VIETNAM PLANT FINAL REPORT Active-Activa One Piece Fire Loss Reduction BB Project #AP 2165 1 Asia Pacific B. Executive Summary 1. Problem Description Baseline statistics of production loss: • • • • FF Loss: 26.7% (Y1) RF Loss: 29.1% (y2) TTL: 45.3% (y) Major defects includes 70.5% over total FF defects: -Clay crack (CC)- loss of total FF pcs (y) -CDT & PHT loss of RF loss … * Shortage Supply vs Demand 15.7% volume around 200 A pcs/month Project targets FF loss : 18 % RF loss : 21% TT loss : 29.9 2 Asia Pacific Baseline study - FFL and RFL& Shortage Demand in 2009 D M A I C Trend of RFL Trend of FFL 100 50 80 40 60 % % 30 20 40 10 20 0 Jan Feb Mar May Jun Jul Aug Sep Oct Nov 0 Dec Jan BSL FFL Target FFL FF Loss% Feb Mar May BSL RFL Jun Jul Target RFL Aug Sep Oct Nov Dec RF Loss % Trend of TTL 80 % 60 40 20 0 Jan Feb Mar May Target TTL 3 Jun Jul BSL TTL Aug Sep Oct Nov Dec Total Loss% Asia Pacific Active-va Fire Loss Reduction - Vietnam D M A Key Initiative: Total Fire Loss Reduction for Category Heroes Active & Activa (Vietnam Operations) I C AP2165 Key Metrics: Primary metric (Y): First Fire Loss = First fire loss (pieces) / total inspection (pieces) Baseline = 26.76% Target = 18% (AOP = Baseline) Secondary metric (y1): Re-fire Loss = Re-fire loss (pieces) / total inspection in re-fire (pieces) Baseline = 29.19% Target = 21% (AOP = Baseline ) TTL Baseline = 45.37% Target = 29.90% (vs AOP = 36.2%) = (Clay loss + FF loss + RF loss) / total casting Measure: 31/03/10 Control: 31/07/10 RF loss by defect RF Loss - Pareto Active -va fire loss 90 100 80 30 80 70 25 60 50 40 40 30 20 20 10 0 RF loss Defect Count Percent Cum % 15 CDT 38.27 42.5 42.5 Matt 9.16 10.2 84.2 LE 8.85 9.8 94.1 PRT 3.60 4.0 98.1 Other 1.75 1.9 100.0 100 20 5 RF loss 29.19 43.0 43.0 FF loss 26.76 39.4 82.4 CL loss 11.94 17.6 100.0 0 FF loss by defect FF Loss - Pareto 25 10 PHT 28.40 31.5 74.1 80 15 60 10 40 5 20 0 FF loss defect Count Percent Cum % CCT 16.96 70.5 70.5 LE 3.02 12.6 83.1 PF 2.01 8.4 91.4 WH 1.12 4.6 96.1 Other 0.94 3.9 100.0 Percent 20 0 Fire loss Count Percent Cum % Percent 60 Count Business Errors: - Active –Activa are key models with high/stable demand. They are strategic products with big benefit. Due to their high losses, supply of these models could not support with high demand. - Sale revenues loss in EBIT about $ 117k annually. We have shortage around 200 pcs of Active-va / month which are caused by high fire loss. - Price increase pressure to maintain GM will impact in lost sales & revenue Barriers: - New model & complex - Lack of skilled workers – Design not complete- Mold technical & mold condition not stable Benefits: (in K USD @ AOP rate) Direct Saving = 214k USD (98 + 117) - Production losses = 98k USD - Sales losses = 117k USD (short supply models) Cash Flow Improvement = 214K USD Indirect Saving = Nil Program Budget: = Nil 4 Analyze: 31/05/10 Close: 30/09/10 Count Define: 01/03/10 Improve: 30/06/10 Count Completion Schedule: DMAIC 0 High Fire Losses on Active-Va one piece cause high cost and unstable supply Project Leader: Nguyen Phuong (BB) Project Champion: Prasanna MBB: Pittaya Team Members: Do Anh Thai, Nguyen Ngoc Cang, Nguyen Lam , Nguyen T Phong , Nguyen Thanh Hai,,Huynh Anh Tuan ,Le Ngoc Hung Asia Pacific Active-va Fire Loss Reduction - Vietnam D Project # AP2165 Project Title Active-Va One piece Fire loss reduction BB Nguyen Phuong Champion Prasanna M A I MBB Techarungnirun Pittaya Benefit Category (K USD) Detailed Computation by Category DIRECT BENEFIT Direct Saving from reduction loss = ((Total pcs FF* cost *( Y BSL- Ytarget) + Total pcs RF *cost *( YBSL - Ytarget))*11.5 month =97690$ A. Cost Takeout = 97.6 C B. Incremental Margin (1) Net Increased Capacity = 116.6 (2) New Product Net Sales = Direct saving from Incremental margin= (Total pcs selling * price selling *%margin *% improve)11.5 month =116649$ C. Inc Cost Exec = AOP Exchange rate = 17,819.65 VND / USD Total Direct Saving Project = Direct saving from reduction loss + Direct saving from Incremental margin=214340$ Total Direct Benefit = 214.3 CASH FLOW BENEFIT Net Cash = 214.3 INDIRECT BENEFIT A.Cost Avoidance = B. Cost Savings = C. Net NVA Reduction = D. Non Financial Benefit= 5 Asia Pacific B. Executive Summary 2. Solution Strategy Statistical study high-loss models and defect locations Focus