The temperature-based structural health monitoring (SHM) baseline of Yarnold and Moon (2015) relates uniform temperature change, local axial mechanical strain, and global unrestrained displacement of a long-span bridge in a three-dimensional surface whose orientation tracks elastic properties and whose bounds track inelastic (slip-force) properties. On the 168 m tied-arch span of the Tacony-Palmyra Bridge it showed far higher sensitivity to realistic damage scenarios than vibration-based baselines, at much lower instrumentation cost. The source paper itself identified decisive unresolved issues: the analytics assumed uniform temperature change, axial-only response, and elastic-perfectly plastic behavior; field data were restricted to evening/night measurement windows with spatial averaging; validation covered a single structure; and, most critically, neither the time needed to accumulate a baseline nor the detection lag after a damage event was quantified. Later literature has generalized the concept (temperature-driven SHM with three-dimensional signatures), introduced minimal-thermal-gradient data selection and data-driven thermal-response modeling, and applied related temperature-based reasoning to additional structures, but the timing quantification and the extension of the mechanistic 3D surface to gradient, bending, and general hysteretic conditions remain unresolved. The problem of determining, with quantified evidence, the deployment readiness of the temperature-based baseline — baseline-accumulation time, detection lag, thermal realism, and generalizability — remains open.

Background

Structural health monitoring (SHM) equips a bridge with sensors so that changes in its condition can be detected and traced to their causes. The classical approach is vibration-based: natural frequencies and mode shapes (the spatial vibration patterns at each frequency) are tracked as a damage-sensitive baseline, because damage reduces stiffness and should shift these global dynamic properties. Two long-standing frustrations with this approach on long-span bridges are that localized damage — a failed connection, a degraded member — often changes global modal properties too little to detect, and that synchronized dynamic measurement requires expensive data-acquisition, processing, and communication infrastructure.

Ambient temperature offers an alternative driving force. A bridge responds quasi-statically to daily and seasonal heating: members expand and contract, and where this deformation is resisted (at bearings, connections, and supports) measurable mechanical strains and displacements develop. Yarnold and Moon (2015) proposed exploiting this directly: relate, at each instrumented cross-section, the local axial mechanical strain εM\varepsilon_M (measured strain with the free thermal-expansion part removed), the global unrestrained displacement δU\delta_U of the span, and the uniform temperature change ΔT\Delta T (a temperature change assumed identical across the cross-section and along the structure). Plotting these three measured quantities against one another over the full seasonal temperature range traces out a surface in three-dimensional space — the temperature-based baseline. In analytic benchmark studies of a simply supported beam with elastic-perfectly plastic springs (springs that respond linearly until a yield slip force is reached and then slip without additional force) representing boundary and continuity nonlinearities, the orientation of this surface is governed solely by elastic stiffness properties while its bounds are governed solely by the inelastic slip forces, so elastic damage and inelastic damage leave distinct, interpretable signatures. In a case study of the 168 m steel tied-arch span of the Tacony-Palmyra Bridge (a tied-arch bridge carries its deck on hangers from an arch whose thrust is tied by the deck itself), a nonlinear finite-element model calibrated on three years of monitoring data showed that realistic damage scenarios — bearing lock-up, slotted-connection failure, and lower-chord stiffness loss — produced large changes in the baseline surface (for example, direction-cosine changes up to 125% at measurement locations near the damage), while natural frequencies changed by at most 6% and mode shapes were essentially unaffected; the approach also needs only slow, unsynchronized measurements, making it far cheaper than vibration-based monitoring.

The same paper was explicit that these results rest on simplifying conditions, and the limitations it listed define the open problem. The benchmark analytics assumed uniform temperature change, axial-only deformation, and elastic-perfectly plastic behavior, whereas real bridges experience cross-sectional temperature gradients (sunlit versus shaded surfaces), bending deformations, and more complex hysteretic boundary behavior. The field data were restricted to measurement windows — evening/night periods chosen to limit solar-radiation-induced gradients, with spatial averaging used to approximate axial response — so real-time diurnal performance is untested. The demonstration covered a single tied-arch span. Most critically, the authors identified the method's primary drawback as time: a complete baseline requires large seasonal temperature swings to accumulate, and after a structural change certain damage scenarios can only be diagnosed once a sufficiently large new temperature excursion occurs — an inherent alert latency — and neither the baseline-establishment time nor the detection lag had been quantified.

Problem Statement

Determine, with quantified evidence, whether the temperature-based three-dimensional baseline is a practically deployable damage-detection method for long-span bridges. The overarching objective is deployment readiness; it decomposes into three jointly necessary components.

(1) Timing. Quantify, for representative bridge types, climates, and damage mechanisms: (a) the monitoring duration required to accumulate sufficient thermal variation to establish the baseline surface to usable accuracy, and (b) the detection lag between a structural change and the first subsequent thermal excursion large enough to make that change diagnosable from the baseline. Both quantities should be reported as functions of structure, damage class, and temperature history, since the lag after an event depends on the stochastic timing of future weather.

(2) Thermal and mechanical realism. Determine whether the baseline framework can be extended beyond its original assumptions — uniform temperature change, axial-only deformation, and elastic-perfectly plastic nonlinearity — so that baselines can be constructed and damage diagnosed under realistic cross-sectional temperature gradients, bending deformation, and general hysteretic behavior. This includes establishing whether continuous diurnal data (daytime, gradient-laden measurements) can be used, rather than only evening/night windows with spatial averaging.

