Financial Management DFM at Nike

We conduct the first analysis of our model based on only financial management of closed loop supply chain.

Parameters:

Lead Time  3 days Demand D Neg. Binom(7.91391, 0.000040812) Life Cycle  90 days ELC Capacity X 400.000 units

Stock Capacity S 400.000 units Vehicle Capacity V 400.000 units

MTBF1 Exprnd(30) MTBF2 Exprnd(30)

MTTR1 1 day MTTR2 1 day

U1 Max. Capacity 200.000 units U1 Max. Capacity 200.000 units

cu1 10 cu2 4

Cw 2 Cs 2

Cr 1 Cz 1

Cl 250 Cq 1

Table 4.1. Parameters of DFM- Nike case study

As clarified earlier, we search for the optimal value of percentage, to understand the relation between the percentage of satisfied demand returning to manufacturing and the average cost values of 1000 periods. In Nike case, we have changed some of the parameters from the example in section 2.2. to make the model more realistic. For instance, lead time is chosen to be 3 days. Useful life cycle of a product is chosen to be 90 days, which may be controversial for several products, however, majority of products have a season of 3 months, which is why we used 90 days for life cycle. Furthermore, manufacturing inventory capacity S, vehicle capacity V and ELC capacity X are chosen to be 400,000 units. This number is rather high considering the mean demand value for Nike (193.903/day), however, since demand data is fitted to negative binomial distribution, high variance in data is a factor to be considered, which led us to have a considerably high stock capacity in different locations. Modeling the failure is also considered differently compared to example; we chose both manufacturing facilities to have a failure with exponential distribution having the mean value of 30 days. On

36

the other hand, repair time of a manufacturing facility is set to 1 day deterministically to simplify the model. As can be seen from table 4.2., average cost per day seems to have relatively low values between 35% and 45%.

In the example from section 2.2 we report that optimal values are achieved with a return percentage of 20%, which creates a discrepancy to Nike case. That difference could be attributed to the difference in demand distribution (deterministic vs. Negative Binomial) and other parameter differences such as manufacturing failure behavior.

P 5% 10% 15% 20% 25%

Avg.Cost 10888193 9026655 7996250 5916036 5438089

P 30% 35% 40% 45% 50%

Avg.Cost 4843432 4029212 4792494 6184476 10488585

P 55% 60% 65% 70% 75%

Avg.Cost 28445240 32212978 50482659 62389849 74662954

P 80% 85% 90% 95% 100%

Avg.Cost 84370140 92465156 108347413 125065063 134583031

Table 4.2. Average cost values based on percentage of satisfied demand returning to remanufacturing.

Figure 4.1. Average cost incurring due to return percentage of sold products (nbin)

Average Cost

Return Percentage

37

Figure 4.2. Average cost incurring due to return percentage of sold products (Norm) As can be seen from two above graphs, Models that adopt Negative Binomial and Normal distribution seem to have similar patterns toward cost structure. Figure 4.1. suggests that approximately 40% return appears to create the most advantageous results for demand data that follow Negative Binomial Distribution.

When we deploy Poisson Distribution to represent demand data, we obtain the following graph of cost vs return percentage:

Figure 4.3. Average cost incurring due to return percentage of sold products (Poi)

Average Cost

Return Percentage

Average Cost

Return Percentage

38

Holding each parameter constant, when we only change the demand distribution, we see differences between the cost structure of model with Negative Binomial distribution and model with Poisson distribution. This indicates that demand distribution is a slight

determinant for closed-loop supply chains only for the same data. That is to say, if we fit the same demand data to several distributions, based on those distributions, optimal percentage would change. However, demand distribution does not necessarily cause a pattern as in the above-mentioned graph.

Figure 4.4. Average Cost incurring due to return percentage (Poi, lambda=100.000) As can be seen from both figures 4.3 and 4.4, parameter (lambda) may change the cost pattern.

Nevertheless, we can draw conclusions considering the same demand data; for industries subject to Negative Binomial demand behavior, return percentage seems to be advantageous for a relatively tight interval compared to Poisson Demand behavior. For instance, if there were a legislation enforcing companies to remanufacture products with a 50% return rate of sold products, companies with a Negative Binomial demand distribution with short lead times would severely suffer from the legislation.

As can be noticed in Figure 4.1., Figure 4.2., Figure 4.3., when returned percentage p increases, average cost values tend to decrease to a certain level. Returned inventory level

Average Cost

Return Percentage

39

provides benefit up to a certain point (40%-50% return) due to increased F2 utilization, which reduces the number of lost sales because manufacturing warehouse (MW) is fulfilled more steadily due to both manufacturing facilities running.

Figure 4.5. Lost Sales occurring based on return percentage(x) (nbin)

In figure 4.1., dramatic increase in average cost could be explained by the accumulation of returned product inventory r, which seems to have prevented the number of lost sales as it can be seen in figure 4.5.

Figure 4.6. R inventory level(y) of returned products (nbin)

Lost Sles

Return Percentage

R inventory Level

Return Percentage

40

As can be seen from figure 4.6., inventory of returned products is fully utilized in

remanufacturing until 40-50%. We see an increasing pattern of returned products’ inventory after 50%, which appears to constitute an increase in overall cost as well (see figure 4.1.) Figure 4.5. and Figure 4.6. provide insights to how lost sales and R inventory level impact the overall cost structure that was shown in Figure 4.1. There is intuitive relation between the number of lost sales and number of returned inventory; once the lost sales start decreasing, returned products’ inventory starts increasing.

Although number of lost sales decrease with increased percentage p, inventory holding cost increase for manufacturing warehouse, returned products’ warehouse and ELC warehouse.