team assignment by model Beware of new comers (PF, HL,WH,LE) Apply Six Sigma tools to identify vital factors (Xs) for process control Key factors to study: Slip formula, Aging time Mold life , mold condition and design variation Casting system, tools, skill and procedure variations Daily communications among key processes 6 Asia Pacific B. Executive Summary Support Activities Daily loss court meeting and trouble shooting - FF major defects - Defect by models - Relevant defect locations - Focus team assignment by product group - Trouble shooting database and follow up - Defect traceability road-map and data collection - Defect location map recording by location 7 Asia Pacific B. Executive Summary 3. Summary of Six Sigma Tool Application Measurement Phase CTQ flow chart Process flow chart Process mapping (KPIVs, KPOVs) C&E Matrix GR&R (A-RF-L) Baseline data (FFL, RFL, TTL, % defects) Analysis Phase FMEA summary SPC charts Multi-vari studies( T-test, One way Anova, …) Paretos and location 8 Asia Pacific B. Executive Summary 3.Six Sigma Tool Application Improvement Phase GR&R (FF result: A,defect CC & LE) Data mining experiment (interactions of Humidity & Mold life , OPT, Aging time…). DOE & RSM Visual communications (Defect location updates) Training (caster master,) Casting procedure modification 9 Asia Pacific B. Executive Summary 3.Six Sigma Tool Application Control Phase A. Control Plan B. Defect tracing and feedback daily C. Slip SPC and control sheet D. Humidity & temperature record E. POKAYOKE & Visual factory F. Casting procedure follow-up G. Mold status record H. Casting procedures follow-up Performance trend charts - FFL,RFL, TTL (Ys) - % of cast loss, CC, PF and new comers (ys) 10 Asia Pacific B. Executive Summary 4. Results and Conclusions Active –va one piece performance has been improved from w4.Apr 2010 and achieved stable results through Jul.2010 Trend Chart RF Loss Time FF loss % BSL FF Ju n' 10 Ju l'1 W 0 1. Au g' 10 ar '1 0 Ap r'1 0 M ay '10 M Ja n' 10 Fe b' 10 O ct '0 9 No v'0 9 De c'0 9 Au g' 09 Se p' 09 40.00 30.00 20.00 10.00 0.00 W 1. Au g'1 0 Ju l'1 0 Ju n'1 0 M ay '10 Ap r'1 0 M ar '10 Fe b'1 0 % 40.00 30.00 20.00 10.00 0.00 Ja n'1 0 Time Target FF RF loss % BSL RF Target RF Trend Chart TT Loss 11 ug '1 0 W 1. A Target % Ju l'1 0 Tim% e BSL ay '1 0 Ju n' 10 TTL% M pr '1 0 A ar '1 0 M Fe b' 10 60.00 50.00 40.00 30.00 20.00 10.00 0.00 Ja n' 10 % % Trend Chart FF Loss Asia Pacific Monthly saving & YTD summary OUT PUT INCREASE DIRECT & INDIRECT SAVING USD 60000.0 Total A pcs increase saving 500 Transfer A 0 Jan'1 Feb' Mar' Apr' May' Jun'1 Jul'1 0 10 10 10 10 0 0 Total A pcs increase saving 129 34 42 55 210 216 129 Transfer A 556 221 576 498 741 835 723 40000.0 Total Direct saving $ $ $ 1,000 Indirect Saving $ 20000.0 0.0 Nov' Dec' Jan' Feb' Mar' Apr' May' Jun' Jul' Total Direct saving $ 56023005176620153304 322131471178 7048 Monthly Indirect Saving $ 27601738153619332141 263619923814 5338 Monthly Early Project Saving 12 Asia Pacific D M A I C II. MEASUREMENT PHASE A. Process Description B. Process Map Summary C. Cause & Effects Fishbone Diagram D. Cause & Effects Matrix Summary E. GR& R 13 Asia Pacific A. Process Description D M A I C Process Flow Chart Generator Air Compressure Slip Making Critical Process Mold Making Model/Mold design Mold fit Surface quality Thixo Residue Slip pumping Mold installation Circulating time Slip line design Leakage Mold maintenace Bench design CASTING Mold prep. Cleaning Drying Mold assembling Bench prep. Pouring Slip rate Time Casters Demolding Punching Predrying Draining Feedback Feedback Sponging Green finishing & inspection Drying Trap glaze White inspection Pass Casting Environment : Humidity & temperature Fail Loss Fai Loss Spraying Pass Loss Glaze dryness Loading Firing Rework Glost inspcection VC Plumbing Vietnam bird-view • Plant Operation: started Jan. 1997 • Headcount: approx. 