4.1. Mechanical Tests and Materials Properties
Mechanics of materials is dealing with the behave of materials which subject to stresses, the mechanics of materials can be investigated in the fresh state or hardened state. Further, the mechanical strength test in the presented paper was conducted to measure the fresh state slurry of the mortar and hardened state of the mortar.
There are two major factors that have a significant effect on the final shape result of the product are (a) the number of printed layers, and (b) the discharging of mortar slurry through different nozzle shapes and sizes.
Figure 6 explains the change in width, which is specified as (
W + ∆
W), while the thickness is expressed as (
T − ∆
T).
Figure 7 shows that the extruded layers have changed into their original shapes. The first printed layer would also change in width and thickness after loading the next layers. Once again, the first layer will face a change in its shape after the printing of the third layer. This change is possibly continuous until the shape has reached stability in its form and has set enough. The last layer of the component did not encounter any modifications and changes due to no further layers being added, so it retained its own shape,
Figure 7. This is the nozzle, Ø50 mm, that has been used for the printed specimens in
Figure 7.
Where
W is the true width and ∆
W is the trace width error.
T is the true thickness and ∆
T is the trace thickness error. Therefore, the area of the object will change at the pre-compression and post-compression stages according to the rheology of the mix proportions and the forced impact by the next layers which are printed.
Figure 6 shows that the area of the layers will vary according to the printing height, nozzle types, the mixing time, and the setting time of the materials. Therefore, the true area of each cross-section printed layer is equal to (
W + ∆
W) × (
T − ∆
T).
However, there could be a different result when the time intervals between layers changes. When a slight decrease in the time intervals happens between layers the rate of penetration between layers increases due to the viscosity and thixotropy (shear thickening) properties in the concrete. This is consistent with research by [
35], which found that the shear thinning of the concrete could be changed to shear thickening by adding superplasticizer to the paste of the cementitious materials. Shear thickening is defined as the proportion of the shear stresses to the viscosity of materials which can be increased gradually. This phenomenon emerged during the pumping process of the mortar [
36]. The mortar had resisted downward pressure and the viscosity also greatly increased. For this to occur, a mixer in the hopper needs to make a consistent movement in the container. In this study, a different ratio of superplasticizer was used. This had a significant influence on the setting time of the mortar and the viscosity of the mortar. The ratio of superplasticizer to cement materials was (0.67% to 1%).
The shape of the nozzles influences the printability, shape, and flowability of the slurry. The study, Li et al. [
37] asserted that intercepting shocks are significantly changed according to the shape of the nozzles. The study found that square shaped orifices are faced with a higher interception than the other jets due to the four corners exiting at the nozzle. In addition, the penetration between two layers increases while the
w/c and the number of layers increment proportionally. Moreover, the shape of the nozzles also affects the percentage of penetration between the two layers. If the nozzle shape is circular, the penetration rate increased slightly. This is due to less flatness of the previous layer which is laid down as a concave shape. The application of the spherical particles and square particles theory could be applied to the circle and square nozzles in terms of the shape and load applications.
Figure 8 shows that the load in the square nozzles is distributed equally due to the radius of distribution in the square shapes. This distribution area is smaller in circular nozzles [
38]. Böhmer et al. [
39] found that in the inkjet printing technique the diameter of the droplet, which contains 0.3% polyvinyl alcohol solutions, would be larger than the diameter of the nozzle. They used a different concentration of polymer solution with the three different nozzle diameters. As a result, at a constant polymer concentration, smaller initial droplets were produced by the smallest nozzle diameter that, in turn, leads to smaller particles as well. Consequently, this could be similar to concrete slurry, where a higher flow of slurry is produced with increasing nozzle sizes.
Cwalina, Harrison and Wagner [
38] stated that the particles with spherical and cubic shapes produce different results. The squeeze flow and load distribution between two cubic particles and two spherical particles with equivalent radii-lengths are illustrated in
Figure 8.
