ABSTRACT AN DWESTIGATION TO DETERMINE THE RELIABILITY OF THE BRINELL TEST AND SOME NONDESTRUCTIVE TESTS IN QUALITY TESTING CRAY IRON GASTJNGS by Jams ‘1‘. Webster Demands by the customers of foundries for tighter quality standards due to increased warranty periods coupled with typical problems associated with producing gray iron castings often results in Brinell testing 100 percent of the castings. For pesent production line operations, 100 percent Brinell testing is a rather inefficient process. These problems pranpted an effort to find a more efficient quality testing process. The first step was an investigd: ion to determine the reliability :1 the Brinell test and sane nondestructive tests in quality testing grey irm castings. The nondestructive tests investigated were the resonance test, static magnetic tests, and eddy current tests. First, the theoretical aspects of each test were investigated. No fundamental law which can be expressed mathematically independently of the measuring process has been developed for hardness. A theoretical relationship does exist which relates resonant frequency to modulus of elasticity. The modulus depends upon the graphite phase in gray iron. Static magnetic properties, magnetic retentivity and coercive fcrce, are definite points on the cyclic, direct current, meteresis curve obtained by saturating a ferrcmagnetic sample. Coercive fcrce is independent of mass and shape. m current distributions and alternating magnetic field distributions CE 51 m 1.1 t« James T. Webster can be predicted frcm mathematical considerations for test samples of simple gemetrical shape. A number of test bars and as cast parts of the same type were furnished by Central Foundry fcr conducting nondestructive tests. Several Brinell impressions were taken over the surface of the test bars and cast parts and averaged. In order to evaluate each nondestructive test, tensile strength, carbon equivalent, and mierostructures were determined for the test bars. Olly Brinell impressions were taken for evaluation of the nondestructive tests on the as cast parts. The nondestructive tests were classified as resonant frequency tests, static magnetic tests, and eddy current tests. Resonant frequency tests were cmducted with Magnaflmc's SR-lOO tester. Static magnetic field tests were oondmted with a Foerster Coercive Fa-ce Meter. Edcv current tests were conducted with Magnaflux's BID-300 Eddy cit-rent Probe Tester md Foerster-Hoover's QC—lOOO Comparator. The Retentivity test and the I-300 Eddy Current tests were eliminated from consideration due to inaccuracies and duplication of the other tests. The the magnetic test cmsidered was the coercive fcrce test, and the edcw current test considered was the QC-IOOO Conparator test. It was concluded that the Brinell test is not a sufficient quality test for gray iron. The resonant frequency test evaluates the graphite phase in gray irm independently of the matrix. The coercive farce test indicates the amount a? ferrite in tl'e matrix independently of the graphite phase, mass, and shape of a test sample. The QC-lOOO Gasparator evaluates tensile strength which is dependent upon both the graphite phase and the metallic matrix (1' a test sample. James T. Webster It is recommended that further laboratory investigation be performed, that application be nude to foundry production, and that establishnent of new standards be initiated. Further laboratory investigation should include investigation to determine the affect of a range of! ferrite contents, a range of pearlite coarseness ,and a range of temperature upon coercive force. The affect of temperature, mass, and shape variations on oc-looo test indications should be investigated. For the resonance test, methods of support, methods of inducing resonance, and the affect of mass and shape variations shoild be investigated. Additional wcrk is required for the design of test systems for particular applications in production testing. Production applications include test set-ups for annealing control and foundry control, and a test set-up for final inspection of the quality of castings before being shipped to customers. Finally, new standards must be set in terms of the improved quality test both in the foundry and at the customer. AN INVESTIGATION TO DETERIENE THE RELIABILITY OF THE EIJNELL TEST AND SOME NONDESTRUCTIVE TESTS IN QIALITY TESTING (RAY IRGW CASTINGS By James Thomas‘webster A THESIS submitted.to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Mechanical Engineering 19614 PREFACE AND ACKNWLEDGWTS This thesis is being submitted in partial fulfillment of the requlrunents for a Master of Science degree in Mechanical Eugineering at Michigan State University and for a Bachelor of Mechanical Engineering degree at General Motors Institute. Under the Bachelor-Master Plan of General Motors Institute, I spent the fifth year working towards a Master of Science degree. The thesis I wrote will serve to canplete the requirements of any graduate work, as well as my undergraduate work. My sponsoring unit, Central Foundry Division, Danville Plant, assigned me the problem to design a mechanized hardness test for gray iron castings. This problem was broken down into two parts: investigation (1' the Brinell test and some nondestructive tests; and the design of test set-ups for specific applications in the foundry. My thesis involves mostly the first part of the problan, since I felt that this part necessarily had to be done first and its scope was sufficient for a thesis. Because of the specialised equipnent required for my investiga- tion, I had to borrow several pieces of equipment from several companies. In practically all the cases, I transported nw test samples to the canpanies, since the equipment involved was in almost continuous use. I am greatly appreciative to the following people and canpanies: Senior Metallurgical Engineer C. L. Langenberg and Experimental Metallurgist L. D. Tankersley of Oldsmobile Division, Plant #1, General Motors Ccrporation. Mr. Art Lindgren, Mr. James Priore, and Mr. J. Baranowski of Magnaflux Corporation. Research Engineers M. J. Diamond and Bob Lut ch of Central Foundry Division, General Motors Corporation. Mr. V. L. Welsh of Foerster-Hoover Electronics, Inc. I am also greatly appreciative to Central Foundry, Danville Plant, for supplying me with the test bars and cast parts necessary for this investigation. Finally, I am deeply indebted to Dr. H. L. Vomochel, Professor (1' Metallurg, in the Metallurgy, Mechanics, and Materials Science Department at Michigan State University for his guidance. -iii- TABLE OF CQ‘I TENTS PREFACE AND ACIQQWLEDGE‘IENTS . . LIST OF ILLUSTRATIONS . . . . . INTRODUCTION . . . . . CHAPTER I. II. III . TI-IE BleEI‘I. mm 0 O O O O 0 O O O O O O O O 0 Theory and Factors Concerning Brinell Testing Preliminary Testing Procedure and Microstructure Brinell Hardness as Related to Physical Properties TBTBilBTCStSe e e e o e o e o 0 Some Foundry Variables Machinability and Microstructure Evaluation of the Brinell Test . . . . THERESONANCETEST.......... Theory and Factors in Resonance Testing and Microstructure . Elastic Modulus . . Tensile Strength . Brin ell Hardness . Microstructure . . Evaluation of the Resonanc STATIC MAGNETIC TESTS AND EDDY (IIRRENT 0 O O O O O O 0 C 0 O O 0 O O 0 O O O O O 0 O O C 6 Test Resonant Frequency as Related to Physical Properties 0 O O O O 0 Principles in Static Magnetic Testing and Eddy Current Testing . . . Static Magnetic Tests Eddy Current Tests Testing Procedure Eddy Current Tests ‘ Physical Propertiiw and Microstructure Tensile Strength . . Brinell Hardness . Carbon Equivalent . Microstructure . . Static Magnetic. Tests -iv- . 0 O 0 O O O 0 0 O O O 0 O O 0 O O O 0 O O O O O O O O O O O O O O 0 00000000... Page ii vi 29 29 30 32 33 3h 37 38 39 b0 CHA CCH vy- -E APE APE TABLE OF CONTENTS (Con't.) CHAPTER Evaluation of Static Magnetic Tests and Eddy CurrentTests.............. CONCLUSIONS RECWIIWS O O O O O O 0 O O 0 Further Investigation . . . Production Line Application Control Device . . . . Final Inspection Device Establishing Standards . . . BIBLIOM O O O O O O O O O O O 0 APPENDIX A O O O O O O O O O O O O 0 ”Pm-DUB. O. O O O O O O .0 O 0. -v- Page h2 us 1:5 1:6 147 1:9 50 51 9O FIGURE 10 LIST or ILLUSTRATIONS TITLE Ch'ayIrOnTestBarS................ Part No. 5692885 Cast Part from Central Foundry MV181m, Baum Plant 0 C O O O C O O O O O O Brinell Hardness Number Vs. Diameter of Indentation Brinell Diameter Vs. Position on Casting for Six Sample Castings, Part No. 5692885 . . . . . may Irm Tangile M O O O O O O O O O 0 O O O O O Brinell Diameter Vs. Tensile Strength . . . . . . . Tensile Strength Vs. Brinell Hardness Nmnber for About 1500 Gray Iron Tensile Bars . . . . . . . Brinell Diameter Vs. Carbon Equivalent fm' Some MCastTestBarS...e........... Sample 1A1 at Center Showing Graphite Size and meeeeeeeeeeeeeeeeeeeeeee Sample 3A1 at Center'Showing Graphite Size and .Typeeeeeeeee-eeeeeeeeeeeeee Sample 2A3 Showing Surface Grahpite and Ferrite . . Sample 3A3 at Centa‘ Showing Ferrite and Pearlite . Sample 2A3 Showing Pearlite at Center . . . . . . . .Typical Pearlite and Steadite Near Center . . . . . FundamentalFlexureMode ............. Fundamental Longitudinal Mode . . . . . . . . . . . FuhdamentachrsionalMode . . . . . . . . . . . . FundamentalDianeterMode............. Fundamental Flexural Mode for a Thin Disc . . . . . FundamentalRadialMode.............. PAGE 52 53 Sh 55 56 S7 58 59 60 61 62 63 6h 65 66 67 67 67 21 22 23 2h 25 26 27 28 29 30 31 32 33 3h 35 37 38 39 to AccmOdatOd Stress-Strain Diagram e e e e e e e e e Schematic Diagram of SR-lOO Resonance Tester . . . Test Set Up for Recording Stress-Strain Cm-ve in Order to Detemine Static Elastic Momlus . . . Tensile Strength Vs. Resonant Frequency for Five ASCaStTeS‘bBarS............... Brinell Diameter Vs. Rescmant Frequency . . . . . . Average Brinell Dimneter Vs. Resonant Frequency fa' sale ‘3 Cast! Part3, Part NO. 5692885 e e e e e e Thermal Arrest Carbon Equivalent Vs. Rosmant Frequency for Some As Cast Test Bus . . . . . . Wet Ca'bon Carbon Eq1ivalent Vs. Resonant Frequency for Sane As Cast Test Bars . . . . . . Direct Ctrrent Magnetization Curve . . . . . . . . Probe Coil and Test Sample Arrangements . . . . . . Schematic Diagram of Coercive Force Meter . . . . . Foerster Magnatest Q Canparator: A Linear Time-Base Iratrunent Similar to The QC-lOOO Comparator . . Circuit Diagram of ED-SOO (Similar to ED-SOO) . . . Relative Magnetic Retentivity Vs. Coercive Force . Surface Indications Vs. Center Indications per Sample fa'ED-BOO................... QC-lOOO Comparator Indications Vs. ED-3OO Center Indications.................. Tensile Strength Vs. QC-lOOO Scope Deflection for mnsamPJ-eseeeeeeeeeeeeeeeee Tensile Strength Vs. Coercive Force . . . . . . . . Brinell Diameter Vs. QC-lOOO Test Indications . . . 