2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

Quantifying Casting Quality Through Filling Conditions

Daniel Hoefert,1 David Weiss,1 Randy Oehrlein,2 Cory Sents,2 Travis Bodick,2 Chris Hastings,3 Jerry Thiel,4 Travis Frush,4 Leah Dunlay,4 Kip Woods,4 Robin Foley,5 John Griffin,5 Kyle Metzloff,6 Henry Frear6

1Eck Industries Inc., Manitowoc, WI; 2Carley Foundry Inc., Blaine, MN; 3Morris Bean & Co., Yellow Springs, OH, 4University of Northern Iowa, Waterloo, IA, 5University of Alabama/Birmingham, Birmingham, AL; 6University of Wisconsin/Platteville, Platteville, WI,

Copyright 2020 American Foundry Society

ABSTRACT This leaves us to wonder, is this also happening below the surface? Focused concern and education related to filling damage and oxide inclusions has been widely promoted among If entrained into the metal stream, even a thin oxide film the foundry industry in the past three decades; with may present a threat to the quality of a casting. Alumina special regards to aluminum.1 However, predicting the skin is insoluble to the melt and has a melting point more quantifiable damage that oxide film may cause to the than double that of molten aluminum.2 The interface of quality of aluminum castings during the filling process the folded skin (bifilm) will not bind to itself; essentially remains largely theoretical, due to a lack of supporting forming a crack. The density of alumina is nearly the data. The purpose of this experiment was to turn on and same as aluminum,3 making them difficult to observe turn off turbulent filling conditions consistent with bifilm with radiography. And, aluminum oxide does not entrainment, in order to obtain the needed data required to fluoresce under ultraviolet light,4 making it difficult to predict the actual porosity and tensile damage these detect with dye penetrant inspections.5 conditions may cause to aluminum castings.

Filling conditions ranging from tranquil to turbulent were generated within a test casting by means of three different gravity-fill gating systems. The test casting is made up of nine bent legs that help establish the varied filling conditions. The molds were constructed of 3D printed sand. Fluid flow simulation was used to identify the tranquil and turbulent filling conditions, which include metal-falls, back-waves and eddies. Four common aluminum alloys were trialed in the study: C355, A356, E357 and A206. Radiography and tensile testing were used to quantify the resulting damage; with SEM analysis of fracture surfaces to inspect for oxide film. Limited leak testing was also performed with a focus on identifying potential oxide related bubble-trail leak paths.

Keywords: oxide skin, metal-falls, back-waves, eddies, bubbles, seams, flow tubes, aluminum alloys

INTRODUCTION AND BACKGROUND

Clear evidence of oxide skin can be observed in many of the processes associated with molten metal handling. The Figure 1. The duller appearance of aluminum oxide reaction of oxidation forms a skin on an aluminum melt skin can be noted in the remains of a melted ingot, on surface almost instantaneously. Its dull gray appearance the melt surface of a pour ladle (skimmed seconds manifests itself quickly at the surface of pour ladles, even earlier) and floating in a pour cup. moments after skimming the initial film from the surface. The remains of deflated ingots and scrap castings can also In bifilm theory, the asserted concern is that surface be seen floating to the melt surface in melting furnaces. turbulence will entrain bits of the oxide skin into the metal stream, allowing it to shred and fold into a cloud of Skins can often be observed accumulating at the melt 6 surface of ladles and pour basins during the pour (Fig. 1). crack-like bifilms of untold numbers. This polluting continues for as long as the surface turbulence exists. The cloud of defects is presumably carried downstream from

Page 1 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages) their turbulent origin, flowing freely throughout the filling CASTING DESIGN process. The density of alumina (slightly higher than that The casting weight is 8.5 kg (18.7 lb) without the gating of aluminum) is said to offset the small amount of air that attached. Each of the nine legs are 82.5mm (3.25in) wide may reside between the folds of the furled bifilm, leaving by 152mm (6in) long by 25mm (1in) thick. Each leg most alumina bifilms neutrally buoyant.7 The final allows for three tensile specimens. The bend between journey of these films may end up gathered in any eddies each leg works in conjunction with each filling systems to created within the cavity, or halted in midstream by establish different filling conditions within casting. solidification, while others may harmlessly exit the cavity by venting into the feed-risers. Bubbles can generate their own special issue, by leaving an oxide bubble-trail. This trail can create a small leak-path to wherever the bubble floated off to.

According to bifilm theory, additional damage becomes evident as the casting solidifies. Bifilms are suspected nucleation sites for porosity.8 Casting areas slow to solidify are said to unfurl the bifilms. As they unfurl, the crack-like film expands along with any precipitating porosity.9 The expanding bifilm and porosity lowers local elongation properties and creates potential leak paths. Conversely, areas well fed10 and quick to solidify11 are said to freeze bifilms before they unfurl, reducing the negative effects to porosity growth and reduced elongation.

