Algorithm of Water Film Identification in Passive Containment Water Distribution Test

Abstract

Water distribution test designed to study the relationship between water film coverage and water flow rate and containment structure, is an important part of the passive containment cooling test. In the process of developing CAP1400 Pressurized Water Reactor (PWR), water distribution test facility was built by SNPTRD (State Nuclear Power Technology Research & Development Center). Water distribution test facility is a separate-effect test facility, based on CAP1400 containment as the prototype. Test body contains 1/8 sector of the containment dome and a certain length (3m) vertical wall, with the ratio of 1:1 of the prototype. Water film identification algorithm is used to extract water film image information from the photo of measured area obtained by Matrix CCD and to determine the effective water film region in order to calculate the water film coverage. The starting point of designing water film identification algorithm was to find out the boundaries between water film zone and no water film region based on the gray difference. In order to ensure that the difference identification can be achieved, median filtering and mean value block were conducted on gray value matrix to remove high-frequency fluctuations in the gray value curve. The algorithm was implemented with MATLAB and sensitivity studies were performed to investigate the effects of block size and difference magnitude on the identified coverage of water film. It was found out that block size and difference magnitude had little influence on the identification of water film coverage. According to the corresponding degree between identified images and original photos, 2-by-2 block and gray-difference function were selected in the algorithm. Then the program identified results were compared with artificial identified results, the results showed that: water film identification algorithm can identify water film effectively. Coverage deviation between calculated and artificial identified results is less than 5%, and nearly not affected by brightness, laying foundation for the analysis of water distribution test.