The Pennsylvania State University. The Graduate School. Department of Industrial and Manufacturing Engineering

Transcription

1 The Pennsylvania State University The Graduate School Department of Industrial and Manufacturing Engineering A PATTERN ALLOWANCE ADVISOR TOOL FOR STEEL CASTINGS A Thesis in Industrial Engineering by Mandar Deo 2005 Mandar Deo Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy August 2005

2 The thesis of Mandar Deo was reviewed and approved* by the following: Robert C. Voigt Professor of Industrial Engineering Thesis Advisor Chair of Committee Paul Cohen Distinguished Professor of Industrial Engineering Timothy W. Simpson Associate Professor of Mechanical Engineering and Industrial Engineering Irene Petrick Assistant Professor of Information Science & Technology Richard J. Koubek Professor of Industrial Engineering Head of the Department of Industrial and Manufacturing Engineering *Signatures are on file in the Graduate School

3 iii ABSTRACT Steel casting features can be expected to have different size than the pattern features from which they were made. This difference between the casting dimension and the pattern dimension is termed as pattern allowance. Currently, no pattern allowance estimation techniques exist. Lack of pattern allowance estimation techniques results in longer lead times, expensive reengineering cycles, and lost market share. Due to inability of the foundries to predict the pattern dimension correctly and casting dimensional variability, castings are designed with extra stock (machining allowance), which is subsequently removed by machining. However, due to increasing machining costs, more and more foundry customers are demanding smaller machining allowance. Casting and pattern dimensional data is collected from production steel foundries along with other process and geometry variables. Statistical analysis is conducted to ensure data integrity through gage repeatability and reproducibility studies and outlier analyses. Dimensional data is then classified according to alloy type, restraint type, location of parting line, and molding method. A new empirical model has been developed to predict pattern allowances for steel castings depending on the abovementioned variables. This model predicts average, minimum, and maximum pattern dimensions based on median, 10 th percentile and 90 th percentile values of an artificial variable called Shrinkage Error. This model is then validated by collecting more pattern allowance data and comparing the predicted pattern dimensions. A software tool called Pattern Allowance Advisor has been developed for the implementation of the pattern dimension prediction model. In addition to helping the

4 iv pattern designers in predicting the pattern dimensions, this software is also designed to serve as a comprehensive casting dimensional data management tool. Furthermore, the software also offers capability of customizing and calibrating the pattern dimension prediction model for a specific foundry through learning from the inspection data. Pattern Allowance Advisor software is acquired by several steel foundries. Application of the new empirical model through Pattern Allowance Advisor has resulted in an increase in first article casting dimensional conformance from 52% to 81% and a reduction in dimensional reengineering cycles from 3.5 to 1.1 during testing in production steel foundries. Overall reduction in total centering errors is also achieved. These improvements have resulted in reduction in lead time and dimensional reengineering costs.

5 v TABLE OF CONTENTS LIST OF FIGURES...x LIST OF TABLES...xviii ACKNOWLEDGEMENTS...xxii Chapter 1 Introduction Motivation Overview of Casting Dimensional Issues Dimensional Variability due to Casting Process Variability Measurement System Errors Pattern Allowance Pattern Allowance Prediction Tool Thesis Statement Thesis Outline...14 Chapter 2 Literature Review Dimensional Tolerance Standards for Steel Castings Dimensional Variability Inspection Errors and Sampling Uncertainty Measurement System Errors Dimensional Uncertainty due to Insufficient Sampling Analytical Modeling of Pattern Allowance in Castings Factors Affecting Pattern Allowances in Steel Casting Effect of Alloy Type on Pattern Allowance Effect of Sand Type on Casting Dimensions Effect of Feature Length on Pattern Allowance Effect of Mold Restraint on Pattern Allowance Effect of Casting Feature Shapes on Pattern Allowance Pattern Allowance in Steel Castings Industry Pattern Allowance Practices Empirical Pattern Allowance Models for Sand Steel Casting Processes...45 Chapter 3 Need for Improved Pattern Allowance Prediction Models Casting Feature Length Variation in Pattern Allowance Non-Normality of Pattern Allowance Data Neighboring Casting Geometry Foundry to Foundry Variation...53

6 3.6 Goals for New Pattern Allowance Model...53 vi Chapter 4 Data Collection and Analysis Point-to-Point 2D Data Collection Data Integrity for Point-to-Point 2D Data Casting Process Variability Measurement System Errors D Point-to-point Data Classification Scheme D Data Summary Features Not Crossing the Parting Line Features Across the Parting Line Linear Scan Measurements Linear Scanning Data Summary Green Sand Molded Low Alloy Steel Casting Linear Scanned Dimensions Nobake Sand Molded Low Alloy Steel Casting Overview of the collected data...84 Chapter 5 Proposed Pattern Allowance Model Constitutive Equations for the Pattern Allowance Model Standard Shrinkage Allowance γ Implementation of New Pattern Allowance Model Low Alloy Casting Features Not Crossing the Parting Line High Alloy Casting Features Not Crossing the Parting Line Low and High Alloy Casting Features Crossing the Parting Line Verification of the New Pattern Allowance Model Chapter 6 Pattern Allowance Advisor Software Database User Types Modules Administrator Module Calibration Module New Part Module Modification Module Inspection Module Reports Module Help Module History Module Dimension Entry Window Casting and Pattern Dimension Entry Area Feature Type Selection Picture Area Pattern Dimension Guide Pictures...143

7 6.5 Calibration Module Optimization Algorithm Golden Interval Method for Minimization of Error Summation Validation of the Optimization Algorithm Software Testing Model Validation Using the Pattern Allowance Advisor Pattern Dimension Prediction Using Pattern Allowance Advisor Inspection Reports and Conformance Comparison Validation of the Pattern Allowance Advisor Calibration Capabilities Chapter 7 Conclusions and Future Work Conclusions Future Work Improvements in the Pattern Allowance Model Improvements in Pattern Allowance Advisor Software References Appendix A Casting Variables Survey Form Appendix B Sample Part File Appendix C Sample Gage R&R Calculation Appendix D Point-to-Point 2D database dimensions D.1 Classified Database of features not crossing the mold parting line D.1.1 Green Sand Molded Low Alloy Steel Castings - Fully Restrained Features D.1.2 Green sand molded low alloy steel castings partially restrained features D.1.3 Green sand molded low alloy steel castings Unrestrained Features D.1.4 Nobake sand molded low alloy steel castings fully restrained features D.1.5 Nobake sand molded low alloy steel castings partially restrained features D.1.6 Nobake sand molded low alloy steel castings unrestrained features D.1.7 Shell molded low alloy steel castings fully restrained features D.1.8 Shell molded low alloy steel castings partially restrained features..215 D.1.9 Shell molded low alloy steel castings unrestrained features D.1.10 Shell molded high alloy steel castings fully restrained features vii

8 D.1.1 Shell molded high alloy steel castings partially restrained features D.1.1 Shell molded high alloy steel castings unrestrained features D.2 Classified Database of features across the mold parting line D.2.1 Green sand molded low alloy steel castings Unrestrained Features D.2.2 Nobake sand molded low alloy steel castings fully restrained features D.2.3 Nobake sand molded low alloy steel castings partially restrained features D.2.4 Nobake sand molded low alloy steel castings unrestrained features D.2.5 Shell molded low alloy steel castings partially restrained features..222 D.2.6 Shell molded low alloy steel castings unrestrained features D.2.7 Shell molded high alloy steel castings unrestrained features Appendix E Database Tables in Pattern Allowance Advisor software E.1 Alloytype E.2 Calibration E.3 Custom E.4 Default E.5 dimarchive E.6 dimensions E.7 inspectied and patinspected E.8 partinfo E.9 patallowhist E.10 processtype E.11 rptinsp E.12 userinfo Appendix F Program steps for the Pattern Allowance Advisor F.1 Administrator module F.1.1 Add User F.1.2 Modify / Delete User F.1.3 Edit / Delete part F.2 New Part Module F.2.1 New part creation stage F.2.2 Dimension specification stage F.3 Modification Module F.4 Inspection module F.4.1 Casting dimension inspection F.4.2 Pattern inspection entry F.5 History module viii

9 F.5.1 Pattern history F.5.2 Pattern allowance history F.6 Reports module F.7 Calibration module Appendix G Pattern Allowance Advisor User Manual G.1 Introduction G.1.1 Computer System Requirements: G.1.2 Additional Requirements to Support the Use of the PAA G.2 Installation instructions G.2.1 Installing from a removable media G.2.2 Installing from an ed or downloaded file G.3 New Part Feature G.4 Modification Feature G.5 Inspection Feature G.6 History Feature G.7 Report Feature G.8 Help Feature G.9 Administrator Feature G.10 Calibration module G.11 Tutorials G.11.1 Tutorial 1: Installation of PA advisor G.11.2 Tutorial 2: Administrative options G.11.3 Tutorial 3: Create A New Part File G.11.4 Tutorial 4: Modifying an existing part G.11.5 Tutorial 5: Entering Measured Casting and Pattern Dimensions G.11.6 Tutorial 6: Reviewing pattern history G.11.7 Tutorial 7: Reviewing pattern allowance history G.11.8 Tutorial 8: Generating Reports G.12 Contact Information ix

10 x LIST OF FIGURES Figure 1-1: US steel Casting Shipments [1]....2 Figure 1-2: Dimensional non-conformance caused by casting process variability...4 Figure 1-3: Factors affecting dimensional variability in steel castings [3]...4 Figure 1-4: Three phases of solidification of molten metal [7]....7 Figure 1-5: Exaggerated size and shape changes in mold during and after pouring [7]...8 Figure 1-6: Pattern dimensional re-engineering cycle for centering casting dimension within customer specification Figure 1-7: Pattern dimension estimation cycle with feedback loop and digital dimensional database used for estimation of pattern dimension Figure 2-1: SFSA T grades [10]...16 Figure 2-2: ISO CT tolerance grades [13]...18 Figure 2-3: Additional component of dimensional variability estimates due to sampling uncertainty errors Figure 2-4: Casting process and geometry variables affecting pattern allowances...27 Figure 2-5: Expected pattern allowance values for various alloys in inches per foot and as a multiplying factor [38]...29 Figure 2-6: Contraction of 0.35C steel cast into greensand molds [43] Figure 2-7: Temperature expansion curves for various mold materials [44]...32 Figure 2-8: Various zones in a sand mold after metal pouring [45]...33 Figure 2-9: Inward expansion of mold wall in convex regions and outward expansion in concave regions [50]...35 Figure 2-10: Effect of feature length on pattern allowance for steel castings [10]...37 Figure 2-11: Effect of feature length on the pattern allowances of investment castings due to scale removal [55]...38 Figure 2-12: Dimension types due to mold restraint [10]...40

11 Figure 2-13: Effect of mold-casting interaction on pattern allowance [57]...41 xi Figure 2-14: Penn State casting feature classification [58]...42 Figure 3-1: Overall distribution of pattern allowances as a function of casting feature length - green sand, nobake, and shell low alloy steel castings [39]...49 Figure 3-2: Distribution of observed pattern allowance values as a function of casting feature length along with average and 80% confidence interval values - green sand low alloy steel casting unrestrained features [39] Figure 3-3: Observed pattern allowance Normal Probability plot for unrestrained features in green sand low alloy steel castings Figure 3-4: Variation in pattern allowance along the length of a casting feature...52 Figure 3-5: Summary of factors influencing casting dimensional variability...54 Figure 4-1: Categories used to classify dimensional data using database SQL queries...58 Figure 4-3: Sample casting for molding method classification of fully and partially restrained features...60 Figure 4-4: Relationship between pattern dimension and casting dimension for all features in the 2D database...63 Figure 4-5: Frequency distribution of observed pattern allowance for all features in the 2D database...64 Figure 4-6: Distribution of pattern allowances with respect to casting feature size for all casting features in the 2D database Figure 4-7: Frequency distribution of observed pattern allowance for low alloy features not crossing the parting line in the 2D database Figure 4-8: Distribution of pattern allowances with respect to casting feature size for low alloy casting features in the 2D database Figure 4-9: Frequency distribution of observed pattern allowance for low alloy features not crossing the parting line in the 2D database Figure 4-10: Distribution of pattern allowances with respect to casting feature size for high alloy casting features in the 2D database...70 Figure 4-11: Variation in pattern allowance with respect to casting feature size in case of low and high alloy steel castings...72

12 Figure 4-12: Effect of neighborhood geometry on pattern allowance in individual feature Figure 4-13: FARO arm Silver Series articulated CMM used for generating linear scans of castings and patterns [67] Figure 4-14: A predetermined, marked scanning path on the casting to be followed by the CMM probe to generate a dimensional point cloud Figure 4-15: Wire frame model built by AnthroCAM for a sample pattern...76 Figure 4-16: Linear scan along with local coordinate system generated in AutoCAD from wire frame model generated by AnthroCAM software...76 Figure 4-17: Relationship between pattern dimension and casting dimension for all features inspected using linear scan measurements...79 Figure 4-18: Distribution of pattern allowances with respect to casting feature size for all casting features measured using linear scanning measurement...80 Figure 4-19: Frequency distribution for observed pattern allowances in features not crossing parting line as well as crossing parting line for green sand low alloy castings inspected using linear scan measurements...81 Figure 4-20: Distribution of pattern allowances with respect to casting feature size in features not crossing parting line as well as crossing parting line for green sand low alloy castings inspected using linear scan measurements Figure 4-21: Frequency distribution for observed pattern allowances in features not crossing parting line as well as crossing parting line for nobake low alloy castings inspected using linear scan measurements...83 Figure 4-22: Distribution of pattern allowances with respect to casting feature size in features not crossing parting line as well as crossing parting line for nobake low alloy castings inspected using linear scan measurements Figure 5-1: Overall distribution of shrinkage error for all casting features in the 2D database. Shrinkage error is calculated using Equation 5.1, the standard shrinkage allowance γ is assumed to be (1/4 in per foot) Figure 5-2: Shrinkage Error Normal Probability plot for unrestrained features in green sand low alloy steel castings from the 2D database...89 Figure 5-3: Observed pattern allowance Normal Probability plot for unrestrained features in green sand low alloy steel castings from the 2D database...90 xii

13 Figure 5-4: Distribution of shrinkage error for green sand low alloy steel castings - unrestrained features along casting feature length. Both the 80% confidence interval and the 10 th and 90 th percentiles for shrinkage error are also indicated...91 Figure 5-5: Change in summation of error with respect to shrinkage allowance γ for low alloy nobake casting fully restrained, partially restrained, and unrestrained features...94 Figure 5-6: Change in summation of square of the error with respect to shrinkage allowance γ for low alloy nobake casting fully restrained, partially restrained, and unrestrained features Figure 5-7: Observed pattern allowance for green sand molded low alloy steel castings - unrestrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves...96 Figure 5-8: Distribution of shrinkage error along the feature size for low alloy fully restrained features not crossing the parting line in green sand molds Figure 5-9: Observed pattern allowance for green sand molded low alloy steel castings - fully restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-10: Distribution of shrinkage error along the feature size for low alloy partially restrained features not crossing the parting line in green sand molds Figure 5-11: Observed pattern allowance for green sand molded low alloy steel castings - partially restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-12: Distribution of shrinkage error along the feature size for low alloy unrestrained features not crossing the parting line in green sand molds Figure 5-13: Distribution of shrinkage error along the feature size for low alloy fully restrained features not crossing the parting line in nobake sand molds Figure 5-14: Observed pattern allowance for nobake sand molded low alloy steel castings - fully restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves xiii

14 Figure 5-15: Distribution of shrinkage error along the feature size for low alloy partially restrained features not crossing the parting line in nobake sand molds Figure 5-16: Observed pattern allowance for nobake sand molded low alloy steel castings - partially restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-17: Distribution of shrinkage error along the feature size for low alloy unrestrained features not crossing the parting line in nobake sand molds Figure 5-18: Observed pattern allowance for nobake sand molded low alloy steel castings - unrestrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-19: Distribution of shrinkage error along the feature size for low alloy fully restrained features not crossing the parting line in shell molds Figure 5-20: Observed pattern allowance for shell molded low alloy steel castings - fully restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-21: Distribution of shrinkage error along the feature size for low alloy partially restrained features not crossing the parting line in shell molds Figure 5-22: Observed pattern allowance for shell molded low alloy steel castings - partially restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-23: Distribution of shrinkage error along the feature size for low alloy unrestrained features not crossing the parting line in shell molds Figure 5-24: Observed pattern allowance for shell molded low alloy steel castings - unrestrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-25: Distribution of shrinkage error along the feature size for high alloy fully restrained features not crossing the parting line in green sand molds Figure 5-26: Observed pattern allowance for high alloy steel castings - fully restrained features not crossing the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves xiv

15 Figure 5-27: Distribution of shrinkage error along the feature size for high alloy partially restrained features not crossing the parting line in green sand molds Figure 5-28: Observed pattern allowance for high alloy steel castings - partially restrained features not crossing the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-29: Distribution of shrinkage error along the feature size for high alloy unrestrained features not crossing the parting line in green sand molds Figure 5-30: Observed pattern allowance for high alloy steel castings - unrestrained features not crossing the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-31: Distribution of shrinkage error along the feature size for low alloy unrestrained features crossing the parting line in green sand molds Figure 5-32: Observed pattern allowance for low alloy steel castings - unrestrained features across the parting line in green sand molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-33: Distribution of shrinkage error along the feature size for low alloy fully restrained features crossing the parting line in nobake molds Figure 5-34: Observed pattern allowance for low alloy steel castings - fully restrained features across the parting line in nobake molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-35: Distribution of shrinkage error along the feature size for low alloy partially restrained features crossing the parting line in nobake molds Figure 5-36: Observed pattern allowance for low alloy steel castings - partially restrained features across the parting line in nobake molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-37: Distribution of shrinkage error along the feature size for low alloy unrestrained features crossing the parting line in nobake molds Figure 5-38: Observed pattern allowance for low alloy steel castings - unrestrained features across the parting line in nobake molds along with xv

16 predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-39: Distribution of shrinkage error along the feature size for low alloy partially restrained features crossing the parting line in shell molds Figure 5-40: Observed pattern allowance for low alloy steel castings - partially restrained features across the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-41: Distribution of shrinkage error along the feature size for low alloy unrestrained features crossing the parting line in shell molds Figure 5-42: Observed pattern allowance for low alloy steel castings - unrestrained features across the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-43: Distribution of shrinkage error along the feature size for high alloy unrestrained features crossing the parting line in shell molds Figure 5-44: Observed pattern allowance for high alloy steel castings - unrestrained features across the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-45: Observed pattern allowance data from linear scanning measurement for low alloy green sand partially restrained features along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 5-46: Observed pattern allowance data from linear scanning measurement for low alloy green sand unrestrained features along with predicted 10th percentile, 90th percentile, and average pattern allowance curves Figure 6-4: New part entry module Figure 6-5: Modification module Figure 6-6: Inspection module Figure 6-7: Report generation scheme Figure 6-8: Pattern tooling history and pattern allowance history module Figure 6-9: Dimension entry window screenshot Figure 6-10: Casting and pattern dimension entry area xvi

17 Figure 6-11: Feature type selection picture Figure 6-12: Pattern dimension guide pictures Figure 6-13: Integrated bracketing and Golden interval algorithm for the minimization of error summation [72] Figure 6-14: Shovel Adaptor molding layout Figure 6-15: Shovel Adaptor casting Figure 6-16: The dimensions entry window when there are no dimensional data entries of the selected feature type in the custom database Figure 6-17: Casting inspection data for LowGreenPR castings is entered using casting inspection module in the Pattern Allowance Advisor Figure 6-18: Calibration module showing the calibration results for low alloy green sand molded partially restrained features Figure 6-19: Dimensions entry window displaying identical custom and default model pattern dimension recommendations. This simulation validates the calibration module used in the Pattern Allowance Advisor Figure C-1: Sample filled out gage R&R data sheet. Template from AIAG manual [31] Figure C-2: Sample filled out gage R&R report. Template from AIAG manual [31] xvii

18 xviii LIST OF TABLES Table 2-1: Equations for SFSA dimensional tolerance specifications T grades [10]...15 Table 2-2: Acceptability criteria for Gage R & R Tests [31]...21 Table 2-3: Statistically determined minimum number of sample castings required for pattern approval process with 95% confidence [4]...22 Table 2-4: Dimensional Variability Multiplying Factors [34]...23 Table 2-5: Shrinkage of various steel alloys for C ( F) temperature range [42]...28 Table 2-6: Permanent linear change in various sand/binder systems [44]...31 Table 2-7: Influence of feature size on the contraction of green sand steel castings [53]...36 Table 2-8: Explanation for various dimension types. Refer to Figure 2-14 [58]...42 Table 2-9: Classification of casting features according to levels of restraint. Refer to Figure 2-14 [58]...42 Table 2-10: Pattern allowance for various casting shapes for steel sand castings in green sand molds [53]...43 Table 2-11: Influence of dimension type on the contraction of green sand steel castings [53]...46 Table 2-12: Average and confidence interval based model for high and low alloy steel castings in green sand, no bake and shell molds [39]...47 Table 4-1: Sample query result for green sand low alloy steel castings fully restrained features not crossing the parting line...61 Table 4-2: Summary of the collected 2D dimensional data for features not crossing the parting line...66 Table 4-3: Summary of the collected 2D dimensional data for features across the parting line Table 4-4: Summary data of castings used for linear scanning measurement...78 Table 4-5: Summary of feature types inspected using linear scan measurements...78

19 Table 4-6: Summary of the collected data...85 Table 5-1: Change in the value of standard shrinkage allowance γ due to minimization of summation of error for green sand molded low alloy steel castings - unrestrained features...96 Table 5-2: Standard shrinkage allowance and changes in sum of error due to minimization for features not across the parting line in low and high alloy steel castings Table 5-3: Standard shrinkage allowance and changes in sum of error due to minimization for features across the parting line in low and high alloy steel castings from the 2D database Table 5-4: Median, Error min (10 th percentile), and Error max (90 th percentile)shrinkage error along with optimized standard shrinkage allowance and observed average pattern allowance values for low alloy casting features not crossing the parting line...99 Table 5-5: Median, minimum, and maximum error along with optimized standard shrinkage allowance and observed average pattern allowance values for high alloy casting features not crossing the parting line Table 5-6: Median, minimum, and maximum error along with optimized standard shrinkage allowance and observed average pattern allowance values for low and high alloy casting features crossing the parting line from the 2D database Table 6-1: List of MS Access database tables included in the Pattern Allowance Advisor Table 6-2: Pattern Allowance Advisor modules along with user types having access privileges Figure 6-2: New user entry by administrator Figure 6-3: User / part deletion / modification by administrator Table 6-3: Comparison of the values of standard shrink and summation of error for all types of features in low alloy castings made in green sand and nobake sand obtained using the Excel solver and the in-house code Table 6-4: Pattern dimensions for shovel adaptor 12N recommended by the Pattern Allowance Advisor compared to conventional pattern allowance rules xix

20 Table 6-5: First article conformance report for Shovel adaptor 12N casting produced using pattern dimensions recommended by Pattern Allowance Advisor Table 6-6: First article conformance report for Shovel adaptor 10N casting produced using pattern dimensions predicted with conventional shrink rules Table 6-7: Comparison of centering errors for 10N casting built using conventional pattern allowances and 12N casting built using pattern allowances recommended by Pattern Allowance Advisor Table C-1: Gage Repeatability and Reproducibility Data Sheet - 2 operators, 2 trials, and 10 parts Table D-1: Complete Database for high and low alloy features Table D-2: Database for green sand low alloy fully restrained features Table D-3: Database for green sand low alloy partially restrained features Table D-5: Database for nobake sand low alloy partially restrained features Table D-6: Database for nobake sand low alloy unrestrained features Table D-7: Database for shell molded low alloy fully restrained features Table D-8: Database for shell molded low alloy unrestrained features Table D-9: Database for shell molded low alloy unrestrained features Table D-10: Database for shell molded high alloy fully restrained features Table D-11: Database for shell molded high alloy partially restrained features Table D-12: Database for shell molded high alloy unrestrained features Table D-13: Database for green sand molded low alloy unrestrained features across the parting line Table D-14: Database for nobake sand molded low alloy fully restrained features across the parting line Table D-15: Database for nobake sand molded low alloy partially restrained features across the parting line Table D-16: Database for nobake sand molded low alloy unrestrained features across the parting line xx

21 Table D-17: Database for shell molded low alloy partially restrained features across the parting line Table D-18: Database for shell molded low alloy unrestrained features across the parting line Table D-19: Database for shell molded high alloy unrestrained features across the parting line xxi

22 xxii ACKNOWLEDGEMENTS I am very grateful to my thesis adviser Dr. Bob Voigt for his patient guidance through my research work. I am indebted to him for helping me complete my program on time through very valuable inputs. Thanks are also due to US Defense Logistics Agency, Steel Founder s Society of America, and American Metalcasting Consortium for their financial support for my research. I wish to express my thanks to Dr. Cohen, Dr. Simpson, and Dr. Petrick for serving on my committee. I am deeply grateful to my MS thesis advisor Dr. Pan Michaleris for working hard on my technical writing skills. I could not have completed this work without constant encouragement and support from my wife Rajashree, my parents Vidya and Vighnahari, and my sisters Medha, Deepa, and Pradnya. I wish to thank numerous undergraduate and graduate students for collecting data for this project and especially Karthik for setting up a good database system. I am deeply grateful to my friends Aamod Sathe, Prasanna Nirgudkar, Angela Wollenburg, and Rajdeep Pradhan for making my days at Penn State very enjoyable. I am very thankful to Drs. Shama and Anil Kulkrani, Drs. Lalita and Ganapati Patil, and Dr. Garga for providing support and welcoming me in their hearts and homes.

23 Chapter 1 Introduction 1.1 Motivation Metal casting processes can produce intricate and complex parts close to the desired size and shape, giving them an inherent advantage over many other manufacturing processes. Furthermore, castings offer several advantages such as more metallurgical versatility, structural integrity through one-piece finished part as well as rapid and economical alterations in shape. More than 90% of all manufactured goods and capital equipment use castings as engineered components or rely on castings for their manufacture [1]. Steel castings in particular are widely used in railroad, valve and pipe fitting, construction, mining, farming equipment, and the defense industry. US steel castings production for the year 2004 is estimated to be 1.5 million tons and is expected to maintain these levels into the next decade as shown in Figure 1-1 [1]. Steel sand castings offer further advantages due to flexible and comparatively inexpensive pattern tooling and the recyclable mold material. This is particularly advantageous for short-run steel castings.