0 Ave. Brinell Hardness Diameter Vs. QC-lOOO Canparator Indications (Part No. 5692885-As Cast Part) . . Brinell Dianeter Vs. Coercive Force . . . . . . . . Analysis of Tensile Strength Vs. Coercive Force . . -—i_Q I 68 69 7O 72 73 7h 75 76 77 78 79 80 81 82 83 8h 85 86 87 88 89 ’43 Possible Locations for Nondestructive waltjey TeSting O O C O C C C 0 O C O O O . O O O O C O 92 M; Method of Inducing Resonance by Automatic Excitation . . . 95 ’45 Synthesis of a Foundry Control Test . . . . . . . . . . . 97 -viii- APPHTDIX A B LIST OF APPENDICES IlluS'bratiOmandGraphS......eee.... Test Set-Up for Foundry Cmtrol . . . . . . . . . PAGE 51 9o INTRODUCTION As a result of increasing the warranty periods on General Motors automobiles, quality standards are being tightened in all phases of autonobile production. This thesis concerns only the narrow aspect of testing gray cast iron parts for quality. Tensile strength, wear resistance, damping capacity, and machinability of parts must be uni- form snd within specifications to insure a high degree of reliability and low production costs. These properties of gray iron are a function of the amount and nature of the graphite phase and the nature of the metallic matrix. In high production founding besides the temperature and chemistry cmtrols, the inoculation of the molten iron before it is poured and the cooling rate of the casting after it is poured are critical. The degree of inoculation and rate of solidification affect the nature and amount of the paphite phase. Since castings are "shaken out" of their molds above 1330 degrees F., the lower austenite transformation range, the rate of cooling through this transformation range governs the nature of the metallic matrix. Central Foundry Division, Danville Plant, at present has no positive control over the rate of cooling of its castings through this very critical transformation range.~ As a result of these and other variables, the structure of gray iron can fluctuate widely. -1- -2- With these existing production problems and requirements of increased reliability, sane parts mist be 100 percent quality tested. The most widely used test for quality of gray iron parts is the Brinell Hardness test. With conventional Brinell machines, a consider- able amount of labor is expended when testing production castings 100 percent. In addition, only a small local area is tested and the area tested is very close to the surface of the casting. Since the "skin condition"1 from casting to casting is variable and since the customer sanetimes machines this "skin" completely off, Brinell hardness readings can be misleading as to the quality of the part. Furtha'more, very small parts which met remain uretressed Q‘ which may be damaged by the Brinell impression, cannot be advantageously tested with the Brinell method. These factors led to the current problem of investigating the Brinell test and other nondestructive tests for gray iron. A survey of the field of nondestructive testing has shown that at present magnetic and sonic methods offer a good potential fcr development. Mr. Charles Walton has said that "the damping of sonic vibration is closely related to the graphitic phase in iron as is the resonant frequency (modulus of elasticity at low stress)" and that "magnetic properties are more clmely related to the silicon content and the matrix structure whereas Brinell hardness is influenced by both the matrix structure and the graphite phase."2 L'Skin condition" is the existence cf a different matrix structure and gaphite type and distribution found at cr near the sur- face than is representative of the casting asa whole. 2Fran a letter from Mr. Charles Walton, Technical Director of the Gray Iron Founder's Society, dated July 6, 19611. -3- Other methods considered were ate-ray and nuclear techniques, infrared analysis, and dynamic hardness tests, such as Shore Sclero- scOpe tests. Jones and Laughlin of Aliquippa, Pennsylvania have developed a device using Beta rays to detect underannealing of a continuous low carbon steel strip.3 At Hennecott Research Center in Salt Lake City, Utah, infared analysis has proven to be "quick and easy," but instrumentation is expensive.h Since vibrations within a molecule are related to the frequencies absorbed, no two canpounds give exactly the same pattern when transmission of radiation is plotted against wave length. Also "the depth of an individual absorption band can be related to the concentration of material responsible ftr it."5 These methods should be fwther investigated to determine whether they can be applied to cast metals containing the yaphitic phase. In order to investigate the sonic and magnetic properties of gray iron, thirty transverse test bars and twenty-six cast parts, Part No. 5692885 pump housings, were obtained from Central Foundry Division, Danville Plant. The cast parts were castings which had to be 100 percent quality tested before shipping them to the customer. Six of the test bars were annealed. For the test bars, Brinell hard- nesses, resonant frequencies, residual induction values, coercive force values, eddy current indications with both a coil and a probe, and tensile strengths were détermined. Only Brinell hardnesses, ——-— fir 3"Hardness Caged 'On The Fly'," ghe Iron 453, (0x011, Sept. 19, 1963), 108-9. b‘l‘uddenham, w., and Yimmerly, s. 11., "Infared Analysis is Quick and Easy," Engineering and_Mining Journal, (July 1960), pp.92-h. SIbid. -u- resonant frequencies, and eddy current indications were obtained for the cast parts. Fimlly, a thorough microstructural analysis was made on eleven of the test bars. With this information, an extensive analysis of the Brinell test and other nondestructive tests was made. The investigation of this problem will be reported in the chapters to come as follows: the Brinell test, the resonance test, static magnetic tests and eddy current tests, conclusions and recomendati ons . CHAPTER I THE BRINELL TEST Brinell hardness testing was developed around 1900 by Dr. J. A. Brinell. Essentially the method Ins remained unchanged since then. As mentioned previously, this method of determining the quality of gray iron parts is standard in General Motors Corporation. This Chapter discusses theory and factors concerning Brinell testing, pre- liminary testing procedure, and the relationship of Brinell hardness to physical properties and microstructure. Theory and Factors Cmceyhgkinell Testing For Brinell testing of gay iron, a ten mm. steel ball and 3000 kg. load must be used because of the inherent inhomogeneity of the material. Brinell Hardness number is expressed by the formula BHN - l_ P l where P is the load in kilograms, D the diameter 1-32 (D JFK—d5) of the ball, and d the diameter of the impression in millimeters. The Brinell Hardness Number is then equal to the applied load divided by the contact area of recovered indentation.6 The permanent impression produced is dependent upon several factors including yield strength, _.4_ A 6Samuel R. Williams, Hardness and Hardness Measurements, (Cleveland: American Society for s, 9 , pp. 3- . -5. -6... ability of the material to flow, the amount and distribution of the gaphite near the impression, and the ability of the material to work- harden upon defamation. It should be noted that the word "hardness" by itself has no concise physical meaning. As S. R. Williams7 points out there are no reliable conversion tables connecting even SOO-kilogam Brinell hardness with 3000-kilogram Brinell hardness. This means that the Brinell hardness number is also dependent upon the size of the penetrating ball and upon the load when testing the same material which is uniform in its physical properties. This combination of applied load and ball diameter have a different effect upon the material in question, such as its effect upon the degree cf cold-working. This is also true when trying to correlate say Vickers or Rockwell hardness with Brinell hardness. Different shaped penetrators thus have an effect upon the hardness value, given the same homogeneous material. Therefore, the "hardness" of a material must be qualified by the method used. As of yet, no fundamental mathematical relationship has been developed for hardness. In Brinell testing and other static indentation tests, the elastic limit of the material is exceeded. In dynamic tests if the elastic limit is exceeded, the rate of deformation plays an important part. If the elastic limit is not exceeded, the hardness number is dependent upon the elastic modulus. Is then hardness a fundamental physical property? If so a fundamental mathematical relationship 7Ibid., 8. s7- applicable to all materials, could be developed. Another factcr in Brinell testing is reading of the Brinell scope. It has been found that human error can be an important factor in Brinell scope reading. At best in high production an accuracy of 1 .05 m. diameter is obtainable with the conventional scope, which corresponds to a nominal accuracy of 1 S in Brinell Hardness Number. In the laboratory, however, better accuracy can be obtained when taking several readings (interpolating to the second decimal place) and averaging these values. Finally, still another factor in Brinell testing is section size. Usually in production Brinell testing, only one Brinell reading is made on a casting to determine if it is acceptable. In crder to investigate the hardness variation over the surface cf a complex cast part with variations in section size, Brinell hardnesses were made at five different locations on the cast parts obtained fran Central Foundry. In Figure 2 in the Appendix, position (1) represents the single position where the part is production Brinell tested, and positions (2) through (6) represent the locations where Brinell hardnesses were taken for laboratcry investigation. The Brinell dia- meters at these different locations for six of these cast parts were plotted in Figure )4 in the Appendix. It can be seen in Figure 14 that quite a variation in Brinell hardness can be obtained over the casting surface. Thus it is not sufficient to represent the hardness of the whole casting by the hardness at one location. -8=- Prel’ginarLTestingProcedure In order to insure varying microstructures among the thirty test bars which were made frcn a standard transverse test bar pattern, three carbon equivalent8 ranges were specified: 3.70 to h.12 percent, b.13 to h.22 percent, and 11.23 to b.60 percent. A group of ten castings were poured in each carbon equivalent range. The first goup with the lowest carbon equivalent range had no inoculant added before pouring. The other two groups had one pound of 912 (Silicm-Manganese-Zirconimn) inoculant added to 800 pounds of iron just before pouring. Also, two castings fran each goup were cooled for forty minutes in their green sand molds. The rest of the castings cooled eleven minutes in their molds and cooled to room temperature on a steel bench. Two castings from each group were annealed at a temperature of 1300 degees F. in air for three hours and then allowed to cool in the furnace with the door partially open for twelve hours. Carbon equivalent was determined initially with a thermal arrest unit ,9 a device that detects the freezing temperature of the iron which is related to carbon equivalent. Pouring temperature and chill depth were taken at the time of pouring. Later carbon, silicon, 8Carbon equivalent is given by the following formula: on - C + 1/3 (Si 4- P). In this study, phoSphorous content was not considered. 