The focus of this experiment is the quantification of casting quality through filling conditions. This was investigated by generating different filling conditions Leg 1 Horizontal Chilled Poorly Fed within a test casting, while maintaining identical Leg 2 Vertical No Chill Poorly Fed solidification conditions. The castings were extensively Leg 3 Horizontal No Chill Well Fed examined through radiography, tensile testing and Leg 4 Horizontal Chilled Well Fed subsequent fracture surface inspections. This was a joint Leg 5 Vertical No Chill Poorly Fed effort between Eck Industries, Inc., Carley Foundry, Leg 6 Horizontal No Chill Poorly Fed Morris Bean & Company, the University of Northern Leg 7 Vertical Chilled Poorly Fed Iowa, UW Platteville and the University of Alabama; with Leg 8 Vertical No Chill Well Fed funding support from AFS. Leg 9 Vertical Chilled Well Fed

Figure 2. The solidification system is composed of EXPERIMENT AND DESIGN two sleeved feed-risers and four iron chills. The legs of the casting are numbered 1–9. The individual legs A test casting composed of nine bent legs was designed to see a wide range of solidification and feeding work in conjunction with three different filling systems. conditions. The filling systems work together with the casting to establish different filling conditions that range from tranquil to turbulent within the casting. SOLIDIFICATION SYSTEM The solidification system is designed to provide a wide The casting orientation and solidification system were variety of thermal and feeding conditions to the nine legs. kept identical between the filling systems in order to Horizontal legs 1 and 4 are chilled from below and maintain identical solidification conditions. Thus, the vertical legs 7 and 9 are chilled from the side. Legs 1 and effect of turning on and turning off filling damage within 7 are located far from the feed-risers, whereas legs 4 and the casting is provided by the three filling systems, while 9 are relatively close to feed-risers. The chills are cast the identical casting orientation and solidification system iron and are approximately 50mm (2in) wide by 70mm maintain solidification conditions. These conditions (2.75in) long by 35mm (1.4in) thick. Feed is provided established differentiation between solidification related from two sleeve insulated feed-risers. The center feed- defects and filling related defects. riser is 38mm (1.5in) in diameter and the other is 63mm (2.5in) in diameter. Both are 152mm (6in) tall. The remaining five non-chilled legs are comprised of three vertical legs (2, 5 and 8) and two that are horizontal (3 and 6) as shown in Figure 2.

Page 2 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

The variety of solidification conditions are for testing slowed to safe filling velocity by means of a tangent inlet whether bifilms unfurl in areas that are poorly fed and step. This allows the metal to vortex under the filter slow to solidify.12 Legs 2, 5 and 6 are of most interest in before transitioning upwards into the surge cylinder and this regard, as they are un-chilled and located relatively vertically slotted ingate. 25 PPI ceramic foam filters were far from the feed-risers. Metal subjected to turbulent used for our trial. Note that two ingate connections were filling conditions should be especially evident within necessary in order to avoid subsequent internal spilling these legs, with the expectations that porosity will within the cavity. Also, the two runners that branch off at increase, and tensile properties will decrease. Last, un- the T-junction were calculated to control the flow such chilled legs 3 and 8 are well fed, but slow to solidify. that legs 5–8 fill evenly with legs 1 and 2. If this is not taken into consideration, legs 1 and 2 will fill The feed-risers were kept relatively small and carefully prematurely, eventually sending metal across leg 3 to spill positioned to minimize the effects of venting bifilms from into leg 5. Fluid flow simulation was used to confirm that the casting cavity. The short height reduces the head tranquil filling conditions were maintained (Fig. 4). pressure, to encourage unfurling of any bifilms existing within the casting.

FILLING DESIGNS The solidification system was applied to three filling systems: providing bottom, side and top filling (Fig. 3).

(a) Figure 4. Tranquil filling noted within the BF system.

