Earth buildings are common types of structures in most rural areas in all developing countries. Catastrophic failure and destruction of these structures under seismic loads always result in loss of human lives and economic losses. Wall is an important load-bearing component of raw soil buildings. In this paper, a novel approach is proposed to improve the strength and ductility of adobe walls. Three types of analyses, material properties, mechanical properties, and dynamic properties, are carried out for the seismic performance assessment of the adobe walls. These performed studies include that, material properties of the earth cylinder block, mechanical properties of adobe walls under quasi-static loads, and dynamic performance of adobe walls excited by seismic waves. On investigation of material properties, eighteen cylindrical specimens with a diameter of 100 mm and a height of 110 mm were divided into three groups for compressive, tensile, and split pull strength tests, respectively. The results of the three groups of tests showed that the yield strength ratios of compressive, tensile, and shear strength were about 1:0.3:0.2. In order to study the performance of structural components, three 1/3 scale model raw soil walls with a dimension of 1,200 mm in width, 1,000 mm in height, and 310 mm in thickness were tested under cyclic loading. The average wall capacity of the wall obtained by the test was about 13.5 kN and the average displacement angle was about 1/135. The numerical simulation experiment is used to explore the mechanism of structural failure. A three-dimensional finite element model is established by choosing the material parameters based on the above test outcomes. The accuracy of the numerical simulation experiment is verified by simulation and comparison of the above quasi-static test results. Further, the collapse process of raw soil wall under a seismic wave is simulated for exploring the response and damage mechanism of structure. Based on those systematically analyzed, some useful suggested guidelines are provided for improving the seismic performance of raw soil buildings.

Soil is a low-cost raw construction material, which can be recycled after demolition and has excellent performance for thermal insulation. This material is still widely used as a construction material in developing countries [

Because the friction between the test specimen and the press constrains the transverse deformation of the specimen, the various parts of the test specimen are in different stress states. The “hoop effect” has a significant influence on the strength of the cubic specimen. In this paper, cylindrical specimens are innovatively fabricated to improve the compressive stress state of the specimens. The cylindrical specimens can be used to reduce the influence of the “hoop effect.” In addition, the specimens are sufficiently compacted in the process of fabrication. The compaction process can significantly reduce the dispersion of experimental results. The experiments are conducted following the standards for the soil test method [

As the significant load-bearing member of raw soil buildings, the adobe wall damage often leads to the overall damage of the structure [

In China, raw soil buildings are mainly distributed in the northwest of the country. Kashgar city and its surrounding areas in Xinjiang are more widely distributed. This area is deep inland, far from the sea, cold and dry in winter, with high temperature and rainless in summer. The survey of 244 rural dwellings has shown that adobe houses and adobe-wood houses account for a considerable proportion of rural dwellings of Kashgar city areas. The distribution of those houses is shown in

Structure type | Adobe house | Adobe-wood house | Brick masonry house |
---|---|---|---|

Quantity (building) | 40 | 118 | 86 |

Proportion | 16% | 49% | 35% |

However, the area is located at the edge of the interaction between the Eurasian and Indian plates, with unique structural sites and strong crustal movements, it makes a region of intense and frequent seismic activity. The earthquake disasters have shown that houses in the areas will suffer severe damage or even collapse. Therefore, the present study is focusing on the seismic performance of the raw soil buildings. The significance of this study ensures the safety of the residents living in the adobe masonry buildings and reduces the loss of property and human lives and improves the quality of life and promotes the stability of the local community.

The soil materials for the specimens were taken from the suburb of Kashgar City, Xinjiang; the main physical properties are shown in

Fine-grain content |
Medium sand content (%) | Coarse sand |
Optimum water content |
Maximum dry density (g/cm^{3}) |
Plasticity index |
Plastic Limit WP (%) | Liquid Limit WL (%) |
---|---|---|---|---|---|---|---|

52.9 | 16.3 | 30.8 | 19.6 | 1.92 | 13.66 | 16.44 | 30 |

Test category | Group | Number | Average density [g/cm^{3}] |
---|---|---|---|

Compressive test | A | 6 | 2.02 |

Tensile test | B | 6 | 2.01 |

A 300t microcomputer servo press is used in this research. For evaluating the compressive strength, the loading method is the standard test method for finding the mechanical properties of concrete. Before the formal experiment, in order to calibrate the compressor, the compressor was pre-stressed. The pre-loading load was 1 kN, which assured that the entire experimental apparatus, i.e., the plate exerting the compressive load, was in close and uniform contact with the specimen before starting the experiment. The experimental process adopts the gradual loading method. The applied load in each stage is 10% of the estimated failure load value, the average loading speed is 0.5 kN/min, and the subsequent stage load is applied after 1 min. After each pressurized specimen is loaded to 85% of the estimated damage load value, it is continuously loaded at the original loading rate until the stress value drops to 85% of the peak stress, then the experiment is ended.