9Milton J. Diamond, "A Summary of Some New Processing and Quality Control Developments in Foundry Technology," General Motors Engineering Journal, (XI, Second Quarter, 1961;), p. 26. -9- manganese, and chromium content were determined.by conventional chemical methods. It was thought that differences in inoculaticn would give the largest possible graphite variation and.that longer cooling in. the mold aid annealing would produce more ferrite in the matrix structure. Four Brinell tests were taken on each test bar. The resulting average Brinell diameters varied from h.0 to h.SS mm. (229 to 17h)10 for the as cast bars and 5.3h to 5.57 mm. (12h to 112) for the annealed bars. (See Figure l in the Appendix for location of Brinell tests) It would have been desirable to have more test bars in the range of h.55 to 5.3h mm. diameter (17h to 12h) but the time required.to select and to anneal these test bars would have been too great. As described earlier, the twenty-six cast production parts fran Central Foundry were Brinell tested here in the laboratory at locations (2) through (6) in Figure 2 in the Appendix. The average Brinell diameter of these cast parts varied fran 3.88 to h.83 mm. (2hh to 15h). The range acceptable to the customer was h.0 to h.7 mm. diameter in position (1) on the casting (See Figure 2). Since Central Foundry's customers normally specify Brinell diameter rather than Brinell Hardness Number, Brinell diameter has been referred to in this paper more than to Brinell Hardness Nunber. There is a possibility of conversion errors when cmverting frcm Brinell diameter to Brinell Hardness Number to Brinell diameter. Figure 3 shows the relationship 10Brinell Hardness Numbers are contained in parenthesis after the Brinell diameter. -iom between Brinell Hardness Number and Brinell diameter. Brinell Hardness as Related to Physical Properties and Microstruct_u£e This section will discuss how Brinell hardness is related to tax sile strength, some foundry variables, machinability, and microstruc- ture. Tensile Tests. Tensile strengh is influenced by both the amount and nature of the gaphite and by the nature of the matrix, as is Brinell hardness. Accuracy of tensile testing gay iron is quite dependent upon the variation in physical factors of the test specimen since gay iron is in fact a brittle material. Such factors as stress raisers produced by machining, geometry, and ary bending during the test will affect the accuracy of the test. Figure 5 shows the standard specimen med. Because of various factors in testing and machining, it was felt that at most only seven out of the eleven desired tests gave accurate results. Figure 6 shows a plot of Brinell diameter versus tensile strength fcr seven specimen. This Figure shows a general correlation of Brinell diameter and tensile strength. Figure 7 which shows tensile strength versus Brinell Hardness Number for 11495 samplesn' shows a considerable spread particularly at higher values of Brinell Hardness. The converging of the upper and lower limits seems significant. Since the Spread of this curve is so llMetals Handbook, ed. Taylor lyman (Metals Park, Ohio: American Society for Metas, I§6I), I, p. 35h. -11, great, the Brinell test is not a reliable indication of tensile strength. Some Falndry Variables. Often Brinell hardness is used as a control in the foundry operation. Chemistry, pouring temperature, inoculation, cooling rate in the mold, and cooling rate after shake-out, all affect Brinell hardness. All of these variables must be controlled in order to control Brinell Hardness. Among the variables examined here are pouring temperature, carbon equivalent, and cooling time in the mold. No correlation could be established concerning the relationship of Brinell hardness and pouring temperature, but all other variables being constant an increase in pouring temperature up to 2725 F. should result in a decrease in Brinell hardness.12 Comparing as cast parts, a much better correlation was obtained between Brinell diameter and carbon equivalent determined by the Thermal arrest unit than with the carbon equivalent determined by the wet chemical method (see Figure 8). No appreciable hardness change resulted when leaving the castings in their molds for forty minutes over the castings cooled for eleven minutes in their molds. Hachinabilitx and Microstructure. Machinability and micro- structure were examined to determine how they are related to Brinell Hardness. Since microstructure tends to vary fran normal to abnormal from the center to the surface and tends to vary scmewhat from point to point, microstructures that were compared were taken at the same distance 12Dr. Dimitri Kececioglu, "Factors Affecting Gray Iron Machin- ability," Foungz (XCI, October 1963), p. 115. alga from the surface and were representative of the structure as a whole at a given depth. Machinability is not an absolute property but a relative one. It is difficult to measure and, like the Brinell test, depends upon the method of measurement and has no fundamental mathematical derivation. Brinell Hardness was related to a machinability index in a series of tests conducted by Dario Fortino of Fiat.13 In these tests a constant-pressure drilling machine was used. The machinability index was determined from a correlation of the time required to drill through a certain thickness of UNI AB hOP steel with that required to drill tIr ough the same thickness of cast iron. A machinability index of 100 was assigned to the steel. It was concluded by the Italian investigator that Brinell hardness permits only an approximate estimate of machin- ability: the lower the Brinell hardness the better the machinability. Upon examining why two castings leving the same Brinell Hardness Number had different machinability indexes, Mr. Fortino found that gaphite in the first casting was mostly of type SA with six to twelve mm. flake length, whereas that in the second casting was of type 1m with twelve to twenty-four mm. flake length. To determine wlnt change in gaphite form and structure results in a change in hardness, six of the thirty test bars were annealed to such an extent that all the pearlite was converted to ferrite. This 13WMachinability of Iron Castings ,‘9 Foundry Trade Journal, (CXII, December 13, 1962), pp. 729-35. — =15- eliminated the matrix as a variable!“ (This heat treatment was explained under "Preliminary Testing Procedure" in this Chapter.) Comparing Figures 9 and 10, it can be seen that the shorter and abnormal gaphite results in a higher Brinell hardness than does the longer randomly oriented flakes. The Brinell varied from 5.146 to 5.80 mm. (118 to approx. 100). The direct influence of the matrix could not be determined in the as cast samples because pearlite was about the same from sample to sample. When ferrite was observed it was always associated with a difference in gaphite. Evaluation of the Brinell Test The following conclusions were arrived at from this study of the Brinell Test: 1. Brinell hardness has not yet been shown mathematically to be a fundamental physical property of metal alloys. 2. Brinell hardness represents only the quality of the material immediately surrounding the test location. A change in section size results in a change in Brinell hardness. 3. The Brinell scope can be source of error. 1:. Some small parts would be destroyed by the Brinell test. 5. Tensile strength cannot be accurately predicted from the Brinell test. 6. A general correlation exists between Thermal arrest carbon equivalent and Brinell hardness. 7. Changes in machinability index due to small changes in gaphite could not be detected with the Brinell test. Brinell hardness is only an approximate estimate of machinability. *The effects that small differences in silicon, manganese, and chromium contents had on matrix hardness among the six castings were neglected. In order to completely eliminate the matrix as a variable, the alloy contents should be held constant. 41,- 8. By examining the microstructures of samples where the matrix was eliminated as a variable, it was shown that larger changes in gaphite could be detected with the Brinell test. The above conclusions indicate that a more sufficient quality test than the Brinell test needs to be developed. A sufficient test should be able to measure both the matrix and the gaphite independently, which is not possible with the Brinell test. CHAPTER II THE RESQMNCE TEST Resonance testing, like indentation hardness testing, is not new to the foundry industry. For years resonance testing has been used to test Arma Steel parts.1h Arma Steel (pearlitic-malleable iron) is resonance tested by first striking the part with a hauler md then using a sensitive microphone to pick up the vibrations. This Chapter will analyze the resonant frequency test in order to determine what application it has in quality testing gray iron. Theory and factors in resonance testing and resonant frequency as related to microstructure and physical prOperties will be discussed. Theory and_ Factor: in ResmanciTestigg Resonance occurrs when the frequency of a periodic exciting fcrce approaches the natural frequency of vibration of a body. 15 This natural frequency of mechanical vibration may be expressed generally as Frequency - (shape factor) 1: (physical - constants factor) (1) £1392 m is a function d the gemetrical design and the dimensions n‘Milton J. Diamond, "The Utilization of Sonic Principles for Application to an Automatic Method of Casting Impection," General _H2tors mgeering JournalI (March-April 1956), pp. 38—142. 15Nondestructive Testin Handbook, ed. Robert 0. McMaster (2 Vols.; New York: The HonaId Hess Company, I559), II, Sec. 51, l. -15- ~16- of a body. The physical;cgn_stants factor includes modulus of elasticity, density, and Poisson's ratio of the material. An important factor which determines resonant frequency is the node of vibration of a given body. Fcr a body of simple shape, the modes of vibration can be induced independently of each other. The fmdsmental modes of vibration include flexural, longitudinal, torsional, diametral, radial, and annular modes of vibration (see Figures 15 thru 20). The nodes as shown are points which remain stationary. Multiples of the fmdamental frequency can also occurr which results in adding md shifting nodal points. Canplex shapes often vibrate with several of the modes cited above. The cylindrical test bar used in this study was vibrated in the longitudinal mode as shown in Figure 16 and supported at the nodal point in the center. The relation for a long cylindrical bar in the longitudinal mode of vibration is r . 