Mechanical behavior analysis of small-scale modeling of ceramic block masonry structures – Geometries effect 1

© 2012 IBRACON This paper presents the experimental results of a research program with ceramic block masonry under compression. Four different block geometries were investigated. Two of them had circular hollows with different net area. The third one had two rectangular hollow and the last block was with rectangular hollows and a double central webs. The prisms and walls were built with two mortar type 1:1:6 (I) and 1:0,5:4 (II) (proportions by volume of cement: lime: sand). One:three small scale blocks were used to test block, prisms and walls on compression. It was possible to conclude that the block with double central webs gave better results of compressive strength showing to be more efficient. The mortar didn ́t influenced the compressive strength of prisms and walls.


Introduction
The red ceramic industry is responsible for creating a number of jobs in Brazil.For instance, in the state of Santa Catarina, there are 742 pottery facilities with a production of approximately 100 million units per month that are responsible for 11,000 direct jobs and 30,000 indirect jobs.Therefore, this industry is socioeconomically very important [1].The pottery sector has approximately the same profile in almost all Brazilian states, showing both a high production potential but also the small-scale technological and investment capacity for creating new products.The red ceramic companies produce various products including tiles, bricks, and both structural and non-structural blocks.The main masonry product used for partitioning space in construction are non-structural blocks containing horizontally oriented circular or rectangular hollows.A Brazilian standard defines the minimum compressive strength of non-structural blocks.There are different shapes and thicknesses of structural blocks that will be further discussed below with regard to their efficiency at stress distribution in masonry walls under compression.The increasing use of ceramic block masonry structures as construction systems in the Brazilian building market has been a factor in generating research projects focusing on development of masonry products that maintain a high efficiency capacity when subjected to external loads.The main goal of this work is experimentally analysing and assessing the influence of ceramic block geometry on the mechanical performance of structural walls under compression on a small-scale, allowing such blocks to potentially become an important component for the Brazilian ceramic industry.

Use of small physical models in structural masonry
One of the greatest challenges in civil engineering is to develop reliable models for representing the behaviour of structures on a full scale, thereby reducing the costs and the difficulties associated with "experimenting" with a full scale.The physical modelling of structures requires understanding of the similarity of the conditions of the models to truly to reproduce the behaviour of full-scale structures with regarding to predicting the ultimate stress, failure modes and stiffness.Physical models therefore should reproduce the fullscale loading, geometry and material properties [2].
One of the first authors to historically depict the use of physical models in structural masonry was ABBOUD et al. [2].ABBOUD et al. reported that VOGT [3] carried out experimental studies on bricks masonry models at 1:4 and 1:10 scales, but failed to obtain consistent data regarding the behaviour of the material.ABBPID also cites that, in the 1960s, studies were performed at Melbourne University with limited success because of difficulties in manufacturing bricks and constructing walls.ABBOUD also mentions that MOHR [4] achieved success in the execution of walls by using commercial units and prefabrication techniques at a 1:6 scale.
The studies carried out by ABBOUD et al. [2] with concrete blocks units showed that there was reliability to be gained from using small models for predicting the complex behaviour of structural masonry.ABBOUD obtained excellent correlation between model results when compared to prototypes, but the standard deviation was smaller in prototypes.This result was obtained by reduction of the effect of stress volume.
In Brazil, CAMACHO [5] was the first to perform studies on the compressive behaviour of block masonry.The author stated that masonry is the oldest and most classical construction method used by man, while the implementation of the small model technique for studying structural behaviour is very recent.CAMACHO [5] affirms that studies were carried out at the University of Bath and Karlsruhe University in Germany, regarding the behaviour of small masonry model walls made with ceramic bricks at scales of 1:2 and 1:4.CAMACHO states that those studies allowed the researchers to determine the strength correlations and deformations and therefore verify the parameters that would be affected by a scaling factor.Based on small-scale tests, it was concluded that masonry models can reproduce the failure mode and the ultimate strength when similar materials between the models and the prototypes are employed.However, the value of the relation between the elasticity modulus based on compression strength was reduced with the decreasing of the scale, as shown in Figure 01.CAMACHO [5] carried out experimental studies with hollow clay masonry blocks on full-scale model sand small scales of 1:3 and 1:5.The author performed compressive strength tests of prisms that were two, three and four blocks high as well as small walls.The compressive strength results for blocks, prisms (two, three and four blocks height) and small walls and the stress/strain results for blocks are shown in Figures 02  and 03, respectively.The author concluded in general that axial compression strengths between the small scale and the full-scale prototypes were similar but that for prisms and small walls, the small scale models presented compressive strength values 1.5 times higher for prisms and 1.3 times higher for small walls compared with the prototype.The strain at failure for the small scale and the prototypes were of the same order of magnitude, but the small-scale prisms had 2.4 times the prototype strain, while the small-scale small walls had strain values that were 4.5 times the prototype.In addition, the rela-