A side-fill (SF) gating system was used to generate semi- tranquil filling conditions. Legs 1-4 experience no turbulence. However, metal spills through legs 5 and 7 at high velocity (Fig. 5). This double spilling “turns on” the surface turbulence, creating concerns of potential filling damage to these legs and downstream into legs 6, 8 and 9. A fill time of 24 seconds was used to calculate the gating system. An offset pour-basin and ceramic-felt stopper was attached to a tapered sprue, that terminated in a well. A 25 (b) PPI filter separates the well from the runner. The ingate is placed on the top of the runner, forcing air to exit the Figure 3. Three gating systems (a) create different filling conditions while maintaining similar runner system. This type of gating is considered a 16 solidification conditions (b). traditional horizontal gating system as used in current foundry practice. A bottom-fill (BF) gating system was selected to provide tranquil filling conditions to “turn off” all surface turbulence associated with filling damage. A fill time of 25 seconds was used to calculate the gating system. An offset pour-basin13 and ceramic-felt stopper was attached to a hyperbolic sprue. The sprue makes a radiused 90- degree transition at the bottom, into the runner, which branches off to terminate at the base of two vortex14ingates. This type of gating is also referred to as a naturally pressurized filling system,15 as it is designed to contain the metal as it falls through the sprue, without exposing the metal front to surface turbulence. While this allows the metal to travel well above .5m/s, it avoids surface turbulence by cradling the natural shape of the metal stream as it is pulled by gravity. The metal is Figure 5. Initial tranquil filling, followed by two turbulent falls into legs 5 and 7 with the SF system.

Page 3 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

The top-fill (TF) gating system17 defies all practical ALLOYS suggestions of bifilm theory, by “turning on” surface Four aluminum alloys were chosen to observe how alloy turbulence throughout the fill. The metal makes a free-fall content may affect the generation of oxide bifilm. The drop of 330mm (13in) through leg 8 before crashing into alloys chosen were A206, C355, A356 and E357. The the bottom of leg 7. As metal floods into leg 6, a back- individual Si, Cu and Mg contents are listed in Table 1. wave occurs, creating turbulence that folds the surface over the incoming flow (Fig. 6a). As metal reaches the top Table 1. Si, Cu and Mg allowances for C355, A356, of leg 5, it races across legs 3 and 4 before falling into leg E357 and A206 (Aluminum Association) 2, crashing at the bottom of leg 1 (Fig. 6b). A subsequent rolling of the surface happens in the angled turn into leg AA Chemistry Si Cu Mg 1, followed by a back-wave as the metal reaches the end Spec: of the bar. As legs 1 and 2 fill, a final back-wave forms in C355 4.50-5.50 1.0-1.50 .40-.60 leg 3. Throughout the fill, prominent eddies are A356 6.50-7.50 0.20 .25-.45 established in chilled legs 1, 4, 7 and 9. The eddies offer E357 6.50-7.50 - .60-.70 the potential to congregate filling related bifilms, and the A206 0.05 4.20-5.00 .15-.35 chills test whether rapid solidification rates can hinder the unfurling of bifilms. MOLDING 3D printed sand cores were chosen for this project. They were all printed from the same source. This removed several variables that may exist in a multi-foundry experiment involving sand molds. The cores all have the same type of sand and same level of resin binder to help ensure the same core density and permeability. Printed molds also ensure dimensional stability as pockets are printed to contain items such as the chills and sleeves.

All other molding components were distributed from one source to ensure all parties involved with mold assembly had access to the same sleeves, filters, chills and stoppers.

MELT PREPERATION Each alloy was rotary degassed with dry argon until 98% density was reached. Grain refinement via 5:1 TiB rod was carefully administrated (adding .015-.025% wt. Ti) 18 (a) with care to allow for oxide sedimentation , while modification with Sr was purposely avoided, as recommended per bifilm theory19 to avoid the potential of introducing hydrogen porosity to the melt. Care was also promoted in the metal handling and processing handling aspects in accordance with commercial best practices and bifilm theory recommendations; dry/clean/pre-heated equipment, avoid splashing, gentle handling of the pour ladle. The melt was held at 750C (1380F).

POURING The molds were poured in melt-heats according to their individual alloys. This included making (1) set of C355 castings, (3) sets of A356 castings, (1) set of E357 castings and (2) sets of A206 castings. The chemistry and specific gravity of each heat was verified prior to pour. Each casting was filled by hand ladle. A filling priority was established to pour the BF mold first, followed by the SF mold, and the TF last. This helped to ensure that the BF mold received the best metal, avoiding the possibility (b) of introducing oxide bifilm contamination from Figure 6. Filling begins with a freefall drop through subsequent pours through the ladle. Each ladle was legs 8 and 7, followed by a back wave in leg 6 (a), and carefully skimmed before each pour. another drop into leg 2 (b) in the turbulent TF system.

Page 4 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

Both the BF and SF involved the use of a stopper in the in legs 6 and 7 of the SF castings and in leg 1 of the TF pour basin. This helps to ensure that any bubbles formed castings; regardless of the alloy (Fig. 8). during the initial splash of metal falling into the pour basin had time to float to the surface of the pour basin; instead of allowing them to be sucked into the sprue. Each system was designed to pour with one full foundry ladle, about 15 kg (33 lb) of metal.