300 employees • Production schedule: 2 shifts x 6 days/week • Plant Capacity: 14 - Output: avg. 35000“A” pcs/ month Asia Pacific Process Map FF Summary D 15 M A I C Asia Pacific CAUSE & EFFECT FISHBONE DIAGRAM D Measurements + Master meter pressure Materials + Raw materials Glaze quality + Mold Life + Slip Quality I C Personnel + Plastic Mold + Humidity Meter M A Caster ,Glost inspector , Kiln operator, Sprayer + Operator skill + Duels Km/l % FF Loss -Humidity -Temperature - Dirty -Bast Wash Environment 16 + Training procedure + Kiln equipment + Standard methods sheet + Spray gun + Slip Formula + Case Mold + handling + casting tool Methods + Glost equipment Machines Asia Pacific C&E Metric Summary FF D 17 M A I C Asia Pacific Process Map RF D 18 M A I C Asia Pacific MEASUREMENT STUDY-GR&R D M 19 A I C Asia Pacific GR&R (A-RF-L) D M A I C Need Improve more MS is Acceptable < 10% 20 Asia Pacific III. ANALYSIS PHASE D M A I C A. FFL% with CCL% correlation B. High volume defect by Location focus C. CCL defect happen high mold life D. Caster performance on FFL defect & impact of Aging time to FFL E. CDT & PHT defect loss high ratio on RFL 21 Asia Pacific FFL% vs CCL % correlation D M A C High correlat ion CCL vs FFL Regression Analysis: FFL % versus CCL % The regression equation is CC Defect happen with high ratio >70%FFL on Active one piece in 2009 Need focus to improve CCL to reduce FFL I FFL % = 5.50 + 1.29 CCL % Predictor Constant CCL % Coef SE Coef T P 5.496 2.373 2.32 0.033 1.2947 0.1680 7.70 0.000 S = 3.56428 R-Sq = 76.7% R-Sq(adj) = 75.4% Analysis of Variance Source DF SS MS F P 1 754.06 754.06 59.36 0.000 Residual Error 18 228.67 12.70 Total 19 982.74 Regression Unusual Observations 22 Obs CCL % FFL % Fit SE Fit Residual 8 11.0 28.194 19.755 0.885 8.439 St Resid 2.44R Asia Pacific CCL % BY HIGH VOLUME LOCATION ON PRODUCT D CCL happen so much on O.A.P.P2,H location A I C Picture to illustrate Product location A H H O M A P2,P Reduce CCL on these location O,A,P,H to improve CCL. 23 Asia Pacific CORRELATION CCL % BY LOCATION VS MOLD LIFE D M A I C Two-Sample T-Test and CI: CCT Loss, Mold Life Two-sample T for CCT Loss ML N Mean StDev SE Mean High 5 56.40 7.83 3.5 Low 5 22.00 4.00 1.8 Difference = mu (High) - mu (Low) Estimate for difference: 95% CI for difference: 44.5073) 34.4000 (24.2927, T-Test of difference = 0 (vs not =): T-Value = 8.75 P-Value = 0.000 DF = 5 Significant Mold life high & Low different CCL CCL by location A, H,P happen on mold Which one have high mold life . Need focus to find out what’s mold life is best setting for improvement 24 Asia Pacific CCL DEFECT WITH CATSERS PERFORMANCE D M A I C CCT Loss Defect by Caster One-way ANOVA: 1070, 1139, 1187, 1293, 1463, 1659, 1666, 1906, 1907 Source DF SS MS F P Factor 8 380.1 47.5 1.38 0.254 Error 25 862.9 34.5 Total 33 1243.0 S = 5.875 R-Sq = 30.58% R-Sq(adj) = 8.36% Individual 95% CIs For Mean Based on Pooled StDev CCT Loss By caster on Mar’10 Level N Mean +---------+ StDev 1070 4 11.000 5.824 (--------*-------) 1139 4 11.675 4.167 (--------*-------) 1187 4 14.825 3.686 1293 4 7.025 4.225 (--------*--------) 1463 4 7.300 3.032 (-------*--------) 1659 4 9.445 7.852 (-------*--------) 1666 4 9.025 6.989 (--------*--------) 1906 3 ------) 18.667 10.599 1907 10.400 3.940 3 ---------+---------+--------- (-------*--------) (---------*--(---------*---------) ---------+---------+--------- +---------+ CCT loss between casters have big variation .Specially on ID 7.0 1659 ,1666,1659,1187 & 1097 with big variation & high mean.On 28.0Pooled StDev = 5.875 Mar’ these ID caster continue with CCT loss high %. Need improve Caster performance 25 14.0 21.0 Asia Pacific CCL DEFECT WITH SLIP AGING TIME D M A I C Test for Equal Variances for CCL Aging time F-Test Test Statistic P-Value 2.5 Lev ene's Test Test Statistic P-Value 2.9 0 Aging time 29.62 0.000 2 4 6 8 95% Bonferroni Confidence Intervals for StDevs 7.33 0.017 10 Significant different 2.5 2.9 2 4 6 8 10 12 14 16 CCL On Low aging time 2.5 have variation so much & Mean % CCT defect higher than High aging time 2.9. So if continue maintain and increase more Aging time to standardization. We can improve more defect performance 26 Asia Pacific % RFL BY CDT ,PHT & MATT LOSS DEFECT D M A I C CDT & PHT loss 75% on RFL 27 Asia Pacific Defect loss by kiln No D M A I C Boxplots of CC% by Kiln (means are indicated by solid circles) CC% 10 5 0 1 2 Kiln P>0.05 No difference between 2 kilns Two-Sample T-Test and CI: CC%, Kiln Two-sample T for CC% Kiln 1 2 N 53 81 Mean 2.56 2.86 StDev 2.10 2.37 SE Mean 0.29 0.26 Difference = mu (1) - mu (2) Estimate for difference: -0.305 95% CI for difference: (-1.078, 0.468) T-Test of difference = 0 (vs not =): T-Value = - 0.78 P-Value = 0.436 DF = 120 28 Asia Pacific FMEA SUMMARY D 29 M A I C Asia Pacific IV. IMPROVEMENT PHASE D M A I C A. Rifle Shot study B. Multi-vari Studies on CCT defect by location C. Improve mold condition/modification & Mold life D. Improve Caster performance/allocation & Aging time E. Multi-vari studies on RF loss CDT& PHT F. Data mining DOE on effects of Mold life & Humidity G. Defects monitoring and daily feedback H. Casting procedures follow-up 30 Asia Pacific CCT LOSS BY LOCATION NEED TO IMPROVE D M A I C I> O location need to improve II> A location need to improve Iii> H location need to improve IV> P2,P location Need to improve 31 Asia Pacific Reduction CCL on location O D Rifle Shot study M A I C Location O : 24% on CCL % -Poor corner need => Change radius dia -Poor moving short pad -=> Change pad by soft pillow -Thickness not enough Make thicker more 3mm Visual oil check Soft pillow Visual Oil check 32 Asia Pacific Reduction CCL on location O D M A I C Significant improve CCL at O location One-way ANOVA: Before, After Source DF SS MS F P Factor 1 293.2 293.2 4.80 0.043 Error 17 1038.1 61.1 Total 18 1331.3 S = 7.814 R-Sq = 22.02% R-Sq(adj) = 17.44% Individual 95% CIs For Mean Based on Pooled StDev Level Before After N Mean StDev 11 12.770 9.552 8 4.813 4.237 --+---------+---------+---------+------(---------*--------) (-----------*----------) --+---------+---------+---------+------0.0 5.0 10.0 15.0 Before W3 of Apr’10 After is from W4 of Apr- W4 of jul’10 Pooled StDev = 7.814 33 Asia Pacific Reduction CCL on location A D M A I C Location A : 20 % on CCT % Rifle Shot study -Humidity at location so hard to control , Especially at high mold life times Cheese cloth cover Air fan to dry Green finishing on transfer dryer date Air Fan to dry Standard loose mold Cheese cloth cover 34 Asia Pacific Reduction CCL on location A D M A I C Significant improve CCL at A location One-way ANOVA: A Before, A After Source DF SS MS F P Factor 1 300.7 300.7 5.84 0.042 Error 8 411.8 51.5 Total 9 712.5 S = 7.174 R-Sq = 42.21% R-Sq(adj) = 34.98% Individual 95% CIs For Mean Based on Pooled StDevLevel +---------+---------+---------+-A Before 5 27.309 7.893 A After 5 16.341 6.376 N Mean StDev ------- (----------*----------) (---------*----------) -------+---------+---------+---------+-14.0 21.0 28.0 35.0 Pooled StDev = 7.174 Before W1 of May After is from w2 of May – w2 of Aug’10 35 Asia Pacific Reduction CCL on location H (Foot core) D M A I C Rifle Shot study =>New comer : need to improve mold design : make spagless for foot core . => Back up foot core ( 2 foot core for 1 mold set 36 Asia Pacific Reduction CCL on location H (Foot core) D M A I C Two-Sample T-Test and CI: Before, After Two-sample T for Before vs After N Mean StDev SE Mean Before 7 13.83 7.03 2.7 After 8 8.75 2.01 0.71 Difference = mu (Before) - mu (After) Estimate for difference: 95% CI for difference: 5.07901 (-1.65354, 11.81156) T-Test of difference = 0 (vs not =): T-Value = 1.85 Value = 0.114 DF = 6 P- Have improve but not so much . NPD research & develop mold technical .How to spagless foot core to control mold humidity 37 Asia Pacific D M A I C •Most of location be reduced CCT loss , just remain P2,P location still not yet improve . •So let take action to improve P2,P location in next step 38 Asia Pacific Reduction CCL on location P2& P D M A I C Rifle Shot study Do experiment on 1 bench with 1 opt . Result after one month Location P2,P: is hollow . => Modified solid Result Significant improve Two-sample T for before_1 vs after_1 N Mean StDev SE Mean before_1 8 0.2275 0.0423 0.015 after_1 4 0.06750 0.00957 0.0048 Difference = mu (before_1) - mu (after_1) Estimate for difference: 95% CI for difference: 0.160000 (0.123757, 0.196243) T-Test of difference = 0 (vs not =): T-Value = 10.18 Value = 0.000 DF = 8 39 P- Asia Pacific Increase Aging time D M A I C I-MR Chart of Aging time1 Individual Value U C L=3.0504 3.0 _ X=2.9263 2.9 LC L=2.8021 2.8 2 4 6 8 O bser v ation 10 12 14 16 Moving Range 0.16 U C L=0.1525 0.12 0.08 __ M R=0.0467 0.04 0.00 LC L=0 2 Before 40 Aging time Ave: 2.92 day 4 6 8 O bser v ation 10 After is from w1 of Jun still Jul’10 12 14 16 After Asia Pacific Improve Caster Performance D M A I C ter, 1463 before, 1463 After, 1293 before, 1293 After, 1632 After, 1070 before, 107 Rifle Shot study 40 One-way ANOVA: 1907 Before, 