|
Inventory management under partial backlog and storage constraints using simulation optimization approach |
|---|---|
| รหัสดีโอไอ | |
| Title | Inventory management under partial backlog and storage constraints using simulation optimization approach |
| Creator | Pijittra Kularbwong |
| Contributor | Pham Duc Tai, Advisor |
| Publisher | Thammasat University |
| Publication Year | 2568 |
| Keyword | Inventory management, Simulation–optimization, Empirical distribution, (Q, R) policy, Demand uncertainty, Lead time variability, Excel VBA, Grid search, Golden section search, Pharmaceutical supply chain |
| Abstract | This research develops a simulation-optimization framework for inventory management in a single-location system with partial backlog and storage constraints. Using empirical data from a Thai hospital, the study examines fast-moving, medium- moving, and slow-moving product categories with distinct demand patterns. The methodology employs a continuous-review (Q, R) inventory policy implemented through Microsoft Excel with VBA macros, making it accessible without specialized software. The simulation incorporates empirical demand distributions from 92 days of historical data, stochastic lead times, and dual storage constraints. A partial backlog policy assumes 50% of unfulfilled demand results in backorders while 50% becomes lost sales, reflecting realistic customer behavior during stockouts. Two optimization techniques are evaluated: grid search and golden section search. Grid search systematically evaluates all parameter combinations to identify the global optimum, while golden section search efficiently converges toward optimal solutions with reduced computational effort. Statistical validation determined 23 to 32 replications were required across product categories to achieve 95% confidence levels. Results show both methods successfully identify effective inventory policies. For fast-moving products, grid search achieved 1,678.10 THB versus 1,698.36 THB for golden section search, only 1.21% difference. Both methods converged to identical solutions for medium-moving and slow-moving products. Golden section search demonstrates significant computational efficiency, requiring 63 to 77% fewer iterations and 44 to 60% less computation time. Cost analysis reveals holding costs dominate total costs at 50 to 80%, indicating where inventory reduction initiatives yield greatest savings. This research integrates multiple realistic constraints: empirical demand distributions, stochastic lead times, partial backlog behavior, and dual storage constraints. The Excel-based implementation provides organizations a cost-effective optimization tool with actionable insights for inventory management. |
