Monitoring Built for New Buildings, Not Old Ones

Structural health monitoring (SHM) systems are overwhelmingly designed for new construction — bridges, high-rises, critical infrastructure — where sensors can be embedded during building. The vast majority of the world's existing building stock, particularly pre-code masonry and concrete structures in seismic zones, lacks any monitoring capability. Post-earthquake rapid assessment still relies on visual inspection by structural engineers — a process that takes weeks to months, is subjective, and puts inspectors at risk in damaged structures. No validated protocol exists for deploying SHM on existing unreinforced masonry buildings, which represent the most vulnerable and most numerous structures in earthquake-prone developing regions.

Earthquakes kill an average of 60,000 people per year, with the vast majority of deaths caused by building collapse. After a major earthquake, the critical question for tens of thousands of buildings is: is this structure safe to reoccupy? Wrong answers in either direction are dangerous — clearance of an unsafe building risks collapse fatalities, while unnecessary condemnation displaces residents and slows recovery. The 2023 Turkey-Syria earthquake damaged or destroyed over 300,000 structures; post-earthquake assessment took months, during which millions lived in uncertainty.

SHM sensor networks require power, connectivity, and physical installation points that existing buildings — especially informal or vernacular construction — lack. Data processing techniques (wavelet transform, FFT, Kalman filter) are tuned for structural models of new buildings with known design specifications, not deteriorated or informally modified older structures whose as-built conditions are unknown. ML-based damage detection models are trained on simulated data from idealized structural finite element models, not real degradation patterns. Retrofit SHM research is concentrated in the Global North and East Asia, while Africa, South America, and Southeast Asia — regions with the largest vulnerable building stocks — remain underrepresented. MEMS accelerometers are cheap enough for wide deployment, but interpreting their data for buildings without structural drawings requires methods that do not yet exist.

Low-cost MEMS accelerometer networks with ambient vibration analysis could characterize building dynamic properties without requiring structural drawings, enabling "blind" monitoring of existing buildings. Smartphone-based sensing — using the accelerometers already in billions of devices — could provide coarse structural assessment at near-zero hardware cost. Transfer learning from well-characterized reference buildings to similar but undocumented structures could overcome the lack of structural models. Crowdsourced post-earthquake damage reports calibrated against professional assessments could bridge the inspection capacity gap.

A team could deploy low-cost MEMS accelerometers (e.g., Raspberry Shake) on 2–3 campus buildings of different construction types, extracting natural frequencies and comparing against finite element model predictions, to quantify the "model uncertainty" for existing buildings. An app development team could prototype a smartphone-based structural motion sensing tool using built-in accelerometers, validating against professional sensors. Relevant disciplines: structural engineering, seismology, embedded systems, machine learning.