This paper investigates computational optimization techniques at the urban design scale, aiming to improve the performance of urban fabric layouts according to predefined evaluation metrics. To this end, this work addresses the use of optimization tools in urban design by comparing various optimization algorithms for generating urban fabrics with improved walkability and by analyzing the outcomes of different urban design rules. These rules formulate orthogonal and non-orthogonal grids from the perspective of transit accessibility (TA), thereby minimizing automobile usage and improving the walkability of neighborhoods. Transit accessibility is also evaluated alongside estimated infrastructure cost to verify the suitability of applying optimization in urban design. Our results suggest that the RBFOpt algorithm performs best for generating urban fabrics according to our quantitative design objectives; more flexible and complex grids in terms of shape and dimensions tend to deliver greater TA than rectangular and uniform-oriented grids; different block patterns can lead to solutions more directed at TA or to infrastructure cost, outlining a trade-off; and multicriteria optimization helped in identifying designs that balanced transit accessibility and infrastructure cost.