Factors that influence masonry strength
The main material responsible for the strength of masonry is the block.There are some Brazilian studies on the mechanical behaviour of clay masonry [7 and 8].In these studies, the blocks have different geometries, and in most cases, the results cannot be compared in terms of mechanical efficiency due to different aspects such as the heterogeneous nature of the mixture, fineness of the clay, hole format and the firing temperature.Furthermore, in Brazil there have been few experiments using small-scale models as a means of interpreting mechanical and physical phenomenon.Thus, it is important to conduct technical studies on material behaviour to support the technical decisions for developing new structural products.

The block influence on masonry strength
Clay blocks are structural masonry components with prismatic or circular hollows aligned in the direction of the load application, so that the block is set with the hollows oriented vertically.Bedding clay blocks are classified as follows: either (a) structural clay tion between the prism and block strength was affected by the scale factor with a value of 0.4, 0.5 and 0.6 for small scales of 1:1, 1:3 and 1:5, respectively.Although there were differences in the strength results, the failure mode presented by the prototype and the small-scale models were similar.CAMACHO concludes that it is possible to use small-scale models to determine the behaviour of clay block masonry.NETO [6] studied theoretical and experimental behaviour of masonry walls with openings, using small-scale models of 1:3.The author determined the mechanical behaviour of the components and elements.The studies used structural clay blocks with two rectangular hollows.The thickness of face shells and cross webs was the same, and the relation between the net area and gross area was 55%.The dimensions of the units were 9.82 cm x 6.59 cm x 4.65 cm (length x height x width).Ta�le � � ��pe�imental �esults of the components and elements of Nascimento ( 2003) Mechanical behavior analysis of small-scale modeling of ceramic block masonry structures -Geometries effect (1:0.5:4.5),iii (1:1:6) and iv (1:2:9).As shown in Figure 06, the ratio of the compressive strength of the walls in relation to the compressive strength of blocks tends to decrease with increasing compressive strength of the block, and this ratio is higher for bricks than for blocks.The BSI 5628-1 [10] considers only the relation of the dimensions (height and width) of the block and does not taking into account the geometry and the arrangement of the hollows.
For walls with relation between height (h) and width (w) of 0.6 to 2.0, the value of the compression strength of masonry should be obtained from Figure 06.

The influence of block geometry on the compressive strength of masonry
During the load application, the quantity and the arrangement of hollows and shapes may lead to a concentration of stress on the block that can decrease the potential strength of masonry, according to work performed by GANESAN and RAMAMURTHY [11].
The authors stated that it is necessary to understand the geometry effect of blocks to increase the efficiency of structural walls.The authors carried out some analytical studies using finite element methods to better understand the behaviour of concrete masonry blocks, taking into account the influence of different geometries, arrangements and properties of mortars.GANESAN and RAMA-MURTHY [11] proposed the use of blocks with three types of geometry, including one with a double central web, that is, where the thickness of the central web was twice the thickness of the face shell that provides the alignment of the hollows.The geometries were modelled with stack and running bond prisms with three courses, using three different geometries of concrete block: blocks with three hollows, blocks with two hollows and blocks with two hollows and a double central web.Four types of mortar were used in the masonry to compare the proportion between the elasticity modulus of blocks (E b ) to that of the mortar (E a ), where the proportions were 1; 1.5; 2.0 and 2.8, while E b was held constant.Establishing a stiffness ratio of E b /E a as a constant was found to affect the mortar and the failure mode of the masonry.It was used a heterogeneous elastic-linear behaviour in the model, by using a solid element of eight (8) nodes for determining stress on the face shells and cross Figure � � Types of the blocks studied by GANESAN and RAMAMURTHY [11], with the dimension in mm tested at Building Technology Laboratory
The results showed that blocks with three vertical hollows produced higher levels of stress than blocks with two vertical hollows.
The stress level remained constant in the region close to the centre of the prism.Regarding the cross webs, the difference in behaviour among the three types of blocks was more evident for running bond prisms.As a conclusion of the work of GANESAN and RA-MAMURTHY [11] on the mechanical behaviour of masonry, it can be verified that the geometry of the block influenced the distribution and the magnitude of the stress level.Moreover, mortar did not influence the behaviour of masonry and stack bond prisms overestimated the masonry strength.Another important conclusion was that the ratio of the compressive strength of walls to that of blocks depended on the block geometry and the type of laying mortar.
The authors found that for some geometries and mortar, there are stress concentrations that reduce the compression strength of masonry.Figure 07 shows the geometries and the compressive strength of blocks and walls along with efficiency factors.For Type-A blocks the mortar was applied only on the faces shells while for others the mortar was applied on the entire surface of the block.

