Lightweight mechanical energy-storage devices or springs with nonlinear strain-energy absorption rate are important building blocks for passive/quasi-passive rehabilitation robotics. They provide support and controllable energy storage/release into the system thereby making daily activities such as walking/running metabolically efficient for the disabled. These devices have stringent footprint constraints and must withstand 10 million cycles of loading for successful implementation on an orthotic device. Currently, there are no off-the-shelf springs or a systematic synthesis methodology that can meet these requirements in a deterministic fashion. In this paper, we demonstrate how existing body of knowledge in compliant mechanisms can be systematically leveraged to design spring geometries with distributed compliance that meet fatigue criteria and weight requirements. Towards this, we implement a strength-based approach to determine feasible initial solutions that upon optimization yield geometries with maximally distributed stresses. Such a framework is general and can be adapted for designing any compliant mechanism.