System Dynamics Models Of Replenishment Decision For Relief Supplies

Thesupplementofemergencymaterialsisveryimportantinthereliefoperationsafterunexpected incidents, and the efficiency of supplement may influence the rescue effectgreatly. With the frequent occurrence of catastrophes and unexpected events in recentyears world widely, the academe focus on relief supply chain continues to rise. But thereis still lack of quantitative description of the extreme uncertainty in the environment ofrelief operations. And the mutual influence between the dynamic operating environmentand the replenishment decision is under studied. Therefore, this dissertation studied theemergency supplies after unexpected incidents, by choosing the System Dynamics (SD)approach and taking the Wenchuan earthquake as the example. Research topics include:1. By introducing the dynamic transport delay and information delay, two SD modelwere implemented to describe the traditional commercial supply chain in stable operationenvironmentandthereliefsupplychainindynamicenvironment. Thedifferencesbetweenthesetwotypesofsupplychainweretestedanddiscussedthroughmultiplesimulations, aswell as their different pattern of responses to the dynamic lead time and random demand.Simulation results and the further analysis confirmed that they have different structure ofinformation feedbacks, hence the behaviors of the supply systems are various. And theimpact of the dynamic delays to the systems is greater than the impact of simply raisingthe delay level.2. The dynamic road capacity was described by a SD model. The researches in post-seismic rapid damage assessment of road networks were referenced, and the intensity dataof Wenchuan earthquake and its aftershocks were used. Several elements were introducedinto the model, including the random secondary disasters, the distance from the affected-village to the epicenter, the mountain geomorphology, the effective road repair capacity.A number of shortcomings of the SD software in dealing with the delays of substantialoscillationwerealsoconsideredandcompensated. Byusingtheparametersofthedifferentgeological conditions of the affected-village, the model was able to simulate the dynamictransportconditionsinthepost-seismicreliefoperations. Thusthequantitativedescription of the dynamic transport delay was achieved.3. A replenishment decision model of the demand-stable supplies was presented, andthe bottled water was chosen as the specific materials. The performances of three types ofstock planning strategies and four lead time prediction methods were tested in simulation,aswellastheirimpulseresponsestodifferentlevelsofdynamicdelays. Simulationresultsand further analysis indicated that the transport delay and the replenishment decisions areenhancing each other due to the heavily loss of road capacity. So the instability of thesupply system is exacerbated. In addition, the information delay has different influencesover the decision-makers at relief head-quarter (the upstream) and the disaster-affectedareas(thedownstream). Adecisiontreewasalsoproposedtohelpdecision-makerschoosethe appropriate stocking strategies.4. A replenishment decision model of the demand-time-varying supplies was pre-sented, and the routine medical supply was chosen as the specific materials. Several ele-ments were introduced into the model, including the casualties of continuous aftershocksand secondary disasters, the cure of the wounded, the outgoing of the wounded and nor-mal victims, the feature of complement the shortage, prediction on both the lead time andthe demand. The structure of the system and simulation results were discussed in the per-spective of system theory, then the impulse-response pattern of the supply system wereanalyzed in detail.5. The subjective adjustments on the prediction results of the lead time were ana-lyzed in a SD model, and an adaptive adjustment method was implemented based on theanticipation of the trends of the lead time. The fixed-value adjustment with different typesof judgmental intervention (optimistic and pessimistic) and the adaptive adjustment withdifferent magnitudes ware tested in multiple scenarios. Other control parameters includedthe level of information delay and the prediction method. The simulation results indicatedthat the fixed-value adjustment with a pessimistic attitude reduces the uncertainty in reliefactivities, and a mixed solution such as optimistic attitude enhanced by the prediction oflead time and further adaptive adjustment may obtain more balanced performance.