| dc.description.abstract | The inventory management question of when and how much to order has presented an
important problem to optimize the product flow in the supply chain while minimizing
costs and lost sales. Such an inventory policy, however, depends heavily on accurate
demand forecasts. This thesis investigates a case study of a coffee supply chain within
a coffee brand which consists of one central warehouse and two coffee stores. To be
specific, this thesis focuses on coffee demand forecast by time-series methods and uses
them as input to the simulation-based optimization method of the inventory system. Two coffee items to be considered are: Dark Roast Espresso Dolce (ED) and Dark
Roast Viennese Blend (VB) which are the best-sellers throughout the supply chain. Three time-series methods considered are Single Exponential Smoothing, Holt
Exponential Smoothing, and ARIMA. The only inventory policy considered is periodic
review policy (T, S) which fits the daily operations at the stores. Inventory system is
simulated by ARENA software in 14 months and validated against the actual data
during the same period, then optimized by OPTQUEST tool integrated in ARENA. Future scenario of demand and corresponding inventory policy are also considered and
optimized. The results show that the simulation reflects the real world sufficiently, and
the optimal system should have saved up to 61,000,000 VNĐ in the past 14 month | en_US |
