« From the comments: "America the Beautiful"? | Main | Hendrik Houthakker dies at 83 »
Demand Response
A large share of the special green issue of the NYTimes Magazine was closely tied to economics. I find this encouraging. Here is one interesting bit:
...demand response has become one of the most powerful green techniques for protecting the nation’s overtaxed power grids. When a blackout looms, utilities call a small coterie of demand-response firms. These firms prearrange for major users of electricity — factories, shopping malls, skyscrapers — to shut down all nonessential electricity in exchange for payments, often totaling tens of thousands of dollars each year. It’s expensive, but far less so than a blackout that cascades across the country....ConsumerPowerline controls 300 huge buildings in New York alone, where hastily brokered turnoffs by Macy’s and major hotels prevented the spread of a 2006 blackout in Queens — a blackout that lasted for more than a week — into Manhattan.
Posted by Alex Tabarrok on April 23, 2008 at 07:05 AM in Economics | Permalink
Comments
Extremely interesting. I have to
say.
I think it is akin to the large
electricity users selling a call
option on the aggregatae power
usage in the same grid. Considering
that this option sounds like a
digital (all or nothing) option, I
wonder how the demand response firm
prices it,i.e assume the excess load
on the grid is 100 units. But the
demand response firm has bought 300
units which are split up like 75, 50,
25x6, 10x2,5. The question is how
does the firm price these units?
differently or same per unit.
Posted by: sa at Apr 23, 2008 8:15:42 AM
I just got back from San Francisco. The power company there will pay you to allow it to install a switch that will allow it to raise the temperature of your A/C a few degrees under certain circumstances.
Posted by: joe at Apr 23, 2008 8:51:00 AM
Any economic theory that is cyclic will predict "{supply response" to ultimately counter balance. The two will simple balance out.
The test is whether the balance will have a longer phase delay with or without the protocol.
The test, of course, is $5, the per capita cost of a digital readout that gives the user his exact (to the penny) immediate electricity bill.
Posted by: Matt at Apr 23, 2008 8:57:41 AM
Yes, but have a look at www.rltec.com
Now that's really really clever...............
Posted by: energetic at Apr 23, 2008 9:16:29 AM
This can work in limited circumstances but in general it is a bad idea.
The principal problem is that you are paying people not to consume power and there is no way to measure the amount of power they didn't consume. If you are paying me $100 dollars not to consume 1 megawatt hour I'll quite happily not consume 2 megawatt hours for $200. How do you know what I would have consumed if you didn't pay me not to consume? In general you can't. The circumstances where it does work is where I'm running some sort of plant whose consumption can be accurately known. The intervals when reduced consumption are required are reasonably predictable, see the live wholesale market data from NSW, Australia at http://www.nemmco.com.au/data/GRAPH_30NSW1.htm. Generally on very hot days in summer about 2.00pm when air conditioners everywhere are going flat out.
The economic incentive can be huge. In NSW the normal wholesale price is 3-4 cents per killowatt hour and it goes to as high as $10.00 per killowatt hour. Consumption is priced in half hour blocks. If you try and predict my usage from historical data and I have some flexibility in my consumption pattern e.g air conditioning or water heating, I can increase consumption during the interval you are using to estimate my consumption and then get paid for not consuming when I wouldn't have consumed anyway. It is even profitable to burn off electricity uselessly for a few hours at my retail rate of 10 cents per killowatt hour so that I can reduce consumption when you are willing to pay me say $8.00 per hour not to consume.
I worked for a while on developing systems for demand response and it became apparent it could never work because the optimum economic outcome for the consumer required behaviour that was destructive for the grid. This is because you can't measure power not consumed. It can only be estimated and the consumer can adjust their behaviour to increase the estimate. You could even make a business out of setting up a big heater in a paddock and occasionally getting paid more to turn it off than it cost to run.
Posted by: Ken Taylor at Apr 23, 2008 9:28:13 AM
My college did this during the California blackouts a few years ago. It was a giant pain in the ass. They would send an email telling you they'd been requested to shut off power, and then within 30-60 minutes it was off. (Often before one received the email, what with class and all.) Several people's computers were fried from the repeated power surges until we all bought redundant surge protection.