1907 After, 1666 before, 1666 After, ... DF SS MS F P Factor 18 2092.3 116.2 3.94 0.000 Error 35 1031.4 29.5 Total 53 3123.7 S = 5.429 30 R-Sq = 66.98% Data Source 10 R-Sq(adj) = 50.0 Individual 95% CIs For Mean Based on 0 Pooled StDev Level N +---------+----- Mean StDev 1907 Before 3 17.937 8.362 1907 After 3 4.127 3.606 1666 before 3 6.754 1.537 1666 After 3 3.733 3.607 1139 before 3 7.810 6.769 1139 After 3 6.746 1.787 1463 before 3 11.333 4.163 1463 After 3 8.238 5.918 20 ----+---------+--------(----*----) re er re er re er r e fter or e fter fter ore fter ore fter or e fter or e fter fo ft f o ft fo ft fo f f a ef A ef A ef A Be 07 A be 66 A be 39 A be 63 A be 93 A 32 A be 70 b 9 B 7 B 4 7 9 66 6 39 1 63 4 93 2 0 0 59 65 87 18 84 18 6 0 7 1 6 1 6 1 1 1 4 1 2 1 1 0 1 1 1 1 1 9 1 1 1 1 1 1 1 1 1 1293 before ---*----) 3 25.317 14.444 1293 After 3 1.111 1.925 (----*----) 1632 After 2 3.095 0.337 (------*-----) 1070 before 3 3.333 5.774 (-----*----) 1070 after 3 3.000 2.646 (----*-----) 1659 before 3 11.746 4.399 1659 After 3 3.333 5.774 1187 Before 3 6.667 2.309 =>Run WCP effectively 1187 After 3 5.333 0.577 -=> Standard casting procedure 1184 Before 1 ---*--------) 25.000 * 7.953 2.228 (----*-----) (-----*----) (----*----) (-----*----) (-----*----) -=> PP program Effectively & Incentive (----*-----) (----*----) 1184 After 3 (- (-----*----) (-----*----) (-----*----) (----*-----) (----(-----*----) Significant improvement Caster Performance 41 Asia Pacific CCL After VS Before Take Action D M A I C Test Mold life Normal distribution Two-Sample T-Test and CI: Before, After Two-sample T for Before vs After N Mean StDev SE Mean Before 19 15.31 3.95 0.91 After 13 8.61 2.14 0.59 Significantly CCL improve before vs After is from May – Jul’10 Difference = mu (Before) - mu (After) Estimate for difference: 95% CI for difference: 6.70320 (4.48224, 8.92417) T-Test of difference = 0 (vs not =): T-Value = 6.18 28 42 P-Value = 0.000 DF = Asia Pacific Reduction CDT & PHT defect Loss on RF Loss D M A I C Rifle Shot study Boxplot of Normal Program, Special program 24 22 Normal Kiln program Data 20 18 16 14 Significa nt different between Kiln program Special Kiln program Set up special kiln program to run for Some heavy & complex model including : Active to reduce CDT ,PHT 43 12 10 Normal Program Special program Two-Sample T-Test and CI: Normal Program, Special program Two-sample T for Normal Program vs Special program N Mean StDev SE Mean Normal Program 7 20.73 1.67 0.63 Special program 7 12.24 1.99 0.75 Difference = mu (Normal Program) - mu (Special program) Estimate for difference: 95% CI for difference: 8.48571 (6.32515, 10.64628) T-Test of difference = 0 (vs not =): T-Value = 8.64 P-Value = 0.000 DF = 11 Asia Pacific Reduction CDT & PHT defect loss on RF Loss D M A I C CDT & PHT Improve from Nov’09 to Jul’10 Two-Sample T-Test and CI: PHT% Before, PHT % After Two-sample T for PHT% Before vs PHT % After Good trend N Mean StDev SE Mean PHT% Before 6 6.92 1.69 0.69 PHT % After 7 4.663 0.833 0.32 Difference = mu (PHT% Before) - mu (PHT % After) Estimate for difference: 95% CI for difference: 2.25212 (0.45997, 4.04427) T-Test of difference = 0 (vs not =): TValue = 2.97 P-Value = 0.021 DF = 7 44 Asia Pacific DOE OBJECTIVE D M A I C The DOE was started from 7-Jun To 12 Jul ’10 Key Variable Used : Mold Life , OPT , Humidity , Slip formula, Slip Aging time Research interaction of Mold life , Humidity impact to CC Loss defect on Active one piece Find out Best combination between mold Life VS humidity with the best result on Active one piece 45 Asia Pacific Data mining –CC% vs FACTORS (DOE) Data mining with coded factors 46 D M A I C Asia Pacific DOE-CCL VS MOLD LIFE & HUMIDITY D M A I C Tabulated statistics: Mold life, Humidity Rows: Mold life off 45 51 57 63 69 75 All on All 0.1633 0.0467 0.1050 0.05774 0.04041 0.07791 3 3 6 0.2033 0.1233 0.1633 0.00577 0.03786 0.05007 3 3 6 0.2233 0.1367 0.1800 0.01155 0.00577 0.04817 3 3 6 0.2833 0.1767 0.2300 0.01155 0.03215 0.06229 3 3 6 0.3033 0.2700 0.2867 0.02309 0.03464 0.03204 3 3 6 0.3267 0.3033 0.3150 0.03512 0.02309 0.02950 3 3 6 0.2506 0.1761 0.2133 0.06485 0.09394 0.08806 18 18 36 Cell Contents: 47 Columns: Humidity Result : Mean Result : Standard deviation Mold life frome 69-75 got high CCL .So that why we just focus mold life from 