Type of Events and Schedules Rescheduling Assumptions and Rules Data Selection Process

Abdul Samad Shibghatullah 114 30052008

5.2 Experiment Plan

This section explains details of types of events, types of schedules, rules and assumptions, the data selection process, and the experimental environment we used for this experiment.

5.2.1 Type of Events and Schedules

The events are the same as discussed in Section 4.2.1 except that there is no delay event. The events are lateness LFSO, LFR, and LFSW and unavailability UNV. The reasons that we do not test delay events is because the duration is quite long and generally it requires more than one round of rescheduling. Two or more than one rounds of rescheduling need manual intervention and adjustments, and when there are many events simultaneously this becomes quite complex. In this experiment, we do not consider more than one round of rescheduling. The types of schedules are the same as in the single event experiments; large, medium and small schedules.

5.2.2 Rescheduling Assumptions and Rules

The rescheduling assumptions and rules are similar to those discussed in Section 4.2.2. The difference is that there is not more than one round of rescheduling because two or more rounds of rescheduling need manual adjustments which are complex. In this experiment, we only consider one round of rescheduling.

5.2.3 Data Selection Process

In these experiments, multiple events take place randomly. Therefore, we should know how to sample the number of events we need and how to select random data. The minimum number of events is 2 and the maximum number is 20 of the total number of duties of the schedule. For a large schedule it is 18 20 of 88 duties, for a medium 10 20 of 51 and for a small 5 20 of 23. To select data randomly, we used the random formula in Microsoft Excel to generate random numbers and the numbers were used to refer to the duty number. For the lateness event we used the formula below: RAND the maximum number-the minimum number + the minimum number Abdul Samad Shibghatullah 115 30052008 The maximum number is the total number of duties and the minimum number is 1. For example, for a medium schedule that has 51 duties the formula will look like this: RAND 51-1 +1 Table 5.1 shows the selected data for a lateness event according to the type of schedule. The full data for the experiments is shown in Appendix G. For the unavailability event, we selected the start time of the event randomly based on the finish work 2 time. Below is the formula. Table 5.2 illustrates the selected data for an unavailability event. RAND the latest time for ‘Finish Work 2’ - the earliest time for ‘Finish Work 2’ + the earliest time for ‘Finish Work 2’ Table 5.1: The Selected Data for Lateness Event Large Medium Small No of Events Crew ID Crew ID Crew ID 1 E K C 2 P O H 3 V Q L 4 W T M 5 X U O 6 AB AG 7 AC AH 8 AD AL 9 AL AP 10 AQ AW 11 AU 12 AW 13 AX 14 BK 15 BN 16 BR 17 BT 18 CA Abdul Samad Shibghatullah 116 30052008 Table 5.2: The Selected Data for UNV Event Large Medium Small No of Events Crew ID Crew ID Crew ID 1 11:48:00 14:30:00 19:12:00 2 12:45:00 14:48:00 20:54:00 3 12:54:00 15:00:00 22:18:00 4 14:24:00 16:30:00 23:54:00 5 15:28:00 17:12:00 6 15:42:00 22:18:00 7 15:54:00 23:36:00 8 16:06:00 23:54:00 9 16:24:00 24:48:00 10 17:24:00 11 17:48:00 12 18:30:00 13 18:54:00 14 19:00:00 15 19:36:00 16 20:12:00 17 20:48:00 In this experiment, we also tested the event with different timing. For lateness except LFSO the event timing started from 15 minutes and was increased by 5 minute intervals until 60 minutes was reached. LFSO is only 15 minutes and 20 minutes because when the timing is more it requires second round rescheduling, which is not considered in this experiment. For unavailability, the event started from 30 minutes increasing by 15 minute intervals until 120 minutes was reached. Table 5.3 demonstrates the overall picture of the experiments. For example, in the event of LFSO- Large schedule there were 34 experiments. The maximum number of duties affected in the large schedule was 18 as discussed in the previous paragraph but the experiment shows only 17 because the event started from 2 events and went to 18 events. The grand total number of the experiments was 870. Abdul Samad Shibghatullah 117 30052008 Table 5.3: The Overall Picture of the Experiments Type of EventsType of SchedulesNumber of Experiments Total Large Schedule 17 x 2 34 Medium Schedule 9 x 2 18 LFSO Small Schedule 4 x 2 8 Large Schedule 17 x 10 170 Medium Schedule 9 x 10 90 LFR Small Schedule 4 x 10 40 Large Schedule 17 x 10 170 Medium Schedule 9 x 10 90 LFSW Small Schedule 4 x 10 40 Large Schedule 17 x 7 119 Medium Schedule 9 x 7 63 UNV Small Schedule 4 x 7 28 Grand Total 870

5.2.4 Experimental Environment