24 2 Figure 1-1: US steel Casting Shipments [1]. The metal casting industry is competing with other manufacturing industry segments for market share, including welded fabrications, machining, plastics, and forging. Within the metal casting industry, different segments also compete with each other for the same components. Due to dimensional variability, castings are designed with extra stock (machining allowance), which is subsequently removed by machining. However, due to high machining costs, more and more foundry customers are demanding smaller machining allowance. Casting customers continue to require castings to be produced to increasingly tighter dimensional tolerance specifications to facilitate subsequent automated machining of the castings after they are purchased. Furthermore, part designers are sometimes hesitant to specify metal castings because of the excessive lead-time required to design and produce a dimensionally acceptable pattern from which targeted casting dimension can be achieved. A survey at a medium production steel foundry has indicated that it can take up to 22 months from the receipt of a new pattern to produce castings that fully

25 3 satisfy the customer s dimensional and geometry requirements. During this period, nearly half of the problems encountered were related to casting dimensional conformance [2]. Therefore, in order to better compete with other processing methods, the steel foundry industry must reduce lead-times for new castings and ensure dimensional integrity of the castings without expensive and time consuming re-engineering steps. 1.2 Overview of Casting Dimensional Issues Dimensional integrity in castings is affected by: 1. Casting dimensional variability 2. Random variation in feature measurements 3. Incorrect pattern dimensions due to the application of improper pattern allowance values. 4. Incorrect pattern dimensions due to errors in pattern production Dimensional Variability due to Casting Process Variability Casting dimensional variability can be caused by random (non-systemic) variation in casting process parameters, such as: core making, mold making, assembly of cores into the mold, mold closing, pouring, heat treating, cleaning, and grinding. Random variations increase the variability of casting dimensions. Even if casting dimensions are correctly centered at the desired nominal size of casting, non-conforming castings can be produced

26 due to high dimensional variability as shown in Figure 1-2. Major sources of nonsystemic random variations in casting process are shown in Figure Lower Tolerance Limit Nominal Dimension Upper Tolerance Limit Desired Variability Achieved Variability Nonconformance Nonconformance Figure 1-2: Dimensional non-conformance caused by casting process variability Molding Method Molding Equipment Metal Condition Core Making Core Type(s) Molding Materials Mold Making Molding Sand Feature Geometry Geometry Core Patterns Overall Dimensional Variability Figure 1-3: Factors affecting dimensional variability in steel castings [3]

27 Measurement System Errors Casting feature measurement errors also cause uncertainty in the estimation of mean casting size and pattern size measurements. These measurement errors are caused by small sample sizes or inadequate gage reproducibility and repeatability (gage R & R). Because the casting feature tolerance range is a function of casting feature dimensional variability, the mean value of a casting feature cannot be estimated by sampling only a few castings. It has been shown that if dimensional variability is greater than 60% of the feature dimensional tolerance, 44 castings are needed to accurately estimate the process mean for subsequent pattern feature dimensional adjustment [4]. Large sample sizes are necessary because smaller sample sizes can lead to inaccurate estimates of casting feature mean dimensions. Measurement system reproducibility and repeatability errors are caused by the equipment and the operator interpretation respectively. Measurement systems errors may confound dimensional data and data interpretation unless those errors are strictly controlled Pattern Allowance As castings cool from the solidification temperature to room temperature, they undergo dimensional changes. Additionally, steel castings may undergo post-processing such as heat treatment, grinding, and cleaning. Therefore, steel casting features can be expected to have different size than the pattern features from which they were made. This difference between the casting dimension and the pattern dimension is termed as pattern allowance and is defined as shown in Equation 1.1 [5].

28 6 (pattern feature size) - (casting feature size) Pattern Allowance[%] = (Casting feature size) In other words, pattern allowance is the amount by which the pattern feature dimension is corrected from the desired casting size to compensate for the dimensional changes of mold and casting during processing. The pattern allowance is traditionally called the pattern shrinkage allowance, the shrink rule, pattern prediction factor or the patternmakers shrink. It is typically expressed in the US in units of inches per foot or as a percentage of the casting feature size. However, names using the term shrink are misleading since casting shrinkage upon cooling is only a part of the dimensional changes that contribute to the total pattern allowance [6]. Liquid metals experience three stages of contraction as they solidify and cool to room temperature: liquid contraction, solidification contraction and solid contraction as shown in Figure 1-4. The contraction taking place during liquid cooling and early stages of solidification can be expected to have very little effect on the final casting dimensions due to risers. However, solid contraction is expected to contribute heavily to the shrinkage of metal affecting final casting dimensions [7]. This solid contraction component of the overall pattern allowance can be analytically predicted and experimentally measured. The volumetric solid contraction can be calculated by using Equation 1.2. V= β T V 1.2

29 where, V is the change in volume, β is the volumetric thermal expansion coefficient of the casting alloy, T is the change in temperature, and V is the original volume of the casting. However, for all practical purposes, this volumetric casting shrinkage is measured and expressed as linear casting shrinkage and is approximated using Equation 1.3. L= α T L where, L is change in casting feature length, α is the linear thermal expansion coefficient of the casting alloy, T is the change in temperature, and L is the original casting feature length. Figure 1-4: Three phases of solidification of molten metal [7].

30 8 Much of the dimensional change between the casting and its pattern is caused by the thermal contraction of solid metal as it cools from solidification temperature to room temperature. In addition, the mold shape and size can be expected to change during and after pouring [7, 8] as shown in Figure 1-5 (mold size and shape changes are exaggerated for illustration). Furthermore, the mold itself can be expected to restrict the contraction of some casting features upon cooling to room temperature. Other factors such as nonuniform cooling, geometric complexities, mold wall movement, mold assembly inaccuracies, mold and core parting lines, casting heat treatment, and oxide scale removal can also be expected to contribute to variations in the expected pattern allowance for any given feature [3]. Figure 1-5: Exaggerated size and shape changes in mold during and after pouring [7]. Due to complex shrinkage phenomenon and the numerous process and geometry variables influencing casting dimensional behavior, prediction of pattern allowance for steel castings and castings in general is difficult [9]. A survey of steel foundries has

31 9 revealed an inconsistent application of pattern allowances for similar casting geometries indicating that the prediction of pattern allowances is complex and not well understood [2]. 1.3 Pattern Allowance Prediction Tool In order to compensate for the difficulties in centering casting feature dimension within the dimensional tolerance limits, casting designers often specify wider dimensional tolerances and then, large machining allowances are necessary. This leads to additional machining costs, which may force part designers to choose alternative manufacturing methods. It is not unusual for new patterns to undergo one or more cycles of dimensional re-engineering, as shown in Figure 1-6, to ensure that all features of a new component are effectively centered within the customer s dimensional tolerance limits. This dimensional re-engineering is estimated to account for 60% of the production lead-time for new casting production [2]. Although, casting dimensions can be effectively centered within the customer specifications through expensive dimensional re-engineering cycles, as shown in Figure 1-6, often times, inadequate inspection sampling sizes lead to incorrect pattern modifications. This is particularly true for short production run castings where it may only be possible to sample a single first article casting to qualify the pattern feature dimensions. Therefore, it is essential to design the pattern with such pattern allowances that casting dimensions are effectively centered within customer tolerance limits without requiring expensive pattern re-engineering.

32 10 Estimate Pattern Dimension and make a Pattern Pour sample casting(s) No Modify pattern/ core box dimensions Casting dimension acceptable? Yes Pour production casting(s) Figure 1-6: Pattern dimensional re-engineering cycle for centering casting dimension within customer specification. Selecting proper pattern allowances is ultimately the responsibility of the foundry; however, experienced patternmakers may often have full or partial responsibility for selecting pattern allowances in some cases. A separate pattern drawing may not be created. This is further complicated by the fact that during their lifetime, patterns may go through numerous dimensional iterations due to dimensional re-engineering. The final pattern dimensions can be very different from the initial dimensions for some features. Pattern dimensional changes during dimensional re-engineering cycles may not be wellarchived. Therefore, initial poor pattern allowance estimates, whether selected by patternmaker or metal caster may be repeated again and again for other similar castings. In order to avoid the wasteful modification cycles, a dimensional management and recordkeeping system for pattern allowance information is essential for foundries. It is more accurate, cheaper, and easier to preserve the critical pattern dimensions and

33 11 dimensional modification data digitally than in the form of paper part files that are currently used in most foundries. This digital dimensional database can be useful in understanding the shrinkage behavior and avoiding repeated selections of bad initial pattern dimensions. Most foundries do not effectively manage their dimensional data in such a way as to allow them to learn to select proper pattern allowances. Typically, pattern allowance adjustments that are necessary for certain types of casting features are made through somewhat random trial and error dimensional re-engineering. However, a digital dimensional database developed over time can assist in learning to select the proper pattern allowances for particular types of casting features. Figure 1-7 shows the pattern dimension estimation cycle using a digital dimensional database with a feedback loop to assist pattern dimension estimations based on the historical pattern allowance data. Estimate Pattern Dimension and make a Pattern Pour sample casting(s) Digital database dimensional Casting dimension acceptable? No Modify pattern dimensions Yes Pour production casting(s) Figure 1-7: Pattern dimension estimation cycle with feedback loop and digital dimensional database used for estimation of pattern dimension.

34 12 The near-net-shape capabilities of metal casting processes cannot be fully exploited due to the lack of pattern allowance prediction tools. This is particularly true when using contemporary CAD/CAM tools for casting design and pattern manufacture. CAD/CAM technologies offer the opportunity for integrated computer-aided engineering from product design to manufacturing. In the metal casting industry, casting drawings or solid models are regularly generated in CAD software. These casting solid models are then analyzed with software tools using solidification simulations for riser and gating design. Pattern solid models can be easily generated from casting solid models by applying a uniform scaling factor for the pattern allowance. Tool paths for CNC machining of patterns can be directly generated from pattern solid models in CAM software. This, at first glance, would provide a seamless flow of information from casting design to pattern manufacturing; however, the pattern allowance is not a simple scaling factor that can be applied to all casting features uniformly. The capability of predicting pattern allowances is lacking in any of the contemporary CAD/CAE tools. Hence these tools cannot be fully exploited for seamless CAE casting production by the foundry industry. In summary, due to lack of effective pattern allowance prediction techniques, the foundry industry has to deal with 1. Long lead-times and expensive pattern dimensional modification cycles that are required for every new pattern. 2. Near-net-shape casting capabilities that cannot be fully exploited resulting in wasteful post-casting machining.

35 13 3. Dimensional non-conformance resulting from pattern errors leading to a perception of poor dimensional control. 4. Loss of market share and missed opportunities to gain new markets. 5. Computer integrated technologies that cannot be fully applied for the improvements in dimensional integrity that they can offer. In order to increase the customer acceptance and market share of steel castings, it is imperative to develop better pattern allowance prediction tools. 1.4 Thesis Statement This thesis investigates key geometric and casting process variables affecting pattern allowances for steel castings. A new empirical model has been developed to predict pattern allowances for steel castings in this thesis. This model predicts pattern allowance for steel casting features depending on following casting process and feature geometry variables: 1. Casting Alloy Type 2. Molding Method 3. Casting Feature Size 4. Casting Feature Type (Mold restraint) 5. Mold and Core Parting Line This model has been validated using extensive pattern allowance data generated from production steel castings. A pattern allowance prediction and dimensional management software tool has been developed. This software is bundled with a

36 customization module to permit learning from the dimensional data of a specific foundry and customize the pattern dimension estimation Thesis Outline Research has been conducted in the past to demonstrate the effects of various process and geometry variables on the pattern allowances in steel castings. Chapter 2 provides a review of relevant pattern allowance prediction literature. Chapter 3 illustrates the shortcomings in existing analytical and statistical pattern allowance prediction models for steel castings and establishes the need for a new pattern allowance prediction model for steel castings. In order to build a new statistical model, dimensional data from twenty different production steel foundries is collected. Chapter 4 presents the detailed data collection procedures, steps taken to ensure dimensional integrity of the data, and summaries of the collected data. Chapter 5 presents the new proposed pattern allowance prediction model along with model validation using production steel foundry data. For easy commissioning of the new model in the production steel foundries, Pattern Allowance Advisor software is developed. Chapter 6 provides detailed logic and flowcharts for the Pattern Allowance Advisor software. Validation of this software through industrial implementation is also presented in Chapter 6. Chapter 7 presents conclusions, contributions to the technology, advantages and limitations of the proposed approach along with future work.

37 Chapter 2 Literature Review 2.1 Dimensional Tolerance Standards for Steel Castings Dimensional tolerance specifications for steel castings are commonly specified in the US using T grades developed by Steel Founders Society of America (SFSA) [10]. American Society for Testing and Materials (ASTM) or US government (MIL-S) dimensional tolerance standards do not exist for steel castings. In the SFSA specification, allowable casting dimensional tolerance depends on both feature dimension and overall casting weight. Figure 2-1 shows tolerance limits for SFSA T Grades as a function of casting weight. Table 2-1 lists the equations used for calculation of tolerance values in inches for SFSA T Grades where W is casting weight in pounds and L is feature length in inches. Table 2-1: Equations for SFSA dimensional tolerance specifications T grades [10] Tolerance Applicability Tolerance Grade T3 Premium sand casting processes TOL(T3) = 0.005W1/ L1/3 T4 - TOL(T4) = 0.010W1/ L1/3 T5 Standard Production techniques TOL(T5) = 0.016W1/ L1/3 T6 - TOL(T6) = 0.027W1/ L1/3 T7 Complex Casting Configurations TOL(T7) = 0.044W1/ L1/3

38 16 Figure 2-1: SFSA T grades [10]. These SFSA standards are based on dimensional tolerance data generated from a 1977 study by SFSA [11]. SFSA-T3 grade represents tolerances that can be achieved by premium sand casting (using shell or nobake sand mold) processes. SFSA-T5 grade represents tolerance values that can be achieved by standard production techniques. SFSA-T7 grade represents tolerance values that can be expected for complex casting configurations. Wider tolerances are suggested for dimensions crossing the parting line and dimensions controlled by two or more cores. The International Standards Organization (ISO) first published casting dimensional tolerance specifications in 1984 [12] ISO , Casting System of Dimensional Tolerances. This standard was first revised in 1994, ISO [13]. Some major revisions and additions were suggested in 1996, ISO [14] but have

39 17 not been adapted yet. This standard provides guidelines describing dimensional capabilities for many casting alloys and casting processes including steel castings. Sixteen tolerance grades CT1 to CT16 (tolerance limits increase with each grade) form the basis of this standard. However, unlike the SFSA casting dimensional tolerance specifications, this standard does not recognize casting weight as a parameter influencing casting dimensional variability. The tolerance grade to be assigned to a particular casting feature is determined based on the feature size, casting process and the alloy being cast. Figure 2-2 shows the total tolerance in inches with respect to feature size for tolerance grades CT4 to CT12. Separate sets of tolerance grades are provided for castings produced in either short production runs or long production runs. Within each grade, specific tolerances are then assigned on the basis of the length of the feature to which the tolerance was being applied.

40 CT Total Tolerance (Inches) CT11 CT10 CT9 CT8 CT Figure 2-2: ISO CT tolerance grades [13] Feature Size (Inches) CT4 CT6 CT5 Geometric dimensioning and tolerancing is increasingly being used in the casting industry. Recommendations on the geometric tolerance capabilities of casting processes were first published as ISO in 1995 [15]. This standard provides 7 geometric tolerance grades (CTG2-CTG8) for casting feature characteristics such as flatness, circularity, straightness, etc. The ISO 8062 specification recognizes that closer tolerances can be specified for long run castings or for steel castings made in shell and nobake sand molds as compared to green sand molds. ISO suggests that casting complexity and length of production run affect the recommended tolerance grade; however neither is precisely defined resulting in tolerance grade selection that is subject to interpretation. German DIN standards also specify dimensional tolerances for steel castings [16].

41 19 SFSA T-5 dimensional tolerance grade has been found to be tighter than ISO and DIN standards [17]. Peters et al. [23] demonstrated that current published dimensional specifications are not realistic and that production foundries can regularly achieve better dimensional performance than SFSA, ISO, and DIN standard specifications suggest. Many of the major steel casting users have developed their own propriety casting dimensional tolerance standards. These standards are generally tighter than any of the published standards [23]. Ross [29, 30] has offered an explanation for overly conservative published dimensional tolerance specifications. Ross has shown that the magnitude of the measurement error in foundries is very high due to poor measurement system gage R & R which would tend to encourage foundries to underestimate their dimensional capabilities. 2.2 Dimensional Variability Dimensional variability in steel castings has been measured by several investigators through data collection in production foundries. Steel and iron castings dimensional capability data was first collected by Institute of British Foundryman in 1969 [19] and 1971 [20]. These studies surveyed a number of production castings in order to determine the dimensional tolerances obtainable by normal production methods. Several variables in the sand casting process that contribute to dimensional variability were identified. The study also developed regression equations to understand the effect of these variables on dimensional variability. Similar studies were carried out by Villner

42 20 [21] and Svenson and Villner [22] as part of an effort to develop an improved dimensional tolerance scheme for castings. Dimensional variability of metal casting processes for various alloys and molding methods has been under investigation for the past decade at The Pennsylvania State University. Dimensional variability in production steel casting has been evaluated and compared with published tolerance standards [23]. Several studies have derived dimensional capability and repeatability from the measured dimensional variability to access true capabilities of production steel foundries [23,24,25]. Vaupel et al. [26] investigated the implications of tolerance system interpretation on dimensional variability studies. Faustine et al. [27] studied dimensional variability of steel castings based on feature type. They characterized geometric variability, in particular for large casting features, for the first time. It was observed that in general flatness, perpendicularity, circularity, and parallelism variations are similar to the variations that are indicated in ISO geometric tolerance specifications [27]. Karve et al. [28] reported important factors influencing dimensional variability of steel investment castings. 2.3 Inspection Errors and Sampling Uncertainty Measurement System Errors Ross has shown that the magnitude of the measurement error in foundries can be on the order of the product variability [29, 30]. According to the Automotive Industry Action Group reference manual on measurement system, for a measurement system to be

43 21 acceptable, the measurement error must be no greater than 30% of the part variability that is being measured [31]. Similarly, Peters et al. [32] developed a simple procedure for pattern shops to check adequacy of pattern measurement systems. Karve et al. [33] have reported the performance of commonly used foundry measurement systems and identified major factors that affect the performance of measurement systems. It is emphasized that any dimensional studies without explicit attention to measurement system gage R&R need to be questioned. Table 2-2 shows the acceptability interpretation of gage R&R test. Table 2-2: Acceptability criteria for Gage R & R Tests [31] Percent Gage R & R Interpretation 10% or less Preferred 10% - 30% Acceptable greater than 30% Not Acceptable Dimensional Uncertainty due to Insufficient Sampling Insufficient sample size for dimensional inspection of casting features during the dimensional re-engineering cycles can introduce significant dimensional errors in pattern modification and hence in the final casting dimensions. This uncertainty error and its effects on foundry industry are investigated in detail by Karve et al. [34] and Potter et al. [4]. A statistical technique was developed by Potter et al. [4] for the determination of sample size for pattern tooling validation based on the process variability and customer tolerance. A process capability ratio as shown in Equation 2.1 was used to derive statistically sufficient sample size.

44 22 Process Process Variation (6σ ) Capability Ratio = 2.1 Total Customer Tolerance Table 2-3 presents statistically determined minimum sample sizes required for pattern approval with 95% confidence, with respect to the process capability ratio [4]. Table 2-3: Statistically determined minimum number of sample castings required for pattern approval process with 95% confidence [4]. Process Capability Ratio Minimum Sample Size Less than More than Table 2-3 suggests that at least 44 castings need to be inspected when process capability ratio is more than 0.6. However, the willingness and ability of the foundries to dimensionally inspect statistically required number of castings depends heavily on the order casting quantity, lead time constraints, and the dimensional complexity of the casting [34]. Especially in case of short-production run castings with complex geometry, it may be feasible to inspect only one casting, introducing uncertainty in the prediction of mean casting feature dimensions. In such cases, it was proposed that an additional compensating component of dimensional error, as shown in Figure 2-3, can be incorporated into the casting dimensional variability estimates. This compensating factor is derived statistically and is expressed as a multiplier for the observed casting dimensional variability [34]. Table 2-4 shows the multiplying factors for observed variation (6σ) tabulated by actual castings inspected and minimum desired sample size.

45 23 Table 2-4: Dimensional Variability Multiplying Factors [34] Actual number of castings sampled Minimum desired sample size shown in Table Dimensional Variability (6σ) Multiplying Factors Figure 2-3: Additional component of dimensional variability estimates due to sampling uncertainty errors.

46 Analytical Modeling of Pattern Allowance in Castings Casting alloys shrink during solidification and during subsequent cooling to room temperature. Attempts to model shrinkage in casting have been directed mainly at the liquid-solid shrinkage during solidification, and not at the subsequent linear contraction that affects final part dimensions. The majority of these modeling efforts have focused on developing models to determine the location and distribution of shrinkage porosity in castings. The International Lead and Zinc Research Organization funded a research project at the University of Swansea, UK, which attempted to model the effect of shrinkage on zinc die casting dimensions [35]. The researchers developed a finite element code that modeled the residual stresses developed in a casting as it cooled and attempted to predict the dimensional change upon cooling based on these stresses. The model accurately predicted the dimensions of simple shapes; however, when this model was validated on more complex test castings, the results were not adequate. At the end of the research program the authors reported that further development of the computer model must be closely paired with extensive experimentation for verification and validation of these models [36]. Validation experiments indicated that the models could predict the trend in dimensions, but the actual shrinkage values were not accurate. The mold dynamics in the rigid metal dies used in this die casting study can be expected to be significantly different than the mold dynamics of sand mold used for steel castings. Finite element models for the prediction of pattern allowances in green sand steel castings have been recently evaluated by Beckermann et al. [37]. Predicted pattern

47 25 allowance was compared with measured pattern allowance values for both simple and complex casting shapes. Pattern allowance data for complex and simple casting shapes was collected from production foundries and from the Penn State University foundry laboratory respectively [37]. The work concluded that pattern allowance trends could be predicted reasonably if the casting shape was simple and accurate thermal expansion properties for the steel, mold, and core materials are used in the simulations. Beckermann suggests that the irreversible expansion of silica sand mold and formation of an air gap between the casting and the mold during cooling cannot be accounted for in the stress model developed in his work. Predicted pattern allowance values for complex production foundry steel castings were, therefore, different from the observed pattern allowance values [37]. The reliability of such pattern allowance prediction models is strongly dependent on the use of good thermo-physical property data for casting alloys and molding materials at or near the solidification temperature. These data are largely non-existent. Furthermore, these models do not account for other post-processing such as heat treatment and oxide scale removal that can be expected to influence pattern allowance values. Production foundries, however, are interested in the final casting dimension after heat treatment and oxide scale removal. Currently, no models for prediction of pattern allowance using analytical casting simulations have been developed [37]. Work is on going at University of Iowa and other places in this direction. Complex casting geometries, air gaps between mold and casting, irreversible expansion of sand, molding parting line and lack of material properties at high temperatures are just some of the most difficult problems hindering analytical

48 prediction of the pattern allowance values using widely used heat transfer and residual stress / distortion models [37] Factors Affecting Pattern Allowances in Steel Casting Various casting and process variables can be expected to affect pattern allowances. Molding process, casting alloy, molding material, and even pattern material can be expected to affect pattern allowances [10, 38]. Castings can be generally classified according to these factors for pattern allowance prediction. Furthermore, other variables such as feature length, casting shape, and molding restraint are reported to affect pattern allowance [3, 39]. Ivey [40] conducted regression analysis to determine statistically significant process and geometry variables affecting casting dimensions in sand castings. Okhyusen [41] conducted similar studies for investment castings. Figure 2-4 summarizes the casting process and geometry variables whose effects on pattern allowance have been statistically shown to be significant by various researchers.

49 27 Mold Distortion Mold Restraint Unrestrained Features Casting Alloy Mold Type Fully Restrained Features Core Type Mold Parting Lines Partly Restrained Features Pattern Allowance Heat Treatment Post Processing Feature Size and Shape Grinding and Cleaning Geometry Neighboring Casting Geometry Figure 2-4: Casting process and geometry variables affecting pattern allowances Effect of Alloy Type on Pattern Allowance Solidification contraction is a major component of pattern allowance. The amount of solidification contraction is directly proportional to thermal expansion coefficient of the casting alloy as illustrated in Equations 1.2 and 1.3. The thermal expansion coefficient is a thermo-physical property of a metal or alloy. Therefore different pattern allowances should be used for different ferrous casting alloys as shown in Table 2-5. The patternmaking guide [38] provides general pattern allowance estimates in the units of inches per feet (for British Units) as well as a multiplying factor (for SI units) for

50 various alloys as shown in Figure 2-5. Various Penn State researchers [3, 39] have also attributed different pattern allowances to different alloy types. 28 Table 2-5: Shrinkage of various steel alloys for C ( F) temperature range [42] Carbon and Low alloy Steels Shrinkage Allowance % 0.23%C %C %C %Mn 1.86 High Alloy Steels 13% Mn %Ni 1% Cr-Mo %Ni % Cr 1.62

51 29 Figure 2-5: Expected pattern allowance values for various alloys in inches per foot and as a multiplying factor [38]. This general shrinkage behavior is even more complex for alloys such as steel that go through a phase change after solidification. Briggs et al. [43] demonstrated the effect of phase change on steel casting contractions for various restraint levels as shown in Figure 2-6. Furthermore, the density of the casting alloy may affect the mold wall displacement that also influences final overall casting feature size and shape.

52 30 Figure 2-6: Contraction of 0.35C steel cast into greensand molds [43] Effect of Sand Type on Casting Dimensions Variation in mold dimensions during and after pouring plays an important role in the determination of pattern allowances as shown in Figure 2-7. Ward [44] investigated changes in mold dimensions and classified mold displacements that occur during casting into permanent linear changes and mold dilation. The permanent linear change of the mold was defined as the irreversible dimensional change of the mold when heated and then cooled. Mold dilation was defined as the mold displacement due to the outward pressure exerted by dense metal on the mold wall. Permanent linear change was

53 31 determined by measuring the expansion of samples of the sand / binder system after heating for 2 hours at 1550 C (2822 F). Table 2-6 shows the magnitude of permanent linear changes in samples of mold binder systems. The molds were measured, metal was poured in them, and the castings were measured at the same locations. The mold displacement was calculated directly from the difference of the feature sizes in molds and castings. Ward concluded that the binder systems that exhibited the greater permanent linear change had the smallest mold displacement caused by the metal. Figure 2-7 shows temperature expansion curves for various mold materials. Table 2-6: Permanent linear change in various sand/binder systems [44] Sand / Binder Permanent Linear Change % Silica / Sodium Silicate Silica, Natural / Core Oil 7.2 Silica / Bentonite -1.54

54 32 Figure 2-7: Temperature expansion curves for various mold materials [44] Sosman [45] observed four distinct regions in green molding sand adjacent to steel casting as shown in Figure 2-8. There is a dry sand zone adjacent to the steel followed by a vapor transport zone where steam is migrating. Water vapor from this region travels into the sand, where it cools, condenses, causing a low strength condensation zone [46] or 'mud layer'. Outside of this layer is the external zone where temperature and water content remain unchanged.

