Data holds a wealth of information. It can provide rich insights and be the foundation of...
Accelerated concept design redefines the optimization process of concept designs. The accelerated concept design framework uses simulation software in combination with in-house developed algorithms to generate optimized concept designs. This unique solution for the maritime industry increases creative solutions while reducing concept design delivery times.
What is accelerated concept design?
Accelerated concept design is a holistic design framework that allows quick design iterations while optimizing ships in an efficient way. Traditionally, a concept design is developed following the classic design spiral. With this process, the naval architect needs to answer many complex design questions in a short time while also taking into account decision variables, constraints, and multiple objects. Decisions taken at this stage are difficult to change in a later stage without massive delays and increasing cost.
Following research on how to improve this process, C-Job implemented a different approach as visualized in the accelerated concept design circle. The new method considers all design aspects simultaneously while generating multiple optimized concept designs. Besides optimization, the framework is set up to automatically check designs for feasibility.
Accelerated concept design not only produces more effective and optimized ship designs, due to automation it achieves this in a shorter amount of time. This gives naval architects more time to consider more innovative, creative – even out-of-the-box – ideas. Additionally, the optimized concept design can be presented to the client much earlier than with traditional methods.
Because it yields accurate information about stability, construction weight, building costs, and vessel performance, accelerated concept design removes uncertainty from the whole ship design and construction process. This reduces risks and gives the ship-owner more confidence in the design process.
Accelerated concept design and algorithms
C-Job’s research into accelerated concept design initially used genetic algorithms. These algorithms use a ‘survival of the fittest’ method to reach optimized designs over progressive generations of designs. As each generation passed, only the fittest designs passed on their design ‘genes’.
As the research project continued, C-Job developed the CEGO algorithm. The algorithm has taken Accelerated Concept Design to the next level.
The CEGO algorithm is much faster and more efficient. This is because they do not need as many function evaluations as the genetic algorithms. With a comparable concept design, the genetic algorithmic method would need to evaluate 200,000 different design variations while CEGO only requires 200 function evaluations to find the optimal solutions.
Constrained Multi-Objective Efficient Global Optimization (CEGO) in short
CEGO is an optimization algorithm that can be used to optimize constrained multi-objective optimization problems using a limited number of function evaluations.
The advantage of CEGO is that it uses surrogate models for both the constraints and the objectives to learn from the evaluations it has made so far. For the constraints, Cubic Radial Basis Functions are used and for the objectives we use Gaussian Process Regression in combination with the S-Metric Selection criterion.
Artificial intelligence
The CEGO algorithm learns from the mistakes it has made in the past. This way the algorithm learns where the boundary is between feasible and infeasible solutions. This can then be exploited by concentrating only on the feasible designs. The CEGO algorithm is a surrogate-assisted optimization algorithm. By modeling the objectives and constraints, the algorithm considers the entire design space instead of only a portion of it.
“Our research into accelerated concept design framework has resulted in a method which considers both capital and operational expenses. By searching for the global best solution, we can reduce total cost of ownership of any vessel.”
- Roy de Winter, Data Scientist
Accelerated concept design is developed together with C-Job partners NAPA, the Finnish independent software house, and the Leiden Institute of Advanced Computer Science in the Netherlands.
Together with our partners, C-Job presented its paper titled ‘Optimizing ships using the holistic accelerated concept design methodology’ at PRADS 2019 in Japan.
The paper is available upon request. If you wish to receive a copy, please contact Roy.