The mortar influence on masonry strength
Development of units (blocks) with greater compression capacities requires a proportional strength increase in the mortar joint, due to a failure mechanism of masonry that is closely related to interaction among these components, as it is shown in Figure 08.Several studies were carried out in Brazil to determine the influence of mortar, in which the studies carried out by GOMES [7] stand out.GOMES concludes that mortar strength should be between 0.7 to 1.0 times the block strength measured over the gross area.GOMES state that when mortars with a compression strength close to that of the block are used, the masonry will display an excessively fragile failure with subsequent instability of the structure.MENDES [8] also conducted studies on hollow clay block prisms that were 140 mm wide x 290 mm long x 190 mm high (shape of Figure 05-b), where the relationship between the net and gross areas was 0.52.Experiments were conducted on grouted and non-grouted prisms with two compression strengths of mortar.Based on the studies of MENDES [8], it can be observed that failure of non-grouted prisms is due to the crushing of mortar joints generating tensile concentrations in the blocks and splitting the contact surface between the block and mortar.The failure types of non-grouted prisms were fragile for prisms with mortar A1 and the crushing of the block lateral walls for mortar A3.For grouted prisms, all walls of the block (face shell and cross webs) were separated.The separation was caused by the lateral expansion of the grout creating tensile concentrations that separated the face shell and the cross webs.Figure 09 presents the individual results for the block (B1), mortars (A1 and A3), grouts (G1, G2 and G3) and the different strength combinations between non-grouted and grouted prisms.The failure modes of grouted and non-grouted prisms are presented in Figure 10, as well as block geometry and the failure process for grouted prisms.Regarding the recommendations of BSI-5628-1 [10] in Figure 06 and the experimental results of GOMES [7] and MENDES [8], it can be concluded that mortar strength did not significantly influence the compressive strength of masonry for block strengths from 2.5 to 10 MPa.However, for blocks with compressive strength greater than 10 MPa it was verified that the mortar influenced the compressive strength of masonry.

Experimental Program
An experimental program was carried out with prepared clay unit blocks and masonry components using small-scale models with proportions of 1:3.

Clay for laboratory production of ceramic units
One of the first challenges of this work was to study the ideal clay composition for block fabrication.The clay mixture should have plasticity when mixed in water, so it can be shaped, contain sufficient strength for keeping that shape and be able to fuse particles at high temperature.The plasticity of the clay and the influence of the drying and burning protocols depended on the particle size and the minerals present in the clay.To produce units on a small scale, clays were composed of colloidal particles with diameter smaller than 0.005 mm.The final product, (i.e., the clay blocks) should have physical properties such as aspect, dimension, squareness and flatness that meet the according to the standardized recommendations presented in Table 02 of NBR 15270-2 [9].LINDNER [12] helped to develop a clay mixture for these studies.Two types of clay were used for fabrication of the units.The clays were subjected to blending, grinding and homogenization.
Ta�ela � � Dimensional tolerances related to the average of effective dimensions Mechanical behavior analysis of small-scale modeling of ceramic block masonry structures -Geometries effect fluorescence carried out at "Centro de Tecnologia em Cerâmica" (Masonry Technology Centre) in Criciúma, Santa Catarina state.In this study, blocks were fabricated with different geometries on a scale of 1:3.The ceramic mass was conformed through an extruder, where the mass is pushed through an opening called a mouthpiece in the geometry of the desired block shape.The extruder was equipped with a vacuum chamber to facilitate removing air from the block.

