THE UNIVERSITY OF MICHIGAN COLLEGE OF ENGINEERING Cast Metals Laboratory Final Report CASTING TECHNIQUES FOR CUNISIBE METAL D. R. Askeland PO K. Trojan J. Lo Herron R. A. Flinn ORA Project 01086 under contract with: OFFICE OF NAVAL RESEARCH CONTRACT NO. N00140-67-C-0503 CHICAGO, ILLINOIS administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR August 1968

TABLE OF CONTENTS Page LIST OF TABLES v LIST OF FIGURES vi SUMMARY x INTRODUCTION 1 EXPERIMENTAL PROCEDURE 2 RAW MATERIALS 2 MELTING AND POURING 2 HOLDING 3 INSPECTION 3 SPECIAL CASTINGS4 1. Fluidity4 2. Hot Tear Susceptability4 3. Thermal Gradients (a) Tangent (b) Secant 4. Riser Size and. Riser Placement 6 5. Production Casting 6 RESULTS AND DISCUSSION 8 MELTING, POURING, AND SAFETY PRECAUTIONS 8 GENERAL CASTABILITY OF CUNISIBE 8 1. Fluidity 8 2. Hot Tears 9 3. Thermal Gradients 9 RISER SIZE AND PLACEMENT 11 1. Feeding Distance 11 (a) Bar Sections 11 (b) Plate Sections 12 (c) Pressure Testing 13 2. Riser Size 14 (a) Side Riser -Insulating Topping 15 (b) Top Riser - Insulating Topping 15 (c) Top Riser - Insulating Topping and. Insulating Sleeves 15 (d) Top Riser - Exothermic Topping 16 (e) Top Riser - Exothermic Topping and Exothermic Sleeves 16 PRODUCTION OR EXAMPLE CASTING 17 iii

TABLE OF CONTENTS (Concluded) Page CONCLUSIONS 20 BIBLIOGRAPHY 21 iv

LIST OF TABLES Table Page I. Chemical Analysis of Cunisibe Metal Ingot 22 II. Summary of Heats 23 III. Feeding Distance Data for Bar Castings of Cunisibe 27 IV. Feeding Distance Data for Plate Castings of Cunisibe 29 V. Riser Size Data: Side Risering, Insulating Topping 30 VI. Riser Size Data: Top Risering, Insulating Topping 31 VII. Riser Size Data: Top Risering, Insulating Topping and Insulating Sleeves 32 VIII. Riser Size Data: Top Risering, Exothermic Topping 33 IX. Riser Size Data: Top Risered, Exothermic Topping and Exothermic Sleeves 34 V

LIST OF FIGURES Figure Page 1. Plate casting with a large top riser which shows the slice used for radiographic examination. 35 2. Section from a side risered bar with a molybdenum mesh dross trap between the casting and riser. If the mesh was inserted. in the gating system the shrinkage defect would. not occur. 36 3. A typical fluidity spiral cast in 70-30 cupro-nickel. The particular spiral was cast with 250~F superheat with a resultant length of 25". 37 4. A typical hot tear casting assembly. One end of each leg has a 3/4" radius while the other radius progressively varies from 1/16" to 3/4" dependent upon the casting position. Each leg is 7" long. 38 5. 1-1/2" x 1-1/2" x 10" end. chilled bar. Gating and risering were similar to that used in the standard 2" x 2" x 12" castability bar (thermal gradient investigation). 39 6. Cooling curves for a 2" x 2" x 12" bar; end. chilled. Distances are referenced to the chilled end. No. 1 station is 1" from the chill or 11" from the riser. (Heat No. 319) 40 7. 1-1/4"-90~ elbow drag cavity. The end chills are in place however the cores are removed to show the gating system. 41 8. Open cope portion of the production casting. Anchors are attached to the cope chills to prevent their movement during mold handling. The top riser neck is attached at the reentrant corner. 42 9. Drag section of the 1-1/4"-90~ ell casting with all cores in place. Cores were made of CO2 sand. while the mold itself was dry sand. 43 10. Fluidity Curves for Cunisibe and 70-30 cupro-nickel. Cunisibe liquidus - 2110~F; Cu-Ni liquidus - 2240~F. 44 vi

LIST OF FIGURES (Continued) Figure Page 11. X-ray of a Cunisibe hot tear casting with a 1/16" radius cast in a C02 mold. The casting did not tear even though the fillet was smaller than would be considered, good foundry practice. 45 12. Cooling curves for 70-30 cupro-nickel cast in a 2" x 2" x 12" bar with an end. chill. (Heat 320) 46 135 Thermal gradient curves for a 2" x 2" x 12" bar chilled. at one end. The gradient has been determined by the secant method, at two temperatures. 47 14o Radiograph of 1/2" vertical thick slice taken from the center of the bar. (X1) TCx refers to the thermocouple station number. 48 15. Cooling curves for a 2" x 2" x 10" end. chilled bar cast from Cunisibe alloy. The more rapid. cooling rates can be compared. with a 2" x 2" x 12" bar as in Figure 6. 49 16. Thermal gradients for a 2" x 2" x 10" end chilled bar of Cunisibe. The data have been calculated, by the secant method. from the cooling curves in Figure 15. 50 17. Feed. distance castings poured when small bars were studied (less than 1/2" x 1/2" cross section). 51 18a, Feeding distances in unchilled bars for Cunisibe alloy and. 7050 cupro-nickelo 52 18b. Feeding distances in end. chilled, bars for Cunisibe alloy and. 70-30 cupro-nickel. 55 19. Feed distance castings poured, when small plates were studied (for 1/2" thicknesses only). 54 20a. Feeding distance in unchilled. plates of Cunisibe alloy. 55 20b. Feeding distance in end. chilled, plates of Cunisibe alloy. 56 21. Summary of feeding distance in plates and. bars in Cunisibe alloy. Bar data for 70-30 cupro-nickel have also been included. from reference (7). 57 vii

LIST OF FIGURES (Continued) Figure Page 22. Pressure test of selected bars used. in the feed. distance study. The leak rate was determined at lOOpsig dry nitrogen pressure. 58 23. Volume ratio (Vr/Vc) vs the riser height to diameter ratio (H/D) for a 2" x 2" x 8" side risered bar. 59 24. Volume ratio (Vr/Vc) vs the riser height to diameter ratio (H/D) for a 2" x 2" x 8" top risered bar. 60 25. Risering curve for plates and. bars of Cunisibe alloy cast with an insulating cover on open risers. 61 26. Top risered bar which shows the flow-off channel used to maintain a constant (H/D) ratio for the riser. 62 27. Top risered plate which shows gating and the use of the flowoff technique. 63 28. Volume ratio vs H/D of the top riser when insulating topping and insulating sleeves have been used. 64 29. Risering curves for plates and bars when insulating topping and insulating sleeves are used. H/D values were 0.50. 65 30. Volume ratio vs H/D ratio for castings top risered with exothermic topping only. 66 31. Risering curve for top risered casting made with exothermic topping. H/D = 0.50. 67 32. Volume ratio vs H/D ratio of the riser when both exothermic topping and exothermic sleeves are used. 68 33. Risering curve for use when exothermic sleeves and. exothermic topping are used. H/D = 0.50 69 34. Summary of the volume ratio (Vr/Vc) vs height to diameter ratio of the riser (H/D for several riser treatments. 70 35. Summary of risering curves for both Cunisibe alloy and 7030 cupro-nickel. 71 viii

LIST OF FIGURES (Concluded.) Figure Page 36. 1" x 5" x 5" plate cast with insulating riser topping. The casting was unsound. as can be seen visually and verified. radiographically. 72 37. 1" x 5" x 5" plate cast with insulating topping plus insulating sleeves. A shrinkage pipe was not evident however the casting is radiographically unsound. 73 38. 90~ elbow and its equivalent section used. as an illustration of the riser data. 74 39. 1-1/4"-90 elbow cast in three different ways to illustrate use of the feeding distance and risering curves. 75 40. Sectioned. 90~ elbow which was risered at one end. and chilled. at the other. The surface shrinkage in the cope portion was due to inadequate feeding. 76 41. A riser adequate for exothermic topping is inadequate in height for insulating topping. The riser shrinkage has been pulled down into the casting 77 42. A correctly risered casting as also shown by radiography. The surface blemishes were due to some metal reaction with the uncoated CO2 sand core. 78 ix

SUMMARY Many cast shapes are produced. by an expensive time-consuming cut and try method. With the development of the new Cunisibe alloy, a complete analysis of the castability was desirable in order to give the foundryman the information necessary to produce high integrity castings. The present research has provided the necessary castability data which ranged from melting practice to riser size determinations. Since Cunisibe is found. to be an alloyed. version of cupro-nickel, a comparison has been made wherever possible. In almost every instance the general castability of Cunisibe has been found. to be somewhat superior. Finally the data have been applied. to a cast pressure fitting to indicate the production utility of the information under various foundry practices. The data can therefore be immediately used to produce radiographically sound. cast components of Cunisibe metal. x

INTRODUCTION In the quest for a higher strength copper base alloy, the Navy developed a modified cupro-nickel alloy with the following composition: Cu Ni Si Be Fe Mn e Rem. 29-50 0.55/0.65 0.40/0.6 0.080/1.00 1.25/1.15 C P S P Weight %o 0.-15 0.005 0.01.max. 002 max. max. max o Since the initial intent for the alloy was as a cast material, it was necessary to develop casting design''information and compare it to the better known 70Cu-30Ni alloys. The object of the research was therefore to provide castability information on Cunisibe metal in order to allow the foundry to produce high integrity castings. The report evaluates melting and. pouring practice, solidification characteristics, fluidity, hot tear susceptability, riser size, riser placement, and. use of various exothermic componds. Finally the results have been applied to a small ell casting to evaluate the data utility in a production situation. 1

