Presentation - MIT Center for Transportation and Logistics
Transcription
Presentation - MIT Center for Transportation and Logistics
Analysis of an International Distribution Hub for Fast Moving Consumer Goods Conducted by: Sebastian Ortiz Duran & Rich Hawks Advised by: Dr. Stephen Graves Agenda • Review of Consumer Co.’s Current Import Supply Chain • Proposed Solutions • Mixed Integer Programming Model of Current Supply Chain with Hub Capability • Solution Analysis • Sensitivity Analysis 2 Key Insights • Hub capability in an international supply chain may provide higher flexibility and lower cost. • Logistics infrastructure and economies of scale impact international transportation costs more than total distance travelled. • Postponing final allocation to regional distribution centers until containers reach a hub enables the use of more accurate forecasts. 3 Review of Current Supply Chain • 11 Countries each with one Regional Distribution Center (RDC’s) • Each imports from 3 Production Countries – Argentina – Brazil – Mexico • Average Lead Time 35 days Standard Deviation 16 days 4 Current Network Design Node j Guatemala Node i Argentina El Salvador Honduras Nicaragua Costa Rica Brazil Panama Colombia Venezuela Mexico Ecuador Peru Bolivia 5 Company Proposed Network Design Node j Guatemala El Salvador Node i Honduras Argentina Nicaragua Node z Brazil Mexico Costa Rica Panama Panama Colon Free Trade Zone Hub Colombia Venezuela Ecuador Peru Bolivia 6 Postponement Benefits Mean Absolute Percentage Error (MAPE) decreases as the days between shipment and actual sales decrease Postponing final allocation to regional DCs until containers reach a hub enables the use of more accurate forecasts Example 44% MAPE 66 days Argentina 33% MAPE 35 days Argentina Costa Rica 4 days Panama Hub Costa Rica 7 Alternative Proposed Network Design Node j Guatemala El Salvador Node i Honduras Argentina Nicaragua Node z Brazil Mexico Costa Rica Panama Panama Colon Free Trade Zone Hub Colombia Venezuela Ecuador Peru Bolivia 8 Alternative Proposed Network Design Node j Guatemala El Salvador Node i Honduras Argentina Nicaragua Node z Brazil Panama Colombia Free Trade Zone Hub Mexico Costa Rica Colombia Venezuela Ecuador Peru Bolivia 9 Expected Hub Results Advantages 1. Reduced Inventory 2. Reduced Order Variability from Risk Pooling 3. Increased Container Utilization 4. Increased Flexibility Disadvantages 1. Increased Transportation Distance and Costs 2. Increased Handling Costs 3. Increased Overhead Costs 4. Increased Risk of Theft/Damage 10 Mixed Integer Model Setup Objective Function: Minimize Total Relevant Costs Total Relevant Costs = Transportation + Holding + Hub Handling Transportation Costs = Direct + Hub Inbound + Hub Outbound Holding Costs = Pipeline + Cycle + Safety Stock Binary Decision Variables H ij Hub Flow M 1 when node i ships to node j through node z (hub) 0 when node i ships to node j directly Hub Safety Stock 1 when hub holds safety stock 0 when hub holds no safety stock (Cross Dock) Constraint: Maintain Current Fill Rates at RDC’s 11 Transportation Costs Transportation Costs = Annual Shipments x Cost per Shipment Variable Description D Annual demand in pallets from node i at node j L Pallet Capacity per Container U Container Utilization R Cost per Shipment from node to node, includes inland freight, FOB expenses, ocean freight, duties on shipping Direct 3 ∑ i =1 ⎛ (1 − H ij ) * Dij ⎞ ⎜ * Rij ⎟ ∑ ⎜ ⎟ Li * U ij j =1 ⎝ ⎠ 11 Hub Inbound ⎛ 11 ⎞ ( H * D ) ⎜ ⎟ ∑ ij ij 3 ⎜ j =1 ⎟ * R ∑ iz ⎟ ⎜ L *U i =1 i iz ⎜⎜ ⎟⎟ ⎝ ⎠ Hub Outbound ⎞ ⎛ 3 ( H * D ) ⎟ ⎜ ∑ ij ij 11 ⎜ i=1 * Rzj ⎟ ∑ ⎟ ⎜ Lzj *U zj j =1 ⎟ ⎜ ⎠ ⎝ 12 Pipeline Holding Costs Pipeline Holding Cost = Lead Time x Landed Cost x Carrying Charge Variable Description W Lead Time from node to node B Landed Cost per Pallet (Transfer Price + Shipping Cost) C Carrying Charge at Hub ⎛ ⎛ Dij ⎞ ⎞ ⎜ ⎟⎟ ⎟] C z * ∑ [ Biz *Wiz * ∑ ⎜ H ij * ⎜⎜ ⎟ i =1 j =1 ⎝ ⎝ 365 ⎠ ⎠ 3 at RDC from Hub 11 at RDC Direct 3 ⎛ Dij ⎞ ⎞ ⎛ 3 ⎛ Dij ⎞ ⎞ ⎛ ⎜ ⎟ ⎜ ⎟⎟ ⎟] ⎟⎟ + ∑ ⎜⎜ Bij *Wij * (1 − H ij ) * C j * [⎜ Bzj *Wzj * ∑ ⎜⎜ H ij * ∑ ⎜ ⎟ 365 ⎠ ⎠ ⎝ i =1 ⎝ 365 ⎠ ⎟⎠ j =1 i =1 ⎝ ⎝ 11 13 Cycle Holding Costs Cycle Holding Costs = (Order Quantity / 2) x Landed Cost x Carrying Charge Variable Description Q Order Quantity (dependant on demand volume) B Landed Cost per Pallet C Carrying Charge at Hub Qiz ⎞ ⎛ C z * ∑ ⎜ Biz * ⎟ 2 ⎠ i =1 ⎝ 3 at RDC from Hub Qzj ⎛ C j *[⎜⎜ Bzj * ∑ 2 j =1 ⎝ 11 at RDC Direct ⎞ 3 Bij * Qij ⎟⎟ + ∑ ] 2 ⎠ i =1 14 Safety Stock Holding Costs Safety Stock Holding Cost = Std Dev of Forecast Errors x k factor x Landed Cost x Carrying Charge Variable Description C Carrying Charge B Landed Cost per Pallet σ Standard Deviation of Forecast Errors over a Variable Lead Time k # of Standard Deviations worth of Safety Stock required to at Hub provide a given Fill Rate 3 at Hub C z * ∑ (Biz * kiz * σ iz ) i =1 at RDC from Hub 11 ∑C j =1 j at RDC Direct * [Bzj * k zj *σ zj+∑ (Bij * kij *σ ij )] 3 i =1 15 Safety Stock Considerations σ at RDC’s when M = 0 (Cross Dock) ‐ Lack of protection at hub requires RDCs’ include Standard Deviation of Forecast Errors over a Variable Lead Time to the Hub (Eppen, 1981) ‐ Due to risk pooling and postponement benefits, only a fraction α of this variability must be accounted for, where α is approximately 1/(# countries served) LT2 Production Country Destination 1 Destination 2 LT1 Destination 3 Hub Destination 4 k when Q is small relative to σ ‐ Backorders are double counted and safety stock is inflated unnecessarily σ ≥2 ‐ When , correct k is that which satisfies the following equation (Silver, 1970): Q G (k ) − G (k + Q σ )= Q σ * (1 − E ) 16 Model Example Solution Solution Analysis $30 • Flowing all channels through hub in either Panama or Colombia increased total supply chain costs $25 Supplt Chain Cost (millions) • Hybrid Panama achieved >4% total supply chain savings $20 Hub Handling $15 $10 Holding $5 Shipping $‐ Baseline Hybrid Panama Total Hub Panama Total Hub Colombia 18 Transportation Cost Analysis • Panama’s infra‐ structure efficiency reduces transportation cost per mile $24,000,000 Bubble Size= Cost per Mile $2.5 $22,000,000 Transportation Cost • Each of scenarios tested increased distance travelled over current supply chain $20,000,000 Baseline $18,000,000 Hybrid Panama $1.7 $16,000,000 $14,000,000 $1.9 Total Hub Panama Total Hub Colombia $1.6 $12,000,000 7,000,000 8,000,000 9,000,000 10,000,000 Miles Travelled 19 Sensitivity Analysis – Key Variable Impact on Savings 100% 80% Change in Total Cost Savings 60% Avg. Yearly Demand 40% Fill Rate 20% Avg. Lead Time Hub Handling Cost 0% ‐20% ‐10% 0% 10% 20% Cost per Mile ‐20% ‐40% ‐60% Change in Variable Value from Baseline 20 Conclusions • Direct shipping not always more economic • Hub allows postponement of allocating decision to distribute product with a better forecast • Hybrid Solution provides reduced overall cost • Potential for greater savings in reduced overstock and write‐offs • Hub capability provides supply chain flexibility to quickly adapt to changes in the environment 21 Thank you! Questions? 22