PROCESSING Once cooled, the castings were extracted from the sand molds. To maintain the original surface integrity, the common practice of grit or sand blasting of the castings was not permitted. The gating was removed, with exception to the center feed-riser. The castings were then sent to one source for heat treatment. The 300-series aluminum alloys were heat treated to the T6 condition. The 206 alloy castings were heat treated to the T4 condition, to magnify tensile elongation sensitivity. The castings were then treated with alkaline-water solution, followed by a nitric acid etch and water rinsing, to clean the castings prior to dye penetrant inspections.

Figure 8. Dye penetrant inspection of the drag (bottom VISUAL AND DYE PENETRANT INSPECTIONS view) of a SF casting highlight flow lines and areas Visual and dye penetrant inspections were performed on where metal fell in concentrated location. each set of castings. External evidence of flow lines confirmed the effects of the different filling systems. The BF castings showed no indications of flow lines. The SF SECTIONING castings exhibited flow lines consistent with what has Following dye penetrant inspection, each set of castings been termed oxide flow tubes,20 including a patch of mold were carefully marked for sectioning (Fig. 9). Each leg erosion in leg 6 related to the fall of metal from leg 3. (1-9) was sectioned into three separate bars and identified Likewise, the TF castings exhibited flow lines; including as A, B and C. The A and B bars were reserved for tensile a patch of mold erosion in leg 1 related to the fall of metal testing, per ASTM-B26. The C bars were reserved as from leg 3 (Fig. 7). spares, surface inspections and for limited leak testing.

Figure 7. Visual inspections reveal flow lines (white Figure 9. Marking of bars; Alloy-1 (BF SF TF) – 3 A arrows) and erosion at the base of flow tubes (blue). represents Alloy-Set, Fill System, Leg & Bar.

Dye penetrant inspection helped highlight the existence of RADIOGRAPHY flow lines and patches, where sand grains were attached The sectioned bars were digitally radiographed in two to casting; apparently from the concentration of heat orientations (top-view and side-view) to observe the depth associated with an oxide flow tube. Consistent with their locations of any defects noted. An additional image was respective filling systems, these patches only showed up taken after turning the A and B bars, along with the

Page 5 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages) unmachined C bars, for reference (Fig. 10). The unmachined bars were plotted against their individual frame grade per AMS-2175A, gas porosity round, shrinkage cavity, shrinkage sponge & foreign material less dense. For X-ray technique, see (Fig. 11).

Figure 11. Digital radiography X-ray technique.

TENSILE TESTING All A and B bars from the first four sets were CNC turned and pulled to obtain tensile results: 216 bars total. In addition, a second set of A bars was pulled from duplicate sets in A356 and A206. After pulling the bars, they were sent to UAB for subsequent fracture surface inspections and SEM analysis.

LEAK TESTING Leak testing was done to select C bars of the C355 and A206-2 castings, with a focus on looking for a correlation to leak paths and filling damage. Bifilms are said to unfurl in conditions where slow solidification exists, and bubbles are said to leave a fine leak path in the form of an oxide bubble trail that follows in the wake of the bubbles’ buoyant journey to the surface. With these assumptions, select bars were chosen for leak testing, by means of machining a center hole and fitting the open end with an air-tight fitting. The bars were subjected to 20 PSI air pressure and inspected under water. Areas that leak are evident by bubbles that exit the sample (Fig. 12).

(a)

Figure 10. Five images were used to capture each (b) casting, with the bars in two orientations; followed by a post machining x-ray image. Figure 12. Machining (a) for leak testing select bars, leak noted as bubbles on bar A206-2 TF 5C (b).

Page 6 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

C355 RESULTS

Table 2. Composition (wt%), Temperature and Fill Times

C355 Si Fe Cu Mn Mg Zn Ti A-A Standard 4.5-5.5 0.2 1.0-1.50 0.1 .40-.60 0.1 0.2 ALCAN C355 ingot 5.00 0.06 1.30 0.00 0.60 <0.01 0.10 Post-pour spectro. 5.132 0.07 1.313 0.009 0.510 0.005 0.152

Holding Temp 1380F 23s BF fill time 748.9C 27s SF fill time 18s TF fill time

Figure 15. Quality index plot of C355-T6 A and B bars, according to BF, SF and TF systems.

Table 2 lists the general reference of melt chemistry, melt temperature and the fill times as recorded for each pour.

Figure 13 is a radiographic frame-grade plotting of the defects noted for each bar within each of the individual filling systems. It can be noted in the C355 BF system that bars 2B, 2C, 3A, 3B and 3C contained significant amounts of shrinkage cavity. Bar 8C contained significant amounts of shrinkage sponge. All the bars contained a small amount of round gas porosity.

The C355 SF system also contained shrinkage in bars 3A, 3B, 3C and 5B; with some sponge in 5A & 5B.

