Advanced Analytics to Mitigate and Manage Supply Chain Effects of Demand Variations Communicated in Delivery Schedules A Case Study of a Supplier in the Automotive Industry Master’s Thesis in the Programme of Supply Chain Management LINUS HANSEN FREDRIK SVEIDE Department of Technology Management and Economics Division of Supply and Operations Management CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2020 Report No. E2020:069 MASTER’S THESIS E2020:069 Advanced Analytics to Mitigate and Manage Supply Chain Effects of Demand Variations Communicated in Delivery Schedules A Case Study of a Supplier in the Automotive Industry LINUS HANSEN FREDRIK SVEIDE Tutor, Chalmers: Patrik Jonsson Tutor, Company: Johan Bystedt Department of Technology Management and Economics Division of Supply Chain and Operations Management CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2020 Advanced Analytics to Mitigate and Manage Supply Chain Effects of Demand Variations Communicated in Delivery Schedules A Case Study of a Supplier in the Automotive Industry LINUS HANSEN, 2020 FREDRIK SVEIDE, 2020 Master’s Thesis E2020:069 Department of Technology Management and Economics Division of Supply Chain and Operations Management CHALMERS UNIVERSITY OF TECHNOLOGY SE-412 96 Gothenburg, Sweden Telephone: + 46 (0)31-772 1000 Cover: A MAPE-profile of forecast accuracy scores of the case company, 5.1.2 Chalmers digitaltryck Gothenburg, Sweden 2020 2.2.1 2.2.2 • • • • 2.2.1 0% 20% 40% 60% 80% 100% 120% 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 Mape diagram Mape 1 Mape 6 Random variation area Late variation 2.2.2 𝑀𝐴𝑃𝐸 = 1 𝑛 ∑ | 𝐴𝑡 − 𝐹𝑡 𝐴𝑡 | 𝑛 𝑡=1 𝑤ℎ𝑒𝑟𝑒 𝐴𝑡 = 𝐴𝑐𝑡𝑢𝑎𝑙 𝑑𝑒𝑚𝑎𝑛𝑑 𝐹𝑡 = 𝐹𝑜𝑟𝑐𝑎𝑠𝑡𝑒𝑑 𝑑𝑒𝑚𝑎𝑛𝑑 𝑄𝑢𝑎𝑙𝑖𝑡𝑦 × 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 × 𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑖𝑙𝑖𝑡𝑦 = 𝑂𝐸𝐸 (%) • • • • • • • • • FAI WTS Item Group C OEM 2 47.70% 0.16 Item Group A OEM 2 49.30% 0.22 Item Group D OEM 3 56.20% 0.13 Item Group B OEM 2 58.10% 0.16 Item Group E OEM 2 58.40% 0.1 Item Group H OEM 2 59.70% -0.18 Item Group D OEM 1 61.00% 0.06 Item Group D Total 61.40% 0.1 Item Group I Total 62.50% 0.06 Item Group I Subassembly 1 62.50% 0.06 Item Group H Total 64.70% 0.07 Item Group D OEM 2 65.90% 0.24 Item Group I OEM 2 66.60% 0.02 Item Group F OEM 1 67.30% 0.09 Item Group G OEM 1 68.20% 0.08 Item Group H Subassembly 3 72.20% -0.06 Item Group H Subassembly 2 75.20% 0.15 Item Group H Subassembly 1 78.20% 0.23 Item Group I Subassembly 4 81.60% 0.07 FAI WTS Item Group D OEM 2 65.90% 0.24 Item Group H Subassembly 1 78.20% 0.23 Item Group A OEM 2 49.30% 0.22 Item Group C OEM 2 47.70% 0.16 Item Group B OEM 2 58.10% 0.16 Item Group H Subassembly 2 75.20% 0.15 Item Group D OEM 3 56.20% 0.13 Item Group E OEM 2 58.40% 0.1 Item Group D Total 61.40% 0.1 Item Group F OEM 1 67.30% 0.09 Item Group G OEM 1 68.20% 0.08 Item Group H Total 64.70% 0.07 Item Group I Subassembly 4 81.60% 0.07 Item Group D OEM 1 61.00% 0.06 Item Group I Total 62.50% 0.06 Item Group I Subassembly 1 62.50% 0.06 Item Group I OEM 2 66.60% 0.02 Item Group H Subassembly 3 72.20% -0.06 Item Group H OEM 2 59.70% -0.18 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 10 w 9 w 8 w 7 w 6 w 5 w 4 w 3 w 2 w 1 w Four relevant MAPE profiles connected to a delivery schedule Customer ItemNo Item Group WVMAPE ∆ ∆ ∆ ∆ ∆ 𝐷𝑎𝑦𝑠 𝑜𝑓 𝑖𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 𝑜𝑛 𝐻𝑎𝑛𝑑 = 𝐼𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 𝐶𝑂𝐺𝑆 ∗ 365 TOTAL DAYS ON HAND AVERAGE 2018/19 RAW MATERIAL 22 WORK IN PROGRESS 2 FINISHED PRODUCTS 14 TOTAL DAYS ON HAND RAW MATERIAL WIP FINISHED GOODS CASE COMPANY 22 2 14 COMPETITOR 1 17 4 13 COMPETITOR 2 17 2 11 https://www.odette.org/news/story/collaborative-forecasting-guidelines https://www.odette.org/news/story/collaborative-forecasting-guidelines 