Enough sunlight strikes unshaded U.S. rooftops to replace all the coal and some of the natural gas we use to make electricity. Backup via ground source heat pumps, and smart grid technology would allow this variable energy source to displace base-load coal with today’s technology. Whether this is the most cost effective way to displace coal is another question. Also rooftop solar is a silver BB rather than a silver bullet: Even after massive efficiency improvements we will need to get many times the power from non-rooftop sources than from rooftops.

According to a 2003 study by the Energy Foundation (PDF), solar PV that converts 15 percent of sunlight to electricity could produce 710,000 Megawatts on rooftops that will be available in 2050. Doug Wood thinks that with concentrating PV using advanced aerospace quality cells we could convert solar at 30 percent rather than 15 percent efficiency. Scaling back to rooftops available today (using 2003 numbers from the same study and extrapolating forward) we could produce around 1.05 billion megawatts today. We normally assume 22 percent capacity factor (PDF) for PV. So that would give us about 2.3 billion megawatt hours, or around 56 percent of today’s electrical production — more than coal provides.

Further, waste heat from this process could provide much of our heating and cooling needs as well. The EF study I cited suggests that about 65 percent of commercial roof space is unshaded compared to about 22 percent of residential roof space. Since some commercial scale chillers run on low to medium temp heat today, with enough storage solar CHP could provide close to 100 percent of commercial heating and cooling. But that much storage takes a lot of capital for a small incremental gain. So more realistically, we would put 16 to 24 hours of low temp Phase Change Material storage and use ground source heat pumps to provide the other 15 percent of low temp needs. As a side effect, the overnight storage would let us run those heat pumps when the electricity was cheapest — which will prove more important than it might appear at first glance.

The comparatively low mount of residential roof space available means you will only have 200 to 300 square feet of unshaded roof space available per home on average. I don’t know if this means some houses provide a lot of available solar space, and others with none or if this is distributed more or less evenly per home. Even in the latter case, you will have a certain number of unshaded south walls, and a certain amount of yard space that could be devoted to solar generation. To be conservative, let’s say that 40 percent of residential space heat and hot water could be provided as a side effect of concentrating PV electricity generation. Again, let’s add PCM storage and the other 60 percent with ground source heat pumps.

This is for existing buildings. New buildings could be designed to use 70 percent to 80 percent less energy through a combination of better insulation and sealing, passive solar, more efficient air exchange, and more efficient lighting and appliances. New buildings could also optimize the amount of solar oriented unshaded roof space.

Now, since we are talking concentrating PV, we would have almost no control of when it would be generated — mostly during the five peak hours of sunlight except when it was cloudy. However, we are assuming all space heating and cooling other than solar is switched to ground source heat pumps with PCM thermal storage. So when solar electricity was produced at a time it was not needed, it could run heat pumps to generate heat or cold in storage for climate control systems to draw on later.

Sixty percent of 2005 residential heating and cooling was about 5.7 quad. Fifteen percent of 2005 commercial heating and cooling was about .6 quad. Without climate control efficiency improvements a climate control storage would let a smart grid absorb around 85 percent of solar electricity. If insulation and other improvements reduced climate control in existing buildings about 40 percent, climate control needs could still absorb about half. If we reduce climate control demand further, industry could add PCM and heat pumps to absorb pretty much as much of this as needed. Industry consumes about 33 quads. About 70 percent of this is used by boilers, and about 35 percent of boiler energy is used to produce process heat below 700 degrees Fahrenheit. That is 8 quads, or more than the total solar energy that would be produced. Because lower temperature heat is cheaper to store than high temperature, the high-end of this would be a last resort; climate control would be the cheapest form of energy to store, followed by hot water at or below the boiling point, followed by hot water at not too much above the boiling point. Between residential, commercial, and industrial hot water I suspect we could place most of what space heating did not require without ever needing storage above the boiling point or below the freezing point of water.

What is the bottom line on coal displacement? It would save slightly more electricity than we currently produce via coal, plus around as much again in displace climate control, hot water and low temp water heat. A small amount of this could displace coal use directly. But the majority would displace natural gas currently used for electricity production, climate control, hot water, and process heat. That natural gas in turn could be used to replace coal for base load, as a first step towards phasing out all fossil fuels, with a bit left over to contribute to phasing out oil. In other words total non-solar electricity generation would go down, even with increased demand to run heat pumps, while natural gas currently used to heat buildings would be available for electricity generation. We could completely replace coal based electricity generation with natural gas, and have some natural gas left over.

A very productive short-term use for this extra natural gas would be to make diesels more efficient. Running diesels mostly on natural gas has been very disappointing in practice. Natural gas powered buses, trucks and so on have lost so much in efficiency that you end up gaining little if anything in vehicle miles per Btu. But something has proven better: Take a diesel engine, and inject natural gas equal to 5 percent or 10 percent of the diesel fuel used into the engine. A minor effect is direct fuel replacement. But a major effect is that this natural gas helps burn the diesel fuel much more completely; the oil is more completely consumed. This can boost total energy efficiency by 30 percent to 40 percent. From a greenhouse standpoint the results are even better. Black carbon from diesel is essentially unconsumed diesel. Burning diesel fuel more completely reduces black carbon by much a much higher percent than it reduces energy consumption. Natural gas injection is a comparatively inexpensive modification of a diesel engine — one we could make to trucks or better still to rail locomotives.

Now physically possible is not the same as feasible. Doug Wood makes a good case for concentrating PV being pretty cheap in actual cell and parts cost. But he does not estimate labor, which in conventional PV is around 2 dollars a watt, or framing material for attaching PV to a roof, which can often run a dollar a watt. Installing a ground source heat pump under favorable circumstances (meaning you have enough land to only need to bury it at a depth of five feet) in a typical U.S. home is around $15,000. Commercial ground source heat pumps can take advantage of certain economies of scale, but they also almost always have to be buried deeper.

Rooftops are not the only pervious surface, but they are the main ones that let us do co-generation. What about parking lots? Well the problem with parking lots is that since you don’t already have a roof, you need to build one. And since existing parking lights are installed on the assumption of no roof, you need to install new parking lights as well. A roof strong enough to hold concentrators and panels (meaning not a low end carport), plus reinstallation lighting would have to run $15-$20 per square foot before you put in mirrors and PV cells. But maybe I’m overlooking some way of doing parking lot power (and road power) cheaply.

Right now it looks to me like two things are true. Roof top power has a lot more commercial potential, potential to be done cheaply, than is apparent at first glance. But rooftop power will never come close to providing the majority of our needs.