Tham khảo tài liệu 'supply chain 2012 part 19', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | New Measures for Supply Chain Vulnerability Characterizing the Issue of Friction in the Modelling and Practice of Procurement 531 costs respectively. Each point in this surface represents the average friction for n 280 experiments across all factor levels of s1 and D for given levels of p and e and this response surface is very closely modeled by the expression z average friction - r2 .843 . friction Figure 3. Response surface formed by average friction levels across environmental factor levels p and e for two-stage test bed. Observation of the three-stage test results suggests similar relationships between these parameters and the broader effect of chain friction. In the case of the three-stage results the response surface is closely modelled by z average friction - - r2 .879 . Furthermore the percent increase between three-stage chain friction instances and their corresponding two-stage counterparts describes a highly similar response surface with respect to p and e as shown in Fig. 4. Further insight can be gained by looking at these two environmental parameters in isolation. Fig. 5 shows the distinctly inflammatory effect of the inventory parameter e the higher the levels of this factor the higher the observed levels of friction. Fig. 5 also indicates the intuitive result that this phenomenon is compounded by the addition of another level of planning as the disparity between the average friction in the two-stage and three-stage experiments widens with increasing values of e. 532 Supply Chain Theory and Applications Figure 4. Response surface formed by average percent increase in friction levels when comparing three-stage to two-stage experimental instances across environmental factor levels p and e. c c inventory parameter e 2-stage 3-stage Figure 5. Average friction levels across factor level e for the two-stage versus three-stage results n 3 360 instances for each data point . Fig. 6 illustrates .