The multi-objective optimum design of building thermal systems

The thermal design of buildings as a multi-criterion optimisation process since there is always a pay-off (balance) to be made between capital expenditure and the operating cost of the building. This thesis investigates an approach to solving 'whole building' optimisation problems. In particular simultaneous optimisation of the plant size for a fixed arrangement of air conditioning equipment, and the control schedule for its operation to condition the space within a discrete selection of building envelopes. The optimisation is achieved by examining a combination of the cost of operating the plant, the capital cost of the plant and building construction, and maximum percentage people dissatisfied during the occupation of the building. More that one criterion is examined at a time by using multi-criteria optimisation methods. Therefore rather than a single optimum, a payoff between the solutions is sort. The benefit of this is that it provides a more detailed information about the characteristics of the problem and more design solutions available to the end user. The optimisation is achieved using a modified genetic algorithm using Pareto ranking selection to provide the multi-criterion fitness selection. Specific methods for handling the high number of constraints within the problem are examined. A specific operator is designed and demonstrated to deal with the discontinuous effects of the three separate seasons, which are used for the plant selection and for the three separate control schedules. Conclusions are made with respect to the specific application of the multi-criterion optimisation to, building services systems, their control, and the viability of 'whole building design' optimisation.