55 33 Figure 2-8: Various zones in a sand mold after metal pouring [45] Bates [47] concluded that hydrostatic pressure causes a biaxial tension in the dry layer. The dry area is in tension because there is no support provided by the mud layer. Mold displacement was measured using a probe implanted into the mold. Expansion of the mold occurs due to the thermal expansion of the refractory mold material. For sodium silicate bonded molds, outward mold wall expansion was limited by the high strength of the bond and the lack of a weak high moisture layer. Because of the rigidly bonded sand, the expansion causes an inward displacement of the sodium silicate molds. This displacement occurred before a sufficient skin of solidified steel was formed. The

56 34 inward mold wall displacement due to thermal expansion was on the order of 0.89 mm (0.035 inches) for the sodium silicate bonded molds [47]. Green sand molds made with zircon sand did not exhibit as much mold wall displacement as those with silica sand. It has been hypothesized that the increased heat extraction capacity of the zircon sand chilled the steel faster, resulting in the development of a solidified skin before additional mold wall displacement occurred. Increased hot strength of the sodium silicate bonded molds decreased the magnitude of all mold wall displacements [47]. Engler [48] compared the mold wall displacement for green sand molds made with silica and zircon sands, and the effect of other sand additions. Radiation heated green sand molds exhibited mold wall displacements similar to molds in which steel was poured. This indicates that the hydrostatic pressure from the liquid steel was not an important variable affecting mold wall displacement for highly compacted molds. Hence it was concluded that mold wall displacement may be dependent on the compressive strength of the sand. Mold wall displacement for green sand and chemically bonded sand molds was measured using linear transducers by Rickards [8]. For the chemically bonded sand molds, the cylindrical mold cavity contracted during solidification. The green sand mold cavities experienced dilation. An inward movement of flat mold wall was measured prior to a subsequent outward movement. However, Winter et al. [49] did not measure any inward movement of dry sand molds. This contradiction is explained in work carried out by Levelink et al. [50], which proposes that inward expansion may happen when the mold surface is convex towards the casting as shown in Figure 2-9. Nishida et al. [51] confirmed this conclusion by carrying out casting and mold measurements.

57 35 Figure 2-9: Inward expansion of mold wall in convex regions and outward expansion in concave regions [50] Bertolino [52] performed fundamental studies on the solidification of steel for different mold geometries. Bertolino found that the steel formed a skin that took the shape of the mold cavity. After a skin of sufficient thickness was formed, any subsequent mold wall changes were reported to not affect the shape of the casting. Mold wall displacements for cube, cylinder, and plate-shaped castings were studied in green sand, dry sand, and sodium silicate bonded sand molds. Castings of iron, steel, aluminum, and copper were poured. The dry sand and sodium silicate bonded sand molds exhibited less overall dimensional mold change. This was explained by the ability of the mold to resist outward wall displacement and the occurrence of inward displacement due to thermal expansion. It was hypothesized that the inward displacement occurred because the

58 36 outward displacement was restricted by the rigidly bonded sand. Increased riser height produced a greater hydrostatic head pressure that resulted in increased casting size. Increased pouring temperature increased the size of the castings made in green sand molds but did not have a significant effect on the casting dimensions for the other molding materials [52] Effect of Feature Length on Pattern Allowance Feature length affects pattern allowance as it affects dimensional variability. Lucking and Pacyana [53] proposed different shrink rates for castings of different length in green sand steel castings as shown in Table 2-7. A 1977 SFSA study [9] also suggested that steel casting shrinkage values depend on the feature dimension changes as shown in Figure Table 2-7: Influence of feature size on the contraction of green sand steel castings [53] Dimension Median Shrink (%) Median Deviation (%) D> < D < >D

59 37 Figure 2-10: Effect of feature length on pattern allowance for steel castings [10] Explanation for change in the shrinkage allowance along with the length was not provided; however, it can be hypothesized that this change can be partly attributed to the post processing of castings. Vaupel [54] investigated the effects of post processing including cleaning, heat treatment, and grinding on the dimensional variability of sand steel castings. Vaupel showed that the scale thickness on a casting surface is relatively uniform. Therefore, the resultant effect of scale removal on pattern allowance is not uniform and varies according to the feature dimension [54]. Okhuysen et al. [55] investigated the effect of heat treatment on pattern allowances for steel investment casting. This work illustrates that heat treatment and scale removal result in variable pattern allowance for different feature lengths as shown in Figure 2-11.

60 38 Figure 2-11: Effect of feature length on the pattern allowances of investment castings due to scale removal [55] Effect of Mold Restraint on Pattern Allowance Researchers have shown that the restraint provided by internal cores can influence the resulting casting size; however, the amount of restraint can be expected to be dependant on many variables. Various researchers have demonstrated the complex nature of the problem. Briggs [43] conducted experiments on the contraction of steel castings for differing levels of casting geometry constraints. Briggs noted that the overall amount of contraction decreased with increasing constraint as shown in Figure 2-6. Furthermore, Briggs demonstrated and documented the influence of material properties on the amount of contraction along with the amount of restraint. Lucking [53] investigated the shrinkage of casting features from machine molded production steel castings. Three hundred features were measured from 47 different

61 39 casting designs. For each feature, 50 castings were measured to establish the median casting dimension value. Lucking s work noted that the pattern allowance was a function of feature length, mold restraint (termed as internal and external dimensions) and shape of the casting. Lucking concluded that the internal dimensions with mold restraint showed reduced shrinkage. The following reasons were presented by Lucking as possible causes for this reduced shrinkage in internal dimensions: The general shrinkage toward the geometric center of the casting is accompanied by a contraction of the casting to the mass center in the corresponding section. For inside dimensions, this will be a positive dimensional change which counteracts the shrinkage to the geometric center of the casting. Oxidation losses will also increase the dimensions of internal features [53]. Nyichomba [56] used a bar casting with flanges to study the relationship between mold constraint, mold type, and casting contraction. Mold types strongly influenced mold displacements and hence the casting contraction. It was further demonstrated that the mold type strongly influenced casting dimensions for some materials such as grey iron and copper alloys, whereas white iron castings experienced the same casting contraction, regardless of mold type. It was suggested that during initial cooling of a constrained casting, plastic flow will relieve some of the forces which develop due to metal contraction. Insulating molds resulted in slower cooling which permit more time for metal plastic flow and hence exhibited less contraction. The Steel Castings Handbook [10] classifies casting dimension types as shown in Figure 2-12 according to mold restraint. It is suggested that the dimensional tolerance and pattern allowance values are different for each dimension type due to these mold relationships and differences in mold restraints [9, 10]. In addition to offering restraint to

62 40 shrinking metal, molds or cores may themselves expand leading to reduced or even negative pattern allowance (feature expansion) for the solidifying metal as shown in Figure Figure 2-12: Dimension types due to mold restraint [10]

63 41 Figure 2-13: Effect of mold-casting interaction on pattern allowance [57] Researchers at Penn State devised a simpler and more specific classification scheme for casting dimension types as shown in Figure Table 2-8 describes the different types of dimensions as indicated by the letters A through I. This easier scheme was used for dimensional studies of complex shaped casting features at production foundries. Karve et al. [58] investigated the mold restraint in detail and classified all casting features into three categories according to the mold restraint fully restrained features, partially restrained features and unrestrained features. Table 2-9 illustrates the classification into these categories with reference to Figure This classification scheme has been used by several researchers [58, 59, 60] for pattern allowance prediction for all types of castings. Voigt et al. [39] applied this scheme of feature classification to present pattern allowance values for steel castings in sand molds.

64 42 D B C A H F G E MOLD Casting MOLD Casting MOLD Casting Core Casting MOLD Figure 2-14: Penn State casting feature classification [58] Table 2-8: Explanation for various dimension types. Refer to Figure 2-14 [58] A mold to mold across casting F Core to core across core B mold to mold across mold G Mold to core across casting and core C mold to mold across mold and casting H Mold to mold across casting/core/ Casting D mold to mold across I Core to core across casting casting/mold/casting E mold to core across casting O Other configurations Table 2-9: Classification of casting features according to levels of restraint. Refer to Figure 2-14 [58] Feature Type in Qualitative Level of Quantitative Level of Restraint Figure 2-14 Restraint A, E, I Unrestrained 1 C,D,G, H Partly Restrained 0 B, F Fully Restrained -1 This three-level classification scheme is simple and easy to use in the production foundry environment. However, all partially restrained features are assigned the same level of restraint in spite of the fact that some partially restrained features may have

65 43 considerably more restraint than others. This leads to a lack of resolution in the model, as the amount of restraint acting on the feature may not be the same for all partially restrained features Effect of Casting Feature Shapes on Pattern Allowance There have been no systematic efforts to model the effect of casting shape on casting dimensional variability or pattern allowances. Lucking and Pacyana [53] described shrinkage factors for sand steel castings based on the shape of the part. Casting shapes were classified as solids, rings, frames and U-shapes. Pattern allowances for the various shapes were reported as shown in Table There is no explanation provided in their work for the differences in shrinkage observed as a function of feature geometry. Furthermore, classification depending on casting shapes that are not clearly defined and hence are open to interpretation. Table 2-10: Pattern allowance for various casting shapes for steel sand castings in green sand molds [53] Shape Median pattern allowance (%) Deviation from Median pattern allowance (%) Solid Ring Frame U-shape Bertolino [52] has speculated on the effect of casting geometry on dimensional changes of metal castings based on limited mold displacement measurements on the

66 44 largest side for some common cast shapes. For each of the metals cast, the amount of mold wall displacement decreased as a function of the logarithm of SA 2 /V 2 for the green sand and sodium silicate bonded sand molds (where, SA and V are the surface area and volume of the casting, respectively). Bertolino s explanation for these results was that for 'chunkier' castings (lower SA 2 /V 2 ratios), the sand is heated more and the mold wall is subjected to hydrostatic head pressure longer [52]. Henschel [61] conducted similar research with green sand molds and made similar conclusions on the effect of SA 2 /V 2 on mold displacements for aluminum and iron castings. 2.6 Pattern Allowance in Steel Castings A number of researchers have investigated aspects of pattern allowance in steel castings. Several empirical and statistical models for pattern allowances in sand steel castings have been approximated. Most of the existing models are empirical; however, some attempts have been made to develop analytical pattern allowance models using finite element or finite difference methods Industry Pattern Allowance Practices To determine how industry experts apply pattern allowances to steel castings, Voigt [17] surveyed several production steel foundries. Pattern allowance values ranged from 0.8 to 2.6% for common casting features. Some foundries used the same pattern allowance value for all features on a casting, regardless of the casting size and feature

67 45 geometry. The foundries that used different pattern allowance values for different casting feature types did not apply the pattern allowance factors consistently from foundry to foundry. Wieser [9] has also presented production pattern allowance data based only on the overall casting length as shown Figure Variation from the standard pattern allowance (hence difficulties in pattern allowance prediction) were observed especially for small casting features. Wagner [62] conducted a similar survey in Germany which revealed very similar conclusions Empirical Pattern Allowance Models for Sand Steel Casting Processes The American Foundry Society Patternmaking Guide [38] published in the literature provides a rudimentary list of standard pattern allowance values for various alloys as shown in Figure 2-5. Uniform pattern allowance values of 1/4 in/ft, 5/16 in/ft and 3/16 in/ft are suggested for all types of steel casting features, irrespective of other casting geometry and process variables. The 1980 SFSA steel castings handbook [10] classifies casting features into various categories with respect to the mold restraint as shown in Figure According to this, different pattern allowance values need to be used for the various pattern features based on mold restraint and casting shape. It also suggests that different pattern allowance values need to be used with different types of steel, casting design and molding methods. It further suggests that two different foundries may need to use different pattern allowances for the same casting depending upon molding process variables. A pattern allowance of 3/16 in/ft (1.56%) for carbon and low alloy steels is suggested as most

68 46 commonly used whereas pattern allowances of 7/32-11/32 in/ft ( %) are recommended for high alloy steels [10]. Lucking and Pacyna [53] have presented a feature length dependent pattern allowance model for steel castings made in green sand molds in terms of median and deviation from the median pattern allowances as shown in Table 2-7. However, the variation in pattern allowance suggested due to feature length in this model is based on a stepwise function. No explanation for the step function is provided. This model also suggests variation in pattern allowances based on mold restraint by classifying dimension types as inside or outside dimensions as shown in Table Furthermore, this model suggests that different pattern allowances should be used depending on the casting shape as shown in Table Table 2-11: Influence of dimension type on the contraction of green sand steel castings [53] Dimension Type Median pattern allowance (%) Deviation from the Median (%) Inside Outside Researchers at Penn State University conducted an extensive survey of steel castings production foundries and established a pattern allowance model for low and high alloy steel castings [39, 59]. Different pattern allowance values were suggested for green sand, nobake and shell castings for fully restrained, partially restrained, and unrestrained features that did not cross mold or core parting line. Average pattern allowance values were suggested along with 80% confidence limits as shown in Table A software The Dimensional Engineering Software was developed to implement this model in the

69 production steel foundries [74]. Average and 80% confidence interval pattern allowance values reported did not change with feature length, casting weight, or casting shape. 47 Table 2-12: Average and confidence interval based model for high and low alloy steel castings in green sand, no bake and shell molds [39] Condition Average Pattern Allowance 80% Confidence Interval for Pattern Allowances Low Alloy Steel Overall 1.96% 1.85 to 2.07% Green sand molding, overall 1.60% 1.43 to 1.77% Un-restrained features 1.56% 1.15 to 1.97% Partially restrained features 1.74% 1.56 to 1.92% Fully restrained features 1.61% 1.48 to 1.74% No bake molding, overall 2.39% 2.20 to 2.58% Un-restrained features 2.33% 1.94 to 2.74% Partially restrained features 2.32% 2.06 to 2.59% Fully restrained features 2.03% 1.75 to 2.30% Shell molding, overall 2.31% 2.10 to 2.51% Un-restrained features 2.87% 2.58 to 3.16% Partially restrained features 2.31% 2.13 to 2.48% Fully restrained features 1.27% 0.91 to 1.63% High Alloy Steel Overall 2.92% 2.72 to 3.11% Green sand molding, overall 4.21% 3.82 to 4.59% Un-restrained features 3.62% 3.34 to 3.83% Partially restrained features Fully restrained features 5.37% 4.98 to 5.76% No bake molding, overall 3.50% 3.08 to 3.92% Un-restrained features 4.04% 3.46 to 4.63% Partially restrained features* Fully restrained features* Shell molding, overall 2.58% 2.35 to 2.81% Un-restrained features 2.90% 2.57 to 3.24% Partially restrained features 2.42% 2.29 to 2.54% Fully restrained features 1.57% 1.28 to 1.85% * These categories contained insufficient data to estimate pattern allowances

70 48 Chapter 3 Need for Improved Pattern Allowance Prediction Models This chapter presents the need for improved pattern allowance prediction models by illustrating various shortcomings in the existing models. Additionally, the goals for a pattern allowance model are presented. 3.1 Casting Feature Length Figure 3-1 shows the pattern allowance data used to build a recent pattern allowance model [39]. Variation observed in pattern allowance values for green sand, nobake, and shell low alloy steel castings with respect to casting feature length is notable, especially in case of small casting features. The variation in pattern allowances grows wider as the feature size decreases; however, recently developed pattern allowance prediction models [39, 59] recommend the same pattern allowance values for all casting feature sizes. The Lucking pattern allowance model [53] recommends stepwise changes in pattern allowance with respect to feature length, which have not been observed in production castings, see Figure 3-1.

71 49 Figure 3-1: Overall distribution of pattern allowances as a function of casting feature length - green sand, nobake, and shell low alloy steel castings [39] 3.2 Variation in Pattern Allowance Figure 3-2 shows the distribution of pattern allowance with respect to casting feature length for unrestrained features in green sand low alloy steel castings from a recent pattern allowance model [39]. Dotted line in the figure indicates the average value of the pattern allowance, whereas, the dark lines indicate upper and lower estimates of 80% confidence interval as suggested by the pattern allowance model [39]. Due to the variation in pattern allowance as a function of casting feature length, observed pattern allowance values are not adequately described using standard 80% confidence intervals, as shown in Figure 3-2. Furthermore, average pattern allowance values can be heavily influenced by very high or very low pattern allowance values observed for small casting

72 feature lengths. Therefore, confidence interval limits for average pattern allowance are not adequate to express the pattern allowance values % 6.00% 4.00% Observed PA 80% Confidence Interval Average PA 2.00% PA (%) 0.00% -2.00% % -6.00% -8.00% Feature Length (Inches) Figure 3-2: Distribution of observed pattern allowance values as a function of casting feature length along with average and 80% confidence interval values - green sand low alloy steel casting unrestrained features [39]. 3.3 Non-Normality of Pattern Allowance Data In order to build an empirical model, the main assumption is that the underlying data is normally distributed. Figure 3-3 shows the normal probability plot of the observed pattern allowances from a recent pattern allowance model [39] for unrestrained features in green sand molded low alloy steel castings. Anderson-Darling normality test [63] is

73 51 applied with α = 0.05 to check the normality of the pattern allowance data. In order for the data to be normally distributed, the P-value from the normality test has to be greater than α = 0.05; however, P-value for pattern allowance data is only as shown in Figure 3-3. Figure 3-3: Observed pattern allowance Normal Probability plot for unrestrained features in green sand low alloy steel castings. Since pattern allowance data cannot be used to build an empirical model, both due to non-normality as shown in Figure 3-3 and confounding with casting feature size as shown in Figure 3-1, pattern allowance data needs to be transformed into a variable that is normally distributes and independent for modeling purposes.

74 Neighboring Casting Geometry For long casting features, camber distortion was suggested to influence casting dimension and pattern allowance values by Longden [64]. The amount of camber distortion was quantified for grey iron castings. Deo et al. [65] observed variation in pattern allowance values along the length of a steel casting feature as shown in Figure 3-4. It was suggested that in addition to camber distortion, neighboring geometry may affect the pattern allowance values. This influence is not completely understood, qualified, or quantified. The range of variation in pattern allowance values is considerable in some long and heavily influenced features as shown in Figure 3-4. Therefore, pattern allowance model should include the influence of neighboring geometries and distortion due to non-uniform casting cooling rates in pattern allowance prediction. 3.00% 2.50% C Pattern Allowance (%) 2.00% 1.50% 1.00% 0.50% Feature C % Distance from Origin (Inches) Figure 3-4: Variation in pattern allowance along the length of a casting feature

75 Foundry to Foundry Variation According to the Steel Castings Handbook [10] and a production foundry survey by Voigt [17], a pattern producing dimensionally correct castings in one foundry may not produce dimensionally correct castings in another foundry due to the process variations from foundry to foundry. Figure 1-3 illustrates the main causes of such variation. If a pattern allowance model has the capability to be calibrated and adapted to specific foundry process settings, this customized pattern allowance model will yield better results for individual foundries. 3.6 Goals for New Pattern Allowance Model Figure 3-5 summarizes the factors influencing total casting dimension variability discussed in detail in chapter 1 and chapter 3. As discussed previously, tight process control and qualified measurement systems help reduce the variability due to casting process and measurement systems. Effective pattern dimension estimation will reduce the uncertainty due to pattern dimension estimation. Pattern error ( centering error ) will then consume smaller amount of total available tolerance. Reduction in pattern dimension errors will, hence, help improve the casting dimensional conformance.

76 54 Lower Tolerance Limit Upper Tolerance Limit Pattern Making Uncertainty Inspection Uncertainty Pattern Estimation Uncertainty Casting Process Variability Figure 3-5: Summary of factors influencing casting dimensional variability In summary, the development of a new pattern allowance model with following characteristics is needed: 1. Pattern allowances are primarily influenced by alloy type, sand type, mold restraint, and mold or core parting line. 2. Casting feature length influences pattern allowance prediction in a linear fashion rather than a stepped fashion. 3. Pattern allowance variation predictions reflect the wider variation in percent pattern allowances observed for small casting feature lengths. 4. Pattern allowance prediction model can be customized to a specific foundry. 5. Pattern allowance prediction models should include the pattern allowance variations due to casting shape, neighboring casting geometry, and distortion influences due to non-uniform casting cooling rates.

77 Chapter 4 Data Collection and Analysis In order to build and validate an empirical model for the prediction of pattern allowances, casting and pattern dimensional data is collected from more than twenty production steel foundries using two different methods. Point-to-point 2D dimensional data was collected using conventional gages. Linear scanning measurements were conducted using articulated arm CMM 4.1 Point-to-Point 2D Data Collection The 2D data set used to develop this pattern allowance prediction model was collected over past 10 years, and includes casting and pattern dimension measurements for 526 different steel casting features. Casting inspection procedures have been developed and validated by researchers at Penn State [66]. These procedures were followed during the collection of all 2D data. All casting features were measured after heat treatment (if any) and shot blasting but, before any grinding or straightening operation. 77 other casting process and geometry variables were recorded along with the pattern and casting dimensional data. Casting variables survey form, as shown in Appendix A, was used to collect this information. Production foundry data was collected in the form of part files. A part file was generated for each part number and consisted of the following:

78 56 Casting Variables Survey Form Casting Feature Variables Survey Form Pattern Measurement Sheet Casting Measurement Sheet Part drawing with feature numbers Detailed instruction sheet for inspection of each feature. Additionally, the part file may have pictures of pattern and casting. Appendix B shows a sample part file. Each casting feature was assigned a feature type represented by alphabets from A to I following the scheme shown in Figure Measurements were made at the identical location for each of the thirty castings being measured. Measurements were always taken at a specified location on a feature, even if that feature had a constant dimension over a wide range. This permitted the measured casting dimension to be later compared to corresponding pattern feature measurement. Dimensions on drafted surfaces needed special measurement techniques to ensure measurement consistency. Data collection procedures are presented with further details in previous works [39, 40, 66] 4.2 Data Integrity for Point-to-Point 2D Data In order to build a sound empirical model, it is imperative to ensure integrity of dimensional data used to develop the model. As discussed in section 1.2, if casting process variability and measurement system errors are analyzed and controlled during the data collection, only then can the collected dimensional data be used to build a reliable

79 57 empirical model. This section discusses the steps taken to ensure the data integrity through analysis of casting process dimensional variability and measurement system errors Casting Process Variability As shown in Section 2.3.2, when the process capability ratio for the castings is less than 0.6, the statistically required minimum number of samples is eleven for 95% confidence. Casting and pattern dimensions were accepted in the dimensional database only if the observed casting process capability ratio with respect to the appropriate ISO CT grade, calculated using Equation 2.1, was less than 0.6. In order to eliminate inspection uncertainty due to insufficient sample sizes, thirty castings for each part number were measured. Since the process capability ratio for all the castings is less than 0.6, the statistically required minimum number of samples is eleven for 95% confidence. Therefore, thirty castings, randomly selected from two different batches, were measured for each part number. When possible, the castings were selected from production runs that were more than a month apart Measurement System Errors Gage Repeatability and Reproducibility (gage R&R) analysis was carried out to validate the measurement systems before any measurements were made. Measurement system analysis techniques used were based on procedures outlined by AIAG [31]. A

80 58 measurement system was considered acceptable if the measurement variability was less than 30% of the tolerance designated by the corresponding current ISO CT tolerance grade for the feature being analyzed. All measurement instruments used in this study had also been calibrated by an outside source within one year before use. A sample Gage R&R study is included in Appendix C. Database High Alloy Features Low Alloy Features Features crossing the Parting Line Features NOT crossing the Parting Line Green Sand Molded Nobake Sand Molded Shell Molded Unrestrained Features Partly Restrained Features Fully Restrained Features Figure 4-1: Categories used to classify dimensional data using database SQL queries

81 D Point-to-point Data Classification Scheme Casting and pattern dimensions along with other process and geometry variables were entered in a relational database created in MS Access. SQL queries were generated to classify the data in separate categories according to the statistically significant geometry and process variables - casting alloy, mold parting line, molding method, and feature types, as shown in Figure 4-2. This data was then exported to MS Excel and Minitab for subsequent data analysis. Alloy steels containing alloying elements, in addition to carbon, up to a total alloy content of 8% were classified as low alloy castings [10]. If the casting feature is measured perpendicular to the mold or core parting line, it is classified as a feature crossing the mold parting line. Chemically bonded sand molding is referred to as no-bake molding. Shell molding is also a chemically bonded sand system; however, shell molds are produced by thermally activating the binder through the use of a heated pattern and can be expected to have different pattern allowance values [10]. Casting features were classified into fully restrained (FR), partially restrained (PR), and unrestrained features (UR), following the alphabetical feature type scheme shown in Table 2-9. In case of unrestrained features, the designated molding method for the feature is same as the molding method for the casting. However, in case of fully restrained and partially restrained features, core type is essential in deciding the mold method classification for the feature. For example, in Figure 4-3, feature P is fully restrained by

82 60 green sand mold in a green sand molded casting. Therefore, it is classified as a fully restrained green sand feature. On the other hand, feature T is fully restrained by nobake core in a green sand molded casting. Therefore, although the casting is green sand molded, feature T is classified as a fully restrained nobake feature. Similarly, features W and V are partially restrained nobake features. S P R Q W T V U Green Sand MOLD Casting Green Sand MOLD Casting Green Sand MOLD Casting Nobake Core Casting Green Sand MOLD Figure 4-3: Sample casting for molding method classification of fully and partially restrained features.

83 Table 4-1: Sample query result for green sand low alloy steel castings fully restrained features not crossing the parting line Part Name Feature Name Dimension Type (Refer Figure 2- Pattern Dimension (PD) Casting Dimension (CD) 61 Pattern Allowance 14) PD-CD B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B %

84 62 Table 4-1 shows a sample SQL query result from the relational database. The complete database categorized according to the abovementioned classes is presented in Appendix D D Data Summary Comprehensive data was collected from green sand, no-bake, and shell molding processes from 150 different parts made in 20 different production foundries. Thirty samples of each part number were measured from two different batches made at different times. A total of 15,780 measurements were made for 526 casting features. Casting feature dimensions ranged from 0.25 to 35 and total casting weights ranged from 1.4 lbs to 2506 lbs. Figure 4-12 shows the relationship between measured pattern dimensions and measured casting dimensions for all features in the 2D database, including high alloy, low alloy, across the parting line, and not across the parting line features. A very high (0.9999) R-squared value for the regression equation, Casting Dimension = m X Pattern Dimension, indicates very strong relationship between pattern dimension and casting dimension.

85 63 40 Casting Dimension (Inches) y = x R 2 = Pattern Dimension (Inches) Figure 4-4: Relationship between pattern dimension and casting dimension for all features in the 2D database. Observed pattern allowances for the overall 2D database are shown in a frequency histogram format in Figure 4-5 and in a scatter plot format in Figure 4-6. Although observed pattern allowance values ranged approximately from -10% to 10%, the majority of the casting features had pattern allowances in the range of 1% to 5% as shown in Figure 4-5. Furthermore, as shown in earlier research works [53, 39, 59], variation in pattern allowances widened as the casting feature size decreased as shown in Figure 4-6.

