Quantifying the Impact of Deployment Practices on Interplant Freight Volatility Kurn Ma Manish Kumar
Agenda Project Motivation Research Problem Methodology Results Conclusion & Future Considerations
Project Motivation Over 5,000 trucking companies (~400,000 trucks) went out of business in 2012. There are about 8,000 fewer trucks available nationwide on any given day. (ATA) Lack of replacement of the retiring drivers
Sponsor Company 60% Typical Day in the Supply Chain Cost Drivers Description (‘000) per day Orders 1 Shipments 2 Tenders 3 Cases picked 325 Cases moved in warehouse 6,000 Potential Lane Combinations 23,000 Pallet-Miles 30,000 Logistics Raw Materials Manufacturing 60% 12%
Sponsor Company Direct Plant Fulfillment Plants Customer & Warehouse Near Plant Warehouses (Full Pallet and Picked Pallets) Customer Store (Consumer) Direct Plant Fulfillment Distribution Centers (Full Pallet and Picked Pallets) DC Shipments Distributors Direct to Consumer
Monthly shipments from Plant to DC- # of pallets Thesis Problem Identify levers that impact this volatility: endogenous & exogenous How can we mitigate this volatility through internal decisions? Recommend deployment practices to reduce this volatility Monthly shipments from Plant to DC- # of pallets
Methodology Data Analysis Simulation Model Forecasted Demand Actual Demand Production Data Simulation Model Discrete Event Simulation Platform: Visual Basic in MS Excel
Project Scope One year time horizon Single plant to single DC 15 product groups analyzed (44% of overall freight volume) Truckload volume analyzed at weekly level
Assumptions Entirely pull-based deployment from plant All products have same MAPE (variable across scenarios) All products have same reorder and target levels 7% inventory holding cost
* Production Schedule Demand Simulation Production Variability Formulation Production Schedule 2 week inventory position Based on system DOS target Demand Simulation Random Distribution based on MAPE Back Calculation of daily forecast error * Production Variability Distribution fit Production Factor *
Framework
Results: Unmanaged scenario Model outputs consistently show that bi-weekly deployment generates lowest volatility It provides 100% stock service level at the lowest average inventory at DC Changes in forecast accuracy do not impact the volatility (only size of shipments) The randomness in production output is very low to have any impact Daily Bi-weekly Weekly # of weekly truckloads for each deployment frequency
Results: Unmanaged scenario It provides 100% stock service level at the lowest average inventory at DC Changes in forecast accuracy do not impact the volatility (only impacts the size of shipments) The randomness in production output is very low to have any impact Bi-weekly Weekly Daily Inventory at DC for each deployment frequency
Results: Managed scenario Eliminates the need for spot market trucks Loads are delayed and evenly distributed the following week
Conclusion Further Research Bi-weekly deployment schedule performs better both with respect to shipment volatility and inventory holding Management of shipments by delaying them and forcing them to be exactly as per forecasted loads provides desired service level Change in demand accuracy does not impact the volatility Further Research