Solar Electric Generation in Austin Texas

Is it cost effective to generate solar electricity in urban Austin Texas? This page estimates the cost of buying and installing a utility-grid connected photo-voltaic (solar electric) system. By connecting to the grid, you can save considerable expense by avoiding the need for batteries to store the generated electricity. By being grid connected, your electric meter simply runs more slowly, or even backwards, thus lowering your electric bill. Below, after we do our calculations, we conclude that its is still not quite cost-effective, even if you buy the cheapest components, and install it yourself. But its not out of reach: it is cheap enough that if you care about the environment today, you can afford to get this stuff. And it's reasonable to think that in ten or twenty years, it might be cost effective to put solar panels on all new homes, just as a matter of routine.

How much sun does Austin Texas get? Annually, on average, about 3.1 KWh/m^2/day of direct sunlight, and an additional 1.7 KWh/m^2/day of indirect light. (Both components can be used to generate electricity, but only the direct component is useful for solar heating. Austin gets about 1000 BTU per square foot per day direct sunlight. 1 KWh=3,413 BTU. 1 BTU will raise 1 pound of water one degree Fahrenheit). See Sustainable Building Sourcebook, Appendix C (mirror) also: Renewable Energy in Texas and Texas Solar (mirror).

Photovoltaic modules cost about $5 per watt. See Photovoltaic Modules Comparisons (mirror) and Photovoltaic Module Manufacturers. Utility tie-in power inverters will cost another dollar per watt; see SolarElectric's Product page. Mounting brackets, wiring, panel modifications, etc. might run another $4 per watt, if you do-it-yourself: net-net, expect an installed cost of about $10/watt for a 2000 watt system. (Total cost: $20,000).

What sort of power output can we expect? Assuming a polycrystalline PV system with 11% efficiency, then Austin's 5KWh/m^2/day insolation equals 550Wh/m^2/day of usable electricity generated. But most panels generate 125 Watts peak per square meter: thus, for Austin, we have 550/125=4.4 hours a day of electrical generating capacity.

(This number may sound weird to you, so lets review its meaning. There are 24 hours a day, so if we could run our system 24 hours a day, we could produce 550/24 = 23 watts per sq. meter round the clock. But the sun doesn't shine 24 hours a day. If we assume it shines only for 12 hours, then we'd get 550/12 = 46 watts during the daytime. But direct sunlight is stronger mid-day than evening or morning, so we might get nearly 100 watts midday, and a few dozen morning/evening. It averages out to the same thing. All that the above division accomplishes is to point out that the panel, running at half-power, for half-the day, is the same as running full-power for 4.4 hours, and then being turned off. The average works out to be the same. Why is it that only the average counts? Imagine pushing that electricity into batteries, and then doling it out evenly during 24 hours ... )

These actual figures depend on the actual efficiency of the panel selected (monocrystalline are more efficient, amorphous are less so) and the power density (I don't know how that varies for monocrystalline and amorphous). Assuming that peak power scales as efficiency, then we have a crude rule of thumb: the usable hours/day is roughly equal to insolation/day, when insolation is measured in KWh/m^2. In other words, we can expect a single 100W-peak panel to generate 100*4.4 = 440 Wh/day. Thus, our hypothetical 2000W-peak system might generate 8.8KWh/day.

How should we amortize such a system? Electricity from the City of Austin costs about 7.6 cents per KWh: the system would save us about 7.6*8.8=67 cents a day. To amortize the $20,000 cost, we'd have to run it for 30,000 days, or about 81 years. Ouch! (Your mileage may vary. Wholesale spot prices in California linger near 18 cents per KWh, and, after retail markups, this might make solar electric far more appealing.)

Suppose we wanted to amortize the system over 15 years, which seems not unreasonable: one expects the solar cells to 'fade' in the hot sun, and have other problems in weathering the elements. The math, then: $20,000 dollars / (8.8 KWh/day * 15 years * 365 days/year) = 41.5 cents per KWh. This compares to the Wall Street Journal's estimates of 22 to 40 cents per KWh for solar, and 6.9 cents per KWh national utility average. (WSJ, 17 Sept 2001 issue supplement).

To make it worthwhile for common deployment, we'd have to shrink the amortization schedule from 80 years to 15 years. To do this, the installed cost would have to drop to under $2.5 per peak-watt. Note also, I failed to take into account interest rates: instead of installing a PV system, one might take that money and invest it in stocks and bonds, and reasonably expect to quadruple it or better in 20 years. Taking interest rates into account, the installed cost needs to be even less to make the system cost-effective. So it looks like we're not there yet, although maybe in another decade or two, we can hope that PV prices should be down to competitive levels.

Other References

Solar Thermal and Photovoltaic Collector Manufacturing Activities provides detailed industry economic information for solar manufacturers for the years 1987-1997.

See the LCRA's A Handbook for Assessing the Market Potential of Off-Grid PV Power Systems in Utility Service Territories) (mirror of google's mirror)

California consumers can make use of a program that provides a rebate of about 50% of the installation costs.


July 2006
There's a world-wide solar panel shortage: prices have not changed at all, still at $5 per Watt. Meanwhile, electric rates have gone up: 9.5 cents per KWh (instead of 7.6 center per KWh in 2000): they're significantly higher, but not enough to change the above calculation by much.


September 2000
updated September 2001
Last updated July 2006 by Linas Vepstas (

Copyright (c) 2000, 2001, 2006 Linas Vepstas.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.1; with no Invariant Sections, with no Front-Cover Texts, and with no Back-Cover Texts. A copy of the license is included at the URL, the web page titled "GNU Free Documentation License".