The 355 TF system showed shrinkage in bars 2B, 5B, 5C, and 8C: and sponge in 7B. More notable, is that many of the bars in the C355 TF system contain foreign material less dense. The side view of these bars showed that this defect was confined to the cope surfaces, consistent with Figure 13. Frame-grade plot of un-machined bars. filling related damage. Figure 14 compares the C355-T6 elongation averages of the chilled and un-chilled bars within the individual filling systems. It can be noted that TF chilled and un-chilled average E% drops slightly to the BF and SF bars.

Figure 15 is a tensile plotting of the C355-T6 quality index of the A and B bars for each of the filling systems (Eqn. 1). In general, spikes in radiographic defects related to shrink, sponge and gas tended to correspond well with the noted decline in the tensile quality of individual bars. However, upticks in less dense foreign material did not always relate to tensile results. It is believed that the subsequent machining of the bars removed much of the

buoyant filling damage noted as less dense foreign material. Fig. 14 Elongation average of chilled and un-chilled C355-T6 bars per BF, SF and TF systems. Eqn. 1

Page 7 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

A356-1 RESULTS

Table 3. Composition (wt%), Temperature and Fill Times

A356-1 Si Fe Cu Mn Mg Zn Ti A-A Standard 6.5-7.5 0.2 0.2 0.1 .25-.45 0.10 0.2 Post-pour spectro. 7.180 0.093 0.013 0.004 0.368 0.134

Holding Temp 1376F 27s BF fill time 746.7C 28s SF fill time 15s TF fill time

Figure 18. Quality index plot of A356-1 T6 A and B bars, according to BF, SF and TF systems.

Table 3 lists the general reference of melt chemistry, melt temperature and the fill times as recorded for each pour.

Figure 16 is a radiographic frame-grade plotting of the defects noted for each bar within each of the individual filling systems. It can be noted in the A356 BF system that bars 2B, 3B and 8B contained a mix of minor defects. Bars 6B and 8C had trace amounts of foreign material less dense. All the bars contained a small amount of gas.

The A356 SF system also contained shrinkage in bars 3B; with trace defects in 1C, 3A, 3C and 6B.

The A356 TF system showed trace shrinkage in bar 2C and sponge in 6B and 6C; with trace defects in 2C, 3B, 3C, 4C, 6A.and 8A. More of the TF system contain foreign material less dense. The side view of these bars showed that this defect was confined to the cope surfaces, consistent with filling related damage.

Figure 16. Frame-grade plot of un-machined bars. Figure 17 compares the A356-T6 elongation averages of the chilled and un-chilled bars within the individual filling systems. It can be noted that the TF chilled average E% drops slightly as you compare with the BF and SF bars, while the un-chilled TF bars are slightly improved over the BF and SF bars. Figure 18 is a tensile plotting of the A356-T6 quality index of the A and B bars for each of the filling systems. In general, spikes in radiographic defects related to shrink, sponge and gas tended to correspond well with the noted decline in the tensile quality of individual bars. However, upticks in less dense foreign material did not always relate well to the quality of the tensile results. It is interesting to note that 3A (positioned below a feed-riser) is of better quality than adjacent 3C. It

is believed that the subsequent machining of the bars Figure 17. Elongation average of chilled and un-chilled A356-1 T6 bars according to BF, SF and TF systems. removed much of the buoyant filling damage noted as less dense foreign material.

Page 8 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

E357 RESULTS

Table 4. Composition (wt%), Temperature and Fill Times

E357 Si Fe Cu Mn Mg Zn Ti A-A Standard 6.5-7.5 0.07 - 0.10 .60-.70 - .10-.20 Trialco E357.2 ingot 7.36 0.050 0.00 0.00 0.61 0.01 0.12 Post-pour spectro. 7.38 0.056 0.021 0.004 0.576 0.018 0.136

Holding Temp 1383F 25s BF fill time 750.6C 24s SF fill time 11s TF fill time

Figure 21. Quality index plot of E357-T6 A and B bars, according to BF, SF and TF systems.

Table 4 lists the general reference of melt chemistry, melt temperature and the fill times as recorded for each pour.

Figure 19 is a radiographic frame-grade plotting of the defects noted for each bar within each of the individual filling systems. It can be noted in the E357 BF system that several bars include a mix of minor foreign material less dense and sponge porosity defects: showing up in bars 1A, 1B, 1C, 2A, 2B, 2C, 3A, 3B, 3C, 5A, 5B, 5C, 6A, 6B, 6C, 8B and 8C. All the bars contained a small amount of gas.

The E357 SF system contained notable shrinkage in bars 2A, 2B, 2C, 3A, 3B and 3C; with trace amounts of foreign material less dense within most of the bars.

