Hybrid bat flight optimization algorithm applied to complex wellbore trajectories highlights the relative contributions of metaheuristic components

• Application of a hybrid bat-flight optimization algorithm to wellbore trajectory design.
• Verifying the efficient performance of the proposed hybrid bat-flight algorithm.
• Detailed description and analysis of metaheuristic components constituting the algorithm.
• Balancing performance of metaheuristics focused on local and global searches of the solution space.
• Presentation of the novel techniques of metaheuristic profiling to aid algorithm design.
• VBA-driven Excel models to perform and interrogate optimization algorithm's performance are useful for applying metaheuristic profiling.

A hybrid bat-flight optimization (BFO) algorithm is described and developed in comparison with the original bat-inspired algorithm (Yang, 2010). The changes made remove the need to evaluate and store velocities from previous iterations to calculate new solutions, thereby reducing the computational requirements without negatively impacting the performance of the algorithm as an efficient optimizer. The hybrid BFO consists of six metaheuristic components that complement each other in their contributions to global and local search of solution spaces. The hybrid BFO algorithm is applied to a well-reported complex wellbore trajectory optimization problem previously used to evaluate the performance of evolutionary optimization algorithms. The hybrid BFO is shown to work effectively and efficiently in finding the optimum solution space, requiring significantly less iterations to do so than a hybrid genetic algorithm applied to the same problem, both developed in VBA code.The performance of the hybrid BFO algorithm is further evaluated by a novel technique of metaheuristic profiling introduced in this work. By recording the origin of each solution generated in each iteration of the algorithm, in terms of which metaheuristic component is responsible for producing it, a profile of the origin of the ten highest-ranking solutions in each iteration is constructed. This profile reveals that the metaheuristic components driven by the frequency, loudness and pulse rate (i.e., the bat-echolocation-inspired metrics used to drive the algorithm) contribute to the solutions derived in complementary, but varying ways as the iterations of the algorithm progress. Metaheuristic profiling is considered to be a promising technique for design, performance comparison, improvement and customization of evolutionary algorithms.

Metaheuristic profile diagram for hybrid BFO applied to the wellbore trajectory example for solutions rank#1 to rank#10 for each of 250 iterations for which the algorithm was run. The symbols are layered from rank#1 downwards, hence, ranks below about rank#3 are obscured in the display. However, it is overall density of symbols against each metaheuristic code (i.e., 1 to 6 on the vertical scale) that identifies the significance of that metaheuristic to solutions derived at various iterations. The technique of metaheuristic profiling is proposed here as a useful approach in characterizing the relative contributions of metaheuristics to the overall performance of optimization algorithms. This approach is proving helping in designing, improving and customizing hybrid evolutionaryoptimization algorithms for specific applications.Figure optionsDownload high-quality image (236 K)Download as PowerPoint slide