|
 |
| Fig. 1: Bhunga, Budj – India | Fig. 2 Interior view of a Bhunga under construction |
Name of Author
Patrik K. Meyer
Co-author: Professor Khalid Mosalam
Research Sponsor (PI)
Professor Khalid Mosalam
Major, Year, Department
Senior, Civil Engineering
Keywords
Masonry, arch, historical buildings, shear, FRP, India, Bhunga,
Contact Information
Patrik K. