【正文】 Effect of pressure = + + + + + + = 4 4 To pare the mean values where pressure is high and low EN/FAD 109 0015 169。 Ola Johansson, 1999 Ericsson Quality Management Institute Creating amp。 Ola Johansson, 1999 Ericsson Quality Management Institute Replicated Factorial Designs with Text Variables ? Problem : You want to study the effect of oil viscosity, temperature, and a special additive on engine wear. ? Procedure: You are looking for the treatment that produces the least wear. You replicate the experiments so you will have two runs for each treatment and will be able to test all possible interactions. All experiments can be run in a single day. ? Factor levels : Viscosity (30,40)。 Ola Johansson, 1999 Ericsson Quality Management Institute 2k Factorial Design Description ? Specialised design consisting of K factors, each observed at only two levels ? Twolevel full factorial designs can have up to 7 factors ? Twolevel fractional designs can have up to 15 factors (Resolution IV) or 47 factors (Resolution III, Plackett Burman) ? Can be quantitative, such as two levels of temperature or pressure ? or qualitative , such as two machines or two locations ? a full factorial 2K is efficient efficient design requires the least number of runs of any full factorial design Because of its simplicity and efficiency, it is the experimental design most monly used in industry. Both full and fractional factorial designs can be used to screen important factors in early stages of experimentation 2K Factorial Design Factor A, B, C = 2 Levels 1 2 3 4 5 6 7 8 B A C EN/FAD 109 0015 169。M Nuisance variables Control variables Constant variables Independent variables Influential Measurable Controllable Low B High B Low A High A Response Low B High B Low A High A Response Interaction No interaction Low A High A 5 6 4 3 Low A High A 5 4 3 Low B High B Interactions Control variables Trial number D of E Array Qualitative amp。 effect’ plexity using DoE Reduce variability and target process further using SPC and SIC 。 Ola Johansson, 1999 Ericsson Quality Management Institute Planning for Success Structured Problem Solving List the Problems Select the Problem Analyse the Problem Decide the Facts Gather Data Interpret Data Devise Solution Present Findings Implement Monitor Action Evaluation amp。 Ola Johansson, 1999 Ericsson Quality Management Institute Planning for Success designing / doing trials Objectives Define the problem clear, concise and measurable People Engineering production technical / finance Ramp。 Additive ( A, B)。 Time (10, 30 sec)。 Ola Johansson, 1999 Ericsson Quality Management Institute A Two Level Factorial Design ? Coefficients determined by dividing effect by 2 ? Constant found by the mean of all responses Pressure Heat Time Pr*He Pr*Ti He*Ti Pr*He*Ti + + + + + + + + Coefficient = Effect / 2 Mean Value Of All Responses (+++++++) / 8 + Modelling the response variable Response = + * Pressure + * Heat * Time + * Pressure * Heat + * Pressure * Time * Heat * Time * Pressure * Heat * Time Linear model with two levels. Curvilinear effect