerations.
Here I have energy devices set up to turn on and off as a function of the home energy consumption meter. They could easily be programmed to turn on and off as a function of time of day, or excessive solar production, or a combination of all of these options. So not only is the home monitoring its electrical consumption, but it is also using that data to improve the economics of the building’s electrical system, without requiring solar or batteries.
Because digital energy controls can be put to use economically, at a lower price point than solar or batteries, with a quicker payback than solar or batteries, with a broader audience of uses as well, they should probably be done first, before planning a broader energy overhaul of a facility.
Home Assistant has a popular chat room on Discord full of voluntary programmers helping you get started. If you are interested in learning more about home energy automation I suggest joining that group. https://discord.gg/c5DvZ4e
With that we are out of time today. I hope you gain some insight into solar design and what encouraged you to continue your solar education. You can find more in-depth material at my YouTube channel, www.youtube.com/c/communitysolar