Objective To explore the application of improved clustering methods in the site selection concern of the emergency material reserve of chemical enterprises.

Methods Taking the chemical enterprises planned to be constructed in a certain area as the research object, the layout of emergency material reserves within the plant area was determined according to the improved RK-medoids method based on the existing clustering method (K-medoids method clustering). The results derived from these two methods were compared.

Results A total of 18 major risks were identified in the plant area, including dust and chemical toxins, mechanical injuries, abnormal high and low temperatures, etc. There were 27 specific risk points, with the highest risk level as Level 4. The "elbow rule" determined that the number of clusters for the risk points was three. Compared with the K-medoids method, the three risk point cluster centers calculated by the RK-medoids method were slightly different. The total distance from each risk point to the three emergency material reserve warehouses decreased from 16.975 km to 16.083 km with a reduction of 5.2%. The total distance from high-risk points to the three emergency material reserve warehouses decreased from 9.463 km to 8.970 km with a reduction of 5.2%, which was beneficial for rapid emergency rescue after an accident.

Conclusions The RK-medoids algorithm is simple to operate and has a fast calculation speed. It is suitable for the design stage of new chemical enterprises and can provide a reference for the site selection of emergency material reserve warehouses.