I. Shakeri
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
1 records found
1
Master thesis
(2026)
-
N.W. Brightman, Dr. Florentia Kavoura, F.P. van der Meer, F. Messali, R.P.H. Vergoossen, I. Shakeri
This thesis develops a numerical framework for predicting the high-cycle fatigue life of demountable steel shear connectors in timber-concrete composite (TCC) road bridges. Demountable TCC decks cut self-weight and embodied carbon relative to conventional systems, but their durability under millions of traffic cycles is the open question, and the steel connector — a dowel bearing on timber through a concrete notch — is its fatigue-critical link.
The methods in routine use, from the Eurocode S-N detail categories to commercial fatigue tools, read life from an empirical stress-life curve: they neither resolve the stress that drives the damage at the thread root nor transfer to an untested geometry. Resolving every one of the 10^5–10^7 traffic cycles directly is, in turn, computationally infeasible.
The framework developed here, “2SCJ”, pairs Lemaitre two-scale continuum damage mechanics on the dowel with a cycle-jump scheme around Abaqus/Standard, advancing a per-element damage field from intact material through mesocrack initiation to a propagating crack in adaptive blocks of cycles. The per-cycle damage rate is read from converged solver output and never fed back, so the scheme runs as an external loop around a sealed commercial code, with no user subroutine or remeshing.
Reading the mesoscopic stress as a centroid average over an element sized at the steel representative volume element supplies the notch-support effect from the material card (the full parameter set) alone, without a notch-specific constant. The framework is verified against closed-form solutions and validated blind against measured coupon data: on a circumferentially notched round bar it reproduces two notch severities to within a factor of two at every amplitude, from a single material card carried over untouched, and a notched plate then calibrates the one mean-stress parameter that fully-reversed data cannot inform.
Applied to the M12 notch-stud push-out specimen of Zheng et al. (2025), it places first mesocrack nucleation at the first engaged thread root, matching both the experimental fracture surface and an independent Smith–Watson–Topper critical-plane cross-check, and returns a per-element nucleation life inside the experimental scatter band at the closest load level. A sensitivity study over the model inputs ranks their influence on the predicted life and singles out the substrate-driven thread-root stress amplitude, not any steel property, as its dominant source of uncertainty.
Because the dowel steel’s own fatigue data were unavailable, this application borrows a material card from the literature and is a demonstration rather than a calibrated prediction: the S-N slope and endurance level do not match the experiment, a shortfall that traces to those borrowed inputs. The dependable result is the geometry-set failure location, and the framework stands as a verified, mechanism-based predictor of where and roughly when a connector fails and which design choices move its life — a virtual test bench for the geometries and loadings that are slow or costly to test physically. ...
The methods in routine use, from the Eurocode S-N detail categories to commercial fatigue tools, read life from an empirical stress-life curve: they neither resolve the stress that drives the damage at the thread root nor transfer to an untested geometry. Resolving every one of the 10^5–10^7 traffic cycles directly is, in turn, computationally infeasible.
The framework developed here, “2SCJ”, pairs Lemaitre two-scale continuum damage mechanics on the dowel with a cycle-jump scheme around Abaqus/Standard, advancing a per-element damage field from intact material through mesocrack initiation to a propagating crack in adaptive blocks of cycles. The per-cycle damage rate is read from converged solver output and never fed back, so the scheme runs as an external loop around a sealed commercial code, with no user subroutine or remeshing.
Reading the mesoscopic stress as a centroid average over an element sized at the steel representative volume element supplies the notch-support effect from the material card (the full parameter set) alone, without a notch-specific constant. The framework is verified against closed-form solutions and validated blind against measured coupon data: on a circumferentially notched round bar it reproduces two notch severities to within a factor of two at every amplitude, from a single material card carried over untouched, and a notched plate then calibrates the one mean-stress parameter that fully-reversed data cannot inform.
Applied to the M12 notch-stud push-out specimen of Zheng et al. (2025), it places first mesocrack nucleation at the first engaged thread root, matching both the experimental fracture surface and an independent Smith–Watson–Topper critical-plane cross-check, and returns a per-element nucleation life inside the experimental scatter band at the closest load level. A sensitivity study over the model inputs ranks their influence on the predicted life and singles out the substrate-driven thread-root stress amplitude, not any steel property, as its dominant source of uncertainty.
Because the dowel steel’s own fatigue data were unavailable, this application borrows a material card from the literature and is a demonstration rather than a calibrated prediction: the S-N slope and endurance level do not match the experiment, a shortfall that traces to those borrowed inputs. The dependable result is the geometry-set failure location, and the framework stands as a verified, mechanism-based predictor of where and roughly when a connector fails and which design choices move its life — a virtual test bench for the geometries and loadings that are slow or costly to test physically. ...