57-63 Asia Pacific DOE-CCL VS MOLD LIFE & HUMIDITY D M A I C ANOVA: Result versus Mold life, Humidity Factor Type Levels Values Mold life fixed 6 45, 51, 57, 63, 69, 75 Humidity fixed 2 off, on Analysis of Variance for Result Source DF SS MS F P Mold life 5 0.188033 0.037607 32.57 0.000 Humidity 1 0.049878 0.049878 43.19 0.000 Error 29 0.033489 0.001155 Total 35 0.271400 S = 0.0339822 48 R-Sq = 87.66% Mold life & Humidity impact so much to CC defect loss 70% & 18.3% R-Sq(adj) = 85.11% Asia Pacific DOE – CC% vs Mold life ,humidity ,OPT ( 3 input, 2 level ) D M A I C Data mining work sheet ( Red- code) Model Cast date cast pcs/day Mold life Humidity Humidity1 OPT Slip formulaAging time loss pcs Result StdOrder RunOrder Blocks CenterPt StdOrder_1RunOrder_1 2010 7-Jun 13 45 off -1 1632 B-36 68 2 0.23 1 1 1 1 1 1 2010 8-Jun 14 51 off -1 1632 B-36 68 3 0.21 2 2 1 1 2 2 2010 9-Jun 13 57 off -1 1632 B-36 68 3 0.23 3 3 1 1 3 3 2010 10-Jun 14 63 off -1 1632 B-36 68 4 0.29 4 4 1 1 4 4 2010 11-Jun 15 69 off -1 1632 B-36 68 4 0.33 5 5 1 1 5 5 2010 12-Jun 15 75 off -1 1632 B-36 68 4 0.33 6 6 1 1 6 6 2010 13-Jun 14 45 on 1 1632 B-36 68 0 0.07 7 7 1 1 7 7 2010 14-Jun 14 51 on 1 1632 B-36 68 2 0.14 8 8 1 1 8 8 2010 15-Jun 15 57 on 1 1632 B-36 68 2 0.13 9 9 1 1 9 9 2010 16-Jun 14 63 on 1 1632 B-36 68 3 0.14 10 10 1 1 10 10 2010 17-Jun 13 69 on 1 1632 B-36 68 3 0.23 11 11 1 1 11 11 2010 18-Jun 14 75 on 1 1632 B-36 68 4 0.29 12 12 1 1 12 12 2010 19-Jun 15 45 off -1 1632 B-36 68 1 0.13 13 13 1 1 13 13 2010 20-Jun 15 51 off -1 1632 B-36 68 2 0.2 14 14 1 1 14 14 2010 21-Jun 13 57 off -1 1632 B-36 68 2 0.23 15 15 1 1 15 15 2010 22-Jun 14 63 off -1 1632 B-36 68 4 0.29 16 16 1 1 16 16 2010 23-Jun 14 69 off -1 1632 B-36 68 4 0.29 17 17 1 1 17 17 2010 24-Jun 14 75 off -1 1632 B-36 68 5 0.36 18 18 1 1 18 18 2010 25-Jun 15 45 on 1 1632 B-36 68 1 0.07 19 19 1 1 19 19 2010 26-Jun 13 51 on 1 1632 B-36 68 1 0.08 20 20 1 1 20 20 2010 27-Jun 14 57 on 1 1632 B-36 68 2 0.14 21 21 1 1 21 21 2010 28-Jun 15 63 on 1 1632 B-36 68 3 0.19 22 22 1 1 22 22 2010 29-Jun 14 69 on 1 1632 B-36 68 4 0.29 23 23 1 1 23 23 2010 30-Jun 15 75 on 1 1632 B-36 68 5 0.33 24 24 1 1 24 24 2010 1-Jul 15 45 off -1 1632 B-36 68 2 0.13 25 25 1 1 25 25 2010 2-Jul 15 51 off -1 1632 B-36 68 2 0.2 26 26 1 1 26 26 2010 3-Jul 14 57 off -1 1632 B-36 68 2 0.21 27 27 1 1 27 27 2010 4-Jul 15 63 off -1 1632 B-36 68 4 0.27 28 28 1 1 28 28 2010 5-Jul 14 69 off -1 1632 B-36 68 4 0.29 29 29 1 1 29 29 2010 6-Jul 14 75 off -1 1632 B-36 68 4 0.29 30 30 1 1 30 30 2010 7-Jul 14 45 on 1 1632 B-36 68 0 0 31 31 1 1 31 31 2010 8-Jul 13 51 on 1 1632 B-36 68 2 0.15 32 32 1 1 32 32 2010 9-Jul 14 57 on 1 1632 B-36 68 2 0.14 33 33 1 1 33 33 2010 10-Jul 14 63 on 1 1632 B-36 68 3 0.2 34 34 1 1 34 34 2010 11-Jul 14 69 on 1 1632 B-36 68 4 0.29 35 35 1 1 35 35 2010 12-Jul 14 75 on 1 1632 B-36 68 4 0.29 36 36 1 1 36 36 49 Asia Pacific DOE – CC% vs Mold life ,humidity ,OPT ( 3 input, 2 level ) D M A I C Factorial Fit: Result versus Mold life, Humidity, OPT NOTE * This design has some botched runs. It will be analyzed using a regression approach. Estimated Effects and Coefficients for Result (coded units) Term Effect Coef SE Coef T P 0.20979 0.004923 42.61 0.000 0.04046 0.02023 0.001441 14.04 0.000 -0.07208 -0.03604 0.004923 -7.32 0.000 OPT 0.00708 0.00354 0.004923 0.72 0.478 Mold life*Humidity 0.00718 0.00359 0.001441 2.49 0.019 Mold life*OPT -0.00361 -0.00180 0.001441 -1.25 0.221 Humidity*OPT -0.02125 -0.01062 0.004923 -2.16 0.040 Mold life*Humidity*OPT -0.00461 -0.00230 0.001441 -1.60 0.121 Constant Mold life Humidity S = 0.0278501 R-Sq = 92.00% R-Sq(adj) = 90.00% Analysis of Variance for Result (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 3 0.235499 0.203099 0.0676998 87.28 0.000 2-Way Interactions 3 0.012202 0.012202 0.0040674 5.24 0.005 3-Way Interactions 1 0.001981 0.001981 0.0019811 2.55 0.121 28 0.021718 0.021718 0.0007756 Lack of Fit 16 0.009668 0.009668 0.0006042 0.60 0.830 Pure Error 12 0.012050 0.012050 0.0010042 35 0.271400 Residual Error Total 50 Main effect Mold life , Humidity Interaction : Mold life & Humidity Asia Pacific Reduce model DOE – red code . OPT exclude D M A I C Factorial Fit: Result versus Mold life, Humidity * NOTE * This design has some botched runs. using a It will be analyzed regression approach. Estimated Effects