86 64 Frequency % -6% -5% -4% -3% -2% -1% 0% 1% 2% 3% 4% 5% 6% 7% 8% Pattern Allowance (%) Figure 4-5: Frequency distribution of observed pattern allowance for all features in the 2D database.

87 % 8.00% 6.00% 4.00% Pattern Allowance (%) 2.00% 0.00% % -4.00% -6.00% -8.00% % Feature Length (Inches) Figure 4-6: Distribution of pattern allowances with respect to casting feature size for all casting features in the 2D database Features Not Crossing the Parting Line Table 4-2 presents the summary of the 2D classified dimensional data used to build the model for features not crossing the parting line. The number of feature measurements, average, minimum, and maximum casting feature size for every dimension type for features not crossing the parting line are presented in Table 4-2.

88 Table 4-2: Summary of the collected 2D dimensional data for features not crossing the parting line. 66 Condition Number of Features Average Feature Size Minimum Feature Size Maximum Feature Size Low Alloy Steel (Inches) (Inches) (Inches) Green sand molding Un-restrained features Partially restrained features Fully restrained features No bake molding Un-restrained features Partially restrained features Fully restrained features Shell molding Un-restrained features Partially restrained features Fully restrained features High Alloy Steel Shell molding Un-restrained features Partially restrained features Fully restrained features

89 A frequency distribution for low alloy casting features not crossing the parting line is shown in Figure As mentioned in the SFSA and ISO tolerance specifications [11, 14], pattern allowances in green sand molded casting features demonstrate a wider variation than pattern allowances in nobake or shell molded casting features. Figure 4-8 shows the variation in pattern allowance with respect to casting feature size for low alloy casting features not crossing the parting line Green Sand Nobake Shell Frequency % -6% -5% -4% -3% -2% -1% 0% 1% 2% 3% 4% 5% 6% 7% 8% Pattern Allowance (%) Figure 4-7: Frequency distribution of observed pattern allowance for low alloy features not crossing the parting line in the 2D database.

90 % 8.00% Green Sand Nobake Shell 6.00% 4.00% Pattern Allowance (%) 2.00% 0.00% % -4.00% -6.00% -8.00% % Feature Length (Inches) Figure 4-8: Distribution of pattern allowances with respect to casting feature size for low alloy casting features in the 2D database. The frequency distribution for high alloy casting features not crossing the parting line is shown in Figure 4-9. Green and nobake sand molded high alloy casting features in the 2D database were insufficient in numbers to build a robust model. Figure 4-10 shows the variation in pattern allowance with respect to casting feature size for high alloy casting features not crossing the parting line.

91 Green Sand Nobake Shell 20 Frequency % 1% 2% 3% 4% 5% 6% 7% 8% Pattern Allowance (%) Figure 4-9: Frequency distribution of observed pattern allowance for low alloy features not crossing the parting line in the 2D database.

92 % 7.00% Green Sand Nobake Shell 6.00% 5.00% Pattern Allowance (%) 4.00% 3.00% 2.00% 1.00% 0.00% % -2.00% Feature Length (Inches) Figure 4-10: Distribution of pattern allowances with respect to casting feature size for high alloy casting features in the 2D database Features Across the Parting Line Table 4-3 presents the summary of the 2D dimensional data used to build the model for features not crossing the parting line. The number of feature measurements, average, minimum, and maximum casting feature size in a dimension type for features not crossing the parting line are presented in Table 4-3

93 Table 4-3: Summary of the collected 2D dimensional data for features across the parting line. Condition Number of Features Average Feature Size Minimum Feature Size 71 Maximum Feature Size Low Alloy Steel (Inches) (Inches) (Inches) Green sand molding Un-restrained features No bake molding Un-restrained features Partially restrained features Fully restrained features Shell molding Un-restrained features Partially restrained features High Alloy Steel Shell molding Un-restrained features Figure 4-11 shows the variation in pattern allowance with respect to casting feature size for low and high alloy casting features crossing the parting line.

94 % 6.00% Green Nobake Shell Pattern Allowance (%) 4.00% 2.00% 0.00% -2.00% % Feature Size (Inches) Figure 4-11: Variation in pattern allowance with respect to casting feature size in case of low and high alloy steel castings 4.5 Linear Scan Measurements In order to investigate the influence of neighboring geometry on the pattern allowance along the length of the feature, variations in the pattern allowance for an individual feature need to be captured. An example of this pattern allowance variation along the feature length is shown in Figure 4-12 for a feature with neighboring heavy features. Multiple measurements at various locations along the feature length, as shown in Figure 4-12 by position markers 2 inches apart from origin up to 25 inches, need to be collected for investigating the variation in pattern allowance. Using measurement gauges such as vernier calipers for this data collection is prohibitively expensive, laborious, time-

95 73 consuming, and possibly inaccurate. Furthermore, other conventional inspection systems such as CMM, layout machines pose other challenges with regards to portability, casting size and weight, time and cost for this type of data collection. Therefore, articulated coordinate measuring machine along with a dimensional data acquisition computer system was used to generate linear scans of casting and pattern features. The CMM probe installed on an articulated arm generates a point cloud and the computer data acquisition system software creates an approximate linear scan by joining those points [67]. Figure 4-12: Effect of neighborhood geometry on pattern allowance in individual feature. A FARO arm Silver Series articulated CMM as shown in Figure 4-13 was used for this purpose. This S06-01 Silver Series model had a six-foot diameter, spherical measurement envelope. The articulated measuring arm is made from anodized aircraft aluminum, with precision bearings and rotary transducers at each of its six joints [67]. The base of the arm is a mounting plate that permitted direct attachment to a stable base. A point probe or ball probe was attached to the handle on the end of the arm to collect

96 and record measurement data, either one point at a time (point-to-point measurements), or in a continuous stream of points as the probe is moved along a surface [67]. 74 Figure 4-13: FARO arm Silver Series articulated CMM used for generating linear scans of castings and patterns [67]. A predetermined, marked path on the castings, patterns, and coreboxes, as shown in Figure 4-14, was followed by the probe to generate the point cloud. This was done to ensure that the same scan path was followed in case of multiple measurements of the

97 75 same casting and measurements of multiple castings. This predetermined path on the pattern also helped to ensure that the pattern dimensional data was collected from corresponding locations on patterns or coreboxes. Detailed data collection procedures and related information has been previously reported by Khanolkar [68]. Figure 4-14: A predetermined, marked scanning path on the casting to be followed by the CMM probe to generate a dimensional point cloud. Measurement data was then fed to a controller box that converted each recorded probe location to a precise location in 3D space. The CMM controller included various add-in tools such as ambient temperature compensation, and part coordinate system for data integrity [67]. The dedicated AnthroCAM software reads the continuous stream of points in 3D space and generates either a linear scanned wire frame model or a surface model of the part being probed. For this study, wire frame models as shown in Figure 4-

98 76 15 were built for all the measured pattern, castings, and coreboxes using AnthroCAM software. These wire frame models were then imported into AutoCAD, in order to create accurate linear scans of patterns, castings, and coreboxes as shown in Figure The AutoCAD linear scans were then used to extract feature dimensions and conduct dimensional analysis. Figure 4-15: Wire frame model built by AnthroCAM for a sample pattern. Figure 4-16: Linear scan along with local coordinate system generated in AutoCAD from wire frame model generated by AnthroCAM software.

99 77 The dimensional data collected from the linear scan measurements was then imported into the MS Access relational database and classified into a pattern allowance scheme identical to the 2D pattern allowance classification (see 4.3) using SQL queries for further analysis. 4.6 Linear Scanning Data Summary Thirty-five features from eight green sand low alloy steel castings, listed in Table 4-4, along with their pattern and coreboxes were inspected using the Faro Arm articulated CMM. Five sample castings for each part number were measured before and after heat treatment. Every sample casting, pattern, and corebox was scanned three times to generate one average dimensional scan. Linear scans were dimensionally analyzed at various positions on a single feature as shown in Figure In all, 323 data points were generated from 35 casting features for pattern allowance calculations. Each pattern allowance calculation is based on the average of 15 casting feature measurements (5 samples X 3 scans of each sample) and 3 pattern feature measurements. Table 4-5 presents a summary of feature types inspected using linear scan measurements.

100 78 Table 4-4: Summary data of castings used for linear scanning measurement Casting Serial Number Casting Name Casting Part Number Casting Weight (lbs) Number of Features - [APL/NPL] Number of data points 1 Tooth 6NSL3HL [3/2] 31 2 Flight Bar C [2/4] 38 3 Clog CLG [2/1] 40 4 Ripper Tooth RT30SL3KP 68 6 [2/4] 40 5 Pocket Yoke SY45AE [2/3] 80 6 I-Section VS312N 27 5 [2/3] 41 7 Shovel Adapter 10N [0/4] 32 8 Suspension Arm SY [0/1] 21 Table 4-5: Summary of feature types inspected using linear scan measurements Sand Green Sand Nobake Features Across the Parting Line Number of Features Number of Data Points Average Feature Size (Inches) Minimum Feature Size (Inches) Maximum Feature Size (Inches) Yes No Yes No Figure 4-17 shows the relationship between measured pattern and casting dimensions for all features, including green sand, nobake sand molded, across the parting line, and not across the parting line casting features. A very high (0.9991) R-squared value for the regression equation indicates a very strong correlation between pattern dimension and casting dimension measured using linear scan measurements. Figure 4-18

101 shows overall distribution of observed pattern allowances with respect to casting feature size for linear scanning measured castings. 79 Casting Dimension (Inches) y = x R 2 = Pattern Dimension (Inches) Figure 4-17: Relationship between pattern dimension and casting dimension for all features inspected using linear scan measurements.

102 % 6.00% Pattern Allowance (%) 2.00% % -6.00% % Feature Length (Inches) Figure 4-18: Distribution of pattern allowances with respect to casting feature size for all casting features measured using linear scanning measurement Green Sand Molded Low Alloy Steel Casting Linear Scanned Dimensions Figures 4-19 and 4-20 show pattern allowance frequency distribution and pattern allowance with respect to casting feature size respectively, for green sand low alloy steel castings inspected using linear scan measurements. Pattern allowance values for features crossing the parting line are in the range of -5% to 2%. For features not crossing the parting line, the pattern allowance values ranged from 0% to 6%. A higher variation with left-skewed distribution was observed in the pattern allowance values for features across the parting line.

103 Not Across Parting Line Across Parting Line 60 Frequency % -4.00% -3.00% -2.00% -1.00% 0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% Pattern Allowance Figure 4-19: Frequency distribution for observed pattern allowances in features not crossing parting line as well as crossing parting line for green sand low alloy castings inspected using linear scan measurements.

104 % 6.00% Not Across Parting Line Across Parting Line 4.00% Pattern Allowance 2.00% 0.00% % -4.00% -6.00% -8.00% Feature Size (Inches) Figure 4-20: Distribution of pattern allowances with respect to casting feature size in features not crossing parting line as well as crossing parting line for green sand low alloy castings inspected using linear scan measurements Nobake Sand Molded Low Alloy Steel Casting Figures 4-21 and 4-22 show pattern allowance frequency distribution and pattern allowance with respect to casting feature size respectively, for nobake low alloy steel castings inspected using linear scan measurements.

105 83 25 Not Across Parting Line Across Parting Line Frequency % -2.00% -1.00% 0.00% 1.00% 2.00% 3.00% 4.00% Pattern Allowance Figure 4-21: Frequency distribution for observed pattern allowances in features not crossing parting line as well as crossing parting line for nobake low alloy castings inspected using linear scan measurements.

106 % 3.00% Not Across Parting Line Across The Parting Line 2.00% Pattern Allowance 1.00% 0.00% % -2.00% -3.00% -4.00% Feature Size (Inches) Figure 4-22: Distribution of pattern allowances with respect to casting feature size in features not crossing parting line as well as crossing parting line for nobake low alloy castings inspected using linear scan measurements. 4.7 Overview of the collected data Dimensional data collected using both 2D and linear scanning measurement techniques indicated similar trends. Features molded in green sand and across the parting line demonstrate a wider variation in pattern allowances than features not across the parting line and features molded in nobake or shell molds. Table 4-6 shows the average pattern allowance observed along with the range of pattern allowances in the collected dimensional data.

107 Table 4-6: Summary of the collected data Condition Pattern Allowance Across the Parting Line (%) 85 Pattern Allowance Not Across the Parting Line (%) Low Alloy Steel Average Min - Max Average Min - Max Green sand molding Un-restrained features Partially restrained features Fully restrained features No bake molding Un-restrained features Partially restrained features Fully restrained features Shell molding Un-restrained features Partially restrained features Fully restrained features High Alloy Steel Green sand molding Fully restrained features Shell molding Un-restrained features Partially restrained features Fully restrained features

108 86 Chapter 5 Proposed Pattern Allowance Model As discussed in Section 3.3, pattern allowance data, which is not normally distributed, and is confounding with casting feature size, cannot be used to build an empirical model for the prediction of pattern allowances. However, transforming the pattern allowance data into a variable which is normally distributed as well as nonconfounding with casting feature size permits a robust model to be created. The variable called shrinkage error is proposed for this purpose. Shrinkage error is defined as follows: Shrinkage error = Actual Shrinkage Ideal Shrinkage Actual Shrinkage = P C Ideal Shrinkage = P I - C = (1 + γ) C C where P is the actual pattern dimension, C is the casting dimension, P I is the ideal pattern dimension assuming a standard shrinkage allowance, and γ is the standard shrinkage allowance for a particular casting feature depending on casting process and geometry variables. Therefore, Shrinkage Error = P C - γ C 5.1

109 87 It is proposed that the standard shrinkage allowance γ is constant for a particular feature type determined by mold restraint, mold type, parting line, and casting alloy. Therefore, the value of the standard shrinkage allowance γ is dependant on Mold type and mold restraint Casting alloy Parting line The procedure for the determination of standard shrink value γ for a particular feature type is discussed in detail in Section Error (Inches) Feature Length (Inches) Figure 5-1: Overall distribution of shrinkage error for all casting features in the 2D database. Shrinkage error is calculated using Equation 5.1, the standard shrinkage allowance γ is assumed to be (1/4 in per foot).

110 88 Figure 5-1 shows the shrinkage error for all features in the 2D database for low alloy and high alloy castings. Shrinkage error is calculated using Equation 5.1, where the standard shrinkage allowance γ is assumed to be (2.08% or 1/4 in/ft) for all casting feature types. Unlike pattern allowance values, shrinkage error values are not confounding with casting feature size. Figures 5-2 and 5-3 respectively show the normal probability plot for shrinkage error and observed pattern allowance for unrestrained features in low alloy green sand steel castings. Unlike the pattern allowance data (see Figure 5-3); shrinkage error data is normally distributed with 95% confidence, as indicated by the normal probability plot as well as the P-value of for Anderson-Darling test [63] (see Figure 5-2). Therefore, the pattern allowance data transformed into shrinkage error can be effectively used to develop an empirical model. Since the goal for a new pattern allowance model is to accurately represent both central tendency and the variation in pattern allowances, 10 th and 90 th percentiles were chosen to represent the variation in pattern allowances instead of the 80% confidence interval scheme used in earlier models [39], see Figure 5-4.

111 Figure 5-2: Shrinkage Error Normal Probability plot for unrestrained features in green sand low alloy steel castings from the 2D database. 89

112 Figure 5-3: Observed pattern allowance Normal Probability plot for unrestrained features in green sand low alloy steel castings from the 2D database. 90

113 91 Figure 5-4: Distribution of shrinkage error for green sand low alloy steel castings - unrestrained features along casting feature length. Both the 80% confidence interval and the 10 th and 90 th percentiles for shrinkage error are also indicated. 5.1 Constitutive Equations for the Pattern Allowance Model Pattern allowance can be derived from shrinkage error using the following equations. From Equation 5.1, Shrinkage Error = P C - γ C Therefore, P C = Shrinkage Error + γ C 5.2

114 Pattern allowance according to Equation 1.1 is, 92 Pattern Allowance= P - C C Substituting expression for P C from Equation 5.2, pattern allowance can be expressed in terms of shrinkage error as Shrinkage Error +γ C Pattern Allowance= 5.3 C In order to represent the wide variation observed in pattern allowances; average, minimum, and maximum pattern allowance values for a particular casting feature length can be calculated by substituting median, 10 th percentile, and 90 th percentile of the shrinkage error in Equation 5.3. Therefore, Median Shrinkage Error +γ C Pattern Allowance50th = C Shrinkage ErrorMin +γ C Pattern Allowance10th = C Shrinkage ErrorMax +γ C Pattern Allowance90th = C 5.4 where Shrinkage Error Min and Shrinkage Error Max are the 10 th and 90 th percentile shrinkage error values, respectively. Equation 5.4 illustrates the pattern allowance prediction model using casting feature length, shrinkage error and standard shrinkage allowance for a casting feature type. For a particular type of casting feature, standard shrinkage allowance, the minimum and maximum values of shrinkage error are constant and are derived from the 2D database. Since the casting feature length appears in the denominator in Equation 5.4, the pattern allowance range will increase as casting feature

115 size reduces. This prediction model behavior closely approximates the wider distribution of pattern allowance values as the casting feature length decreases Standard Shrinkage Allowance γ Precise determination of the standard shrinkage allowance value γ is very important for the success of the empirical model shown in Equation 5.4. All data points used in this section are from the 2D database. As discussed earlier, (1/4 in per foot) is the most commonly used industry standard shrinkage allowance for all low alloy steel casting features, irrespective of feature type or molding method. However, it has been shown by various researchers [34, 53, 58] that the pattern allowance is strongly influenced by casting feature type. Therefore, it would be desirable to use feature-specific standard shrinkage allowances. Figure 5-5 shows the relationship between the summation of error and shrinkage allowance γ for low alloy nobake castings with fully restrained, partially restrained, and unrestrained features. When shrinkage allowance is 0.03, summation error for all feature types is negative. As the shrinkage allowance is increased, the summation error increases, becomes zero and keeps increasing in the positive direction. Figure 4-19 shows the relationship between summation of square of the error and shrinkage allowance γ for low alloy nobake castings with fully restrained, partially restrained, and unrestrained features.

116 94 Summation Of Error (Inches) LowAlloyNobakeFR LowAlloyNobakeUR LowAlloyNobakePR Shrink Rate Figure 5-5: Change in summation of error with respect to shrinkage allowance γ for low alloy nobake casting fully restrained, partially restrained, and unrestrained features. Summation of Square of Error LowAlloyNobakeFR LowAlloyNobakeUR LowAlloyNobakePR Shrink Rate Figure 5-6: Change in summation of square of the error with respect to shrinkage allowance γ for low alloy nobake casting fully restrained, partially restrained, and unrestrained features.

117 95 As shown in Figures 5-5 and 4-19 summation of square of the error and summation of error follow interrelated trends when the value of shrinkage allowance is changed. The summation of the square of the error is minimum when the value of shrinkage allowance results in zero summation of error. However, the summation of error initially has a negative value and crosses zero line to become positive as the value of shrinkage allowance is changed. This gives the summation of error a distinct computational advantage over the summation of the square of error for the optimization. Therefore, it is proposed that the standard shrinkage allowance is calculated by minimizing the summation of error for a particular casting feature type. Minimize n 1 Pj- Cj- γc j 5.5 Equation 5.5 shows the minimization equation where n is the number of features of a particular feature type in the database, P and C are pattern and casting dimensions respectively, and γ is the shrinkage allowance for that particular casting feature type. Summation of error is minimized by changing the value of shrinkage allowance γ. The value of shrinkage allowance yielding minimum value of summation of error for a casting feature type is called the standard shrinkage allowance for that feature type. Microsoft Excel solver was used for the minimization of error sum by changing the value of shrinkage allowance. Table 5-1 shows the effect of minimization of error sum on shrinkage allowance for unrestrained low alloy steel casting features molded in green sand. The number of such features in the database is 86, therefore, for Equation 5.5, n = 86. Initially, the value of shrinkage allowance is assumed to be and the summation of error is calculated to be inches. After optimization, the standard

118 96 shrinkage allowance value γ is calculated to be for green sand molded low alloy steel casting unrestrained features and the summation of error is approximately zero (see Table 5-1). Table 5-1: Change in the value of standard shrinkage allowance γ due to minimization of summation of error for green sand molded low alloy steel castings - unrestrained features. Standard Shrinkage allowance Summation of Error (in) Before Minimization After Minimization E-16 Figure 5-7 shows the observed pattern allowance along with the predicted minimum, average, and maximum pattern allowances for green sand molded low alloy steel casting unrestrained features. Predicted pattern allowance, shown in Figure 5-7, is predicted using the standard shrinkage allowance γ = % Pattern Allowance (%) 4.00% 0.00% -4.00% -8.00% Observed PA Predicted PA 10th Predicted PA 90th Avg PA Feature Length (Inches) Figure 5-7: Observed pattern allowance for green sand molded low alloy steel castings - unrestrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

119 97 Table 5-2 shows standard shrinkage allowances and change in the summation of error for unrestrained, partially restrained, and fully restrained features not crossing the parting line for low and high alloy steel castings molded in green sand, nobake, and shell sand. Standard shrinkage allowance of is assumed to calculate the sum of error before optimization. Except for the green sand low alloy fully restrained features, the standard shrinkage allowance increased as the amount of restraint decreased. This indicates that the unrestrained features shrink more than the partially restrained features and partially restrained features shrink more than the fully restrained features. This conclusion is similar to previous findings and is rationalized in Section 2.5.4, validating the proposed optimization approach. Table 5-2: Standard shrinkage allowance and changes in sum of error due to minimization for features not across the parting line in low and high alloy steel castings. Feature Type Readings Before Optimization After Optimization Standard Standard Shrinkage allowance Summation of Error (in) Shrinkage allowance Summation of Error (in) LowAlloyGreenSandFR E-16 LowAlloyGreenSandPR LowAlloyGreenSandUR E-16 LowAlloyNoBakeFR E-15 LowAlloyNoBakePR E-16 LowAlloyNoBakeUR E-17 LowAlloyShellFR LowAlloyShellPR E-17 LowAlloyShellUR E-17 HighAlloyShellFR HighAlloyShellPR E-17 HighAlloyShellUR E-17 UR = Unrestrained, FR = Fully restrained, PR = Partially Restrained

120 98 Limited numbers of features crossing the parting line are measured using the 2D measurement scheme and are shown in Table 5-3. Standard shrinkage allowance of is assumed to calculate the sum of error before optimization. Similar to features not crossing the parting line, standard shrinkage allowance increased as the mold restraint reduced. Table 5-3: Standard shrinkage allowance and changes in sum of error due to minimization for features across the parting line in low and high alloy steel castings from the 2D database. Feature Type Readings Before Optimization After Optimization Standard Standard Shrinkage allowance Summation of Error (in) Shrinkage allowance Summation of Error (in) LowAlloyGreenSandURPL E-17 LowAlloyNoBakeFRPL E-17 LowAlloyNoBakePRPL E-17 LowAlloyNoBakeURPL E-17 LowAlloyShellPRPL LowAlloyShellURPL HighAlloyShellURPL E-18 UR = Unrestrained, FR = Fully restrained, PR = Partially Restrained, PL = Crossing parting line 5.3 Implementation of New Pattern Allowance Model Median, 10 th percentile, and 90 th percentile error values and standard shrinkage allowances for each feature type from the 2D database are presented in this section. These values can be substituted in Equation 5.4 to predict the range of pattern allowance values for given feature type and casting feature size. Appendix E presents complete the 2D database and individual error values used to build this model.

121 Low Alloy Casting Features Not Crossing the Parting Line Table 5-4: Median, Error min (10 th percentile), and Error max (90 th percentile)shrinkage error along with optimized standard shrinkage allowance and observed average pattern allowance values for low alloy casting features not crossing the parting line. Error min Standard Shrinkage allowance Average Pattern Allowance Feature Type Median Error (Inches) (Inches) Error max (Inches) LowAlloyGreenSandFR % LowAlloyGreenSandPR % LowAlloyGreenSandUR % LowAlloyNoBakeFR % LowAlloyNoBakePR % LowAlloyNoBakeUR % LowAlloyShellFR % LowAlloyShellPR % LowAlloyShellUR % Table 5-4 summarizes the values of median, 10 th percentile, and 90 th percentile shrink error values along with standard shrinkage allowance γ and average pattern allowance for low alloy steel casting features not crossing the parting line made in shell, nobake, and green sand molds. Due to the minimization of the summation of shrinkage error, the median error values are very close to zero. Fully restrained features demonstrated a larger variation in the shrink error than partially or unrestrained features. This can be attributed to the complex mold-metal interaction in fully restrained features due to the restraint offered to the contracting metal by the expanding mold as shown in Figure Figures 5-8 to 5-12 show the shrinkage error, and predicted and observed pattern allowances for low alloy casting features made in green sand molds. Figures 5-8 and 5-9 show the shrinkage error, and predicted and observed pattern allowances respectively for

122 fully restrained features, Figures 5-10 and 5-11 show those for the partially restrained features, and Figures 5-12 and 5-7 show those for the unrestrained features Error (Inches) Feature Length (Inches) Figure 5-8: Distribution of shrinkage error along the feature size for low alloy fully restrained features not crossing the parting line in green sand molds. Pattern Allowance (%) 6.00% 4.00% 2.00% 0.00% -2.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA % Feature Length (Inches) Figure 5-9: Observed pattern allowance for green sand molded low alloy steel castings - fully restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

123 Error (Inches) Feture Length (Inches) Figure 5-10: Distribution of shrinkage error along the feature size for low alloy partially restrained features not crossing the parting line in green sand molds. 5.00% Pattern Allowance (%) 4.00% 3.00% 2.00% 1.00% 0.00% -1.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA % Feature Length (Inches) Figure 5-11: Observed pattern allowance for green sand molded low alloy steel castings - partially restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

124 Error (Inches) Feature Length (Inches) Figure 5-12: Distribution of shrinkage error along the feature size for low alloy unrestrained features not crossing the parting line in green sand molds. Figures 5-13 to 5-18 show the shrinkage error, and predicted and observed pattern allowances for low alloy features in nobake sand molds. Figures 5-13 and 5-14 show the fully restrained features, Figures 5-15 and 5-16 show the partially restrained features, and Figures 5-17 and 5-18 show the shrinkage error, and predicted and observed pattern allowances for unrestrained features. Similarly, Figures 5-19 to 5-24 show the shrinkage error, and predicted and observed pattern allowances for low alloy features in shell molds. Figures 5-19 and 5-20 show the fully restrained features, Figures 5-21 and 5-22 show the partially restrained features, and Figures 5-23 and 5-24 show the shrinkage error, and predicted and observed pattern allowances for unrestrained features.

