The Case for Dumb, Part II

by Michael Cain

This is the second of two guest posts on the topic of smart electric grids in the United States. The first postdiscussed how the term applies to the transmission network connecting generators to substations. This post discusses the distribution network from the substation to the consumers and small-scale generators. The distribution part of the smart grid is, at least in my opinion, much more interesting and controversial than transmission. The gadget pictured here is the heart of the smart distribution network: a smart meter. Different kinds of smart meters are the enablers for all kinds of new things.

One of the simplest things that smart meters enable is time-of-day pricing. The goal of time-of-day pricing is to smooth out the daily pattern of electricity consumption. As a nation, we use much more electricity during the day than we do at night [1]. Because storing electricity on a large scale is an unsolved problem, the grid’s generating capacity has to be sized for the peak hours (peak summer hours in particular). Lowering the peak by shifting demand into hours that currently have less demand reduces the need for new generating capacity. Smart meters alone aren’t enough for time-of-day; states also need to approve tariffs that include that option. Side note: everywhere that smart meters are introduced, there seem to be a significant number of customers complaining that they are suddenly being billed for much higher electricity usage. My own theory on this experience is that, in light of historical regulatory penalties for overbilling, those customers’ old dumb meters have been reading low for the last few (and possibly many) years.

Another step up on the smartness scale is to connect the meters to a data communications network, preferably two-way. There are a number of different technologies being used or at least proposed for the data network: the City of Austin uses a proprietary mesh radio network; the cell phone companies would like to the utilities to use the cell data network; and there are proposals to use the wires in the electric grid to carry data. Once a network is in place, a wider range of capabilities are possible. Meters can be updated remotely to reflect changes in time-of-day tariffs. Polling the meters in an area make it possible to isolate almost the exact point where a fallen tree limb has broken a cable. More disturbing, perhaps, is that the meters can be used to monitor power usage by a customer on a near-continuous basis. There are some relatively harmless uses for such data: “We see that you run your clothes dryer at 6:30 PM on Mondays; if you would wait until 8:00, you would get a cheaper rate that would save you some money [2].” Civil libertarians suggest that there are more sinister uses for such data.

Many utilities are interested in pushing the data communication inside the dwelling to smart devices there. One of the primary goals is to allow intelligent load shedding when the network is heavily congested. If you or I temporarily raise our thermostat a few degrees, or increase the target temperature for our refrigerator for a few hours, it doesn’t make a lot of difference. If the utility can do that in a million households and commercial establishments when a power plant has been forced to reduce its output, by signaling to the thermostats and refrigerators, a considerable amount of demand can be curtailed. At some point, when everything is smart enough, you can turn a lot of decision making over to the software. If adding more megawatts of available power during the next hour is going to cost the utility a lot, it can notify all of the smart gear that the customer will be charged a higher rate if they increase their usage from its level at the current moment. Then the equipment can decide, (hopefully) based on parameters set by their owner, which things will run and which will not [3].

Along a different axis are the new devices that support distributed grid-connected power generation. The standard example is the house with the big array of solar panels on the roof. On a bright sunny afternoon while everyone in the household is at work or school, the panels produce more power than is needed in the house. The excess can be transferred to the grid, reducing the need for power from other sources. Feeding power into the grid this way is not a trivial problem. The home generator must be synchronized with the network (see the first post for a bit more discussion of that). The amount of power flowing into the grid must be measured. For safety purposes, if the main commercial source of power to the local distribution grid fails, the home generator must be automatically disconnected from the grid [4]. The simplest arrangement for such a connection is a version of the dumb meter than can run both forwards and backwards, depending on which way power is flowing (net metering). Smart meters enable the same kind of complex arrangements for generators that they do for consumers: feed-in tariffs at a rate different than the retail price, time-of-day pricing, etc.

So, why aren’t smart meters widely deployed?

• Cost. A smart meter costs on the order of a thousand dollars. Assume 150 million households in the US and half that many commercial customers and you’re looking at a price tag around $225 billion. Then add the infrastructure necessary to support the data communications network (essentially, fiber to the substation or beyond). It’s a potentially very large bill, and it has to come out of the ratepayers’ pocketbooks.

• Lack of standards. Or in some cases, an excess of standards. Especially when the intelligence is pushed into consumer devices. Such problems generally shake out eventually, but can be frustrating for some period of time.

• Privacy. As I mentioned above, a smart meter with the right software can infer a lotabout activities inside the household. IANAL, but I do wonder about how the courts might classify access to such surveillance data. Do you need a warrant? Can you only access data that is collected after the warrant is issued? Or is it a “pen register” deal and once the warrant is granted you can examine both new and historical data?

• Security. Every type of network I’ve ever worked with seems to have had some sort of global “everything shut down now” command. Sometimes by design, sometimes by accident (never attribute to malice that which can be adequately explained by buggy software). In at least one case, there was no corresponding “everything come back up now” command and it took hours to bring all of the equipment back on line, one device at a time. I have occasional unpleasant visions of some tech typing the wrong command and blacking out Cleveland for three days.

One thing that seems to be seldom discussed with respect to smart meters is the issue of equity. Many of the advantages that smart meters are supposed to deliver to consumers will not be universal. It is difficult for people who work evenings and nights to take advantage of cheap electricity under time-of-day tariffs. Poor people are less likely to be able to afford new smart appliances. Apartment dwellers often have little or no choice about their appliances. People who live in dense urban cores are less likely to ever be able to take advantage of grid-connected local generation. Like broadband data service generally, it will be more difficult to connect smart meters in rural areas to a data network.

[1] This pattern is at the heart of studies showing that powering large numbers of new electric cars won’t require the addition of new generators: they assume the cars will be charged at night when much of the current capacity is idle.

[2] Depending on the sophistication of the sensor and software doing the monitoring at the meter, the turn-on and turn-off spikes produced by various large appliances are quite different. Enough so that it is possible to recognize the refrigerator versus the dryer versus the air conditioner.

[3] A group of negotiating appliances is not as far-fetched as some people might think. Essentially all modern devices are built around a microprocessor that provides the control logic, and wifi chips are getting cheaper all the time. I don’t know if the typical consumer will be happy when his or her smartphone beeps with a message that the air conditioner and the dryer are bickering over who has priority.

[4] The safety issue involves the utility’s workers repairing damage. In some situations, they need the local grid to be reliably “dead” while they work on it.