Evaluating the Effectiveness of Pre-Positioning Policies in Response to Natural Disasters
EXPERIMENTAL DESIGN
Objectives
The intent of this research is to understand the impact of implementing a local pre-positioning strategy when the stored relief supplies may be near the event center. Specifically, we are interested in the following:
1. Quantifying the benefit of pre-positioning in terms of cost and unmet demand quantity;
2. Understanding the sensitivity of the solution to the initial supply level and storage capacities at the supply nodes.
Summarizes the test cases con- structed to address the objectives posed above. The model parameters affected by each experi- ment are shown as well.
The benefit of implementing a pre-posi- tioning strategy is determined by comparing the difference in cost and unmet demand quantity between the pre-positioning policy obtained from the model, and no pre-positioning policy.
RESULTS AND DISCUSSION
Experiment 1: Base Model SLP Policy Evaluation
Figure 2 illustrates the capability of the supply networktosatisfydemandineachofthedefined scenarios. The fraction of unmet demand per scenario is lower when a pre-positioning policy is used. The largest benefits correspond to the scenarios where supply is reduced as a result of damage to the supply nodes. The results also indicate increasing the forecasted demand with no resulting damage to supply yields no differ- ence in the fraction of unmet demand between the 2 policies (Scenarios 5, 9, 13, 17). The total cost associated with using the pre-positioning policy is $120,630 consisting of a first-stage pre-positioning cost of $4,760. The expected cost associated with not using a pre-positioning policy is $242,200 (Table 6). Utilizing the pre- positioning policy results in a 49.8% reduction in expected system costs and improves the expected demand fulfillment. Figure 3 shows the resulting supply and demand network when bothnopolicyandthepre-positioningpolicyare implemented. The figure illustrates that using
Experiment 2: Initial Inventory Variation
In an actual event, pre-positioned supply is potentially damaged and there is some change in demand. With this, it would be impractical to assume that the actual demand would be met. Increasing initial inventory would yield a higher resultant supply able to fulfill more ac- tual demand. Conversely, decreasing the initial supply would have a two-fold impact on the ability of the network to satisfy demand. First, the difference in initial supply and forecasted demand would already ensure there will be unmet demand. Second, the changes in demand and available supply due to the event would further hinder responsiveness.
Using the parameters of experiment 1.2 as a base, the initial inventory values per node are changed to determine the effect on the pre-positioning solution.
Experiment 3: Supply Node Capacity Variation
Experiment 3 considers an increase in the stor- age capacity of each node. The initial supply level is defined from experiment 1.2. Two cases are considered: (1) a general network-wide increase of 20% for each supply node's storage capacity; and (2) increasing the capacity of the safest nodes. Relative to the base model, both experiments are increased by the same percent- age.
Attachment:- IJORIS_Final_Proof.rar