Investigation and Improvement of Pushing Dislocation in Ceramsite Sand Three-Dimensional Printing

Abstract

Three-dimensional printing (3DP) is considered to be one of the important technologies for a new manufacturing mode. When ceramsite sand is used as a 3DP material to produce a mold (core), the printed layer is prone to deviation from the original location. In this study, the continuous stacking of the printed part deviation was termed as pushing dislocation, and a physical model was designed to investigate the pushing dislocation mechanism. When the gravity of the printing layer and the pressure of the sand scraper decreased, or when the supporting force increased, the angle of the sand scraper and the maximum friction of the prelaying layer on the printed part will reduce the pushing dislocation. To optimize the quality of the ceramsite sand mold, experiments on the pushing dislocation were conducted by altering the recoater speed, layer thickness, and bottom support condition (with or without bottom supporting plate). The sample dimensions were obtained by a 3D imaging scanner, and the gas evolution and ignition loss were measured. The results revealed that the dimensional difference of samples continuously decreased and the pushing dislocation was gradually reduced as the recoater speed and layer thickness increased. The pushing dislocation of the X-direction sample was more severe compared with that of the Y-direction sample. Increasing the layer thickness is an effective way of reducing the pushing dislocation. The bottom supporting plate can reduce the pushing dislocation, but the effect was insignificant.

Introduction

Nowadays, in the aerospace and automotive industry, the sand molds and cores that were traditionally used for metal casting are gradually being replaced by additive sand molds, which are conducive to the production of high-quality castings with complex shapes.^1–5 This additive manufacturing technology for manufacturing sand molds and cores in a foundry is called the powder-binder-jetting process or three-dimensional printing (3DP).^6–9

The 3DP technology is used to obtain the final printed parts through layered processing, layer-by-layer printing, and stacking based on the principle of discrete accumulation.^10–12 This solves the problems of a long development cycle and low production efficiency in a traditional foundry due to avoiding the design and manufacture of the molds, while realizing the full utilization of raw materials.^13–15 Compared with the direct printing metal parts, 3DP sand mold printing is more widely used in casting production owing to its advantages of higher speed, larger forming size, and lower cost.^16,17 In the production and development of complex structural parts such as the engine block and cylinder head, the 3DP process greatly reduces the design time and development cycle of the molds.^18–20

Presently, the 3DP process mostly uses silica sand as raw sand, owing to its advantages of high-temperature resistance and low cost.²¹ However, because the crystal lattice transition of silica from α-quartz to β-quartz sand occurs at 573°C, rapid expansion occurs, which makes the silica sand mold (core) prone to the occurrence of veining defects. This affects the surface quality of the castings, and renders the complex parts incapable of surface cleaning and satisfying the requirements of high surface quality.^22,23 In the 3DP sand mold (core) printing process, silica sand can be replaced by ceramsite sand with a smaller coefficient of thermal expansion and higher surface roundness, which reduces the veining of the castings and improves their surface quality. Zhao et al.²⁴ investigated the relationship between the diameter of coated ceramsite sand and the resin coating thickness, established a particle packing model that describes the particles' achievement of a stable state, and obtained the shrinkage ratios at different particle sizes.

The process parameters of 3DP not only affect the performance of the sand mold (core) but also play an important role in their dimensional accuracy. Miyanaji et al.²⁵ investigated the effect of the printing speed on the dimensional accuracy of the printed samples, and the experimental results revealed that the increase of the printing speed reduced the accuracy, owing to the enhanced spreading of droplets under a more significant inertia force. Lee et al.²⁶ investigated the influence of compressibility and load on layer displacement and concluded that the powder layer position and the powder thickness affect the dimensional accuracy in the vertical direction. Martinez et al.¹⁴ studied the effect of furan binder content on 3D sand printing molds to determine changes in dimensional accuracy. Hackney and Wooldridge²⁷ investigated the direct 3DP process developed by ExOne for the production of a sand mold and evaluated the part characteristics, including the dimensional accuracy, tensile and compressive strength, and high temperature resistance. They concluded that the 3DP process can be used to manufacture sand molds with an accuracy of ±0.5 mm. The 3DP method of printing and layer-by-layer stacking makes it difficult to observe the forming process in real time. The complete shape of the sand mold (core) was observed only after the printing had finished.^4,28–30 In summary, existing research has mainly focused on the dimensional accuracy of the printed parts, while the deformation of the sand mold (core) has rarely been investigated. In this study, the phenomenon of pushing dislocation in 3DP ceramsite sand printing was considered, and its formation was investigated using a mechanical analysis model. The effect of the 3DP process parameters on the pushing dislocation was obtained through experiments. The results revealed that the best process parameters can be selected, and the printing quality of the sand mold (core) was greatly improved.

