KNOWLEDGE-BASED MODELLING FOR OFFSHORE HEAVY LIFT DESIGN

Abstract

Heavy lift is an important operation in the construction of offshore structures. A large number of interdependent parameters and requirements need to be appraised by experienced engineers to arrive at an optimized heavy lift design. Proper planning and evaluation prior to an actual lift operation has a major impact on the cost-effectiveness and safety of the project. A knowledge-based heavy lift design system can serve as a design assistant for an engineer to evaluate the alternatives before the final selection of a solution based on cost and safety considerations. This research project aims to investigate into the structure and organization of knowledge-based approaches for heavy lift design, identify the appropriate strategies for problem solving, and propose, develop and implement a suitable knowledge-based modelling technique for offshore heavy lift design. The understanding of concepts for heavy lift design prior to software implementation is essential to the success of any knowledge-based system development. The interdependent design constraints and parameters related to heavy lift, such as the environmental conditions, availability of barges, types and sequences of possible lift operations, and configuration of rigging components are analyzed. Several issues related to implementing knowledge-based design in heavy lift design are investigated. These issues include a consistent and rational design model, spatial and geometric reasoning requirements, and different problem-solving approaches for handling various design aspects of heavy lift design. An object-oriented system, C-LIFT has been developed using ART*Enterprise, a toolkit for developing knowledge-based systems. The system provides the facilities to evaluate the liftability and sequence of installation of modules by a crane barge onto a platform. An object-oriented approach which facilitates the evolution of design objects and consistency of design specifications through the principles of abstraction, encapsulation, modularity and hierarchy is adopted. This approach is effective for representing object and concept hierarchy with varying levels of specifications for the appropriate extraction of design information in heavy lift design. Flexible geometric representations are important for software modelling of heavy lift design. The spatial and topological abstractions or physical entities which facilitate the definition of physical entities for geometric modelling during the preliminary design are adopted. The implementation of locational dependency hierarchy which utilizes a tree structure to represent the hierarchical assembly structure as well as the locational dependencies amongst the components of the structure, is important for top-down approach in heavy lift design. The combination of these representations forms the geometric modelling backbone for C-LIFT. A problem-solving approach using case-based reasoning is discussed to solve problems where the knowledge for problem solving cannot be conveniently represented in the form of rules or models. It involves retrieving relevant cases, adapting the solution from a previous case if necessary to solve the new problem, and storing the current new case. The organization of the case-based model and its generation process with respect to the selection of lift configuration are presented. To verify the validity of results generated by C-LIFT using the above implemented methodologies, a computer Iiftability study is performed based on an actual platform with multiple modules in the North Sea. The lift operation (single/dual-crane lift) and sequence of lift installation of modules generated by C-LIFT are tabulated and compared. These results are found to be consistent with the actual execution of lift installation.