12-1 (211:5)é (2) where E is modulus of elasticity, f is fundamental resonant frequency, 6. is density, g is acceleration due to gravity, and L is the length of the cylindrical bar. It is generally known that the modulus of elasticity of gray iron is closely related to the tensile strength providing the matrix structure remains constant. The curved nature of the stress-strain curve of gray iron in tension is also well known, so the modulus referred to above, of course, is the modulus determined at a specific point on the stress-strain diagram. Some investigators measure the -17- tangent elastic modulus at a point on the stress-astrain curve that is one-fmrth the value of the tensile strength. However, others such as Dr. E. I’lenard16 of the Centre Technique des Industries de la Fondsris, Paris, France insist that the modulm of elasticity of gray iron should be measured at the origin of the stress-strain curve after accanodation has occurred since this value is truly a constant for a particular iron. Accomodation is the process of arriving at a stable stress-strain cm which is cyclic when applying cyclic stresses which do not case yielding of the material. (See Figure 21) Resonant frequencywas determined at low stress levels in this investigation. So the modulus at the origin of the stress-strain curve would apply here . Resonant F‘regiengy as Related to msical Properties and Microstructure Before examining physical moperties and microstructure the testing procedure will be discussed. As can be seen in equation (2), the important variables are length, density, and elastic modulus. The density of gray iron was assumed to be constant in this investigation. Believer, the higher the carbon equivalent resulting in a higher free graphite content, the lower would be the average density, and a g 16Elizabeth Plenard, "The Elastic Behavior of Cast Iron," {gfiyblished pmer presented at the A. S. M. Cast Iran Saninar, J1me . p- 8- -18- corresponding tendency to increase the resonant frequency would result. Length was very'closely controlled. The castings were machined.to a length of 7.958 inches in a lathe. The maximum deviation from.this value did not affect the resonant frequency significantly. Resonant frequencies of the thirty test bars and twentybsix castings were measured with Magnaflux's SR-lOO, an instrunent developed by the British Cast Iron Research Association. Figure 22 shows a schematic diagram.of the instrument. The bars were supported on three pins near the nodal point and.vibrated longitudinally‘with an electromagnetic transducer. (The detecting and exciting transducers were (I the same type.) A piece of styrofoam was also used. This worked'well as a support near'the center,‘when the parts were grounded. Styrofoam was used to support the cast part as well. Several trial and error positions were tried.before a position could.be fonnd.in which tie casting wcnld resonate. The fundamental frequency was found by mans of a lissojous figure on a cathode ray oscilloscope. Repeatibility was within 1 1 CPS even with different transducers, including a.mechanieal transducer and a small transducer of the type used.tc measure angular speed of a gear. 'Hhen the two SR-lOO trans- dmers were within about three inches of each other and the air-gap was from one-eighth to one-fourth of an inch, interference of the scope. pattern resulted fran direct interaction of the two transducers. It'was also found.that a temperature increase of 26 F. degrees in the test bar resulted in a decrease of twentyhseven CPS in the resmant frequency of the test bar. -19- In the following discussion the relationship of resonant fre- quency to tensile strength, Brinell hardness, carbon equivalent, and microstructure will be established. But first the procedure used to determine static modulus will be discussed. Since Resonant frequency is theoretically related to the modulm by equation (2) and modulus (E) determined in the tensile test has been found to correlate with tensile strength, it would be desirable to canpare the modulus found by equation (2) and that determined with the tensile test. The former modulus is referred to as cwnamic modulus, and the latter is referred to as static modulus. Elastic Modulus. Static modulus is determined in the tensile test fran a stress-strain curve. In the conventional method a drum with graph paper rotates through an mgle that is directly proportional to the load applied to the tensile specimen. A recording pen which moves along the axis of the drum is displaced parallel to the axis of the drmn with a movement directly proportional to the strain in the sample as determined by an extensometer. n This method is good for d)serving the entire stress-strain curve up to fracture. But at low loads (and low values of tensile stress) this method has proven to be inaccurate. This and other factors praupted the use of a more sensitive stress detecting method at low stress levels. It was decided to use some $44 strain gages to detect both stress and strain. Became Duco cment pulled gaphite from the pores of the metal, Eastman 910 cement was used to attach the gages. See Figure 23 for a description of the test set-up. To measure strain two gages were attached to the -20- tensile specimen as shown. In order to compensate for bending, the gages were placed in opposite arms of a bridge. Temperature canpensation was accomplished by placing "dumny' gages in adjacent arm of the bridge. (See Figure 23) To record stress, two SR-h gages were cemented to a steel swivel shaft and fed. into a bridge in the same manner as for recording strain of the gray iron specimen. Strain in the steel shaft theoretically is proportional to the stress in the shaft. Neglecting reduction of area in the steel shaft and in the gray iron specimen, the tensile stress in the gray iron sample is found to be: S - 95% e, E, (3) and the strain is e e " 125 (h) where es is the sum of the strains fran the gages on the steel swivel shaft, Es is the elastic modulus fcr steel in tension which is a constant, A3 and A3 are the cross-sectional areas of the steel shaft and grey iron specimen, cg is the sum of the strains from the two gages on the gray iron sample, and e is the average tensile strain in the gray iron sample. Next, the two bridge output voltages, one fcr stress and one for strain in the gay iron tensile sample, were each fed into 9. Brush carrier amplifier whose outputs were each fed into a preamplifier of a Sanborn optical X-Y recorder. The stress was recorded on the I-axis -21- and the strain on the X-axis. The X-I recorder recorded the stress- strain curve by means of a sharply focused highly intense light beam and a photographic paper. The X and Y axes were calibrated in terms of microinches per inch of strain. Stress-strain curves were obtained with this method. But be- cmse of a faulty amplifier discovered in the circuit too late, the stress-strain curves were not accurate. Since available samples for tensile testing were exhausted, no results could be reported here. However, if the method were perfected, it would be ideal for analysis of the stress-strain curve of cast irons at any stress level below the stress at which plastic deformations begin because of the high sensitivity which is possible. Tensile Strgfl. The relationship between tensile strength and resonant frequency for 5 ix as cast bars is shown in Figure 2“. It was felt that sample 2A 3, the sample which did not fall close to the straight line in Figure 2'; had an erronous resonant frequency due to unknown factors. 7 These bars were poured in groups of two per mold and the resonant frequencies of the two bars per mold were found to be within 100 CPS of each other in every case except this bar which had a resonant frequency. of 965h CPS while its mate had a resonant frequency of 9908 CPS. Previous studies lave been made in which many more samples have been tensile tested and the results plotted against resonant -22a frequency. In an article by A. Ge Fuller18 and others, a similar correlation was found as in Figure 2h. Hr. Fuller's data showed that the tensile strength did not exceed‘: 2000 Psi about the mean line. Brinell Hardness. ‘Hhen plotting average Brinell diameter against resonant frequency, data for the as cast test bars were considered separately from.data for the annealed test bars. In Figure 25, the upper curve represents six annealed test bars and the lower curve represents twentybfour as cast test bars. As was cited earlier, the matrix structure of each of the annealed.bars were found to be all ferrite. This'was discovered.by etching the sample in a two percent nital solution. The lower curve shows another relation, but all of the points except one are grouped within _+_ 1.5 m. of the mean line. In comparing average Brinell.diameter to resonant frequenoy of the twenty-six as cast parts shown in Figure 2, a considerable spreadis found above a Brinell diameter of in} m. (197) (See Figure 26). THhen observing the non-uniformity of castings in Brinell hardness, it is evident in Figure )4 that at 11.3 mm. diameter and later the Brinell diameter is much more uniform. Carbon Equivalent. As shown in Figures 27 and.28, a good correlation between carbon equivalent and resonant frequenoy is not 18A. o.’ Fuller, et al, "Sonic Testing: A simple Non-Destructive Test for Verifying Casting Quality," BCIRA Journal, (VII,IMay 1963), p. 372. -23- obtained. But the correlation between Thermal arrest carbon equivalent and resonant frequency is better than the correlation between wet carbon equivalent and resonant frequency. Microstructure. Modulus of elasticity has been found to be related to the average flake length of the graphite in the microstruc- ture which is in turn a function cf the cooling rate and gaphite " content. It has been found also that the coarser the graphite, the poster the systematic deviation between the dynamic modulus determined by equation (2) and the static modulus.” It is suggested by Dr. Plénnrd that this systematic deviation might be a result of the hetero- geneous nature of the structure of gray iron since there should be no distinction between static and dynamic modulus. Equation (2), then, might also need to consider a factor of graphite flake size for coarser structure. These considerations indicate a relationship between graphite quantity and length of flakes and resonant frequency. A photomicrograph tcward the center of sample 1A1 (Figure 9) which shows a relatively short graphite length and sane abnormal graphite has a resonant frequency of 10052 CPS, while sample 3A1 which has a much greater average graphite length has a resonant frequency of 9202 CPS. Upon examining the microstructures of samples 2A1 and 2A2 (Figure 25) which had resonant frequencies in between samples 1A1 and 19Pltnard, pp. 15 and 16. azha 3A1, it was found that the average flake length was between that of samples 1A1 and 3A1. Also some abnormal graphite was found in samples 2A1 and 2A2. This suggests a relationship between flake size and resonant frequency. A further investigation was trade with samples 15, 3A7, and 3A3 (see Figure 25). Amity sample 1A5 having a high resonant frequency had a shorter flake length and abnormal graphite. Sample 3A3 having a low resonant frequency had coarser graphite flakes. And sample 3A7 having