Pushing Dislocation in Ceramsite Sand 3DP

The pushing dislocation of the 3DP samples of ceramsite sand is shown in Figure 1. According to the statistics, the height difference of the pressing end from the preset value is 0.5–1.0 mm, while the height of the warping end is essentially the same as or slightly less than the preset value.

FIG. 1.

Pushing dislocation that occurred in the ceramsite sand 3DP sample; the white dots indicate the reflective stickers for 3D image scanning. 3DP, three-dimensional printing.

The analysis of the formation of the pushing dislocation is expressed as follows. Figure 2 shows the relative position between ceramsite sand and the recoater during the 3DP process, which can be simplified as the mechanical analysis diagram of sand particles in the 3DP process.

FIG. 2.

Schematic diagram of ceramsite sand 3DP process.

Figure 3a shows the schematic diagram of the ceramsite sand state during 3DP, and can be divided into two parts, A and B, along the boundary between the printed and printing layers. In the process of sand scraping shown in Figure 3b, part A is subjected to its own gravity G_A, the pressure F of the sand scraper, the friction f_A, and the supporting force F_NA of the bottom prelaying sand layer. In Figure 3c, part B is subjected to its own gravity G_B, the pressure F₁ of the printing layer, the supporting force F_NB of the printed layer, the reverse force $f{\prime }_A$ of friction f_A, and the friction f_B of the printed layer. According to the force equilibrium, the pressure F₁ can be expressed as follows:

FIG. 3.

Mechanical analysis and pushing dislocation formation in ceramsite sand 3DP process. (a) State of ceramsite sand, which can be divided into parts A and B. (b) Force analysis of part A. (c) Force analysis of part B. (d) Schematic diagram of sample warping. (e) Pushing dislocation formation between A and B. (f) Separation of A and B.

$$F_1=G_A+Fcos\theta$$ (1)

where θ is the angle of the sand scraper.

When there is no relative movement between part A and part B, part B remains stationary and is exerted by the static friction f_B of the printing layer, which is expressed as follows:

$$f_B=f{\prime }_A={\mu }_1\left(G_A+Fcos\theta -F_{NA}\right)$$ (2)

If f_B is too large and f_B>f_Bmax, the printed part will tilt up and tend to move toward the recoater movement direction, as shown in Figure 3d. Conversely, the unformed part will keep being printed at the preset position. Consequently, the pushing dislocation will begin to form between part A and part B. In this case, the friction of the printed layer changes into the sliding friction f_BH and can be expressed as follows:

$$f_{BH}={\mu }_2\left(G_B+F_1-F_{NB}\right)={\mu }_2\left(G_B+G_A+Fcos\theta -F_{NB}\right)$$ (3)

The corresponding f_Bmax depends on the maximum of the sliding friction f_BH. As the sand-laying process continues, the printed part B gradually increases and the gravity G_B also continuously increases. Accordingly, the sliding friction f_BH and thus f_Bmax continue to increase until the static friction f_B is not sufficient for pushing the printed part. At this time, parts A and B are not in relative motion, and the formation of pushing dislocation ends.

If the adhesive force is sufficiently large for making part B move as a whole, complete pushing dislocation will form between the printed layer and the printing layer, as shown in Figure 3e. If the adhesive force is not sufficiently large, under the joint action of f_B and f_BH, part B will elongate or even fracture during the sand-laying process, as shown in Figure 3f.

Through analysis, it was found that the formation of pushing dislocation occurs because the static friction f_B of the printing layer is greater than the maximum static friction f_Bmax, which makes the printed part tilt and offset from the preset position. With continuous printing, the layer-by-layer sand causes the gradual increase of the deformation of the sand mold (core). The formation of the pushing dislocation will stop until the force of a certain layer is not sufficient for making the printed part move. The final pushing dislocation is shown in Figure 1, and has great impact on the quality of the castings.

Experiments

It is concluded that by reducing the printing layer gravity G_A and sand scraper pressure F, or by increasing the supporting force F_NA, the sand scraper angle θ and the maximum friction f_Bmax can achieve the effect of reducing the occurrence probability of pushing dislocation.