125 103 Error (Inches) Feature Length (Inches) Figure 5-13: Distribution of shrinkage error along the feature size for low alloy fully restrained features not crossing the parting line in nobake sand molds. 8.00% Pattern Allowance (%) 4.00% 0.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA % Feature Length (Inches) Figure 5-14: Observed pattern allowance for nobake sand molded low alloy steel castings - fully restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

126 104 Error (Inches) Feature Length (Inches) Figure 5-15: Distribution of shrinkage error along the feature size for low alloy partially restrained features not crossing the parting line in nobake sand molds. 3.50% 3.00% Pattern Allowance (%) 2.50% 2.00% 1.50% 1.00% 0.50% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA 0.00% Feature Length (Inches) Figure 5-16: Observed pattern allowance for nobake sand molded low alloy steel castings - partially restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

127 105 Error (Inches) Feature Length (Inches) Figure 5-17: Distribution of shrinkage error along the feature size for low alloy unrestrained features not crossing the parting line in nobake sand molds. 8.00% Pattern Allowance (%) 6.00% 4.00% 2.00% 0.00% O bserved P A P redicted P A 10th P redicted P A 90th A vg P redicted P A % -4.00% F e atu re Length (In ch es ) Figure 5-18: Observed pattern allowance for nobake sand molded low alloy steel castings - unrestrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

128 106 Error (Inches) Feature Length (Inches) Figure 5-19: Distribution of shrinkage error along the feature size for low alloy fully restrained features not crossing the parting line in shell molds. Pattern Allowance (%) 5.00% 4.00% 3.00% 2.00% 1.00% 0.00% -1.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA % Feature Length (Inches) Figure 5-20: Observed pattern allowance for shell molded low alloy steel castings - fully restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

129 Error (Inches) Feature Length (Inches) Figure 5-21: Distribution of shrinkage error along the feature size for low alloy partially restrained features not crossing the parting line in shell molds. 4.00% Pattern Allowance (%) 3.00% 2.00% 1.00% 0.00% Feature Length (Inches) Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA Figure 5-22: Observed pattern allowance for shell molded low alloy steel castings - partially restrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

130 108 Error (Inches) Feature Length (Inches) Figure 5-23: Distribution of shrinkage error along the feature size for low alloy unrestrained features not crossing the parting line in shell molds. Pattern Allowance (%) 10.00% 8.00% 6.00% 4.00% 2.00% 0.00% -2.00% -4.00% -6.00% -8.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA Feature Length (Inches) Figure 5-24: Observed pattern allowance for shell molded low alloy steel castings - unrestrained features not crossing the parting line along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

131 High Alloy Casting Features Not Crossing the Parting Line Table 5-5 summarizes median, minimum, and maximum shrink error along with standard shrinkage allowance γ and average pattern allowance for high alloy steel casting features not crossing the parting line made in shell molds. Due to the minimization of the summation of shrinkage error, the median error values are very close to zero. Fully restrained features demonstrated a larger variation in the shrinkage error than partially or unrestrained features. Table 5-5: Median, minimum, and maximum error along with optimized standard shrinkage allowance and observed average pattern allowance values for high alloy casting features not crossing the parting line. Median Error (Inches) Minimum Error (Inches) Maximum Error (Inches) Standard Shrinkage allowance Average Pattern Allowance Feature Type HighAlloyShellFR % HighAlloyShellPR % HighAlloyShellUR % Figures 5-25 to 5-30 show the shrinkage error, and predicted and observed pattern allowances for low alloy features in green sand molds. Figures 5-25 and 5-26 show the shrinkage error, and predicted and observed pattern allowances respectively for fully restrained features, Figures 5-27 and 5-28 show those for the partially restrained features, and Figures 5-29 and 5-30 show the shrinkage error, and predicted and observed pattern allowances respectively for unrestrained features.

132 110 Error (Inches) Feature Length (Inches) Figure 5-25: Distribution of shrinkage error along the feature size for high alloy fully restrained features not crossing the parting line in green sand molds. 8.00% Pattern Allowance (%) 6.00% 4.00% 2.00% 0.00% -2.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA % -6.00% Feature Length (Inches) Figure 5-26: Observed pattern allowance for high alloy steel castings - fully restrained features not crossing the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

133 Error (Inches) Feature Length (Inches) Figure 5-27: Distribution of shrinkage error along the feature size for high alloy partially restrained features not crossing the parting line in green sand molds. 3.50% 3.00% Pattern Allowance (%) 2.50% 2.00% 1.50% 1.00% 0.50% O bserved PA Predicted PA 10th Predicted PA 90th Avg Predicted PA 0.00% Feature Length (Inches) Figure 5-28: Observed pattern allowance for high alloy steel castings - partially restrained features not crossing the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

134 112 Error (Inches) Feature Length (Inches) Figure 5-29: Distribution of shrinkage error along the feature size for high alloy unrestrained features not crossing the parting line in green sand molds. Pattern Allowance (%) 6.00% 5.00% 4.00% 3.00% 2.00% 1.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA 0.00% Feature Length (Inches) Figure 5-30: Observed pattern allowance for high alloy steel castings - unrestrained features not crossing the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

135 Low and High Alloy Casting Features Crossing the Parting Line Table 5-6 summarizes median, minimum, and maximum shrink error along with standard shrinkage allowance γ and average pattern allowance for low and high alloy steel casting features crossing the parting line. Due to the minimization of the summation of shrinkage error, the median error values are very close to zero. In general, fully restrained features demonstrated a larger variation in the shrink error than partially or unrestrained features. Table 5-6: Median, minimum, and maximum error along with optimized standard shrinkage allowance and observed average pattern allowance values for low and high alloy casting features crossing the parting line from the 2D database. Feature Type Median Error Minimum Error Maximum Error Standard Shrinkage allowance Average Pattern Allowance LowAlloyGreenSandURPL % LowAlloyNoBakeFRPL % LowAlloyNoBakePRPL % LowAlloyNoBakeURPL % LowAlloyShellPRPL % LowAlloyShellURPL % HighAlloyShellURPL % Figures 5-31 and 5-32 show the shrinkage error, and predicted and observed pattern allowances for low alloy unrestrained features across the parting line in green sand molds. Figure 5-33, 5-34 and Figures 5-35, 5-36 and Figures 5-37, 5-38 show the shrinkage error, and predicted and observed pattern allowances for low alloy features across the parting line in nobake molds that are fully restrained, partially restrained, and unrestrained respectively.

136 114 Error (Inches) Feature Length (Inches) Figure 5-31: Distribution of shrinkage error along the feature size for low alloy unrestrained features crossing the parting line in green sand molds. Pattern Allowance (%) % % % % % O b s e rv e d P A P re d ic te d P A 1 0 th P re d ic te d P A 9 0 th A v g P re d ic te d P A % F e a tu re L e n g th (In c h e s ) Figure 5-32: Observed pattern allowance for low alloy steel castings - unrestrained features across the parting line in green sand molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

137 115 Error (Inches) Feature Length (Inches) Figure 5-33: Distribution of shrinkage error along the feature size for low alloy fully restrained features crossing the parting line in nobake molds. 8.00% Pattern Allowance (%) 6.00% 4.00% 2.00% 0.00% O bserved PA Predicted P A 10th Predicted P A 90th Avg P redicted PA % -4.00% Feature Length (Inches) Figure 5-34: Observed pattern allowance for low alloy steel castings - fully restrained features across the parting line in nobake molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

138 Error (Inches) F eatu re L en g th (Inches ) Figure 5-35: Distribution of shrinkage error along the feature size for low alloy partially restrained features crossing the parting line in nobake molds. 3.50% 3.00% O bserved PA Predicted PA 10th Predicted PA 90th Avg Predicted PA Pattern Allowance (% 2.50% 2.00% 1.50% 1.00% Feature Leng th (Inches) Figure 5-36: Observed pattern allowance for low alloy steel castings - partially restrained features across the parting line in nobake molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

139 Error (Inches) Feature Length (Inches) Figure 5-37: Distribution of shrinkage error along the feature size for low alloy unrestrained features crossing the parting line in nobake molds.

140 % 4.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA Pattern Allowance (%) 3.00% 2.00% 1.00% 0.00% Feature Length (Inches) Figure 5-38: Observed pattern allowance for low alloy steel castings - unrestrained features across the parting line in nobake molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves. Figures 5-39, 5-40 and Figures 5-41, 5-42 show the shrinkage error, and predicted and observed pattern allowances for low alloy partially restrained and unrestrained features respectively across the parting line in shell molds. Figures 5-43 and 5-44 show the shrinkage error, and predicted and observed pattern allowances for high alloy unrestrained features crossing the parting line in shell molds.

141 Error (Inches) Feature Length (Inches) Figure 5-39: Distribution of shrinkage error along the feature size for low alloy partially restrained features crossing the parting line in shell molds. 4.00% Pattern Allowance (%) 3.00% 2.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA 1.00% Feature Length (Inches) Figure 5-40: Observed pattern allowance for low alloy steel castings - partially restrained features across the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

142 Error (Inches) Feature Length (Inches) Figure 5-41: Distribution of shrinkage error along the feature size for low alloy unrestrained features crossing the parting line in shell molds. 8.00% Pattern Allowance (%) 6.00% 4.00% 2.00% Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA 0.00% Feature Length (Inches) Figure 5-42: Observed pattern allowance for low alloy steel castings - unrestrained features across the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

143 121 Error (Inches) Feature Length (Inches) Figure 5-43: Distribution of shrinkage error along the feature size for high alloy unrestrained features crossing the parting line in shell molds.

144 122 Pattern Allowance (%) 4.50% 4.00% 3.50% 3.00% 2.50% 2.00% 1.50% 1.00% 0.50% 0.00% Feature Length (Inches) Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA Figure 5-44: Observed pattern allowance for high alloy steel castings - unrestrained features across the parting line in shell molds along with predicted 10th percentile, 90th percentile, and average pattern allowance curves. 5.4 Verification of the New Pattern Allowance Model In the previous section, observed pattern allowance values in production steel castings are shown to be within the predicted minimum and maximum range of pattern allowance. However, this good agreement is only demonstrated between the model and the data used to build this model. Further verification of this model was conducted by comparing observed pattern allowance data not used to build this model (the linear scanning pattern allowance data). Linear scanned data is termed as 2-1/2D data as opposed to the 2D data and results are presented in this section. Figures 5-45 and 5-46 present a comparison between the predicted and observed pattern allowance values. Pattern allowance values are predicted by the model built using

145 123 the 2D database. Observed pattern allowance values are derived from linear scanning measurements of patterns and castings. For low alloy casting features made in green sand and nobake molds, all of the observed pattern allowance values are within the predicted pattern allowance range. Good agreement between the predicted and observed pattern allowance values is demonstrated. This good model agreement along with agreement demonstrated in section 5.3 validate the proposed pattern allowance model. Pattern Allowance 4.00% 3.00% 2.00% 1.00% 2-1/2D Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA 0.00% Feature Length (Inches) Figure 5-45: Observed pattern allowance data from linear scanning measurement for low alloy green sand partially restrained features along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

146 % 5.00% 4.00% 2-1/2D Observed PA Predicted PA 10th Predicted PA 90th Avg Predicted PA Pattern Allowance 3.00% 2.00% 1.00% 0.00% -1.00% % Feature Length (Inches) Figure 5-46: Observed pattern allowance data from linear scanning measurement for low alloy green sand unrestrained features along with predicted 10th percentile, 90th percentile, and average pattern allowance curves.

147 125 Chapter 6 Pattern Allowance Advisor Software As mentioned in Section 3.6, the newly developed pattern allowance model should be easily commissionable in a production foundry setting. In order to take full advantage of various capabilities of the new model and to introduce customization capability, a software tool is designed. This software uses a Visual Basic front-end and new statistical pattern allowance prediction model as the backend. In addition to pattern dimension prediction, this software is designed to serve as a comprehensive dimensional management tool. This chapter discusses the software architecture, flowcharts, implementation, and validation in detail. The following goals were set for the new software tool. This software should effectively advise the pattern designer on appropriate pattern allowances based on the new statistical model. The pattern designer should not be required to have statistical expertise or make complex calculations in order to predict pattern dimensions. In order to accommodate for the inherent variability in the pattern allowance, the pattern designer should be provided with a range of pattern dimensions instead of a single prediction. The user should be provided with effective assistance when choosing the final pattern dimension in that range. The pattern dimension estimation process should be very transparent and the designer should be free to choose any pattern dimension even outside the predicted range.

148 126 A simple user interface should be provided for the inspector to enter casting and pattern dimensions in order to build a dimensional database. The software should provide comprehensive casting dimensional data management system so that the past dimensional data can be effectively used to learn and to better predict pattern dimensions. The software should provide simple tools to customize the pattern dimension prediction for a particular foundry while not changing the default model. This is essential to ensure that bad dimensional data will not destroy the pattern dimension prediction capability.

149 127 Calibration Module New part entry module DATABASE Modification module History module Inspection module USER INTERFACE Help module Reports module Reports Administrator module Figure 6-1: Pattern Allowance Advisor software architecture. Microsoft Visual Basic 6 was used as the front end of the software and MS Access database is used as the backend. Figure 6-1 shows a schematic representation of the Pattern Allowance Advisor software architecture. Twenty-three Visual Basic forms and 4 class modules are used to generate an executable file for Microsoft Windows operating systems. The user accesses eight different modules through a user interface. The software generates various reports for inspectors and presents statistically-based pattern dimension estimates to advise the pattern designer based on a default database and a custom database.

150 Database The Microsoft Access database provides PSU-built default and user-built custom lookup tables for pattern dimension prediction and stores dimensional inspection data for castings and patterns. Different queries can be performed to effectively access and use the dimensional data. Table 6-1 shows a complete list of table names along with a brief description for each table contained in the Pattern Allowance Advisor software databse. A complete list of database tables, field names, data types, and their functions is provided in Appendix F.

151 Table 6-1: List of MS Access database tables included in the Pattern Allowance Advisor. 129 No. Table Name Description 1 AlloyType List of available alloy types 2 Calibration Stores query results for a feature type during the calibration 3 Custom Stores pattern allowance values for the Custom model 4 Default Stores pattern allowance values for the Default Penn State Model 5 dimarchive Stores archived dimensional data 6 dimensions Stores current dimensional data for parts in the database 7 inspected Stores dimensional data for the inspected castings 8 partinfo Stores information about current parts in the database 9 patallowhist Stores query results for pattern allowance history of a feature type 10 patinspected Stores dimensional data for the inspected patterns 11 processtype List of available molding methods 12 rptinsp Stores the results of query for report generation 13 userinfo List of current users along with their types 6.2 User Types In order to maintain the security of the dimensional data, the Pattern Allowance Advisor software requires the users to login before they can use the software. Four different user types are assigned in Pattern Allowance Advisor by the administrator with each user type having access privileges only to certain modules. Four user types are: 1. Administrator accesses all the Pattern Allowance Advisor modules

152 Pattern Designer accesses new part, modification, history, help, and reports modules 3. Inspector access limited to inspection, reports, help, and history module 4. General User accesses all modules except for the administrator and calibration module 6.3 Modules The Pattern Allowance Advisor software is organized into eight basic modules. Depending on the user access privileges, some of those modules may not be accessible to some users. Table 6-2 lists the Pattern Allowance Advisor modules along with a brief description and user access permissions to the modules. This section provides a detailed description along with flowcharts for each module. A complete user manual for the Pattern Allowance Advisor is included in Appendix G which shows the details of all modules along with selected screenshots.

153 Table 6-2: Pattern Allowance Advisor modules along with user types having access privileges. No. Module Description User Type 1 Administrator To add, delete or modify users or parts A 2 Calibration To calibrate the custom model A 3 New Part To add a new part in the database A, G, P 4 Modification To modify a current part A, G, P 5 Inspection To enter casting/pattern inspection data A, G, I 6 Report To view reports All 7 Help To provide help All History To view dimensional data and pattern allowance history All A = Administrator, G = General User, P = Pattern Designer, I = Inspector Administrator Module The Pattern Allowance Advisor administrator uses the password protected administrator module to add new users, modify current user settings, delete a user, edit part details, and delete parts from the database. Using add user task, the administrator enters pertinent user information and specifies the user type as shown schematically in Figure 6-2.

154 132 Administrator entry Administrator enters user details Administrator clicks save option Database Table: userinfo Figure 6-2: New user entry by administrator. When editing the user information, the administrator selects a user name and the user information is shown. The administrator can make any necessary changes and save them. To delete a user, the administrator selects a user name from the list and clicks the delete option. These user information manipulation functions are shown schematically in Figure 6-3. When editing part details, the administrator selects a part name from the list, and the details of the selected part are displayed. After making the required modifications, changes can be saved using the save option. To delete a part from the software, the administrator selects a part name from the list and clicks delete. This task must be used cautiously; because the part records will be deleted from the database along with all other archived dimensions for the part name. Part modification / deletion operations are shown schematically in Figure 6-3.

155 133 Administrator selection User / part name Show selected user / part details Administrator makes necessary changes Database Table: userinfo / partinfo Figure 6-3: User / part deletion / modification by administrator Calibration Module Calibration module is accessible only to the administrator and can be used to customize the embedded pattern dimension prediction model for a particular foundry. This module can only be used after sufficient pattern and casting feature dimensional data is entered in the database by the inspection department. In the calibration module, the user selects the type of feature for which the pattern allowance model is to be calibrated. The software then checks if there is sufficient dimensional data available in the database. If sufficient data is available in the database, the data integrity is first checked using process capability ratio. If the data is within specifications, then an optimization algorithm based on the golden interval [72] is used to minimize the summation of error. (This optimization algorithm is discussed in detail in section 6.5). New standard shrinkage allowance along with minimum summation of error values,

156 median, 10 th, and 90 th percentile shrinkage errors are displayed to the user. If user accepts this new calibration, the custom database is updated with new pattern allowance model New Part Module The new part entry module is used to generate new parts. New part generation is done in two stages. In the first stage; the part name, alloy type, molding method, and unit system are specified. At the completion of the first stage, entries made by user are stored in the database partinfo table. In the second stage, individual casting features are specified with respect to size and restraint type using the dimension entry window. (The dimension entry window is discussed in detail in section 6.4). At the completion of the second stage, the user hits the Save button and the entries are stored in dimensions table of the database as shown in Figure 6-4 User Entry: Part name Alloy type Molding method Units Database Tables: partinfo dimensions User Entry: dimensional details for each feature Figure 6-4: New part entry module.

157 Modification Module The modification module is used to modify the dimensional details of existing parts. In this module, the user selects the part name and the existing dimensional details are presented through an SQL query result. The user can then make any necessary modifications to the selected part features and save them. When the user saves the modifications, the existing dimensions are archived and modified dimensions are stored as the current dimensions. A revision number is generated to assist in dimensional tracking as shown in Figure 6-5. User selection Part name SQL Query: Current dimensional details displayed User Entry: modifications save button Database Tables: dimensions dimarchive Current dimensional detail are archived Modifications made current Figure 6-5: Modification module Inspection Module The inspection module is used to enter measured casting and pattern feature dimensions and for generating conformance reports. Measured casting or pattern dimensions are entered by the inspector by selecting the appropriate part name and revision dates. SQL queries are generated and dimensional data corresponding to the

158 136 selection are then displayed to the user. The user then enters the measured dimension to check for conformance and measured dimensions are saved to the database as shown in Figure 6-6. User Selection Part Name Revision Date SQL Query Dimensional data and inspection history displayed Database Table: Inspected User Entry Measured dimensions are entered, conformance checked Figure 6-6: Inspection module Reports Module The reports module is used to view or print various reports. These reports are: 1. Current recommended pattern dimensions 2. Tooling history indicating actual pattern dimensions if measured 3. Conformance reports for measured castings 4. Conformance reports for measured patterns 5. Combined current measured pattern dimensions and measured casting dimensions. 6. Pattern allowance history for a particular feature type Microsoft Word template files have been created for all reports. When the user selects a report type, the template for that report is opened and saved as a temporary file.

159 137 Then the required data is accessed from the database through an SQL query and inserted into the temporary Word file. After inserting all the data into the temporary word file, it is presented to the user. The user can save or print the report using Microsoft Word menus. This report generation procedure is shown schematically in Figure 6-7. Reports module operation instructions are presented in detail in the user manual in Appendix G. User selection Report type Part name Revision date Open word template file SQL Query Database Save template file as temporary word file Insert data into temporary word file Show temporary word file to user Figure 6-7: Report generation scheme Help Module The help module is designed to provide detailed operating instructions and explanations for both new and experienced Pattern Allowance Advisor users. Users can easily browse and select the help files included with the software. The selected help items will be shown to the user within the help module window. The help module uses a simple

160 138 Visual Basic user form. The user can use back and forward buttons to browse through the help file list or the can simply click on the list and browse. All help files are in HTML format. Once the user clicks on a particular help topic, the help file containing selected the topic is shown in the web browser within the help module window. Additionally, simple tutorials are included in the help module for new users. Furthermore, a user manual in PDF format is also available through menu shortcuts, which is also attached in Appendix G History Module The history module is used to review two types of history: pattern tooling history and pattern allowance history. Any changes made in pattern tooling dimensions during dimensional reengineering cycles are stored in the Pattern Allowance Advisor. The pattern tooling history module shows a trail of dimensional changes made in a pattern tool along with dates and optional comments explaining the reasons for modifications. In this module, current and archived dimensional details are displayed for parts using an SQL query as shown in Figure 6-8. The pattern allowance history module could be used by pattern designer for making decisions regarding new pattern dimensions. User selects the alloy type, molding method, and restraint type and pattern allowance history module then displays the actual historic pattern allowance data for that feature type from the database. This data then can be used by the pattern designer to better predict pattern dimensions for current part features.

161 139 User selection Part name or Feature type SQL Query Database Tables: dimensions and archive dimension Pattern allowance or pattern dimension history is shown Figure 6-8: Pattern tooling history and pattern allowance history module. 6.4 Dimension Entry Window For a pattern designer using the Pattern Allowance Advisor for pattern dimension prediction, the dimension entry window is the most important user entry form. This window has been continually improved since it s original development based on user friendliness, feedback from production foundry users, transparency in pattern dimension estimation process, and greater flexibility with respect to pattern dimension estimation to the user. Since this form is a key user interface component, it is discussed in detail in this section.

162 140 Figure 6-9: Dimension entry window screenshot. The dimension Entry Form is accessed while defining a new part or modifying an existing part. This screen is divided in 3 key areas indicated by the numbered circles in Figure 6-9: 1. Pattern dimension guide picture area 2. Feature type selection picture area 3. Casting and pattern dimension entry area

163 Casting and Pattern Dimension Entry Area Figure 6-10 shows a zoomed screenshot of the casting and pattern dimension entry area. The pattern designer uses the first three columns to enter desired casting dimensions along with upper and lower control limits. Dimension type and core type columns are filled in automatically by software depending on user selections in the feature type selection picture. The pattern dimension is filled in automatically by the software after the feature type is selected; however, the user can enter any other number in that column.

164 142 Figure 6-10: Casting and pattern dimension entry area Feature Type Selection Picture Area The user has to specify several process and geometry variables such as molding method, core type, restraint type, and parting line location in order for the software to predict the pattern dimension from the casting dimension. In order to simplify feature type specification, a feature type selection picture is created as shown in Figure The user clicks different buttons to choose the desired feature type. Once the selection is made, dimension type and core type columns in the dimension entry areas are automatically filled in. The Pattern Allowance Advisor then uses default PSU model and custom model (if available) to calculate average, minimum, and maximum pattern dimensions. The average pattern dimension is filled in column six of the dimension entry table depending on user s choice of the model.

165 143 Figure 6-11: Feature type selection picture Pattern Dimension Guide Pictures In order to represent the variability in the pattern dimension, the pattern dimension prediction model generates a scaled average, minimum, and maximum pattern dimension image when the user selects the dimension type. Predicted pattern dimension values using default and custom models along with 1/4 and 5/16 shrink rule are then presented pictorially to the user as shown in Figure In order to help the user choose between default and custom models, model details such as number of data points, average, minimum, and maximum feature size values from custom and default model databases are presented to the user. Using custom and default model pattern dimension

166 144 pictures and other details about these models, the user can select the most appropriate pattern dimension and enters it in the pattern dimension column (if different from average values that is automatically entered). Figure 6-12: Pattern dimension guide pictures. 6.5 Calibration Module Optimization Algorithm Minimization of error summation is carried out in order to calculate the standard shrinkage allowance for each feature type. The Microsoft Excel solver has been used for this purpose when developing new pattern allowance models. These new pattern allowance models serve as the default models in Pattern Allowance Advisor. However, the Pattern Allowance Advisor also provides customization capabilities for developing and calibrating a custom pattern allowance model suited for a specific foundry based on their historic inspection data. In such case, Microsoft Excel solver cannot be used for the purpose of minimization due to data security reasons. Therefore, an algorithm for

167 145 minimization of the summation of error from newly entered inspection data has to be developed and embedded into the Pattern Allowance Advisor. Since the optimization objective function is continuous and linear as shown in Figure 5-5, a simple interval search algorithm called the Golden Interval algorithm [71, 72] integrated with zero bracketing is used for minimization of error summation Golden Interval Method for Minimization of Error Summation Since the behavior of the objective function is well known and the targeted optimum value for error summation is zero, a simple bracketing algorithm can be used in the first step to find a small interval of uncertainty within which the optimum solution exists. The Golden Interval method is then used to converge to the solution within that bracket of uncertainty. This method uses the continuous interval reduction strategy by reducing the interval by the Golden section ratio (τ) of [73]. Figure 6-13 shows the algorithm used for integrated bracketing and Golden Interval method for the minimization of shrinkage error [72].

168 146 Calculate summation of error F1 using initial shrinkage allowance X1 = X2 = X1 + step; evaluate F2 Yes Is F2 > F1? No Interchange X1 and X2, step = -step X4 = X2 + step Yes X3 = τ*x4 + (1 - τ)*x1 Is F4 > F2? No Rename X2,X4 as X1,X2 Yes Is F3 > F2? No Rename X1,X3 as X4,X2 Rename X2,X3 as X1,X2 Stopping Criteria Met? Yes Stop Figure 6-13: Integrated bracketing and Golden interval algorithm for the minimization of error summation [72]. The stopping criteria used for this minimization is twofold. First, the difference between the targeted value of objective function (absolute zero) and the actual value should be within the tolerance of 1E-10 Inches. A second criterion is based on the change in the value of shrinkage allowance γ in successive iterations. Difference between the values of shrinkage allowance for successive iterations should be within the tolerance of

169 147 1E-10. If both of these criteria are met, then the optimization algorithm stops. Furthermore, there is an upper limit on the number of iterations so as to avoid entering an endless loop. If this limit is reached, the optimization loop stops with a message alerting the user of inability of convergence within specified tolerance limits Validation of the Optimization Algorithm In order to validate the optimization algorithm and the in-house code, optimization results obtained using Microsoft Excel solver were compared with those obtained using the Golden Interval optimization algorithm. Table 6-3 compares the values of standard shrink and summation of error obtained using the Excel solver and the in-house optimization code for all features types in low alloy castings made in green sand and nobake sand. Although there are differences in the values of summation error due to the differences in the respective stopping criteria, those differences are insignificant due to the fact that all error sum values are practically zero. Standard shrinkage allowances predicted by both the solvers agree with each other thus validating the in-house code for obtaining standard shrinkage allowances.

