Author Identifier

Hossein Mirzaaghabeik: http://orcid.org/0000-0002-7068-6015

Date of Award

2025

Document Type

Thesis - ECU Access Only

Publisher

Edith Cowan University

Degree Name

Doctor of Philosophy

School

School of Engineering

First Supervisor

Sanjay Kumar Shukla

Second Supervisor

Nuha S. Mashaan

Abstract

Ultra-high-performance concrete (UHPC) is a highly advanced composite material recognized for its exceptional mechanical properties, including high compressive strength and ductility. UHPC deep beams (UHPC-DBs) are structural elements well-suited for short spans, transfer girders, pile caps, offshore platforms, and bridge applications, particularly where heavy loads are involved. Key factors influencing the shear behaviour of UHPC-DBs include UHPC compressive strength, vertical web reinforcement (ρsv), horizontal web reinforcement (ρsh), longitudinal reinforcement (ρs), shear span-to depth ratio (λ), fibre type, fibre content (FC), and geometrical dimensions. This research begins with a comprehensive literature review to evaluate the factors affecting the shear performance of UHPC-DBs. The review aimed to identify research gaps and deepen understanding of the influence of these variables. The findings were systematically analysed and categorized to highlight the impact and trends associated with each parameter. The results indicate that increasing compressive strength, FC, ρsv, ρs, and ρsh can enhance the shear capacity of UHPC-DBs by up to 63.36%, 63.24%, 38.14%, 19.02%, and 38.14%, respectively. Furthermore, reducing λ by 61.29% led to a maximum shear capacity increase of 49.29%.

In the next phase, a predictive modelling approach was developed using a novel hybrid algorithm combining an artificial neural network (ANN) with an adaptive neuro-fuzzy inference system (ANFIS) to estimate the shear capacity of UHPC deep beams. Initially, the ANN and ANFIS models were trained separately using available experimental data. A hybrid ANN–ANFIS model was then proposed to enhance prediction accuracy using numerical input data. Model performance was evaluated using R² and RMSE metrics. The R² values obtained were 0.95, 0.99, and 0.90 for ANN, ANFIS, and the hybrid ANN–ANFIS model, respectively, demonstrating the robustness of the hybrid model, even without being trained on experimental data. Compared to the models trained with experimental results, ANFIS and ANN, the ANN–ANFIS model achieved accuracies of 90.90% and 94.74%, respectively, validating its practical applicability. Finally, the predicted shear capacities from the ANN, ANFIS, and ANN ANFIS models were compared with ACI 318-19 design code values. A novel reliability factor was proposed to safely estimate the shear capacity of fibre-reinforced UHPC deep beams with a safety margin of 0.66, enabling potential application in structural design.

Given the known benefits of fibre reinforcement, this study also addresses the concerns related to corrosion when steel fibres are used in corrosive environments. The literature review indicated that non metallic fibres, such as synthetic fibres, provide a corrosion-resistant alternative to traditional steel fibres. Although prior research has extensively examined the impact of steel fibres on the shear performance of UHPC deep beams (UHPC-DBs), limited attention has been given to the role of synthetic fibres in UHPC applications. To address this gap, a numerical study was conducted to compare the effects of 5D steel fibres, hooked-end (HE) steel fibres, and Forta-Ferro (FF) synthetic fibres on the shear performance of UHPC-DBs, considering the influence of geometrical dimensions. Finite element analysis (FEA) was conducted in ABAQUS using the concrete damage plasticity (CDP) model, with results validated against experimental data from four tested specimens. Key parameters assessed included fibre type, shear span-to-depth ratio (λ), length-to-depth ratio (l/d), and width-to-depth ratio (b/d). Design-oriented results indicated that the safe zone for maximizing shear capacity corresponds to b/d ratios between 0.19 and 0.25 and l/d ratios between 2.0 and 3.0. Conversely, the risky zone, where shear capacity is minimized, lies between b/d = 0.3–0.375 and l/d = 2.25–2.75. These insights, currently absent from ACI 318-19, could inform future design standards. Moreover, beams reinforced with 0.11% FF synthetic fibres achieved 85.97% of the shear capacity of beams with 0.76% 5D steel fibres, highlighting FF fibres as a viable and corrosion-resistant alternative.

