Monte Carlo simulations and sensitivity analysis for developing control strategies in hybrid ventilation systems

• Hybrid ventilation control strategies developed via Monte Carlo simulations • Method demonstrated on a three-room Oslo office case study • Mechanical ventilation airflow ranked as most critical input in sensitivity analyses • Interactive parallel coordinates plots enable exploration of the control space • Stakeholder input and multi-criteria decision-making support tailored solutions Hybrid ventilation hold substantial potential for minimising energy use while maintaining acceptable indoor air quality and thermal comfort, yet developing effective control strategies is challenging because many variables interact and objectives conflict. Widely used heuristics (rule-based, fuzzy logic) are simple but inflexible; optimal approaches (Model Predictive Control, Reinforcement Learning) need extensive high-quality data that are rarely available in early design. This study proposes a Monte-Carlo framework that couples global and local sensitivity analysis with multi-criteria decision-making to develop hybrid ventilation control strategies from limited early-stage information. A generic Oslo office building with three room types (open-plan office, conference room, cell office) serves as the case study. Inputs such as temperature setpoints, airflow rates, night ventilation schedules, are sampled within stakeholder-defined ranges; 10000 simulations per room map the control space. Sensitivity analysis identifies mechanical ventilation airflow as the dominant driver of energy use, indoor air quality and thermal comfort metrics across all rooms. Interactive parallel coordinates plots with real-time sensitivity feedback allow stakeholders to filter and explore solutions, visualise trade-offs and converge on a small set of high-performing strategies. Imposing a single supply air temperature for all rooms worsens the cell office weighted score by 5 /cent to 12 /cent, quantifying the trade-off of a centralised air handling unit constraint. The framework delivers a structured yet flexible method, requiring only envelope data and broad setpoint ranges. It can be rerun as design evolves, providing consultants with a transparent alternative to rule-of-thumb tuning or data reliant optimal control methods.