The E357 TF system showed notable amounts of foreign material less dense in bars 5A, 5B, 5C and 6B, with trace amounts in bars 1A, 1B, 2B, 3A, 3B, 3C, 4A, 4B, 6C, 9A Figure 19. Frame-grade plot of un-machined bars. and 9C. The side view of these bars showed that this defect was confined to the cope surfaces, consistent with filling related damage. Figure 20 compares the E357-T6 elongation averages of the chilled and un-chilled bars within the individual filling systems. It can be noted that TF and SF chilled bars average E% improved slightly as you compare to the BF bars, while the un-chilled results appear to be statistically insignificant.

Figure 21 is a tensile plotting of the E357-T6 quality index of the A and B bars for each of the filling systems. In general, spikes in radiographic defects related to shrink, sponge and gas tended to correspond well with the noted decline in the tensile quality of individual bars. It is believed that the subsequent machining of the bars removed much of the buoyant filling damage noted as less dense foreign material. Figure 20. Elongation average of chilled and un-chilled E357-T6 bars according to BF, SF and TF systems.

Page 9 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

A206-1 RESULTS

Table 5. Composition (wt%), Temperature and Fill Times

A206-1 Si Fe Cu Mn Mg Zn Ti A-A Standard 0.05 0.1 4.2-5.0 .20-.50 .15-.35 0.10 .15-.30 Trialco A206.2 ingot 0.023 0.030 4.64 0.34 0.29 0.01 0.19 Post-pour spectro. 0.039 0.041 4.78 0.348 0.245 0.017 0.228

Holding Temp 1376F 25s BF fill time 746.7C 23s SF fill time 10s TF fill time

Figure 24. Elongation average of chilled and un-chilled A206-T4 bars according to BF, SF and TF systems.

Table 5 lists the general reference of melt chemistry, melt temperature and the recorded fill times for each pour.

Figure 22 is a radiographic frame-grade plotting of the defects noted for each bar within each of the individual filling systems. It can be noted in the A206 BF system that several bars include a mix of shrinkage cavity, shrinkage sponge and gas porosity round. However, the chilled bars in legs 1, 4, 7 and 9 are nearly free of defects. Also, none of the bars in any of the systems show foreign material less dense.

Several bars in the A206 SF system contain a mix of shrinkage cavity, shrinkage sponge and round gas porosity. However, the chilled bars in legs 1, 4, 7 and 9 are nearly free of defects.

The A206 TF system showed trace amounts of shrinkage cavity, shrinkage sponge and gas porosity round. The side view of these bars showed that this defect was confined to Figure 22. Frame-grade plot of un-machined bars. the cope surfaces, consistent with filling related damage. As with the A206 BF and SF systems, the chilled bars in legs 1, 4, 7 and 9 are nearly free of defects. Figure 23 compares the A206-T4 elongation averages of the chilled and un-chilled bars within the individual filling systems. It can be noted that TF and SF chilled average E% are both notably improved over the BF system. The SF un- chilled bars were slightly improved over the TF bars, followed by the lowest average in the BF bars.

Figure 24 is a tensile plotting of the A206-T4 quality index of the A and B bars for each of the filling systems. In general, spikes in radiographic defects related to shrink, sponge and gas tended to correspond well with the noted decline in the tensile quality of individual bars.

Figure 23. Quality index plot of A206-1 T4 A and B bars, according to BF, SF and TF systems.

Page 10 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

SUPPLIMENTARY INSPECTION RESULTS Set A356-3 was heat treated to the T4 condition. Elongation results showed the SF and TF systems Set A356-2 was cut into bars for radiography purposes, outperformed the BF system (Table 6 and Fig. 27). and not turned into tensile specimens. The cope surfaces of the TF casting are of primary interest, as they contain Table 6. Composition (wt%), Temperature and Fill pits consistent with the other samples (Fig. 25). Times A356-3 Si Fe Cu Mn Mg Zn Ti A-A Standard 6.5-7.5 0.2 0.2 0.1 .25-.45 0.10 0.2 Post-pour spectro. 7.170 0.117 0.021 0.017 0.407 0.006 0.156

Holding Temp 1376F 27s BF fill time 746.7C 28s SF fill time 15s TF fill time

Figure 25. A356-2 TF Bar 6B viewed in two positions, show the voids resting tight to the cope surface.

Set A356-3 was sent out for CT surface scanning, after digital radiography noted sub-surface voids in the other TF samples. The CT surface scans (complimentary of

YXLON) clearly revealed the sub-surface voids lying just below the surface of the TF system (Fig. 26). This Figure 27. Elongation average of chilled and un- confirmed the buoyant nature of the voids created in chilled A356-3 bars according to T4; BF, SF, TF. turbulent filling conditions. By comparison, BF and SF systems were nearly free of sub-surface voids. Set A206-2 was heat treated to the T4 condition. Similar to the A206-1 and A356-3 findings, the elongation results showed the SF and TF systems unexpectedly outperformed the BF system (Table 7 and Figure 28).