EXPERIMENTAL PROCEDURE The general procedure is detailed, under the following major headings: (1) Raw Materials (2) Melting and Pouring (3) Molding (4) Inspection (5) Special Castings RAW MATERIALS The initial experimentation was conducted from Cunisibe metal ingot supplied by the Naval Applied. Science Laboratory. Three separate ingot casts were forwarded as indicated in Table I. After several heats the virgin material had. been exhausted. which required. a charge mad.e up of remelted. stock. A complete list of input materials, castings poured, and. the ultimate chemical analyses has been included. in Table II. Since over 4000 lb of castings were produced. from the original 500 lb of ingot, a large number of remelt cycles were required.. The chemical analyses of several cupro-nickel heats have also been included in Table II. MELTING AND POURING A 3000-cycle, induction heated, lift coil was used with a number 60 graphite crucible. Melting and. pouring was therefore conducted in the same crucible. Throughout the melting cycle, dry nitrogen was introduced through a hole in the graphite crucible coyer in order to minimize the oxidation and. vaporization loss of beryllium. The nitrogen flow rate was maintained, at approximately 0~3 standard cubic feet per minute. No deoxidant was used in the melting practice. A standard pouring temperature of from 2300-2350~F was used for all heats. The temperature was recorded. on a strip chart with an immersion Pt-Pt 10% Rh thermocouple. Expendable fused silica protected thermocouples were found, to offer the most economical means for temperature measurement within the designated. temperature range. Since different castings were poured from each heat, the pouring times necessarily varied. However, 100-lb melts were generally poured off into 3 or 2

4 molds in approximately 30 secso The short pouring times were required in order to minimize temperature loss from the beginning to the end of the pour. HOLDING Dry sand. molds have been used. throughout the investigation and. had, the following composition for each batch: 140 mesh N.J. washed silica sand. 152 lb Western bentonite 8 lb (4. 0) Dextrine 2 lb (1.20o) Water 8 lb (4. 0) New sand. was mixed. each time to be used for facing; old. sand. was tempered. with water only and was used, for backing sarido Several of the specialty castings had. aluminum patterns and, were hand, molded. All bars and. plates poured for the riser size and placement used, wood patterns and. were also hand, molded. After ramming the molds were baked, for 12 hours at 300~F. When insulating and. exothermic riser sleeves were employed, there were rammed into place during the molding sequence and. were also exposed to the baking cycle. This was considered, to be advantageous since these materials have a tendency to pick up moisture upon prolonged. storage. In the event chills were used, they were machined from graphite and were also rammed into place. Further explanation of the procedure is included under "Special Castings". INSPECTION In numerous instances, the casting soundness could. be ascertained from a simple visual inspection of a slice through the casting. However borderline riser sizes and, feed. distances required examination by radiographic techniques. A 1/2" vertical slice was first cut from the centerline of these castings with a large hydraulically-operated cut off wheel (see Figure 1). The pieces were then radiographed at 180 KV and 10 ma. for 15 sec. Although Cunisibe alloy has not been found. to be a mushy alloy (very large liquidus to solidus temperature difference), there is often difficulty encountered in radiography interpretation due to microshrinkage. In order to evaluate the radiography sensitivity, samples were also pressure tested. Several 1/2" thick pieces were machinecl to approximately o.060" thicknesses first. (Although. 030'" thickness is more desirable, machinability difficulty with a shaper prevented thinner sections. ) Dry nitrogen gas at 100 psig was then introduced to one side of the sample through an "0" ring seal. Any gas leakage was detected. by passage through a gas burette. In this way a leak rate could be determined for various portions of the castings and, correlated with shrinkage detected in 5

the radiograph. The area pressure tested. was 3/8" or 3/4" in diameter, depending upon the casting tested. A complete discussion of the technique and its sensitivities is contained in the literature (1) (2). In some of the preliminary heats, difficulty was encountered with dross. A molybdenum mesh screen was inserted between the riser and. the casting which resulted. in shrinkage (Figure 2). The screen would undoubtedly work better in the gating system although dross was found to cause no difficulty if the metal was skimmed before pouring SPECIAL CASTINGS A number of special techniques were applied to various castings in ord.er to study the several indices of castability such as fluidity, hot tears, thermal gradients, riser size, riser placement, and production shapes. These casting procedures have been included separately due to their unique applications. 1o Fluidity A standard fluidity spiral as shown in Figure 3 has been cast for both Cunisibe and 70-30 cupro-nickel. The spiral casting has been designed to maintain a constant head. upon pouring. Metal is poured into one side of the dual pouring basin where metal in excess of that required, to fill the spiral "flows off" into the other pouring basin. Thus a constant metal head has been maintained with the length of the spiral determined only by the metal "flowability." The spiral length is measured in inches from the 2" bosses cast into the spiral. The length can then be correlated with degrees super heat above the liquidus. In all cases the pouring temperatures and liquidus temperatures have been measured. with a Pt-Pt 10% Rh immersion thermocouple protected. within a fused. silica tube. Again a permanent temperature record. was made on a strip chart potentiometer. Although a drawing of the fluidity spiral has not been includ.ed in the report, the particular spiral used. has been completely described in conjunction with other copper base founding principles (3). 2o Hot Tear Susceptability A standard technique for aluminum alloys has been employed for both Cunisibe and. 70-30 cupro-nickel to evaluate the susceptability to hot cracking (4). A sample set of castings is shown in Figure 4. The "U" shaped. series of bars are identical except for a change in the radius at the reentrant angle closest to the sprue. The radii have been varied from 1/16" to 3/4". The mold material between the legs of the "U" inhibit contraction of the metal

during solidification of the last liquid. Since the strength of the composite liquid-coherent solid has been found. to be low, a small constraint could. cause hot tears. For a given casting material, the degree of cracking will be depend.ent upon the stress concentration provided by the reentrant angle and the constraint provided by the mold. material. A hot crack index is then the smallest radius of the fillet at which the metal will indicate a crack. All radii smaller than this value would show a crack whereas larger radii should inhibit the hot crack formation. Since mold. materials offer various degrees of constraint, CO2 sand. was used. for one series in order to compare the results with the normal dry sand molds previously described. The 140 mesh sand. was bonded with 50 sodium silicate and. rammed. in the normal was. The mold. was not gassed. with C02 however, but rather was dried. for 2 hours at 250~F just before pouring. The resultant mold. was very hard with high room temperature strength. 35 Thermal Gradients It has long been appreciated that directional solidification toward a riser results in a thermal gradient which reaches a minimum somewhere within the casting geometry. In an attempt to correlate cast material, thermal gradient and shrinkage occurrence, a castability bar 2" x 2" x 12" chilled on one end has been developed (2) (5) (6). A large riser (4" diax 6" high) has been provided with bottom gating through a tapered. sprue. The casting configuration and gating system were very much similar to those shown for the smaller bar in Figure 5. Six fine chromel-alumel (28 gauge) thermocouples protected from the liquid. metal by 1 mm bore fused silica tubes are placed. at various distances from the chilled. end. The 6 cooling curves are then continuously recorded at the centerline but at different distances from the chilled. end. (as an example ", 3", 5", 7", 9", and 11" from the chilled end. Since the bar is 12" long, 11" from the chilled end is 1" from the riser). From these cooling curves, the thermal gradient can be obtained at either the solidus temperature or the midpoint between the liquidus and solidus (both temperatures have found favor in the literature as being the point where restriction to liquid metal flow results in shrinkage). Two techniques have been used. in the past to determine the thermal gradient. (a) Tangent. By definition the thermal gradient has units of ~F/in. or is the slope dT/dXo Therefore the cooling curves must be replotted as temperature versus thermocouple position where time is a constant parameter. The thermal gradient is then the slope of the iso-time curves at the temperature of interest (solidus or solidus-liquidus midpoint) (5). The technique, though exact, is extremely time consuming and incorporates further error due to the need for the replots of the data. (b) Secant. A much more simple method requires direct use of the cooling curves with no replot of data. As an example, the thermal gradient at the 5

solidus for a given position is obtained. by determination of the temperature at the nearest hotter position and. dividing by the distance between the two positions. Assume a solidus of 1940~F (Cunisibe) (see Figure 6), the thermocouple at 3" from the chill reached 1940~F at approximately 400 sec. At the same time (400 sec.) the next hottest couple was at 1970~F (5" from the chill). The thermal gradient at the 3" position can then be approximated as: 1970~F - 1940~F = 15~F/in. 5"? - 3" This technique gives a secant rather than a true tangent, however due to its simplicity the method has been employed for the present investigation, to represent the thermal gradient. 4o Riser Size and. Riser Placement Development of feed distance for plates and bars has been conducted with and without chills. Graphite chills were used with a thickness equal to the thickness of the cross-section of the casting. As an example, a 2" x 2" crosssection bar had a chill 2" x 2" x 2" whereas a 6" x 1" cross-section plate used. a chill 1" x 1" x 6". A bar has been defined as a section where the width is less than 3 time the thickness (W < 3T), whereas in a plate, W > 3T. Open risers were used throughout the study and were covered. with an insulating compound after pouring (Cuprit; Foseco EP 2534 B). The risers were larger than necessary in order to insure that shrinkage occurred due to "choking off" of feed. metal rather than an "inadequate supply" of feed metal. The minimum riser size was determined for bars and plates for several different sets of conditions as outlined below: (a) Side riser -insulating topping (Cuprit) (b) Top riser - insulating topping (Cuprit) (c) Top riser - insulating topping (Cuprit) and. insulating sleeves (Kalmin; Foseco precast sleeves) (d.) Top riser - exothermic topping (Feedol 9; Foseco) (e) Top riser - exothermic topping (Feedol 9; Foseco) and exthermic sleeves (Feedol 9; Foseco) It was then possible to evaluate the effectiveness of insulating and. exothermic compounds in order to maximize the casting yield.. Previous data from 70-30 cupro-nickel also allowed. a comparison with Cunisibe (7) (8). 5o Production Casting As a check on the applicability of the riser size and placement data, a small 1-1/4" - 90~ elbow was selected as an example of a production shape. 6

The casting was initially specified, to be cast in Navy "M1 and. withstand. 1200 psi gas or water pressure. Riser calculation was based, upon visualization of a lengthwise cut along the centerline and. flattening out as would be d.one if a tin can was cut along the seam and unrolled. to provid.e a single sheet. The resultant section was approximately 5-1/2" x 5-5/8" x 5/8" (a plate section). Three castings were made as outlined below. (In each case the riser had. the height equal to the diameter (2-1/8") and was selected to be 1/4" larger in both the height and. diameter than given by the minimum riser size curves for exothermic topping). (a) Top riser; both ends chilled; bottom gated to the reentrant corner; exothermic topping (Feedol 9); see Figures 7, 8, and 9 (b) End riser; one end, chilled; gated into riser; exothermic topping (Feedol 9) (c) Top riser; both ends chilled; bottom gated to the reentrant corner; insulating topping (Cuprit) 7