170 148 Table 6-3: Comparison of the values of standard shrink and summation of error for all types of features in low alloy castings made in green sand and nobake sand obtained using the Excel solver and the in-house code. Type Values obtained by Excel solver Values obtained by inhouse solver Shrink ErrorSum Shrink ErrorSum LowAlloyGreenSandFR 2.12E E E E-14 LowAlloyGreenSandPR 1.58E E E E-14 LowAlloyGreenSandUR 1.93E E E-13 LowAlloyNoBakeFR 1.52E E E E-13 LowAlloyNoBakePR 2.08E E E E-13 LowAlloyNoBakeUR 2.55E E E E Software Testing Extensive software testing was conducted both at Penn State and in the field in production foundries. Pattern Allowance Advisor Version 1.0 has been released and is officially recommended for pattern dimension prediction and dimensional data management by the Steel Founder s Society of America (SFSA) [69, 70]. Fifty SFSA member foundries have had access to the Pattern Allowance Advisor software for beta testing. Feedback has been obtained from three foundries after extensive in-plant testing. Inputs from McConway and Torley of Pittsburg, PA, Atchison Steel Casting & Machining of Atchison, KS, and Sivyer Steel Corporation of Bettendorf, Iowa along with members of SFSA research committee have played very important role in improving the software user interface. The graphical user interface, the efficiency of the software, and other features were improved due to this extensive beta-testing. These improvements and newly developed pattern allowance model customization capabilities will be released soon as part of Pattern Allowance Advisor Version 2.0.

171 Model Validation Using the Pattern Allowance Advisor In order to validate the pattern allowance model, the Pattern Allowance Advisor was used to predict pattern dimensions for seven steel castings at various foundries. Detailed inspection and casting dimensional conformance report for one of those castings is provided in this section along with conformance results summary for the other six castings Pattern Dimension Prediction Using Pattern Allowance Advisor A 700 lb low alloy steel shovel adapter casting (part number 12N) molded in green sand as shown in Figure 6-14 was selected for the model validation. Dimension types for each key casting feature were specified using the dimension entry window in the software. The pattern dimensions predicted using the Pattern Allowance Advisor were then used to build a wood pattern using conventional manual patternmaking techniques. At the same time, a sister shovel adapter part 10N from the same product family was built using conventional shrink rules. Table 6-4 shows pattern dimension values recommended by Pattern Allowance Advisor along with pattern dimension values calculated using the conventional 1/4, 5/16, and 3/16 per foot pattern allowances. As shown in Table 6-4, pattern dimension values recommended by Pattern Allowance Advisor are sometimes very different from the standard shrink rules. Furthermore, depending on mold restraint, parting line location, and core type, the Pattern Allowance Advisor recommends different pattern feature dimensions for casting features with the same nominal dimensions.

172 150 Cold-box core S49 S52 S53 S48 Green sand core Figure 6-14: Shovel Adaptor molding layout.

173 Table 6-4: Pattern dimensions for shovel adaptor 12N recommended by the Pattern Allowance Advisor compared to conventional pattern allowance rules. No. Casting Dimension Restraint Type Recommended by Pattern Allowance Advisor Pattern Dimension (Inches) 1/4 in/ft (2.08 %) 3/16 in/ft (1.526 %) 151 5/16 in ft. (2.60 %) 1 3 FR UR UR FR UR UR UR UR PR FR PR PR UR UR UR UR FR FR FR UR FR UR UR UR UR UR UR UR UR UR UR

174 Inspection Reports and Conformance Comparison The shovel adaptor casting is shown in Figure Both 10N version (produced from the new pattern built using conventional shrink rules) and 12N version (produced from the new pattern built using pattern dimensions recommended by Pattern Allowance Advisor) were inspected using articulated CMM. Tables 6-5 and 6-6 show dimensional conformance and centering errors (difference between actual casting dimension and targeted nominal casting dimension) for the first article casting features for 12N and 10N castings respectively. For the first article shovel adopter 12N casting, out of 30 features inspected, 27 casting features were within the dimensional tolerance limits as shown in Table 6-5. In case of 10N casting, however, of total 31 critical casting feature dimensions inspected, 16 (51.6%) were within the specified tolerance limits as shown in Table 6-6. Figure 6-15: Shovel Adaptor casting.

175 Table 6-5: First article conformance report for Shovel adaptor 12N casting produced using pattern dimensions recommended by Pattern Allowance Advisor. Casting Dimension (Inches) Nominal Minimum Maximum Actual Conformance Centering Error Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Not Inspected NA No

176 Table 6-6: First article conformance report for Shovel adaptor 10N casting produced using pattern dimensions predicted with conventional shrink rules. Casting Dimension (Inches) Nominal Minimum Maximum Actual Conformance Centering Error Yes No No No No No Yes Yes Yes No Yes Yes Yes No No No Yes Yes Yes Yes Yes No Yes No Yes No Yes No No Yes No Centering errors in terms of difference between measured casting feature dimension and specified nominal casting dimension were calculated for 10N and 12N shovel adaptor castings. Table 6-7 summarizes the centering errors observed in both

177 155 the castings. The centering errors of the casting features built using pattern dimensions recommended by Pattern Allowance Advisor have a lower mean centering error and error variance than casting features built using standard pattern allowances. Table 6-7: Comparison of centering errors for 10N casting built using conventional pattern allowances and 12N casting built using pattern allowances recommended by Pattern Allowance Advisor. Casting Centering error (Inches) Mean Standard Deviation Variance 10N casting using conventional PA N casting using Pattern Allowance Advisor Totally, 165 casting features were inspected from the seven castings built using pattern dimensions recommended by Pattern Allowance Advisor. Out of those 165 features, 133 casting features (80.6%) were within specified dimensional tolerances. Historically, first article dimensional conformance for castings built using conventional shrink rules has been approximately 50% [2, 57]. The casting dimensional conformance rate observed for the 10N casting inspected in this study also agrees with the historical data. A 30% improvement in first article casting dimensional conformance has been achieved using pattern dimensions recommended by the Pattern Allowance Advisor. Application of pattern dimensions recommended by Pattern Allowance Advisor has not only increased the first article dimensional conformance but has also reduced the total amount of centering error for all casting dimensions. Even better first article conformance

178 156 can be expected using improved pattern allowance models for features crossing the parting line that are part of the current Pattern Allowance Advisor version. Due to significant improvements in the dimensional conformance and centering error of the first article casting, dimensional reengineering cycles required to produce a dimensionally acceptable casting are reduced from a historical 3.5 [2] to 1.1. Reduction in number of dimensional reengineering cycles has obviously resulted in reduction in dimensional lead time. 6.8 Validation of the Pattern Allowance Advisor Calibration Capabilities In order to validate the customization and calibration capabilities in the Pattern Allowance Advisor, a new part called LowGreenPR was created. Specifications for sixteen features from the Penn State dimensional database for low alloy green sand molded partially restrained features were entered in the Pattern Allowance Advisor using the dimensions entry window. Figure 6-16 shows the dimensions entry window for the LowGreenPR part with model details turned ON. Since the custom database does not have any inspection data for pattern or casting dimensions for the selected feature type, the custom pattern dimension estimation picture is not drawn. Furthermore, model details indicate that there are no data points of this type in the database. Pattern and casting inspection dimensions for these features were then manually entered in the dimensional database of the Pattern Allowance Advisor using casting and pattern inspection modules as shown in Figure 6-17.

179 Figure 6-16: The dimensions entry window when there are no dimensional data entries of the selected feature type in the custom database. 157

180 158 Figure 6-17: Casting inspection data for LowGreenPR castings is entered using casting inspection module in the Pattern Allowance Advisor. The calibration module was then executed for low alloy green sand molded partially restrained features as shown in Figure Once the calibration results were reviewed, they were accepted into the custom database.

181 159 Figure 6-18: Calibration module showing the calibration results for low alloy green sand molded partially restrained features. Figure 6-19 shows the dimensions entry window accessed from the modify existing part module. Since casting and pattern dimensions for low alloy green sand molded casting partially restrained features from the same database were entered to build the custom model, both custom and default models suggest identical pattern dimensions. Information displayed in the model details grid is also identical for this same reason. Because the calibration routine converged to the same pattern dimension recommendations as in the default model, the overall calibration algorithm has been validated.

182 160 Figure 6-19: Dimensions entry window displaying identical custom and default model pattern dimension recommendations. This simulation validates the calibration module used in the Pattern Allowance Advisor.

183 161 Chapter 7 Conclusions and Future Work An empirical model was developed and validated to predict steel casting pattern dimensions, based on casting alloy type, feature type, and feature size,. To build the empirical model, dimensional data for casting and pattern feature dimensions was collected from twenty production steel foundries. Gage R&R and process capability ratios were checked to ensure adequate sampling and data integrity. This dimensional data was then entered into a relational database. Following conclusions were made from the analysis of the pattern allowance data: Pattern allowance values calculated using (Pattern Size Casting Size) / Casting Size demonstrated large variations. This variability increased as the casting feature size decreased. Pattern allowance values could not be directly used to build an empirical model since they were not independent and normally distributed. A data transformation methodology using a Shrinkage Error term was developed to build an empirical model. Shrinkage error was calculated using the standard shrinkage allowance. Following conclusions can be drawn with respect to the shrinkage error: Standard shrinkage allowance values are dependant on casting alloy, molding method, restraint type, and parting line location.

184 162 Standard shrinkage allowances can be calculated by minimizing the value of summation of error for a particular feature type. The standard shrinkage allowance increases as the mold restraint decreases. Values of shrinkage error are both independent and normally distributed. Fully restrained features demonstrate a larger variation in the shrink error than partially restrained or unrestrained features. This can be attributed to the complex mold-metal interaction common in fully restrained features. 7.1 Conclusions A pattern dimension prediction model was then developed based on median, 10 th percentile, and 90 th percentile shrinkage error values. Median error values were used to calculate the central tendencies of the pattern allowance; whereas, percentile values were used to capture the variation observed in actual pattern allowances. A software tool called Pattern Allowance Advisor was developed to assist pattern designers in estimating pattern dimensions. This tool is also designed to manage the pattern and casting dimensional data. Furthermore, a customization and calibration module was added to the software so that the pattern allowance model used for pattern dimension prediction can be calibrated to a specific foundry. Following conclusions were drawn after testing the pattern allowance model along with Pattern Allowance Advisor tool and the calibration module.

185 163 The new pattern allowance prediction model uses casting alloy type, molding method, feature size, mold or core restraint type, and parting line location for the prediction of the pattern allowances. Pattern allowance predictions made using the new model agree well with actual observed pattern allowance values. The new pattern allowance prediction model successfully expresses the wide variation observed in actual pattern allowance values. Pattern Allowance Advisor software was used to implement this new model in production steel foundries. First article casting dimensional conformance was improved from a historic 51% to 81% using this software. Furthermore, dimensional reengineering cycles were reduced from a historic 3.5 to 1.1 resulting in reduction in dimensional lead time. Pattern Allowance Advisor Model calibration and customization capabilities were validated using simulated inspection data. 7.2 Future Work Although the industrial validation of Pattern Allowance Advisor and the underlying pattern dimension prediction model showed significant improvements in first article dimensional conformance as compared to the conventional methods, further improvements both in the pattern allowance model and the Pattern Allowance Advisor can be made.

186 Improvements in the Pattern Allowance Model According to the current scheme, the casting feature restraint type is classified into only one of three categories - fully restrained, partially restrained, and unrestrained features. Although this simple scheme is very useful, this scheme does not provide adequate resolution. For example, as shown in Figure 7-1, the current model classifies both features A and B as fully restrained features in spite of the differences between the amounts of surrounding metal. Furthermore, as shown in Figure 7-2, features C and D are both classified as partially restrained features and the same pattern allowance is applied to both the features. In order to improve the pattern allowance prediction and first article conformance, the feature type classification scheme should be improved by refining the resolution. This may be achieved by incorporating the ratio of fully restrained feature size to unrestrained feature size in the classification of features. this adds significant complexity to feature description, but this may be the only way to further improve pattern allowance predictions.

187 165 Figure 7-1: Resolution of the feature restraint type fully retrained Figure 7-2: Resolution of the feature restraint type partially retrained

188 166 Additionally, the new pattern allowance model does not incorporate neighborhood effects in predicting the pattern allowance. It has been shown that pattern allowances vary across the length of feature due to the impact of neighboring casting geometry. Therefore, it would be more precise and accurate to predict the pattern allowance depending on the neighboring casting feature geometries. Once the influence of neighboring geometry, non-uniform cooling, camber distortion, and other factors are incorporated in the model, it may be possible to guide the pattern designer to a more precise pattern dimension rather than the current range. For example, as shown in Figure 7-3, if the desired casting dimension is four inches, the current model provides a range of pattern allowances to be used. However, by developing a regression equation as shown in Equation 7.1, a more precise estimation of pattern dimension could be made.

189 167 Pattern Allow ance (% ) 8.00% 6.00% 4.00% 2.00% 0.00% -2.00% -4.00% -6.00% Observed PA Min PA Max PA Avg PA % Feature Length (Inches) Figure 7-3: Pattern allowance prediction within the specified range φ n = µ 1 i Xi 7.1 In Equation 7.1, φ is the precise pattern allowance within the range predicted by the current model. Values of µ are the multiplication constants for various independent variables Xs.

190 168 Although currently the pattern allowance model has been developed and used for steel castings only, with the embedded calibration techniques, pattern allowance models for other types of castings and processes can be easily developed using the same Pattern Allowance Advisor software platform Improvements in Pattern Allowance Advisor Software Currently, pattern designers enter all the casting feature dimensions manually. A new procedure can be added to the current software to link the pattern allowance prediction model to 2D CAD software such as AutoCAD. When such a procedure is developed, the pattern designers will be able to input the 2D casting drawing into the Pattern Allowance Advisor software directly, saving the time required for entering the dimensions. Extending this logic further, procedures could be developed to import a 3D casting model into the Pattern Allowance Advisor software and generate all the required pattern dimension values. Then a 3D pattern model can be exported from the Pattern Allowance Advisor to CAM software to generate the CNC tool paths for patternmaking. The Pattern Allowance Advisor software is currently a stand-alone application. With some modifications, this tool can be converted into a web-based tool with proper user access settings. If this tool is available as a web-based tool, individual foundry users will not have to install newer versions of the software as improvements are made and all the inspection data from various foundry users will be stored in a big central database. Furthermore, it will be easier to develop models for various alloys and feature types

191 because every user will have access to selected information from a very huge central database. 169

192 170 References 1. American Foundry Society Foundry Trends Schaumburg, IL, Accessed 10 August 2004, Available at: 2. Peters, F.E.; Unpublished Survey of Lead Time at a Steel Foundry Peters F. E., Pattern Allowance Prediction for Steel Castings, PhD Dissertation, The Pennsylvania State University, University Park PA, Potter L., Voigt R., Peters F., Lies J and Chandra M., A statistically based pattern approval Process, AFS transactions, Volume 104, pp , Voigt, R.C. and F.E. Peters; "Dimensional Tolerances and Shrinkage Allowances for Steel Castings," Proceedings of the Steel Founders' Society of America T & O Conference, Chicago, IL, Peters, F., and R. Voigt, Experiments on the Dimensional Changes of Steel Castings during Solidification, Proceedings, Modeling of Casting, Welding and Advanced Solidification Processes VII, TMS, pp , Campbell, J.; Castings, Butterworth-Heinemann Ltd., Oxford, U.K., Rickards, P.J.; Factors Affecting the Soundness and Dimensions of Iron Castings Made in Cold-curing Chemically Bonded Sand Moulds, British Foundryman, volume 75, part 11, p. 213 (1982). 9. Wieser P. F., Aubery L. S., Rowe C., Dimensional tolerances of Production castings J. Steel Casting Research, No. 85, (Dec 1975) 10. Steel Founders Society of America, Steel Casting Handbook Steel Founders Society of America, Des Plaines, IL (1980) 11. Steel Founders Society of America, Dimensional Tolerances Steel Founders Society of America, Report No. 84, (1977) 12. International Standards Organization; Castings system of dimensional tolerances and machining allowances, ISO 8062 (1984) 13. International Standards Organization; Castings system of dimensional tolerances and machining allowances, Second Edition, ISO 8062 (1994)

193 14. International Standards Organization; Castings system of dimensional tolerances and machining allowances, Third Edition, ISO 8062 (1996) 15. International Standards Organization; Castings system of Geometric Tolerances, ISO (1995) 16. Steel Raw Castings general tolerances and machining allowances DIN standard No. 1683, Part1 (1980) Voigt, R. C., and F. E. Peters, Dimensional Tolerances and Shrinkage Allowances for Steel Castings Proceedings, Steel Founders' Society of America T&O Conference, Des Plaines, IL (1992) 18. Peters, F. E., J. Ristey, W. Vaupel, E. C. De Meter, and R. C. Voigt, "Dimensional Variability of Production Steel Castings," Proceedings, Steel Founders' Society of America T&O Conference, Chicago, IL, IBF technical subcommittee TS71, First Report of Technical Subcommittee TS71 Dimensional Tolerances in castings, The British Foundryman, Vol 62 Part5, pp. 179, IBF technical subcommittee TS71, Second Report of Technical Subcommittee TS71 Dimensional Tolerances in castings, The British Foundryman, Vol 64 Part10, pp. 364, Villner, L., A General System of Dimensional Tolerances for Castings, The British Foundryman, vol. 62, no 12, Svensson, I., L. Villner, Dimensional Accuracy of Castings, The British Foundryman, vol. 67, no 10, Peters, F. E., and R. C. Voigt, Dimensional Capabilities of Steel Castings" Proceedings of the Near-Net-Shape Manufacturing Conference, ASM International, Metals Park, OH, Peters, F. E., R. C. Voigt, L. A. Potter, and E. C. DeMeter, Initial Study of the Dimensional Variability of Production Steel Castings Research Report No. 109, Steel Founders Society of America, Des Plaines, IL, Peters, F. E., R. C. Voigt, L. A. Potter, and E. C. De Meter, "Dimensional Repeatability of Steel Castings: An Update," Proceedings, Steel Founders' Society of America T&O Conference, Chicago, IL, Vaupel, Wayne G., Edward C. DeMeter, Frank E. Peters, Robert C. Voigt, The Implications of Tolerance System Interpretation on past and present Dimensional

194 Variability Studies. Proceedings, Steel Founders Society of America T&O Conference, Des Plaines, IL, Faustine, W. C., N. Vanikar and R. C. Voigt "Feature and Geometric Variability of Production Steel Castings," Proceedings Steel Founders' Society of America T&O Conference, Chicago, IL, Karve, A., Padmanabhan, and R. C. Voigt, 1997, "Factors Influencing the Dimensional Variability of Investment Castings," Proceedings 45th Investment Casting Institute Conference, Atlanta, GA, Vol. 3, pp [Republished in Incast, 1997, Vol. X, No. 7, pp ] 29. Ross, P.J.; "Measurement System Capability Project," Proceedings of the Steel Founders' Society of America T & O Conference, Chicago, IL, November Ross, P.J.; "Measurement System Capability Project Update," Proceedings of the Steel Founders' Society of America T & O Conference, Chicago, IL, November AIAG, Measurement System Analysis reference manual, Automotive Industry Action Group, Southfield, MI, Peters, F. E. and R. C. Voigt, Assessing the Capabilities of Pattern Shop Measurement Systems AFS Transactions Vol. 103, pp , Karve, A.A., L.A. Potter, R.C. Voigt, Selection and Use of Steel Foundry Dimensional Inspection Equipment Based on Measurement Systems Analysis Studies, Proceedings of the 50th Steel Founders Society of America T & O Conference, Karve, A., M. J. Chandra, and R. C. Voigt, "Determining Dimensional Capabilities from Short Run Sample Casting Inspection," AFS Transactions, Vol. 106, pp , Goodwin, F.E., S.G.R. Brown, J.A. Spittle, Development of Easy-to-Use Die Design Programs for Zinc Die Casting, Die Casting Engineer, vol. 38, no 5, pp , Spittle, J.A., S.G.R. Brown, ILZRO Project ZM 363 : Effect of Die Cavity Size on Casting Size - Annual Progress Report for the Period 1st January st December 1992, University College Swansea, Swansea, Ou S., Beckermann C., Simulation of dimensional changes in steel castings, Proceedings of the Steel Founders' Society of America T & O Conference, Chicago, IL (2003).

195 38. Patternmaking Guide, Second Edition, American Foundrymen s Society, Des Plaines, IL, Voigt, R.C., N. Ivey, S. Halbe, Dimensional Control of Steel Castings, Proceedings, Steel Founders Society of America, T&O Conference, Chicago, IL, (2001) 40. Ivey N. J., Understanding the Dimensional Variability of Steel Castings MS Thesis, The Pennsylvania State University (1999) 41. Okhuysen V. F., Determination of Tooling Allowances in Investment Castings. PhD Dissertation, The Pennsylvania State University(1998) 42. Griffiths, E., ed., Physical Constants of Some Commercial Steels at Elevated Temperatures, Butterworth Scientific Publications, London Briggs, C.W. and R.A. Gezelius; "Studies on Solidification and Contraction in Steel Castings -II --Free and Hindered Contraction of Cast Carbon Steel," AFS Transactions, vol 42, p 449 (1934) 44. Ward, E.; "High-Temperature Properties of Moulding-Materials as Related to Dilation of Moulds for Steel-Castings," Foundry Trade Journal, vol 120, p 601 (May 1966) 45. Sosman R. B., The properties of Silica, The Chemical Catalog Co. USA, American Chemical Society Monograph Series, p. iv-45, Kubo K., Pehlke R. D., Metallurgical Transactions, 17B, , Bates, C. and J.F. Wallace; "An Investigation of Thermal and Mechanical Stability of Mold Materials for Steel Castings," Modern Casting, vol 49, p 216 (May 1966) 48. Engler, S., D. Boenisch, and B. Kohler; "Metal and Mold Wall Movements During Solidification of Cast Iron," Cast Metals Research Journal AFS, vol 9, no 1, p 20 (1973) 49. Winter B. P., Ostrom T. R., Sleder T. A., Trojan P. K., Pehlke R. D., American Foundrymen Society Transactions, 95, , Levelink H. G., Julien F. P. M. A., AFS Cast Metals Research Journal, 56-63, June Nishida Y., Droste W., Engler S., Meta Transactions, 17B, , (1986)

196 Bertolino, M.F. and J.F. Wallace; "Influence of Metal Factors, Molding Sand Composition and Casting Shape on Mold Cavity Enlargement," AFS Transactions, vol 75, p 708 (1967) 53. Lucking, H.J. and H. Pacyna; Dimensional Differences Between Patterns and Machine Molded Steel Castings, Giesserei, vol 57, p 320 (1970). 54. Vaupel W. G., The effects of final stage processing on the dimensional and geometric variability of production steel castings, MS thesis, The Pennsylvania State University (1995) 55. Okhuysen, V.F., and Robert C. Voigt, Heat-Treatment Effects on Tooling Allowance in Investment Casting, AFS Transactions, 69:25-28 (1999) 56. Nyichomba, B.B., V. Kondic, and G.H.J. Bennett; "Linear Contraction of Grey Cast-Iron Sand Castings," Cast Metals, vol 4, no 4, p 195 (1992) 57. Kochar V. M., Variation in Pattern allowances MS Thesis, The Pennsylvania State University (2002) 58. Karve, A., Dimensional Control of Die Casting, PhD Dissertation, The Pennsylvania State University (1996) 59. Voigt, Robert, Raymond Monroe, Gary DiSpensa, Charles Monroe, Frank Peters, Pattern Allowances Predicting Casting Dimensions from Tool Dimensions, May Halbe S. S., Dimensional Control of production Aluminum sand Casting, MS Thesis, The Pennsylvania State University (2002) 61. Henschel, C., R.W. Heine, and J.S. Schumacher; "Casting Dimensions and Mold Dilation," AFS Transactions, vol 74, p 357 (1966) 62. Wagner, K.; "Studies on the Use of Shrink Rules in Foundries," Giesserei, no 4 (Feb 1990) 63. Stephens, M. A. (1974). EDF Statistics for Goodness of Fit and Some Comparisons, Journal of the American Statistical Association, Vol. 69, pp Longden E., AFS Transactions, Vol 56, pp Deo M. and Voigt R.: Two and Half Dimensional Analysis for Large Green Sand (10N Shovel) Casting Pattern Allowance Progress Report, Penn State University, 2002

197 Peters, F. E., and R. C. Voigt, Casting Inspection Strategies for Determining Dimensional Variability Proceedings, Steel Founders' Society of America T&O Conference, Steel Founders Society of America, Des Plaines, IL ( Online Product manuals 68. Khanolakar A. A., Variation in Pattern allowances MS Thesis, The Pennsylvania State University (2005) 69. Steel Founder s Society of America American Metalcasting Consortium Annul Report on Casting Solutions for Military Readiness and Sustainment (2005) 71. Brent, Richard P., Algorithms of Minimization without Derivatives, Prentice Hall, Englewood Cliffs, NJ, Belegundu Ashok D., Chandrupatla Tirupathi R., Optimization Concepts and Applications in Engineering, Prentice Hall, Upper Saddle River, NJ, Chandrupatla T.R., Sectioning Algorithm for Minimization without Derivatives, Computer Methods in Applied Mechanics and Engineering, 152, , (1998) 74. Umapathy K., A Software Guidance Tool for Pattern Allowance Prediction and Dimensional Data Integration of Steel Castings MS Thesis, The Pennsylvania State University (2002)

198 Appendix A Casting Variables Survey Form Please circle or fill in the most appropriate answer. Add additional comments wherever necessary. Foundry Name Part Number Approximate number of castings produced each year Approximate number of castings produced per lot Number of castings per mold Cavity number measured If more than one, what is the minimum between spacing

199 177 What is the casting used for: automotive, railroad, pump/valve, other Type of flask: tight, slip, flaskless, other Mold size (inch x inch) Cope height Drag height Are jackets used during pouring yes / no Alloy or grade of metal used Pouring temperature range specified High Low How are the symbols weighted during pouring Type of heat treatment Is this casting typically upgraded by straightening yes / no What is the total pour weight units?

