This study presents a methodology based on process integration techniques to improve the energy efficiency of a large-scale chemical plant. The key to the approach is to represent the energy requirements with different heat transfer interfaces. Considering difficulties of data extraction for a large-scale plant, a multi-level data extraction scheme is introduced based on different heat transfer interfaces and includes five levels of growing complexity: black-box, grey-box, white-box, simple-model and detailed-model analysis. A combination of these levels instead of a single definition for the energy requirement has been applied on an industrial case study. Different steps of the approach are explained in detail and their potential are highlighted. The Single Process Integration (SPI) and Total Site Integration (TSI) has been performed and revealed that a higher potential of heat recovery could be driven through the TSI. The optimized site utility integration together with heat recovery improvement scenarios have considerably increased the energy saving potential in our case study. A multi-objective optimization has also been performed to find the optimum combination of units with different energy requirement levels. In conclusion, results from our case study have indicated that using a combination of different energy requirement levels will reduce the required modification of the actual site configuration. (C) 2014 Elsevier Ltd. All rights reserved.