In the final days of the 20th Century Ireland had a problem. It was under obligations with the EU to deregulate its Energy Market. Fines were about to be imposed unless the system was up and running by March 2001. PwC were brought in to help, both with the design and implementation of the market rules and the software system to support those rules and perform the settlement calculations.
My role on this project was that of a lead developer in a team of around 10 developers. In particular I focussed on the meat of the solution, the technical implementation of the market rules for how the energy market would be settled. This matter was complicated by the fact that the market rules were in a draft form, submitted to the marketplace for review and comment. Indeed, as part of analysing the proposed market rules I identified some technical loopholes which could enable the market to be “gamed”. Flagging these up to the working group responsible for defining the rules enabled these loopholes to be closed prior to the market being opened.
We designed and implemented the front half of the solution, to collect data from the generation companies regarding their proposed energy supply levels in advance of the back half of the system to actually settle the market. Indeed, the market went live before the ability to settle it had been developed and tested! Through what can only be described as a hurclean effort by a fantastic young team of consultants from PwC’s UK and Irish practices the settlement system was delivered, and the first settlement calculations were delivered on-time.
Whilst the details are starting to get murky (it was 20 years ago after all) - I recall that the market worked in 30 second increments. And for each time point, for each power generation unit (e.g. a power station, a wind farm etc), there were 4 key pieces of information:
- The proposed supply: how much energy that PGU wanted to commit to provide in that time period
- The proposed demand: how much energy Eirgrid wanted the PGU to supply
- The actual demand: How much energy was actually demanded at each time period.
- The actual supply: When all was said and done, how much energy did the PGU actually provide?
Every day the proposed supply would arrive from each provider and they would later be provided with the actual demand (i.e. what they were being asked to provide). They would be paid a certain amount for the energy they supplied. But they would also be paid (a smaller compensatory amount) for any excess they had asked to supply but hadn’t been asked to provide. Eventually we’d find out how much energy has actually been supplied. If the PGU had oversupplied against their demand they would be paid a small amount for that excess. If they had underdelivered they would be penalised etc. But there is a delicious twist. The proposed demand may not be technically achievable by the PGU. For example, you can’t asked a Coal-powered station to ramp up from 0MW to 1000MW in 30 seconds. And it wasn’t reasonable to penalise a PGU if it was being asked to deliver something that wasn’t possible. Therefore the settlement system had a built in power station simulator which, given the characteristics of the power generation technology, would determine what was actually achievable given the actual demand. The PGUs actual supply would be compared to the achievable supply and not the actual demand for the purposes of penalties.
So there was an additional data point - The achievable supply: Given the actual demand, and the characteristics of the particular generation technology, what was technically possible for the PGU to supply
I was responsible for the implementation of the power station simulator and also the reporting solution which consolidated the four data points per PGU per time period. It was always said that you could tell when favourite TV programs ended because you’d see an energy surge as kettles were turned on and it was pleasing to see be able to see these spikes actually coming out in our data.
Subsequent to the market going live I worked to further refine and improve the settlement system - in particular by optimising much of the SQL. I also worked as a system analyst helping the energy generation companies understand the settlements being generated by the system.