For the particular 3DP and raw materials, the influence of f_Bmax and θ on the pushing dislocation can be ignored. Owing to the printing layer gravity G_A, the sand scraper pressure F and supporting force F_NA are related to the recoater speed, ceramsite sand layer thickness, and bottom supporting condition. The influence of these three parameters on the pushing dislocation was investigated through experiments, respectively.

The mesh number of the ceramsite sand used in the experiments was 70/100, and the main composition was Al₂O₃. Furan resin and P-hydroxybenzenesulfonic acid were used as the binder and activator, respectively. Figure 4 shows the scanning electron image of the ceramsite sand, which was observed using scanning electron microscopy. Compared with silica sand, ceramsite sand has better roundness, larger fluidity, and smaller expansion rate, and can thus effectively reduce the casting defects caused by the high-temperature expansion of the molding materials. However, the surface of the ceramsite sand is rough and there are obvious missing particles/vacant.

FIG. 4.

Scanning electron image of ceramsite sand: good surface roundness with poor roughness and many obvious missing particles/vacant.

The position arrangement of the samples is shown in Figure 5. The samples were prepared using the EXOne S-MAX printer. The moving direction of the printhead is along the x-axis, and the recoater moving direction is along the y-axis. The samples at position 5 were selected for the experiments, and the dimensions of the designed samples were 22.4 mm × 22.4 mm × 174 mm.

FIG. 5.

Position arrangement of samples in 3DP: the x-axis is the moving direction of the print head; the y-axis is the moving direction of the recoater. The No. 5 samples were used in the experiment, including X-direction samples and Y-direction samples.

Because the pushing dislocation has different effects on the sample dimensions, the error caused by the sample dimensions affects the accuracy of the strength measurement. Therefore, the sample dimensions were directly used to characterize the effect of different printing parameters on the pushing dislocation. Moreover, the gas evolution and ignition loss were used to indirectly reflect the sample strength in the sense that their dependence on the binder content is similar to the dependence of the strength on the binder. The sample was scanned by a 3D imaging scanner, and the dimension difference between the actual sample and designed sample can be got. According to the GB/T 2684 standard of China, the gas evolution was measured by the sample ignition at 1000°C, and the ignition loss was expressed as the percentage of mass loss of the sample before and after ignition. In the experiments, the mean value $\overline{x}$ of three measurements was considered the effective result, and the standard deviation s was used to characterize the dispersion degree of three measurements.

Results and Discussions

Influence of recoater speed

The dimensional mean value and standard deviation of the samples, which are printed at the recoater speed of 140 mm/s to 220 mm/s, are shown in Figure 6.

FIG. 6.

Mean value $\overline{x}$ and standard deviation s of sample dimensions under different recoater speeds. (a) x-Axis direction sample with recoater speed of 140 mm/s. (b) y-Axis direction sample with recoater speed of 140 mm/s. (c) x-Axis direction sample with recoater speed of 180 mm/s. (d) y-Axis direction sample with recoater speed of 180 mm/s. (e) x-Axis direction sample with recoater speed of 220 mm/s. (f) y-Axis direction sample with recoater speed of 220 mm/s.

As can be seen, as the recoater speed increased, the dimension difference of the sample gradually decreased, the difference of the X-direction sample decreased from the maximum value of 1.9094 mm to the minimum value of 0.0723 mm, and the difference of the Y-direction sample decreased from the maximum value of 0.6828 mm to the minimum value of 0.0756 mm, which indicates that the pushing dislocation gradually decreased. This can be explained as follows. At the beginning of sand-laying, the compactness and smoothness were poor due to the thin sand layer, and thus, the bottom ceramsite sand did not have sufficient support for the forming part during the printing process. Therefore, one side of the forming part that contacted the recoater sagged to the bottom sand, which resulted in the inclination of the forming part. With the increase of the recoater speed, the sand layer became compact and smooth; therefore, the sagging degree significantly decreased. Hence, the dimension difference of the sample gradually decreased. Compared with silica sand, the higher roundness of the ceramsite sand leads to less friction between the printing layer and the printed layer. The sand flow flux is constant in unit time. The increasing recoater speed is equivalent to reducing the time of the recoater at the same position, which in turn reduces the amount of sand accumulated in front of the sand scraper. Therefore, not only was the friction of the printed layer reduced, but the pressure of the sand scraper also decreased to reduce the pushing dislocation.

