CoChat

Field monitoring and assessment of reinforced concrete building collapse using remote sensing and digital twins

Open access
Bai, Yongsheng, Sezen, Halil, Yilmaz, Alper, Qin, Rongjun · 2026 · Journal of Civil Structural Health Monitoring

The DOI matches its Crossref record; 40 of the 40 references it cites were found in a registry or an index; 2 of the 12 sources its claims cite were checked for support.

Abstract

This paper introduces a novel application of remote sensing systems and digital twins to investigate the collapse potential of existing buildings... offering practical solutions for remotely monitoring full-scale RC structures and evaluating their real performance.

Overview

CoChat Summary

TL;DR

This paper reports a field experiment on a six-story reinforced concrete parking garage where columns and slabs were removed, using cameras, drones, wireless accelerometers, and LiDAR-based digital twins to monitor structural response. The authors show that remote sensing could track damage, vibration, beam deflection, and helipad deformation, with results generally matching accelerometer or analysis-based references, while also exposing practical field limitations.

Key findings

  • Damage detection with A-PaNet Mask R-CNN varied strongly by camera type: inside cameras achieved 99.4% accuracy (3224/3244 images), while outside camera 1 reached 76.3% (363/476), outside camera 2 30.4% (284/943), cell phone 78.8% (172/220), and drone 56.8% (172/303).
  • Remote-camera vibration measurements matched accelerometers closely in many cases; for the targeted facade and helipad, frequency differences were reported as less than 4.7% and 9.5% in two directions using Mask R-CNN + SIFT.
  • The frequencies extracted from cameras were around 13.9 Hz horizontally and 11.2 Hz vertically, with no significant increase or decrease after successive slab removals.
  • Beam and slab deformation was measurable with vision and digital twins: the beam above the second removed column had a maximum measured deflection of about 75 mm at 360 s, compared with 16 mm from nonlinear structural analysis; the helipad’s maximum downward displacement was around 50 mm at the right tip, versus 28.5 to 39.8 mm from nonlinear dynamic analysis.

Methods

  • Field experiment on a six-story RC-framed parking garage at Ohio State University, conducted in January 2021 over two days, during which cantilever slabs were removed and two columns were cut/removed in sequence.
  • Multi-sensor monitoring used in-situ and remote cameras, drones, cell-phone video, three wireless accelerometers, and LiDAR; LiDAR scans before and after removal were aligned with ICP to build digital twins.
  • Main analyses included A-PaNet Mask R-CNN for damage detection, Mask R-CNN + SIFT and Lucas-Kanade (LK) tracking for vibration/displacement measurement, FFT for frequency extraction, and point-cloud patch comparison for digital-twin-based deflection estimation.

Limitations

  • The paper states that field experiments are inherently difficult because of dust obstructing visibility, risk of sensor damage, and the need for careful timing, weather, and sensor placement.
  • The authors state that in-situ LVDTs and strain gauges were damaged before useful data could be gathered, preventing direct verification with those traditional sensors.
  • Evident from the text: some measurements could not be cross-checked because other sensors were destroyed during the experiment, and LiDAR could not reliably capture very small beam deflections (less than 5 mm) at the long scanning distance.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1007/s13349-025-01044-3

