DESCRIPTIVE STATISTICAL ANALYSIS OF FOUNDRY PROPERTIES OF
SAND PARTS PRODUCED BY THREE-DIMENSIONAL PRINTING
K. Nyembwe
University of Johannesburg, Johannesburg, South Africa M. Motadi
Metal Casting Technology Station
University of Johannesburg, Johannesburg, South Africa E.M. Gonya
University of Johannesburg, Johannesburg, South Africa
ABSTRACT
Rapid sand casting technology is a growing field of additive manufacturing (AM) or three-dimensional printing. Applications of rapid sand casting include amongst other the production of foundry sand moulds and cores. The properties of parts produced by rapid sand casting technologies have been assessed in previous studies including dimensional accuracy, surface finish and strength. However, very little attention has been devoted to the statistical variability of mechanical properties of three-dimensional printed sand components. This knowledge is critical to ascertain the process capability of rapid casting technologies for consistent production of foundry parts and ultimately the quality of castings. To that end, the statistical variability of foundry properties of sand parts produced on a three- dimensional printer was assessed in this study. Descriptive statistical analysis results show that sand samples produced by three-dimensional printing are not always consistent in terms of mechanical properties. The highest variability of
mechanical properties in sand samples was found with the tensile strength. The bend strength in sand samples was found to have the lowest variability with a symmetrical normal distribution of data.
Key words: Three Dimensional Printing, Rapid sand casting, Statistical variability.
INTRODUCTION
Rapid sand casting refers to the direct fabrication of sand moulds and cores through the use of additive
manufacturing (AM) techniques. AM is based on layer by layer production of parts from three-dimensional digital
files. The process of AM is often cited as one of the innovative features of the fourth industrial revolution.1 Benefits of AM to foundry technology include the ability of building geometrical complex patterns, cores and moulds that could not have been easily produced through traditional manufacturing methods.2 An additional advantage of rapid sand casting is the reduction of the overall lead design and manufacturing time allowing castings to quickly reach the market.3, 4, 5 ,6 ,17. Numerous AM systems are commercially available for rapid sand casting applications including Zcast process, Exone and Voxeljet technologies. 7, 8, 16
Foundry properties of the sand moulds and cores
manufactured through AM processes have been the object of various studies found in the literature.5, 9 These properties include amongst other the dimensional accuracy, surface finish and mechanical properties. One study analysed the tensile strength and permeability of baked samples produced by the ZCast process.10 A mathematical model was then developed that could predict these properties as a function of the temperature and time of baking of the printed moulds. Snelling and his colleagues have investigated in detail the properties of ZCast moulds using different types of powders and binders. 11, 12 The effects of these moulding materials on the casting properties in terms of surface roughness, density, hardness, and porosity were also assessed. From these studies, several recommendations were made on the performance of commercially available raw materials. Other recent studies have looked at mechanical properties of AM suitable for aluminium cast alloys13
Although the foundries properties of sand parts produced through AM processes have been evaluated and compared
to tradidional manufacturing processes as highlighted in the paragrapgh above, the statistical variability of these properties has not received sufficient attention at least as far as the mechanical properties are concerned. This question is important in order to ascertain the consistency of sand parts produced by rapid sand casting processes. Such an analysis will contribute to a better understanding of properties of AM sand parts for the production of defect free castings. The study could also open the way for effective process control of rapid sand casting processes at a time when these technologies are
considered as promising alternative to traditional foundry processes . This study makes use of descriptive statistics to assess the variability of sand parts produced by three-dimensional printing.
DESIGN OF EXPERIMENTS
The methodology of the study consisted of three steps namely:
‐ Manufacturing of sand test specimens by three-dimensional (3D) printing
‐ Sand testing of mechanical propertities ‐ Descriptive statistical analysis
Manufacturing of sand test specimens by three dimensional (3D) printing
The three-dimensional printer used in this study was a Voxeljet VX 1000 available at Vaal University of Technology (South Africa). Standard operating parameters were set on the printer including a layer thickness set at 300 µm. Silica sand recommended by the equipment manufacturer with the characteristics shown in Table 1 was used for three-dimensional printing of sand specimens. The sand was supplied pre-coated with sulphonic acid activator at 0.3% per weight of sand. Furfuryl alcohol based resin was dispensed through the printer head at 1% per weight of sand during the layer-by-layer manufacturing process. 3D printed sand specimens were subsequently oven-cured 24 hours after printing at a temperature of 110 ͦC for a duration of 2 hours.
