A prototype artificial intelligence driven marine propulsor design tool

This paper describes the implementation of a prototype Artificial Intelligence (A.I.)/Numerical Optimization (N.O.) system for marine propeller preliminary design. Engineers from CDNSWC and GE-CRD collaborated on the system implementation. Existing CDNSWC propulsor preliminary design software was integrated with the GE-CRD program ENGINEOUS, a general purpose A.I.DJ.0. shell which has helped GE engineers produce improved product designs at significant cost savings. Propeller lifting line, blade stress, cavitation inception, and unsteady force prediction codes were included in the A.I.-driven system, called PADS (Propeller Automated Design System). Propeller diameter; operating rpm; and pitch, chord, thickness, and skew distributions were among the design parameters automatically varied by PADS. Expert knowledge from the CDNSWC propeller design community was also input to PADS in the form of Knowledge Bases (K.B.s), which are accessed by the system for guidance on varying the design parameters. The K.B.s are supplemented by the N.O. algorithms to maximize further propeller performance. A patrol craft propeller was used as a test case for the automated design system. PADS was able to meet the performance requirements of the propeller within a few hours of computer time. Implementing and exercising the prototype system made apparent the significant potential benefits of its long-term use, which include: increasing the number of propeller alternatives that may be evaluated during the design process by orders of magnitude; reducing preliminary design times from months to weeks; and retaining and supplementing expert knowledge associated with propeller design in the integrated system. Plans for future PADS applications include blade section shape design and advanced propulsor concept design. Also, PADS may be applied potentially to industrial and commercial propeller design problems.