presentasi isiem dsg

Upload: faishal-makarim-kamali

Post on 03-Apr-2018

228 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/28/2019 Presentasi Isiem dsg

    1/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    TEACHING DESIGN OF EXPERIMENT FORINDUSTRIAL STATISTICS LABORATORY CLASS

    Dedy [email protected] / [email protected]

    Jakarta, 2007

    mailto:[email protected]:[email protected]:[email protected]:[email protected]
  • 7/28/2019 Presentasi Isiem dsg

    2/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Problem :

    Books on design of experiments (DOE) havemany exercises at the end of chapters that givestudents practise in the analysis of completed experiments, but students often receive littleexperience in DOE

  • 7/28/2019 Presentasi Isiem dsg

    3/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Objective :

    to share some ideas about teaching DOE by giving examples of simple experiments for laboratory class that integrates practice in

    designing realistic experiments, running theexperiments, and also practice analyzing data insuch a way that is easy to learn, fun, challenging,and memorable.

  • 7/28/2019 Presentasi Isiem dsg

    4/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Where do the ideas come from ?

    Antony J, N Capon. Teaching Experimental Design Techniques toIndustrial Engineers. Int. J. Engng Ed . Vol. 14, No. 5, , 1998 pp.335343

    Antony J, N Capon. Some key things industrial engineers shouldknow about experimental design. Logistics Information Management Volume 11 Number 6 1998 pp. 386 392Hunter WG. 101 Ways to Design an Experiment, or Some Ideas

    About Teaching Design of Experiments . 1975.http://williamghunter.net/articles/101doe.cfm Lye LM. Tools and toys for teaching design of experimentsmethodology . 33rd Annual General Conference of the CanadianSociety for Civil Engineering. 2005Martinez-Dawson R. Incorporating Laboratory Experiments in anIntroductory Statistics Course. Journal of Statistics Education Volume

    11, Number 1, 2003.http://www.amstat.org/publications/jse/v11n1/martinez-dawson.html

    http://williamghunter.net/articles/101doe.cfmhttp://williamghunter.net/articles/101doe.cfm
  • 7/28/2019 Presentasi Isiem dsg

    5/30

  • 7/28/2019 Presentasi Isiem dsg

    6/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Some examples of simple experiments from Lye LM(2005), Mackisack M (1994) and Hunter WG (1975):

    No Response or output Factors or input variables

    1 taste (maximize), un- popped kernels

    (minimize) of Microwave popcorns

    brand, time, power, height (on bottom or raised)

    2 virus scan time RAM cache, program size, operatingsystem

    3 time to boil water pan type, burner size, cover, amount of water, lid on or off, size of pan

  • 7/28/2019 Presentasi Isiem dsg

    7/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    No Response Factors

    4 distance paper aeroplaneflew

    design, paper weight, angle

    5 blending time for soy beans blending speed, amount of water,temperature of water, soaking time

    before blending6 height of cake oven temperature, length of heating, amount

    of water

    7 length of rubber band before it broke

    brand of rubber band, size, temperature

  • 7/28/2019 Presentasi Isiem dsg

    8/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Industrial Statistics LaboratoryIndustrial Eng. Dept Trisakti UniversityLab. Form for DOE Module :

    Names:____________________ Date:_______________ Period:______________ Purpose of experiment :

    Hypothesis:

    Materials:

    Procedures:

    Results and Analysis using MINITAB :

    Conclusion:

  • 7/28/2019 Presentasi Isiem dsg

    9/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Case 1 : Pop corn experiment

    For example, we may want to investigate theinfluence of pop corn brands on the proportion of un-popped kernels (minimize). We use

    completely randomize design or without blockingof experimental unit for this single factor experiment. There are tree levels for brand (A, Band C) and tree replications for each lavel. We

    use one hundred kernels for each trial and 3,5minutes to make pop corn on stove.

  • 7/28/2019 Presentasi Isiem dsg

    10/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    We wish to test hypotheses about thetreatment means, and our conclusion willapply only to the factor levels considered inthe analysis

    Ho : 1 = 2 = . = a H1 : i j for at least one pair (i,j)

  • 7/28/2019 Presentasi Isiem dsg

    11/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Randomization using Minitab :

    Run order Un-popped kernels

    proportion

    Brand C

    Brand C

    Brand B

    Brand A

    Brand C

    Brand B

    Brand ABrand A

    Brand B

  • 7/28/2019 Presentasi Isiem dsg

    12/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Picture 1. Three brands of pop corn

    Picture 2. Processing of pop corn

  • 7/28/2019 Presentasi Isiem dsg

    13/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    The results of experiment :

    Picture 3. Popped and un-

    popped kernelsfrom tree brands

    Run order

    Un-poppedkernels

    proportion

    Brand C 0,04

    Brand C 0,05

    Brand B 0,11

    Brand A 0,00

    Brand C 0,08

    Brand B 0,13

    Brand A 0,03

    Brand A 0,03Brand B 0,08

  • 7/28/2019 Presentasi Isiem dsg

    14/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Minitab Output :One-way ANOVA:Pooled StDev = 0,02134Source DF SS MS F PBrand 2 0,011356 0,005678 12,46 0,007Error 6 0,002733 0,000456Total 8 0,014089

    S = 0,02134 R-Sq = 80,60% R-Sq(adj) = 74,13%Individual 95% CIs For Mean Based onPooled StDev

    Level N Mean StDev ---+---------+---------+---------+------Brand A 3 0,02000 0,01732 (-------*-------)Brand B 3 0,10667 0,02517 (-------*------)Brand C 3 0,05667 0,02082 (------*-------)

    ---+---------+---------+---------+------

    0,000 0,040 0,080 0,120

  • 7/28/2019 Presentasi Isiem dsg

    15/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Interpreting the Results :

    The small p-values for the brand (p =0.007) that lower than ( 0.05) suggestthere is significant effect of brand onproportion of un-popped kernels. Individual95% confidence interval for mean of threebrand suggest that brand A has significantly

    difference with brand B.

