A substructure shake table test (SSTT) based on the branch mode method was performed to reveal the mechanism and rules of equipment-adjacent structure-soil interaction (EASSI) under a seismic effect. EASSI system was divided into three substructures, namely, equipment-single structure, foundation soil, and adjacent structure. The coupling terms of interaction among the substructures were proposed.The branch mode method was effectively applied to the SSTT by decomposing and transforming the dynamic equation of the entire system and utilizing the coupling terms of interaction for data exchange among substructures. The degree of freedom was reduced for the linear substructures. Experiments indicated that in EASSI, the presence of soil magnified the flexibility and equivalent damping of the entire system.The overall effect was presented as a reduction in the dynamic response of the system. The dynamic feedback of the equipment inhibited the dynamic response of the main structure, which intensified the rate of vibration attenuation of the system.The seismic response analysis was also performed for the system when the mass ratio and frequency ratio between the equipment and the main structure and the position of the equipment in the main structure varied. KEYWORDS branch mode method, equipment-adjacent structure-soil interaction (EASSI), seismic response, shake table, soil-structure interaction, substructure test 1 | INTRODUCTION Along with the rapid social and economic development, the costs of equipment add to total costs of construction. A seismic action may cause great damage to the equipment, leading to a growing concern on antiseismic properties of equipment and pipelines. These equipments and structures are considered as a system engaged in dynamic interaction. The methods for analyzing the dynamic interactions between the equipment and the structure include the theoretical and experimental approaches. The floor response spectrum is a typical theoretical approach. In the conventional floor response spectrum, forced decoupling is performed, with the response of the main structure as the seismic input of the equipment.However, the tuning effect and interactions between the equipment and the main structure, as well as nonclassical damping and space coupling, are neglected when computing the floor response spectrum. As a result, the computational result is relatively conservative and does not satisfy the economical requirements. This method was applied to determine the design peak acceleration of the equipment located in the nonelastic structure. [1] The modified floor response spectrum considers the dynamic features of equipment and main structure as well as the interactions between ---