RESULTS AND DISCUSSION For ease of presentation, the data have been divided, into the following categories: 1. Melting, Pouring, and Safety Precautions 2. General Castability-Fluidity, Hot Tears, Thermal Gradients 3. Riser Size and Placement-Feed Distance for Bar Sand. Plates, Evaluation of Insulating and Exothermic Riser Compounds 4. Production Casting MELTING, POURING, AND SAFETY PRECAUTIONS The normal procedure has been described previously; however since beryllium is known to constitute a health hazard the initial experiments were monitored. by Professor W. A. Cook; Industrial Health; School of Public Health at The University of Michigan. The beryllium content in the various experimental stages is given below: Operation (Concentration micrograms/meter3) Melting (covered crucible with N2) 3 Pouring (through air) 70 Cutting (wet abrasive wheel) 300 It was interesting that melting constituted the least hazard whereas cutting was most hazardous. An attempt was therefore made to use other cutting techniques such as power hacksaws; however the operation was too slow. Therefore wet abrasive cut off wheels were used for all sectioning operations even though the beryllium toxicity level was high. Since the normal tolerance of 5 micrograms per miter3 was well exceeded in most instances, any person in the area has been required. to wear a filter type respirator approved by the U. S. Bureau of Mines for protection against fume. As an added precaution exhaust ventilation was used if the operation was sufficiently close to the intake system. GENERAL CASTABILITY OF CUNISIBE 1. Fluidity Fluidity curves for both Cunisibe and 70-50 cupro-nickel are shown in 8

Figure 10 (Heats 320-523; Table II). It can be seen that Cunisibe has slightly higher fluidity than cupro-nickel. The equations of the curves are: Cunisibe - Spiral length = 0.120x (~F Superheat) Cu-Ni - Spiral length = 0ol02x (~F Superheat) where ~F Superheat = pouring temperature -2110~F (Cunisibe) = pouring temperature - 2240~F (Cu-Ni) The slight difference in fluidity may possibly be attributed to the occurrence of an eutectic in Cunisibe. It is a well-known phenomenon that an eutectic and a pure material will generally show higher fluidity than alloys which exhibit a broad liquidus to solidus distance. However the difference between the two alloys as tested is so slight as to be insignificant from the point of view of castability. 2. Hot Tears The hot tear castings as shown in Figure 4 did not indicate cracks for either of the alloys for even the smallest radius of 1/16". Such was the case for both dry sand. and the more rigid C02 sand molds. This does not mean the alloys will not hot tear, since the use of still more rigid, mold materials may cause the experimental castings to crack. However a separate investigation of various investment mold materials would. be required to completely evaluate the hot tearing tendency under the most severe casting conditions. In conclusion it is of importance to realize that semi-rigid mold. materials will not cause hot tears in a "U" shaped casting with a 7" span with a poor foundry practice of less than generous fillets. Therefore, with normal practice, hot tearing should not be a problem as generous fillets, streamlined gates and runners, and streamlined casting geometry are most often used. As a matter of record an x-ray of the 1/16" radius Cunisibe casting in a CO2 mold is included in Figure 11 (Heat 361, Table II). It can be noted that shrinkage has occurred along the centerline due to inadequate feeding and at the 3/4" radius hot spot. It should also be pointed out that shrinkage would promote the occurrence of hot tearing; however a crack is not evident in this case. 3- Thermal Gradients The cooling curve for the 2" x 2"' x 12" bar cast of Cunisibe alloy was previously shown in Figure 6. Figure 12 shows the corresponding set of cooling 9

curves for 70-30 cupro-nickel (Heat 320). The liquidus and solidus temperatures for Cunisibe were approximately 2110~F and 1940~F, respectively (Figure 6). There was of course, a variation in these "arrest" temperatures from one thermocouple to the other due to the interdependence of constitutional supercooling and thermal gradient. The "arrest" at 1940~F was probably due to an eutectic as seen in a brief metallographic examination of the casting. When compared. to the cupro-nickel with a liquidus of 2240~F and, a 102~F liquidus to solidus range, the broader (1700F) range for the Cunisibe suggested. greater difficulty with soundness. This was not borne out by the feed distance relations as will be pointed out later in the report. Two temperatures of interest have been used in the literature; the thermal gradients at the solidus and half way between the liquidus and solidus. The practical temperature at which the thermal gradient would be most important would. be the point where the liquid. can no longer feed through the d.end.rites and shrinkage results. Since the degree of liquid-solid coherence as related to shrinkage is unknown and is undoubtedly dependent upon the alloy system, the solidus and liquid.us-solid.us midpoints have been assumed. to be of greatest interest. It would seem logical that somewhere between these two temperatures is the point of most significance which as yet has not been defined. In any event the thermal gradients at the two temperatures as determined by the secant method. have been summarized. in Figure 13. Therefore, the table below provides interesting comparison between the two alloys as obtained from the thermal gradient curves and X-rays. 2" x 2' x 12" Bar-End Chilled Cunisibe 70-30 Cupro-Nickel Liquidus Temperature 2110~F 2240~F Solidus Temperature 1940~F 2138~F Freezing Range 170~F 102~F Liquidus-Solidus Midpoint 2025~F 2189~F Gross Shrinkage (Measured from chill) 4-3/4" 6" Minor Shrinkage (Measured. from chill) None 4" Thermal Gradients Gross Shrinkage - Solid.us 7.25~F/in. 8.4~F/in. Gross Shrinkage - Liq.-Sol. Midpt. 2.5~F/in. 1.8~F/in. Minor Shrinkage - Solidus None 18.5~F/in. Minor Shrinkage - Liq..Sol. Midpt. None 5.5~F/in. As pointed. out previously, it appeared. that Cunisibe should. be more difficult to cast since it had a longer freezing range (170~F). However the Xrays indicate that the length of sound bar (2" x 2" cross section) is slightly 10

greater for Cunisibe than 70-30 (Figure 14). On the other hand the thermal gradient developed for Cunisibe is also greater which should, result in greater soundness. Figure 13 and the previous table suggested that a shorter bar would be sound. due to higher thermal gradient. Therefore a 2" x 2" x 10", end chilled castability bar was poured in Cunisiby alloy. The cooling curves and thermal gradients are given in Figures 15 and 16. Radiographs of the casting have indicated it to be sound which of course would be due to the higher thermal gradients. It can be seen that the minimum thermal gradient in Figure 16 is well above that where shrinkage occurred in the 2" x 2" x 12" casting discussed in the previous table (2.5~F/in. for shrinkage in the longer bar versus 7~F/in. in the shorter, sound bar at the solidusliquidus midpoint). Although the reasons for casting soundness or lack of it is dependent upon the thermal gradient, a more rapid utilization of the data is generally found through feeding distance relationships as presented in the following section. RISER SIZE AND PLACEMENT In a normal risering problem, two questions must be answered. First how large must the riser be in order that it freeze after the casting? Secondly, where should the riser be placed, in order that dendritic solidification of the casting does not choke off the feeding of liquid metal from the reservoir or riser? Since the preceding section dealt with the directional solidification or feeding distance, the latter question will be discussed first. 1. Feeding Distance (a) Bar Sections. A complete list of castings poured for the bar feeding distance has been included in Table III. In all cases the riser was larger than necessary in order to be assured that it froze last. Again insulating compound (Cuprit) was applied to the open risers, when small bars were poured, several where radiated from the same riser as shown in Figure 17. Graphical representation of the results are included in Figures 18a (unchilled bars) and. 18b (chilled bars). The maximum feed distance can best be summarized below in equation form: 11

Feed. distance in bars (FD) (T = bar thickness in inches) Cunisibe - Unchilled FD = 2T + 5" for T > 1" Cunisibe - Unchilled FD = 7T for T < 1" Cunisibe - End. Chilled FD = 2T + 6" for T > 1" Cunisibe - End. Chilled. FD = 8T for T < 1" 70-30 CuNi - Unchilled. (7) FD = 2T + 4" for T > 1" 70-30 CuNi - End. Chilled. (7) FD = 2T + 5" for T > 1" The maximum length bar which can be mad.e sound. in Cunisibe is therefore 1" longer than in the same cross section cast in 70-30 CuNi. Since the increased. feed. distance for the addition of an end. chill is 1", and end. chilled. 70-30 CuNi bar will give the same feed. distance as an unchilled Cunisibe casting. It is also note worthy that feed distance data for bars has seldom been carried below a thickness of 1" for any of the more common cast metals. However it is most logical that the upper linear portion of the curve must break and. proceed to the origin when the thickness has become sufficiently small. The reasoning is based. upon the extensive feeding resistance imposed by the dendrites rapidly growing from four sides. The best curve for the small thicknesses again appeared. to be linear as shown in Figures 18a and 18b with the break point at approximately 1" thicknesses. Since cupro-nickel data have not been carried below the 1" thicknesses, the Cunisibe advantage over cupronickel cannot be evaluated at the smaller sizes (7). Finally an attempt has been made to divide the feed distance relationships into an "end." and "riser" contribution as shown in the figures. The total feed distance is then the sum of the two individual contributions. The use of radiographs to determine these cutoff points does not allow for maximum accuracy, therefore the end. and riser contributions are shown as crosshatched. influences. That is to say the accuracy associated with the individual contributions were a degree of unsoundness must be determined, is not as well d.efined as the more simple choice; is it or is it not sound? As might be anticipated however, the addition of an end chill essentially only increases the end contribution and has little or no effect on the riser contribution. (b) Plate Sections. The summary of plate castings poured is given in Table IV. Plates smaller than 1/2" thick were not poured due to the small feed distances at these thicknesses. However when small thicknesses were poured, a gating technique similar to that for small bars was used (Figure 19). Again large risers and. riser insulating compound were used. The graphical results are shown in Figures 20a and. 20b. The corresponding equations for the straight lines are given below 12