an intermediate resonant frequency had graphite intermediate in size. This effect can be observed in the photomicrographs in Figures9 and 10. Figure 9 showing the, fine graphite flakes was associated with a high resonant frequency, and Figure 10 showing the coarse graphite flakes was associated with a low resonant frequency. Graphite flake size, then, determines resonant frequency. Anothm' significance concerning resonant frequency is the ' pronounced shift of the rescnant frequency of the annealed samples toward the lower frequencies in Figure 25.. The carbon equivalents and chill depths of the samples 3A1, 3A3, and 3A).; are the same, and the carbm equivalents and chill depths of samples 2A1, 2A2, and 3A? are nearly the same. Since it was determined that the matrix was nearly the same fran sanple to sample, then the above grouped samples would have all been similar in Brinell hardness, microstructure, and resonant frequency in the as cast condition. The pronounced shift which is about 300 cps to the left, could be a result of both matrix change and a result of secondary graphitization occurring upon annealing. -25.. It should be noted that pearlite and ferrite have slightly different elastic moduli which might affect resonant frequency. At tin surface, all the samples had the rozstte pattern of grqzihite as shown in Figure 11 but again varying in fineness. The finer surface graphite was associated with a finer graphite at the center and the coarser surface graphite was associated with the coarser grqahite at the center. In addition sane ferrite which varied in quantity from sample to sample was associated with the finer surface graphite (see Figure 11). Again the greater quantities of ferrite at both the surface and interior seemed to be associated with the larger gaphite flakes as represented by Figure 10 and, therefore, with lower resonant frequencies. Figure 12 shows the ferrite patterns associated with the coarser graphite flakes represented by Figure 10. This could mean that at lower resonant frequencies (coarser graphite flakes) one walld expect to find sane ferrite. Figure 13 shows a photomicrograph of the typical pearlite found in all the as cast samples examined. No pronounced variation in pearlite was found. This means that the pearlite was eliminated as a variable. Finally, Figure 114 shows evidence of still another constituent found in all tin samples examined in slightly varying amounts. Since this material was located in between eutectic cells and was not continuous when examined at higher magnifications, it was concluded to be steadite. Steadite is a hard brittle constituent that consists of a binary eutectic of ferrite (containing sone phOSphorous in solution) and iron phosphide (FeBP). Steadite becomes visible when the phosphorous content ~26- exceeds 0.1 percent. Unfortmately, the phosphorous content was not analyzed for these samples, and no definite decision could be arrived at as to effects of phosphorous. It was thought, though, that tin mount of steadite was too small to affect the test indications. Evaluation (1’ the Resonance Test The conclusions arrived at in the analysis of the resonance test are as follows: 1. 2. 3. 6. A theoretical relationship exists between resonant frequency and modulus of elasticity of gray iron at ltw stress lemls. Since modulus has been found to be related to the amount and nature cf the gaphite (if modulus is defined properly) and nearly independent of the nature of the matrix, resonant frequency should also be related to the amount and nature of the graphite. Elastic modulus was not accurately determined. However, a pranising method using strain gages for analysis of the stress-strain curve at low stress levels was found, but the method needs perfecting. Relationship between resonant frequency and tensile strength among six as-cast test bars was good but not enough tensile tests could be obtained. Others have found similar relationships with an accuracy (1' + 2000 Psi about the mean curve through the points. The matrix Inst be the same from sample to sample when comparing tensile strength since resonant frequency is a function ofE which is a function only if the gaphite. Average Brinell hardness and resonant frequency seem to have a good correlation when the variation of Brinell hardness does not vary significantly frau point to point within a cast part. A general correlation was found to exist between Themal arrest carbon equivalent and resonant frequency, but was not significant enough to use as a melting cmtrol. kamination of microstructures of eleven selected saples revealed a relationship between resonant frequency and graphite size and type. The coarser flakes structure having -27- very little type D graphite was associated with a lower resonant frequency. The finer flake structure with much more type D and E graphite was associated with a higher resonant frequency. Greater quantities of free ferrite seemed to be associated with the coarser structured graphite and with the lower resonant frequencies. The resonance test, then, offers a precise means of analyzing the influence of the graphite phase nearly independent of the type of matrix structure. The next step would be to find a method which would analyze the matrix structure independently of the amount and nature of the gaphite. This attempt is made in tin next chapter. CHAPTER III STATIC MAGNETIC TESTS AN D EDD! CIRRENT TESTS Magnetic properties were known to be associated with hardness of steel for many years. A magnetically "hard" material is usually “hard“ physically. Hard steels will, in general, give meteresis curves of large areas and soft steels, small areas.20 H. J. Diamond has applied magnetic retentivity to sort Arma Steel rocker arms 21 Thus, a relationship of mechanical hardness according to hardness. to magnetic properties of ferranagnetic materials is indicated. The tests analyzed in this investigation my be classified into two groups: static magnetic tests and eddy current tests. The static magnetic tests include the magnetic retentivity test and the coercive force test. These tests were conducted with a Foerster Coercive Force Meter. The eddy current tests include a comparative test and a probe coil test. The canparative test was conducted with a Foerster-Hoover Pbdel 00-1000 comparator which used two transfomer coils. A standard of the same kind as the part to be tested was placed in me coil, and the part to be tested was placed in the other coil. The probe coil test where the axis of the coil was placed perpendicular 20Williams, p. 383. 2lDiemmd, "A Summary ...." p. 26. -2 8.. 129- to the test surface was conducted with Magnaflux's model ED-300 Eddy Current Tester. These static manetic tests and eddy current tests will be discussed as follows: principles in static magnetic testing and edw current testing, testing procedure, microstructure and physical proper- ties, and evaluation of these tests. Principles in Static Magetic Testing and Eddy Currept Testing?22 Both static magnetic tests and eddy current tests have been aplayed in industry to nondestructively test ferrous parts. The principles upon which these tests are based will be discussed below. Static Maeetic Test . The advantage of static magnetic tests is that test indications are representative of the whole cross-section rather than mainly the surface of the part. Retentivity is that value of residual induction on the direct-current magnetization curve which is obtained by saturating a ferranagnetic sample and then reducing the magnetizing force to zero. Coercive force is the amount of magnetizing force required to reduce the residual induction to zero after saturation of the material. (See Figure 29.) An advantage of the coercive force method is that it represents a factor which can be measured independently of the shape and mass of the material. This means that the coercive force of two different parts could be compared, if desired. Coercive - w—i 22mm Handbook, II, 3h.1 to 112.714. -30- force has been related to sudh properties as hardness, tensile strength, depth of case, allqy content, and aging conditions. ‘Wbsn correlating these properties to coercive force, two other factors must be considered. It is important that no reversal appear in the function relating coercive face to material properties. It is stated in the Nondestructive Testing Handbook, Vol. II, that suchhreversals may appear with several alloys, particularly after repeated heat treatment. This reference also states that coercive force is dependent upon the temperature of the material being tested. Eddy Current Tests. Eddy current testing was developed.by Dr. Foerster in Germany around the end of World War II. The principles developed then form the basis of present-day eddy current testers. A schematic representation of the two types of probes used in this study are presented in Figure 30. The QC-lOOO Comparator uses the transformer coil'where a part to be tested is the core (Figure 30A), and the ED-BOO lhwy'Current Tester uses the single inductor coil (Figure 30B). An induced.magnetic field in either probe produces eddy currents which in turn produce an opposing alternating field in the sample. The change in impedance due to the presence of the part in both eddy current methods is dependent upon electrical conductivity, dimensions of the part, magnetic permeability, presence of discontinuities,.frequenqy of the test coil, the size and shape of the coil, and the coupling between the coil and.the sample. A facta in eda current testing which must be considered is that the alternating fields are stronger at the surface than at the center of the part. The penetration depth, P, is defined as the depth -31. below the surface at which the field hm decreased to 36.8 percent of the field strength. In order to determine the effective depth, the mathmatical solution of the general case of a cylindrical test object had to be obtained.23 In this solution the cylinder was assumed to have a unifam permeability over its cross-section designated as effective pemeability (Ueff) which is related to electrical conductivity, relative permeability, diameter of the test object, and test frequency. Fran this solution which occurred as the argument A of the Bessel function, the values (1' frequency defined as the limit frequency, fg, was obtained by setting A equal to unity. The value for fg is: Ural c D2 where Ural is the relative permeability of tha test mterial, C is the electrical conductivity. (1' the test material in meter/ohm-mz, and D is the diameter of the test sample. The ratio of f/fg determines the value ani phase angle of the field strength at a given point below the surface of the test sample, f being the test frequency. The effective depth fa a particular f/fg ratio is then determinable.”4 For a hanogeneous iron sample of the same nominal diameter as the test bars used in this study, 1.2 inches diameter, the effective depth was computed to be approximately 0.15 inches fa a coil frequency of 60 CPS and roughly 0.0!; inches fa a coil frequency of 360‘CPS. These 231b1d., II, 36.13. 