For particles with an identical characteristic half-width,
R, moving along their lines of the centre at a relative velocity,
V, in a Newtonian fluid of viscosity,
ηf, the lubrication force between the spherical and cubic particles is given, respectively, as [
38]:
In the two Equations (2) and (3), it is obvious that the reaction force between two particles increases when the shapes change from cubes to spheres. Therefore, this result will be similar for printing when square or circular nozzles are used. The printed shape will replicate the shape of the nozzles. The particles used in this experimental test were mostly spherical with some of the irregular shapes. As a result, the printed slurry will be a similar shape as it passes through the nozzle. The different shapes and sizes of the nozzles were also investigated in this study. The shapes used were circular and rectangular, with sizes of (20) mm in diameter and (35 × 10) mm. The forces were distributed evenly over the greater surface area in the square and rectangular shapes than in circular shapes. Consequently, the printed layers of the square or rectangular nozzle shapes withstand more layers than the circular nozzle shapes. It should be noted that the same mix ratio was used for the printed object utilizing different nozzle types. It was found that the nominal width in a rectangular shape was larger by 2 ± 0.85 mm than its reduced width (layer surface contact). Conversely, the nominal width of a circular shape was larger by 3.1 ± 0.75 mm than its reduced width.
Considering the forces applying to differently shaped particles, the higher forces emerged between flat cubic particle surfaces compared with the curved surfaces characteristic of spherical particles.
Figure 9 shows an object where one layer has six layers printed onto it. It has been printed to measure its dimensional geometry and test its mechanical behaviour.
For a printed object of over 120 mm (more than 7 layers), the oscillation at the arm of the robot increased in the end-tip of the arm which is most related to joint 4 and joint 6 in the robot [
28].
Figure 10 shows how the printed layers collapsed after 10 layers of printing.
Another challenge that it faced during the printing process was the use of a flat-based hopper, as shown in
Figure 11, where core-flow (rat-holing) occurred during the printing of the specimens. It is observed that some of the slurry close to the wall of the bucket is in a static state, while other parts of the slurry are in a mobile state.
Figure 12 shows a modified hopper angle of Ø45° to improve the flowability in the hopper. Generally, fresh concrete or mortar during poured-in-place behaves as a liquid slurry (a viscoplastic fluid with high yield stress). However, the internal structure of slow casting concrete or when in a rest state leads it to flocculate. It also has the ability to resist the load from concrete cast over it without increasing lateral stress, despite the nature of the mould. Feys et al. [
41] explained that the (hydro-) clusters are assembled together and become moulded from certain shear stress on the critical shear stress. By increasing the shear rate, the viscosity of concrete increases proportionally. This state of fresh concrete is called shear thickening. For the concrete properties, it is noticeable that when the temperature rises, the workability and slump of the concrete decreases. This is another reason that the mortar could not pass through the hopper effortlessly. Apparently, the longer concrete or mortar remains in the hopper, the more advanced the reaction and the higher the increase in temperature, which subsequently leads to an increase in the viscosity of the mortar. A temperature increase from 21–35 °C was measured after 30 min of the mixing process occurring.
Earlier studies explained the two types of powder flow pattern in the hopper: Mass-flow and core-flow [
40]. The most noteworthy is the core-flow that emerges while feeding the concrete through the pump (deliver) to the robot. According to Fitzpatrick et al. [
42], the moisture content has a high impact on powder particles in terms of flowability. Consequently, the surface forces between the powder particles or slurry and the wall surface play a major role in shaping the nature of the powder flow (see
Figure 11).
Figure 12 shows a hopper with an angle of Ø45°, which reduced rat-holing but did not eliminate it completely. This rat-holing phenomenon happens due to the flocculates of the particles and maintains particles in the static state. Improving this situation requires consistent mixing in the hopper.
For most of the trials that were prepared in the experimental program, a set of prisms and cubes were prepared to validate the mechanical behaviour of the mixing property. The printed prisms have been created by selected trial 5 cementitious mortar as a mix of proportions. So, the printed mortar has been tested for each of the layers from (1–4) by caulking gun.
The outcomes suggest that waiting time between layers of 10 min is necessary, as it has been proofed by [
43] and the ratios of water to admixture need further investigation.
The printed part is shown in
Figure 13. Shrinkage cracks appeared on the printed part after one day of curing in the laboratory temperature. To reduce the cracks so the printed materials are stronger and exhibit fewer shrinkage cracks, chopped strand fibre was introduced.
