To boost economic development, promote community revitalization, and create social harmony, the Hong Kong government has accelerated the long-term development of major infrastructures. One of the critical issues with respect to achieving sustainable infrastructures is making wise decisions in the early integrated design stages. This decision-making, referred to as Multidisciplinary Decision-Making (MDM), is a process of integrating knowledge from multidisciplinary professionals. The MDM process aims to optimize the evaluation and selection of sustainable infrastructure designs, called Design Alternatives (DAs). During the evaluation process, Decision Makers (DMs) evaluate DAs based on their respective decision benchmarks, called the Design Criteria (DC). In the selection process, DMs select the preferred DAs by comparing their potentials to achieve sustainability, namely by their Sustainability Performance Index (SPI). To guarantee that the MDM process is effective and efficient, reasonable DA(s) that satisfy all of the DMs should be consistently achieved.

In the pursuit of consistent MDM processes, two kinds of uncertainty arise in DA evaluation and DA selection. The first uncertainty is the preference uncertainty across DMs occurred in the DA evaluation process. The second uncertainty is the outcome uncertainty across DAs occurred in the DA selection process. To mitigate the first uncertainty, heterogeneous DC and their distinguishable impacts on SPIs must be explicitly represented in DA evaluation. To manage the second uncertainty, inconsistencies across individual, disciplinary, and multidisciplinary DMs should be measured and decision spaces should be enlarged to generate newly optimized DAs.

Previous studies have been done to resolve the above uncertainty problems. To mitigate uncertainty in DA evaluation, heterogeneous DC with multidisciplinary knowledge and ranges have not been explicitly specified in the MDM context. To manage the uncertainty in DA selection, multi-level decision inconsistencies have not been dynamically measured and optimized DAs cannot be automatically generated. To this end, this research aims to propose an integrated MDM (iMDM) system that is capable of mitigating and managing uncertainties in both DA evaluation and selection in a concurrent manner. To achieve this aim, a conceptual framework consisting of a representation framework and a modeling framework was constructed. The representation framework mitigates uncertainty in DA evaluation by specifying MDM information, including DM, DC, DA, and SPI. The modeling framework manages uncertainty in DA selection by integrating and optimizing MDM information. This research has four objectives: (1) to establish a multidisciplinary metric system; (2) to develop two MDM representation models; (3) to formalize an MDM modeling method; and (4) to design and validate an iMDM system.

This research is based on statistical analyses and modeling developments to achieve above objectives. To begin with, two MDM representation models were developed to specify multidisciplinary DMs with heterogeneous DC and infrastructure DAs with predictable SPIs. In addition, an MDM modeling method integrating the complex and multi-attribute group decision-making modeling and the multi-objective harmony search optimization programing was developed to rank and generate DAs. Finally, an iMDM system integrating the representation models and the modeling method was established and tested through Python programming, which is to underpin the mitigation and management of MDM uncertainties in both DA evaluation and selection.

This research further validated the research results based on case studies and experiments of professionals. The MDM representation models were validated by representing MDM information (i.e., multidisciplinary DMs, heterogeneous DC, evaluated DAs, and predicted SPIs) in three case studies. The MDM modeling method was then validated through adopting experiments of professionals evaluating the integration and optimization of MDM information. The iMDM system was finally validated by trial runs evaluating and selecting sustainable infrastructure designs. The validation evidence indicates that the iMDM system integrating the representation models and the modeling method enables efficient and effective mitigation and management of uncertainties in both DA evaluation and DA selection.

To conclude, this research contributes to both knowledge and practice in integrated design of sustainable infrastructures. Theoretically, the MDM representation models contribute to the Triple Bottom Line theory by specifying heterogeneous DC preferences in evaluating sustainable infrastructure designs. The MDM modeling method contributes to the group decision-making theory by integrating a multi-objective optimization process leveraged by heterogeneous DC. The iMDM system integrates into its MDM modeling process uncertainty measurement based on the social choice theory, and into uncertainty management through the generation of new DAs based on the uncertainty management theory. Practically, the iMDM system supports early decision workshops for sustainable infrastructure designs in which multidisciplinary knowledge is collectively integrated. The system is generalizable to structuring MDM rationales at the product (DAs with SPIs), process (evaluation and selection), and organizational (DMs with DC) levels.