Building performance simulation — energy consumption, structural behavior, daylighting, thermal comfort, acoustics — generates data that should inform every significant design decision. In practice, it typically happens too late and too rarely: as a compliance check at permit stage rather than an exploratory tool during design. AI changes this by connecting simulation engines directly to the parametric model and automating the interpretation of results.
The Simulation Gap in Architectural Practice
The barrier to simulation is not the simulation engines — EnergyPlus, Radiance, OpenFOAM, and SAP2000 are powerful and free. The barrier is the workflow: manually preparing the geometry, configuring the simulation parameters, running the solver, and interpreting numerical output requires specialist knowledge that most architects do not have and cannot justify acquiring for each project phase.
AI dissolves this barrier in two directions: upstream (automated geometry preparation and parameter configuration from natural language) and downstream (AI interpretation of simulation results as design recommendations rather than tables of numbers).
Energy Simulation Integrated into Design
An AI-connected energy simulation pipeline receives the current design geometry from the parametric model, configures an EnergyPlus or similar model with the climate data and building use profile, runs the simulation, and returns results as narrative interpretation: 'The current south facade glazing ratio produces a 23% overheating risk in summer. Reducing glazing to 40% with external shading improves thermal comfort significantly while maintaining the design's transparency intent. The following three alternatives balance comfort, energy performance, and light quality differently — here is the comparison.'
The architect responds to a design recommendation, not to a CSV file of hourly energy loads.
Structural and Acoustic Simulation
The same pipeline pattern applies across simulation domains: connect geometry to solver, run simulation, interpret results as design feedback. For structural analysis, AI can prepare a schematic FEM model from the conceptual design, identify overstressed members, and suggest geometric modifications that improve structural efficiency without departing from the design intent. For daylighting, AI interprets annual Useful Daylight Illuminance maps and recommends window positioning changes. For acoustics, AI reads reverberation time predictions and suggests material or geometry modifications.
The key is that simulation becomes a conversation rather than a specialist report — integrated into the design process rather than delivered as a periodic external review.
MIAW Module F4 — Simulation with AI
Building simulation with AI is the subject of MIAW Module F4. Students build a simulation pipeline applied to at least two relevant performance domains for their own project — typically one bioclimatic and one structural or acoustic. The deliverable is an AI-generated results report with design recommendations, produced directly from the project geometry. The pipeline is reusable: configured once, it can re-simulate the project at any design stage with minimal additional work.