Numerical modelling of unbonded post tensioned concrete structures in fire including explicit modelling of creep in prestressing steel tendons
Lee, James Alistair
Due to the unbonded nature of tendons to the slab within Unbonded Post Tensioned (UPT) concrete structures, tendon stress relaxation under heating affects all regions of the slab spanned by the tendon; not just in the locality of the fire. The numerical modelling of bonded and unbonded post tensioned concrete structures in fire has been performed to some degree, notably by Bailey and Ellobody. The consideration of elevated temperature creep to the relaxation of tendon prestress however, has not been considered. This thesis attempts to incorporate a uniaxial creep strain rate function of stress and temperature into the commercial FE software package Abaqus, compatible for use within the in-built multiaxial metal plasticity constitutive framework. What follows is a validation study of the Harmathy’s uniaxial creep strain accumulation function via the modelling of stress relaxation in isolated, tensioned and heated prestressing steel tendons, against experimental data. From here, UPT concrete slab models are analysed whilst exposed to a standard fire temperature-time curve and subsequently allowed to cool. Tendon prestress relaxation and resulting UPT concrete slab deflection is compared, where tendon creep is explicitly modelled, as opposed to implicitly covered by Eurocode 2 determined temperature dependent stress-strain curves. Following this, a large scale continuous one-way spanning UPT concrete structural model is developed to consider global structural behaviour resulting from localised fire, where realistic boundary conditions such as beam rotation and deflection are permitted. The ignorance of explicit elevated temperature creep consideration, in prestressing steel tendons, is commonly justified through the implicit accountability stated within Eurocode 2 temperature dependent stress-strain curves. This however is not completely true; Eurocode 2 states implicit accountability only holds should the tendon be heating at a rate within the bounds of 2⁰C/min to 50⁰C/min. Where only heating of a UPT concrete slab is considered, evidence from this thesis suggests Eurocode 2 determined stress-strain curves can implicitly account for accumulated creep strain up to limited temperatures. Prestressing steel tendons are however embedded within a concrete slab through which thermal gradients build up during fire. This means heat transfer can continue to the tendon, increasing its temperature postfire at an ever decreasing rate until it reaches its peak. Should post-fire cooling behaviour not be considered, continued tendon heating and subsequent creep strain accumulation will be ignored. Further, during the transition from heating to cooling within the tendon, it will be exposed to elevated temperatures with a rate of change below 2⁰C/min, whereby Eurocode 2, as stated cannot implicitly account for creep. It is shown, a significant degree of subsequent relaxation of prestress, UPT concrete slab deflection and concrete damage in hogging can occur during this phase of postfire behaviour, where the tendon temperature peaks during its transition from heating to cooling. In order to justify non consideration of creep, it should be shown tendon temperature will remain suitably low throughout the entire heating-cooling regime to which the UPT concrete slab is exposed. This must be achieved through adequate specification of minimum concrete cover to tendons to limit tendon temperature exposure for a given parametric fire curve duration, including the potential continued rise post-fire. Evidence within this thesis identifies 350⁰C as a critical temperature whereby the explicit consideration of tendon creep does not significantly increase predicted prestress relaxation and subsequent UPT concrete slab deformation, compared to implicit creep consideration from Eurocode 2. The manufacturing standard to which prestressing steel tendon strands are produced has been shown experimentally by Gales to significantly influence their susceptibility to elevated temperature creep. This is reflected by Gales determining differing creep parameters as a function of stress for incorporation in Harmathy’s uniaxial creep strain function. Modelled prestress relaxation of isolated, tensioned and heated tendons within this thesis is therefore significantly reduced when tendons are manufactured to a yield stress of 1860MPa according to the BS 5896 standard, as opposed to the ASTM A416 standard. As a result Eurocode 2 determined stress-strain curves implicitly account for accumulated creep strain during heating, at 10⁰C per minute, up to approximately 400⁰C for grade 1860 ASTM A416 manufactured tendons and 500⁰C for grade 1860 BS 5896 standard tendons. The aforementioned critical temperature of 350⁰C does not in actuality apply to necessary explicit creep consideration for UPT concrete slabs modelled with grade 1860 BS 5896 standard tendons. This temperature however remains a design temperature limit, owing to the potential onset of microstructural recrystallization beyond 400⁰C and the associated degradation of mechanical properties that coincides. The reasons for such differing elevated temperature creep and stress relaxation behaviour between the two manufacturing standards of prestressing steel wires and strands has been postulated within this thesis to be due to differing chemical compositions. This relates specifically to large relative differences of phosphorus and sulphur found in wires manufactures to each standard as tested by Gales.
The following license files are associated with this item: