1、Experiment Design ProcessDetermine problemsDetermine ObjectivesPreparation for experimentDesign ExperimentConduct Experiment & Collect dataAnalyze dataInterpret resultsVerify Predicted results第1頁，共21頁。cementDetermine problem In conventional processes, workers apply cement on uppers or bottoms manual

2、ly. For decades, different workers apply cement with different quantity. Due to the use of brush to apply cement, the cement thickness is also related to brushas below photo. With automation project, we have opportunities to make the thickness more uniform. Since we hypothesize that uniform thicknes

3、s could make the cementing boundaries more intimate . In this way, we believe that bonding strength would be more stably passed bonding standard.第2頁，共21頁。Objective of experiment To find the optimal spray system setup for target thickness to meet requirements as below so as to ensure stable bonding s

4、trength.Stable average cement thickness(properly dried after oven) Ideal cement profileLowest deviation of thickness Stable bonding strength by NIKE standard第3頁，共21頁。-Ideal profile for cementing thicknessCurrentcementObjective of experimentcementIdealAdhesive thicknessBonding score (Kg/cm2)The best

5、range2.5/ 3 Kg/cm2第4頁，共21頁。NoProcessCheck pointsPhoto1Mark the measure check point on fiberglassFollow the pattern2Measure the thickness of the of each check pointFiberglass shall be pressed while measuring3Auto spray on fiberglass along a straight line4Oven curing50 , 210 sec5Cool the fiberglass fo

6、r 3 minutes6Measure the thickness of each check points Fiberglass shall be pressed while measuring-cement thickness measurement procedurePreparation for experiment第5頁，共21頁。Name/ Location:Otis Chen / Lasted upperA3 Project Name:Auto cementingProcess:Spray stabilityModel/ Size:NADate/ Revision:01/28/2

7、013Experiment Objective/ Goal: To test lasted upper cement thickness with different nozzle distance and tank pressureIdentify Output (Y) & Input (X): (Use Process Map, FMEA, Fish Bone Diagram, Pareto, C & E Metric, etc. to identify inputs and outputs)Nozzle distanceAutoCementingThickness avgXYCement

8、 viscosityNozzle pressureRobot speed2900 cps30 psi800 mm/sIdentify Experiment Conditions:Fixed VariablesValue(Use ANOVA, Correlation, Hypothesis Testing, etc. to identify critical Inputs and highlight in red Experiment Environmental Conditions:Temperaturehumidity25.254%VariablesValueTank pressureThi

9、ckness std Cement widthIdentify Experiment Plan:Tank pressure18Design Factor(Current setting)LowHigh36Nozzle distance (cm)24第6頁，共21頁。Result第7頁，共21頁。Data Analysis Cement WidthCement width=9.15+1.3833*nozzle distance+1.15 tank pressure +0.45*nozzle distance*tank pressureP-value Tank pressure interacti

10、on effect of 2 factors Data Analysis Cement Width第9頁，共21頁。Data Analysis Average cement thicknessAverage thickness=22.858-7.642*nozzle distance+3.942 tank pressureP-value 0.200第10頁，共21頁。Main Effect PlotSlope of Two factors are both high two factors have significant main effects on average thickness P

11、areto chartEffect on cement width:nozzle distance Tank pressure interaction effect of 2 factorsData Analysis Average cement thickness第11頁，共21頁。Data Analysis Cement thickness deviationCement thickness deviation=8.817-3.55*nozzle distance-1.2*tank pressure + 3.3* nozzle distance*tank pressureP-value i

12、nteraction effect of 2 factorsTank pressureData Analysis Cement thickness deviation第13頁，共21頁。Interpret Result:1. Significant factors for each output are listed as below 2. For average thickness, as nozzle distance increases, more atomized cement is released to the air, so the cement thickness decrea

13、ses.3. For thickness deviation, nozzle distance has higher effect. As nozzle distance increases, the deviation decreases. We believe that the nozzle air pressure would push more cement to 2 cement boundaries as the nozzle get closer to the surface. Nozzle distanceTank pressureInteraction of bothCeme

14、nt widthAverage thicknessThickness deviation : output increases as factor increase : Output decreases as factor increase第14頁，共21頁。Optimization:Expected cement profile -Cement width =10 mm -cement thickness =25 um -cement deviation 8 um第15頁，共21頁。Verify Predicted results:Conclusion:Precise pressure re

15、gulator is valid for pressure control.Its valid to dip spray gun into water to reduce the clog problem. However, the clog condition could happen as the spray gun is exposed to the air for more than 5 minutes.We could find optimal value within the low/high level of factors with good R-sq. 第16頁，共21頁。N

16、ame/ Location:Otis Chen / Lasted upperA3 Project Name:Auto cementingProcess:Cement bondingModel/ Size:NADate/ Revision:02/21/2013Experiment Objective/ Goal: To test lasted upper cement bonding strength for different materials with different cement thicknessIdentify Output (Y) & Input (X): (Use Proce

17、ss Map, FMEA, Fish Bone Diagram, Pareto, C & E Metric, etc. to identify inputs and outputs)materialsAutoCementing2D bondingXYCement viscosityNozzle pressureRobot speed2900 cps30 psi800 mm/sIdentify Experiment Conditions:Fixed VariablesValue(Use ANOVA, Correlation, Hypothesis Testing, etc. to identif

18、y critical Inputs and highlight in red Experiment Environmental Conditions:Temperaturehumidity25.254%VariablesValueTank pressure Identify Experiment Plan:Design Factor(Current setting)LowHighNozzle distance (cm)1060Nozzle distance 3 cmSpecimen prep. By std processes第17頁，共21頁。 All bonding passes exce

19、pt 1 point failed for synthetic leather due to material failure.b1Spray direction15 cm15 cm1.0 cmRBSynthetic/Leather/Suede/Meshb2b3b4b5Remarks: b1 means the bonding of 1 section in a specimen which is divided into 5 sections as belowFindings:第18頁，共21頁。Cement DOE experience share :Spray cement profil

20、e (width, avg thickness, thickness dev.) would vary due to: (a) cement viscosity air temperature (b) cement pot life (c) clog condition (tube, filter, spray gun.) (d) cement tube spec (e) spray gun spec (f) tank pressure / nozzle pressure / spray gun distance So, it seems that its essential to have

21、check cement profile for each day or even each shift.2. Bonding strength is certainly a key indictor for cementing projects. Its hard to control all the bonding processes to be homogeneous but we could analyze to see if the bonding issue is due to cement spray issue or others.3. Gage R&R is critical

22、 to ensure measurement method to be reliable.4. Clog issue could happen at any time, please monitor spray often.5. Its important to discuss with engineers about experiment design.第19頁，共21頁。Future Offshore DOE deployment :Q: Whats the plan to apply DoE method to optimize the parameters in offshore factory(sometime may different from RD phase) and how to co-work with factory IE?A: (1) Auto