Feeding Distance in Plates (T = Plate thickness in inches) Cunisibe - Chilled FD = T + 2-1/4" for T > 1/2" Cunisibe - Unchilled FD = 1/2T + 1-1/4" for T > 1/2" Unfortunately published data for plate sections are unavailable for 70-30 cupro-nickel. Therefore a comparison for the two alloys cannot be made. However, there is no reason to believe that Cunisibe will be poorer than 70-530 CuNi; since most previous information on the castability of the two materials has indicated, the opposite to be true. Again an attempt was made to divide the total feed distance into and end. and riser contribution with the associated difficulty. It should. be noted. that 1/2" thicknesses have not resulted in the break toward the origin as was the case with bars; therefore no attempt should be made to use the data below 1/2" sections. The addition of an end. chill has increased, the feed distance by approximately 1" as with bars, where the increased feeding is felt basically in an increased end contribution. A comparison between plates and bars can best be made from the summary curves in Figure 21. The plates have somewhat less than one half the feed. distance of that found. in bars of the same thickness. This general trend is obtained in most cast materials due to the dendritic growth which chokes off feeding in the relatively thin plate sections. (c) Pressure Testing. As pointed out previously, the radiography sensitivity was checked by pressure testing 0.060" thick pieces cut from the 1/2" thick radiographic bar samples. The results of the pressure test are included in Figure 22. Radiographs of 2" x 2" x 9" unchilled and 1" x 1" x 8" chilled bars had. shown the castings to have marginal soundness. These castings were just below or on the feed distance curves shown in Figures 18a and 18bo Even though a cupro-nickel 0.0140" penetrameter with a 0.0310" hole was distinguishable in the radiograph, the possibility of shrinkage placed the castings in the marginal category. The pressure test results have verified the choice and the leak rate was smallo On the other hand the existence of easily distinguishable shrinkage as in the 1" x 1" x 8" unchilled bar resulted in much higher leak rates as compared to the chilled bar. Since some difficulty was anticipated with dross in these two bars, both a 3/8" and. 3/4" diameter area was pressure tested. It should be noted that the maxima occurred, at the same distance from the riser with the larger area giving a higher leak rate. Therefore gross radiographic unsoundness has given very high leak rateso As the radiograph became more difficult to interpret the leak rate decreased. The "soundness'V in the present study is that determined by radiography. It is entirely possible that pressure testing in thin sections may result in leakage 13

even though the casting appeared. radiographically sound. However, radiographs of the 0.060" sections all showed unsoundness as shown in the pressure tests. Soundness then has become a problem in definition. It would appear that Cunisibe has a tendency to disperse shrinkage, which made radiography of the 1/2" sections difficult to interpret, when the castings approached, the minimum feed. distance. Therefore the feed. distance curves should not be used without the addition of a margin of safety. Finally it should be mentioned. that 70-30 cupro-nickel castings are somewhat easier to treat radiographically. The more clear cut difference between shrinkage and soundness may be due to the shorter liquidus to solidus temperature range (102~F vs 170~F for Cunisibe). 2. Riser Size All of the risers used. in the feed. distance study have been larger than necessary in order to guarantee a high thermal gradient at the riser end. Therefore shrinkage occurrence would, be due to dendrite choking of metal flow within the casting and. not due to an inadequate supply of liquid. metal in the riser. The feed. distance curves then provide information as to riser placement. The present information deals with optimum riser size. Before proceeding to the risering curves, several terms are defined below: Vr - Riser Volume Vc - Casting Volume H - Riser Height D - Riser Diameter L - Casting Length W - Casting Width T - Casting Thickness Vr/Vc - Volume Ratio (L+W)/T - Casting Shape Factor Several variables which effect the riser size have been investigated. as pointed out in the experimental procedure. These will be discussed in the following sequence: (a) Side riser - insulating topping (b) Top riser - insulating topping (c) Top riser - insulating topping and insulating sleeves 14

(d) Top riser - exothermic topping (e) Top riser - exothermic topping and. exothermic sleeves (a) Side riser - insulating topping. In the past, various researchers have pointed, out the importance of the height to diameter ratio of the riser (H/D). Unfortunately there has been very little consistency as to how it should be measured. Since the volume ratio (Vr/Vc) should be minimum for maximum efficiency and. economy, the H/D ratio should. be determined as a function of the volume ratio. The dependence of height to diameter ratio upon the volume ratio is shown in Figure 23. The corresponding castings poured. are given in Table V. Shrinkage occurred in these side risered bars (2" x 2" x 8") when the riser became too small to maintain a sufficiently high thermal gradient. The pipe seldom ended up in the casting, but rather centerline shrinkage appeared in the casting in very much the same way as with inadequate feeding distance. It should be pointed. out that a 2" x 2" x 8" long bar is within the feed. distance relationship if the riser is sufficiently large to maintain the necessary thermal gradient (greater than 5~F/in. at the solidus-liquidus midpoint). Here again a higher H/D ratio for the riser would provide the necessary gradient. The shape factor versus the volume ratio for side risered castings was not determined since it became evident that the shape factor could not be varied over a significant range due to the feeding distance limitations. Therefore top risered castings were used. for the remainder of the riser size determinat ion. (b) Top riser - insulating topping. The interrelationship between the volume ratio and (H/D) is shown in Figure 24. The limiting value for soundness would be when the shrinkage pipe is pulled down into the casting. Therefore, higher H/D ratios would inhibit this occurrence as shown. In general the most economical and practical H/D ratios fall between 0.5 and l 0 as seen in Figure 24. Lower values decrease casting yield whereas higher ratios begin to lose their effectiveness as seen by the decreasing slope at H/D values greater than 1.0 Therefore the actual riser size can be determined from the curves in Figure 25 for H/D values of either 0.5 or 1.0. The data can be used. for either plates or bars where insulating topping is used. on open risers. The complete summary of castings poured in this series has been given in Table VI. It should be pointed out that H/D ratios were maintained. by the use of "flow-offs." The casting rigging is shown in Figures 26 and. 27. (c) Top riser - insulating topping and insulating sleeves. The volume ratio versus H/D and the volume ratio versus the shape factor curves are given in Figures 28 and. 29 with the castings poured enumerated in Table VII. It will be noted. that the conjunctive use of insulating topping plus sleeves has shown a lesser interdependence between the volume ratio and the H/D ratio than when 15

insulating topping was used. alone (Figures 24 and 28). Therefore an H/D value of 0.50 was used to establish the risering curve shown in Figure 290 (d.) Top riser - exothermic topping. The working graphs and tabular data are included in Figures 30 and. 31 and. Table VIII. Again (H/D) values of 0.50 were chosen to establish the risering curve. As a point of interest, a casting with a shape factor of 5 0, Vr/Vc = 0.6, and H/D = 0.50 was made in the normal manner; however Feedol 1 (less active) was used for the exothermic topping rather than the normal Feedol 9 (more active). Whereas the risering curve (Figure 31) would show the Feedol 9 casting to be marginal, the Feedol 1 casting was definitely unsound with rather gross shrinkage. Therefore all exothermic compounds are found. not to be the same, and difficulty may be anticipated. when risers are selected which fall close to the minimum sizes advised. by the curves. Furthermore, past experience has shown active exothermic compounds to become more inactive with long storage, possibly due to moisture pick up. (e) Top riser - exothermic topping and exothermic sleeves. Graphical representation of the results are presented. in Figures 32 and 33. Tabulation of the castings poured, is given in Table IX. The general trends are seen to be quite similar to those previously discussed. Perhaps the best comparison of the experimental variables can be appreciated from Figures 34 and 55 which summarize all of the riser size data. The experimental points have not been plotted but rather only the average curves derived from the data presented in Figures 23-33. Data have also been plotted for 70-30 cupro-nickel (8), which provided an interesting comparison with Cunisibe alloyo Many conclusions could be drawn from the summary curves, however the following are considered to be of greatest significance: 1. Side risers are not considered very economical since the feed distances can easily be exceeded and large volume ratios are required. 2. For Cunisibe, exothermic topping gives better yield than if only insulating topping is used. Insulating topping plus insulating sleeves is as good. if not better than exothermic topping and exothermic sleeves. (The possible production of toxic beryllium fume is also decreased. when insulating compounds are used. ) 3. Little is gained by the addition of exothermic sleeves to exothermic topping for Cunisibe whereas a large increase in yield. is to be appreciated for 70-30 cupro-nickel. Therefore a higher yield is evident for Cu-Ni when exothermic are used except when only an exothermic topping is used which provided Cunisibe with a higher casting yield (a lower volume ratio for a given casing geometry~ 16

These results are difficult to rationalize since the cast materials possess different thermal blocks due to different phase equilibria upon solidification. It is also possible that the exothermic compounds used in the cupronickel study (8) were of a different activity level than those in the present work. Finally, the solidification and occurrence of a riser pipe suggested a source of error as shown in Figures 36 and. 37. The use of only insulating topping gave a pipe in the riser from which shrinkage could. be easily found. radiographically if not by visual examination (Figure 36). On the other hand, piping was not evident for the other combinations of riser compounds as exemplified by the insulating topping and insulating sleeves in Figure 37. The casting was radiographically unsound, however a casting which had been radiographically sound may still not be pressure tight in very thin machined sections as discussed previously. Therefore some variation in the risering curves (Figures 34 and 35) should be anticipated. It would. be possible to have particular combinations of riser compounds disperse the shrinkage to the point where radiography would, not reveal a defective casting. These interpretations could only be evaluated through an extensive evaluation of pressure testing, thermal measurements and radiography. However if the next riser size higher is used. as determined from the risering curves, the castings would indicate X-ray quality sound castings. PRODUCTION OR EXAMPLE CASTING The incorporation of this section has been to illustrate the utility and flexability of the data; in particular the riser size and placement information. Therefore a rather complete analysis of the problem has been included. First, consider the casting geometry as shown in Figure 38. Through straightening of the 90~ ell and. then unrolling of the cylinder, a plate has been approximated with dimensions of 5-1/2" x 5-5/8" x 5/8". Equivalent length = 4-1/8" + 1-1/2" = 5-5/8" Equivalent width = median circumference = 5-1/2" The volume of the casting (Vc) and the shape factor ((L+W)/T) can then be calculated. Secondly, it was decided to use H/D values for the riser which would be equal to one. Although the curves have been given for insulating topping (Figures 24 and 25), they had not been provided for exothermic topping which was to be used for two of the castings. However a replot of points from Figure 30 for H/D = 1.0 on Figure 31 would give a new risering curve for exothermic topping. It should be noted that this is only possible because the risering curves are parallel when the H/D value is changed, as seen in Figure 25. The summary of data used for the minimum riser determination is therefore 17

given below when (L+W)/T is extrapolated to 17.8. Type HD Vr/Vc H and D Feeding Distance* Insulating Topping 1.0 0.73 2.6" 1.5" Exothermic Topping 1.0 0.265 1.9" 1.9" Three castings were poured from the same heat to illustrate the following: 1. Correct riser size with inadequate feeding due to riser placement at one end. of the ell with the other end chilled. 2. Adequate feeding with a top riser at the -reentrant corner, however with an inadequate size riser. 3. Correct riser size and placement. The three finished castings with the gates removed and the risers still in place are shown in Figure 39. In each case the riser attachment area was decreased. to facilitate riser removal. 1. Consider first the correct riser size with poor riser position. The riser was to be covered with exothermic topping; therefore with (L+W)/T = 17.8 (Figure 38) and H/D = 1.0 (interpolation from Figures 30 and 31) the. volume ratio Vr/Vc is equal to 0.265. Since the Vc = 19.4 in., Vr = 5.14 in. or H = D - 1.9 which is the minimum riser size. If a safety factor of 1/4" is added t6 this riser, the size would be H = D = 2-1/8". The feeding distance of a riser placed at one end of a 5/8" thick plate chilled on one end would be: FD = T + 2-1/4" h 2-7/8". The total feeding from the edge of the riser would then be: D + 2-7/8" = 2-1/8" + 2-7/8 = 5" whereas the casting is 5-5/8" long. Therefore the feeding would be inadequate. A section of the casting is shown in Figure 40. As might be expected, the shrinkage has broken through to the cope surface of the cored section and is most prevalent at the hot spot or reentrant corner. 2. Since the riser placement above was incorrect, assume the same riser size (H = D = 2-1/8"); however centrally located the riser at the corner. The potential feeding distance is 2-7/8" if both ends are chilled. The riser would feed 2-7/8" + 2-1/8"/2 or approximately 4" in either direction which of course is much longer than required. *Feeding Distance = 5-5/8" - D/2 for the top riser centrally placed (see Figure 38). 18

