6. Conclusion and outlook
This article has presented a unifying method for routing of flexible one-dimensional components with geometric design constraints. In a segregated approach, a resolution complete A ∗ based grid search combined with a lower dimensional sweep procedure first generates feasible nominal configurations. Secondly, a simulation model based on Cosserat rod theory is coupled to the Nelder–Mead algorithm to ensure that the flexible component is still feasible in mechanical equilibrium by locally adjusting a set of tunable design variables. Test results show that the method is able to efficiently generate global solutions for different objectives for industrial scenarios involving complex geometries and real constraints. To extend upon this work in the future there are several possibilities:• The method should be extended to flexible 1D components with multiple branching and break out locations and non-circular cross sections. • The design of the feedback update in Section 4.4 should be investigated more in detail and the simple scheme in Algorithm 2 could potentially be improved. • As mentioned in Section 4.3, it should be looked into how to gradually increase the gravity field during local refinement to avoid local minima where the configuration is in contact. • The cost field φ defined in Eq. (9) should be extended to give preference to more specific areas of space, e.g. favouring closeness to a certain surface. Also, additional via points could be introduced in order to further help the designer guide the routing through preferred regions. • The structure of the uniform grid sometimes permits optimal nominal configurations with an unnecessary high number of turns if the grid is too coarse. This issue can certainly be resolved by increasing the grid resolution, however a smoothing of the nominal configuration in a post-processing step could be an alternative.