21mm Handbook, II, 37.9. -32- calculations were made assming that the coil axis was coincident with tie axis of the cylindrical test bar. The main significance of the general mathanatical soluticn is that effective depth is predictable knowing test frequency (1‘), limit frequency (fg), and the diameter of a cylindrical test bar. Another significant principle worthy of mention is the law of similarity in eddy current testing. This fundamental law is stated as follows: "The effective permeability, as well as the geanetrical distributions of the field strength and eda current densities, is the same fa two different test objects,2%f the ratio f/fg,is the same for each test object." This law is significant in comparative testing, as with the QC-lOOO Casparetor. Since fg is a function of U rel: C, and D, a comparative test would indicate a difference in relative permeability, electrical wnductivity, and size. Size was considered as a constant in this investigation. So a difference in test indications with the QC-lOOO Cmparata was a measure of the relative pemeability and electrical conductivity with respect to a standard. Testing Procedure The static magnetic test instrument investigated was the Foerster 251m Handbook, II, 37.10. -33- coercive face meter which measures magnetic retentivity and coercive face. The eda current instruments investigated were the QC-lOOO Comparator and the ED-300 Eddy Current Tester. The thirty test bars that were resonance tested were first tested with the QC-lOOO Canparator. Then three-inch bars from the longer bars tested above were machined to length in order to measure coercive face and retentivity. The ED-300 tests were also made on the three-inch test bars. (See Figure 1.) Only QC-lOOO and nil-300 indica- tions were taken on the cast parts from Central Foundry since their physical size would not allow them to be placed in the Coercive Face Hater Test Coil. The testing procedure fa static magietic tests and eddy current tests will be presented. Static logistic Tests. A diagram of the Coercive Face Meter is shown in Figure Bl The high-sensitivity field probes are aligned so that the tangential field from the test coil does not affect the probes when reversing the field to measure coercive face. This unit must also be canpensated fa the earth's magnetic field. Testing procedure involved these steps: ‘ (1) Making preliminary adjustments and canpensating fa the earth's magnetic field. (2) Slowly and steadily, without stopping, increasing the magnetizing face until the sample was saturated. (3) Slowly and steadily, without stopping, decreasing the magnetizing face to zero. (14) Reading the relative magnetic retentivity. -3h- (5) Reversing the current and thus the direction of the magnetizing force; and increasing the magaetizing force until the value of relative magnetic induction is zero. The value of magnetizing force required to perform step (5), then, is the coercive face of the material tested. Eddy Cur_rent Tests. With the QC—lOOO Comparata, mapstization variations, permeability variations, or the variations of the curvature of the hysteresis loop of a sample with respect to a standard are measured. The above instrument is similar to the Rhgnatest Q instrument26'which is a linear time-base instrument. The variation in magnetic properties between two parts is represented by the relative change in the vertical position of a trace displayed on a cathode ray oscilloscope. The linear time-base is applied to the horizontal plates of the oscilloscope. (Figure 32) Tie QC-lOOO unit was only one test frequency, 60 CPS. It would be desirable at times to neasure magnetic noperties at even lower frequencies, for two reasons: a greater penetraticn depth can be obtained, and the effects of inhomogeneous internal stresses27 can be eliminated. Inhomogeneous internal stresses tend to result in an inseparable pattern on the oscilloscope screen. ‘ Testing procedure with the QC-lOOO Conparator was as follows: (1) Making preliminary adjmtments accading to the instruction manu‘1u 26unr Handbook, II, no.29. 27NDT Handbook, II, 142.1;3. -35- (2) Choosing a standard sample and placing it into one of the two coils (see Figure 32), the standard being midway between the extremes of Brinell hardness. (3) Adjmting the scope display to obtain a haizontal line then properly placing other samples of identical Brinell hardness into the second coil one at a time. (h) Adjusting vertical sensitivity, such that a full scale difference in scope diSplay results, when examining a number of samples on the "hard" side if the hardness range and on the "soft" side of the hardness range. (5) Making a "phase" adjustment in order to locate the maximum deflection of the trace at 'the center of the screen. (6) Placing a aid over the oscilloscope screen ad taking readings of the trace position near the center of the screen, considering ary phase shift. Next, testing cmsiderations fa the ED-BCO tester will be examined and the test procedure presented. Essentially, the ED-BOO tester which Operates at a frequency of 360 CPS, is a power-loss measuring device for ferromagnetic materials. The power-loss is due to hysteresis losses and edcv current losses in the material. The following excerpt serves to explain how the ED-300 operates: "When an alternating current is made to flu: through a coil held in close proximity to a ferranagletic metal, alternating magnetic fields are induced in the metal. These fields, in turn, will induce circular counter-currents (eddy currents) within the metal. The reSpective count er field developed by the eddy cm'rents will oppose the applied field with a magnitude and phase dependent on the resistivity and permeability of the metal. Both of these characteristics vary with analysis and . structure, pemeability being by far the greatest affected. The losses associated with each cannot be measured independently of each other using alternating magnetic field techniques, but their sun can be indicated on tge BID-300 meter as It: nmbers or power-loss probe measurements."2 28 eratin Manual fa Ma atest ED-300 Low Fre uen Measurin Instrument, gn 1n: apora on , - . -36- The ED-300, like the QC-lOOO, is a special application of general eddy current principles. The QC-lOOO compares a standard with an 11111010!!! test specimen, while the ED-BOO reads directly the power losses in different parts. The methods would be comparable providing the effective penetration of both were the same. (Figure 33) Preparation of the FIB-300 for use after initial want-up was found to be very simple and rapid. Initial warm-up requiring about an hair takes the most time. After the instrument Ind warmed up, the sensitivity was set at minimum and the probe was cmpensated for lift- off with a test bar intermediate in Brinell hardness. Lift-off is a term med to designate adjmtment of the instrument so that no difference in reading is obtained when the probe is lifted a few thousandths of an inch off the surface of a test sample. When coupe-sated for lift-off, oxide scale or burnt-in sand on a casting a few thousandths thick will not affect the reading. To compensate for lift-off, a piece of ordinary writing paper was placed between the probe and the test sample and adjustments were made until the presence of the paper did not affect the restate. Then the sanples were tested. First, the two ends of the cylindrical test specimen which were machined surfaces were tested at the center. Then lift-off was readjusted so the contour of the as cast cylindrical surface could be measured. Since the as cast surface was of different geometry, an overall average value could not be datained. When condmting tests cm the thirty test bars and twenty-six cast parts, the following observations were made: (1) Gemietry d the test sm'face affects the results. (2) Mass affects the results. -37- (3) A maximum down-scale jump of ten percent of the total scale occurred while taking measurements. Readings at maximum scale reading were taken after observing scale indications fcr approximately one minute. The first two observations could, of course, be predicted from the earlier theoretical considerations. flgsical Prgperti es andficrostructure Before, considering each test separately, a comparism will be made between the test indications. The coercive force test and the magnetic retentivity test should be related since they are both static tests; and the QC-IOOO and the HID-300 eddy current tests should be related. The comparison of static magnetic tests should indicate whether to eliminate one of tie tests from consideration. Neglecting saple 2A? in Figure 3h a maximal: deviation about the mean line of the plot of residual magnetism against coercive fcrce was 3 12 units of residual magnetism. This deviaticn was probably due to errors in determining residual magnetism. This was concluded after considering the following. In the Nondestructive TestELHandbook, Vol. II, it was reported that “a rod or small grindings of a given material will give, without re- calculation, the same measured values" of coercive force regardless of weight. Retentivity, however, would change since the orientation of the material to the detecting probe would change (see Figure 31) . Thus, errors in retentivity might result due to positioning of the sample. It would seem, then, that coercive force as measured by a Coercive che Meter would be less subject to variations due to ~38~ extraneous factors. Thus, one of the static magnetic tests, the retentivity test, was eliminated. Next, when plotting surface indications against center indica- tions per sample fcr the ED-300, a considerable spread was obtained (see Figure 35). This is not surprising since the effective penetra- tion depth for a steel cylinder of the same diameter as the test bars was only o.oh inches. when plotting QC-lOOO Comparator test indications against ED-300 center indie ations,a rather narrow relative deviation about the mean line resulted(see Figure 36). Surface conditicns appeared to have a mcre pronounced effect upon the results of the ED-300 test than upon the results of the QC-lOOO test due to the higher frequency involved with the EDa-BOO. Also the relationship between ED-300 (center indications) and QC-lOOO tests has the correlation expected, since both are eddy current tests. As a result, than, the less accurate tests were eliminated and the magnetic tests under consideration were reduced to a static field measurement, coercive force; and an eddy current canparison of two samples, the QC-lOOO Canparator indications. The relationship of coercive force and QC—lOOO indications to tensile strength, Brinell hardness ani carbon equivalent, and microstructure are reported here. TensilgLStrengh. As can be seen in Figure 37, the correlation between tensile strength and the QC-lOOO indications for seven tensile tests is good. However, more tensile tests should be made in order to establish the standard deviation. It is not surprising to find a correlation between eddy current indications and tensile strength since tensile strength is a function