The objective of this study is to improve the applicability and efficiency of the gas engine driven heat pump with energy storage system when supplying heating and cooling to the different types of buildings. The physical models of four types of buildings, including residential, hotel, office and university building, were sampled and simulated by the DesignBuilder software using the weather data of typical weather year in Tianjin. The building cooling and heating load characteristics of the typical air-conditioning day (17/07/2016) and the typical heating day (07/01/2017) was chosen as the research data. It was found that the four types of buildings had different load demand due to the different function and characteristics of the buildings. The control strategy of the gas engine driven heat pump with energy storage system with four modes, such as Mode L, Mode P, Mode F and Mode O, was developed to meet the various building heating and cooling load demand. The simulation results showed that the annual primary energy ratio of the gas engine driven heat pump with energy storage system is about 21.4% (residential), 35.2% (hotel), 23% (office)and 26.6% (university) higher than traditional gas engine heat pump system. And a compared experiment was given, which demonstrated that the simulation relative errors of the university building cooling load, the cooling/heating capacity of the gas engine driven heat pump with energy storage system, fuel-consumption and primary energy ratio were lower than 12%, except the heating load with the relative error of 35.6%. From these results, the gas engine driven heat pump with energy storage system could be used in different types of buildings with a more stable operation characteristic, higher primary energy ratio and more suitable performance. The results presented in this paper is of some reference basis for optimizing the gas engine driven heat pump with energy storage system and developing a new route for raising cooling/heating quality, energy saving and consumption reducing. (C) 2018 Elsevier B.V. All rights reserved.