construction · family: designed for a world that doesn’t exist here
monitoring built for newbuildings, not old ones
Structural health monitoring cannot scale to existing building stock
Problem statement
Structural health monitoring (SHM) may be cheaper to install when sensors go in during construction. One of the two reviews behind this write-up notes that sensors embedded in structural components during construction may have lower installation costs than traditional ones, and that integrating these technologies into existing infrastructure presents challenges, including compatibility with conventional construction practices, cost-effectiveness, and long-term durability. The same review finds gaps in the development of self-sensing materials for building structures and historic buildings (and for bridges and metal structures). The other review, written from Ecuador, reports that no system of any type is currently installed in buildings in that country that allows their structural state to be analyzed. Without monitoring, assessment falls back on physical inspection, and the review lists its limits: some places cannot be inspected directly, material may have to be removed to check the structure's state, and the work adds cost and takes considerable time. The review does describe SHM applications on older structures (a historic masonry tower, a masonry church in Portugal, an adobe church in Peru), and it presents itself as a first step toward SHM-based technical procedures in a future version of Ecuador's construction standard.
Why this matters
Ecuador sits on the Pacific Ring of Fire, where, the first review says, 85% of the planet's total seismic energy is released; a 7.8 Mw earthquake struck its central coast on 16 April 2016. After a major earthquake, the critical question for each building 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. Without a monitoring system, the review notes, serious problems become visible only when the situation is excessively critical, leading in the worst case to demolition because the damage cannot be repaired. Its stated aim is to contribute to standardizing SHM for post-earthquake assessment of strategic buildings such as hospitals and schools.
What’s been tried and why it hasn’t worked
All commercially viable sensors require an external power source, either battery or solar, and many have shown excessive cost, low durability, and limited detection capacity and area. The first review finds the field moving toward cost-effective wireless systems in which MEMS sensors are standard, with three-axis MEMS accelerometers the usual choice in the studies reviewed, but it lists what remains unsolved: lower energy consumption, synchronization of signals across wireless nodes, sensor precision, real-time data processing, and damage identification that is accurate without heavy computation. The standard data processing techniques (wavelet transform, FFT, Kalman filter) each carry drawbacks the review tabulates, such as noise sensitivity and the need for calibration. ML-based damage detection algorithms are trained on records of past earthquakes together with digital and mathematical models of structures, and are then evaluated with artificially induced vibrations or movements that simulate an earthquake. Research on self-sensing materials is led by a few countries: China, the United States, and Italy contributed 15%, 13%, and 13% of the articles in the second review, and no African country appears in its chart of authors' countries.
What would unlock progress
The first review names what has to improve: minimum energy consumption, higher-precision sensors, and more accurate damage identification algorithms with less computational consumption. It points to tiny machine learning, where inference runs on the embedded device, as a way to cut the computing burden. Low-cost accelerometer nodes can identify a structure's dynamic properties: one study in the review, using a low-cost accelerometer-based wireless sensor node, identified the first three vibration frequencies and the most relevant damping ratios of a structure. Three further directions are this write-up's own suggestions and are not drawn from either review: smartphone-based sensing, using the accelerometers already in phones, for coarse structural assessment at very low hardware cost; transfer learning from well-characterized reference buildings to similar but undocumented structures; and crowdsourced post-earthquake damage reports calibrated against professional assessments.
Entry points for student teams
A team could deploy low-cost MEMS accelerometers (e.g., an ADXL345 or MPU6050 sensor on a Raspberry Pi or ESP8266 board, parts the first review names) 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.
Genes
Structural cousins: stuck the same way, in other fields
Sources
López-Castro, B., Haro-Baez, A.G., Arcos-Aviles, D., Barreno-Riera, M., and Landázuri-Avilés, B., "A Systematic Review of Structural Health Monitoring Systems to Strengthen Post-Earthquake Assessment Procedures," Sensors, 22(23):9206, 2022, Vasconcelos, A.R.C., Matos, R.A., Silveira, M.V., and Mesquita, E., "Applications of Smart and Self-Sensing Materials for Structural Health Monitoring in Civil Engineering: A Systematic Review," Buildings, 14(8):2345, 2024, accessed 2026-02-20 go to source 1 ↗ go to source 2 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Two reviews. The first (Sensors, 2022) reviews SHM technologies, architectures, data processing and damage identification for post-earthquake assessment, with Ecuador as its case, and describes a three-node prototype in a university building; it does not state how many studies it covered. The second (Buildings, 2024) is a systematic review of 33 articles on self-sensing materials, from a database search in February 2024. Neither review states a gap between new and existing buildings as its finding; that framing is this write-up's, resting on the second review's remarks about embedding sensors during construction and the challenges of integrating them into existing infrastructure. Related briefs: infrastructure-cascading-failure-modeling (infrastructure system failure), wildfire-wui-fire-codes-unproven (building code gap for existing structures in fire zones). The smartphone-based sensing entry point is this write-up's own suggestion; neither source discusses it.
