Common use of Verification and Validation Clause in Contracts

Verification and Validation. Verification is done for both models. Verification is to test whether the model does what we expect it does. The verification is divided into two parts: the extreme value tests and compares the output of the simulations with exact calculations. In Appendix H the verification of the Min-max model is given. In Appendix I, the verification of the Lead-time model is provided. For the extreme value test, extreme values are assigned to the input parameters; respec- tively to the turn-around stock, the number of servers, the minimum level or threshold, the lead-times or total requirements and the repair times. For example the turn-around stock is set equal to 0 and 100 while the values of the other input parameters remain unchanged. For every case one run is simulated and it is checked whether the output of the simulations was as expected. The results clearly show that the model works as expected. For the com- parison with the exact calculation for one parameter setting, different KPIs are calculated using formulas. For this parameter setting 20 simulations are run. The values of the exact calculation are compared with the values of the 20 simulation runs using a statistical t-test. In the appendices the methods are explained in more detail. During the validation we tested whether the outcome of the model correspond to the outcome in the real situation. To test whether this is the case the Min-max As-Is situation is simulated. As this situation is most similar to the current situation at NedTrain. The average fill rate obtained from the simulation is equal to 0.84 (independent from the uti- lization rate), See Appendix L and Table L.1. The fill rate of the model should be equal to the network service rate at NedTrain, because they measure the same. The network service rate of the 99 parts used in this simulation, in the year 2011 is equal to 0.9681. The differences between the fill rate of the simulation and the network service rate is quite large. As discussed in the subsection 3.3.3 the assumption of independent lead-times might be responsible for this large difference. But due to lack of the right data it was not possible to determine whether the lead-times are dependent from their priority. Therefore this large difference is taken for granted.