Towards Net Zero

Watch us as we transform our 1920's Urbana home.

How much do we produce?

The installation process.

Our photovoltaic array, completed.

Our array was completed and we pulled the lever to turn everything on right around the winter solstice, Dec 2014. I wasn't able to keep up with blogging about the whole installation (Assistant Professors, alas, are not known for having copious amounts of free time), but I do want to show some pictures describing the whole process. Especially since this coming Saturday, October 3 2015, we will be participating in the 2015 Illinois Solar Tour. You can come visit my house, and other solar installations in the area, for free between 10am-2pm.

Our array is probably one of the more complex installations you are going to come across. We were ready for and expecting that. The roof is what is known as a "hip roof", meaning that it is has steep walls shaped like a pyramid. There are four faces, each facing north, east, south, and west. To add to the complexity, you can see the "dormer" on the roof as well, which is where the attic partially sticks out from the top. Ideally, for a solar array you want a large, flat south facing roof to mount the panels so that they can point towards the sun. Only one of our hips points towards south, the terrain is complicated by the dormer, and on top of that there is a beautiful, gigantic poplar tree in front of the south facing hip.

Even though I knew our array installation would be more complicated than most, at this point, I seriously considered whether having a solar array made sense for our location. It wasn't clear how much usable area we had on the roof, and it wasn't clear how much the tree would block the light. I got lots of helpful advice, especially from Scott Willenbrock, who put me in touh with John Hammons, who runs the company Oy Not Solar. The process turned out to be very easy. I emailed John a picture of my roof (thank you Google Earth). A few days later John emailed me back a picture of my roof, now with a solar array on it.

In fact, it was better than that. He was even able to make some videos that estimate the shading losses I would have because of the tree, on the south facing panels. You can see the videos here.

The first one shows the shading effects of the tree in the summer time when the sun is high in the sky, and the second one shows the same thing in the winter, when the sun is lower on the horizon. With these movies, it was possible to estimate our monthly production accounting for the effects of shading. I'm not sure what software John used to produce this, but it turns out that there are several options in case you wish to do the same for your location. One option is called Solmetric.

From the videos, it was clear that the south part of the roof would be shaded in the winter time by the big tree. On the other hand in the summer, the sun would be high enough in the sky that the tree wasn't a problem. I was encouraged, and kept moving forward. In fact, the winter time movies were a worst case scenario, since the tree will lose its leaves in the winter anyway.

Apart from the tree, the second issue we faced is that there simply isn't much south facing roof. I threw out the crazy idea: why not just cover everything we could on the east, south, and west facing hips? There was a time, in the not so distant past, when this would not have been possible. But thanks to a recent development, known as a microinverter, this approach is now feasible. Solar panels produce DC current but to connect to the grid the currnet needs to be converted to AC. That's what the inverter does, and it used to be that all the current produced from the entire array would be sent to a large inverter to be converted. Uniform illumination of the entire array was important for this (some of the modules are connected in series and there can be "weakest link" problems). Instead, our array uses microinverters. Each module has its own microinverter clipped into the backside, and can more or less operate independently of the others. So putting the panels to cover east, south, and west was doable. Here is a picture of the microinverter, mounted to the railing on the roof. The module will be placed on top of it.

One question that I had was how much would I take a hit by mounting the panels eastward/westward, rather than towards south. It turns out that, for my location, not much. In fact for the modules that we have placed on the roof, the biggest predictor of how much they produce is how high up they are on the roof (rather than which direction they face).

How much electricity will we need to generate?

An important question that I wanted to answer in order to plan the design of my solar array was simply "how much electricity do we need to produce?". The amount of electricity we need to produce determines how large the solar array should be. To answer this question, I dug up our electricity usage records from our utility company. Here's what is shows, for the twelve month period from April 2013 until March 2014.

The y-axis shows how much electricity we used, measured in "kWh", or "kilowatt hours", which is a unit of energy. We used the most during the summer, largely from running the air conditioner. On the other hand, we use gas rather than electricity to heat in the winter, and our winter electricity use is much less (but our winter gas usage is higher). One of the nice things about a solar array is that it produces the most in the summer, when the demand is the highest. In the summer, the daylight hours are long and the sun is high in the sky. Peak production matches peak demand naturally.