Reconciliation 2026-10-01: Fact repair after reading the whole write-up against both sources. Both were read in full on 2026-10-01, and a second reader went over every repair against the same two sources on the same day; all fifteen stand, with six sentences tightened so they say no more than the papers do. The publisher pages at mdpi.com refused automated access, so the first paper was read from the Europe PMC full text (https://www.ebi.ac.uk/europepmc/webservices/rest/PMC9736237/fullTextXML) and the second from the publisher's own PDF (https://mdpi-res.com/d_attachment/buildings/buildings-14-02345/article_deploy/buildings-14-02345.pdf); the citation records were read at https://api.crossref.org/works/10.3390/s22239206 and https://api.crossref.org/works/10.3390/buildings14082345. Most of the write-up's specific claims were not in either paper. (1) The Source line gave the first paper's authors as "Bhuiyan, M.Z.A. et al." The record lists López-Castro, Haro-Baez, Arcos-Aviles, Barreno-Riera and Landázuri-Avilés; no Bhuiyan is among them. It gave the second paper's title as "Smart materials for SHM" with no authors. The record gives "Applications of Smart and Self-Sensing Materials for Structural Health Monitoring in Civil Engineering: A Systematic Review" by Vasconcelos, Matos, Silveira and Mesquita. Both are now copied from the records; journal, volume, issue, article number and year were already right. (2) The text said SHM systems are "overwhelmingly designed for new construction" and that the vast majority of the world's existing buildings lack any monitoring. Neither paper says this. The second says sensors embedded during construction may have lower installation costs than traditional ones, that integrating these technologies into existing infrastructure presents challenges, and that there are gaps for bridges and metal structures, building structures and historic buildings; the first says no such system is installed in buildings in Ecuador. The paragraph now says that. (3) The text said post-earthquake assessment by visual inspection takes weeks to months, is subjective and puts inspectors at risk. The first paper says only that physical inspection cannot reach some places, may need material removed, costs more and takes considerable time; the sentence now follows it. (4) The text said no validated protocol exists for monitoring existing unreinforced masonry buildings and called them the most vulnerable and most numerous structures in developing regions. Neither paper says this, and the first describes SHM applications on a historic masonry tower and on masonry and adobe churches; the claim was removed and those applications are now mentioned. (5) Removed because neither paper contains them: an average of 60,000 earthquake deaths a year; "tens of thousands of buildings"; and the 2023 Turkey-Syria earthquake figures (over 300,000 structures, months of assessment, millions affected), which postdate the first paper. The paragraph now uses the first paper's own facts: the Pacific Ring of Fire releasing 85% of the planet's seismic energy, the 7.8 Mw earthquake of 16 April 2016, and its stated aim of standardizing SHM for strategic buildings such as hospitals and schools. (6) The text said existing, informal or vernacular buildings lack the power, connectivity and installation points sensors need. The papers say sensors need an external power source and list energy use, signal synchronization across wireless nodes, precision, real-time processing and computing cost as open challenges, without tying them to older or informal buildings; the sentence now says that, and the first paper's finding that cost-effective wireless systems with MEMS sensors are standard replaces the write-up's earlier wording. (7) The text said the wavelet transform, FFT and Kalman filter are "tuned for structural models of new buildings." The first paper names these three as the standard techniques and lists their drawbacks; it says nothing about new buildings. (8) The text said machine-learning models are trained on simulated data from idealized finite element models, not real degradation. The first paper says the algorithms are trained on records of past earthquakes and on models of structures, and then evaluated with artificially induced vibrations that simulate an earthquake; the sentence now says that. (9) The text said retrofit research is concentrated in the Global North and East Asia while Africa, South America and Southeast Asia are underrepresented. The second paper, which covers self-sensing materials only, says China, the United States and Italy contributed 15%, 13% and 13% of its articles and that the articles span continents including South America; its chart of authors' countries includes Brazil, Colombia, Vietnam and Singapore and no African country. The sentence now reports that. (10) The text said interpreting MEMS data for buildings without structural drawings "requires methods that do not yet exist." Not in either paper; removed. That MEMS accelerometers are low-cost and the usual choice is confirmed in the first paper. (11) In the unlock section, monitoring "without requiring structural drawings" and phones "in billions of devices" were removed. The smartphone, transfer-learning and crowdsourcing ideas are in neither paper; they are kept but now labelled as this write-up's own suggestions, and the section leads with the improvements the first paper names, its mention of tiny machine learning, and the one study it cites in which a low-cost accelerometer-based wireless sensor node identified a structure's first three vibration frequencies and damping ratios. (12) The entry point named "Raspberry Shake" as an example sensor; neither paper mentions it. The example now names parts the first paper does (ADXL345, MPU6050, Raspberry Pi, ESP8266). Both entry points are otherwise unchanged and neither needs a special facility. (13) The notes above said the review covered "publications 2010–2024, 217 studies" and that the new-versus-existing gap is "the central finding." The first paper gives no count of studies and no date range (it was published in 2022 and has 84 references); the second reviewed 33 articles found in a February 2024 search; neither states that gap as a finding. The notes now say so. No quotation attributed to a source was found in the write-up. The long title and the display title were not changed, but neither paper states that monitoring cannot scale to existing buildings or that it is built for new ones; the titles carry the write-up's framing, not a finding of the sources. The sentence about wrong answers in either direction after an earthquake is the write-up's own reasoning and was left. Genome tags untouched.