Millimeter Precision in the Factory, Gaps on Site

Modular construction achieves millimeter-level precision in factory conditions, but site foundations and previously placed modules accumulate dimensional errors that compound with each stacked floor. A building designed for factory precision encounters site-level tolerances that are 5–10× looser, forcing crews to spend hours shimming, trimming, and wet-grouting interfaces that were supposed to be bolt-together connections. In high-rise modular (8+ stories), accumulated geometric drift can exceed structural connection tolerances, risking slip failure under lateral loads. This erodes the 50% time savings that motivated offsite fabrication.

Modular construction is projected to capture 5–10% of new building starts globally by 2030, driven by labor shortages and housing demand. But rework at module interfaces currently consumes 15–25% of on-site labor, undercutting the economic case. If the tolerance management problem is not solved, modular construction will remain limited to low-rise, non-structural applications where dimensional errors are forgiving.

3D laser scanning of foundations before module placement can detect as-built deviations, but no closed-loop system feeds scan data back to factory production fast enough to adjust modules before they ship. Current tolerance management relies on oversized connection details that waste material and add weight. Adjustable base connections exist but increase cost and complexity per joint. The McKinsey framing treats this as a "project management" problem, but the real barrier is metrology and feedback control: no real-time measurement-to-fabrication loop exists between site and factory.

A real-time dimensional feedback system that scans each placed module, computes cumulative geometric error, and transmits corrected fabrication dimensions to the factory for subsequent modules — essentially a closed-loop manufacturing tolerance control system spanning the factory-to-site boundary. This also requires adaptive connection details adjustable in three axes that can absorb residual mismatch without compromising structural performance. Analogous systems exist in aerospace fuselage assembly; the adaptation challenge is cost and environmental robustness.

A student team could instrument a small-scale modular mock-up with photogrammetry or LiDAR to measure cumulative dimensional error across 3–5 stacked units, then prototype a feedback algorithm that adjusts subsequent unit dimensions. Mechanical engineering, construction management, and controls/robotics skills would be most relevant.