Reducing the amount of the accumulation sand in front of the sand scraper can decrease the formation of the pushing dislocation. In the practical printing process, the recoater speed can be appropriately increased to inhibit the pushing dislocation, and the speed of 220 mm/s was selected in this experiment.

In addition, Figure 6 shows that the pushing dislocation of the X-direction sample was more severe compared with that of the Y-direction sample. When the recoater speed was 140 mm/s, the maximum difference of the X-direction sample was 1.9094 mm, while that of the Y-direction sample was 0.6828 mm. This is mainly attributed to the fact of the axis of the X-direction sample being perpendicular to the recoater and the axis of the Y-direction sample being parallel to the recoater. Therefore, the pushing dislocation could more easily form for the X-direction sample and its accumulation was larger.

The experimental results of gas evolution and ignition loss under different recoater speeds are presented in Table 1.

Table 1.

Experimental Results of Gas Evolution and Ignition Loss Under Different Recoater Speeds

No.	Printing parameters				Gas evolution (mL/g)				Ignition loss (%)
	Activator (%)	Resolution X (mm)	Layer thickness (mm)	Recoater speed (mm/s)	x1	x2	x3	$\overline{x}\pm s$	x1	x2	x3	$\overline{x}\pm s$
1	0.24	0.1	0.24	140	7.23	7.61	7.37	7.40 ± 0.19	1.51	1.68	1.71	1.63 ± 0.11
2	0.24	0.1	0.24	180	8.52	8.69	8.81	8.67 ± 0.15	1.63	1.78	1.61	1.67 ± 0.09
3	0.24	0.1	0.24	220	10.68	10.18	10.35	10.40 ± 0.25	2.01	2.10	1.87	1.99 ± 0.12

Table 1 shows that with the increase of the recoater speed, the gas evolution and ignition loss gradually increased; the gas evolution increased from 7.4 to 10.4 mL/g and the ignition loss increased from 1.63% to 1.99%. The reason for this is described below. The amount of single-layer sand is inversely proportional to the recoater speed. As the recoater speed increased, the single-layer sand amount decreased and the number of sand layers for printing the same sample increased. However, the binder amount for the single-layer sand was fixed. Therefore, the binder percentage in the sample increased with the recoater speed, which resulted in the increase of gas generation, ignition loss, and sample strength, and the decrease of pushing dislocation.

Influence of layer thickness

The dimensional mean value and standard deviation of the samples, which are printed at the layer thickness of 0.26–0.30 mm, are shown in Figure 7.

FIG. 7.

Mean value $\overline{x}$ and standard deviation s of sample dimensions under different layer thicknesses. (a) x-Axis direction sample with layer thickness of 0.26 mm. (b) y-Axis direction sample with layer thickness of 0.26 mm. (c) x-Axis direction sample with layer thickness of 0.28 mm. (d) y-Axis direction sample with layer thickness of 0.28 mm. (e) x-Axis direction sample with layer thickness of 0.30 mm. (f) y-Axis direction sample with layer thickness of 0.30 mm.

As can be clearly seen, the layer thickness affected the degree of the pushing dislocation for the printed samples in both directions, and the dimensional difference decreased in both directions with the increase of the layer thickness. The difference of the X-direction sample decreased from a maximum of 0.76959 mm to a minimum of 0.0179 mm, and the difference of the Y-direction sample changed from −0.8018 to 0.0229 mm. Although the maximum difference in the Y-direction sample was larger than that in the X-direction sample, the X-direction sample was more significantly influenced by the layer thickness compared with the Y-direction sample, as can be seen from the difference distribution of the integral sample. In other words, the pushing dislocation in the X-direction sample was more severe compared with that in the Y-direction sample. Both samples with the largest layer thickness of 0.30 mm had the smallest difference, and thus, the lowest pushing dislocation.

The sand diameter of the ceramsite sand was mainly distributed between 0.15 and 0.21 mm, corresponding to a 70/100 mesh. When the layer thickness increased, it was approximately equal to two times the sand diameter, which is equivalent to adding the rolling friction between the ceramsite sand at the front of the sand scraper and the surface of the printed part. The increase of the layer thickness changed the force situation in the printing process, and greatly reduced the force exerted on the printing part, which resulted in the decrease of the pushing dislocation. Compared with Figure 6, the decrease of the pushing dislocation caused by the increase of the layer thickness was more significant than that caused by the increase of the recoater speed. Therefore, the increase of the layer thickness is an effective approach for reducing the pushing dislocation.

