This thesis aims to explore two ways in which the electrical infrastructure of a gigawattscale offshore wind farm may be optimised to reduce costs and aid the increasing deployment of renewable energy capacity. The two areas considered are (1) the integration of
a cable layout optimisation with a turbine layout optimisation forming a novel concurrent optimisation framework, and (2) the integration of energy storage systems into a
cable layout optimisation for the purpose of peak-shaving power in the cables enabling
alternative cable sizes to be used and reducing peak electrical losses.
While there are a wide variety of approaches to turbine layout optimisation, the
particle swarm optimisation (PSO) method offers a good balance between accuracy and
computational expense, enabling large-scale problems to be handled by standard desktop PCs. The turbine layout problem is formulated as a grid-based layout, fully defined
by eight variables, to comply with maritime navigation and search and rescue regulations, while allowing some deviation to maximise energy capture. The eight variables
defining the grid are optimised by means of PSO, followed by a novel micro-siting function to move individual turbines and increase energy yield. The method was compared
to SSEs in-house method, matching the energy capture of a case study of their Berwick
Bank site to within 0.3%.
Two cable layout optimisation methods from the literature are selected, which are
the widely used mixed-integer linear programming (MILP) method, and ant colony optimisation (ACO) representing the increasing use of heuristic approaches. These are
compared to a novel ACO-based method, ACOsp, that employs a decomposition strategy. The ACOsp method is shown to maintain the good quality solutions of the MILP
approach, with solutions 0.0-1.4% more expensive than optimal, while also demonstratii
ing the computational efficiency of heuristic approaches, useful for large-scale problems.
An optimisation framework considering the concurrent optimisation of turbine and
cable layouts is proposed, with comparison made to a sequentially optimised solutions,
isolating the impact of this integration. Solutions of the integrated, concurrent, approach show improved objective values where the increase is statistically significant.
For a case study site with 164-165 turbines, the approach increases the objective value
(for this maximisation problem) by 0.45%, which is slightly less than the increase found
by the addition of one further turbine at 0.55-0.57%.
Considering the limitations of the investigated cable layout optimisation approaches,
a following study proposed a MILP-based optimisation in combination with a decomposition strategy, MILPsp. The MILPsp method maintained the accuracy of the MILP
method and reduced computational expense, improving on the earlier ACOsp method.
Variables describing ESS are integrated into the MILPsp algorithm to determine the
impact of using co-located ESS on the array cable network. It was found that very few
charging strategies were able to deliver meaningful peak shaving to the power in the array cables and those that did required a very large ESS capacity to do so (3MW/64MWh
for a site using 8MW turbines). Further, the required cost of the co-located ESS was
prohibitively low, at
| Date of Award | 18 Nov 2024 |
|---|
| Original language | English |
|---|
| Awarding Institution | - University Of Strathclyde
|
|---|
| Sponsors | EPSRC (Engineering and Physical Sciences Research Council) |
|---|
| Supervisor | Olimpo Anaya-Lara (Supervisor) & David Campos-Gaona (Supervisor) |
|---|