However assume insulating topping to be substituted. for the exothermic. From Figure 25 at H/D = 1.0 Vr/Vc = 0.73-with Vc = 19.4 in. 3; H = D = 2.6" whereas only a 2-1/8" riser has been used. The result is clearly shown in Figure 41. The shrinkage pipe has been drawn down into the casting. A 2-3/4" riser (H = D) with insulating topping would probably have given a sound casting. 3. Finally, the use of a 2-1/8" diameter riser top placed. at the corner with both ends chilled. would give a sound casting if exothermic riser topping was used. The resultant casting is shown in Figure 42. A radiograph of the sectioned component also indicated it to be sound. Some surface imperfection was evid.ent at the cored surface, however a core wash would. undoubtedly alleviate the situation. It is therefore evident that the castability data obtained. for Conisibe alloy can be applied. to a production shape. It is then possible to consistently produce high integrity castings from this material. 19

CONCLUSIONS The comparison between Cunisibe alloy and 70-30 cupro-nickel has indicated. a similarity between the two materials. In almost every instance the castability for Cunisibe has been shown to be somewhat superior. These indices have included: 1. Slightly higher fluidity. 2. Low susceptability to hot tears. 3. Lower thermal gradients to produce radiographically sound section. 4. Longer feeding distance in bars, 5.Smaller size risers to produce a radiographically sound section (dependent upon riser compounds). Supplementary data have also been generated. for melting and. pouring practices, feeding distances in plate sections, riser sizes for various combinations of insulating and exothermic compounds, and pressure testing of selected thin sections. Finally all of these data have been applied to a small illustrative casting. Therefore a complete set of castability data for Cunisibe alloy has resulted which would direct the foundry man to the production of high integrety castings. The authors wish to acknowledge the excellent laison with Mr. M. L. Foster and Mr. J. R. Crisci of the U. S. Naval Applied Science Laboratory. 20

BIBLIOGRAPHY 1. Trojan, P. K. and. Flinn, R. A., "Pressure Tightness in 85-5-5-5 Bronze Castings," Trans. AFS, 64, 339-343 (1956). 2. Trojan, P. K. and Flinn, R. A., "Pressure Tightness in 85-5-5-5 Bronze Castings," Trans. AFS, 65, 238-246 (1957). 3. Flinn, R. A., "Copper, Brass, and. Bronze Castings; Their Structures, Properties, and Applications," Non-Ferrous Founders Society, 121 (1963). 4. Gamber, E. J., "Hot Cracking Test for Light Metal Alloys, " Trans AFS, 67, 237-241 (1959). 5. Flinn, R. A. and Mielke, C. R., "Pressure Tightness of 85-5-5-5 Bronze," Trans. AFS, 67, 385-392 (1959). 6. Flinn, R. A. and Kunsmann, H., "Copper Base Casting Alloys-Physical Properties and. Void Volume Correlation with Solidification," Trans. AFS, 69, 208-220 (1961), 7. Weins, M. J., de Botton, J.L.S., and Flinn, R. A., "Data for Calculation of Riser Placement in Copper Alloy Castings," Trans. AFS, 72, 832-839 (1964). 8. Rote, F. E., Guichelaar, P. J., and Flinn, R. A., "Riser Design for Copper Alloys of Narrow and Extended Freezing Ranges," Trans. AFS, 74, 380-388 (1966). 21

TABLE I CHEMICAL ANALYSIS OF CUNISIBE METAL INGOT Ingot Identification BRO - 101 BRO - 102 BRO - 103 Element Avg. Range' Avg. Range Avg. Range Cu 65.42 65.36-65.47 65.47 65.26-65.64 65.92 65.88-65.99 Ni 30.60 30.48-30.82 303.2 30.04-30.70 30.11 30.00-30.22 Mn 1.32 1.30- 1.34 1.35 1.33- 1.37 1.29 1.28- 1.31 Fe 1.07.99- 1.11 1.01 1.00- 1.02.99.94- 1.04 Pb <. 01 <. 01 <. 01 <. <. 01 <. 01 P <. 01 <. 01 <. 01 <. 01 <. 01 <. 01 S.006.003-.009.005.00-.007.006.003-.009 Si 1.02.98- 1.14 1.08 1.05- 1.11 1.09 1.05- 1.13 Be.54.53-. 56.57.51-.62.52.48-.54 C.02.02-.03.02.02-.03.02.02-.0 Cb <.01 < 01 <.01 < 01 <.01 <.01 Amount Shipped 167 165 170 (lb) 22

\10 L(I\ r-H CU Q) C CM 0 6C6 6 6666 V LI\ _ co o \ oC K\ COJ C 00 00 00 00 o o oo66 oo66 c O 6 r6o r r- U) HCM O\OD o?i K\ C.11m r< Cn N\ CN _C C r K\ CM D 6 (3 H~ r6 6 H rH HH 1-A C WI\\.CC )- \ t-. cC.- K\ H C) C O ~ *H Lr\L\ -Kt\- HOHH\ HH 00000 00 ONO <i.H ir l _t- 4 t - ri ojr-Ir- o b o oo o oo \ C 66 H HHlrHi rHi r4 ri r i r4 r4 r4r r4 rH r4 66 0 0\\ CJ-t Ck Oi-:0 f o c — K\-,i- LC\ iCO r- mq 66 o6666666666666 666o o o d oo ioo 00 00 00 f \\ ON st Hl (V-D Hl rl-L0C nH cocc 000-3t_ P1 0 0\C o \o, r4 Ea (la VI Ea F F. O I4t 4r Cl rd rq VIU) d t Cd Cd Cd Cd C Cr3 co d rr3 H ) H C cr r H a3 ( 4I C 0r3 pq l 4) H H cd * (l qi ^ ^ a( af fl 19 F- HH;j p -e-N - S ^0. - Ea X Cd 43 -P 4O 3 O bl b D t)D bdO t O D a ) A D 0- co P ~4Cdpi a 3 a H H ) a ~ (1 a CJ CJ MP E V) T P Ea H *H r H *H *H *H *H *H *H OJ CM U *H rI.H *H- rd d d rd rd T r d n r ed rI,Cr3 d r3 HH - 4OH U) U) U) U) U) U) U) U) C r-f U) OJ^OJCMC g E F f E P C4 F P4 4 ~4 1- 4 CMPQ1 P rD * — cd Ci3 U/) C' H bn K\ CMo 0 0 C OJ H \ -Dnt — C\.O cO H 0 0 U 0 \ O l — CO)J r-l r —l lC-\ L \D r-ON C\OJL O 00 tD"S iO \ O H^ r-l C i\ C 0 0 0 tC- 0o ON OO CW a r-H CM CM CMM M j C Cj CM CM CM CMj CKN C\ K 25

r-lf-I n n -=t _t _ _t_-t - -t L n nn - n L K\ K^ J KO\ _ t _:- _: 0 00000000 00 00 00 00 0000 r \ MICJMK j C jC -zt K\ l\ _t MN\ WC\ K r _ ^ -r \ 00 00 00 00 00 00 00 00 0000 0 00 0 0 0 0 0 0 0 0 0 000 00 0 0 00 00 O\No 0 CjM \O i\'CM 1O \ G\ CMH rJ C'jCj O\O\ co o\ O\ o\ ) O rH 00 0000 00 oo 0\ 0\ 0C\ O\o 0\ -i r- rir HiH i HrH HH Hi H H H 00 C6666 Ma 0\O CO O o0\O\ -C\ 00 ^D _0 I'o- OI \\ \' 0 I\ *H Fr COJ K- CM CUj CUJ r< \ NK\ KN NlK KIN N r\K\ MK KIN -0t 0 C\ A 0CH O~ (t-i~~ 1- 0CM CM0 lO H rC H.C:3" r*H H r- 0 0 ON ON \ 0\'-\O \ 0 0 00\ 0 \O 0 OO 00 0 0 OO LOO0 0 00 C0 0000 000000 00 00 00 00 0000 (OC KrO\ICOM 0\ O l 0O \ 0 0 \ O-'O:mI U C Ol - C - 5 o od66oooo i-16 o oo d6 d6d6d t-0 O 01t t-O \ -. 4 HcO K\4- CMcO 0\ 1f\Lf\ 00 o C) iU\ NW\ W\\.O lNi *\LN\ Lf1I\ in\\f U\& c o \O\^0 \.O\O'-.D \0\D \ \ \ \\.10 \0'O\D \O\ D O'0 F-4 CV VI Co H U) U) U) U) ) Cd CO CO H, C.. O o5 U)i U) U) U) CoCo co a)c EE co C) CV-) C)- C) r. 0CCd Cd w Cd Cd Cd Cd CdP CdP pq p q pq pq to. -P -CC -43'cp fl I I (Zt N NN N N N N N mI VA *H *H *H * *H *H Cd bO boh O CO CO Co CO C Co C d d Id <d a ) U4' U)43 U43 U U U) ) U) Uo) C M to E EC a M ) U U) U) *H H *H *H H *H *H F4 Fr F Fr l Pip Pi PF P^ Pipi Pi A Pp P^ a) co c6<L) O~l^ OJK^O OI t, 1 kkk~~~~ 0 0 0 0 0 1n O0 Cd U) CM dL\ CM -4 zt C WI43 K\ D I\0 K'O M t- -CM KN\CO \0 t- O cIO< ONH K U) C M rCKK CJMt M MI KU K tr CJC K OJ OD\O 0MJKq \\D l C \ C C K \ _ O O K rEc\\ C\MC\K\IC\ K M c\l WN C\ CMC K C\M-' ( CM KC- \ M C C\ \\w\ -CK \ ~ r C\C M CM \W\&\U\CM 0O (2)CNONCMCM I C MO 1Z-4 I Hti-: 0K \0WA\CM0 I \O 43'(D c bo 4 I-dO cc O0 Hoi CM 1 4 X)^ CC) - IC- l ~ r 0\ 4 I0 -4 -4 K'~!x KI N. r r S~~~~~~~~~. 24~~~~~~~~~~~~~~