𝑑0 ≠ 0 FAI ∶= ∑ 𝑎𝑖 𝑚𝑎𝑥 {0; 1 − |∆𝑖| 𝑑0 } 𝑛 𝑖=0 𝑑0 = 0 FAI ∶= ∑ 𝑎𝑖 𝑤ℎ𝑒𝑟𝑒 𝐼𝑛 = {𝑖|∆𝑖= 0; 𝑖 = 1, … , 𝑛} 𝑛 𝑖=0 𝑓𝑜𝑟 𝐼 = 0: 𝐹𝐴𝐼 ∶= 0 ∑ 𝑎𝑖 𝑛 𝑖=1 |∆𝑖| ≠ 0 𝑊𝑇𝑆 ∶= ∑ 𝑎𝑖 𝑛 𝑖=−1 ∆𝑖 ∑ 𝑎𝑖 𝑛 𝑖=1 |∆𝑖| ∑ 𝑎𝑖 𝑛 𝑖=1 |∆𝑖| = 0 𝑊𝑇𝑆 ∶= 0 Equation 1, Mean absolute percentage error Equation 2, overall equipment effectiveness Equation 3, days of inventory on hand Figure 1, an illustration of the literature background part of the report. Figure 2, A 2x2 matrix to decide whether a variation should be red or green (Ekberg, Raju, Bahsson, & Jirholm, 2019) Figure 3, An example of the dashboard with alerts (Ekberg, Raju, Bahsson, & Jirholm, 2019) Figure 4, six examples of MAPE-profiles Figure 5, a framework of resource and capacity planning (Jonsson & Mattsson, 2009) Figure 6, Difference between forecast and customer-order driven production (Berry, Vollmann, & Whybark, 1988) Figure 7, aggregating multiple forecasts hides variations Figure 8, relations of s&op, capacity and demand (Jonsson & Mattsson, 2009) Figure 9, model of qualitative data collection through interviews and observations Figure 10, the analysis model for the mixed method analysis Figure 11, a model of the production line of product A Figure 12, the sales forecast planning process Figure 13, The Planning horizons at the Case company Figure 14, The current deviation model from the case company Figure 15, Actual production capacity after variations in production rate Figure 16, Size comparison between the supplier, case company and the customer Figure 17, Size comparison between two buying companies of the same supplier, where one of them are the case company https://studentchalmersse-my.sharepoint.com/personal/linhanse_net_chalmers_se/Documents/Master%20Thesis%202020,%20Meridion/Masters_Thesis_Meridion_2020.docx#_Toc39490354 Figure 18, % of average FAI and WTS (over or underestimations) in the highest volume items from an average of 2w, 4w and 6w prior to delivery Figure 19,four MAPE profiles connected to one item Figure 20, a volume weighted MAPE-profile which is higher than the average MAPE profile Figure 21, a volume weighted MAPE-profile which is higher than the average MAPE profile Figure 22, a graph of the individual items MAPE-profiles within the item group Y Figure 23, a graph of the demand of four items within an item group Figure 24, a graph of the total demand of the products within the item group Figure 25, A model of the production flow within the case company Figure 26, an example of a dashboard controlling the production model Figure 27, an Example graph generated by the model of capacity, demand and buffer levels Figure 28, 3 different scenarios of a demand change and production capacity Figure 29, areas of demand with high stress for the production Figure 30, the different levels of capacity at the suppliers Figure 31, an example scenario of combining mape-profiles and bias Figure 32, limits for when action has to be taken for a planner Figure 33, an example of a dynamic deviation model Figure 34, comparison of the model estimations and related mape profiles Figure 35, five items plotted as MPE and MAPE Figure 36, the relationship between capital tied up and service level (Jonsson & Mattsson, 2009) Table 1, The results of how how forecast kpis affect supply chain visibility (Ekberg, Raju, Bahsson, & Jirholm, 2019) Table 2, planning functions and general attributes (Jonsson & Mattsson, 2009) Table 3, the six big losses of oee Table 4, Item Groups FAI and WTS sorted by FAI, lowest to highest Table 5, Item Groups FAI and WTS sorted by WTS, highest to lowest Table 6, examples of shift levels and forecasted demand Table 7, backtesting of the model estimates compared to actual demand Table 8, total days on hand of inventory of the case company Table 9, competition comparison on days on hand