Mechanical characterisation of blocks and mortars
In the evaluation of the influence of block geometry on the mechanical behaviour of masonry, experimental studies on the compression strength of units (blocks), prisms and walls at a small scale of 1:3 were carried out using two types of mortar.The small-scale blocks were 4.67 x 6.33 x 9.67 cm. Figure 11 shows the different geometries of the blocks as well as the dimension and an image of the small-scale prisms.The main goal of the experimental program was to use the small-scale models to investigate the influence of the block geometry on structural masonry when submitted to compressive stress, and to determine the potential use of the small scale to represent masonry behaviour.Four different types of block geometry designated type A, B, C, and D were used.Reduction of the geometric scale was applied for the bedding mortar joint and for the vertical joints of prisms and walls.To keep the properties of the joint equal to the real scale, a reduction in the particle size distribution of the mortar sand was performed.A sand was selected that best fit the particle size limits in the British The two clays were dosed and blended in the feeder, which breaks up the mixture prior to the horizontal mixer.In the final phase of the production process, water was added to adjust the moisture content for optimum extrusion.Table 03 presents the clay chemical composition determined by chemical analyses by using X-ray  The bedding mortar was prepared using a mixer with a vertical axis.For each mortar, six cylindrical specimen 5 cm in diameter and 10 cm in height were moulded for 28-day compression testing following procedures in NBR 13279 [13].The specimens were cured in a laboratory environment for 28 days to reproduce the conditions of prisms and walls.

Geometries effect
The granulometric distribution of sand used in the experimental tests followed the recommendations of BS 1200 [14].Portland cements CP II-F-32 and CH-III-type hydrated lime were used.Determination of the unitary mass of the cement and the lime followed the procedures described by NBR 7251 [15].Table 04 shows the values of the unit masses of cement, lime and sand.The geometries of the blocks had the following characteristics: 1 -Block type A -a model with two rectangular hollows similar to the format of concrete block; 2 -Block types B and C -both types with two circular hollows.
The block type B maintains the same thickness for the face shell and cross webs, resulting in a higher net area.For block type C the net area was maintained equal to the block type A; 3 -Block type D -a model with two rectangular hollows.The internal cross webs thickness is double the thickness of the face shell plus that of the mortar joint.This causes a meeting on the vertical joints of the mortar.The relationships between the different net areas of the blocks are presented in Figure 13, where it can be observed that the relationship between the net and gross area of blocks A and C (BA/BC)

Figure �� � �risms and �alls for e�aluation of the influence of geometry on masonry (dimensions in cm)
are 1.0, that is, both blocks have the same relationship between the net and gross area.The walls were built with an apparatus that ensured that the blocks were level, aligned and vertical in each row, following the recommendation of NBR 8949 [16].Table 14 presents the first and second rows with the apparatus for execution of the wall.For the different types of blocks, five prisms with and without vertical joint and three walls were built for each type of mortar, as shown in Figure 15.Table 05 presents the descriptions of the different tests of blocks, prisms and walls.The designation PA1 indicate the prisms with block type A and mortar I while the designation PPB2 indicates a masonry wall built with block B and mortar type II.The designation code is as follows: A, B, C, D = block, P = prism; PP = masonry wall; 1 = mortar type 1:1:6; and 2 = mortar type 1:0.5:4.Due to the difficulties of implementing tensile tests on the blocks, it was decided to obtain the tensile strengths of the blocks indirectly by diametric compression as shown in American Standard ASTM C1006-84 [17].The cylindrical steel bars required for the tests were between 1/8 and 1/12 of the height of the sample and had lengths greater than their widths.The bars were aligned with the central crossing web in each block.The load applied at a rate of 0.33 MPa/min.The tensile strength was then determined by using Equation 01. (1)