Summing all of our usage for the whole year, we used 7145 kWh of energy for electricity. It's not always easy to get a good sense of just how much energy this is. I converted it to an average energy "burn rate" to see how much energy per unit time we use on average. That is, 7145 kWh/year is around 20 kWh/day, which is 20 kWh/(24 hours) or something like 0.8 kW. That is our house's overall energy burn rate, which means that we are on average burning eight 100 Watt lightbulbs at any given time. Of course, sometimes we're using more (when the air conditioner is running), and sometimes less (at night when the lights are out), but that's the average. For the sake of comparison, the average US household yearly consumption was 10,800 kWh in 2012 according to the US Energy Information Administration although it varies a lot from state to state depending on climate.

In order to meet our current needs, the solar array would need to provide 7200 kWh of electricity each year, or a time averaged production of 0.8 kW. The next question I wanted to answer is "how much real estate is required to produce that much electricity from a solar array?". We need two pieces of information to answer this question. The first is the local energy input from the sun. How much sunlight energy does my location get on average? If I lived in Nevada where the sun shines all the time, the size of the array I need to meet my needs would be smaller than if I lived in a cloudy location. Then the second piece of information is the overall system efficiency. How efficiently will the solar system as a whole convert that incoming sunlight energy into electrical energy output.

The average "INcoming SOLar radIATION", or "insolation", accounts for your local climate: is it normally sunny or cloudy? How long are the days in the summer and how short are they in the winter? You can quickly estimate the local insolation at your location using NREL's solar maps. Choose the map for "photovoltaics" and estimate the insolation at your location. For Urbana, IL we'll estimate the insolation to be 5 kWh/m^2/day. That means that every square meter of land receives 5 kWh of sunlight energy each day.

That's the input energy, and then output energy is (output) = (input)*(conversion efficiency). I'm going to assume an overall solar array conversion efficiency of 12%. If you want to know where I got that number, I just estimated. An average performance silicon solar module is probably around 15%-17% efficient these days (they get up to around 23% for the high end modules, but I'll go with the average). Then when you factor in other system-scale losses that might arise from the inverter, from shading due to trees, etc, then we'll go with 12%. We could do better if we really tried -- using the best modules available and working hard to minimize other losses. But I live in a tree-filled neighborhood, and using a lower estimate of 12% conversion efficiency seems reasonable. That means that every square meter of solar collector will produce (5 kWh/day)*(0.12) = 0.60 kWh/day. Since I need 20 kWh/day, I will need around 33 m^2 of total collector area.


A few months ago, we finally purchased our home. It's one of those old Urbana "state street" homes that was built back in the 1920's. I like it for many reasons - old bookshelves built into the walls, the lively and warm reds and purples that Cope and Walter painted the inside walls with some time ago, and all the big old trees in the neighborhood. Not only that, its close to campus which means that I can bicycle or walk to work year round. Since we bought the house, nothing about my life has changed, practically speaking (apart from now having a mortgage, that is). We'd already been living there for a few years anyway, renting the home that we eventually bought.

The house is typical of the west urbana area. It was built with no insulation in the walls or the attic, single pane windows. By now most of the windows are replaced and the house has been insulated. At the time it was built, there was little consideration for the environmental costs of burning coal and little understanding of climate change. So, the experiment that I'm conducting -- and will document here -- is to see how close we can get to being "net zero". We'll be installing solar panels on the roof to generate our electricity, and we'll be putting in a ground source heat pump for heating and cooling. (Don't take "net zero" too literally by the way. Being "net zero" would be nice, but is by no means necessary.) This is an experiment to see how close we can get (i) using an old house that wasn't designed with a solar array or geothermal in mind, (ii) in a neighborhood full of huge old trees (beautiful, but difficult for the solar array), (iii) in what is one of the more challenging climates out there (cold winters, hot summers). Since first moving here a few years ago, I've been keeping track of our energy footprint - both electricity and gas. So let's see. (You might notice that there are people on the roof - more on that later.)

I'll be documenting the process here for many reasons. The first is that I get a lot of questions about how to go about this from many people: how does the solar array work? what if you don't generate enough from the sun? how much does it cost? I hope to answer a lot of those questions here. The second reason is to show that a solar array and a geothermal system are becoming easier and easier to install. A lot of my research involves understanding and improving the performance of materials for solar energy conversion. At the scale of atoms and electrons, I could tell you a lot more than you probably want to know about how photovoltaics work. You might think that means I could install a solar array on the roof in my sleep. Quite the contrary. I am manifestally less "handy" around the house than average, and practical home installation is quite different from the research that I do. The point is to show that nowadays the technology has established itself so much that yes even I can do it.

Many thanks need to go out. The first of which belong to Scott Willenbrock, who motivated me to go through with this. See his awesome work on the Colonial Solar house; I'll be referring to these pages a lot myself. Another big thanks to Ty Newell, whose Equinox House is another great local example.