Table 1: Properties of imported silica sand
Tensile, bend and friability sand test specimens were printed in one day. The z-direction was used as geometric orientation of samples during additive manufacturing of sand samples. The total number of the different sand test samples are shown in Table 2. Two-dimensional drawings of the test specimens as per American Foundry Society (AFS)’s recommendations14 were used to produce the CAD files for the three-dimensional printing.
Table 2: Number of printed sand test specimens
Sand testing of mechanical properties
Mechanical testing of cured sand test specimens was carried out as per the AFS procedures.14 Appropriate foundry sand testing equipment from Ridsdale and Dietert were used for the determination of tensile and bend strengths and friability resistance.15 Sand testing was conducted at room temperature 24 hours after oven-curing of specimens as explained in the section above. Test results were then recorded for statistical analysis. Statistical analysis
Statistical Package for the Social Sciences (SPSS) software was used for descriptive statistical analysis. Data analysis included the determination of central tendency, measures of variability or dispersion and summary tables as shown in Table 3. The statistical data were also
Sand Parameter
Refractory sand as received
Acid demand Value [ml] 0.00
AFS Grain Fineness 65.45
Average Grain Size [µm] 211.63 Bulk Density [g/cm3] 1.5291 Loss on Ignition [%] 0.53 pH 4.68 SiO2 [%] 97.08 Sinter Point [oC] 1500 Specific Gravity [g/cc] 2.63 Total Clay Content [%] 0.56
Sand specimen Number printed
Tensile 10
Bend 10
graphically presented in the form of box plots and histograms.
Table 3: Data analysis methods
EXPERIMENTAL RESULTS Central tendency
Table 4 shows the average and mean values of the sand testing results for the tensile and bend strengths and friability resistance. The average values of tensile and bend strengths and friability are in line with previous studies that were obtained from three samples.7 These values were discussed and found to generally be lower than the ones achieved with traditional sand moulding methods. In addition, the AM sand samples have poor friability resistance higher than the recommended 10%. The median values of tensile and bend strengths are equal to the average values thus suggesting a symmetrical distribution of data with negligible skewness values. The skewness for the tensile and bend strength were found to respectively be 0.222 and 0.000. There is a non-negligible difference between the median and the average in the case of friability resistance implying a non-symmetric data distribution curve. In the latter case, the skewness was found to be equal to – 0.610.
Table 4: Central tendency of data
Tensile [N/cm2] Bend[N/cm2] Friability[%]
Mean 86.00 294.00 15.90
Median 86.00 294.00 16.00
In terms of three-dimensional printed samples produced, the above observations could indicate that 50 % of sand samples will have their tensile and bend strengths higher than the average while another 50 % of sand samples will have the same mechanical properties lower than the average. This equal distribution around the average in the case of bend and tensile strength does not applied for the friability resistance of sand samples which indicates a left skewness with more than 50% of sand sample having this
mechanical property higher than the average friability resistance.
Box Plots
Figures 2 and 3 show the variation of mechanical properties in AM sand samples in the form of box plots respectively for the strength data and friability data. The range of mechanical properties in sand samples appears to be higher in the case of the bend (38 N/cm2) compared to the tensile strength (25 N/cm2). The range of friability resistance in sand sample was found equal to 3%. All box plots seem to indicate non-presence of outliers. The box plot related to the friability resistance confirmed the left skewness of data in the sand samples.
Figure 2: Tensile and bend strengths box plots
Figure 3: Friability resistance box plot
From the above observations, it does appear that no outliers sand samples in terms of mechanical properties is produced during three-dimensional printing. The
variability of mechanical properties in sand samples
Data analysis Central tendency Mean
Median
Variability Range Inter-quartile range
Standard deviation
Summary tables Histograms Skewness
looking at the range appears to be largest for the bend strength.
Histograms
Figure 4, 5 and 6 show the histograms for the tensile, bend and friability data in the sand samples. These figures are in agreement with the discussion above with regards to symmetry, skewness and spread of mechanical properties in sand samples. The histogram charts provide additional information related to the shape. The tensile and friability data distribution appears to be unimodal following a normal distribution while the bend test data distribution could characterize as bimodal.
Figure 4: Histogram and frequency distribution for tensile strength data
Figure 5: Histogram and frequency distribution for bend strength data
Figure 6: Histogram and frequency distribution for friability data
The figures also indicate the standard deviations for the different mechanical properties in the sand samples. Using these data, the coefficients of variations as the ratio of the standard deviation to the mean were computed and shown in Table 5. These coefficients of variations appear to be of non-negligible magnitude unlikely to be due to testing method but rather induced during the three-dimensional printing. The variability in the tensile strength is the highest followed by the bend and followed by the friability test data.
