Example of Delaunay triangulation-based sub-sample generation

the M-marked point represents the performance vector for which the corresponding configuration is to be determined. All other points represent the set of Pareto-optimal solutions Z ll. A-marked points are adjacent to y. The corresponding simplices incident to f are highlighted in blue. 

To illustrate a more realistic decision problem, assume a stock holding cost rate for Naphtha 55 year as calculated in example 9 and average ordering 

The corresponding configuration is calculated as = (15, 591 29,667 76,538). I. e. based on the given cost parameters and the results of the simulation study, the optimal inventory parameters can be derived as shown in Table 4-14- They differ from the baseline values derived at the end of example 12 by component-wise deductions. 

Clearly, the optimal values are smaller than the baseline parameters (except for Sy) indicating a smaller average stock level. Moreover, for the optimal configuration, the order frequency at the harbour is smaller since Sp/ -sff . This implies fewer 

240 phis is caused by rounding errors during simulation as well as the natural stochasticity of the responses. I.e. it cannot be decided whether a (marginal) difference in a response is caused by noise or an advantageous constellation of parameters. 

---

Figure 4.17 Example of Delaunay triangulation-based sub-sample generation...

---