Posted by: Amber at Apr 23, 2008 9:42:45 AM
Chris Galvin (former CEO of Motorola) has an initiative dealing with these power system issues.
http://www.galvinpower.org/
Posted by: Paul F. Dietz at Apr 23, 2008 12:04:16 PM
An electric utility's generation and transmission capacity must be built to support the maximum expected load (the middle of the day in the summer), which means that it is significantly under-used most of the time. The goal of demand response is to essentially flatten out the peaks by encouraging users to time-shift their use of power. The examples above of making money running a heater or shutting off power to a dorm seem to me like problems with the design of specific demand-response schemes, rather than problems with the concept itself. As an engineer in the automated metering industry, where all of the industry players are working on demand response products for utilities, my impression is that while this concept has been around for a while, we're just at the beginning of figuring out how to design these schemes to have the intended effect. I think to make it work, the price of electricity must go high enough to motivate users to change their habits, and the infrastructure needs to provide real-time pricing information and simple ways for users to respond to it.
Posted by: Quentin at Apr 23, 2008 12:23:49 PM
Because the demand-response groups are being charged, overall, less than others, are they being "subsidized"?
(Of course, I'm talking about discounted prices for agricultural water in California, not power.)
Posted by: Francis at Apr 23, 2008 1:23:06 PM
Yes, but have a look at www.rltec.com. Now that's really really clever.
It may well be, but it would be cleverer still if they could describe what it is somewhere in the first two web pages.
Posted by: Mitch at Apr 23, 2008 1:37:36 PM
Pricing electricity according to demand at the retail level can reduce demand and as Quentin says the industry is moving that way. It is not easy though. The metering infrastructure is expensive and consumers, and ideally appliances themselves need to know the current price and an estimate of the future price and then trade off value of whatever the appliance is doing against price. Often the optimum strategy is increase demand now to reduce it later when the price is high. E.g use the air conditioning to make the building cold now and let it warm up a bit later on. Probably the biggest problem though is psychology. There is a lot of consumer resistance to variable pricing. The idea that my power can cost 100 times as much this hour as it did a few hours ago appalls many. Being paid not to consume is psychologically appealing.
Demand response is usually defined as paying consumers to switch off and that is what Tyler is describing above. Power is sold at the retail level for the usual price of say, 10 cents per kilowatt hour and during a high price event the retailer is purchasing it from the wholesaler at $10.00 per kilowatt hour. At first glance it makes sense to pay consumers $6.00 for every kilowatt hour they don't consume as the supplier is $4.90 better off.
The flaw is that you can not measure power not consumed therefore there is no way to know who should be paid $6.00. It would be like a crime control scheme where people are paid not to commit murder. You get the payment by stating I would have committed murder but I will not if you pay me. A person who doesn't intend to murder can take the money safe in the knowledge that their fraud is undetectable.
Posted by: Ken Taylor at Apr 23, 2008 5:44:21 PM
Just to keep capital costs in perspective.
I can measure power usage with a $3 instrument. I can buy a meter for $8 off the store shelf, and it would have five times the functionality we need.
What is the cost of transmitting current price info. If prices change every hour and I have eight price levels, then I need 8 bits/hour. I can do this with a an analog tone if I add a two dollar AM tone detector in my device.
Current payoff for the device? About a week.
Posted by: Matt at Apr 24, 2008 10:06:04 PM
Well, I no doubt am talking to no one, but this device can even be made simpler.
We can measure voltage drop over time, for a buck. Price should vary monotonically with price, so just measuring the changes in voltage is enough to print out the relative price.
So, my device now costs of $2. It needs no communication, and you just plug it into a socket like a night lite (at about the same cost).
Posted by: Matt at Apr 25, 2008 12:57:47 AM
Folks you have this whole program wrong!!! I am writeing a small thesis to help all understand and will post it in the morning. For a quick understanding goto www.esgna.com and look at the XSITE portion of the site.
Mike
Posted by: Michael J. Verkuylen at Apr 25, 2008 11:35:58 PM
Even easier if you're a business getting paid to implement a demand response system. http://www.enernoc.com
Posted by: George at May 1, 2008 8:50:27 PM
Energy management is an important issue these days. Thanks!
Posted by: George at May 22, 2008 11:18:51 AM
Energy management is an important issue these days. Thanks!
Posted by: George at May 22, 2008 11:21:05 AM
Ken, you do not know what you are talking about. one uses meters to see how much energy was reduced during emergencies and each kW is worth about $100 in nyc so it is VERY lucrative + much cheaper than a blackout + you don't have to build additional peaker power plants for increasing energy usage.
Posted by: luke at Aug 6, 2008 3:28:10 PM