  • 7/28/2019 Presentasi Isiem dsg

    16/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Case 2 : Time to boil water experiment

    For example, we may want to investigate theinfluence of pan size and cover on the time inminute to boil water. We use general full factorialdesign and completely randomize design or without blocking of experimental unit. Single.There are two-level for each factor and 3replications for each combination. Dimensions of small pan is 14 cm for diameter and 10 cm for height. Dimensions of medium pan is 18 cm for diameter and 11 cm for height. Volume of water is600 ml.

  • 7/28/2019 Presentasi Isiem dsg

    17/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Randomization using Minitab :RunOrder pan size cover time (minutes)

    1 medium no2 small yes

    3 small yes

    4 small no

    5 medium no

    6 medium yes

    7 medium yes

    8 small yes

    9 small no

    10 medium yes

    11 medium no

    12 small no

  • 7/28/2019 Presentasi Isiem dsg

    18/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Picture 4. Small pan, medium

    pan and glasscover

    Picture 5. Medium pan with cover on

    stove

  • 7/28/2019 Presentasi Isiem dsg

    19/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    The results of experiment :

    RunOrder pan size cover time(minutes)

    1 medium no 3.92

    2 small yes 3.07

    3 small yes 3.00

    4 small no 3.53

    5 medium no 3.63

    6 medium yes 3.83

    7 medium yes 3.20

    8 small yes 3.22

    9 small no 3.75

    10 medium yes 3.9811 medium no 3.50

    12 small no 3.60

  • 7/28/2019 Presentasi Isiem dsg

    20/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Minitab Output :General Linear Model: time (minutes) versus pan size, coverFactor Type Levels Valuespan size fixed 2 small, mediumcover fixed 2 yes, noAnalysis of Variance for time (minutes), using Adjusted SS for

    TestsSource DF Seq SS Adj SS Adj MS F Ppan size 1 0.29768 0.29768 0.29768 4.90 0.058

    cover 1 0.22141 0.22141 0.22141 3.65 0.093pan size*cover 1 0.20021 0.20021 0.20021 3.30 0.107Error 8 0.48560 0.48560 0.06070Total 11 1.20489S = 0.246374 R-Sq = 59.70% R-Sq(adj) = 44.58%Unusual Observations for time (minutes)

    time

    Obs (minutes) Fit SE Fit Residual St Resid7 3.20000 3.67000 0.14224 -0.47000 -2.34 R

    R denotes an observation with a large standardized residu

  • 7/28/2019 Presentasi Isiem dsg

    21/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    M e a n o f t

    i m e

    ( m i n u

    t e s )

    mediumsmall

    3.70

    3.65

    3.60

    3.55

    3.50

    3.45

    3.40

    3.35noyes

    pan size cover

    Main Effects Plot (data means) for time (minutes)

  • 7/28/2019 Presentasi Isiem dsg

    22/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Interpreting the Results :

    The small p-values for the pan size (p =0.058) and cover (p = 0.093) that lower than (0.10) suggest there is enoughsignificant effect of pan size and cover ontime to boil water. Interaction of pan sizeand cover is not significant. Mean plot of

    rensponse suggests that small and cover (yes) give lower time to boil water .

  • 7/28/2019 Presentasi Isiem dsg

    23/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Case 3 : Painting experiment

    We may want to investigate the influence of paintingmethods (dipping, spray and brush) and brand of paint onvisual quality using 10 point scale. As a standard for point10, we use sample product from PT. Safira Tumbuh

    Berkembang (wooden toys producer). General fullfactorial design and completely randomize design used inthis experiment. There are tree levels for painting methodsand two levels for paint brands and three replications for each combination. This experiment is part of

    Manufacturing Industrial Design Lab. in IndustrialEngineering Department Trisakti University. The endproduct is wooden toy train.

  • 7/28/2019 Presentasi Isiem dsg

    24/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Our products :Wooden toy train

  • 7/28/2019 Presentasi Isiem dsg

    25/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Picture 7. Materials and equipmentsof paintingexperiment

    Picture 8. Preparation of experimental unit

  • 7/28/2019 Presentasi Isiem dsg

    26/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Picture 9. Dipping method for painting

    Picture 10. Spraying methodfor painting

  • 7/28/2019 Presentasi Isiem dsg

    27/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    easy to learn, fun, challenging, and memorable.

  • 7/28/2019 Presentasi Isiem dsg

    28/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    Picture 11. Drying

  • 7/28/2019 Presentasi Isiem dsg

    29/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University

    CONCLUSIONS

    1. This paper has shown the benefit of employing asystematic approach to simple experimentation usingDOE, rather than utilising a trial and error approach

    2. The paper has also illustrated some simple experiments

    that can be used as a powerful teaching and learningtool in industrial statistics laboratoy.

    3. These simple experiments will form a student foundationfor studying DOE for the wider application in real-lifesituations or using other techniques of DOE like

    response surface, taguchi or mixture experiments.

  • 7/28/2019 Presentasi Isiem dsg

    30/30

    Industrial Statistics Laboratory Industrial Engineering DepartmentTrisakti University