Abstract:

The challenges of repetitive construction projects are addressed that frequently encounter uncertainties such as extreme weather and resource shortages， leading to schedule delays and budget overruns. Current approaches like stochastic and fuzzy optimization face limitations in practical applications due to distribution information requirements， while existing robust scheduling methods lack quantitative analysis of uncertainty propagation. The time-cost trade-off problem is investigated through a robust optimization lens， focusing on developing scheduling plans resistant to duration fluctuations in repetitive activities.The proposed framework integrates geometric analysis of work continuity constraints with robust optimization theory. The core problem is formulated using a budget uncertainty set to bound deviations in activity durations， where each sub-activity \left(i,j\right) has an uncertain duration d_{i,j} within the interval \left[\underline{d_{i,j}},\overline{d_{i,j}}\right]. Deviation variables {\epsilon }_{i,j} are constrained by {\displaystyle \sum _{i=\mathrm{1}}^n}{\displaystyle \sum _{j=\mathrm{1}}^m}{\epsilon }_{i,j}\le T， where T is a parameter controlling the solution's conservatism. A key innovation is the development of a general algorithm for identifying controlling paths within the Line of Balance （LOB） framework. This algorithm dynamically traces delay propagation routes across repetitive units based on geometric workflow continuity and resource transfer logic， enabling the construction of a restricted uncertainty set that prioritizes disruptions along these controlling paths to reduce unnecessary buffers. The resulting finite-domain robust optimization model minimizes the worst-case project duration while incorporating constraints for work continuity， precedence relationships， and budgeted cost thresholds. A two-stage optimization architecture combines budget-constrained uncertainty modeling for worst-case scenario analysis with restricted uncertainty sets that prioritize critical path disruptions. The hybrid Genetic Algorithm-Binary Particle Swarm Optimization （GA-BPSO） algorithm implements a nested optimization logic where outer-layer genetic operations optimize crew deployment while inner-layer particle swarms simulate adversarial duration scenarios， effectively balancing solution robustness against implementation costs.Validation employs a bridge construction prototype featuring 5 sequential processes with 24 non-uniform units， incorporating real-world parameters including crew productivity thresholds， equipment cost gradients， and material expenditure patterns. The testbed models duration uncertainties through engineer-calibrated deviation ranges and right-skewed Beta distributions reflecting typical construction risk profiles. Results demonstrate that the controlling path mechanism successfully contains delay propagation while reducing idle time buffers by 18%-25% compared to conventional robust approaches. The restricted uncertainty sets prove particularly effective in high-risk scenarios， maintaining 97%+ on-time completion rates despite clustered disturbances. The methodology shows strong adaptability across risk levels， enabling managers to strategically balance schedule protection levels against budget constraints through adjustable conservatism parameters.It provides practitioners with a decision-support framework that systematically balances schedule robustness against resource costs in repetitive projects. The controlling path-driven approach offers infrastructure managers an effective tool for risk-informed scheduling， particularly valuable for capital-intensive projects with strict deadline constraints. The quantitative comparison of robustness costs under varying uncertainty levels enables adaptive strategy selection based on project risk profiles. These advancements bridge the gap between theoretical robust optimization and practical construction scheduling needs， extending the application scope of robustness analysis in project management.

Key words: repetitive projects, robust optimization, time-cost trade-off problem, GA-BPSO algorithm