A multi-objective decision-support framework for early-stage cost–carbon–process-hour screening of prefabricated cold-formed steel residential framing in New Zealand

Abstract

Early-stage configuration decisions in prefabricated cold-formed steel (CFS) residential framing can influence process cost, A1–A5 embodied carbon, and construction process-hours before detailed structural verification, procurement, and construction planning are complete. In the New Zealand residential construction context, these decisions are important because panel rationalisation, truss rationalisation, opening/detailing complexity, and prefabrication level affect how work is distributed across factory production, transport, logistics, and site installation. As a result, single-objective assessment is insufficient for comparing alternative framing configurations.

This thesis develops a multi-objective decision-support framework for early-stage cost–carbon–process-hour screening of prefabricated CFS residential framing in New Zealand. A project-derived residential framing package is represented using the decision vector 𝐱=[𝑅𝑝,𝑅𝑡,𝑂𝐶,𝑃𝑓], where 𝑅𝑝 and 𝑅𝑡 represent panel and truss rationalisation intensity, 𝑂𝐶 represents opening/detailing complexity, and 𝑃𝑓 represents prefabrication level. Three objectives are minimised simultaneously: CFS framing process cost, A1–A5 embodied carbon, and CFS framing process-hours. A model-calculated S2 reference configuration is used for normalisation, comparison, and robustness assessment.

An LHS-initialised NSGA-II optimisation procedure was used to generate nondominated cost–carbon–process-hour alternatives within the fixed case-study boundary. The reported optimisation run produced 8,302 feasible nondominated configurations without hard constraint violations, of which 2,695 achieved simultaneous reductions in predicted process cost, embodied carbon, and process-hours relative to S2. The remaining nondominated configurations are interpreted as trade-off alternatives rather than universal improvements over the reference case.

Model credibility was assessed through S2 reasonableness checking, coefficient-audit traceability, convergence diagnostics, sensitivity analysis, constraint testing, multi-seed stability checking, and Monte Carlo robustness assessment. The results indicate that improved performance is achieved through coordinated rationalisation, detailing simplification, and prefabrication control rather than by maximising any single variable independently. The contribution of this thesis is an applied-methodological decision-support workflow for screening prefabricated CFS residential framing configurations before detailed structural design, contractor pricing, life-cycle assessment, and construction scheduling.