σ represents the test strength, and σ_{max} represents the ultimate strength. The testing procedure and results can be observed. When σ < 0.55 σ_{max}, the increase of stress is approximately proportional, the load and displacement curves are linear elastic, and there are no visible cracks on the specimen’s surface. As the load continued to increase, small cracks at both ends of the specimen began to appear. These tiny cracks quickly propagate into the middle of the specimen, and it is a brittle failure process, so the crack development process is very fast, as shown in _{max}, the through cracks began to appear, the edge of the specimen is slumped, as shown in

Number | Ultimate stress [MPa] | Ultimate strain | Elastic Modulus [MPa] |
---|---|---|---|

A1 | 1.177 | 0.025 | 47.1 |

A2 | 0.811 | 0.024 | 34.0 |

A3 | 1.209 | 0.025 | 46.6 |

A4 | 1.055 | 0.031 | 34.9 |

A5 | 1.024 | 0.035 | 32.5 |

A6 | 1.057 | 0.023 | 37.8 |

The tensile test was conducted using the indirect tensile method, the split pull test was carried out, and the normal stress for the test results was processed by the splitting test formula, _{t} < 0.45 σ_{tmax}. The increase of stress was approximately proportional. When σ_{t} ≈ 0.45 σ_{tmax}, small cracks began to appear on the specimen. When σ_{t} ≈ (0.45~0.60) σ_{tmax}, the specimen was broken, as shown in

Number | Ultimate stress [MPa] | Ultimate strain | Elastic Modulus [MPa] |
---|---|---|---|

B1 | 0.335 | 0.028 | 10.3 |

B2 | 0.307 | 0.019 | 10.65 |

B3 | 0.381 | 0.026 | 12.3 |

B4 | 0.375 | 0.021 | 14.9 |

B5 | 0.371 | 0.027 | 11.4 |

B6 | 0.24 | 0.028 | 9.42 |

The results of the compressive and tensile tests are shown in

Ultimate compressive stress [MPa] | Mean square deviation | Mean [MPa] | Coefficient of variation | Ultimate tensile stress [MPa] | Mean square deviation | Mean [MPa] | Coefficient of variation | Tension to compression ratio |
---|---|---|---|---|---|---|---|---|

1.177 | 0.1408 | 1.055 | 0.133 | 0.335 | ||||

0.811 | 0.307 | |||||||

1.209 | 0.381 | |||||||

1.055 | 0.375 | 0.054 | 0.335 | 0.162 | 1/3 | |||

1.024 | 0.371 | |||||||

1.057 | 0.24 |

After compaction and curing, the dense cylinder specimen bear uniform force in the experiment, and the experimental results show a low coefficient of variation. We assume that the cylinder adobe specimen is an anisotropic material. Von Mises yield criterion [

For the tensile test, the values for

For the case of a pure shear state (

Compressive yield stress [MPa] | Tensile yield stress [MPa] | Shear yield stress [MPa] |
---|---|---|

1.055 | 0.335 | 0.193 |

Depending on the ratio of normal stress to shear stress, three types of damage may lead to the destruction of the wall: shear friction failure, shear compression failure, and oblique compression failure. Since the failure of the single-layer wall under the action of the earthquake is primarily shear failure, the shear stress yield stress 0.193 MPa is utilized as the yield point in the numerical analysis.

Three pieces of raw soil wall are produced. Its numbered are SQT1, SQT2, SQT3. The designed wall size is 3600 mm × 3000 mm × 310 mm (width × height × thickness). However, a 1/3 scale model is used in the actual test, and the scale model of adobe wall size is 1200 mm × 1000 mm × 310 mm (width × height × thickness), as shown in

A concrete beam is placed on the top of the adobe wall during the test. The horizontal load and a vertical load of the simulated seismic force are applied to the top concrete beam. The vertical load is calculated according to the actual upper load on the wall, and the value is 0.06 MPa. The test measurements included horizontal displacement of the top beam, the horizontal displacement of the top and bottom of the wall, horizontal displacement of the bottom beam, and corresponding load data.

where: 1—Test wall, 2—Concrete bottom beam, 3—MTS actuator, 4—Fixed support, 5—Jack, 6—Loading frame, 7—Concrete crest beam, 8—displacement meter 1, 9—displacement meter 2, 10—displacement meter 3, 11—displacement meter 4.