This thesis develops a numerical framework for predicting the high-cycle fatigue life of demountable steel shear connectors in timber-concrete composite (TCC) road bridges. Demountable TCC decks cut self-weight and embodied carbon relative to conventional systems, but their durability under millions of traffic cycles is the open question, and the steel connector — a dowel bearing on timber through a concrete notch — is its fatigue-critical link.
The methods in routine use, from the Eurocode S-N detail categories to commercial fatigue tools, read life from an empirical stress-life curve: they neither resolve the stress that drives the damage at the thread root nor transfer to an untested geometry. Resolving every one of the 10^5–10^7 traffic cycles directly is, in turn, computationally infeasible.
The framework developed here, “2SCJ”, pairs Lemaitre two-scale continuum damage mechanics on the dowel with a cycle-jump scheme around Abaqus/Standard, advancing a per-element damage field from intact material through mesocrack initiation to a propagating crack in adaptive blocks of cycles. The per-cycle damage rate is read from converged solver output and never fed back, so the scheme runs as an external loop around a sealed commercial code, with no user subroutine or remeshing.
Reading the mesoscopic stress as a centroid average over an element sized at the steel representative volume element supplies the notch-support effect from the material card (the full parameter set) alone, without a notch-specific constant. The framework is verified against closed-form solutions and validated blind against measured coupon data: on a circumferentially notched round bar it reproduces two notch severities to within a factor of two at every amplitude, from a single material card carried over untouched, and a notched plate then calibrates the one mean-stress parameter that fully-reversed data cannot inform.
Applied to the M12 notch-stud push-out specimen of Zheng et al. (2025), it places first mesocrack nucleation at the first engaged thread root, matching both the experimental fracture surface and an independent Smith–Watson–Topper critical-plane cross-check, and returns a per-element nucleation life inside the experimental scatter band at the closest load level. A sensitivity study over the model inputs ranks their influence on the predicted life and singles out the substrate-driven thread-root stress amplitude, not any steel property, as its dominant source of uncertainty.
Because the dowel steel’s own fatigue data were unavailable, this application borrows a material card from the literature and is a demonstration rather than a calibrated prediction: the S-N slope and endurance level do not match the experiment, a shortfall that traces to those borrowed inputs. The dependable result is the geometry-set failure location, and the framework stands as a verified, mechanism-based predictor of where and roughly when a connector fails and which design choices move its life — a virtual test bench for the geometries and loadings that are slow or costly to test physically.
The methods in routine use, from the Eurocode S-N detail categories to commercial fatigue tools, read life from an empirical stress-life curve: they neither resolve the stress that drives the damage at the thread root nor transfer to an untested geometry. Resolving every one of the 10^5–10^7 traffic cycles directly is, in turn, computationally infeasible.
The framework developed here, “2SCJ”, pairs Lemaitre two-scale continuum damage mechanics on the dowel with a cycle-jump scheme around Abaqus/Standard, advancing a per-element damage field from intact material through mesocrack initiation to a propagating crack in adaptive blocks of cycles. The per-cycle damage rate is read from converged solver output and never fed back, so the scheme runs as an external loop around a sealed commercial code, with no user subroutine or remeshing.
Reading the mesoscopic stress as a centroid average over an element sized at the steel representative volume element supplies the notch-support effect from the material card (the full parameter set) alone, without a notch-specific constant. The framework is verified against closed-form solutions and validated blind against measured coupon data: on a circumferentially notched round bar it reproduces two notch severities to within a factor of two at every amplitude, from a single material card carried over untouched, and a notched plate then calibrates the one mean-stress parameter that fully-reversed data cannot inform.
Applied to the M12 notch-stud push-out specimen of Zheng et al. (2025), it places first mesocrack nucleation at the first engaged thread root, matching both the experimental fracture surface and an independent Smith–Watson–Topper critical-plane cross-check, and returns a per-element nucleation life inside the experimental scatter band at the closest load level. A sensitivity study over the model inputs ranks their influence on the predicted life and singles out the substrate-driven thread-root stress amplitude, not any steel property, as its dominant source of uncertainty.
Because the dowel steel’s own fatigue data were unavailable, this application borrows a material card from the literature and is a demonstration rather than a calibrated prediction: the S-N slope and endurance level do not match the experiment, a shortfall that traces to those borrowed inputs. The dependable result is the geometry-set failure location, and the framework stands as a verified, mechanism-based predictor of where and roughly when a connector fails and which design choices move its life — a virtual test bench for the geometries and loadings that are slow or costly to test physically.