Bounded Rationality in Construction-Phase Change Impact Assesment

Abstract

Construction-phase Change Impact Assessment (CIA) is routinely formally complete but remains se lectively narrow. Impacts not surfaced before commitment often resurface later as corrective work. This thesis asks how bounded rationality affects the quality of construction-phase CIA and explains how that selective narrowing arises in complex engineering projects. The framework is empirically tested on six CIA episodes in a single industrial construction complex and triangulated with six semi-structured interviews: two with practitioners involved in the case, two with cross-project procedural experts, and two with practitioners working on adjacent projects. The findings are reported throughout at calibrated evidence weights (Strong, Moderate, Preliminary). Time pressure is confirmed as the precipitating condition that determines whether an assessment enters a bounded rational regime at all. Information uncertainty and cognitive load determine which mechanism becomes salient within a bounded rational regime. This is the baseline condition, not a variable to be reduced. Four refinements follow. A meta-BR layer at the contracting stage results in an inherited assessment environment through a design-maturity pathway (Strong) and a procurement fragmentation pathway (Moderate). Early closure operates on two tiers, per change and at portfolio level via bundled settlement, expanding the scope from individual to decision portfolios. Four protect ive conditions bounded the mechanisms. Three were derived from the framework and one, relational backing, was inductively derived as the meta-condition that supports the others under sustained time pressure. Finally, the three constraints work as a self-reinforcing system rather than as parallel inputs, with each decision reducing the resources available for the next. Read through an ecological account of heuristics, the mechanisms turn out to be ambivalent. The samestrategies that narrowed assessment and corrective work in four episodes also closed two others with none. Bounded rationality thus both supports and undermines assessment quality, depending on whether a protective condition makes the residue of a simplified assessment visible and recoverable. Three of the six tools are used within a single assessment cycle, and three are used across cycles. A workflow specification carrying the tools into the documentation environment of the case organization is also presented. The results shift CIA’s practical goal from completeness to managed incompleteness: knowing what was not assessed and having a plan for what to watch for during implementation.