Table 7. Composition (wt%), Temperature and Fill Times A206-2 Si Fe Cu Mn Mg Zn Ti A-A Standard 0.05 0.1 4.2-5.0 .20-.50 .15-.35 0.1 .15-.30 Trialco A206.2 ingot 0.020 0.030 4.800 0.380 0.320 0.010 0.180 Post-pour spectro. 0.025 0.050 4.518 0.384 0.267 0.014 0.207

Holding Temp 1380F 25s BF fill time 748.9C 25s SF fill time 14s TF fill time

Figure 26. CT surface probe reveals the cope surface of TF is pitted, whereas BF & SF are free of voids. Figure 28. Composition (wt%), temperature, fill times and elongation averages for A206-2 T4; BF, SF, TF.

Page 11 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

STATISTICAL ANALYSIS FRACTURE AND RAW SURFACE ANALYSIS

Statistical analysis of the tensile results from the first Several metallographic techniques were used at UAB to four alloy sets were compiled at UAB by means of inspect the fracture and casting surfaces. Oxide film was ANOVA (analysis of variation). Comparisons were first noted on casting surfaces, on dendrite porosity and on at made between the six A and B bars of each leg set. least one tensile bar fracture surface (Fig. 30). These results were then compared within their respective alloy on the basis of chilled versus un-chilled, and then in regards to their respective filling systems. Figure 29 lists the results of the A356-1 set. No statistical significance was noted in between the three filling systems, but chilled versus un-chilled bars consistantly showed marked improvements. This was also noted with C355 and E357 samples. The A206-1 showed the BF system actually had lower tensile properties than the TF and SF filling systems.

Figure 30. Examples showing an oxide film on the fracture surface of A356-1 SF 5B and wrinkled skin noted on the cope casting surface of A206-1 SF 6C.

A summary of these statistical inspections was shared at the AFS Aluminum Conference of 2018; concluding:

• Turbulence had no consistent effect on the C355, A356, E357 or A206 tensile properties • 3XX-series showed no statistical difference of tensile properties between the gating systems • A206 actually had lower properties with the BF gating system samples • Elongation, UTS and QI was reduced by bigger flaws on fracture surface

The porosity, silicon and brittle Fe intermetallics noted on the fracture surfaces had a bigger impact on tensile Figure 29. ANOVA comparison of the A356-1 Quality results than “oxide films.” Turbulence had no consistent Index results of chilled to un-chilled, and below the effect on the porosity levels within the C355, A356, comparison between TF, SF and BF filling systems. E357 and A206 fracture surfaces.

Page 12 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

LEAK TESTING THERMAL CONDITIONS REVISITED

Leak testing was performed on select bars to look for Initial expectations were that bottom filling would evidence that oxide skins and bubble trails cause leak reduce porosity (shrinkage and microshrinkage) and an paths. Bars 4C 5C and 6C were inspected from the three improve tensile properties; especially in elongation. filling systems of the 355 series. Only 355-TF 5C and Following the unanticipated results, simulation was 6C leaked, midlegth along the cut surface (Fig. 31). revisited for possible explinations. The first clue came from noting how the BF system establishes an unfavorable temperature profile, leaving the center feed- riser cooler than in the SF and TF systems (Fig. 33). This establishes a shallower thermal gradient, which in turn increases porosity and lowers tensile properties.

(a)

(b) Figure 31. C355 bars TF 5C (a) and TF 6C (b) leaked.

Select bars from the A206-2 TF were chosen to capture Figure 33. End of filling temperatures show the center a thorough inspection of what is happening within the riser of the BF system is cooler (light blue) than the SF (dark blue) and TF (orange) systems. most turbulent filling system. Center B bars were chosen from 1B, 2B, 4B and 7B. C bars were chosen Niyama criterion is useful for noting shallow thermal from 3C, 5C 8C and 9C. Bars from leg 6 were reserved gradients (Eqn. 2). With the scale set to .2 maximum, for tensile and surface inspections. Only 5C leaked, and areas in red have a shallow thermal gradient (Fig. 34a). did so through the cast surface near the inside bend Additionally, hot spots differences can noted (Fig. 34b). toward leg 3 (Fig 32). Based on assumptions that the These subtle difference in solidification explain the leak paths caused by turbulence would follow the differences in porosity and tensile elongation. general filling path, no conclusive connections were made to filling conditions.

Eqn. 2 Where: G = thermal gradient, R = cooling rate

(a)

(a)

(b)

(b) Figure 34. Niyama Criterion (a) highlighting shallow thermal gradients in red and Hot Spots FS Time (b) Figure 32. Of the eight A206-2 TF bars tested (a), only showing isolated hot spots within the cavity in blue; 5C leaked and did so near transition to leg 3 (b). scales set at .2 maximum and 50-400s respectively.

