In many chemical plants such as LNG, refineries, air enrichment, and ammonia compression is a major consumer of energy. In such plants, some process streams may need compression, while others may need expansion. Optimal integration of these streams may yield major savings in energy. In this work, we introduce a work exchanger network synthesis problem that is analogous to the well-known heat exchange network synthesis problem. We model the details of compressor and turbine operating curves to identify high-pressure and low-pressure streams that should be matched for work exchange via compression/turbine stages located on a single shaft. We propose a superstructure for the work exchange network configuration and develop a mixed-integer nonlinear program (MINLP) to minimize the total annualized cost for a constant speed of the single shaft. We use this preliminary model to optimize shaft speed, and demonstrate the benefits of an optimized network via an illustrative case study. (C) 2011 Elsevier Ltd. All rights reserved.