Introduction Optimizing Estate Space Needs Through Centralization and Consolidation ThuBa T. NGUYEN, Sijie HO, Mark GOH, Robert de SOUZA Motivation • Land for industrial estate development is scarce in Singapore. • Need methodology for industrial warehouse clustering and freight distribution within a cluster. Introduce approaches to design capacity of Multi-User Distribution Centre (MUDC) for a cluster, and to simulate and evaluate feasibility of an Estate- wide Goods Mover System (EGMS). Objectives • To formulate mathematical model for determining the capacity of automated mover distribution centre in an urban context. • To develop simulator to evaluate the feasibility of EGMS. Organization • Mathematical and simulation approaches: o Problem definition o Model development. o Findings o Summary I. Mathematical model: Capacity Planning for Multi-Storeyed Shared Facilities II. Simulation model: Simulate Operations of EGMS Main assumptions : 1. Raw materials (RM) & finished-goods (FGs) are shipped first-in-first-out. 2. One type of raw material and FG for a given user. 3. Deterministic arrivals of RM to MUDC. 4. Deterministic order quantities of RM from the factories. 5. Only one product is stored on each rack. 6. A product can be stored only in a storey. 7. Outbound rate (number of pallets/day) of the FGs corresponds to the outbound rate of the raw materials requested for delivery from MUDC to users. Figure I.1: Storage structure on one storey of MUDC R1 R2 R3 Rt u RT 0 RT 1 F1 F2 F3 Ft u C1 C2 C3 Ct u FT 0 FT 1 CT 0 CT 1 ….. ….. ….. ….. ….. ….. ….. ….. ….. FT 2 CT ….. CT 2 ..... ..... Planning horizon Figure I.2: Network flow for MUDC resource planning Problem definition Objective function: Minimize construction cost which includes cost of building the floors and opening the goods lanes Subject to: • Constraints of building a floor, • Constraints of opening a lane, • Constraints of allocating a lane/ storey to each type of product and to each user, • Constraints of delivery time and shipping time for RM and FGs respectively. Model development 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0 1 2 Uniform Triangular Normal Poisson Delay time (days) Portion of RM delivered (%) 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0 1 2 3 Uniform Triangular Normal Poisson Delay time (days) Percentage of finished goods shipped (%) Figure I.3: Pattern of delay time for delivering raw materials Figure I.4: Pattern of delay time for shipping FG Number of storeys Required capacity (number of lanes/storey) Delay time (days) Input combinations of maximum delays (days) RM 0 2 5 0 2 5 0 2 5 FG 0 3 5 0 3 5 0 3 5 Distribution of inbound & outbound RM rates Uniform Min 5 5 5 22 22 22 0 0 0 Median 5 5 5 23 22 23 0 0 0 95%ile 5 5 5 23 23 23 0 1 1 Max 5 5 5 23 23 23 0 3 3 Triangular Min 5 5 5 22 22 22 0 0 0 Median 5 5 5 23 22 22 0 0 0 95%ile 5 5 5 23 22 22 0 1 1 Max 5 5 5 23 22 22 0 2 3 Normal Min 5 5 5 22 22 22 0 0 0 Median 5 5 5 23 23 22 0 0 0 95%ile 5 5 5 23 23 23 0 1 1 Max 5 5 5 23 23 23 0 2 2 Poisson Min 5 5 5 22 22 22 0 0 0 Median 5 5 5 22 22 22 0 0 0 95%ile 5 5 5 23 22 22 0 1 2 Max 5 5 5 23 22 22 0 1 3 Table I.1: Sensitivity analysis on change in service level • Maximum number of storeys needs to be built regardless of changing the service level. • Same day delivery should be implemented and can be achieved with the presence of automated MUDC. Table II.1: Consolidated demand/supply data (PE) manufacturing facilities Figure II.2: Schematic layout of manufacturing facilities in Wenya industrial estate Figure II.3: EGMS estate simulation model for Precision Engineering (PE) cluster using SIMIO software Figure II.1: Schematic goods flow in EGMS model Problem definition Model development Figure II.5: RM count in system (average + maximum) Figure II.6: FG count in system (average + maximum) Figure II.4: ASRS goods count (average) with capacity reached at X = 2.1 • Marked improvement via significant reduction in the average number of both RM and FG lingering in EGMS estate as the number of operating AGVs increased. • As ASRS proposed design parameter allows for 32,000 pallets space, the maximum increment in demand/supply is capped at X=2.1. • Linear increments in the goods count in system as the demand/supply rate increases. • Dedicated AGVs to single tenanted facilities to minimise number of FGs piling up at their locations Objective: Investigate feasibility of proposed set-up of EGMS estate Table II.2: Simulation results with Automated Guided Vehicles (AGVs) in operation Capacity Planning for Multi-Storeyed Shared Facilities: • Systematic way to design required resource and to allocate users’ products to minimize land used. • An automated MUDC can save up to 80% of land used. • JIT delivery and shipment should be considered when planning required capacity of an automated MUDC Implementation of EGMS: • Provide useful insights for EGMS estate operational planning and supports important decision making on actual implementation of EGMS estate Limitations: • Current MUDC model considers only one product; putting multiple products on a rack and dedicated zones for certain types of products require further investigation. • EGMS estate model analysis can be studied with more operational data and exact designated plot Conclusion