and Coefficients for Result (coded units) Term Effect Coef SE Coef T P 0.21333 0.005011 42.57 0.000 0.04200 0.02100 0.001467 14.31 0.000 -0.07444 -0.03722 0.005011 -7.43 0.000 0.00838 0.00419 0.001467 2.86 0.007 Constant Mold life Humidity Mold life*Humidity S = 0.0300661 R-Sq = 89.34% R-Sq(adj) = 88.34% Analysis of Variance for Result (coded units) Source P DF Seq SS Adj SS Adj MS F Main Effects 0.000 2 0.235098 0.235098 0.117549 130.04 2-Way Interactions 0.007 1 0.007375 0.007375 0.007375 8.16 32 0.028927 0.028927 0.000904 8 0.006394 0.006394 0.000799 24 0.022533 0.022533 0.000939 35 0.271400 Residual Error Lack of Fit 0.569 Pure Error Total 51 0.85 Mold life, Humidity & Interaction ML & Humidity impact to CC loss Asia Pacific DOE – Fractional factorial ( red code ) D M A I C Interaction Plot (data means) for Result 0.300 Mold life 57 63 0.275 Mean 0.250 0.225 0.200 0.175 0.150 off on Humidity Mold life , humidity effect so much to CC% Let do with 2 factors , 3 replicates with 3 center point ( mold life 57& 63) 52 Asia Pacific DOE –Full factorial s D Factorial Fit: Result versus mold life, humidity M A I C Estimated Effects and Coefficients for Result (coded units) Term Effect Coef SE Coef T P 0.19417 0.011286 17.20 0.000 0.01167 0.00583 0.011286 0.52 0.614 humidity -0.06444 -0.03222 0.009215 -3.50 0.004 mold life*humidity -0.00833 -0.00417 0.011286 -0.37 0.718 -0.02250 0.019547 -1.15 0.270 Constant mold life Ct Pt S = 0.0390950 R-Sq = 51.77% Ct Pt not significant & curvature not significant R-Sq(adj) = 36.93% Analysis of Variance for Result (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 2 0.0190972 0.0190972 0.0095486 6.25 0.013 2-Way Interactions 1 0.0002083 0.0002083 0.0002083 0.14 0.718 1 0.0020250 0.0020250 0.0020250 1.32 0.270 13 0.0198694 0.0198694 0.0015284 1 0.0017361 0.0017361 0.0017361 1.15 0.305 12 0.0181333 0.0181333 0.0015111 17 0.0412000 Curvature Residual Error Lack of Fit Pure Error Total 53 Let use RSM to analysis Asia Pacific DOE – RSM ( contour plot) D M A I C Response Surface Regression: Result versus mold life, Humidity1 The analysis was done using coded units. Estimated Regression Coefficients for Result Term Coef SE Coef T P Constant 0.171667 0.015960 10.756 0.000 mold life 0.005833 0.011286 0.517 0.614 Humidity1 -0.032222 0.009215 -3.497 0.004 mold life*mold life 0.022500 0.019547 1.151 0.270 mold life*Humidity1 -0.004167 0.011286 -0.369 0.718 S = 0.03909 R-Sq = 51.8% R-Sq(adj) = 36.9% Analysis of Variance for Result Source DF Seq SS Adj SS Adj MS F P Regression 4 0.021331 0.021331 0.005333 3.49 0.038 Linear 2 0.019097 0.019097 0.009549 6.25 0.013 Square 1 0.002025 0.002025 0.002025 1.32 0.270 Interaction 1 0.000208 0.000208 0.000208 0.14 0.718 13 0.019869 0.019869 0.001528 1 0.001736 0.001736 0.001736 1.15 0.305 12 0.018133 0.018133 0.001511 17 0.041200 Residual Error Lack-of-Fit Pure Error Total Not different with inferential applied 54 Asia Pacific Conclusion of DOE D Tabulated statistics: mold life, Humidity Rows: mold life M A I C Monthly cost comparison between mold life and loss reduction Columns: Humidity compare cost by Mold life used $4,000 60 63 All All 0.2300 0.1350 0.1825 0.00000 0.00707 0.05500 2 2 4 0.1850 0.1500 0.1675 0.02665 0.00894 0.02633 Cost of mold 2127.26 mold life used 55 2088.5 420.85 391.42 2900.81 Average cost of CCT due to mold life > 55 casting times . This will happen only when we use mold life > 55 times (not happen every month) 0.03536 0.07500 0.00 2 2 4 0.2150 0.1500 0.1825 0.04790 0.01700 0.04833 10 10 20 Count Best Setting : Here We see at Mold life 60 result still in target , so we select mold life 60 we will saving mold life cost 90 $ / 2 months . So Mold life 60 with Humidity on is best setting Microsoft Excel Worksheet Compare loss cost CCT loss on ML >55 ML used at 55 2000.00 0.00000 Standard Total cost 287.04 1000.00 : Labour cost for Labour cost for mold making operation 308.62 0.2275 result $287$391 1531.6 0.1650 Mean mold life used 75 mold life used 55 mold life used 75 0.2900 : $309$421 $0 12 result $2,127 $2,089 $1,532 $1,000 6 deviation 55 $2,000 6 Cell Contents: $2,901 $3,000 1459.53 500.00 This incremental cost happens every month with the same amount 1,460 1500.00 773.55 $ 57 on $ off 774 Average monthly saving = $ 686 Loss cost Annual cost comparison between mold life and