P . 2 
Where: f t = tensile strength by diametric compression (MPa); P = applied load (kN); L = length (mm); and H = height of the sample.Values of the tensile strength determined by diametric compression are presented in Table 06, together with a depiction of the test device.Sixteen blocks of each geometry were randomly selected for the compression tests.Blocks were prepared for testing by the following procedure: -the top and bottom of the blocks were covered with a mixture of 70% cement paste plus 30% sand retained in the 0.15 mm sieve to avoid cracking caused by shrinkage; -after the capping of the top and bottom of the blocks the specimens were immersed in water for 24 hours; -excess water was removed with a dry rag before the tests Compression tests were performed by incrementally applying the load at a rate of 0.5 MPa/second.The compressive strength of the gross area gives a standard strength for a constant area that is independent of geometry effects.

Mechanical characterization of prisms and walls
Five stack bond prisms that were three blocks high were built with each of the two types of mortar (I and II) and five running bond prisms with an intermediate row composed of two half-blocks and a vertical joint were built only with mortar type I.The three-block height was selected because of the effect of confinement stress produced by the plate so that the intermediate block did not develop shear stress.The prisms have a full mortar bedding (face shell and cross webs) and were built on a levelled granite table and covered with a plastic with oil.The thickness of the mortar joint remained constant on the order of 3 ± 0.1 mm.Levelling was maintained during prism construction while plummeting was maintained during construction of masonry walls.The prisms were tested 28 days after construction.Before the compression tests the prisms were capped with a mixture of 70% cement paste and 30% sand retained on 0.15 mm sieves.Prisms and walls were tested using a servo-controlled machine (SHIMADZU, series UH) with a 200 ton capacity, at a loading rate of 0.05 ± 0.01 MPa/s.Table 07 shows the compression strength results for six mortar type I and II specimens.The experimental results of compression tests of stack and running bond prisms with mortar types I and II are presented in Table 08 along with the standard deviations and coefficients of variation.
The compressive strength results for prisms and blocks were obtained for both the net and gross areas (ES 772-1 [18]).There were no significant differences in the compressive strength results for prisms using the different block types A, B, C and D. The small differences in the strength values are likely due to the superposition of the face-shell and the cross webs of the blocks for the two types of prisms.Table 08 presents the efficiency factor between the compressive strength of prisms and blocks (f p /f b ).It was observed that there was a reduction in the compressive strength between prisms and blocks of approximately 55% to 65%.The failure mode of prisms were similar to those obtained by MENDES [8] who observed a failure caused by crushing of the bedding mortar joint and the splitting of the surfaces between blocks and mortar.The circles in Figure 17 depict the failure modes of prisms.The tests showed that scale-reduced prisms had failure modes similar to those obtained for full-scale prisms.
For each type of block and mortar, three walls were built and tested.The blocks were wetted before bedding so that water would not be removed from the mortar, making it available to hydrate the cement.The blocks at the top and bottom of the walls were capped with the cement paste and sand mixture described above before testing.Walls were tested 28 days after construction, and remained in the laboratory environment between construction and testing.Table 09 presents the compressive strength results for walls with different block geometries and two types of mortar along with the standard deviations and coefficients of variation measured on both the net and gross areas.efficiency factor of prisms and walls with different clay blocks.For the walls using the blocks of types A, B, C and D, the efficiency factors did not depend on the type of mortar (I and II).According to the experimental results, the geometry of block D presented the best efficiency, close to 0.25.The improvement in the vertical distribution of stress over the face shell and the cross web due to the geometry of block D, where the longitudinal wall was twice as thick as the block's wall thickness plus the thickness of the mortar joint, increased the compression efficiency of the masonry.No differences were found for the compressive strength of prisms for different block geometries either with or without the presence of half blocks at an intermediate course.That is, the prisms failed to show influence of block geometry.Thus, it is possible to conclude that the geometry of block D presents a better compression performance compared to the others.Figure 20 shows the failure mode under compression of walls built with different types of blocks.No differences were observed in the failure mode of walls with the block type.The cracks were basically vertical with failure caused by the crushing of the bedding mortar joint and splitting the surface of the block.The axial strain was measured with a mechanical extensometer, namely a "demec-gauge" following the procedures of NBR 8522 [19], as shown in Figure 21.The experimental results are the averages for three samples for each type of block.The results led to the relationship between the elasticity modulus and the compressive strength of masonry, the socalled "Ritter constant" (k) for different block geometries as shown in Equation ( 02).
(2) Ewall= k.Fwall  stress level of 30% of the compression strength.For measurements of the initial stress and strain, different characteristics of elasticity modulus of the walls depending of the block geometry were observed, especially for block types B and C.This difference between blocks B and C was a factor of 1.8.This difference is not thought to be due to the geometric shape of the block but is rather due to the level of confining stress in the mortar caused by friction between the block surface and the bedding mortar joint.For this level of stress, the mortar joint is the main cause for the deformation of walls.The value of "Ritter constant" was lower than those specified in Brazilian and international standards.The Brazilian Standard NBR 15812-1 [20], shows a Ritter constant of 600 while ES 6 [21] recommends a value of 1000.This fact may be related to the degree of compaction of the mortar joint during the bedding of the units, as suggested by ABBOUD [2] and CAMACHO [5].

