RELIABILITY BASED ROBUST DESIGN OPTIMIZATION OF A FREE-FALL-LIFE-BOAT

Author(s): 
Alberto Clarich, Dimitrios Drougkas

CHALLENGE - Combined CFD and FEA algorithms are used to analyze a vessel’s behavior, where numerous iterations are needed to reach convergence, including interpolation of the dynamic loads from the CFD to the FEA. However, the use of an FSI algorithm can produce results much faster and thus the design optimization can become a feasible and cost effective approach. In this paper, a case study of the FFLB optimization is presented using an FSI solver. A reliability-based optimization problem is defined to statistically increase the safety of free fall lifeboats (FFLB), which are typically used to evacuate passengers in oil platforms and large transport vessels.

SOLUTION - The free lifeboat studied in this work is 10.2 meters long and 3.4 meters wide, weighing 9,517 kg, made of GFRP (Glass Fiber Reinforced Plastic ) and able to carry up to 30 people. The model pre-processing has been done in ANSA. The responses which are of interest in this problem are: the distance from the host structure reached by the boat when it emerges out of the water, and the CAR or combined acceleration ratio. The setup of the optimization process was completed in a process workflow in modeFRONTIER. The ANSA model including the original mesh can be automatically updated for each different configuration proposed by the optimization algorithm accordingly to the values of the input variables, and the updated mesh model (.key file) is then transferred to the following application, a shell script which launches the LS-DYNA simulation for the stress analysis. 

BENEFITS - The application of MOGA-II algorithm in modeFRONTIER brought optimal results with an overall number of required simulation designs equal to 15. Results show the constraint respected, with a significant reduction of the highest percentile of CAR acceleration (29%), and an increase of 10% of the mean distance, and an increase of even 31% relatively to the lowest percentile, confirming the robustness of the solution.