There have been very few advanced control algorithms developed and deployed for small/medium commercial buildings due to practical difficulties such as the lack of Building Management System (BMS), significant disturbances, high sensor costs and high cost of site-specific engineering solutions. These high implementation costs have been a major impediment to successful market penetration. The focus of this work is to develop and demonstrate a practical control algorithm for the coordination of multiple roof top units (RTUs). The goal is to minimize sensor and configuration requirements in order to enable a more cost effective control implementation for small/medium commercial building applications. The algorithm characterizes the responses of the thermostat sensors to RTU cycling and uses a model within an optimization algorithm to determine the best combination of RTUs to operate in order to minimize power and maintain comfort conditions over a short prediction horizon. The algorithm was first tested in simulation for an existing sit-down restaurant and then at a field site that is a gymnasium. The estimated energy savings were greater than 20% for the restaurant and about 8% for the gym. Peak demand savings of about 40% were demonstrated for the gym. The algorithm is termed "plug-and-play" because no additional sensors are needed for implementation and the thermostat response models and optimization require minimal configuration. (C) 2015 Elsevier B.V. All rights reserved.