Complex engineering systems, such as power plants, deliver their best performance when operating along a designed range of some priority parameters. However, plant field operation may deviate from design conditions, and new references must be identified. Actions towards high-quality operation can be supported by fine modeling, which helps building decision support tools. The present work proposes a standardization strategy for the operation of an actual coal-fired power plant based on a Design of Experiment approach, partially tested onsite and finally accomplished with surrogate models built upon a 2 year long database. Artificial Neural Networks (ANNs) and Mass and Energy balances (M&Es) are used to represent the plant’s steam generator and its mills subset, which is the core of an operational guide to increase system efficiency under actual operation. Primary and secondary air flows, pulverized coal outlet temperature, speed of the dynamic classifier, primary air flow, excess O, primary and secondary air pressures are the seven controllable factors selected as the most relevant ones among an extensive set of parameters, able to perform effective maneuvers. The application of the operational guide indicates combinations of ranges of the seven controllable parameters that allow for achieving steam generator efficiency within the 84.0% to 88.92% range. The proposed methodology aims as well to improve safe and stable conditions to a system that undergoes operation different than the one prescribed by the original design. The study case results show an opportunity to raise efficiency by up to 2.28% during operation, which represents a reduction in coal consumption by 3.1 t/h and above 6% on CO emissions.