The ORC (organic Rankine cycle) is an established technology for converting low temperature heat to electricity. Knowing that most of the commercially available ORCs are of the subcritical type, there is potential for improvement by implementing new cycle architectures. The cycles under consideration are: the SCORC (subcritical ORC), the TCORC (transcritical ORC) and the PEORC (partial evaporation ORC). Care is taken to develop an optimization strategy considering various boundary conditions. The analysis and comparison is based on an exergy approach. Initially 67 possible working fluids are investigated. In successive stages design constraints are added. First, only environmentally friendly working fluids are retained. Next, the turbine outlet is constrained to a superheated state. Finally, the heat carrier exit temperature is restricted and addition of a recuperator is considered. Regression models with low computational cost are provided to quickly evaluate each design implications. The results indicate that the PEORC clearly outperforms the TCORC by up to 25.6% in second law efficiency, while the TCORC outperforms the SCORC by up to 10.8%. For high waste heat carrier inlet temperatures the performance gain becomes small. Additionally, a high performing environmentally friendly working fluid for the TCORC is missing at low heat carrier temperatures (100 degrees C). (C) 2015 Elsevier Ltd. All rights reserved.