200 178 What is the finished casting weight units? What is the largest dimension on the casting units? What is the wall thickness, if applicable units? What is the projected area of casting? units? Bounding box that contains casting? units? How many cores are used What is the total volume of internal coring (inch 3 ) If shell cores are used, what is the specified shell core thickness units? Type(s) of pattern: aluminum, epoxy, wood, iron, steel, bronze, other Condition of pattern: excellent, very good, good, fair, poor How old is the pattern equipment (if known)

201 179 Type of pattern set-up: matchplate, separate cope & drag, loose, other Type of molding sand: green, no-bake, shell, other What is the assigned Sand System Number (SSN) Is facing sand used yes / no If yes, what is the assigned Facing Sand System Number (FSSN) If flask used, how often are the pins and brushings on flasks checked Molding method: jolt squeeze, slinger, hand rammed, automatic (type ), other What type of chills used: none used, nails, block (size), other

202 180 Is mold wash used yes / no Please add any comments about binders, additives, process controls, etc., or anything of relevance that has not been included: A mold to mold across casting B mold to mold across mold C mold to mold across mold and casting D mold to mold across casting/mold/casting E mold to core across casting F core to core across core G mold to core across casting and core H mold to mold across casting/core/casting I core to core across casting O other configurations?

203 181 H D G C F E B A MOLD CASTING CORE CASTING MOLD MOLD CASTING MOLD CASTING MOLD

204 Casting Feature Variables Survey Form 182 Feature number Dimension location cope, drag, across PL Direction of the dimension Do chills affect this feature dimension Does draft affect this feature Does grinding affect this feature* Does mold wash affect this feature Is the dimension upgraded by straightening* What is the nominal dimension What is the total tolerance What is the mold/core relationship Type of core used for this feature What is the Core System Number How are the cores made Are the cores lighted How are the cores set Does dimension cross core parting line Is dimension part of a core assembly Does dimension cross core assembly joint Is a core assembly fixture used Does core wash affect this feature Type of core box used to manufacture core Condition of core box

205 Appendix B Sample Part File

206 184

207 185

208 186

209 187

210 188

211 Appendix C Sample Gage R&R Calculation Table C-1: Gage Repeatability and Reproducibility Data Sheet - 2 operators, 2 trials, and 10 parts APPRAISER/ PART TRIAL # Average A Average Range B Average Range Part Average: Average of the Average Ranges: Difference between the Averages Upper Control Limit on Range: Range of the Part (Rp) Repeatability - Equipment EV = E-05 Total Part Tolerance (PT) Variation (EV): = 3E-06 Reproducibility - Appraiser AV = E-05 % R&R = R&R/(PT) 18.06% Variation (AV): = Repeatability and Reproducibility R & R = Total Variation (TV) (R & R): = Part Variation (PV): PV = % R&R = R&R/(TV) 21.89% = Total Variation (TV): TV =

212 Figure C-1: Sample filled out gage R&R data sheet. Template from AIAG manual [31] 190

213 Figure C-2: Sample filled out gage R&R report. Template from AIAG manual [31] 191

214 Appendix D Point-to-Point 2D database dimensions Sr. No Table D-1: Complete Database for high and low alloy features. Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type F LOW A27 GREEN NOBAKE N F LOW A27 GREEN NOBAKE N F LOW A27 GREEN NOBAKE N F LOW A27 GREEN NOBAKE N B LOW M201 GREEN N/A N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N B LOW M201 GREEN N/A N B LOW M201 GREEN N/A N B LOW M201 GREEN N/A N H LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N I LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N I LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N Dim. Across Parting Line?

215 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N F LOW M201 GREEN NOBAKE N B LOW NBAKE N/A N B LOW NBAKE N/A N B LOW NBAKE N/A N B LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N F LOW NBAKE NOBAKE N F LOW NBAKE NOBAKE N F LOW NBAKE NOBAKE N F LOW NBAKE NOBAKE N A LOW A148 NBAKE N/A Y A LOW A148 NBAKE N/A Y A LOW A148 NBAKE N/A Y A LOW A148 NBAKE N/A Y F LOW NBAKE NOBAKE N A LOW NBAKE N/A N A LOW NBAKE N/A N F LOW NBAKE NOBAKE N A LOW NBAKE N/A N A LOW NBAKE N/A N B LOW A148 NBAKE N/A N 193 Dim. Across Parting Line?

216 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type F LOW A148 NBAKE NOBAKE N F LOW A148 NBAKE NOBAKE N F LOW A148 NBAKE NOBAKE N F LOW A148 NBAKE NOBAKE N I LOW A148 NBAKE NOBAKE N I LOW A148 NBAKE NOBAKE N I LOW A148 NBAKE NOBAKE N O LOW A148 NBAKE NOBAKE N I LOW A148 NBAKE NOBAKE N I LOW A148 NBAKE NOBAKE N I LOW A148 NBAKE NOBAKE N O LOW A148 NBAKE NOBAKE N F LOW A148 NBAKE NOBAKE N F LOW A148 NBAKE NOBAKE N F LOW A148 NBAKE NOBAKE N F LOW A148 NBAKE NOBAKE N H LOW A148 NBAKE NOBAKE N H LOW A148 NBAKE NOBAKE N A LOW A148 NBAKE N/A N A LOW A148 NBAKE N/A N H LOW A148 NBAKE NOBAKE N H LOW A148 NBAKE NOBAKE N A LOW A148 NBAKE N/A N A LOW A148 NBAKE N/A N H LOW A148 NBAKE NOBAKE Y H LOW A148 NBAKE NOBAKE Y A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N/A Y A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW NBAKE N/A Y A LOW NBAKE N/A N A LOW NBAKE N/A N A LOW 8630 NBAKE N/A N A LOW 8630 NBAKE N/A Y A LOW 8630 NBAKE N/A Y A LOW 8630 NBAKE N/A N A LOW 8630 NBAKE N/A N A LOW 8630 NBAKE N/A N H LOW 4130 NBAKE NOBAKE N 194 Dim. Across Parting Line?

217 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type A LOW 4130 NBAKE N/A Y A LOW 4130 NBAKE N/A Y A LOW 4130 NBAKE N/A N H LOW 1026 NBAKE NOBAKE Y A LOW 1026 NBAKE N/A N H LOW 1026 NBAKE NOBAKE Y D LOW 1026 NBAKE N/A N F LOW 1019 NBAKE NOBAKE N H LOW 1019 NBAKE NOBAKE Y F LOW 1019 NBAKE NOBAKE N I LOW 1525 NBAKE NOBAKE N F LOW 1525 NBAKE NOBAKE N O LOW 1525 NBAKE N/A N F LOW 8628 NBAKE NOBAKE N H LOW 8628 NBAKE N/A N F LOW 8628 NBAKE NOBAKE N H LOW 8628 NBAKE NOBAKE N C LOW 8628 NBAKE N/A N E LOW 8628 NBAKE NOBAKE N I LOW 8628 NBAKE NOBAKE N H LOW 1022 NBAKE NOBAKE Y E LOW 1022 NBAKE NOBAKE N E LOW 1022 NBAKE NOBAKE N A LOW 1022 NBAKE N/A N A LOW 1022 NBAKE N/A N H LOW 1022 NBAKE NOBAKE Y H LOW 1022 NBAKE NOBAKE Y F LOW 1022 NBAKE NOBAKE N A LOW 1022 NBAKE N/A N A LOW 1022 NBAKE N/A N A LOW 1022 NBAKE N/A N A LOW A216 SHELL N/A N A D LOW A216 SHELL N/A N B D LOW A216 SHELL N/A N C D LOW A216 SHELL N/A N D D LOW A216 SHELL N/A N A A LOW A216 GREEN N/A N B A LOW A216 GREEN N/A N C A LOW A216 GREEN N/A N D A LOW A216 GREEN N/A N F LOW A216 SHELL SHELL N F LOW A216 SHELL SHELL N F LOW A216 SHELL SHELL N 195 Dim. Across Parting Line?

218 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type F LOW A216 SHELL SHELL N A LOW A148 GREEN N A LOW A148 GREEN N F LOW A216 GREEN SHELL N F LOW A487 GREEN SHELL N G LOW A216 GREEN SHELL N A LOW A216 GREEN N/A N F LOW A148 GREEN NOBAKE N F LOW A148 GREEN NOBAKE N F LOW A148 GREEN NOBAKE N F LOW A148 GREEN NOBAKE N F LOW A216 GREEN SHELL N A LOW A216 GREEN N/A Y A LOW A216 SHELL N/A N E LOW A216 GREEN NOBAKE N F LOW A216 GREEN NOBAKE N A LOW A216 GREEN N/A N F LOW A216 SHELL SHELL N F LOW A148 SHELL SHELL N F LOW A148 SHELL SHELL N A LOW NBAKE N/A N B LOW NBAKE N/A N O LOW NBAKE N/A N D LOW NBAKE N/A N B LOW GREEN N/A N H LOW GREEN N/A N A LOW GREEN N/A Y A LOW SHELL N/A N A LOW SHELL N/A N F LOW GREEN SHELL N B LOW GREEN N/A N B LOW GREEN N/A N D LOW GREEN N/A N D LOW GREEN N/A N A LOW GREEN N/A N B LOW GREEN N/A N A LOW WCB SHELL N A LOW WCB SHELL N D HIGH CF8M SHELL N D HIGH CF8M SHELL B HIGH CF8M SHELL N D LOW WCB SHELL N B LOW WCB SHELL N 196 Dim. Across Parting Line?

219 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type A HIGH CF8M SHELL N A HIGH CF8M SHELL N A HIGH CF8M SHELL SHELL N A HIGH CF8M SHELL N A HIGH CF8M SHELL N A LOW WCB SHELL SHELL N A LOW WCB SHELL N A LOW WCB SHELL N D HIGH CF8M SHELL N B HIGH CF8M SHELL N D HIGH CF8M SHELL N B LOW WCB SHELL N A LOW WCB NBAKE N D HIGH CF8M NBAKE N D HIGH CF8M NBAKE N B HIGH CF8M NBAKE N B HIGH CF8M NBAKE N A HIGH CF8M NBAKE N E LOW WCB SHELL SHELL N A LOW WCB SHELL N A LOW WCB SHELL N A HIGH CF8M SHELL SHELL N A HIGH CF8M SHELL N A HIGH CF8M SHELL N A HIGH CF8M SHELL N A HIGH CF8M SHELL N A LOW 8620 SHELL N/A N H LOW 8620 SHELL SHELL N D LOW 8620 SHELL N/A N B LOW 8620 SHELL N/A N B LOW 8620 SHELL N/A N A LOW 8620 SHELL N/A N A LOW 8620 SHELL N/A N A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N D LOW P6 GREEN N/A N E LOW P6 GREEN SHELL N A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N A LOW P6 GREEN N/A Y 197 Dim. Across Parting Line?

220 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type A LOW P6 GREEN N/A Y A LOW P6 GREEN N/A N F LOW 1025 GREEN CO2 N A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N E LOW P6 GREEN OIL Y A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N E LOW P6 GREEN OIL Y F LOW 1025 GREEN CO2 N O LOW 1025 GREEN CO2 N A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A Y A LOW 1025 GREEN N/A Y F LOW 1025 GREEN OIL N A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A N I LOW 8620 GREEN OIL N I LOW 8620 GREEN OIL N A LOW 8620 GREEN N/A N I LOW 8620 GREEN OIL N I LOW 1020 GREEN SHELL N B LOW 1020 GREEN N/A N A LOW 8630 GREEN N/A Y A LOW 8630 GREEN N/A Y A LOW 1025 GREEN N/A N A LOW 1025 GREEN N/A N F LOW 1025 GREEN OIL N E LOW 1025 GREEN OIL N E LOW 1025 GREEN OIL N O LOW 8630 GREEN SHELL N A LOW P5 SHELL none N A LOW P5 SHELL none N A LOW P5 GREEN none Y A LOW P5 GREEN none Y A LOW P5 GREEN none Y A LOW P5 GREEN nono N 198 Dim. Across Parting Line?

221 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type A LOW P5 GREEN none N A LOW P5 GREEN none N A LOW P5 GREEN none N A LOW P5 GREEN none N A LOW P5 GREEN none N A LOW P5 GREEN none N A LOW P5 GREEN none N A LOW P5 GREEN none N A LOW P5 GREEN none N H LOW P25 GREEN N/A N B LOW P6 GREEN N/A N C LOW P6 GREEN N/A N A LOW P6 GREEN N/A Y A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N A LOW P6 GREEN N/A N A LOW OTHR SHELL N/A Y A LOW OTHR SHELL N/A N A LOW P25 SHELL N/A N A LOW P25 SHELL N/A N A LOW P25 SHELL N/A N B LOW P25 SHELL N/A N B LOW P25 SHELL N/A N B LOW P25 SHELL N/A N A HIGH CF8M SHELL NA Y A HIGH CF8M SHELL NA Y A HIGH CF8M SHELL NA Y A HIGH CF8M SHELL NA Y A HIGH CF8M SHELL NA Y A LOW WCB SHELL NA Y D LOW WCB SHELL NA Y A LOW WCB SHELL NA Y A LOW WCB SHELL NA Y D LOW WCB SHELL NA Y A HIGH CB7CU1 SHELL NA Y A HIGH CB7CU1 SHELL NA Y A HIGH CB7CU1 SHELL NA Y A HIGH CB7CU1 SHELL NA Y A HIGH CF8M NBAKE NA Y A HIGH CF8M NBAKE NA Y A HIGH CF8M NBAKE NA Y A HIGH CF8M NBAKE NA Y A HIGH CF8M NBAKE NA Y 199 Dim. Across Parting Line?

222 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type A HIGH CF8M NBAKE NA Y A HIGH CF8M SHELL NA Y A LOW WCB SHELL NA Y A LOW WCB SHELL NA Y A HIGH CF8M SHELL NA Y B HIGH CF8M SHELL NA Y A HIGH CF8M SHELL NA Y B HIGH CF8M SHELL NA Y A HIGH CF8M SHELL NA Y B HIGH CF8M SHELL NA Y A HIGH CF8M SHELL NA Y B HIGH CF8M SHELL NA Y F LOW 1025 SHELL SHELL N A LOW 1025 SHELL N/A N A LOW 1025 SHELL N/A N A LOW 1025 SHELL N/A N F LOW 1025 NBAKE OIL N A LOW 1025 NBAKE N/A N A LOW 1025 NBAKE N/A N A LOW 1025 NBAKE N/A N D LOW 1025 NBAKE N/A N H LOW 1025 SHELL SHELL Y A LOW 1025 SHELL N/A N A LOW 1025 SHELL N/A N B LOW 1025 SHELL N/A N H LOW 1025 NBAKE NOBAKE N H LOW 1025 NBAKE NOBAKE N F LOW 1025 NBAKE OIL N F LOW 1025 NBAKE OIL N G LOW 1025 NBAKE OIL N B LOW 1025 NBAKE N/A N B LOW 1025 NBAKE N/A N F HIGH CF8M NBAKE NOBAKE N F HIGH CF8M NBAKE NOBAKE N H HIGH CF8M NBAKE NOBAKE N H HIGH CF8M NBAKE NOBAKE N A HIGH CF8M NBAKE N/A N A HIGH CF8M NBAKE N/A N B HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N D HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N B HIGH CF8M SHELL N/A N 200 Dim. Across Parting Line?

223 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type H LOW 1025 SHELL SHELL Y D LOW 1025 SHELL N/A N A LOW 1025 SHELL N/A N B LOW 1025 SHELL N/A N A LOW 1025 SHELL N/A N H HIGH CF8M SHELL SHELL N H HIGH CF8M SHELL SHELL N F HIGH CF8M SHELL SHELL N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N H HIGH CF8M SHELL SHELL N H HIGH CF8M SHELL SHELL N F HIGH CF8M SHELL SHELL N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N D HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N D HIGH CF8M SHELL N/A N B HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A Y B LOW WCB GREEN N/A N A LOW WCB GREEN N/A N B LOW WCB GREEN N/A N A LOW WCB GREEN N/A N B LOW 8625 GREEN N/A N A LOW 8625 GREEN N/A N E LOW WCB SHELL SHELL N E LOW WCB GREEN SHELL N H LOW WCB GREEN SHELL N B LOW WCB GREEN N/A N B LOW WCB GREEN N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A N B LOW WCB GREEN N/A N B LOW WCB GREEN N/A N D LOW WCB GREEN N/A N 201 Dim. Across Parting Line?

224 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type B LOW WCB GREEN N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A N B LOW WCB GREEN N/A N D LOW WCB GREEN N/A N D LOW WCB GREEN N/A N B LOW WCB GREEN N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A N B LOW WCB GREEN N/A N D LOW WCB GREEN N/A N D LOW B GREEN N/A N H LOW B NBAKE NOBAKE N H LOW B NBAKE NOBAKE N A LOW EMS-15 GREEN N/A N F LOW EMS-15 GREEN NOBAKE N A LOW EMS-15 GREEN NOBAKE N H LOW EMS-15 GREEN NOBAKE N A LOW EMS-15 GREEN N/A N A LOW U45 NBAKE N/A N A LOW U45B6 NBAKE N/A N A LOW U40 NBAKE N/A N A LOW U40 NBAKE N/A N A HIGH CF3M GREEN N/A N A HIGH CF3M GREEN N/A N A HIGH CF3M GREEN N/A N A HIGH CF3M GREEN N/A N F HIGH CF8M GREEN * N F HIGH CF8M GREEN * N F HIGH CF8M GREEN * N A HIGH CF8M GREEN N/A N A HIGH CF10M GREEN N N B HIGH 4320 SHELL N/A N B HIGH 4320 SHELL N/A N B HIGH 4320 SHELL N/A N B HIGH 4320 SHELL N/A N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N A HIGH CF8M SHELL N/A N B LOW WCB SHELL N/A N 202 Dim. Across Parting Line?

225 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type B LOW WCB SHELL N/A N A LOW WCB SHELL N/A N B HIGH 4320 GREEN N/A N B HIGH 4320 GREEN N/A N B HIGH 4320 GREEN N/A N D HIGH 4320 GREEN N/A N B HIGH 4320 GREEN N/A N B HIGH 4320 GREEN N/A N B HIGH 4320 GREEN N/A N D HIGH 4320 GREEN N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A N A LOW WCB GREEN N/A N H LOW WCB SHELL SHELL N H LOW WCB SHELL SHELL N A LOW WCB SHELL N/A N A LOW WCB SHELL N/A N A LOW WCB SHELL N/A N B LOW WCB SHELL N/A N D LOW A148 SHELL N/A Y B LOW A148 SHELL N/A Y A LOW A148 GREEN N/A Y H LOW A148 SHELL SHELL Y A LOW A148 SHELL N/A Y H LOW A148 SHELL SHELL Y B LOW A216 NBAKE N/A N D LOW A487 NBAKE N/A Y C LOW A487 NBAKE N/A Y O LOW A487 NBAKE NOBAKE Y A LOW 4130 A GREEN N B LOW 4130 A GREEN N/A N D LOW 4130 A GREEN N/A N B LOW 4130 A GREEN N/A N A LOW 4130 A GREEN N/A N A LOW 4130 A GREEN N/A N D LOW 4130 A GREEN N/A N A LOW 4130 A GREEN N/A N A LOW 4130 A GREEN N/A N B LOW 4130 A GREEN N/A N A LOW 4130 A GREEN N/A N A LOW 4130 GREEN N/A N D LOW 4130 GREEN N/A N 203 Dim. Across Parting Line?

226 Sr. No Part Name Feature Name Feature Type Pattern Dim. (Inches) Casting Dim. (Inches) Alloy Type Alloy Name Molding Sand Type Core Type A LOW 4130 GREEN N/A N A LOW 4130 GREEN N/A N B LOW 4130 GREEN N/A N A LOW 4130 GREEN N/A N D LOW 4130 GREEN N/A N B LOW 4130 GREEN N/A N D LOW 4130 GREEN N/A N B LOW 4130 GREEN N/A N A LOW 4130 GREEN N/A N A LOW 4130 GREEN N/A N A LOW 4130 GREEN N/A N D LOW 4130 GREEN N/A N B LOW 4130 GREEN N/A N B LOW 4130 GREEN N/A N D LOW 4130 GREEN N/A N A LOW 4130 GREEN N/A N A LOW 4130 GREEN N/A N A LOW 4130 GREEN N/A N F LOW 8514 GREEN NOBAKE N 204 Dim. Across Parting Line?

227 205 D.1 Classified Database of features not crossing the mold parting line D.1.1 Green Sand Molded Low Alloy Steel Castings - Fully Restrained Features Table D-2: Database for green sand low alloy fully restrained features. Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance (%) # Part Name Feature Name Feature Type Shrinkage Error B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B % B %

228 206 D.1.2 Green sand molded low alloy steel castings partially restrained features Table D-3: Database for green sand low alloy partially restrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error C D D D D D D D D D D D D D D D D.1.3 Green sand molded low alloy steel castings Unrestrained Features Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance (%) # Part Name Feature Name Feature Type Shrinkage Error A % A % E % E % E % A % A % A % A % A % A % A % A %

229 Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance (%) # Part Name Feature Name Feature Type Shrinkage Error A % A % A % A % A % A % A % A % A % A % A % A % A % E % A % A % A % A % A % E % A % A % A % A % A % A % A % C A % A % D A % A % A A % B A % A % A % A % A % A % A % A % A % A % A % A %

230 Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance (%) # Part Name Feature Name Feature Type Shrinkage Error A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % A % D.1.4 Nobake sand molded low alloy steel castings fully restrained features Table D-4: Database for nobake sand low alloy fully restrained features. # Pattern Dimension (Inches) Casting Dimension (Inches) Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error F F F F F

231 Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error F F F F F F F F F F F F F B F F F F F F B B F F B F F F F F F F F F F F F F F F F F F B

232 Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error F F F F F F F F F F F F F F F F F F F F B B F F F F B B F F F F

233 211 D.1.5 Nobake sand molded low alloy steel castings partially restrained features Table D-5: Database for nobake sand low alloy partially restrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error H D C H H H H D D H H H H H H H

234 212 D.1.6 Nobake sand molded low alloy steel castings unrestrained features Table D-6: Database for nobake sand low alloy unrestrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error E E A A A A A A A A A A A A A A A A A A A A A A A E A A A A A A A A A A

235 Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error A A A A A A A A A

236 214 D.1.7 Shell molded low alloy steel castings fully restrained features Table D-7: Database for shell molded low alloy fully restrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error F F F F F B B B B B B F B B B F B F F F F B F B

237 215 D.1.8 Shell molded low alloy steel castings partially restrained features Table D-8: Database for shell molded low alloy unrestrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error G H D D D C D B D A D H H D D H

238 216 D.1.9 Shell molded low alloy steel castings unrestrained features Table D-9: Database for shell molded low alloy unrestrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error A A A A A A A A A A A A A A A E A A A A A A A A A A A E A A A A A

239 217 D.1.10 Shell molded high alloy steel castings fully restrained features Table D-10: Database for shell molded high alloy fully restrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error B B B B B B B B B F F D.1.1 Shell molded high alloy steel castings partially restrained features Table D-11: Database for shell molded high alloy partially restrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error D D D D D D H H H H

240 218 D.1.1 Shell molded high alloy steel castings unrestrained features Table D-12: Database for shell molded high alloy unrestrained features. Pattern Dimension (Inches) Casting Dimension (Inches) # Part Name Feature Name Feature Type Pattern Allowance Shrinkage Error A A A A A A A A A A A A A A A A A A A A A A A A A A A

241 219 D.2 Classified Database of features across the mold parting line D.2.1 Green sand molded low alloy steel castings Unrestrained Features Table D-13: Database for green sand molded low alloy unrestrained features across the parting line. Part Name Feature Name Feature Type Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance Shrinkage Error # A E E A A A A A A A A A A A A A D.2.2 Nobake sand molded low alloy steel castings fully restrained features

242 Table D-14: Database for nobake sand molded low alloy fully restrained features across the parting line. # Part Name Feature Name Feature Type Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance 220 Shrinkage Error 1 6NSL C F NSL C F NSL C F NSL C F CLG03 C F CLG03 C F CLG03 C F SY45AE E F SY45AE E F SY45AE E F SY45AE E F RT30 D F RT30 D F RT30 D F RT30 D F RT30 D F SY45AE E F NSL D F NSL D F NSL D F NSL D F NSL D F NSL D F D.2.3 Nobake sand molded low alloy steel castings partially restrained features

243 Table D-15: Database for nobake sand molded low alloy partially restrained features across the parting line. # Part Name Feature Name Feature Type Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance 221 Shrinkag e Error H H H C H D H H H H D.2.4 Nobake sand molded low alloy steel castings unrestrained features

244 Table D-16: Database for nobake sand molded low alloy unrestrained features across the parting line. # Part Name Feature Name Feature Type Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance 222 Shrinkag e Error A A A A A A A A A A D.2.5 Shell molded low alloy steel castings partially restrained features

245 Table D-17: Database for shell molded low alloy partially restrained features across the parting line. # Part Name Feature Name Feature Type Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance 223 Shrinkag e Error D H H H D D H D.2.6 Shell molded low alloy steel castings unrestrained features

246 Table D-18: Database for shell molded low alloy unrestrained features across the parting line. # Part Name Feature Name Feature Type Pattern Dimension (Inches) Casting Dimension (Inches) Pattern Allowance 224 Shrinkag e Error A A A A A A A D.2.7 Shell molded high alloy steel castings unrestrained features Table D-19: Database for shell molded high alloy unrestrained features across the parting line. Pattern Dimension (Inches) Casting Dimensio n (Inches) Pattern Allowanc e # Part Name Feature Name Featur e Type Shrinkag e Error A A A A A A A A A A A A A A

247 Appendix E Database Tables in Pattern Allowance Advisor software E.1 Alloytype Field name Data type Description txtalloy Text Type of alloy E.2 Calibration Field name Data type Description partnm Text Part Name dimnum Number Dimension Number dimrest Text Restraint Type pavalue Number recommended pattern dimension meapatdim Number measured pattern dimension meacastdim Number measured casting dimension Units Text Unit system error Number shrinkage error E.3 Custom Field Data type Description name ID Number ID Number Type Text Alloy and restraint type Readings Number Number of data points Shrink Number Standard shrink rate ErrorSum Number Summation of shrinkage error AvgError Number Median Error MinError Number 10th Percentile of shrinkage error MaxError Number 90th Percentile of shrinkage error AvgSize Number Average feature size

248 226 MinSize Number Minimum feature size MaxSize Number Maximum feature size E.4 Default Field Data type Description name ID Number ID Number Type Text Alloy and restraint type Readings Number Number of data points Shrink Number Standard shrink rate ErrorSum Number Summation of shrinkage error AvgError Number Median Error MinError Number 10th Percentile of shrinkage error MaxError Number 90th Percentile of shrinkage error AvgSize Number Average feature size MinSize Number Minimum feature size MaxSize Number Maximum feature size E.5 dimarchive Field name Data type Description ID Number Primary key value of dimensions table dimnum Text Dimension serial number dimval Number Casting dimension ucldbl Number Upper specific limit lcldbl Number Lower specific limit dimrest Text Restraint pamethod Text Pattern allowance method pavalue Number Pattern dimension partnm Text Casting part name rev Number Pattern revision number revdate Date Pattern revision date comments Memo Comments for the dimension

249 227 E.6 dimensions Field name Data type Description dimnum Text Dimension serial number dimval Number Casting dimension ucldbl Number Upper specific limit lcldbl Number Lower specific limit dimrest Text Restraint pamethod Text Pattern allowance method pavalue Number Pattern dimension partnm Text Casting part name rev Number Pattern revision number revdate Date Pattern revision date comments Memo Comments for the dimension E.7 inspectied and patinspected Field name Data type Description dimnum Text Serial number for dimensions mdim1 Number Measured dimension partnm Text Casting part name conformance Text Conformance for whether measured dimension is within tolerance date date Date of entry for measured casting patrevdate date Revision date of the pattern comments Memo Comment for measured dimension shell Number Average for shell process E.8 partinfo Field name Data type Description Partnm Text Casting part name Txtalloytype Text Type of alloy Txtprocess Text Type of process method Units Text Type of unit for the dimensions

250 228 E.9 patallowhist Field name Data type Description ID Number ID Number dimnum Number Dimension Number dimval Number Desired casting dimension dimrest Number Restraint Type coretype Text Core Type pavalue Number Pattern Dimension partnm Text Part Name rev Number Revision Number revdate Number Revision Date meacastdim Number measured casting dimension meapatdim Number measured pattern dimension E.10 processtype Field name Data type Description txtprocessmethod Text Type of process method E.11 rptinsp Field name Data type Description dimnum Text Dimension serial number dimval Number Casting dimension ucldbl Number Upper specific limit lcldbl Number Lower specific limit pavalue Number Pattern dimension mdim1 Number Measured dimension partnm Text Casting part name conformance Number Conformance for whether measured dimension is within tolerance date Date Inspection date comments Memo Comments for the dimension

251 229 E.12 userinfo Field name Data type Description Username Text Login name for the user Fullname Text Full name of the user department Text Department name employid Text Employee number pwd Text Password usertype Text Software user type

252 Appendix F Program steps for the Pattern Allowance Advisor F.1 Administrator module Administrator module provides access to various operations in the Pattern Allowance Advisor. F.1.1 Add User 1. Add user window is displayed. 2. Administrator types in username, password, department, and user type and clicks the Save button. 3. If all the required information is not entered, alert the user and go to If the user name is already in use, alert the user and go to Open the userinfo database table and add all the information about the new user. F.1.2 Modify / Delete User 1. Modify / delete user window is displayed. 2. The user selects the user name from the dropdown list. 3. If selected user is the current user, alert the user that changes can t be made to the current user and go to 2.