C.) U CO ill U) C) a 4-U) 0Du a a a, m I rJ i I a'd co 0 co m d m ed md ro m m m r-i *rl a) OU) O *H Od O H 0 E Cd a PL I E Ea E Ea 4) Ea a Ea co EaP -1 9 O U9 0 O' O 03 ~ C 0 4' CF 3 r V *Hp a, I a O w aO aU ) aU U * U U ) a aU U aU U aU) U)) U) U -r4 Nc IQ' N q 3N N N N N +N N N N N m ( ) H *rl *H *H H- H *H H *H *H H *HO a *H. 0 id co C 0 COC c CO C CO CO CO - CO CO CO C CO CO CO C S5 cM m OM' > X w cr o oI co rl to to Ea c t r0. C C a)H C V) i a) -H Ct C) a) ) a) i CL /) ) ) a) a) C aC) rl *Hr -1 H H *r H *H HH 00H r H H rr4 H r r 00 rCHl U 00 H ) a 3 P& P 3 r P O4 C O rO CO COC PC PO P-C P; P D P 4 Pi P-i Pi Pi p 0 bJ D O C2 Cid a)O K- CC O) \ 4-U ) -:t 1 K CC C C eiifc\ zoj- (U) KL\ C~ P _: - \,D t ONOO CM C t-t_ \ \ Io DO OD H i- 0-t t-~- -t K\ -z CM 4P- \I-D-t -:J-LCI\&\ LC ( \n n n\ n n \Lf n LCN Ln Ln W H H-q 0 0 HH i oW C H) 0 - I ON\- 0 00 \O - H H-q — FO H O, c4 > CM CMONV t-HGrDCOco-CM\ -Zj t zt_-tO O\ n CM oC ON c OD L - Cc 0\ 0 H CM K o\ L Ln \I - CC) ON aU )J K - O -K _N UN r LC n (NLLn r n\ n n Ln L\'^^ i'o\ i rc\ Kf\ =4N K'4 K\ Kl\ i^ WI z I -:: 0 t^ I I II^ ^:::;^ ^ ^ ^:lt:tS f t::^ ^ ^ ^ 1^ 0^ -O rl -- 1 0\>->- ^^ OO OO-? 250

O H d) 0 ^ a c C11 *rl C)V D r HO *rl Co- 0o -P -P 0 03 O a) )U n *rlU) E-r a oo C OO0 r-I 0 bOD Hc () rM C aPC Ho -P U)C 0 a pq Cd (Uq P4 CE' c ) bP.0 ~ coE' r-U)U)I bO )ri) H^ coH HU -,U.P'.C E. rQ 0\ PCO 03 rrbO.j U C H3 - H U 0) *H ) U M X CH O;0 26 E-' 3 0 U)H *N r' H OH r Ef4 Hq c-iq'H C)0UC CU 4- 0* co ON 0 H R R'H (1).\ f \H P \(0 c, ij 0~~~~ CO (U ^0 t>- M +3 2

TABLE III FEEDING DISTANCE DATA FOR BAR CASTINGS OF CUNISIBE Casting Casting. Distance Sound (in.) in- i..* - - \Condition Soundness Number Size (in.) Cit From End From Riser une 346-3 1/4 x 1/4 x 4 3/4 5/8 No 346-3 1/4 x 1/4 x 2-7/8 Chilled. 5/8 /8 No 346-3 1/4 x 1/4 x 2 - - Yes 346-3 1/4 x 1/4 x 3 5/8 3/8 No 346-3 1/4 x 1/4 x 2-1/2 Unchilled. 1/2 1/4 No 346-3 1/4 x 1/4 x 1-7/8- - Yes 336-5 1/2 x 1/2 x 7 1-1/4 2 No 336-5 1/2 x 1/2 x 6 Chilled 1-3/8 1-3/4 No 346-3 1/2 x 1/2 x 5 1-3/4 1 No 346-3 1/2 x 1/2 x 4 - - Yes 336-5 1/2 x 1/2 x 6 1 3/4 No 336-5 1/2 x 1/2 x 5 Unchi1 3/4 No 1546-5 1/2 x 1/2 x 4 Unchll1 1/2 No 346-3 1/2 x 1/2 x 3- - Yes 324-3 1 x 1 x 10* * No 325-5, x 1x 9 5-1/2 0 No 326-1 1 xx 8 Chilled. - Marginal 324-1 1 x 1 x7 - - Yes 325-3 1 x 1 6 - - Yes 325-1 1x 1 x 5 - Yes 324-4 1 x 1 x 106 * No 325-6 1 x 1 x 9 * 0 No 326-2 1 x 1 x 8 Unchilled 1 No 324-2 1 x 1 x 7 - - Yes 325-4 1 x 1 x 6 - - Yes 325-2 1 x 1x 5 - - Yes 327-2 1-1/2 x 1-1/2 x 10 * * No 326-5 1-1/2 x 1-1/2 x 9 Chilled - - Yes 326-3 1-1/2 x 1-1/2 x 8- - Yes *Clearcut contribution from the riser or end effect was difficult to interpret from the radiographs. 27

Table III (Concluded) Casting Castingit Distance Sound (in.),., _. /. - Condition Ditne,,- _.' Soundness Number Size (in.) From End From Riser Soundnes 526-6 1-1/2 x 1-1/2 x 9 3-1/4 2 No 526-4 1-1/2 x 1-1/2 x 8 Unchilled 4 1 No 527-1 1-1/2 x 1-1/2 - 7 - - Yes 327-3 2 x 2 x 10 Chilled Yes Chilled. Yes 328-1 2 x 2 x 9 Yes 327-4 2 x 2 x 9hil No 329-1 2 x 2 x 8 - - Yes 328-2 5 x 3 x 12 - - Yes Chilled. 330-1 3 x 3 x 11 - Yes 329-2 3 x 3 x 11hill - - Marginal 330-2 3 x 3 x 10 - - Yes *Clearcut contribution from the riser or end effect was difficult to interpret from the radiographs. 28

TABLE IV FEEDING DISTANCE DATA FOR PLATE CASTINGS OF CUNISIBE Casting Casting Distance Sound (in.) Condition Soundness Number Size (in.) From End From Riser 332-2 1/2 x 7 x 7 1/4 No 333-1 1/2 x 6 x 6 2-1/4 1/2 No 333-2 1/2 x 5 x 5 1-3/4 1/2 No 335-4 1/2 x 4 x 4 Chilled 2-1/8 1/8 No 336-3 1/2 x x 21/2 1/16 No 362-4 1/2 x 2 x 2 - Yes 362-4 1/2 x 1-1/2 x 1-1/2 -- Yes 333-3 1/2 x 5 x 5 1-1/2 1/4 No 333-4 1/2 x 4 x 4 1-1/2 1/8 No 355-3 1/2 x 3x 3Unchilled 1-1/2 0 No 362-4 1/2 x 2-1/2 x 2-1/2 * * No 336-4 1/2 x 2 x 2 1-5/8 1/8 No 362-4 1/2 x 1 x 1- - Yes 335-2 1 x 6 x 6 3 1/2 No 336-2 1 x 5 x 5 Chlld 3-1/2 1/8 No Chilled 360-1 1 x 4 x 4 2-5/8 1/16 No 361-4 1 x 3 x 3- - Yes 335-1 1 x 5 x 2-/8 /8 /8 No 337-3 1 x 4 x 4 Uhil 1-9/16 0 No Unchilled 348-1 1 x 3 x 3 115/16 3/16 No 361-4 1 x 2 x 2 11/4 1/8 No 334-2 1-1/2 x 7 x 7 4 5/8 No 336-1 1-1/2 x 6 x 6 3-3/4 0 No 337-1 1-1/2 x 6 x 6 Chilled 3-3/4 0 No 359-4 1-1/2 x 5 x 5 3-1/4 0 No 360-2 1-1/2 x 4 x 4 3-3/8 1/8 No 361-3 1-1/2 x 3 x 3- - Yes 334-1 1-1/2 x 6 x 6 3 1/2 No 337-2 1-1/2 x 5 x 5 2-1/2 0 No 359-3 1-1/2 x 4 x 4 Unchilled 1-7/8 3/16 No 360-53 1-1/2 x 3 x 3 1-1/4 1/4 No 361-3 1-1/2 x 2 x 2 1-1/8 3/16 No *Clearcut contribution from the riser or end effect was difficult to interpret from the radiographs. 29

TABLE V RISER SIZE DATA: SIDE RISERING, INSULATING TOPPING Casting Casting NbCasting Csing) H D H/D Vr/Vc (L+W)/T Condition Number Size (in. ) 338-4 2 x 2 x 8 3-1/4 6 0.54 2.87 5 Sound 343-1 2 x 2 x 8 3-1/4 5-3/8 O.605 2.60 5 Unsound 338-3 2 x 2 x 8 4-3/8 4-3/4 0.92 2.43 5 Sound 329-1 2 x 2 x 8 6 4 1,50 2.10 5 Sound 338-2 2 x 2 x 8 4-3/16 4-3/8 0.96 1.96 5 Sound 546-2 2 x 2 x 8 4 4 1.00 1.58 5 Unsound. 339-4 2 x 2 x 8 3-1/2 4 0.875 1.38 5 Unsound 338-1 2 x 2 x 8 5-1/8 3-1/8 1.54 1.22 5 Sound 341-4 2 x 2 x 8 4-1/8 3 1.38 0.91 5 Unsound 50