A further investigation examined the influence of vertical web reinforcement (ρsv) on the shear performance of UHPC deep beams (UHPC-DBs) reinforced with 2% HE steel fibres, 0.76% 5D steel fibres, and 0.11% FF synthetic fibres. The study evaluated the effects of different ρsv levels, fibre types, and shear span-to-depth ratios (λ) on the shear capacity of UHPC deep beams by analysing load deflection behaviour, crack patterns, and reinforcement strain. Using the particle swarm optimization (PSO) algorithm, the optimal ρsv was identified. Shear capacity improvements of 12.42%, 15%, and 16.46% were achieved for beams reinforced with HE, 5D, and FF fibres, respectively, as ρsv increased from 0% to 0.38%. Vertical reinforcement strains increased notably with higher ρsv, confirming its significant contribution to improved stiffness and ductility. UHPC deep beams reinforced with 2% HE steel fibres demonstrated greater ductility compared to those reinforced with 0.76% 5D steel fibres. At shear span-to-depth ratios (λ) of 0.923 and 0.739, all specimens exhibited shear-compression failure modes, except for those reinforced with 2% HE fibres combined with higher ρsv values, which failed in diagonal tension. Crack widths were reduced with increased ρsv, and 5D-reinforced beams had narrower cracks than those reinforced with FF. Three-dimensional surface plots were created to illustrate the combined effects of λ and ρsv, offering tools to guide future design and revisions to ACI 318-19. Notably, an optimal ρsv value of 0.19% achieved 90.47–98.42% of the shear capacity of UHPC deep beams reinforced with 0.38% ρsv, while reducing reinforcement costs by 46%.

The next stage focused on the combined effects of longitudinal reinforcement (ρs) and horizontal web reinforcement (ρsh) on the shear performance of UHPC deep beams. The same three fibre types, HE steel, 5D steel, and FF synthetic, were studied using validated finite element models in ABAQUS with the CDP model. The interaction of ρs and ρsh on shear capacity, load–deflection behaviour, crack formation, and reinforcement strain was thoroughly assessed.

The PSO algorithm was again employed to determine the optimal reinforcement values: longitudinal reinforcement (ρs) at 3.21% for three fibre types, and horizontal web reinforcement (ρsh) at 0.16% for HE and 5D fibres, and 0.19% for FF fibres. These values maximized shear performance while minimizing cost. The PSO algorithm was used again to determine optimal reinforcement values: longitudinal reinforcement (ρs) at 3.21% for all three fibre types, and horizontal web reinforcement (ρsh) at 0.16% for HE and 5D fibres, and 0.19% for FF fibres. These values maximize shear performance while minimizing cost. Final assessments included shear ductility, defined as the area under the load deflection curve to collapse (Ac) over that to ultimate load (Au), and energy absorption, providing vital data for refining current design guidelines.

Finally, fibre content (FC) emerged as a crucial factor influencing the shear performance of UHPC deep beams (UHPC-DBs). Although several studies have examined the impact of individual fibre types, particularly steel fibres, on the shear capacity of UHPC-DBs, a significant gap remains regarding the role of synthetic fibres and hybrid fibre systems, especially combinations of steel and synthetic fibres. This final phase evaluated the shear performance of UHPC-DBs with varying contents of 5D steel fibres and FF synthetic fibres, both individually and in hybrid configurations. The CDP model in ABAQUS was used for FEA, validated against five tested beams. Key outputs included shear capacity, mid-span deflection, crack patterns, fibre stress, and ductility. PSO was applied to determine the optimal fibre content, which was found to be 1.0% 5D steel fibres, 0.11% FF synthetic fibres, and a hybrid combination of 1.0% 5D steel with 0.11% FF synthetic fibres. The results provide practical insights for optimizing hybrid fibre designs, including synthetic fibres, an area currently unaddressed by ACI 318 19, highlighting the need to update existing standards to reflect modern material systems.

Access Note

Access to this thesis is embargoed until 18th December 2030

DOI

10.25958/d1sy-3046

Download

Available for download on Wednesday, December 18, 2030

Share

 
COinS