According to the “Building Seismic Test Method” (JGJ 101-96), the adobe wall is subjected to a cyclic load. The predicted lateral bearing capacity was 37.7 kN, and the predicted axial pressure is 22.6 kN. During the pre-loading phase, 30% of the predicted axial pressure is applied first, and 30% of the predicted shear capacity is applied for 2 min and repeated twice. The loading position has to be strictly centered on avoiding out-of-plane instability. The predicted pressure value of the vertical load and horizontal load is applied for the actual loading phase. The horizontal load is gradually loaded by displacement control in a sinusoidal pattern. The specification stipulated that each stage should not exceed 10% of the standard load. The highest frequency was 0.006 Hz, first push and then pull back. The peaks of the sine waves are 1 mm, 3 mm, 5 mm, 7 mm, 9 mm, and 11 mm, respectively. Each load is cycled twice and stayed for 50 s at the end of each cycle before the first crack occurred.

Ensure that the vertical load is applied and the measuring instrument is working normally. The first level horizontal load is applied, the surface of the adobe wall dose not change significantly. When the second-level horizontal load is applied, the surface of the adobe wall is showed slag destruction partially, and some of the vertical cracks are slightly propagated. Continue increasing the displacement, the number and the width of fine cracks are increased continuously. Most of the cracks are oblique cracks, and a few are vertical cracks. When visible through cracks began to appear, the test was stopped. The failure mode of the crack is shown in

Number | Cracking load [kN] | Cracking displacement [mm] | Peak load [kN] | Peak Displacement [mm] | Damage load [kN] | Destruction displacement [mm] | Displacement angle |
---|---|---|---|---|---|---|---|

STQ-1 | 7.9 | 2.9 | 13.1 | 4.0 | 11.1 | 5.8 | 1/106 |

STQ-2 | 8.8 | 1.9 | 13.6 | 4.1 | 11.5 | 5.2 | 1/105 |

STQ-3 | 9.3 | 1.7 | 13.8 | 4.1 | 11.7 | 6.1 | 1/107 |

Number | STQ-1 | STQ-2 | STQ-3 | Average value |
---|---|---|---|---|

Displacement angle | 1/195 | 1/105.6 | 1/106.5 | 1/135.6 |

The failure mechanism of adobe walls is similar to masonry structures. Based on the formula for calculating the bearing capacity of the masonry structure [

where

Due to the difficulty in ensuring the vertical mud can not contain any voids during the construction process, the influence coefficient κ is 0.6 in this test. The masonry shear strength design value

A numerical model developed by using the software program ANSYS/LS-DYNA is presented in this section. Similar to the experiments, the loading conditions of the simulation experiment include both static and dynamic loads. Furthermore, the numerical results are compared with the experimental results to verify the accuracy and the reliability of the numerical model [

LS-DYNA is a nonlinear finite element solver that mainly uses the explicit method. The reason for considering the explicit method is that the implicit method can only simulate the structural behavior before the collapse, but not during the collapse process since it is based on an iterative solution, and the global stiffness needs to be evaluated at every step. When the structure stiffness deteriorates or the force-deformation relationship shows a degradation behavior, the global stiffness matrix becomes a singular matrix, leading to non-convergence. However, the explicit method is based on dynamic equations, and thus, it has better stability. The inverse calculation of the stiffness matrix can be avoided by utilizing this method, which is the most challenging part. Hence, the simulation cost can be effectively reduced. As long as the time step is sufficiently small, the explicit method can simulate the dynamic behavior during the structure’s collapse. Presently, the explicit method is widely used in conducting shock and explosion research. The central difference method is used in the corresponding explicit dynamic solution [

There are three frequently used models—the integral model, decoupled model, and coupled model—each of which has respective advantages and disadvantages. The integral model, in which the reinforcements are equivalent calculated in the concrete structure, based on certain principles, and the elements are a single material. It is easy to carry out a fast calculation with low accuracy. The decoupled model treats the brick and mortar joint separately and considers the slip between them, which is more similar to the actual situation. Unfortunately, due to its high level of accuracy, the decoupled model requires high computational capacity due to many elements. Thus, it is time-consuming. The coupled model combines the contribution of the reinforcements of the element stiffness matrix with that of the concrete, assuming no slip at all between these two materials. This model is selected since the actual performance of a brick and mortar joint in raw soil building is similar to what is depicted in the integral model.