TABLE VI RISER SIZE DATA: TOP RISERING, INSULATING TOPPING Casting Casting Casting Casting DH D H/D Vr/Vc (L+W)/T Condition Number Size (in. ) 358-4 3 x 3 x 3 2-3/8 4-3/4 0.50 o.575 2 Unsound 345-4 2 x 2 x 8 1-5/8 6-5/8 0.25 1.76 5 Sound 343-3 2 x 2 x 8 2-1/2 5 0.50 1.54 5 Sound 347-1 2 x 2 x 8 1-9/16 6-1/4 0.25 1.50 5 Unsound 342-1 2 x 2 x 8 4 3-13/16 1.05 1.425 5 Sound 340-2 2 x 2 x 8 3 4-1/4 0.706 1.33 5 Sound 339-2 2 x 2 x 8 2-3/8 4-3/4 0.50 1.32 5 Unsound 343-2 2 x 2 x 8 4-3/16 3-3/8 1.24 1.17 5 Sound 340-1 2 x 2 x 8 3-5/8 3-5/8 1 00 1.165 5 Unsound 341-1 2 x 2 x 8 4-1/6 3-1/4 1.25 1.065 5 Unsound 339-3 2 x 2 x 8 3-1/4 3-1/4 1. 00. 845 5 Unsound 339-1 2 x 2 x 8 2 4 0.50 0.785 5 Unsound 342-4 1 x 5 x 5 3 4-1/2 0.67 1.90 10 Sound 345-2 1 x 5 x 5 2-1/4 4-1/2 0.50 1.43 10 Sound 341-2 1 x 5 x 5 2-1/8 4-1/4 0.50 1.205 10 Unsound. 345-1 1 x 5 x 5 3-1/4 3-1/4 1.00 1.08 10 Sound 342-2, 1 x 5 x 5 3-1/8 3-1/8 1. 00 0.958 10 Unsound 340-3 1 x 5 x 5 2-7/8 2-7/8 1.00 0.745 10 Unsound 347-1 1 x 6 x 6 2-7/8 4-3/4 0.50 1.17 12 Sound 345-3 1 x 7 x 7 2-11/16 5-3/8 0.50 1.24 14 Sound 341-3 1 x 7 x 7 2-1/2 5 0.50 1.005 14 Unsound 342-3 1 x 7 x 7 3-13/16 3-1/16 1.00 0.886 14 Sound 340 1 x 7 x 7 3-3/8 3-3/8 1.00. 615 14 Unsound 31

TABLE VII RISER SIZE DATA: TOP RISERING, INSULATING TOPPING AND INSULATING SLEEVES Casting Casting casNumbr asting H D H/D Vr/Vc (L+W)/T Condition Number Size (in.) 358-2 3 x 3 x 3 1-1/2 3 0.50 0.393 2 Unsound 344-2 2 x 2 x 8 2-1/8 4-1/2 0.50 1.05 5 Sound 344-3 2 x 2 x 8 3 3 1. 00 o.665 5 Sound. 344-1 2 x 2 x 8 1-1/16 4-1/2 0.25 0.53 5 Sound 346-4 2 x 2 x 8 2-1/2 2-1/2 1.00 0.381 5 Sound 348-4 2 x 2 x 8 1 3-7/8 0.258 0.69 5 Unsound. 344-4 2 x 2 x 8 1-1/2 3 0.50 0.33 5 Marginal 349-2 2 x 2 x 8 1-7/8 2-1/2 0.75 0.286 5 Sound 349-1 2 x 2 x 8 2-1/4 2-1/4 1,00 0.28 5 Sound 350-3 2 x 2 x 8 2 2 1.00 0.196 5 Unsound 351-3 1 x 5 x 5 1-1/2 2-5/8 0.57 0.324 10 Sound 353-4 1 x 5 x 5 1-5/16 2-5/8 0.50 0.285 10 Unsound 350-5 1 x 5 x 5 1-1/4 2-1/2 0.50 0.245 10 Unsound 350-4 1 x 7 x 7 1-3/8 2-3/4 0.50 0.167 14 Unsound 351-4 1 x 7 x 7 1-9/16 3-1/8 0.50 0.246 14 Unsound 354-4 1 x 7 x 7 1-11/16 3-3/8 0.50 0.306 14 Unsound 358-1 1 x 7 x 7 1-3/4 3-1/2 0.50 0.345 14 Sound 32

TABLE VIII RISER SIZE DATA: TOP RISERING, EXOTHERMIC TOPPING casting Casting Number Size (in.) H D H/D Vr/Vc (L+W4)/T Condition Number Size (in'-'. ) -.,_......_ 357-1 3 x 3 x 3 1-13/16 3-5/8 0 50 0.695 3 Unsound 348-2 2 x 2 x 8 3-3/8 3-3/8 1.00 0.945 5 Sound 352-2 2 x 2 x 8 1-5/16 5 0.26 0 800 5 Sound 350-1 2 x 2 x 8 1-15/16 3-7/8 0.50 0.718 5 Sound 354-1 2 x 2 x 8 1-7/8 3-3/4 0.50 o.647 5 Sound 351-2 2 x 2 x 8 1-13/16 3-5/8 0.50 0.585 5 Unsound 350-2 2 x 2 x 8 2-7/8 2-7/8 1.00 0.585 5 Sound 348-3 2 x 2 x 8 2-3/4 2-3/4 1.00 0.510 5 Unsound 353-1 1 x 5 x 5 1-11/16 3-3/8 0.50 0.605 10 Sound 354-2 1 x 5 x 5 1-5/8 3-1/4 0.50 0.54 10 Sound 356-1 1 x 5 x 5 1-9/16 3-1/8 0.50 0.481 10 Unsound 353-2 1 x 7 x 7 2-1/16 4-1/8 0. 5 0.561 14 Sound 354-3 1 x 7 x 7 2 4 0.50 0.515 14 Sound 356-2 1 x 7 x 7 1-15/16 3-7/8 0.50 0.469 14 Sound 358-3 1 x 7 x 7 17/8 3-3/4 0.50 0.424 14 Sound 360-4 1 x 7 x 7 1-13/16 3-5/8 0.50 0.382 14 Unsound 33

TABLE IX RISER SIZE DATA: TOP RISERED, EXOTHERMIC TOPPING AND EXOTHERMIC SLEEVES Casting Casting Casting Casting H D H/D Vr/Vc (L+W)/T Condition wNumber___Size (in.)... 359-1 3 x 3 x 3 1-3/4 3-1/2 0.50 0.625 2 Unsound 356-4 2 x 2 x 8 1-5/32 4-5/8 0.25 0.608 5 Sound 359-2 2 x 2 x 8 1-3/4 3-1/2 0.50 0.527 5 Sound 357-2 2 x 2 x 8 1-11/16 3-3/8 0.50. 472 5 Unsound 352-4 2 x 2 x 8 1-1/16 4-1/4 0.25 O.470 5 Unsound 352-1 2 x 2 x 8 1-5/8 3-1/4 0.50 0.424 5 Unsound 352-3 2 x 2 x 8 1-3/4 3 0.58 0.388 5 Unsound 355-1 2 x 2 x 8 2-1/2 2-1/2 1.00 0.385 5 Sound 351-1 2 x 2 x 8 2-3/8 2-3/8 1.00 0.329 5 Unsound 349-4 2 x 2 x 8 2-1/8 2-1/8 1.00 0.236 5 Unsound 349-3 2 x 2 x 8 1-3/4 1-3/4 1.00 0 130 5 Unsound 357-3 1 x 5 x 5 1-9/16 3-1/8 0.50 0.479 10 Sound 355-4 1 x 5 x 5 1-1/2 3 0.50 0.424 10 Unsound 355-2 1 x 5 x 5 1-7/16 2-7/8 0.50 0.373 10 Unsound 356-3 1 x 7 x 7 1-7/8 3-3/4 0.50 0 424 14 Sound 357-4 1 x 7 x 7 1-13/16 3-5/8 0.50 0 382 14 Unsound 355-3 1 x 7 x 7 1-3/4 3-1/2 0.50 0.344 14 Unsound 54

j~~~~~~~~~~~~~~~~K~~~~~~~~~~~~~~~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~:::.:.:~:8:::1:.-::::iZ:_ii'-':''''''''''::ii~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~iE'-iiii~~~~~~~~~~~~~~~~~~~~i-F~~~~~~~~~~i-;~~~~~~~~~',s i'::':.i,,.i'~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ii~~~~~~~ii-iiii~I!1!iii K~~~::~~~~~~~~~~:rii'~ii~~~~~~~~~~~~~~~~~~~~~~~~~~~~ii"'`i~~~~~~~~~~~~~~~~~~~~~~ii iiliiiliiiii::-: —:::~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~:i-l~~~~~~~~~~~~~~~~i~~~~~~~i~'~~~~~~~~~~~~~:ii'iil~~~~~~~~iii~iiiiiiii? ii~~~~i:~ii-lii~~~~~~~~iiii-iii~~~~~~~~-~~~i::-:li; —~~~~~~~~~~:-: —:-:c:::i::;:::-:::::::::::::: —-i~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~'-i iii~~~~~~~~i:_-iiiii~~~~~~~~iiiiii ix-i~~~~~~~~~~i-:i-.i:-:~i~~~iii~~iiiii::i Figure 1. Paecsigwt ag o ie hc hw h lc sd o radiographic examination. 55

-.,.!'i'.... Figure 2. Section from a side risered bar with a molybdenum mesh dross trap between the casting and riser. If the mesh was inserted in the gating system the shrinkage defect would not occur. 36

Figure 3. A typical fluidity spiral cast in 70-30 cupro-nickel. The particular spiral was cast with 2500F superheat with a resultant length of 25". 37

Figure 4. A typical hot tear casting assembly. One end of each leg has a 3/4" radius while the other radius progressively varies from 1/16" to 3/4" dependent upon the casting position. Each leg is 7" long. 38

Figure 5. 1-1/2" x 1-1/2" x 10" end chilled bar. Gating and risering were similar to that used in the standard 2" x 2" x 12" castability bar (thermal gradient investigation). 39

0 rd 0 ()0 0 0 "~o ~ ~0 rr E ///// - O ^ / /o i f U) C) O \ -- p r4 0 ~ 0 0 _ 0 *H 0o.o o0'r IT A - 1T n -i -P -T A (illl;4 0

Figure 7. 1-1/4" 90~ elbow drag cavity. The end chills are in place however the cores are removed to show the gating system.

