CARACTERISTICAS BÁSICAS

Waiting for delivery to the customer is also a factor that influences the throughput time. For example, if bearing supports are transported to the customer on for example Monday, and the bearing support is finished on Tuesday, it has to wait for another 6 days before it will be transported. Therefore the delivery day and the frequency might also be an interesting factor to experiment with. Choosing a different delivery and acquiring day does not influence the base model. A different delivery and acquiring day would only mean that the entire production schedule does not start on, for example, Monday, but on a Tuesday, resulting in similar results. It would be interesting to see what the effect would be of delivering and acquiring bearing supports twice per week. Though, to change the base model as such, such that this can be experimented, would take a lot of modeling time. Regarding the time of this thesis, this will not be done since the results would, in my opinion, not be that valuable for the base model. This opinion is based on the following example:

For example, the new and finished bearing supports are delivered and acquired on Monday and Thursday. Delivering twice would mean that 9 bearing supports start on Monday and are acquired on Monday 9 weeks later. The same goes for Thursday. This is because this is simply how the planning is made. The only effect this has, is that the batch sizes are smaller. Therefore, spreading the bearing supports more over the week such that bearing supports are produced not only at the start of the week but also at the end of the week. In this way, queues would probably be lower in longer chains (for example the chain in week 5) such that perhaps less waiting time is observed. Though, the departments still have to do the same amount of work, so queues at bottlenecks would probably still be as long. Lastly, delivering twice would, if the planning was strictly followed, also need transporting to MPP twice. If not, delivering and acquiring is also influenced by the chosen MPP date, adding even more complexity.

To conclude, the delivery and acquiring days and intervals will not be experimented for the base model. Though, to see the influence of transporting on a different day than Monday and of transporting more often, it will be done for the MRP-model. In this model, the issues mentioned

above do not play a role since the planning is not ‘fixed’. The transport day to MPP is set on Monday

for similar reasons as outlined above (interaction effects), just like in the base model.

The optimal delivery and acquiring day in the MRP-model

Regarding the optimal transport day, the following results were acquired.

Table 43: Results of KPI's for a different delivery and acquiring day (MRP-model).

Average Internal throughput time Total average throughput time Total average WIP (pc.)

Avg. Waiting for Delivery Mo 21:23:07:28 41:14:41:39 107,00 05:15:34:11 Tu 22:06:27:35 38:12:19:05 99,04 02:05:51:30 We 22:08:29:38 37:14:53:11 96,74 01:06:23:32 Th 23:14:38:38 39:17:56:06 102,21 02:03:17:28 Fr _{23:17:03:12 40:20:42:55} 105,09 03:03:39:43

91

Figure 15: Confidence intervals of the average total throughput time for a different delivery and acquiring day (MRP-model).

The different days change the throughput time. The confidence interval of Wednesday lays lower than all other days. Moreover, no confidence interval seems to be overlapping. This shows that choosing a different day does change outcomes. As an addition, a paired-T approach was conducted with a confidence level of 95%. This paired-T approach can be found in appendix F.8. Based on this, Monday is significantly larger as all the other days. Moreover, Wednesday outperforms all other delivery and acquiring days. Tuesday outperforms Thursday and Friday, while Thursday outperforms Friday.

Multiple transport days and the best combination

Regarding the optimal combination, if driving two customer X is done two or three times, the following results can be acquired:

Table 44: Results of KPI's for a combination of different delivery and acquiring days (MRP-model).

Internal avg. TT (s) Total avg. TT (s) Total average WIP (pc.) Average waiting for delivery (s) Mo - We 19:14:26:50 34:16:35:52 89 01:02:09:00 Mo - Th 19:16:24:51 35:14:37:14 92 01:22:12:22 Tu - Th 20:06:12:39 35:00:53:55 90 00:18:41:15 Tu - Fr 20:08:09:30 35:10:11:11 91 01:02:01:41 We - Fr 20:12:01:47 35:16:19:56 92 01:04:18:08 Mo - We - Fr 19:20:05:56 34:21:00:08 90 01:00:54:11

With the corresponding confidence intervals (95%) regarding the total average throughput time:

3200000 3300000 3400000 3500000 3600000 Mo Tu We Th Fr Av era ge to ta l th ro u gh p u t time (s ec) Day Interval Left bound Interval right bound Average

92

Figure 16: Confidence intervals of the average total throughput time for a combination of different delivery and acquiring days (MRP-model).

