Loughborough University
Browse
PhDThesis_RT_final_submission.pdf (18.51 MB)

The sequential design optimization of building performance

Download (18.51 MB)
thesis
posted on 2022-11-16, 14:43 authored by Riccardo Talami

Computational methods of design optimization offer great potential in solving the challenges posed by the 

design of high-performing buildings. However, numerical optimization in building design industry is not 

sufficiently exploited and has not yet emerged as part of the design process. This reveals the need to develop 

optimization methods to support the identification of well performing designs while overcoming the long 

computational time of performance evaluation due to the multi-dimensionality of real-world problems. In 

response, this thesis aims to evaluate a sequential approach in the multi-objective design optimization of 

building geometry, fabric, HVAC system and its controls. 

An algorithm based on individual optimization steps conducted in sequence while concurrently considering 

building design elements is developed to ensure a low computational burden and a whole-building holistic 

approach. Three metrics are developed to evaluate the sequential design optimization: the performance of 

the sequential search, the reliability (robustness) to variations of the same problem formulation with respect 

to uncertain boundary conditions (weather and internal loads) and scale (four scales with increasing number 

of solutions), and the efficiency of the search as compared performance with a NSGA-II algorithm. Six criteria 

pertaining to three categories are applied to the three metrics for the evaluation: search effectiveness in 

finding solutions with respect to the search and global optima, computational performance of the search with 

respect to the computational load and computational savings compared to a full factorial search, and solution 

optimality with respect to the objective and variable space. 24 different configurations of the search derived 

from 4 initial starting points, 3 categories of grouping of the building design elements and 2 runs of the 

algorithm are investigated. 

The results are presented for the landscape of configurations of the sequential search and benchmarked 

against the global set of optimal solutions derived from a full factorial search. The obtained results displayed 

different magnitudes of performance, reliability, and efficiency depending on the configurations of the 

search. Therefore, according to the purpose of the optimization, different configurations of the sequential 

search could be preferred for different stages of the building design process. Among them, the iterative run 

of the search in a 2-stage sequential process mapping the architectural and engineering disciplines (field 

grouped) configured as (building geometry + fabric) and (HVAC system + controls) matched the results 

obtained from a full factorial search in terms of effectiveness in finding solutions and solution optimality, 

with a computational saving of function evaluations of 91.2% compared to the full factorial. Additionally, the 

sequential search proved to be reliable to variations with respect to uncertain boundary conditions and scale 

of the problem formulation. Finally, for an equal number of function evaluations, the sequential search 

outperformed the NSGA-II in terms of effectiveness in finding solutions and solution optimality.

Funding

Engineering and Physical Sciences Research Council (EPSRC)

History

School

  • Architecture, Building and Civil Engineering

Publisher

Loughborough University

Rights holder

© Riccardo Talami

Publication date

2022

Notes

A Doctoral Thesis. Submitted in partial fulfilment of the requirements for the award of the degree of Doctor of Philosophy of Loughborough University.

Language

  • en

Supervisor(s)

Jonathan Wright ; Bianca Howard

Qualification name

  • PhD

Qualification level

  • Doctoral

This submission includes a signed certificate in addition to the thesis file(s)

  • I have submitted a signed certificate