Consider a practical rammed earth wall with the dimension of 4 meters in length, 3 meters in height, and 24 centimeters in thickness. The ultimate bearing capacity of the wall is formulated as

where F is the ultimate bearing capacity of the wall, σ is the ultimate compressive stress, W is the width of the wall, and t the thickness of the wall. Then the similarity principle in Equation 1 is formulated as

where

Variables with the subscript of “m” refer to the scaled model, while “p” represents the practical wall model. Substitute Equation 3 into Equation 2, the similarity relationship S_F is obtained, which shows that the ultimate bearing capacity of the practical wall is two times that of the scaled model. It is noted that the boundary of the model wall is constrained by the concrete frame, and thus the similarity relationship in Equation 2 is still an approximation. However, the error caused by the similarity calculation is acceptable in engineering application.

Three pieces of rammed earth walls were fabricated. The structural configuration of each wall model including component material proportion and structural dimension etc. is listed in Table 1. Three reinforced concrete frames were precasted to simulate the constraint of each wall boundaries. The general configuration of the experimental device of a rammed earth wall model is shown in Figure 1.

The original three wall models (W1, W2 and W3) were loaded with the uniformly-distributed vertical loading in steps, and their ultimate bearing capacities were obtained. The results are shown in Table 2. After the first round of loading, cement mortar-steel fiber reinforcement was performed on the rammed earth walls. Steel webs of Φ4 in diameter with 300 mm-interval in both horizontal and vertical direction were fixed on the surface of each wall model. Tie bars went through the wall with an interval of 600 mm. Mortar of M10 (Chinese Masonry Design Code GB50003) was laid homogeneously on both sides of each wall with a thickness of 30 mm. The wall models W1, W2 and W3 after reinforcement, are labeled as W1’, W2’, and W3’, respectively.

Vertical load generated by a hydraulic jack was delivered on top of the wall model through two stiffening-ribbed H-shaped steel beams. A force transducer was put between the jack and the upper H-steel. Fine sand was spread on the top of the wall homogenously to apply uniform vertical load approximately. Numerical simulations of the rammed earth wall models were carried out before the experiment to simulate the structural properties of the wall subjected to vertical loading. However, it is still difficult to estimate the ultimate bearing capacity of the wall models accurately due to various influencing factors as material discreteness. In consideration of the reason above, load from 5 to 20 kN in each step was applied gradually according to the structural response of the wall models. Each load step lasts for some time until the meter readings no longer change. Dial gauges were distributed at certain positions based on numerical simulation results to measure the strains in the middle and two sides of the wall model, as shown in Figure 2(a,b). The one perpendicular to the wall model (B8) was used to observe the out-of-plane displacement. Local strain of the wall was measured by strain rosettes which were distributed as shown in Figure 2(c).

It is noted that a rammed earth wall shows a behavior somewhat different to brick masonry. Brick masonry generally fails at much lower loads than brick compressive strengths due to interaction between the bricks and mortar. The absence of mortar joints makes the failure of rammed earth more of a crushing failure than a tensile failure. Therefore, a rammed earth wall may not give an adequate warning in the form of cracks. Hence, it is recommended to use an adequate factor of safety against failure in practical rammed wall structural application.

In this study of the three wall models, the W1 wall model suddenly cracked in the upper boundary when jack force reached 30 kN. The crack grew rapidly both in width and length, while a large part of the rammed earth on the corner was to drop down (Figure 3(a)). The W3 model, which contained more lime than W1, had swellings partially on the surface when the loading force reached 50 kN. Then cracks appeared on both sides of the upper corner. Applying load continually till 73 kN, the deformation increased rapidly, and then the wall model crashed at the loading of 90 kN, and the boundary of W3 separated from the RC frame (Figure 3(b)). The failure characteristics of W2 were almost the same as W3. After the first round of loading, the three wall models were reinforced using the aforesaid CMSF method, and similar experimental results were observed in the second round of loading, It is noted that the typical failure characteristics of the reinforced wall models was the separation of the mortar layers to the original rammed earth wall models (Figure 3(c)), and all the three models showed the cohesive failure of the rammed earth wall, while little failure was observed on the mortar layer.

Cracking load and ultimate bearing capacity of each wall model is shown in Table 2. From Table 2, it is noted that the method of cement mortar-steel fiber reinforcement enhances the strength of W1 and W2 greatly, while the enhancement of the wall W3 is not obvious, which is probably because the original ultimate bearing capacity of W3 is much greater than those of W1 and W2 since more lime is contained in W3. The strength of W1’ even reached about four times that of W1, and the ultimate bearing capacities of W2’ and W3’ are both greater than those of W2 and W3, respectively.

Wall model W3 was taken as a typical example to investigate the deformation of the wall models. A loading-deformation curve (Figure 4(a)) was obtained based on the experimental results, which shows that the deformation increases with the increasing of the vertical force. From Figure 4(a), it is noted that in the plastic stage, the deformation of wall model W3 increases much even though loading increases little. When the load reached the peak point, it was gradually unloaded to zero. In this process, part of the elastic deformation recovered. However, loading track could not come back to the origin which obviously showed the elastoplasticity of the rammed earth wall.

A load-strain curve of W2 is shown in Figure 4(b) to investigate the plane strain of the wall models. All the strains increase with the loading. Different parts of the wall have different strains which indicate that the strain of the wall has become a 2-dimensional stress problem under the special boundary of RC frame. Strain 5 on the lower corner is greater than strains in other positions which indicate that arching above the window of W2 enhances the compressive strength of the whole wall by distributing the stress to the boundary on the two sides. Strain 7 and Strain 8 in Figure 4(b) illustrate that the rammed earth both above the window and below the window is in compressive status.