1. Automatic Design System Using Artificial Intelligence to Design Piping, Air Conditioning Ducts, and Cable Trays
Due to current reductions in the cost of renewable energy, nuclear power generation is faced with a demand for improvements in various areas of competitiveness. In addition to the approximately 100 km of piping used for each advanced boiling water reactor (ABWR) plant, this automatic design system using artificial intelligence (AI) was also developed to support high-speed route plans and mutual arrangement adjustment for air conditioning ducts and cable trays. By utilizing this system, it is possible to shorten processes, reduce amounts of materials, and dramatically reduce the cost of plant construction. This system uses a hybrid method that combines a route search program employing genetic algorithms with deep learning in order to learn the three-dimensional computer-aided design (3D-CAD) data of existing plants, thereby enabling the design of piping routes while taking advantage of previous design know-how.
Another one of this system’s features is its ability to plan piping routes while giving consideration to thermal expansion, vibrations, and the fluid phenomena that previous design processes had particular trouble dealing with, such as turbulence and swirling flow. As described in the example of the article “[3] Design Rationalization through the Application of Measurement Accuracy Evaluation Technology Using Fluid Analysis within Piping” below, by teaching this system with the input conditions (fluid conditions and route shapes) and output results (swirl intensity) analyzed using this technology, it is now possible to create piping route plans that reduce the occurrence of flow rate measurement flaws caused by swirling flow.
(Hitachi-GE Nuclear Energy, Ltd.)
[01]Result of piping route plan created by automatic design system