Conclusions
Based on the results, the following can be concluded: n Blocks with a double central web presented the best performance in wall compression compared to the others.This conclusion is confirmed by the efficiency factor of the set (block and mortar); n Compression tests on stack-bonded prisms with a half-block in the intermediate course were not adequate for checking the influence of the geometry.For the cases studied, the efficiency factor ranged randomly from 0.35 to 0.47 without demonstra-tion of similar behaviours for the same shapes of hollows; n There were no significant differences in the compressive strength of walls caused by increasing mortar strength, with it likely the case that the scale reduction decreased the influence on the strength most likely the reduction on the scale decreased this influence; n The strength and the corresponding efficiency factors showed that strength potential among prisms and walls decreased; n Three-block prisms with an intermediate layer consisting of two half-blocks did not produce lower efficiency factors compared to three-whole-block prisms; n D-shaped blocks provided a more uniform distribution of traction tension than the other geometry shapes due to the coincidence of the webs in transversal walls of the block; n It was found by the results between the deformation modulus by compression strength ("constant of ritter") that the values were significantly lower on the small scale.This finding was demonstrated by the mortar bedding degree at the moment of unit bedding; n The failure mode of prisms and walls on a small scale were similar to the ones found in the literature cited in this work.It can therefore be stated that the study of small models is able to reproduce trials at a scale that is an efficient, practical alternative to full-size trials.The failure mode of prisms and walls was the crushing of bedding mortar joints followed by contact cracking of blocks and bending mortar.

Figure � �
Figure � � Failure mode of prisms and small walls of NETO [6]

Figure
Figure �� � �o��ressive strength on blocks in the net and gross area

Figure 19
presents the individual compressive strength results for blocks, mortars, prisms and walls.

Figure
Figure �� � F�ilure mode of three-block prism

FigureFigure 19
Figure �� � �ompression strength of blocks, mortars, prisms and walls in the net area

Figure
Figure �� � Failure mode of wall under compression Table 01 presents the experimental test results and the corresponding strengths.The failure modes of prisms and small walls presented by NETO [6] showed crushing of the bending mortar joint combined with a splitting displacement at the contact point with mortar, with tensile cracks induced by the application of compression stress.Combined with the compressive strength results for prisms and blocks, it was possible to obtain the efficiency factors of experimental tests of NETO [6], whose values were f p / f b = 0.55 MPa and f small walls / f b = 0.38 MPa.

Table � �
Tensile strength by diametral compression E. RIZZATTI | H. R. ROMAN | G. MOHAMAD | E.Y. NAKANISHI -s.d is the standard deviation and c.v (%) is the coefficient of variation in percentage.
MPa); c.v is the coeficiente of variation (%); f A is the strength in the net area; f A is the strength in the gross area.
net gross

Table � �
Results of co��ression strength of structural wallsWhere: f is the wall strength (MPa); f /f is efficiency factor between wall strength in relation to the block; s.d. is the standard MPa); c.v is the coeficiente of variation (%); f A is the strength in the net area; f A is the strength in the gross area.

Table 10 presents
the average results of the elasticity modulus of walls built with mortar type I.The elasticity was obtained at a