Addition of Features to an Existing MDO Model for Containerships
Traditionally, the âDesign Spiralâ is used for the design of ships. The design spiral endorses the concept that the design process is sequential and iterative. Though this procedure was very effective over the years, the current trend of engineering demands that more stress be put on the exploration of optimum design. With the advancement of computing technologies, the onus has shifted from finding better calculation schemes to formulating an economically viable design scheme. One of the objects of the FIRST project funded by MARITECH was to develop a computer tool to give the best ship design using optimization techniques. This was entrusted to the Department of Aerospace and Ocean Engineering at Virginia Polytechnic Institute and State University in Blacksburg, Virginia. A container ship was chosen as the test case. The problem was tackled from an ownerâs point of view. Hence, the required freight rate was chosen as the objective.
To achieve that goal, the team developed a package that consists of three modules: optimization, geometric and a performance evaluation module. Though these modules are essentially independent, the user has control over an overall manager. He can change the initial value of design parameters, set bounds and vary constraint bounds as per his needs. Though he does not know what goes on behind the user interface, he still feels secure with the design process because he has overall control. This sense of security breaks down when he has access to limited variables and constraints.
A prototype MDO tool is developed based on Microsoftâs COM framework using ATL. With this design, the modules can be modified with minimum programming effort. The user interface gives the user flexibility to manipulate relevant parameters that affect the design. A geometric shape manipulation scheme is developed in which the hull form was generated by blending two hull forms. This MDO tool is used to design a container ship with the required freight rate as the objective to be minimized. It is noticed that without a structural constraint, the design tends towards one with maximum length and beam. This led to unreasonably large ratios of B/D and L/D.
A B/D constraint is applied to the design to get a better structural design. Results with this constraint enabled have pointed in the direction of adding two other design variables. This constraint increases the depth of the ship. With the increase in depth, the center of gravity of the ship also rises decreasing the GM of the ship. This lowering of GM adversely affects the GM constraint. The number of tiers on deck (NTd) is made a design variable to enable the optimizer to have the flexibility of manipulating the cargo carrying capacity. It was noticed that the ship is unable to have a high NTd because of the violation of the GM constraint. Hence, ballast has also been added as a design variable to reduce the center of gravity of the ship increasing the GM of the ship. This feature enables the optimizer to carry greater cargo on deck improving the objective function.
An effort is made to analyze the efficacy of the MDO tool by varying various parameters that affect the design. Technology factors have been introduced which give an insight on effect of key parameters. They also reflect on future design trends. Three evaluation tools: sensitivity analysis, alpha plots and restart option have been incorporated in the design process to gauge the results of optimization.
The effect of another structural constraint L/D was also investigated. This constraint tends to bring down the overall length and is inconclusive in its results. Further analysis of this constraint is needed to draw usable conclusions. The linear response surface approximation was eliminated and the original stepwise discontinuous TEU capacity function is employed in the later examples. It was found that the minimum of the required fright rate occurred at the lower limits of length and beam on each TEU capacity platform. A systematic search of TEU plateaus in the vicinity of the primary optimum was necessary to define the secondary optimum