loss reduction Change mold every 75 casting times Average CCT defects of 52% happen during mold life of 55 – 75 75 casting times 55 casting times Month 0 Month 2.3 Mold life line Assumptions 1.Avg volume = 1170 casting / month or 1076 firing / month 2. Timing of high CCT defect = 0.84 month / each 75-casting time mold life Month 3 Month 9 Month 6 CCT reduction 100% from cutting mold life to 55 CCT loss Incremental Monthly Mold life on Mold monthly cost cost from life >55 from mold CCT 75 52% 55 0% Month 12 100% # Month impact / year 5,237 773.55 Saving from reducing mold life to 55 casting times Annual Impact 3.34 17,514.33 12.00 9,282.58 8,231.75 = 3.34 months / year Asia Pacific V. CONTROL PHASE D M A I C C C A. Control Plan B. Defect tracing and feedback daily C. Slip SPC and control sheet D. Humidity & temperature record E. POKAYOKE & Visual factory F. Casting procedure follow-up G. Mold status record H. Caster Performance record 56 Asia Pacific CONTROL PLAN D 57 M A I C C Asia Pacific DEFECT TRACEBILITY DATABASE D 58 M A I C C Asia Pacific DEFECT TRACEBILITY SHEET D M A I C C Daily check and record all defective pcs for quick tracing, actions and feedback 59 Asia Pacific DEFECT TRACING AND FEEDBACK D M A I C C Trouble shooting Analysis 60 Asia Pacific MOLD STATUS DAILY MONITORING D M A I C C Broken mold 61 Asia Pacific DAILY SLIP CONTROL SHEET D 62 M A I C C Asia Pacific SLIP ONLINE CONTROL D 63 M A I C C Asia Pacific HUMIDITY & TEMPERATURE Record D Boxplot of DA 9h Aug, DA 9h Jul, DA 9h jun, DA 9h May Boxplot of DA 14H Aug, DA 14h jul, DA 14H Jun, DA 14h May 80 M A I C C Boxplot of DA 5h Aug, DA 5h JUl, DA 5h JUn, DA 5h May 90 80.0 80 77.5 70 70 50 75.0 60 Data Data Data 60 50 72.5 70.0 40 67.5 40 30 30 65.0 20 DA 9h Aug 64 DA 9h Jul DA 9h jun DA 9h May DA 14H Aug DA 14h jul DA 14H Jun DA 14h May DA 5h Aug DA 5h JUl DA 5h JUn DA 5h May Asia Pacific POKAYOKE APPLY & VISUAL FACTORY D M A I C C Preventiv e risk CC defect while trap glaze at Spray Dept Fix location when run Air Fan 65 Asia Pacific DAILY CASTING PROCEDURE FOLLOW-UP D M A I C C Base on Slip property , mold life to set up casting record according daily 66 Asia Pacific Caster Performance Tracking D M A I C C Daily ,Weekly & monthly Individual caster performance be tracked & feedbacked 67 Asia Pacific FMEA SUMMARY D 68 M A I C C Asia Pacific VI. RESULTS SUMMARY A. FFL ,RFL & TTL% PERFORMANCE B. PROJECT SAVING 69 Asia Pacific PROJECT RESULTS Performance FFL % compare before Action One-way ANOVA: FFL % before, FFL% After Normal distribution Source DF SS MS F P Factor 1 463.2 463.2 44.66 0.000 Error 10 103.7 10.4 Total 11 567.0 S = 3.221 R-Sq = 81.70% R-Sq(adj) = 79.87% Individual 95% CIs For Mean Based on Pooled StDev Level N Mean --------+---------+ StDev FFL % before (----*-----) 7 26.116 3.910 FFL% After 5 13.513 1.733 ---------+---------+- (-----*-----) ---------+---------+- --------+---------+ 15.0 25.0 30.0 Pooled StDev = 3.221 20.0 Significantly Improved from Apr-jul’10 Target 18% Significantly archived 70 Asia Pacific PROJECT RESULTS Result RF Loss after took action One-way ANOVA: RFloss Bfeore_1, RFlossSignificant After_1 improved from Nov’09 to jul’10 Normal distribution Source DF SS MS F P Factor 1 2046 2046 11.88 0.004 Error 13 2239 172 Total 14 4286 S = 13.12 R-Sq = 47.74% R-Sq(adj) = 43.73% Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev ------+---------+---------+ RFloss Bfeore_1 10 (-------*------) 39.09 15.71 RFloss After_1 --------) 14.31 2.21 5 ---------+--- (----------*---------+--- ------+---------+---------+ 12 24 36 48 Pooled StDevarchived = 13.12 Target 21% Significantly Boxplot of RFloss Bfeore_1, RFloss After_1 71 Asia Pacific PROJECT RESULTS Significant improved from W1 of Apr to w4 of Jul’10 Result TT Loss after took action One-way ANOVA: Before, After Normal distribution Source DF SS MS F P Factor 1 2487.4 2487.4 34.67 0.000 Error 22 1578.4 71.7 Total 23 4065.7 S = 8.470 59.41% R-Sq = 61.18% R-Sq(adj) = Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev ------+---------+-----Before 14 47.838 (-----*-----) After 10 27.188 ---+---------+--- 10.639 3.446 (------*------) ---+---------+--- ------+---------+-----24.0 32.0 Target 29.1%40.0 Significantly 48.0 archived Pooled StDev = 8.470 72 Asia Pacific SAVING RESULT AP2165 - MONTHLY PROJECT SAVING 73 Asia Pacific