Cost optimization of deck girder system of post tensioned i-girder bridge for different number of lanes

Abstract

Cost optimization of deck girder system of post-tensioned I-girder bridge for different number of lanes is presented in this thesis. The main objective is to find the optimized deck girder configurations which satisfy all the safety and serviceability criteria while considering the cost of materials, fabrication installation, and labor. For a particular bridge superstructure with specific span length, lane width and footpath width, the design variables considered for this thesis are number of girder, c/c spacing of girder, various cross-sectional dimensions of the deck slab and girder, type and number of cables, applied force on cables, and ordinary reinforcement for deck slab and girder. Bounds are applied to the design variables based on code-specified guidelines and geometric requirements. Constraints for the design are applied using AASHTO LRFD Bridge Design Specifications, 2017, and geometric requirements.

The bridge design optimization is a very complex problem with lots of design variables, non-linear constraints long calculations, and lots of local minimums, making it very difficult to find the global optimum solution. So, to solve this complex problem Differential Evolution (DE) algorithm from the SciPy library is used, which is a widely used global optimization algorithm used in different fields and very easy to implement. A spreadsheet is used to calculate everything involved in design calculation and to input design variables, bounds, constraints, and objective functions. Then a computer program written in Python is used to link the spreadsheet with the DE algorithm and solve the problem.

The formulated spreadsheet and program were then used to optimize conventionally used design where the output results show around 15~20% cost reduction in optimized design from unoptimized design for 100 ft, 120 ft, 140 ft, and 160 ft span, for both 2-lane and 4-lane bridges. The optimized design also shows around 11~13% cost improvement from a 100 ft span where, a few constraints are applied to resemble the field condition (maximum height limit 0.055L for deck plus girder, maximum applicable stress on cable 0.75fpu, allowable maximum compressive stress at transfer 0.60f’ci, allowable bottom stress at service limit state -0.109 ksi (compression)). The study found that the cost per unit area for 2-lane bridges is 1.83~2.15% more than the cost per unit area for 4-lane bridges. This study also compares Ahsan et al. (2012) results, where it shows that cost reduction of 10% from existing design using DE and 21% cost increase from EVOP optimization results. The study also covered a detailed parametric study on different design variables and how every variable affects the overall superstructure design which will be very helpful for practical design.