Figure 8. Open cope portion of the production casting. Anchors are attached to the cope chills to prevent their movement during mold handling. The top riser neck is attached at the reentrant corner. 42

Figure 9. Drag section of the 1-1/4"-90~ ell casting with all cores in place. Cores were made of C02 sand while the mold itself was dry sand. 43

Cunisibe 40- x s 0 / Cu-Ni 30 c liquidus - 2110yF; Cu-Ni liquidus - 2240~F. 100 200 300 400 Superheat (~F) Figure 10. Fluidity Curves for Cunisibe and. 70-50 cupro-nickel. Cunisibe liquidus - 2110~F; Cu-Ni liquidus - 2240~F.

Figure 11. X-ray of a Cunisibe hot tear casting with a 1/16" radius cast in a C02 mold. The casting did not tear even though the fillet was smaller than would be considered good foundry practice. 45

aidnoaowJou.Lji CM ia ~ LO.rl rd Od,-t Cd c-,C),-O o 0 0 0 0 H Ir) ) IC )u IM CMn- -

80 70 Temperature Gradient at the Solidus Temperature. Gradient. 60- at the Solidus- Liquidus.N:Midpoint 0 ) 50 30 10 /'940-. / riser 2 4 6 8 10 chil Distance from Riser (inches) Figure 13. Thermal gradient curves for a 2" x 2" x 12" bar chilled at one end. The gradient has been determined by the secant method at two temperatures. end. Tegain a endtrie ytescn ehda w eprtrs

0 0) N H qv o 0 d 0rl~E E- E 0) r4 r-4~rt ~ ~ ~ ~ ~ ~ r ri d 0 4). | | r~~~l | _. I _ a0, CMC E4 +2 CH 0)4 rd * I k 1 IllI 1, _. _ l. _ l _ l. _ _~. _l. _.. l _ l. _. _~~~~~~~F - * l l l || l_ a

UOI!7SOd pv7 Li N. ajdnoDowJq r C L..^ | 1a) 0 I 0,-I,d 0 H — () 0 0 ro r#)- ci Pq oo l.49 0 C) 0 cI) V 03 0 Ob I"-. If) n-,0 br0 g9

100 Temperature Gradient at the Solidus Temperature Gradient at the Solidus - Liquidus Midpoint 80 eC _0 40 E 20 / 2 4 6 8 10 Distance from riser (inches) Figure 16. Thermal gradients for a 2" x 2" x 10" end chilled bar of Cunisibe. The data have been calculated by the secant method from the Cooling curves in Figure 15. 5o

Figure 17. Feed distance castings poured when small bars were studied (less than 1/2" x 1/2" cross section). 51

Cunisibe 0 Sound * Unsound e Marginal - --- Cupronickel 12 2 10 riser contributio n 1 2 3 Section Size (inches) Unchilled Figure 18a. Feeding distances in unchilled bars for Cunisibe alloy and 70-30 cupro-nickel. 52

-Cunisi be 0 Sound * Unsound e Marginal Cupronickel E: 10., \ _CZ- 8/ -J 6 co L. 2A\ / /riser contribution / / / 1 2 3 Section Size (inches) Chilled Fi ure 18b. Feeding distances in end chilled bars for Cunisibe alloy and 70-30 cupro-nickel. 553

Figure 19. Feed distance castings poured when small plates were studied (for 1/2" thicknesses only). 54

0 Sound * Unsound 6 6,.. 1, "' ii 5 " c __ ______//////i eco iluo// rU 4 2 end contribution riser contr but ion 0.5 1.0 1.5 Plate Thickness (inches) Unchilled Figure 20a. Feeding distance in unchilled plates of Cunisibe alloy. 55

0 Sound * Unsound 6;s 5 a)3 33t| c a_ / /;/ ////,riser contribution/// 0.5 1.0 1.5 Plate Thickness (inches) Chilled Figure 20b. Feeding distance in end chilled plates of Cunisibe alloy. 6

1 2 3 c 4 1 2 3 Section Size (inches) 1 2 3 Figure 21. Summary of feeding distance in plates and bars in Cunisibel aloy. Bar data for 70-30 cupro-nickel have also been included from reference (7). 57

30 KEY: 0 44 0- ring 20. * 3/8 0-ring 2x2x9 unchilled (feeding distance bar) 10 _0: 1 2 3 4 5 6 7 8 9 E L 30._ 02o0.1 x 1 x 10 chilled 20 2 2 O.ij ~|::X / ^.\ \ (feeding distance bar) _.10 * a \'s' 0 C- I 2 3 4 5 6 7 8 9 10 4 —0 I I 30 (D~L 10. x3 1 x x8 unchilled 20 4^ / rU~~/ \ \ (feeding distance bar) t /2\ - 10 \ - lo30 L 30 20 1 x 1 x 8 chilled ~~~~~~~~~~0 ~~~(feeding distance bar) 10 \ 1 2 3 4 5 6 7 30 20 2 x 2 x 8 side-risered 10 (/~ ~ ~~~~~~~ = 0.605 2.6 10 1 2 3 4 5 6 7 8 Distance from riser (inches) Figure 22. Pressure test of selected bars used in the feed distance study. The leak rate was determined at lOOpsig dry nitrogen pressure. 58

rd 000 oO c / (d r I - r eo O c0 (O 0.0D 0 0 4/k 0 CD0 U crc ~ ~ ~ ~ ~ ~~5 ________/ a____________ o o'E r~~~bI> Q- U) bDO c 00~5 D3 - I-|

rd -P CCF 0 co "0 c3" c 0 00E p Q r 1 -— 19 CH / ~ ~I0 k /, " 0 o7__ _ _ _ ICo U) LO/~~ r4 U~~ ~~~~~~~~~~0 -~a)!'-0 - 0''~) C)..., c ri 0. 10 0 _t) I j -I- CO 0~ ~ ~~0 ) I 0 )a Q o. N000 >LI.

tz taken from versus H/D curv/e Insulating Topping *,O H =1.0 0,E1 Sound D *, Unsound D*,D H= 0.5 1.5 Vr 1 O 1.0 0. 5 Top Risering. D.. I v, ] lVr tH plates ~> D> IIVr $H bars 5 10 15 L+A T Figure 25. Risering curve for plates and bars of Cunisibe alloy cast with an insulating cover on open risers. 61

Figure 26. Top risered bar which shows the flow-off channel used to maintain a constant (H/D) ratio for the riser. 62

Figure 27. Top risered plate which shows gating and the use of the flow-off technique. 63

70o U p 0 — 0 -r 0 0 0 0C o 0 ~0'4 IhV0 I Ua ~PI 0- 0 ZJ'- CH0 t_ 0 L/ 4,,-. w /aU 0 0f) /0 1 — /<H~

0 Sound * Unsound Insulating Topping & Sleeves * Marainal Q.6 H 0.5 0.4 Vr " 0 0.3 0 0.2 Top Risering 0.1 V^ FiV H plateH r _bar Vcbar 5 10 15 L+W T Figure 29. Risering curves for plates and bars when insulating topping and insulating sleeves are used. H/D values were 0.50.

0 l00 jU)t~~~ o -0 c-. Pi 0 C0 L.) 0 LO LO. o to 4) E ^ 0'I 0 CF4 t1) o __ i ici /LL(~~~ 4o 0 0 OO 66 _ --- -- -- - a

0 Sound Exothermic Toppingq Unsound H -05 0.7 L! 67 0 0.4 0.3 ___ _ Top Risering ~eD~ 0.2 Vc 3" cube I ^ I Vr ItH plate 0.1 I Vr -H bar 5 10 15 L+W T Figure 31. Risering curve for top risered casting made with exothermic topping. H/D = 0.50. 67

(L) _0 Q rC /iiC') D'/ rd 0oc d._j / L~O IIC) ~ 4 0) o (I') Q) 0 0 -dL L 0o I) <- I I~ I' ~0 ~~0 68

0 Sound Exothermic Topping & Sleeves 0 Unsound -D 0.5 0.7 0.6 VN 0.5 vc II 1' ~' ~.._ _0O 0.4 0.3 Top Risering.D* IH 0.2 j3 cube 43. v0. IV, H plate 0.1'' ~Vc bar 5 10 15 L+W T Figure 33. Risering curve for use when exothermic sleeves and exothermic topping are used. H/D = 0.50. 69

Cd Q) tO CH 0~0 ~ 0 0 >C~ ~ o ~0~ ~ ~ ~ ~-C) r> olo I() 0,/ o / C 100 0 C) 0 o 4)c 4 <;- bf - I^l Sl 0 r- "P 0' 4,,a

1.5 yr v 1.0 5 10 15 Figure 35. Summary of risering curves for both Cunisibe alloy and 70-0 CU. ~-l ig i es (H/DO.5) ~ther n/ t~.-_eves (H/D= 0.5) 5 10 15 L +W T Figure 35. Summary of risering curves for both Cunisibe alloy and 70-30 cupro-nickel. 71

Figure 36. 1" x 5" x 5" plate cast with insulating riser topping. The casting was unsound as can be seen visually and verified radiographically. 72

Figure 37. 1" x 5" x 5" plate cast with insulating topping plus insulating sleeves. A shrinkage pipe was not evident however the casting is radiographically unsound. 73

1%'8 Equivalence possible riser positions D. 5/ Circumference (outside) = 7.47 in. Circumference (inside) = 3.54 in. Circumference (median) = 5.50 in. Volume of casting = -(5 8 )(2/8)2 (1/8)2 = 19.4 in.3 Shape factor= L W = 17.8 T 5/% Figure 38. 90~ elbow and its equivalent section used as an illustration of the riser data. 74

Figure 39. 1-1/4"-90~ elbow cast in three different ways to illustrate use of the feeding distance and risering curves. 75

Figure 40. Sectioned 90~ elbow which was risered at one end and chilled at the other. The surface shrinkage in the cope portion was due to inadequate feeding...~ e ~ ~ ~ ~ ~ ~~7

Figure 41. A riser adequate for exothermic topping is inadequate in height for insulating topping. The riser shrinkage hag been pulled down into the casting. 77

Figure 42. A correctly risered casting as also shown by radiography. The surface blemishes were due to some metal reaction with the uncoated CO2 sand core. 78

UNIVERSITY OF MICHIGAN 3 9015 02499 53941111111111111 11111111111111111 3 9015 02499 5394