The effect of the layer thickness on the gas evolution and ignition loss is presented in Table 2.

Table 2.

Experimental Results of Gas Evolution and Ignition Loss Under Different Layer Thicknesses

No.	Printing parameters				Gas evolution (mL/g)				Ignition loss (%)
	Activator (%)	Resolution X (mm)	Recoater speed (mm/s)	Layer thickness (mm)	x1	x2	x3	$\overline{x}\pm s$	x1	x2	x3	$\overline{x}\pm s$
1	0.24	0.1	160	0.26	8.22	8.02	7.98	8.07 ± 0.13	1.88	1.96	1.94	1.93 ± 0.04
2	0.24	0.1	160	0.28	8.05	7.87	8.16	8.03 ± 0.15	1.80	1.89	1.85	1.84 ± 0.05
3	0.24	0.1	160	0.30	7.58	7.80	7.64	7.67 ± 0.11	1.72	1.82	1.76	1.77 ± 0.05

As the layer thickness increased, the gas evolution and ignition loss gradually decreased. The 3DP sample at a layer thickness of 0.26 mm had the highest gas evolution and ignition loss. When the same sample was printed, the number of printed layers decreased as the layer thickness increased. Because the binder amount was the same for each layer, the binder content in the samples decreased with the number of the printed layers. Hence, the gas evolution and ignition loss also decreased as the layer thickness increased.

Influence of bottom supporting condition

A bottom supporting plate was set to provide the support for the printed part. To ensure the consistency of the experimental condition, a 5-mm-thick supporting plate was printed at half of the bottom by laying sand and spraying resin, while a 5-mm-thick sand was set at the other half because the resin was not sprayed. The bottom supporting condition and sample arrangement are shown in Figure 8. The influence of the bottom supporting force on the pushing dislocation was investigated by comparing the 3D scanning difference of the samples with and without a supporting plate. The printing parameters are the X-resolution of 0.1 mm, activator of 0.24%, layer thickness of 0.24 mm, and a recoater speed of 160 mm/s. The experimental results of the influence of the bottom supporting condition on the pushing dislocation are shown in Figure 9.

FIG. 8.

The bottom supporting condition and sample arrangement for bottom support experiment in 3DP: a 5-mm-thick supporting plate was printed at the left half, while a 5-mm-thick sand was set at the right half without resin.

FIG. 9.

Mean value $\overline{x}$ and standard deviation s of sample dimensions with and without the bottom supporting plate. (a) X-direction sample without bottom support. (b) X-direction sample with bottom support. (c) Y-direction sample without bottom support. (d) Y-direction sample with bottom support.

As can be seen, the maximum difference of the X-direction and Y-direction samples is 2.6128 and 2.8977 mm without the bottom support, and 0.2402 and 0.6415 mm with the bottom support. The significant decrease of the difference caused by the addition of the bottom support indicates that the pushing dislocation of the samples with the bottom support was less than that without the bottom support. However, the pushing dislocation was not completely eliminated, particularly for the situation shown in Figure 9d, where it can be seen that the difference of ∼0.6 mm was evenly distributed on the sample.

Although the bottom supporting force reduced the tilt of the printed part, pushing dislocation still existed. The reason for this may be that the bottom supporting plate could have induced stress on both sides of the printed layer and thereby caused extrusion deformation. Hence, the effect of applying a bottom supporting plate to decrease the pushing dislocation is insignificant.

Conclusions

The following conclusions were drawn from this study:

• 1.The printed part will deviate from the preset location due to the force of printing layer and bottom prelaying layer, and the continuous superimposition and stack of the printed part offset result in pushing dislocation.
• 2.The pushing dislocation can be reduced by decreasing the printing layer gravity and the sand scraper pressure, or by increasing the supporting force, sand scraper angle, and maximum friction.
• 3.With the increase of the recoater speed, the dimension difference of samples continuously decreased and the pushing dislocation gradually decreased. The pushing dislocation of the X-direction sample was more severe compared with that of the Y-direction sample.
• 4.The increasing layer thickness reduced the pushing dislocation, which is considered to be an effective approach for reducing the pushing dislocation. By adding the bottom support, the pushing dislocation can be reduced but is not completely eliminated. The effect of the bottom supporting plate was insignificant.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported by the National Natural Science Foundation of China (Grant No. 51975165).