What was done?
DNV wanted to determine the cheapest way to generate electricity, subject to operational and financial constraints, something known as the “Unit Commitment Problem” – a series of mathematical problems to find a low cost operational schedule for power generators.
After receiving training on quantum computing, the DNV team worked with quantum computing experts from Digital Catapult to formulate this problem as a combinatorial optimisation problem, where binary variables represented different combinations of energy sources that could be switched on, or off, at different time periods. In addition, for each time period, the cost of production was optimised subject to constraints including enough energy being produced to meet the demand in each period. A penalty term was added to the cost if the solution did not meet a constraint. The ORCA PT-1 quantum boson sampler (quantum computer) produced binary strings mapped to the relevant binary variables and the boson beam splitter angles were tuned classically to find the binary string which corresponded to the lowest cost.
What were the results?
DNV simulated four different energy sources which could be switched on or off for up to eight time periods. DNV’s team experimented with constraints, such as power supplied from a gas plant being required for at least three consecutive time periods, and the gas plant ramping up to 25%, 50%, 75% and 100% over four time periods. The simulations were generally successful, but in some larger data sets the optimum solution was not found because there were too many possible combinations of input variables. DNV benefited from the flexibility of the boson sampler in allowing different penalty terms.
DNV successfully ran a scaled down version of the problem with three energy sources and two time periods on the ORCA PT-1 quantum computer.
