Click here to see this story as it appears in the September 2017 issue of Modern Casting

The dimensional accuracy at which sand castings can now be produced has approached that of investment castings.  3-D sand printing technologies have greatly improved the dimensional accuracy of molds and cores but have failed to match the surface smoothness of conventional sand castings, let alone investment castings.

Investment casting provides very smooth parts with excellent feature resolution and dimensional accuracy. 3-D printed sand molds and cores may provide a cost-effective alternative to investment casting if the process can meet both dimensional and surface requirements. 

Although many changes and improvements have been made in the area of foundry consumables, sand is the one material that has stayed somewhat constant. After mining and washing, if required, foundry sands are classified into individual or two-mesh groupings and stored. They are combined into normal distributions for shipping to the foundry customer. Although there are many different mine distributions, sand of similar AFS-grain fineness number is supplied in similar distributions. Surface finish is an integral part of casting quality specifications.  Rough internal surface finishes on castings can cause the loss of efficiency for both fluids and high velocity gasses. Such is the case for turbocharger and intake manifold components.  The University of Northern Iowa has been investigating mold material characteristics that affect surface smoothness for castings.  The research was conducted on aluminum castings but has applications and relevance in ferrous alloys that don’t exhibit defects such as penetration or fused sand defects. The study investigates the influence of molding media characteristics such as sand fineness, material type and refractory coating selection. The goal of the project was to accomplish investment casting surface finishes in sand cast parts.

Permeability and Surface Area Results
AFS permeability is defined as the amount of time it takes for a known volume of air to pass through a standard sample at a head of 10 cm of water.  Simply, the AFS permeability represents the amount of open spaces between the aggregate grains that allows air to pass. The GFN of a material significantly changes the permeability until 80 GFN, where the trend appears to level out.

The data shows that the same surface roughness can be achieved with any particle shape at different rates.  The spherical and round grain materials improve casting smoothness at an accelerated rate in comparison to angular and sub-angular aggregates.

Gallium Contact Angle Results
Contact angle measurements were conducted to measure the relative wettability of the bonded molding aggregates with liquid metal using a liquid gallium test. Ceramic sands had the highest contact angle while zircon and olivine shared a similar lower contact angle.  The gallium exhibited hydrophobic behavior on all sand surfaces.  A similar AFS-GFN was used for all of the samples.  The results indicate the contact angle for the sand types depended heavily on the aggregate grain shape as shown on the secondary axis, rather than base material.  The ceramic sands had the roundest shape and olivine sands exhibited a highly angular shape.  While surface wettability of the base aggregate may play a role in casting surface finish, the range of contact angle measurements in the test series was subordinate to grain shape. 

Surface Roughness Results from Test Castings
Surface roughness results were measured using a contact profilometer. There was a significant improvement in surface smoothness from the three-screen 44 GFN silica to the four-screen 67 GFN silica. Changes beyond 67 GFN did not show an impact on surface roughness despite variation in distribution width. The threshold value of 185 RMS is observed.

A large improvement in smoothness can be observed between the 101 and 106 GFN materials. The 106 GFN sand has over 17% more 200 mesh material in the screen distribution. The two-screen 115 and 118 GFN materials resulted in a decrease in smoothness.  The 143 GFN sand resulted in similar readings to the 106 GFN zircon. The threshold value is 200 RMS.

A steady improvement in surface smoothness was observed from the four-screen 49 GFN chromite to the three-screen 73 GFN chromite despite the particle distribution becoming narrower. A 19% increase in retention of the 140-mesh screen was seen in the 73 GFN chromite in comparison to the 49 GFN.  A significant increase in casting smoothness was shown from the three-screen 73 GFN to the four-screen 77 GFN chromite sands regardless of their similar grain fineness numbers.  No change in smoothness was observed between the 77 GFN and 99 GFN chromite materials.  Interestingly, the two sands shared a very similar retention in the 200-mesh screen. The threshold value is 250 RMS.

There is a significant improvement in casting smoothness from the 78 GFN olivine to the 84 GFN olivine despite the narrower distribution.  An increase of 15% retention in the 140-mesh screen was visible in the 84 GFN olivine.  There is significance between the 84 and 85 GFN olivine.  The 85 GFN olivine improved the smoothness by 50. The 85 GFN olivine is a three-screen sand with almost 10% retention in the 200-mesh screen while the 84 GFN olivine is simply a two-screen material.  A steady improvement in smoothness can be observed from the 85 GFN olivine to the 98 GFN olivine.  The screen distribution shows an increase of 5% retention in the 200-mesh screen. No change was seen from the 98 GFN to the 114 GFN olivine despite an increase in 200 mesh retention of nearly 7%.  A threshold value of 244 RMS can be observed.

