Why has Flexitricity never missed a triad?

Another year, another rogue triad. Against most predictions, 25th November 2015 won third prize in the annual winter peak competition which has become central to the security of the national electricity system. The margin of victory was tiny – just 37kW, or 0.00008% of peak demand. Flexitricity called it (we got the first and second as well, for the record).

This is the triad system, a mechanism which applies a relatively minor industry charge (for the upkeep of the transmission network) to three half-hour periods in which National Grid most wants businesses to avoid using electricity. These periods are the triads.

The triad system is exceptionally cunning, for three reasons. First, by concentrating the annual transmission charge onto 90 minutes, there is a much larger incentive to reduce demand than if the charge were spread out across the year. In fact, “triad management” is now one of the most valuable demand response activities on the market; it still beats even the top end of Short-Term Operating Reserve (STOR) by a comfortable margin.

Second, National Grid doesn’t tell you in advance when the triads will be. This is because it doesn’t have a crystal ball. Business energy users have to reduce demand 20-30 times each winter to have a good chance of hitting all three. And third, there’s the shadow. Triads must be separated by ten clear days, which means that a two-week cold snap will almost never produce more than one triad.

So the triad system is set up as a game. If you win, you don’t have to pick up a share of that year’s transmission network costs. But demand will always peak somewhere, so the network will be paid for. Active triad management reduces peak flows, which makes the network cheaper in the long run, and that means National Grid wins progressively, year-on-year. Through the wizardry of Ofgem’s price controls, the customer wins too, through lower overall system cost. Eventually.

Triads occur between November and February. Apart from the shadow, and some details of what’s counted in peak demand, there are very few other rules. Nowhere does it say that triads have to fall on weekdays around 5pm. That’s just what tends to happen. It doesn’t always – in the winter of 1972/73, a triad fell on a Sunday.

Predicting triads

Most suppliers and many energy brokers attempt to predict triads to help customers avoid transmission charges. This year, at least eight established triad predictors missed the 25th of November, including some which we generally rate highly. Some “inner circle” warnings (yes, those do exist) also missed it.

Does this mean that the triad system is now unmanageable? Is it now just a matter of luck? Or are there strategies which will protect triad value for business customers and small generators?

It’s sometimes said that balancing services can be used as a hedge against missing a triad. The argument is that National Grid is more likely to call for STOR during peaks, so if a site is offered for STOR, it could get the triads as a by-product of STOR calls. There is some value in this approach, and in fact that is how all but one of our customers successfully hit the infamous Friday triad in 2014. (For the record, the one customer not called was deployed in another service, which earned him more profit than the triad would have provided.)

However, there is a difference between peak demand and peak stress. If National Grid forecasts demand accurately, and power stations stand up and deliver, there’s no need for a STOR call even in a peak. And the Friday triad had some unique features. In particular, demand was exceptionally above forecast. That made our decision to stick with STOR on that day the right one. In more normal circumstances, hedging with STOR is not a reliable tactic. It’s also not permitted to drop out of STOR at the last minute to manage triads – National Grid is on record as having terminated STOR contracts for doing so. Worse still are frequency response services, which we have never seen used significantly in a triad.

A pattern we see regularly in triad behaviour is response to weather conditions. There is a lot more sense in this approach. Cold, windless days often produce triads, while mild, wet, windy days tend not to. But caution must be exercised here as well. It’s not just today’s weather that matters – it’s tomorrow’s, and the rest of the season’s. One of the best predictors of a triad in November is a mild December. That’s not much use as a real-life prediction.

There’s a sense in which everyone is constantly managing last year’s triads. One very successful triad manager once claimed that he didn’t bother if demand was below 55GW. Nowadays, demand at that level would break the system. It’s also now clear that triad warnings reduce demand, and cause triads to pop up on the less likely days. That’s what happened on 25th November. But we also see consumers or generators leaving their triad decision to the last minute, watching demand closely. Of course, this is part of Flexitricity’s strategy too. If triad warnings drive the triad, they only do so through customer behaviour, which can also respond to real-time information. “Events, dear boy, events”, as Harold Macmillan put it.

Good triad management requires response to information that comes in just minutes away from the peak. But that information must also be understood. National Grid’s demand forecasts and rolling measurements are the triad manager’s favourite resource. But the purpose of these data sources is to enable Grid to balance the system, not to predict triad demand. In fact, out-turn triad demand is calculated by a different team using different meters and subtly different definitions. Live system demand data aren’t wrong; they’re just measuring something a bit different.

Some companies deal with this uncertainty by grading their triad warnings as high/medium/low, or red/amber/green. At Flexitricity, we are perplexed by this. We’re even less convinced by suggestions that a lower level of warning requires a lower level of response. It’s either a triad or it isn’t. While some businesses do indeed suffer disruption if they maximise their triad management, and could perhaps deliver a portion of it with substantially less pain, the high/medium/low strategy only makes sense for the customer if high warnings correlate better with triads than medium or low warnings. We’re not at all convinced that that is the case. With apologies to Yoda, we say, “do or do not; there is no amber”.

An unbeatable triad record

Flexitricity’s triad management record is the best in the business: all available capacity in our portfolio has hit every triad since we started in 2008. We don’t know of any other company which can make that claim. But change is unceasing, so the Flexitricity boffins have recently been working on new tools for triad prediction, with some interesting results. It is surprising how close we came to seeing a triad in the first week of November. That would have been another first for the industry.

Detailed analysis, to-the-minute decisions and active management of balancing services (with strict observance of the rules of those), can provide high confidence of a successful triad season with a low level of wasted runs. However, the potential for surprises is designed into the triad mechanism, so there is one tactic which every triad manager must deploy: play the long game. Finance directors need to account for triad risk over the long term using realism and pragmatism. Missing one triad in three is bad. Missing one in fifteen is less bad.

One day, Flexitricity will miss a triad. “Not on my watch,” is the response from every member of the Flexitricity team. But whether they’re right or wrong, we’ll still lead the market in triad success; we’ll still be pulling down bills by making the whole energy system more efficient; and our customers will still be earning revenue by helping to keep the lights on.