The element of the numerical model adopted Solid164 in ANSYS/LS_DYNA. The material properties selected are MAT_PLASTIC_KINEMATIC. The main parameters are based on the above soil material test, such as density, modulus of elasticity, Poisson’s ratio, yield stress, tangent modulus, limit strain, etc., as shown in

DENS [kg/m^{3}] |
EX [MPa] | NUXY | Yield stress [MPa] | Failure strain | Other values |
---|---|---|---|---|---|

2010 | 37.7 | 0.3 | 0.193 | 0.01 | System default |

The number 16891 in the right of

The value of the displacement angle in the actual test and the numerical results are shown in

Test type | Displacement angle |
---|---|

Actual test | 1/135.6 |

Numerical results | 1/212 |

The collapse of a wall under a seismic wave is a nonlinear system problem due to hysteretic degradation behavior. Assuming that the physical properties remain constant in a short period or in the deformation increment, an incremental equation can be established.

where

The incremental

For the seismic analysis, El Centro is selected. The precautionary seismic intensity of the structure is 8°, equivalent to the design peak acceleration of ground motion (PGA) of 0.32 g. To observe the wall behavior in different levels of earthquake intensity, the ground motions are increased with increments of PGA such that they are representative of the minor (0.1 g), moderate (0.32 g), and major earthquakes (1.15 g), as shown in

Seismic wave type | Acceleration peak (g) | ||
---|---|---|---|

X | Y | ||

Minor | El Centro wave | 0.1 | 0.08 |

Moderate | El Centro wave | 0.32 | 0.26 |

Major | El Centro wave | 1.15 | 0.98 |

Curve A is the acceleration response for the bottom of the wall. Curve B is the top of the acceleration response. It can be seen that there is a significant amplification of the acceleration at the top of the wall relative to the acceleration at the bottom of the wall, especially near the peak of the acceleration. The peak acceleration at the top of the wall is 4.07 m/s^{2}; the peak acceleration at the bottom of the wall is 0.6 m/s^{2}. The acceleration peak ratio of the two is 6.78.

Curve A is the bottom acceleration. Curve B is the top acceleration. It can be seen that the wall top acceleration of the adobe wall is significantly larger relative to the bottom acceleration of the adobe wall, especially at t = 5 s. At this moment, the base of the adobe wall shows a failure as seen from the corresponding elements, as shown in ^{2}, and the peak acceleration at the bottom of the wall is 2.4 m/s^{2}. The acceleration peak ratio of the two is 6.138. The acceleration peak ratio is substantially close to the acceleration peak ratio in minor earthquakes.

The acceleration response of the adobe wall under major earthquakes is shown in ^{2}, and the peak acceleration is 6.5 m/s^{2} at the bottom of the wall. The acceleration peak ratio of both is 10.14. Due to the unrecoverable damage at the bottom of the wall, the structural stiffness mutation and elastic restoring force are not enough to support the superstructure, and the superstructure will undergo acceleration mutation under the action of inertial force, which is similar to the whipping effect of the structure under the action of an earthquake.

By selecting suitable test parameters from experimental studies, developing a corresponding finite element model, numerical simulation of adobe walls under the action of the seismic load is performed in order to find out how the seismic performance of earth adobe walls can be improved. The following conclusions can be drawn from this study:

It can be seen from the equivalent stress cloud diagrams that the stress concentration areas of the bottom of the wall are prone to shear failure. Thus, it is suggested to increase the strength of the connections of this part. For instance, wooden pillars can be installed to improve the integrity of composite earth walls.

From the acceleration response of the wall, under a major earthquake, it is observed that the acceleration value of the wall top is far greater than the acceleration value of the wall bottom. When the plastic damage occurs, the acceleration value of the oscillation is not synchronized. This indicated that the physical integrity of the wall is weak.

During the wall collapse process, it was apparent that the initial local damage occurred at t = 1.3 s, and the wall collapse occurred at t = 1.8 s. This means a destruction time of less than 1 s, which is quite fast. This obviously indicates brittle damage characteristics. It is suggested to mix additional materials in the soil mix to enhance its structural ductility. This can be achieved by adding, for instance, hay straws.

Involvement and contributions of the second author in this research project, who supervises the doctoral research of the first author, was made possible, in part, due to a release time funding from Donald E. Bently Center for Engineering Innovation, Mechanical Engineering, Cal Poly, San Luis Obispo. Herein this support is acknowledged.