of the two variables matrix + graphite, and the eddy current indicat ims are a function of the permeability -39- and conductivity, which in turn are functions of the matrix material and graphite. Next examining coercive force (Figure 38), one finds a much poorer correlation d‘ tensile strength with coercive force. This might be expected when considering the following factors: (1) The coercive force is a property of the material and not dqaendent upon mass or shape. (2) Graphite size shcnld not affect the coercive force of the material appreciably since it is nonmagnetic and would have nearly the same pemeability as air. (3) Coercive force is dependent upon the amount of combined carbon, silicon content, and other alloys in solid solution with the iron. Brinell Hardness. Average Brinell diameter was plotted against the 00-1000 test indications for both the test bar and the as cast parts. in average deviation of less than I 0.20 nnn. Brinell dimeter is indicated for both the test bar ani the as cast part (see Figures 39 and 110, respectively). When ccmparing Figures 6 and 37 , it can be seen that a much better correlation exists between 00-1000 conparator indica- tl. one and tensile strength than between Brinell hardness diameter and tensile strength. Thus a deviation is expected when conparing average Brinell diameter with QC-lOOO test indications. Therefore, if tensile strength is the property important to the customer,the QC-1000 canparator test would be much more reliable than the Brinell test. Coercive force and Brinell hardness seemed to be better related than coercive force and tensile strength. (See Figures 38 and hl) A deviation of _+_ 0.20 m. zesulted about the mean line connecting the two sets of points. Again, more points between these two groups of points would have been desirable. -140. Carbon Equivalent. The correlation between (QC-1000 test indica- tions and carbm equivalent was not good. There was even poorer correla- tion between coercive force and carbm equivalent. It mat be remembered that when equivalent is determined before a casting is poured. The important variables, ladle inoculation which affects the graphite phase, and cooling rate, which affects both the graphite phase and the matrix, are not taken into account when trying to correlate these test indications with carbon equivalent. A test which does adequately evaluate the properties of gray iron, however, can be used in Foundry control. By observing the changes in carbon equivalent before the iron is poured and by observing changes in test indications after gray irm parts are poured and have cooled, changes in the intermediate variables, inoculation and cooling rate, can be observed. morostrligture. Coercive force and QC-lOOO comparator indica- tions of some of the test bars were ccmpared with microstructures to determine if a relationship exists between these test indications and microstructure. Also, tensile strength, resonant frequency, md Brinell hardness were used to supplanent microstructural observations. In the canparing coercive force with tensile strength which is a function of both matrix and grqahite and considering resonant frequency which is a function of graphite only, the relationship between coercive force and microstructtre can be deduced. In closely examining Figure ’42, it can be seen that the effect of ferritising treatment (annealing) is to reduce both the coercive force and the tensile strength. It can be seen that a reduction in resonant frequency -hl- (and therefore a coarsening of the graphite) also causes a reduction in tensile strength (considering either the as cast group or the annealed graup), but does not affect coercive fcrce. Thus, it can be deduced that the charge in grqahite flake size does not affect coercive fa'ce. Cm coercive force detect smaller changes in the matrix than was brought about by the ferritis ing treatment? To answer this question, Figure hl was examined. The lower coercive face of samples 3A3 and 3A1; may be attributable to the presence of some ferrite around the gaphite flakes as shown in Figure 12. Other small changes in coercive force were not explainable on the basis of microstructures, since any changes in pearlite coarseness were not observed. It was found from this investigation, then, that coercive force is related to the amount ct ferrite in the matrix, but is independent of the graphite size. The eddy current test, the 00-1000 comparative test, seans to be a good indicator of tensile strength (see Figure 37). Tensile strength, of course, is related to both the graphitic phase and tln matrix. When analyzing Figure 39 and picking out samples whose microstructures were examined, it can be, seen that the 00-1000 indica- tions are influenced by both the matrix and the graphite. Samples 1A1 and 3A1 have the same type of matrix (ferrite), but sample 1A1 has flakes of shorter length than sample 3A1 (see Figures 9 and 10, respectively). The differences between samples 3A1 and 1A5 in 00-1000 comparator test indications are primarily a result of changes in matrix. Sanple 3A1 has an all-ferrite matrix, and sample 1A5 has an all-pearlite matrix. -hg- Evaluation of Static Magnetic Tests and wCurrent Tests The static magnetic tests considered were magnetic retentivity test am coercive face test which were conducted with a Foerster Coer- cive Face Meter. The magnetic retentivity test was eliminated frm ccn sideration because it depended upon orientation with respect to the measm‘ing probe and upon the geanetry and mass of the test sanple. Coercive face depended upon neither geometry na mass of the test sample. The following conclusions were reached in the investigation of the coercive face test: 1. 2. 3. 7. Coercive face, being a static magnetic field test, is representative 0? the entire cross-section. The coercive forces of two different parts with different masses and shapes can be canpared because coercive face is independent of ness and geanetry of the test sample. It is significant that no reversals appear in the function relating physical properties to coercive face. Coercive face is dependent upon temperature. The coercive face test is not a sufficient test by itself to predict tensile strength a Brinell hardness. Carbon equivalent of gray iron could not be significantly correlated with coercive face. Coercive face seemed to be related to the amount of ferrite and pom-lite in the matrix independent of graphite size. In the samples examined for microstructure, the coarseness of pearlite did not vary. Thus the affect that chages in coarseness of pearlite has on coercive face was not investi- gated. The coercive force test would be the solution to the problem of finding a test which examines the metallic matrix of a casting independently of the graphite size, if differences in pearlite coarseness could be 4,3- detected with the coercive face test. The eddy current tests com idered were the 00-1000 Caparata test and the ED-BOO Eddy Current test. The ED-300 test, nearly the same as the 00-1000 test in principle, was eliminated fran consideration because it depended more upon the abnamal surface conditions than did tl'e 00-1000 Canparata test. The conclusions reached concerning the 00-1000 Comparator test were as follows: 1. 2. 3. The penetration depth (or effective depth), which is de- fined as the depth below the surface of the test sample fa which the field strength drops to 36.8 percent of the surface field strength, can be predicted accurately fa simple gemetrical shapes adaptable to mathaaatical solution. The Law of Similarity, in essence, states that if relative permeability, mass, geanetry, and test frequency of two parts are the same the eddy current distribution and effective permeability in one part will be identical to the eddy current distribution and effective permeability in the other. This law fame the basis of cooperative test- ing. Tensile strength and 00-1000 test indications have a much better correlation than tensile strength and Brinell hardness. A carelation exists between Brinell diameter and 00-1000 Canparata test indications which has an average deviation less than _+_ 0.20 Brinell diameter about a mean line drawn through the test points. No significant carelation was found between carbcn equivalent and 00-1000 Comparata indications. Analysis of microstructures showed that 00-1000 test indications are a function of both the matrix and the graphite. , The 00-1000 Comparator test indication, like Brinell hardness, is a function of both matrix and graphite. But the 00-1000 Comparata test is a more reliable test than the Brinell test in rredicting tensile Strength 0 CON CIUSI'. ONS The goal of this investigation was to examine the Brinell test, the resonance test, static magnetic tests, and eddy current tests in order to find a sufficient test a a canbination of tests which are capable of examining the quality of gray iron. The conclusions re- sulting from this investigation are as follows: 1. The Brinell test is not a sufficient quality test to predict the physical properties and microstructure of gray iron. 2. The resonant frequency test evaltetes the graphite phase in gray iron independently of the matrix. A larger graphite flake size results in a lcwer resonant frequency. Resonant frequency depends upon the graphite phase, dimensions, density, and temperature of the sample being tested. 3. The coercive force test evaluates the amount of ferrite in the matrix independently of the graphite flake size. Coercive force is dependent upon the amount of carbon and other alloying elements in solution and upon temperature. h. The 00-1000 Comparative test, like the Brinell test, evaluates both the metallic matrix and the gaphite phase; however, the 00-1000 test is more reliable than the Brinell test in predicting tensile strength. In summary, the Brinell test is not a sufficient quality test fa gray iron. The resonance test evaluates the graphite phase independently of the matrix. The coercive face test evaluates the mount of ferrite in the matrix independently of the graphite phase. Finally, the 00-1000 test depends upon both the matrix and the graphite phase, but test indications are more clcsely related to tensile strength than to Brinell hardness. 4,1,- RECDMI‘ENDATION S Recamnendations are made as follows: father investigation, production line application, and establishment of standards. Father Investigation Further labaatory investigation should include: 1. 2. 