Table 5: Coefficient of variations in sand test specimens
Mechanical Properties Coefficient of variation [%]
Tensile 6.76
Bend 3.60
Friability 6.25
CONCLUSION
The study attempted to assess the consistency of mechanical properties in sand parts produced by a well-established three-dimensional printing process used for rapid casting application. Under standard opertating conditions of a the three-dimensional printer, the sand samples displayed noticeable variability of mechanical properties. Tensile strength in samples was found to have the highest variability measured as coefficient of variation compared to bend strength and friability resistance. The statistical evaluation of mechanical properties of three-dimensional printed components could contribute to a better understanding of the performance of sand moulds and cores produced by three-dimensional printing and the
casting quality. Future study could include additional properties, comparison of rapid casting technologies, longer time of assessment and bigger sample polpulation. ACKNOWLEDGEMENTS
This project was generously funded by the following organization:
Technology Transfer and Innovation of the Vaal University of Technology
The Metal Casting Technology Station of the University of Johannesburg
REFERENCES
1. W.B. du Preez, and de Beer, D. J. “Implementing the South African additive manufacturing technology roadmap: The role of an additive manufacturing center of competence”,
South African Journal of Industrial Engineering, vol 26, no.
2, pp.85-92, 2015.
2. N. Guo and M. C. Leu, "Additive Manufacturing: technology, applications and research needs," Front Mechanical Engineering, vol. 8, no. 3, pp. 215-243, 2013. 3. D. Dimitrov, W. van Wijck, and N. de Beer, “An
introduction to rapid casting: development and investigation of process chains for sand casting of functional prototypes".
The South African Jounal of Industrial Engineering, vol18,
no.1, pp 157-173, 2007.
4. Stratasys, Sand casting with Polyjet and FDM patterns”, 2017. [online]. Available:
http://www.stratasys.comsolutions/Additive-manufacturing/tools/sand casting. [Accessed 5 July 2017]. 5. D. de Beer, W. du Preez, and W. Greyling, H. et al. “A
south African Additive Manufacturing strategy. Commissioned by the Department of Science and Technology”, 2016. [online]. Available:
http://www.rapdasa.org. [ Accessed 7 July 2017]
6. K. Nyembwe, M. Motadi, P.J.M. van Tonder, D.J. de Beer, and E. Gonya, “Physical properties of sand parts produced using a Voxeljet VX1000 three-dimensional printer”, South
African Journal of Industrial Engineering, vol 27, no.3,
Special edition, pp. 136-142, 2016.
7. ExOne, "Technology Overview," 2016. [Online]. Available: http://www.exone.com/Resources/Technology-Overview. [Accessed 7 July 2017].
8. VoxelJet, "The rapid and economical method for sand casting moulds," 2016. [Online]. Available:
http://www.voxeljet.de/en/services/sand. [Accessed 7 July 2017]
9. Anon, “Moulding material and properties, 2009. [Online]. Available: http://www.scribd.com/mobile/doc/Moulding
sand properties. [Accessed 10 July 2017].
10. S. Singamneni, N. McKenna, O. Diegel, D. Singh, T. Neitzert, J. St Goerge, A. R. Choudhury and P. Yarlagadda, "Rapid Casting: A critical analysis of mould and casting characteristics," Australian Journal of Mechanical Engineering, vol. 7, no. 1, pp. 33-43, 2009
11. D. Snelling; R. Kay, A. Druschitz and C. Williams. (2013). Mitigating Gas Defects in Castings Produced From 3D Printed Molds. Proceedings of the117th MetalCasting Congress, Saint Louis, Missouri.
12. D. Snelling, C. Williams and A. Druschitz, "A Comparison of Binder Burnout and Mechanical Characteristics of Printed and Chemical Bonded Sand Moulds." 2014. [Online]. Available:
http://sffsymposium.engr.utexas.edu/sites/default/files/2014-018-Snelling.pdf [Accessed 21 July 2016].
13. P. Hackney, R. Wooldridge, and C. Connor, “Optimisation of Additive Manufactured Sand Printed Mould Material for Aluminium Castings, 2017. [online]. Available:
www.sciencedirect.com. [Accessed 28 September 2017]
14 American Foudry Society, Mold & Core Test Handbook, United States of America: American Foundry Society, 2006
15. Risdale and Risdale DieterT, Foundry Sand Testing Equipment Operating Instructions,” 2009. [Online]. Available:
http://www.basrid.co.uk/ridsdale/images/pdf/AFS_OIM.pdf. [Accessed 20 July 2016]
16. 3D Systems, "Casting, Patterns & Moulds," 2016. [Online]. Available:
http://www.3dsystems.com/solutions/casting-patterns-and-molds. [Accessed 29 September 2017].