4.1.1. Slump Test
This test showed the different results of the three different trial mixes which have three various slump results. This difference in results was based on their
w/c mix ratio and the cohesiveness of their particles. Consequently, all the trial results have a different slump ratio. The deformation of each slump was 8.5 mm, 8.8 mm, and 12 mm for trials 5, 8 and 12, respectively. It is noteworthy that trial 8 has a greater slump than trial 5, with the difference likely due to the amount of
w/c and superplasticizer in this mortar mix. It is also interesting that the mixing time and the time taken to pour into the slump have a great influence on the resulting mix. Each trial was, therefore, mixed for 5 min. As a result, trial 5 was expected to achieve higher penetration than trial 8, but trial 5 was more coherent than trial 8, as explained in the squeeze flow test, see
Figure 14.
Figure 15 shows the spread-flow test for trial 5 both in the presence and absence of 1% PP fibre.
Table 4 shows the results of the relative slump for all main mixes which were used for the printed samples. The minimum flow and slump occur in trial number 5 with 1% polypropylene fibre, which was expected due to the consistency and cohesiveness of this mix and the fibre content. The use of polypropylene fibre increased the mechanical strength of the structural element (
Section 4.1) and it reduced flowability (
Figure 15) and shrinkage (
Figure 13). In addition, fibre increases the buildability of the mortar and stiffens the mass of the printed layer. For that reason, it is recommended that fibres be used in the printed specimens to increase stiffness and mitigate shrinkage in the printed part. Each test has been repeated three times.
4.1.2. Squeeze Flow Test
The results of the squeeze flow tests showed different values in the single, double and triple layers for each of the selected trial mixtures (5,8,12). Each trial has been conducted 3 times. The maximum result of the three trials is shown in
Figure 16. In the single layer mortar mix test, higher results were obtained in the reaction force value until it reached the required displacement. For instance, at a displacement of 2.99 mm, the required loads in trial 12 (concrete mix with small aggregate) were approximately 956.51 N. While at the same displacement for trials 5 and 8 (cement mortar), the load was approximately 277.82 N and 153.11 N, respectively. An examination of the results of the tests that utilised a double layer found significant differences in comparison to the single layer results. Trial 5 (cement mortar) had a reaction force which reached approximately 622.54 N when the displacement was about 3.99 mm, while for trials 12 and 8 the loads were 536.05 N and 275.75 N, respectively. At the same displacement, the results for the triple layers exhibited a similar pattern.
These results show that the mortar is more coherent than the concrete mixtures in the fresh state. Thus, it does not allow trapped air bubbles to remain in the mortar mixture. This is consistent with the study by Surendra et al. [
44], who discovered that the mortar in the first 24 h had a greater compressive strength by comparison with normal concrete. Hence, a much higher percentage of open pores will exist in the concrete mixtures by comparison with the cementitious mortar due to the presence of aggregates. The larger particle sizes lead to a higher porosity within the concrete mixtures.
The reaction force is dependent on the number of chains (layers) and the force between the particles. Trial 12, a single layer test, showed that the presence of the small aggregate in the mixture can resist more force over the given displacement,
Figure 16. Furthermore,
Figure 16 shows that the mortar mix for more than one layer has better resistance than the concrete mix and that less penetration will occur between layers when the loads are applied
Figure 16. It can be seen that for both double and triple layers, mortar has a better squeeze-flow performance.
4.1.3. Setting Time
In
Table 5, the setting time results for the different trial mixes is shown. There are minimal differences in the initial setting times of these trials.
The setting time has a crucial effect on the bond between printed layers and penetration rate between each printed layer.
The buildability tests, which depend on the setting time, have been applied for each trial by printing using an extrusion caulking gun.
Table 6 relates to trial 5; this trial was based on the original mix design, which used a 1:1 cement to fine sand ratio with a water/cement ratio of 0.33. This is able to be printed with the caulking gun and a circular nozzle of 14 mm diameter. The settling observed was, to some extent, expected with the relatively similar levels of sand and cement in the mix. Not much change was observed in the heights of the sample. Sample number 6 showed a failed extrusion as it had collapsed considerably.
4.1.4. Compressive Strength Test
For each of the optimum trials that were printed successfully, a uniaxial compressive strength test was conducted.
Figure 17 presents the results of the test conducted for the manual concrete mix, which included polypropylene fibres with a different ratio and with a control sample.