Page 13 of 14 2020 AFS Proceedings of the 124th Metalcasting Congress Paper 2020-051 (14 pages)

CONCLUSIONS ACKNOWLEDGEMENTS

Inconsistent with bifilm theory anticipations, the overall This work was inspired by the ongoing work of John porosity21 and tensile quality22 of the BF castings fared Campbell and his generous correspondence over the past similar and even slightly less than the SF and TF castings. several years, along with Bob Puhakka. Generous Directional solidification and cooling rates played a more correspondence was also shared from Raymond Monroe significant role in controlling casting quality. (SFSA) and Christoph Beckermann (UI). Special thanks to the team for all your in-kind support and to the AFS for The turbulent TF filling system demonstrated the ability the supporting funds that made this project possible. to create repeatable filling damage in the forms of Additional in-kind donations came through Richard Cox bubbles, seams and flow tubes. The bubbles associated of Yxlon for the CT scans and through James Schmahl of with the TF system were noted as being very buoyant, Selee Corporation for the filters. clinging to within a few mm of cope surfaces. The buoyant nature of the bubbles indicates that these defects could be removed by venting areas of concern, or by REFERENCES machining cope surfaces. 1. Campbell, J., “Castings, 2nd Edition,” Butterworth- The semi-tranquil SF filling system demonstrated the Heinemann, Oxford, UK (2003); pp. 2; 1st Edition, ability to create filling damage in the forms of flow lines (1991). and flow tubes. Despite the metal-falls that occur in legs 5 2. Campbell, J., “Concise Castings: A Casting and 7, little evidence of bubble formation was noted. Workshop Lecture,” American Foundry Society, Perhaps they vented out the feed-riser above leg 8. Or, Schaumburg, IL, pp. 1 (2010). perhaps the flow tubes offered protection against bubble 3. AFS Div. 2E, “Aluminum Permanent Mold formation. The oxide film noted in A356-1 SF 5B (Fig. Handbook,” American Foundry Society, 26) is most likely a remnant of a flow tube that formed in Schaumburg, IL, pp. 3–25, 18–1 (2010). the drop within SF leg 5. 4. Armao, F., “Got porosity? Do your homework before assessing blame,” The WELDER, (Sept/Oct 2018). The tranquil filling conditions of the BF system 5. Campbell, J., “Concise Castings: A Casting demonstrated the ability to avoid bubble formations, Workshop Lecture,” American Foundry Society, seams and flow tubes. However, it did so at the cost of Schaumburg, IL, p. 16, (2010). establishing adverse thermal gradients by bringing heat 6. Campbell, J., “Complete Casting Handbook, 2nd through the bottom of the casting. These thermal Edition,” Butterworth-Heinemann, Oxford, UK, pp. conditions increased porosity and decreased tensile 17–90 (2015). properties. 7. Ibid. pp. 25 8. Ibid. pp. 373 The oxide film noted in bar A356-1 SF 5B is of serious 9. Ibid. pp. 67–70 concern. Evidence of flow tubes existed in both the TF 10. Ibid. pp. 346 and SF systems. The neutrally buoyant nature of this 11. Ibid. pp. 75 defect highlights the potential to leave unbonded bifilm 12. Ibid. pp. 72 skin within a casting. These unbonded bifilms do reduce 13. Ibid. pp. 662–669 tensile properties and likely create leak paths. The defect 14. Ibid. pp. 723 in A356-1 SF 5B was undetectable in radiography. 15. Ibid. pp. 753 16. AFS 5G Committee, “Basic Principles of Gating & The concluding comments are: Risering 2nd Edition,” American Foundry Society, Schaumburg, IL, pp.26–24 (2008). 1. Turbulent surface filling conditions can create 17. Ibid. pp. 77–79 buoyant bubbles surrounded by oxide skin 18. Campbell, J., “Complete Casting Handbook, 2nd 2. Turbulent surface filling conditions do not Edition,” Butterworth-Heinemann, Oxford, UK, appear to influence bulk tensile results p. 243 (2015) 3. Turbulent surface filling conditions do not 19. Ibid. pp. 254–258 appear to influence shrinkage porosity formation 20. Ibid. pp. 56–57 4. Remnants of flow tube oxide skin can create 21. Ibid. pp. 25, 341 unbonded seams that reduce tensile properties 22. Campbell, J., et al, “Influence of Oxide Film Filling 5. Back waves can create cosmetically undesirable Defects on the Strength of Al-7Si-Mg Alloy oxide skin seams on cope surfaces Castings,” Transactions of the American Foundrymen’s Society, pp. 341–347 (1994).

Page 14 of 14