Analytics’ FocusAnalytics Operations EngineeringSelected Results (Confidential)Presentation OutlineThe ProductBusiness NeedVCD Business StrategyEclipse Project DesignSystems with no buffersSystems with no buffersBusiness Situation StatementBuffer AnalysisBuffer AnalysisBuffer AnalysisBuffer AnalysisBuffer AnalysisImportant System PhenomenaApplication of TechnologyDevelopment of TechnologyKnown Two Stage ModelsDevelopment of DecompositionSolution ApproachImplementation ResultsShort-Term BenefitsLong-Term BenefitsSummaryThe Use of Operations Research Techniques to Improve the Design of a Hewlett-Packard Printer Production LineMitchell BurmanAnalytics’ FocusAnalytics Operations EngineeringQuantitative Analysis– Inventory Optimization– Forecasting & Scheduling– SKU Rationalization– Distribution Efficiency– Quality Engineering– Supply-Chain ManagementOperationalTacticalStrategicSkill LevelLow HighFocus AreaOperationalOperationalTacticalTacticalStrategicStrategicSkill LevelLow HighFocus AreaSelected Results (Confidential)Fleet ManagementFleet Management• United $30m Savings• Flight Options $6m Profit Increase• Active Aero 15% EBITDA Increase*ProductivityProductivity• HP $280m Increase• Truck Mfr. >10% Increase• Baxter $5m Scrap ReductionSKU Rationalization & PricingSKU Rationalization & Pricing• Unisource >$10m Profit Increase* • Broder $30m Inventory Reduction• Superior >10% Profit IncreaseSupply Chain Design & MgtSupply Chain Design & Mgt• IBM $750m Savings• Samsonite $25m Savings • Boeing $9m Inventory ReductionFleet ManagementFleet Management• United $30m Savings• Flight Options $6m Profit Increase• Active Aero 15% EBITDA Increase*ProductivityProductivity• HP $280m Increase• Truck Mfr. >10% Increase• Baxter $5m Scrap ReductionSKU Rationalization & PricingSKU Rationalization & Pricing• Unisource >$10m Profit Increase* • Broder $30m Inventory Reduction• Superior >10% Profit IncreaseSupply Chain Design & MgtSupply Chain Design & Mgt• IBM $750m Savings• Samsonite $25m Savings • Boeing $9m Inventory Reduction* Estimates based on work in progressSelected Analytics Clients Acme Steel Active Aero (Berkshire) Alcan Atkins (Parthenon) Baxter Healthcare BICC General Boeing Boston Scientific Broder Brothers (Bain) Cambridge Industries (Bain) DESA (HIG) Endres (JMH Capital) Flight Options FluidSense GMAC Harley-Davidson Hewlett Packard ICI (Imperial Chemical) 3i (Advent) Intel J.M. Huber Corporation Johnson & Johnson Karsten Textilia (Brazil) Kraton Polymers (TPG) Lockheed Martin Lone Star Industries Motorola Northrop Grumman Nutraceutical (Bain) Palm Coast Data (Tinicum) Primedica Raytheon/Flight Options Samsonite (Bain) SAPPI Paper Superior Essex Team Products (HIG) Thomson Legal & Regulatory Truck Mfr. (BCG) Unisource (Bain) US Can (Berkshire) Western Mining (Australia) Wolverine (Parthenon)Presentation Outline Business Need  Application of Technology  Benefits SummaryThe Product HP Desk Jet Printer >$1,000,000,000 in demand Encroaching competition (Canon)Business Need HP background Vancouver Division (VCD) background VCD Business requirements “HP Way”VCD Business Strategy Productivity improvement to meet business needs Automation for attaining productivity objectivesEclipse Project DesignOriginal SystemFinishedAssemblyTightly Coupled ProcessEmptyPalletSubassemblyCellSubassembliesProductBaseProcessStationBuzacott ApproximationE = ------------ = 0.98MTTFMTTF + MTTR1Single MachineMTTF = Mean Time to FailMTTR = Mean Time to RepairBuzacott ApproximationE = ------------ = 0.98MTTFMTTF + MTTR1Single MachineMTTRMTTFSystems with no buffers= 0.0204E = ---------------- = 0.32811 + 100 MTTRMTTF100 MachinesMTTF = Mean Time to FailMTTR = Mean Time to RepairMTTF = Mean Time to FailMTTR = Mean Time to RepairMTTRMTTF= 0.0204Single MachineE = ------------ = 0.98MTTFMTTF + MTTR1E = ---------------- = 0.32811 + 100 MTTRMTTF100 MachinesBuzacott ApproximationSystems with no buffersOneYearLaterBusiness Situation Statement Eclipse sub-assembly performance insufficient Simulation not working Method needed to quickly assess alternative system architecturesBuffer Analysis Early paradigm:– Large BuffersEfficiencyTotal System Buffer Spaceoo-bufferlimit0-bufferlimitBuffer Analysis Early paradigm:– Large BuffersEfficiencyTotal System Buffer Spaceoo-bufferlimit0-bufferlimitBuffer Analysis Early paradigm:– Large BuffersEfficiencyTotal System Buffer Spaceoo-bufferlimit0-bufferlimitBuffer Analysis Early paradigm:– Large Buffers JIT– Buffers UnnecessaryEfficiencyTotal System Buffer Spaceoo-bufferlimit0-bufferlimitBuffer Analysis Early paradigm– Large Buffers JIT– Buffers Unnecessary Reality– The truth is in betweenEfficiencyTotal System Buffer Spaceoo-bufferlimit0-bufferlimitImportant System Phenomena Long multi-stage process Failing machines Different operation times  Finite storage spaceMachineBufferApplication of TechnologyGoal: Rapidly predict performance of a production system– Many alternatives (695,784,701,952)– Long execution times make simulation awkwardDevelopment of Technology Analytical method needed for rapid system evaluation Difficulty: Large state space makes brute force numerical solution of Markov chain impracticalKnown Two Stage Modelsµ1MTTF1MTTR1µ2MTTF2MTTR2NDevelopment of DecompositionAnalysis of large system by breaking it into smaller systemsSolution Approach Analytical systems performance model was developed Hardware changes Critical operational targetsCost of Changes ~$1,000,000Implementation Results New system architecture System validation New methodology for system optimizationShort-Term Benefits Eclipse system throughput increased 25% $280 million in incremental revenue Incremental pen and media revenue Lower production costsLong-Term Benefits More robustness and predictability for all new manufacturing system designs Faster production ramps Greater throughput and productivityPotential long-term financial benefitsfar outweigh the short-termSummary Realized $280 million incremental revenue Captured incremental market share at critical time which increased Leveraged the methodology into future system