TariffAnalysis 2018Finals Game102

In Game 102, the competitors were AgentUDE, Bunnie, COLDPower18, and EWIIS3. COLDPower did not issue any tariffs, but did make some money trading in the wholesale market. Bunnie won by a large margin, although AgentUDE was far ahead in the first hours. I used the new log analyzer to understand what happened. The report shows several things, some of which appear to be the result of misunderstanding how brokers should interact with the server.

1. EWI's initial tariffs all have the same fixed price of 0.18/kWh, and all of them charge a fixed 0.95/day to all subscribers regardless of energy use. It does not offer production tariffs. Then in timeslot 364 (the simulation starts at timeslot 360) it issues a new set with much higher prices of 0.3/kWh. It also re-issues a new set of identical tariffs every few timeslots, paying the publication fee for each one, and without revoking the earlier ones. It gets a few subscriptions, but the only one that appears to earn a profit is the CONSUMPTION tariff from the first set. See the Game Totals at the end of the file for the summaries.
2. AgentUDE offers a Time-Of-Use CONSUMPTION tariff at the start, with prices varying by hour. It gets some subscriptions early on, but then loses them. This appears to be because it issues a new TOU CONSUMPTION tariff in timeslot 402 and then revokes its original CONSUMPTION tariff in timeslot 403. At this point, virtually all the customers switch over to Bunnie's simpler TOU CONSUMPTION tariff and stay there.
3. AgentUDE also offers a STORAGE tariff and gets a couple of subscriptions. It submits BalancingOrders against this tariff, which results in some regulation, and at the end of the game it has earned more than its initial publication cost.
4. Bunnie offers a suite of simple tariffs at the start of the game, and then leaves them alone. Its CONSUMPTION tariff accounts for most of its total earnings. Its STORAGE tariffs don't attract many subscribers, and it fails to include regulation rates or issue balancing orders against this tariff. Its SOLAR_PRODUCTION and WIND_PRODUCTION tariffs attract a good fraction of the customer population. It's hard to see the profitability of these tariffs without seeing what they earned by selling this energy in the wholesale market.
5. No agents offered regulation rates, and only one offered balancing orders. That only applied to the battery customers, although there is considerable balancing capacity available in the electric vehicle customers and in the various thermal storage customers. The only agent to offer an interruptible tariff was EWIIS3, and it failed to take advantage of it.
6. Other than the default broker, only AgentUDE offered PRODUCTION tariffs, and it only attracted three subscriptions. The default broker bought essentially all the solar production from the small solar producers, but AgentUDE appears to have attracted the three larger ones.

We can see a bit more about where brokers made and lost money in this game using the BrokerAccounting tool, which produces a csv file. The columns, for each broker, represent per-timeslot debits and credits arising from static and usage-related tariff transactions (ttx), market transactions (mtx), balancing transactions (btx), distribution transactions (dtx), capacity transactions (ctx), balancing control events (bce, resulting from balancing orders and from tariffs having regulation rates), and bank interest.

We can observe several interesting bits of broker behavior with this view:

1. COLDBroker18 made a modest profit of 28,118 from buying and selling in the wholesale market, but it appears that its real strategy is to build a market position that puts it on the profitable side of the overall per-hour imbalance between production (including wholesale purchases) and consumption (including wholesale sales). It earned 109,680 on balancing transactions.
2. EWIIS3 spent 484,743 on tariff publication fees, far more than the 293,390 it earned by selling energy. It also lost 194,406 in the balancing market because it did not buy any energy in the wholesale market, and so all the energy used by its customers came at far higher prices through the balancing market. It also spent 108,797 on capacity transactions. So slightly more than a third of what it earned selling energy was spent on balancing because it was not building market positions in each timeslot to match its customer demand.
3. AgentUDE started off with a good customer base, but started losing out to Bunnie around timeslot 404. Its "cost of goods sold" -- net wholesale market costs -- were less than a quarter of its net earnings in the tariff market. It made nearly a third as much money in the balancing market as in the tariff market, probably because of the massive imbalance caused by EWIIS3 buying its energy in the balancing market. It actually earned a bit of money on capacity transactions because it had some production customers who made its net usage negative during a good bit of the game. It had a few battery customers and earned a bit on balancing control events, but this was a minor part of its portfolio.
4. Bunnie benefited when AgentUDE jacked up its rates and canceled its original popular tariff. It earned 11,961,251 in the tariff market, and its cost ratio (tariff earnings / wholesale energy cost_ is around 17%. Its market position seems to have been consistently short because it paid more for energy in the balancing market than it did in the wholesale market. Nearly half its tariff market income was spent on capacity costs, since for most of the game it was almost exclusively responsible for peak demand.