253 The database table userinfo is opened to display all the information about the user. 5. The user makes necessary changes and clicks delete or save button. 6. Selected changes are made in the table userinfo. F.1.3 Edit / Delete part Same as modify / delete user as described in section F.1.2. However, table partinfo is accessed instead of the table userinfo. F.2 New Part Module features. This module has two stages: new part creation and specification of casting F.2.1 New part creation stage 7. New part is created using create new part button 8. Part definition window is displayed. 9. The user enters part name, alloy type, unit system and molding method. Create new part button is clicked. 10. If any required information is missing, display alert and go to Save part name, alloy type, molding method, and units system into partinfo table. 12. Display dimensions entry window.

254 232 F.2.2 Dimension specification stage 1. All the available information for the selected part is displayed from database tables partinfo and dimensions. 2. After casting dimension and specification limits are specified, the user selects the feature type by clicking the appropriate button. 3. If casting dimension is missing, alert the user and go to Depending on the selection of dimension type, fill in the core type and restraint type columns for the current dimensions in the worksheet. 5. Read standard average, minimum, maximum feature size, number of data points, shrink rate, median, 10 th percentile, and 90 th percentile error values from custom and Default tables. Calculate average, minimum, and maximum predicted pattern allowance by PA = (Error + Shrink rate * Casting Dimension) / Casting Dimension. Calculate pattern dimension = Casting Dimension * 1 + PA. 6. If the user has selected Custom model, fill in average pattern dimension calculated using values from Custom model else fill in value from the database table Default in pattern dimension column. 7. Create default and custom model pictures showing the values and lines for average, minimum, and maximum pattern dimension. 8. If model details checkbox is unchecked go to Fill in values of average, maximum, and minimum feature size along with number of data points for both the models in the model details grid.

255 When the user clicks Save button, IF sent here from modification module, go to 3 in F.3 ELSE save all the data entered for the specification of the new part in dimensions database table. 11. Close this form and show the Pattern Allowance Advisor main screen F.3 Modification Module 1. The user selects the part name from the existing part list for modification. 2. Go to dimension specification stage section F If the dimensions are modified on the same date, go to step Archive all the existing dimensional data for this part in the database table dimarchive and generate a new revision number. 5. Save modified dimensional data in the database table dimensions. F.4 Inspection module Inspection module consists of casting inspection and pattern inspection sections. F.4.1 Casting dimension inspection 1. The user selects the part name and revision date for which casting inspection data is to be entered.

256 An SQL query is generated and dimensional data for the selected part is shown from the dimensions database table. Additionally, the current casting sample number for the inspected casting is displayed along with the number of times each feature was inspected. 3. As the user enters the inspected casting dimensions, conformance is checked and displayed. 4. When the user hits Save, the inspected castings number is incremented by one. Casting feature inspected times number is also incremented for all conforming features. These numbers are saved in dimensions database table. 5. All the inspected casting dimensions and conformance information is saved in the inspected database table along with inspection dates. Average values for casting dimensions of each inspected feature are also calculated and stored. 6. The user is asked whether more inspection data for the same casting is to be entered. If yes, go to The casting inspection form is closed and main form is displayed. F.4.2 Pattern inspection entry 1. The user selects the part name and revision date for which pattern inspection data is to be entered. 2. An SQL query is generated and dimensional data for the selected pattern is shown from the dimensions database table.

257 As the user enters the inspected pattern dimensions, conformance is checked and displayed. 4. When the user hits Save, all the inspected pattern dimensions and conformance information is saved in the patinspected database table along with inspection dates. F.5 History module history The history module has two sections: pattern history and pattern allowance F.5.1 Pattern history 1. The user selects the part name for which the pattern tooling history is to be viewed. 2. The database table partinfo is opened and the molding method, alloy type, and unit system are displayed. 3. The database table dimensions is opened and the revision date, print feature dimension, upper and lower control limits, restraint type and suggested pattern dimension for the selected part sorted by revision date are displayed. 4. The database tables patinspected and inspected are opened and average inspected casting and pattern dimensions are displayed.

258 236 F.5.2 Pattern allowance history 1. The user selects the alloy type, molding method, and unit system from the dropdown lists. 2. The user clicks the required dimension type. If the required user inputs are missing, alter the user and go to The database table dimensions is opened and the part name, dimension number, casting dimension, dimension type, recommended pattern dimension, and recommended pattern allowance in percentage is displayed. 4. The database tables patinspected and inspected are opened and actual observed pattern allowance for the selected feature type is displayed. F.6 Reports module Refer to the flowcharts and logic explanation for reports module in Section F.7 Calibration module and 6.5. Refer to calibration module flowchart and detailed logic explanation in Sections

259 Appendix G Pattern Allowance Advisor User Manual User Manual for Pattern Allowance Advisor Version 3.0 Developed by Penn State University Department of Industrial and Manufacturing Engineering Effective date April 23, 2005

260 238 G.1 Introduction This user manual guides the user on the effective use of the Pattern Allowance Advisor software. This user manual covers all eight features of the software, addressing each feature individually. 1. New part entry feature 2. Modification feature 3. Inspection feature 4. Tooling history feature 5. Report feature 6. Help feature 7. Administrator feature 8. Calibration feature G.1.1 Computer System Requirements: 1. Pentium PC with at least 256MB RAM and 50MB free space. 2. Microsoft windows 95/98/2000/ME/NT/XP operating system. 3. Microsoft Office 97 or newer installed on the PC. G.1.2 Additional Requirements to Support the Use of the PAA 1. Paper or electronic casting part drawings with numbered features. (The PAA does not store part drawings electronically.) 2. Pattern and casting feature measurements with adequate gage R&R.

261 239 G.2 Installation instructions G.2.1 Installing from a removable media Step1: Insert the removable media in the computer. Step 2: Go to My Computer and double click on the removable media Step3: Go to step 4 G.2.2 Installing from an ed or downloaded file Step 1: Create a folder called Dimeng on you C drive. Step 2: Save the attached PAadvisor.zip file on your computer in C:/Dimeng. Step 3: Go to My Computer, C:/Dimeng. Unzip the saved file. Go to step 4 Step 4: Double click on installation file PAadvisor**.exe to install. Step 5: Follow through the installation instructions. G.3 New Part Feature The new part feature is used for adding new casting part details to the database. The new part feature option is available in the main screen of the software as the New Pattern option. Following are the tasks in the new part feature: Enter new part details Enter feature dimensional details Enter new part details The user form for new part details can be accessed by clicking the New Pattern button. Following entering new part details are entered or selected in this screen:

262 240 Part name Alloy type Molding method Units system Enter dimensional details Dimensional details entry form is displayed after entering new part details and clicking Next >> option in the new part entry form. Dimension details entry screen is divided in 3 key areas indicated by numbered circles Figure 1: 1. Pattern dimension guide picture area 2. Feature type selection picture area 3. Casting and pattern dimension entry area Figure 1: Dimension details entry form

263 241 Figure 2 shows a zoomed screenshot of the casting and pattern dimension entry area. Only one row is available for entry and new rows are added using the Add new row option and any row can be deleted using the Delete row option. Pattern designer uses first three columns to enter desired casting dimension along with upper and lower control limits. Column four (dimension type) and five (core type) are filled in by software depending on the user selections in feature type selection picture. Column five (pattern dimension) is filled in automatically by the software after feature type is selected; however, the user can type in any other number in that column. Figure 2: Dimensions entry area Pattern designer has to specify several process and geometry variables such as molding method, core type, restraint type, and parting line location in order for the software to predict the pattern dimension from the casting dimension. In order to simplify feature type specification, a feature type selection picture is created as shown in. Pattern designer clicks different buttons to choose required feature types. Once the selection is made, dimension type and core type columns in the dimension entry areas are filled. Pattern Allowance Advisor then uses default PSU model and custom model (if available) to calculate average, minimum, and maximum pattern dimensions. Average pattern

264 242 dimension is filled in column six of the dimension entry table depending on the user s choice of the model. In order to account for the variability in the pattern dimension, the pattern dimension prediction model generates average, minimum, and maximum pattern dimensions when pattern designer selects the dimension type. Predicted pattern dimension values using default and custom models along with 1/4 and 5/16 shrink rule are then presented pictorially to the user as shown in Figure 4. Pattern designer can request further information about Penn State and Custom models by checking Model Details checkbox. In order to help the user choose between default and custom models, model details such as number of data points, average, minimum, and maximum feature size values from databases of both models are presented to the user. Figure 3: Feature type selection

265 243 Figure 4: Pattern dimension guide pictures Using custom and default model pattern dimension pictures and other details about these models, pattern designer estimates appropriate pattern dimension and enters it in the pattern dimension column (if different from average). Dimensional details are then added to the database using the Save option. After saving data to the database, the user is directed back to the main screen. G.4 Modification Feature The modification feature is used to modify the pattern dimensional details of an existing part in the database. Pattern dimension modification form can be accessed by selecting the Existing Pattern option in the main screen of Pattern Allowance Advisor, and then selecting the Modify Existing Patterns option. Following are the tasks in the modification feature: Select part name Modify pattern dimension specifications Add new row (if required) Delete row (if required) Comment on reasons of modifications Save changes

266 244 Selecting part name In this task, the user selects the part name from list box for which dimensional details are to be modified. A worksheet will be visible in the dimensions entry window after the user selects a part name. Modifying pattern dimension Any dimensional details of the selected part can be modified. The user can change pattern dimensions and record any changes in the comments column. The user can also change casting dimension, specification limits and dimension type. The user can add new dimensions by using the add new row option or remove any dimension using the delete row option. G.5 Inspection Feature the pattern. The inspection feature is used to enter the measured dimensions of the casting and

267 245 Figure 5: Casting inspection screen The form to enter measured casting dimensions task can be accessed by clicking the Enter Measured Dimensions of Casting option in the existing part main screen. The inspector enters some or all of the measure casting or pattern dimensions and then save them. Following are the different sub-tasks: Select part name Select revision date Enter measured casting dimensions Check for conformance Save After the part name is selected, all the feature dimensions of that part are shown in the screen as shown in Figure 5. A casting number generated automatically for each inspected casting and the number of times a particular feature was inspected is also

268 246 shown in this window. After entering the measured casting dimensions, the user can click in the conformance column for each row to check for conformance. The user can also enter comments for each dimension if required. To save entered dimensions, the Save button is clicked. The software then checks and displays the conformance of each feature before the data is saved. After saving the inspection data, the user is prompted to check if there is more inspection data to be entered for the same part number. G.6 History Feature Pattern tooling history The tooling history feature is for viewing the entire pattern history of a part. Tooling history feature can be accessed by clicking the Pattern History option in the existing part main screen. To view history of patterns, the user has to select the part name and the pattern tooling history for the selected part is displayed in a worksheet as shown in Figure 6. Pattern allowance history The pattern Allowance history form displays pattern allowance history of a particular dimension type. This task can be reached by selecting option Pattern Allowance History in main screen of existing part category. The user selects alloy type, molding method, and dimension type for which pattern allowance history has to be viewed. Once all the selections are made, Pattern Allowance Advisor shows historic pattern allowance data for that type of feature as shown in Figure 7.

269 247 Figure 6: Pattern history screen Figure 7: Pattern allowance history

270 248 G.7 Report Feature The report feature is for viewing and printing reports of the dimensional details, measured castings, measured patterns and the pattern history. Figure 8 shows the main screen for the report feature. The form for the report feature can be accessed by selecting the Reports option in the main screen of the Dimensional Engineering Software. The following are different tasks in this feature View and print current dimensional details View and print pattern history View and print measured castings View and print measured patterns View and print combined report of existing pattern and measured casting dimension

271 249 Figure 8: Report feature main screen View and print current dimensional details In this task dimensional details are viewed and printed if required. Figure 9 shows the form for this task, which can be accessed by selecting the Current Pattern Dimension option in the main screen of the report feature. The user selects the part name and clicks Get Report option to view the selected report. The report will be displayed in MS Word, as a word file. The user can print or save the report using menu options in MS Word software.

272 250 Figure 9: Report feature - Dimensional detail report View and print pattern history This task is for viewing and printing the history of patterns. Figure 10 shows the form for this task, which can be accessed by selecting the Pattern Tooling history option the main screen of the report feature. Figure 10: Report feature - Tooling history report View and print measured castings This task is for viewing and printing measured casting dimensions. Figure 11 shows the form for this task. After selecting the part name the user has to select the revision date of the casting from the list box. Then click the Get Report option.

273 251 Figure 11: Report feature Measured castings View and print measured patterns This task is for viewing and printing measured pattern dimensions. Figure 12 shows the form for this task. After selecting the part name the user has to select the revision date of the pattern from the list box. Then click the Get Report option. Figure 12: Report feature Measured patterns View and print combined report of existing pattern and measured casting dimension This task is for viewing and printing the existing pattern and measured casting dimensions. Figure 13 shows the form for this task. After selecting the part name

274 the user has to select the revision date of the casting from the list box. Then click the Get Report option. 252 Figure 13: Report feature Existing pattern and measured casting dimension View and print Pattern Allowance history report for one dimension type This task is for viewing and printing existing pattern allowances for a dimension type. Figure 14 shows the form for this task, which can be reached by selecting option Pattern Allowance History Report in main screen of Report category. After selecting alloy type and molding method, click on the dimension type button to view selected report. The report will be shown MS Word, as a word file. The user can print the report using print option in word. The user can save the MSWord report using save as option in the Word software

275 253 Figure 14: Report feature Pattern allowance history for one dimension type G.8 Help Feature Figure 15 shows the Pattern Allowance Advisor software main help screen. On the left hand side of the form, the categories list is displayed. The user can select a help topic from the topics list on the left hand side. Detailed topic information is displayed in the right hand side window. Navigation buttons on the top can also be used to navigate through available help topics.

276 254 Figure 15: Help feature G.9 Administrator Feature The administrator feature is for administrator use only. Only the administrator can access this feature of the software. Following are the list of the tasks in the administrator feature: Add user Delete user Edit user info Delete part Edit part details

277 255 Add user The add user task is to add a user to the software. Figure 16 shows the add user screen. Fields marked with * are necessary fields. All necessary fields must be filled. Following are the sub-tasks: Full name enter the name of the user being added User name enter the login Id for the user. This can be alphanumeric Password enter the password for the user to log into the software Department enter the department of the user Employee ID enter the user s employee ID User type select appropriate user type Figure 16: Administrator feature - Add user Delete user Figure 17 shows the delete user screen. Administrator selects the user name and clicks the Delete User option. An online user cannot be deleted.

278 256 Edit user info Figure 17: Administrator feature Delete user Figure 18 shows the screen for edit the user information. The administrator selects the user name and the user information is displayed. Required changes to the user details are then made and saved by clicking the Save button.

279 257 Figure 18: Administrator feature Edit user info Delete part This task is used to delete obsolete parts so that the database space can be freed. Figure 19 shows the delete part screen. Figure 19: Administrator feature Delete part

280 258 Edit part details Figure 20 shows the screen for edit part details. The administrator selects the part name from the list box and details of the selected part are displayed. Required changes to the part details are then made and saved by clicking the Save. Figure 20: Administrator feature Edit part details G.10 Calibration module The calibration module is only accessible by the administrator and is accessed through the administrator module calibration tab. Figure 21 shows the calibration tab in the administrator module. When the user clicks Calibrate Pattern Allowance Model button, the model calibration module as shown in Figure 22 is displayed. The number of data points in the default PSU model, custom model, and current database of the selected feature type are then displayed to the user. It is advisable to run the calibration module

281 only if the number of data points in the current database is substantially larger than the number of data points in custom model. 259 Figure 21: Calibration tab in administrator module Figure 22: Selection of feature type for calibration

282 260 When the user clicks on Calibrate button, the software checks if there is sufficient good data of the selected type in the database. If there is no sufficient good data, the user is alerted and the calibration module is closed. If there is sufficient good data, Pattern Allowance Advisor shows the results of calibration as shown in Figure 23. The user can accept or reject the calibration from looking at the calibration results. Following are the criteria for accepting new calibration: 1. The number of data points in the model after calibration should be higher than number of data points before calibration. 2. The error sum after calibration should be close to zero or smaller than that before calibration. 3. The median error should be very close to zero. Figure 23: Calibration module showing the results of calibration

283 261 If the new calibration meets all the above criteria, the user accepts the calibration and the custom model is updated with new numbers for shrink rule, median, 10 th percentile, 90 th percentile error, and error sum. G.11 Tutorials G.11.1 Tutorial 1: Installation of PA advisor 1. Start the installation process by double clicking installation file. An installation wizard will guide you through the installation. 2. After clicking Next, the default installation directory - C:Program Files/PA Advisor is displayed. If you wish to change the directory, you can do so at this stage. 3. Click Next. If you do not have a directory named PA Advisor, you will get following message Destination Directory doesn t exist. Do you want it to be created? Click YES. 4. A verification message is displayed. Click the START button in verification window. 5. The user manual can be viewed by clicking the View PAA user manual button. 6. If you do not want to run the PAA program as soon as the installation is complete, uncheck the Launch Pattern Allowance Advisor checkbox. 7. After clicking Next, installation program information is displayed. This information can be ignored. 8. Click the EXIT button. 9. Following message will be displayed: Pattern Allowance Advisor installation is now complete. Administrator ID must be created first. Click OK. 10. Create administrator screen is displayed. Type Administrator ID and Password. Select an administrator ID and password that is easy to remember. Please note down this information. 11. Click the Create Administrator button.

284 262 G.11.2 Tutorial 2: Administrative options 1. Click the Administrator Tools button from the Main Menu or the Admin Options button from the PAA toolbar to access the administrative options. 2. Enter userid and password along with other information as shown in Figure 25. Choose Administrator user type. Click Add User. Wait for confirmation.

285 263 Figure NOT TO SCALE NOT CASTABLE FEATURES DO NOT CROSS MOLD PARTING LINE Figure 25 G.11.3 Tutorial 3: Create A New Part File 1. Click the New Part button from Main Menu or toolbar to create a new part 2. Type part name or number. For this tutorial we will use part name Tutorial3. Choose Units Inches, Alloy Type Low Alloy and Molding Process Green Sand as shown in Figure 26. Click Next. Dimension entry window is now shown. Refer to the appropriate part drawing with numbered features. For this tutorial please refer to Figure 25

286 264 Figure Enter the casting dimension in the Casting dimension box. For Tutorial3 Feature 1, enter as casting dimension. 4. You may also enter Upper and Lower control limit dimensions in UCL and LCL boxes where necessary. Feature # Casting Dimension (Inches) Dimension Type Table 1 Core Restraint Description C Fully restrained Mold to mold across mold A Unrestrained Mold to mold across casting G Nobake Fully restrained Mold to mold across core G Nobake Fully restrained Mold to mold across core A Unrestrained Mold to mold across casting 6 E Partially restrained Mold to mold across mold/core and casting C Fully restrained Mold to mold across mold 5. Click dimension type button C (refer Table 1) in the dimension types figure on the right. The Pattern Allowance Advisor picture in upper right hand corner suggests minimum, maximum and average pattern dimensions. The average recommended pattern dimension is automatically filled in Pattern dimension column. Any other pattern dimension value can be entered manually instead of the average recommended pattern dimension value. 6. Click Add New Row button.

287 Repeat Steps 3 to 6 to add dimension number Enter in Casting Dimension box for feature number Enter UCL and LCL. 10. Because this is a core related features, you must first click on the Dimensions related to core tab in the dimension types figure window. 11. Check the No Bake core type option. 12. Select dimension type G by clicking the G button in dimension types figure. 13. Repeat this procedure to enter all the remaining features. 14. When all dimensions have been entered, save your work by clicking the Save button. Wait for confirmation. Selecting Appropriate Pattern Dimension: Pattern dimension column shows the average pattern dimension recommended by Pattern Allowance Advisor. Any other pattern dimension value can be typed in this column. 1/4 and 5/16 in/ft pattern allowance pattern dimensions as well as recommended minimum and maximum pattern dimensions from the Penn State database are shown in pattern dimension advisor picture for reference. In addition, pattern dimensions calculated using the custom pattern allowance prediction model are also shown Note: Pattern allowance recommendations depend on casting feature size. For small feature sizes in particular, large variation in pattern allowances can be expected. The effect of feature size on the suggested pattern allowance can be observed in pattern allowance advisor picture by clicking on the pattern dimension cells for similar features such as feature 1 and 7. G.11.4 Tutorial 4: Modifying an existing part 1. Click Existing Part in the Main Menu or the Toolbar of the pattern allowance advisor. 2. Click Modify Existing Pattern to open the Modify existing part details window. Any dimensional details can be changed using this screen. 3. Select Tutorial3 in the part name drop down list. All the dimensional details for Tutorial3 are automatically displayed. 4. Click Add new row and enter 5 in casting dimension box in row number Select Dimensions related to core tab and select No Bake core type. 6. Click button L to specify dimension type for dimension number No pattern Allowance is recommended. Please choose your own message will be displayed. Click OK. Type your own pattern dimension in pattern dimension box. 8. Click Save button to save your work. Wait for confirmation.

288 266 Figure 27: Dimension modification screen G.11.5 Tutorial 5: Entering Measured Casting and Pattern Dimensions Entering measured casting and pattern dimension is important for dimensional recordkeeping of castings and for adaptive pattern allowance learning from the data. 1. Click Existing Part in the Main Menu or the Toolbar of the pattern allowance advisor. 2. Click Enter Measured Casting Dimensions to open the Casting Inspection entry window. 3. Select Tutorial3 in the part name drop down list. All feature dimensions along with UCL and LCL for Tutorial3 are displayed. 4. Enter measured casting dimensions from casting inspection after all post processing in the measured casting dimension column. 5. Click on conformance box to check for the conformance. 6. Any comments can be entered in comments column.

289 Save your work. Wait for confirmation. 8. You will receive following message Do you want to enter dimensions for another casting made from the same pattern? Click YES. Note change in casting number and number of times this feature was measured. Figure 28 G.11.6 Tutorial 6: Reviewing pattern history Keeping track of pattern dimensional reengineering is very important to avoid wasteful dimensional reengineering for similar features on other castings. This form in the software shows the entire pattern history for a particular part. 1. Click Existing Part in the main screen of pattern allowance advisor. 2. Click Pattern History to open the Tooling Pattern History window. 3. Select NittanyLion in the part name drop down list. 4. All the pattern history along with revision dates, revision numbers, measured pattern dimensions and measured casting dimensions is displayed.

290 268 Figure 29 Note: Casting dimensions are sorted by revision numbers. Revision numbers are generated automatically if the part modification is made on a date other than previous modification. G.11.7 Tutorial 7: Reviewing pattern allowance history Observed pattern allowance data for a particular type of feature in a specific foundry can be used to help pattern designer in assigning pattern allowances for similar casting features. Pattern allowance history can be viewed for a specific feature type as follows. 1. Click Existing Part in the Main Menu or the Toolbar of the pattern allowance advisor. 2. Click Pattern Allowance History.

291 Select Low Alloy in Alloy Type drop down list and GreenSand in Molding Method drop down list. Select Inches in Units selection. 4. Click dimension type C button in the adjoining picture. Suggested and actual pattern allowances for type C features are shown along with part numbers, dimension numbers and other details sorted by casting feature length. 5. Click Create Report button to display the report in Microsoft word format. Wait till Microsoft Word opens this report 6. Click on Microsoft Word window to view the report. 7. Save the report using Save As in File menu in Microsoft Word and give the report appropriate name. Figure 30 G.11.8 Tutorial 8: Generating Reports Pattern Allowance Advisor can generate various reports for viewing and printing. 1. Click Reports in the main screen of pattern allowance advisor. 2. Click the Recommended Pattern Dimension button. Existing pattern dimension window is opened.

292 Select Tutorial3 in the part name drop down list. 4. Click the Get Report button. Wait till Microsoft Word opens this report. 5. Click on Microsoft Word window to view the report. 6. Save the report using Save As in File menu in Microsoft Word and give the report appropriate name. G.12 Contact Information Mandar Deo 310 Leonard Building Penn State University University Park PA Phone No.: (814) Dr. Robert Voigt 310 Leonard Building Penn State University University Park PA Phone No.: (814)