The surface roughness results for the castings obtained from ceramic cores show slight improvement between the 32 GFN and 41 GFN materials.  There was an increase in retention of the 70-mesh screen by 34% in the 41 GFN sand.  A significant increase in smoothness was observed between the 41 GFN and 54 GFN ceramics.  The 54 GFN material had over 19% greater retention in the 100-mesh screen in comparison to the 41 GFN material.  This improvement occurred despite the distribution narrowing in the 54 GFN material.  The largest impact in the ceramic results was seen between the 54 GFN and 68 GFN sands.  The 68 GFN sand had 15% higher retention in the 140-mesh screen which widened the distribution.  Despite an increase of over 40% retention in the 140-mesh screen, little improvement was observed between the 68 GFN and 92 GFN materials.  The threshold value is 236 RMS.

The surfaces generated by the 3-D printed sands are significantly rougher than a rammed sand surface using the same aggregate.  The samples printed in the XY orientation provided the smoothest test casting surface while those printed in the XZ and YZ orientation resulted in the roughest.

The rammed silica uncoated 83 GFN silica sand resulted in a roughness value of 185 RMS.  Although the castings appeared smoother, the refractory coatings increased the surface roughness as measured by the profilometer. The alcohol-based alumina coating exhibited the best performance while the alcohol based zircon coating resulted in the highest roughness.  The 83 GFN 3-D printed samples showed the opposite effect.  While the uncoated sample printed in the most favorable orientation of XY, the uncoated sample exhibited a casting roughness of 943 RMS.  The coatings smoothed the surface substantially from the uncoated surface finish from a low of 339 to a high of 488 RMS.  It appears the surface finish of the coated sands is somewhat independent of the roughness of the substrate sand and depends heavily on the formulation of the refractory coating.  3-D printed sand, although starting with a much rougher surface finish, can be improved significantly with the use of refractory coatings.

Conclusions
Currently available molding aggregates have the ability to achieve surface roughness values of less than 200 RMS microinches.  These values are slightly within the values associated with investment castings.  For the materials tested, each exhibited a decrease in casting roughness with increasing aggregate AFS grain fineness.  This was true with all materials up to a threshold value, at which time no further decrease in casting roughness was seen with increasing AFS-GFN.  This was supported by previously conducted research. 

Within all material groups, the effect of AFS-GFN was secondary to both calculated surface area and aggregate permeability.  While permeability can be thought to describe the open areas of the compacted sand, the surface area better describes the screen distribution of the sand and corresponding amount of fine particles.  Both permeability and surface area were directly related to casting surface smoothness. It should be noted that this was true for aggregates within a shape group.  Although angular and sub-angular aggregates had high surface areas, their permeability was high and indicated an open surface.  Spherical and rounded aggregates exhibited the smoothest surfaces combining low permeability with high surface area. 

It was originally believed surface wettability as measured by contact angle between liquid metal and the bonded aggregate was a critical factor in the resulting casting surface finish.  While it was shown that contact angle on various materials at similar AFS-GFN was not proportional to the casting roughness, it was confirmed that grain shape was a major factor.  The absence of a relationship between contact angle and casting surface roughness might be explained by the fact that grain shape was seen as a major influence in surface roughness.  There is a significant possibility that the contact angle of various materials was affected more by the grain shape and resulting surface smoothness than that of the wettability of the material alone. 

As with all measuring instruments, artifacts of the test method may influence the results to some degree.  The increase in casting roughness, although visually the castings looked smoother with the application of refractory coating, may be due to the shape of the peaks and valleys created with the coatings.  By definition and measurement, the refractory coatings only increased the surface roughness over non-coated samples.  All of the refractory coatings were very successful in improving the surface roughness of the 3-D printed sands. It appeared that the surface finish of the test castings from coated samples was somewhat independent of the starting substrate sand.  The coatings had a major effect on the surface finish but further work is required to revise the coatings to improve casting finishes.  

This article is based on paper 17-094 that was presented at the 2017 Metalcasting Congress.