Environmental Engineering Reference
In-Depth Information
•
An optimisation tool
An optimisation tool hence should combine forecasts with grid constraints, market
trends and the physical constraints of the storage unit. Congestion on the grid is a typ-
ical reason for higher prices, but mostly to the disadvantage of the energy pool, if it is
dominated by wind power. The optimisation problem is in theory global, but in practise a
limited area problem, because the global problem extends to political decisions that would
have to be modelled by stochastic processes. Instead, optimisation should be applied on
the local grid. For this, time dependent boundary conditions are required. These should in
theory contain the large-scale global trends, but at some stage approach the average trends
for the season. The optimisation tool will have to comprise a set of partial differential equa-
tions including the storage unit, a portion of the grid and the intermittent energy sources.
The partial differential equations describe the total output and exchange between storage
units and the intermittent source. The accuracy of the numerical solution is partially de-
termined by the extent of the domain of dependence for the partial differential equations.
The domain increases with forecast horizon and can well reach part of the boundary, if the
boundary values are accurately predictable by a simple function or some other prediction
tool.
An example of a predictable boundary condition could be the large-scale electricity
demand, which could be approximated with a periodic function to simulate the diurnal
cycle. The large scale demand will after some hours have an influence on how the storage
system should be scheduled, but the time derivatives of our intermittent energy is then likely
to be the dominant forcing term on the storage equation.
The domain of dependence for the solution increases with the forecast horizon, partly
because the dependency domain of the weather forecast increases. However, such an energy
system has as a good approximation no feedback on the weather, thus this system can be
one way coupled. Also, trading of oil and the transmission of gas have both an influence on
the pricing. Conflicts between employers and their labour (e.g. strikes), unavailability of
multiple plants and extreme weather at offshore platforms can cause peaks in the spot mar-
ket prices of gas and oil and consequently also electricity. Such peaks can have a dominant
negative impact on the average market value of wind power, if there would be imbalance in
the pool during an interval with high prices.
An objective optimisation process would hence require an algorithm that carries out
simulations in a closed system, but with the possibility to control the time dependent bound-
ary conditions. Typically, a strike would be known in advance and the time dependent
boundary conditions could be manually adjusted by the user to take account for such ef-
fects. Although the optimisation problem should ideally encompass the entire globe, it
appears that the impact from far distances on the next couple of days can be modelled
equally well by subjective boundary conditions, as with attempts to objectively model such
effects. The conditions that take place at far distances are rather consequences of unpre-
dictable events that have spurious nature. Any objective algorithm would have to be tuned
to discard observations that conflict with the present state of the system to prevent that the
large scale numerical solution would become unstable. Large scale waves would propagate
through the system and trigger new waves on the local scale and the final solution would
destabilise the energy system and the result would be higher balancing costs.
Search WWH ::
Custom Search