Optimalisasi beton prategang PCI girder pada jembatan perning kloji, mojokerto menggunakan metode artificial neural network

Abstract

The Perning Kloji Bridge in Mojokerto Regency is considered unfit for use due to the displacement of one of the pillars due to the collapse of the retaining wall, making it necessary to replace the bridge. The bridge will be replaced with a Prestressed Concrete I Girder (PCI Girder). The bridge replacement has been planned by the service provider at a cost of Rp1,133,300,230.21. It requires an appropriate method to determine a more efficient cost. Existing PCI girder planning with concrete quality f'c 68.60 MPa, span 50 m, girder height 2.1 m, and a total of 8 girders. Bridge structure planning in this study only includes PCI girder calculations. The analysis process was carried out with Microsoft Excel software. By using the same load, existing modeling and 150 trials were carried out in accordance with the design requirements, including deflection, moment stress, and shear stress. Then ANN programming was continued with the input variables used, including span length, concrete quality, steel quality, strand diameter, number of strands, number of girders, and distance between girders. While the outputs used for ANN programming are girder height, moment, shear force, and cost. ANN was used to predict the optimum PCI girder planning and proved by comparing the optimization results with ANN program and existing design. ANN prediction results have met the design requirements. The existing design cost is Rp1,133,300,230.21 and after ANN programming it is Rp 905,184,683.36, 20.13% cheaper than the existing bridge. Comparison between ANN predictions and prediction results based on empirical formulas obtained an error of 7.56% and from these results it is still safe because < 10%, so it can be concluded that the results of ANN predictions are more optimal than the existing.