Multivariable coupled prediction model for particulate matter concentration in the dynamic environment of a biologically clean room

To ensure the cleanliness of a biologically clean room, the air change rate applied by an air conditioning system is usually high, resulting in high energy consumption during its operation. One of the important reasons is that the change law of particle concentration for a clean room is not clear, and the dynamic adjustment of the air volume cannot be performed in the air conditioning system according to the production load. In this study, a biological clean room with a C-class environment was constructed to simulate a pharmaceutical production plant. Under an operating air change rate of 10, 20, 30, and 40 h(-1), 36 groups of particulate matter concentrations corresponding to stationary, mild, and severe activity types and 1-3 people were tested for the frequency conversion adjustment of the air conditioning system. A comparison of the test and standard limit values shows that when the biological clean room runs at 40 h(-1) ventilation times, the test value of the 0.5 mu m particulate concentration is 1/718 of the standard limit, indicating high redundancy in cleanliness. The air conditioning system has great energy-saving potential. The clean redundancy observed for 5 mu m particles is low when the air conditioning system of the biological clean room runs at low air volume. The ratio of 0.5 and 5 mu m particle concentrations to the standard limit should be used as a criterion for determining the air change rate of the air conditioning system. The analysis of the experimental data provided the number of personnel, activity types, air change rate, and the change rule of the clean room particulate matter concentration. The proportional relationship between the personnel growth and particulate matter concentration growth rates under an air change rate of 10-40 h(-1) was given, offering a more accurate reference value for particulate matter concentration in the design stage. Based on the concentration values of 0.5 and 5 mu m particles, the relationship between the concentration of particles in the light and heavy-activity types and that in the stationary activity type was summarized for a group of 1-3 people. Compared with previous studies, which suggested that the concentration of particles in the light and heavy-activity types was 2-5 and 5-10 multiple that in the stationary type, respectively, the data in this study are of more practical reference significance. Based on the measured data, a multiple regression method was adopted to establish two multivariable coupled prediction models for determining particulate matter concentration. Model 1 is a prediction model without any interactions between the variables, while Model 2 is a prediction model that considers the interaction between the variables. In terms of R-2, Model 2 has a higher degree of fitting. Compared with Model 1, Model 2 is used to predict the concentration of particulate matter under different production conditions, which is closer to the real situation. Thus, the air change rate required to maintain the cleanliness of a biologically clean room can be obtained using this prediction model, providing data reference for developing an energy-saving operation strategy for the air conditioning system in such rooms.