The compressive strength result for the 7 layers of the printed mortar is shown in
Figure 18. It can be observed that the highest compression result for printed mortar without curing and after 28 days was 13.45 kN as a maximum load. The printed specimens were left in the control temperature room at 20 ± 2 °C without any additional post-curing. However, the compressive strength increased after using 1% of Polypropylene fibre in the mortar. The maximum strength was 17.65 kN after using fibre. Therefore, the increase in the percentage rate was 31%.
Figure 18 shows the trend of the printed hollow column (7 layers), it is clear from how the trend line (i.e., force versus displacement) varies and has an unstable trend line in both plots. This is more obvious in the printed specimens while using 1% PP fibre. The PP fibre significantly stiffens the structure during the early applied load on the hollow column. Therefore, a bump at the early stage of the printed sample can be observed.
Figure 19 shows the printed sample after drying at the control temperature in the lab. No shrinkage cracks or hairline cracks on the printed specimens resulted. After loading, cracks appeared on the edge of the printed column, as shown in
Figure 19b. The results for the printed column showed that the average rate of the printed column is (12.83 ± 0.54 kN). This is equal to the strength of 2.37 MPa, however, this needs to be improved by post-curing and the use of large particle sizes. It is also obvious that the low strength was due to the number of layers. The printed column was a mortar mix rather than a concrete mix which normally has less resistance than normal concrete after curing for 28 days.
The short column under the uniaxial load cracked and ruptured at the edge of the sample, which showed the weakest part of the column. The edge of the column was revealed as the weakest part of the column due to the irregular movement of the robot while printing the column.
4.1.5. Flexural Strength Test
The three-point bending test was applied for the optimum mix of mortar using a different type of fibre and different ratios. The flexural strength results for the different ratios are shown in
Figure 20.
The three-point bending test and flexural strength for the one, two, three and four layers of cementitious mortar with 1%PP fibre and without fibre are shown in
Figure 21. The maximum result of 5.78 MPa for 28 days curing was observed in the single layer with 1% PP fibre content in the sample. Another high result was achieved for three printed layers which recorded 5.65 MPa. It is worth noting that using PP fibre increased the flexural strength in all variable layers. In addition, it is noticeable that when increasing the number of layers, the flexural strength decreases due to inconsistency, a high ratio of moisture content, and air trapped between layers. The flexural strength of the triple layers with 1% PP fibre is higher than the double layer printed specimens, this may be due to the inconsistency of the printed layers which resulted in higher flexural strength than a double layer. The reason for the flexural strength results being different among layers still requires further investigation. The moisture content on the printed surface layer, surface roughness, and orientational angle of bedded layers have a major contribution in the fluctuation of the results. Overall, the results show a decline in flexural strength as the number of layers in the printed specimen increases.
Figure 22 shows the effect of different nozzles on the result of flexural strength. It is noted that the highest flexural strength is found in the 3rd printed layer of the rectangular nozzle when investigating the highest flexural strength in circular and rectangular nozzles in the wet medium cure. In
Figure 22, all the layers of the rectangular nozzle have a similar result, but for the circular nozzle, each layer has unstable results. Therefore, it shows that the load distribution of the circular nozzle is reduced with the reduced surface area and width. This directly affected the mechanical strength results of the printed object (see
Figure 8). Consequently, the result suggests that a rectangular or square shape has a constant result and a better result than a circular nozzle print. This is consistent with the study of Reference [
22]. Conversely, it has been shown that wet medium curing is better than curing at air-temperature in vitro, as demonstrated by Reference [
45],
Figure 22. In addition, it is highly recommended to use fly ash in the mix to reduce the voids between particles and increase the durability of the mortar [
46].
The caulking gun and cavity pump show differences in the printed specimen results due to differences in the way fresh state material is deposited, e.g., internal pressure, the height of material’s deposition, and the delivery system. First, the pressure in the caulking gun is generated by hand (manual), while in the cavity pump is more uniform and regulated since it is controlled automatically by the control panel. Second, the height of the material’s deposition is another difference in both processes. In the caulking gun, it is hard to control the height distance between the platform and the nozzle, but this process is controlled automatically through the MATLAB software to the end-effector of the robotic arm. The last point, the delivery system in the caulking gun, directly deposits materials from the tube (there is no delivery hose), while in cavity pump the deposition materials pass through a pump and hose before reaching the nozzle.