Cites

40 of 40 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 A workflow for using remote sensing systems and digital twins to monitor the existing building and evaluate its performance
Fig. 1 A workflow for using remote sensing systems and digital twins to monitor the existing building and evaluate its performanceFrom the PDF
Image 2
From the PDF
Fig. 2 An illustration of deep learning methods to automatically detect structural damage and measure the displacement of an object with visual data from cameras. a is the framework of a deep learning
Fig. 2 An illustration of deep learning methods to automatically detect structural damage and measure the displacement of an object with visual data from cameras. a is the framework of a deep learningFrom the PDF
Fig. 3 Translation of a target measured by a bounding box or by match -ing keypoints on an image plane.
Fig. 3 Translation of a target measured by a bounding box or by match -ing keypoints on an image plane.From the PDF
Fig. 4 A process of ICP alignment for two point clouds. The orange solid and blue dash lines represent the reference and new clouds, respectively, whereas the red lines between two matching pairs with the ICP algorithm
Fig. 4 A process of ICP alignment for two point clouds. The orange solid and blue dash lines represent the reference and new clouds, respectively, whereas the red lines between two matching pairs with the ICP algorithmFrom the PDF
Image 6
From the PDF
Image 7
From the PDF
Fig. 8 A drone flying next to the building. a is its deployment on testing day, and b is its captured view of the building
Fig. 8 A drone flying next to the building. a is its deployment on testing day, and b is its captured view of the buildingFrom the PDF
Image 9
From the PDF
Image 10
From the PDF
Image 11
From the PDF
Image 12
From the PDF
Fig. 12 Visual data processing versus accelerometer measurements to capture vibration signals and extract frequencies of the helipad in the horizontal direction at 9:36 AM. a shows two different methods
Fig. 12 Visual data processing versus accelerometer measurements to capture vibration signals and extract frequencies of the helipad in the horizontal direction at 9:36 AM. a shows two different methodsFrom the PDF
Fig. 13 Visual data processing versus accelerometer measurements to capture vibration signals and extract frequencies of the helipad in the vertical direction at 9:36 AM. a shows two different methods to obtain
Fig. 13 Visual data processing versus accelerometer measurements to capture vibration signals and extract frequencies of the helipad in the vertical direction at 9:36 AM. a shows two different methods to obtainFrom the PDF
Fig. 15 Frequencies extracted by visual methods and measured by accelerometers on the helipad in the sixth story at different times during the first testing day. a and b are results in the horizontal and vertical directions, respectively. For ACC2, the frequency around 13.9 Hz in the horizontal direction is not detected from 11:46 AM to 2:10 PM
Fig. 15 Frequencies extracted by visual methods and measured by accelerometers on the helipad in the sixth story at different times during the first testing day. a and b are results in the horizontal and vertical directions, respectively. For ACC2, the frequency around 13.9 Hz in the horizontal direction is not detected from 11:46 AM to 2:10 PMFrom the PDF
Image 16
From the PDF
Image 17
From the PDF
Image 18
From the PDF
Image 19
From the PDF
Fig. 17 Measured and filtered deflections of two selected loca -tions on the beam near the removed column in the fifth story. A positive value means downward deflection
Fig. 17 Measured and filtered deflections of two selected loca -tions on the beam near the removed column in the fifth story. A positive value means downward deflectionFrom the PDF
Image 21
From the PDF
Fig. 18 Views of pre- and post-col -umn-removal point clouds (partial digital twins) for the building. a , b and c , d are elevation and 3D side views of the building before and after the removal of one top corner column and slabs at the floors and helipad, respectively
Fig. 18 Views of pre- and post-col -umn-removal point clouds (partial digital twins) for the building. a , b and c , d are elevation and 3D side views of the building before and after the removal of one top corner column and slabs at the floors and helipad, respectivelyFrom the PDF
Fig. 20 Deformation of the helipad from a plan view a and an exagger -ated elevation view b after average coordinate changes of patches in point clouds are computed. Its range is shown in the color bar and its unit is in meters
Fig. 20 Deformation of the helipad from a plan view a and an exagger -ated elevation view b after average coordinate changes of patches in point clouds are computed. Its range is shown in the color bar and its unit is in metersFrom the PDF

Verification

Claims the paper makes

2 Valid · 1 Not valid · 7 Not checked · 55 Own results and interpretations

36 more claims

Create a free account to see all 65 claims and to verify the 3 references still open.

See all claims — free account
  • It may state two separate claims as one. 11
  • Probably not a substantial claim. 12
  • The quote only partly states this claim. 3

Create a free account to see these claims.

Citations

Cites

Research works printed but not linked: 5

  1. 24. Lodhi MS (2012) Seismic evaluation of reinforced concrete col -umns and collapse of buildings. M.S. Thesis, The Ohio State University
  2. 28. Li K (2017) Collapse experiments and assessment of masonry wall buildings. PhD Dissertation, The Ohio State University
  3. 38. Zha B, Bai Y, Yilmaz A, Sezen H (2019) Deep convolutional neu -ral networks for comprehensive structural health. Int J Comput Vision 115(3):211-252
  4. 43. Bouguet JY (2001) Pyramidal implementation of the affine Lucas Kanade feature tracker description of the algorithm. Intel Corpo -ration 5(1-10):4
  5. 47. Betit BA (2021) Experimental and analytical assessment on the progressive collapse potential of a reinforced concrete building. Master Dissertation, The Ohio State University

Metadata

Authors
Bai, Y., Sezen, H., Yilmaz, A., & Qin, R.
Affiliations
  • Texas State University (US)
  • The Ohio State University (US)
Corresponding author
Sezen, Halil
Venue
Journal of Civil Structural Health Monitoring
Published
2026-02-11
Volume
16
Issue
2
ISSN
  • 2190-5452 (ISSN-L)
  • 2190-5479
OpenAlex
W7128599389
Cited by
0
References
38
FWCI
0.00
Type
journal article
Language
en
Open access
Licence
Locations
Journal of Civil Structural Health Monitoring (publishedVersion, cc-by)
Funders
Directorate for Engineering (2036193)
Topics
  • Physical Sciences › Engineering › Civil and Structural Engineering › Structural Health Monitoring Techniques
  • Physical Sciences › Engineering › Civil and Structural Engineering › Infrastructure Maintenance and Monitoring
  • Physical Sciences › Engineering › Civil and Structural Engineering › Structural Response to Dynamic Loads
Keywords
reinforced concrete buildings, building collapse, digital twin, remote sensing, structural health monitoring, column removal, damage detection, deep learning, vibration measurement, LiDAR, unmanned aerial vehicles, wireless accelerometers, residual deflection, nonlinear structural analysis

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.