3. Carelation of coercive face and ferrite content over a range of ferrite content: In this investigation the matrix had no ferrite, two percent ferrite, or 100 percent ferrite. Ferrite contents of ten, twenty-five, and fifty percent should be investigated in addition. Observation of the affect that differences in pearlite have on coercive force: In this investigation the differences in pearlite coarseness were not detectable under the microscope. As large a range of pearlite as is possible should be examined. The affect of temperatae on coercive face: The temperature of a single gray iron sample could be varied, and successive readings cf coercive force could be taken in ader to establish a1 experimental relationship between coercive face and tanperatae. Temperature of castings varies comiderably in various stages of foundry production. The effect of temperature on 00-1000 test indications: It was not determined in this investigation but 00-1000 test indications vary with the temperature of the test sample. The affect of temperature on 00-1000 canparator indications could be investigated as proposed in (3) above. Methods of suppating samples to be resonance tested: Pins and styrofoam were used to suppat parts at nodal points when they were resonance tested. Pins will likely break-off and styrofoam may not. be durable enough in pro- duction foundry testing. The, better methods of suppat shculd be investigated. Methods of inducing resonance in small parts: In this investigation, when the SR-lOO electromagnetic transducers were within about three inches of one another and loosely coupled with the sample being tested, "cross-talk" a 4.0... ‘f/ 4:6- interference between the two transducers resulted. First, shielding of ore transducer from the other should be tried. If this does not work, different transducers should be tried. 7. Theoretical and experimental investigation of the resonant frequencies of complex parts: A complex part often has nany frequencies at which it resonates. These frequencies must be far enough apart so that no more than me frequency falls in the testing range. 8. The affect of mass and shape variations in production cast- ings upon resonant frequency. 9. The affect of mass and shape variations in production castings upon 00-1000 comparator test indications. Production Line Appligation After investigation of the nine items above, test set-ups for production line testing in tin foundry must be designed. A quality test could be med for two different purposes in a production foundry: as a control device (r as a final inspection device. Figure 1:3 shows the possible test locations in the production routing and Appendix B shows an example of a specific test set-up. Cgtrol Defigg. Used as a control device, the quality test could be used either as a test evaluating heat treatment car as a test evaluating iron pouring and mold line performance. In the heat treatment of gray iron castings at Central Foundry, Danville Plant, the castings are heated to temperature unifomly, held at temperature for a period of time, and then cooled unifomly in a cooling zone in the furnace ani in air. The heat-treatment could be either for stress relief or fcr various degrees of ferritising (anneal- ing). when cooled to a certain temperature, (the temperature must be below the Curie taped-attire) the coercive forces and resonant frequencies of several castings could be obtained. The amount of free ferrite -h7— (coercive face) and the size of the final graphite (resonant frequency) could be determined. This inf amation could then be fed back into the annealing fanace as a control device. A similar control device might be set up to examine a certain percentage of castings prior to annealing. This device would indicate what adjustments on the annealing furnace are required to properly anneal the castings. Used as a control on iron pouring and the mold line, the tests would be perfumed on uncleaned castings which had sufficiently cooled frat their shake-out temperature. The castings could be standard test bars. The coercive face test would evaluate the variables affecting the matrix, nmnely chemistry and cooling rate. Soon chemistry will be determined rapidly enough so that cooling rate will be the only unknown at the time a casting is shaken-out of its mold. If chemistry is known, mold line speed can be adjusted in ader to control the cooling rate of the matrix. At the same tine, the resonant frequency test would evaluate the variables affecting the graphite phase which are inoculation, rate of solidification, and chemistry. The rate of solidification is controlled by chemistry, pouring temperature, and heat transfer. The variables which can be easily adjusted are inoculation, chemistry, and pouring temperature. Thus, a resonant frequency indication might require adjustment of one a mae of these variables in ada to hold the graphite phase within a specified range. Final Inspection 2ev_i_ce. When used as a final inspection device, a quality test shmld segregate the acceptable castings from the unacceptable ones. -hei Suppose, for example, that strength and machinability we the important factors to a custaner. A test would have to be devised which would indicate strength and machinability. One approach might be to use the QC-lOOO comparator to directly evaluate strength and the resonance test to evaluate the graphite phase. With tie combination of these tests, the graphite phase and the matrix could be evaluated. The matrix would be evaluated indirectly since resonant frequency is a function of the graphite phase and the 00-1000 indication is a fumtion of both the natrix and the gaphite phase. A genaal indication of machinability would be detemined since mchinebility is mlated to both the matrix and the graphite phase. Another approach would be to use the coercive force test which determines the amount of ferrite in the matrix and the resonant frequency test which evaluates the graphite phase. Machinability which is a function of the microstructure would be generally indicated. Strength could be deduced frcm a combination of the above tests. To summarize, a quality test might be used as a control a as a final inspection device. When used as a control device, separation of the matrix and graphite factors are important to simplify the cmtrolling of both the annealing cycle and the casting process. When used as a final inspection device, separation of the matrix and the graphite phase are also important. For tensile strength the QC-lOOO test which depends upon both the matrix: and the graphite phase is adequate. -hg- Establishing_8tandards The establishment of standards at the foundry and with the customer are very important items in the application of a nondestruc- tive test to final inspection. In and: to use the matrix and graphite tests separately as controls on final inspection, the customer met accept the new tests adcpted and must be persuaded to give his specifications in terms of'the nondestructive tests. In emery, it is recommended that further laborattry investi- gation be conducted as indicated, that [reduction line application be nude, and that tin proper standards be established in the foundry ad with tln cmtaner. BMIOGIAPHY Diaond, Milton J. "A Smary of Sane New Processing and Quality Control Developments in Foundry Technology," General Motors Eng’neering Journal, XI (Second Quarter, 1961;). Diamond, Milton J. "The Utilization of Sonic Principles for Application to An Automatic Method of Casting Inspection," General Motors Queuing Journa . (March-April, 1956). Fuller, A. G., et a1. "Sonic Testing: A Simple Nondestructive Test for Verifying Casting mality," BCIRA gournal, VII (Hay, 1963). Kececioglu, Dr. Dimitri. "Factors Affecting Gray Iron Machinability," Foundg, XCI (October, 1963). Lyman, Taylor (ed.). Vol. I of Metals Handbook, Metals Park, Ohio: American Society for Me'taIs, IE1. 'Machinability of Iron Castings," Foungz Trade Journal, cm (Dec. 13, 1962). McMaster, Robert C. (ed.). Vol. II of Nondestructive Testin Handbook, New York: The Ronald Press Canpanv, 1959. gratinfi Manual for mustang-200 Low Frgguengz Measuring Instrulent, gn ux rpora ion . Plenard, Elizabeth. The Elastic: Behavior of Cast Iron, (Unpublished paper presented at the 1.3.5. East Tron Seminar, June, 1961;). Williams, Samuel R. Hardness and Hardness Measurements, Cleveland: American Society for mm, Idle. “1 -So- APPENDIX A musmnons ND GRAPES .hpnssao n8 endure 8333 .853er one: 3502“» ed.“ neon £033 .89.“ :0.“me .m .5 89m ... fies: « Wu San senator pure pangs bore 835 as. anon. 83s Baotou .3 a3 eons .m .5 was. .. .333 ”some oonocooom non .aB some .N 68383 fine floranm .H a N m H a m g Edd 33m BE magmas gm moan Nan endgamv gggg H5535 m—c- ERA a: 5393 Egon deems 6E ass. .83 mmmwmom .02 gm N amps: .22 RH ~.HOHH¢BZMQE mo gfléHQ as as an em 33 on as 3 3 3 ms .3 3 as 3 as as an 3 an a _ _ a i _ d e _ _ 4 a _ A. _ _ _ J l . _ ZOHB mmmzbz mmmag Rama—”mm m gefi UMFVP. '\V l\‘ A! f‘ \ ~o BRIDE-EU. HFK’I‘II'IE..." , BRINELL DIAMETER IN NM. 5.0 h.9 h.8 11.7 h.6 h.5 luh h.3 h.2 h.1 h.o 3.9 3.8 FIGURE h 33mm DIAMETER vs. POSITION on casrmo FOR on warm cAsrmos PART NO. 5692885 POSITION ON CASTIN G mam 33sz En has mmmBHm Ii 00.0 5 2 o 0 a 0.50 3:555 m m: leOfiD Bl®\H H mu. 131mm IN HH. 5.50 5.140 5.30 5.20 5.10 5.00 M90 ' h.80 b.70 . h.6o A h.50 him h.30. b.20- 14.10- T T I I r FIGURE 6 m1. DIHIE‘I'FR VS. TRISILE STRHWGI‘H A—ANNEAIEDTESTBARS 0‘” CAST TESI' BARS ..< T > 55%? / , \\ a x \ .- W//////// > Br MEASUREMENT SATURATIm / POSITION, VARIABLE DC POWER SUPPLY FIGURE 31 SC'HEMATIC DIAGRAM 0F COERCIVE FORCE METER Hc - Coercive force B - Relative magnetic retentivity (normal r component of the residual field measured at position shown) _ I! .enaccapm one agoans.puongo ago» on one accede oweaao» asp an gonad» casaomgua 22: .83838 :33 :2: H op mega—339.80 ..os._uu> 0.5339. «0 £50an mofiw no.3: H0380 :oapunflfido .0 H856 139.8» .8... .84..qu .p 825:80 80H I Do Enhances 05.533 3.530930 33965 oncomoduomo 33558 H856 Gasoflgom .8.“ “3.335 399350 $009.33. 532m 32$ coo... HNMJmW may 09 SSH—w Egg @2329 g 4.4 «Saga—Sue 0 Egg Egg NM EB“: m N H\ .Ag Hog 54% 95% > , a r. E0 or \Y o esp \ V m r T as e If SE : fill—lg mug m «a: .mxoo A8 fl who 00: 90 who 00 HO mmom gage OSCILIATCR E , FIGURE 33 CIRCUIT DIAGRAM OF ED—soo (3mm TO BID-300) RELATIVE HAWK: WM! mm ELMO REI'ENTIV'IT! V8. COEROIVE FCRCE 1&0 .. 13o . 120 110 70 50- ’40 100- 80 . 60- FIGURE 3h l l L l l 6 78 910 COERCIVE FOICE IN OMTEDS l L 1112 .mmfiaromvHx 2H mchaquQZH muazuu oomnnu.mc mZCHadoHQZH mu¢mmbm mm 55on owH omH CON OHN CNN as OJN 0mm 00m CNN 0mm omw oom OHM EVE. ED-BOO SURFACE INDICATIONS IN IKICROHAMPS. omm 3m omm own Sm 8m o3 8m o3 8N omw SN 08 0% 03 8a 8H 8H 0:“ o3 omH ofl oQ .mm:EIOMUHz 2H QZOHH<0HQZH muszuu conunuflncdfiflb< u q: aw d u d J- a d d 1 q u q u 7 d "“6392 fig 825 .3 mEHafiHQH §§§8 82.8 on nmucum d1 o.m or: o.m oé o.> o.m o.m 93 o.HH 0.3 o.ma 0.:H V o.mH 0.0H 1“! m: nan-‘1, 1‘“? mm. m‘.‘ FOERSTER~HOOVER QC-lOOO COMPARATOR TENSILE STRMGTH IN 1000 PSI. 1:2 ho- 38 36 32 . 30 28 ,26 2h 22 20 18 FIGURE 37 TENSILE STRENGTH VS. QC-lOOO SCOPE DEFLECTION FOR SEVEN SAMPLES A-ASCASTTESTBARS A-AININEAIEDTESTBARS 1 l L I L 4 L J n 1 j 1 1 L n 3.0 h.O 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.013.0 111.0 15.0 16.0 QC-lOOO SCOPE DEFIECTION IN CM. TENSILE STRENGTH IN 1000 PSI. he ho 38 36 3h 32 30 28 26 2h 22 20 18‘ FIGURE 38 TENSILE STRENGTH VS. COERCIVE FORCE £5— AS CAST TEST BARS <>- ANNEALED TEST BARS l l 1 I 1 J 1 h 5 6 7 8 9 10 COERCIVE FORCE IN OERSTEDS 11 13 .6 2H EOHBHmom‘ 853. 98.78 0.0..” o.m.n 0.4—” 0.9” 0.9” 0.4...” 0.0..” o.m 06 0;. 0.9 o.m 0.: o.m _ - d _ q 1 d _ u d d . _ fi _ . mg Emma. @4513 I q mm