Peak-Demand charges in Power TAC

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Peak-Demand charges in Power TAC

grampajohn
Administrator
Dear colleagues -

One of the important features of real power systems is completely missing from Power TAC: the cost of peak demand, which is basically a portion of the amortized cost of infrastructure allocated to individual customers. Most large users of electricity pay for both energy (kWH) and peak demand (max kW over a billing period). In most areas, peak demand is measured monthly, and billed at a flat rate determined by the tariff. It might make sense to add such a feature to Power TAC tariffs, but only if brokers have an incentive to manage peak demand.

Where would such an incentive come from? I am a member of a rural electric power cooperative in Wisconsin. In our part of the world, demand peaks are almost always in the summer, due to irrigation and air-conditioning loads. Our cooperative, St. Croix Electric, is a distribution entity; it does not own any production capacity, but it does own substations and distribution-level infrastructure. It buys all its energy from a "meta-cooperative" called Dairyland made up of a number of distribution cooperatives. Dairyland does own production capacity, and trades at the wholesale level through the Midwest Independent System Operator (MISO).  Member cooperatives like St. Croix Electric pay for energy at basically the wholesale rate, and they pay Dairyland for the shared infrastructure through demand charges. Basically, Dairyland computes the amortized cost of all its infrastructure and divides it up among its members according to how much of the demand each was responsible for during the n highest overall peak-demand events in the previous year. I believe n = 5. In general, the monthly energy charges and monthly demand charges paid by St. Croix Electric are fairly comparable - the demand charges are a little higher than energy charges in the winter and about the same in the summer. A member cooperative can make itself better off (at the expense of the other members) by doing a better job of managing its own peak demand. It turns out that solar panels typically generate around 50% of their peak capacity during demand peaks, and so owners of solar are paid (in summer months only) a "demand reduction" rate for their surplus production, beyond what is used by the household. For me, that was almost $60 last month.

In Power TAC, the Distribution Utility charges for use of infrastructure as a fixed amount per kWh transmitted over the wires in a given timeslot. But of course the costs borne by the DU are a function of peak kW, not hourly kWh. Ideally, the DU charge would be based on peak demand (for the infrastructure, which handles both power delivered to customers and power sourced by customers). How could this work? We don't want to penalize brokers for acquiring new customers, and we don't have a year of data on which to base overall peak demand. We cannot allocate demand peaks to customers and then charge brokers based on which customers are subscribed, because the point is to give brokers an incentive to use their tariffs to modify the behaviors of their customers. That is made especially difficult for the large-population customers because different subsets will behave differently under different tariffs.

So I am looking for ideas about how we might switch the DU charges from usage-based to demand-based, keeping in mind that populations of subscribed customers can change several times in a single day. It might make sense, for example, to set a daily revenue target for the DU, and compute demand charges based on demand peaks observed during the bootstrap period. We might then allocate demand charges once/day, perhaps at midnight, and ask brokers to pay their respective shares of the daily revenue based on the actual demand peak for the previous 24 hours. I can see potential problems with this scheme, so I'm looking for more ideas and a discussion on how this might work. Please think it over and share your thoughts.

Thanks!

John
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Peak-Demand charges in Power TAC

grampajohn
Administrator
Dear colleagues -

I've been thinking a bit more about this issue, and I needed to see what the overall demand patterns looked like in the 2015 tournament. Here's a contour plot over all the games in the tournament:
2015 demand contours
The contours are [2%, 25%, 50%, 75%, 98%]. You can see the effect of solar production -- sometimes on weekends the net demand (amount supplied through the wholesale market) goes slightly negative.

Data for this plot comes from the aggregation of PRODUCE and CONSUME tariff transactions, which account for both normal consumption and production, and for broker contributions to balancing.

The general idea I'm working on is that the distribution charges would be based partly on transported energy (that's 100% of the charge now) and partly on broker contributions to peak demand as seen from the wholesale market. For the latter, we might use a sliding window of perhaps 2 weeks and charge for broker contributions to the highest peak(s) over that period. Details are still TBD, but I would be very interested in your thoughts.

Thanks.

John
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Peak-Demand charges in Power TAC

grampajohn
Administrator
In reply to this post by grampajohn
Dear colleagues -

I've been thinking further about how we might assess capacity charges, and I'm working up a page on the github wiki to explore the issue. We want to finalize an approach in the next couple of weeks so we can update the specification for the 2016 competition. Our goal is to have the final specification out by the end of December.

My current thinking is that we divide each game into "assessment periods" of perhaps 7 days, find the largest peak(s) during each period, weight them according to the amount by which they exceed a threshold, compute a total charge, and assess brokers according to their (customers') individual contributions to the peaks. This will take some additional bookkeeping in the server along with a new transaction type, but it should not be difficult to implement. The goal is to motivate brokers to manage peak demand. Obvious ways this might be done include time-of-use tariffs and use of curtailments. Note that brokers already have access to total demand and local production through the DistributionReport message.

Please take a look at the page, and share your thoughts over the next week or two.

Thanks much.

John
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Peak-Demand charges in Power TAC

grampajohn
Administrator
Dear colleagues -

I've done quite a bit of work to refine our ideas of how to motivate brokers to try to shape the demand profile. It's written up in the server wiki. See the bottom of the page for an experiment that might give you a better idea of how it would work. As you can see, the code is implemented and working, but it's not yet on the master branch.

The goal is to challenge broker developers to deal with some of the most important issues facing a sustainable energy market: reducing the need for fossil-based balancing resources, and shaping customer demand to better match the availability of renewable resources. So far, there has been no incentive to shape demand, and the cost of balancing energy has been only marginally higher than the wholesale cost. In the example shown in the wiki, the cost of balancing is 80% higher than the wholesale cost.

As always, please share your questions and suggestions.

John
Loading...