Demand Response Systems to help the power grid

I recently read that California is stepping up (or planning to increase) their billing rates for electrical usage during peak hours. There's a fascinating graph (from a study done by UC Davis) showing the amount of power consumed by residential and commercial customers during a typical day. The sum of the two graphs actually looks pretty sinusoidal, with peak demand occurring between 4pm and 7pm. The peak of the commercial/industrial graph is a little sooner, the residential one is a bit later.

In many parts of the country, consumers receive a credit on their utility bill if they install a remote-controlled shutoff device to their air conditioner. Given that the A/C unit on most homes is the largest electrical load in use (particularly during the daytime) it is a good candidate for saving electricity and reducing peak demand on the electrical grid.

There's a downside to this method of demand reduction (not just the fact that your house will get a bit warmer for a while) as it defers, rather than eliminates, the electrical demand. Once the utility allows the A/C unit to turn back on, it will then operate for a longer time to bring the temperature of the house back down. In the end, you only shift the demand, you don't eliminate it.

A better (but more complex) solution
A better method would be to raise the temperature setpoint a few degrees for several hours, assuming that the outdoor temperature will begin dropping, reducing the cooling load later. If discomfort of someone in the residence is an issue, it seems that reversing the logic might actually serve the purpose better: a few hours in advance of peak demand, commanding a longer cooling cycle (to pre-cool the house a few degrees) would reduce the demand that will occur later in the day. Unfortunately, this doesn't prevent the electrical usage, it again shifts it. However, shifting it earlier, rather than later, may be a better solution.

Another study by UC Davis (and supported by others) indicates that decreasing the level of lighting by 15% is imperceptible to people in a building. While that is probably true, I don't think the answer is as simple as dimming the lights. I haven't made measurements to be sure, but I suspect that decreasing the level of a dimmable fluorescent light by 15% results in a power savings less than 15% (probably 5 to 10) so the lighting efficacy is reduced. (Efficacy is the efficiency of a light source in terms of light produced per watt of power consumed - usually specified in lumens/Watt.) Even if a reduced light level is perceived, dimming the lights in a building is certainly preferable to a rolling blackout.

Dimming incandescent bulbs to save energy has another side effect (in most cases). Incandescent dimming is usually accomplished by chopping off part of the electrical power waveform. In other words, the power is turned off for a fraction of every power line cycle. This phase-cut method of dimming places a nasty load (along with lots of electrical harmonics) on the power grid. Dimming a large percentage of a building load in this manner would probably cause power quality problems.

Load Shedding on-demand
Demand response - an action taken by a customer of an electric utility in response to a request by the utility during a period of high demand - is likely to become more important to protect the integrity of the power grid in certain areas. A demand response that eliminates, rather than defers, power consumption is more useful than simply shutting off A/C units for a while. Some critical technologies to make this possible: building-wide lighting and HVAC control systems, dimmable fluorescent ballasts, per-fixture control when dimmable ballasts are not an option (allowing a subset of light fixtures to be turned off), and communication methods to get the demand message from the utility to the fixtures.

I've been experimenting with a dimmable (0-10 V controlled) fluorescent ballast, and I hope to have some efficiency vs. dim level information in the future. I've noticed a small shift in color temperature as the bulbs are dimmed, but nothing even close to the color shift that occurs when dimming incandescents.