4.5 IMPACT OF SENSITIVITY ANALYSIS FACTORS
4.5.3 Level of Availability of Machines
In this sub-section, the effects of the variation in the level of availability of each of the manufacturing stages on the products’ service levels and the system WIP are presented for each of the experimental scenarios. The level of availability/reliability of a machine is derived by dividing its mean time to failure by the sum of its mean time to repair and its mean time to failure (i.e. ܯܶܶܨȀሺܯܴܶܶ ܯܶܶܨሻ). The base level of availability of each machine was set to 90% in Section 3.4; therefore, the aim here is to determine at a 95% confidence level if the ±5% variation in the level of availability of each of the manufacturing stages had statistically significant impact on the products’ service levels and the system WIP.
EKCS
Under the DKAP and the SKAP, it was only in Scenario 1 that the variation in the levels of availability of the three stages had any statistically significant impact on the performance measures, as shown in Table 4-16 and Table 4-17.
90
Table 4-16: Impact of Stage Availability Level under EKCS DKAP
Scenario
Stage 1 Level of Availability
on Stage 2 Level of Availability on Stage 3 Level of Availability on Product 1 SL Product 2 SL WIP Product 1 SL Product 2 SL WIP Product 1 SL Product 2 SL WIP
1 Yes Yes Yes Yes Yes No Yes No No
2 No No No No No No No No No
3 No No No No No No No No No
4 No No No No No No No No No
Table 4-17: Impact of Stage Availability Level under EKCS SKAP
Scenario
Stage 1 Level of Availability
on Stage 2 Level of Availability on Stage 3 Level of Availability on Product 1 SL Product 2 SL WIP Product 1 SL Product 2 SL WIP Product 1 SL Product 2 SL WIP
1 Yes Yes Yes Yes Yes Yes No No No
2 No No No No No No No No No
3 No No No No No No No No No
4 No No No No No No No No No
In Scenario 1, under the DKAP, the variations in the levels of availability of the first two stages had statistically significant impacts on the two products’ service levels, while that of the last stage only had statistically significant impact on Product 1’s service level. Additionally, it was only Stage 1’s variation in the level of availability that had statistically significant impact on the system WIP. Under the SKAP, it was only Stage 3‘s variation in the level of availability that did not have statistically significant impact on any of the three performance measures. Those of Stages 1 and 2 had statistically significant impacts on the products’ service levels and the system WIP
GKCS
Unlike under the EKCS, the variations in the levels of availability of the manufacturing stages – specifically the first two stages – had statistically significant impact on at least one of the performance measures in all the four scenarios, as shown for the DKAP and the SKAP respectively in Table 4-18 and Table 4-19. The variation in the level of availability of Stage 3 did not have any statistically significant impact on the products’ service levels and the system WIP across the four scenarios. This could be attributed to the GKCS’s demand information transmission philosophy which is done stage by stage
91
and thus relies on the availability of the stages. The last stage is not as critical because the demand information transmission process by-passes it.
Table 4-18: Impact of Stage Availability Level under GKCS DKAP
Scenario
Stage 1 Level of Availability
on Stage 2 Level of Availability on Stage 3 Level of Availability on Product 1 SL Product 2 SL WIP Product 1 SL Product 2 SL WIP Product 1 SL Product 2 SL WIP
1 Yes Yes Yes Yes Yes Yes No No No
2 Yes Yes Yes Yes Yes Yes No No No
3 Yes No Yes Yes Yes Yes No No No
4 Yes Yes Yes No No No No No No
Table 4-19: Impact of Stage Availability Level under GKCS SKAP
Scenario
Stage 1 Level of Availability
on Stage 2 Level of Availability on Stage 3 Level of Availability on Product 1 SL Product 2 SL WIP Product 1 SL Product 2 SL WIP Product 1 SL Product 2 SL WIP
1 Yes Yes Yes Yes Yes Yes No No No
2 Yes Yes Yes Yes Yes Yes No No No
3 No No Yes Yes Yes Yes No No No
4 Yes Yes Yes No No No No No No
In Scenarios 1 and 2, under both policies, the variations in the levels of availability of Stages 1 and 2 had statistically significant impacts on the two products’ service levels and the system WIP. Also, the statistically significantly higher service level resulting from an increased level of availability of Stage 1 results in a correspondingly higher WIP level, as shown for Scenario 1 in Figure 4-10. The same was observed in the other scenarios in cases whereby the variation in the level of availability of any stage caused statistically significant difference in the service levels.
92
In Scenario 3, under the DKAP, the variations in the levels of availability of Stages 1 and 2 had statistically significant impacts on the service level of Product 1 and the system WIP. Additionally, that of Stage 2 had statistically significant impact on the service level of Product 2. Under the SKAP, the variations in the levels of availability of Stages 1 and 2 again had statistically significant impacts on the system WIP, while it was only Stage 2 that had statistically significant impact on the service levels of the two products.
In Scenario 4, under the DKAP and the SKAP, only the variation in the level of availability of Stage 1 had statistically significant impact on the products’ service levels and the system WIP. This might be due to the high levels of demand variability of the two products and the fact that those are the two stages to which a demand information is transmitted directly as it arrives to the system. A lower or higher level of availability of those stages might not have made significant difference in their ability to overcome the high levels of product demand variability.
In general, across the four scenarios, the variations in the levels of availability of the stages had more statistically significant impact on the performance measures under the GKCS than under the EKCS. In particular, the first two stages had statistically significant impacts more consistently than the last stage, and this could be explained as follows. Firstly, the impact of stage availability on the GKCS shows that because it relies on a stage by stage demand information transmission, if the manufacturing stages are not reliable to complete the processing of parts on time, the demand information will be frequently delayed waiting for Kanbans to be detached from parts that complete processing. Secondly, under the GKCS, the first two stages are more crucial because the last stage is by-passed in the transmission of demand information upstream. As described in Section 2.2.6, when a customer demand arrives to the last stage, it is transmitted directly to the penultimate stage without having to first couple with a free Kanban, as done at the other stages. Therefore, even if the last stage is less available, the upstream transmission of the demand information is not delayed. The same logic can be applied to understanding why the EKCS’s global demand information transmission ensured that the levels of availability of the stages had lesser impacts on the system’s performance. The EKCS seems to have been affected under Scenario 1 because Product
93
1 with the higher demand arrival rate also had the higher CV level, and this could have had a confounding effect on the products’ service levels. This was also evident in the robustness analysis of Section 4.4.1.
Furthermore, it can be concluded from these observations that the statistically significant impacts of the variations in the levels of availability of the stages on the GKCS’s system WIP was also as a result of the delay in the demand information transmission, especially due to low levels of availability of Stages 1 and 2. The GKCS’s release of new parts into the system only occurs after the demand information has passed through all the stages to reach the upstream raw parts buffer. On the contrary, the EKCS’s demand information reaches the upstream raw parts buffer as soon as it arrives to the last stage and it can release new parts into the system as soon as a first stage Kanban is available.
In conclusion, it seems a study that proposes a flexible routing of information transmission to create entirely new controls as needed is very promising for the customisation of pull strategies to suit specific manufacturing systems [38]. Such approach can be used to by-pass the localised flow of demand information through an unreliable manufacturing stage.