The conceptual design of a petroleum refinery that satisfies multiple economics and operating constraints is a highly complex task. Coupled with the ever-rising cost of designing and constructing a new refinery and the increasing demand for energy and fuels, there is incentive to optimize the energy recovery and energy efficiency of such a facility. This work addresses the flowsheet optimization of the synthesis of a petroleum refinery to attain an optimal heat-integrated configuration or topology. A sequential two-step strategy is employed that first performs simultaneous flowsheet optimization and heat integration to obtain an optimal refinery topology with minimum utility cost. Subsequently, the fixed optimal topology with minimum utility loads is optimized to arrive at a configuration with the fewest heat exchanger units. A mixed-integer linear program (MILP) is formulated based on a superstructure representation that considers many alternative feasible refinery topologies. The computational results show meaningful reduction in the total annualized capital and operating costs as compared to a non-heat-integrated configuration.