In contrast to different delivery days, confidence intervals of combinations of days are overlapping. This is because the averages in the tables are relatively close to each other. The average of the combination Monday-Wednesday seem to lay the lowest. Though, based on the graph it is difficult to say which combination performs the best. Therefore, a paired-T approach was conducted with a confidence level of 95%. This paired-T approach can be found in appendix F.8. With this paired-T approach, Monday and Wednesday have also been compared to driving more often to customer X. Driving two times always outperformed driving one time. Moreover, driving Monday and Wednesday outperformed all other combinations of transport days, including driving three times. The

conclusions for the other configurations can be found in the appendix.

Analysis of the results

Regarding the standard MRP-model, the best day to acquire new bearing supports and send finished bearing supports is on Wednesday. It is interesting to see that the internal throughput time on Monday is lower than on Wednesday. An explanation for this could be that waiting on MPP is lower in the model (included in internal throughput time) but waiting on the customer is longer, resulting in a longer total throughput time.

When transporting is done twice, the best configuration is on Monday and Wednesday. Another interesting configuration is Tuesday and Thursday since it has the lowest waiting on delivery time. Though, with the configuration Monday – Wednesday, average waiting for MPP time is probably lower. This configuration even outperforms driving three times. An explanation for this might be the introduced set-up times. If the batches become smaller, departments finish the smaller batches faster, leaving the department idle. Later, a new batch comes in, and the department needs to be set-up again. In a larger batch, the department only needs set-up once for the entire batch. This disadvantage does not outweigh the difference between driving one or two times but does outweigh the difference between driving two or three times.

Regarding the saving in throughput time, the following conclusions can be drawn based on the paired-T approach (95%). Unfortunately, no costs are known for this transport. It is unclear who will pay what since transporting more often is also interesting for the customer because it acquires the bearing supports faster.

2950000 3000000 3050000 3100000 3150000 Mo - We Mo - Th Tu - Th Tu - Fr We - Fr Mo - We - Fr Av era ge T o ta l T h ro u gh p u t Tim e (s ec) Configuration Interval Left bound Interval right bound Average

93

Table 45: Comparing the different delivery and acquiring days on time gains, WIP, and WIP value.

Configuration Approximate time

saving regarding

delivering and acquiring once on Monday.

Difference in mean WIP regarding the base MRP model (pc.)

Difference in mean WIP value regarding the base MRP model

Monday 0 0 0

Wednesday 3 days and 21 hours to 4 days and 2 hours.

10,26 €51308,-

Monday and Wednesday

6 days and 19 hours to 7 days and 1 hour.

17,80 €88983,-

And the approximate time saving regarding delivering and acquiring once, on Wednesday compared to driving twice on Monday and Wednesday is 2 days and 20 hours to 3 days and 1 hour (paired-T approach – 95%). This corresponds to a difference in mean WIP of approximately 7,54 pc. and a difference in mean WIP value of €37676,-. This experiment shows again that the outsourcing day to MPP and delivery and acquiring day of the customer should be well-aligned.

6.4 Determining the required capacity

This section will return to the main question of this thesis, namely the action problem presented by the company:

How can the throughput time of the bearing support be minimized from 9 weeks to a maximum of 4 weeks?

This section will combine the best results from the experiments to determine the capacity and things that are needed to reduce throughput time to a maximum of 4 weeks. This will be done in a

simulation model, in which the results are combined. With these combined results, the model will be run. Consequently, it will be investigated if the average throughput time is under 4 weeks. If this is true, all the things will be outlined that are needed to produce under 4 weeks. This includes the number of full-time employees (FTE’s), by tracking their occupation and calculating the minimal number of needed FTE’s. FTE’s are based on 40 hours per week.

This is also done because, during this thesis, a sufficient capacity was assumed by calculating the needed capacities. In addition, in the simulation model, some departments were modeled separately because their occupations are not that high, while they are one department in reality. By combining them in this section, the needed capacity can still be determined. So, this solves these two

assumptions that were made when modeling was initiated.