(3) Generalizability. Determine whether the approach generalizes to structural forms and damage mechanisms beyond the single 168 m tied-arch span — for example cable-stayed, suspension, girder, or truss bridges, and damage classes beyond bearing lock-up, slotted-connection failure, and lower-chord stiffness loss — together with the instrumentation requirements (measurement locations relative to expected damage) that follow.

These three components jointly decide deployment: whether, under given conditions of structure, climate, damage class, and elapsed time, the method delivers reliable detection, and where it does not. The instance lists above (introduced by "for example") are illustrative rather than exhaustive: the problem encompasses the structural forms, damage classes, climate conditions, and measurement locations addressed, and the accuracy and latency criteria that define whether the baseline is usable.

References

  1. Yarnold MT, Moon FL (2015). Temperature-based structural health monitoring baseline for long-span bridges. Engineering Structures 86:157-167. doi:10.1016/j.engstruct.2014.12.042, https://doi.org/10.1016/j.engstruct.2014.12.042
  2. Yarnold MT, Moon FL, Aktan AE (2015). Temperature-based structural identification of long-span bridges. Journal of Structural Engineering 141(11):04015027. doi:10.1061/(ASCE)ST.1943-541X.0001270, https://doi.org/10.1061/(ASCE)ST.1943-541X.0001270
  3. Yarnold M, Dubbs NC (2015). Bearing assessment with periodic temperature-based measurements. Transportation Research Record 2481(1):115-123. doi:10.3141/2481-15, https://doi.org/10.3141/2481-15
  4. Reilly J, Abdel-Jaber H, Yarnold M, Glisic B (2016). Identification of steady-state uniform temperature distributions to facilitate a temperature driven method of Structural Health Monitoring. Proceedings of SPIE 9805 (Health Monitoring of Structural and Biological Systems 2016), 980521. doi:10.1117/12.2219334, https://doi.org/10.1117/12.2219334
  5. Reilly J, Abdel-Jaber H, Yarnold M, Glisic B (2017). Evaluating the coefficient of thermal expansion using time periods of minimal thermal gradient for a temperature driven structural health monitoring. Proceedings of SPIE 10169 (Nondestructive Characterization and Monitoring of Advanced Materials, Aerospace, and Civil Infrastructure 2017), 1016929. doi:10.1117/12.2260638, https://doi.org/10.1117/12.2260638
  6. Reilly J, Glisic B (2018). Identifying time periods of minimal thermal gradient for temperature-driven structural health monitoring. Sensors 18(3):734. doi:10.3390/s18030734, https://doi.org/10.3390/s18030734
  7. Reilly J, Glisic B (2019). Thermal behavior of a structure characterized through three-dimensional temperature signatures for a temperature driven method of structural health monitoring. Structural Health Monitoring 2019 (IWSHM proceedings). doi:10.12783/shm2019/32140, https://doi.org/10.12783/shm2019/32140
  8. Murphy B, Yarnold M (2018). Temperature-driven structural identification of a steel girder bridge with an integral abutment. Engineering Structures 155:209-221. doi:10.1016/j.engstruct.2017.10.074, https://doi.org/10.1016/j.engstruct.2017.10.074
  9. Han Q, Ma Q, Xu J, Liu M (2021). Structural health monitoring research under varying temperature condition: a review. Journal of Civil Structural Health Monitoring 11(1):149-173. doi:10.1007/s13349-020-00444-x, https://doi.org/10.1007/s13349-020-00444-x
  10. Rizzo P, Enshaeian A (2021). Challenges in bridge health monitoring: A review. Sensors 21(13):4336. doi:10.3390/s21134336, https://doi.org/10.3390/s21134336
  11. Fu W, Sun B, Wan H-P, Luo Y, Zhao W (2022). A Gaussian processes-based approach for damage detection of concrete structure using temperature-induced strain. Engineering Structures 268:114740. doi:10.1016/j.engstruct.2022.114740, https://doi.org/10.1016/j.engstruct.2022.114740
  12. Xu J, Liu M, Ma Q, Han Q (2022). Temperature-based anomaly diagnosis of truss structure using Markov chain-Monte Carlo method. Journal of Civil Structural Health Monitoring 12(3):705-724. doi:10.1007/s13349-022-00572-6, https://doi.org/10.1007/s13349-022-00572-6
  13. Gong X, Song X, Cai CS, Li G, Xiong W (2024). A temperature-driven approach for quantitative assessment of strengthening effect of continuous bridges using structural health monitoring data. Structural Health Monitoring 23(2):1053-1070. doi:10.1177/14759217231181882, https://doi.org/10.1177/14759217231181882
  14. Glashier T, Kromanis R, Buchanan C (2024). Temperature-based measurement interpretation of the MX3D Bridge. Engineering Structures 305:116736. doi:10.1016/j.engstruct.2023.116736, https://doi.org/10.1016/j.engstruct.2023.116736
  15. Zhu J, Wang Z, Zhang Y, Wang Y (2025). A knowledge foundation and prospect review on thermal response of bridges. Advances in Structural Engineering 28(15):2871-2892. doi:10.1177/13694332251365970, https://doi.org/10.1177/13694332251365970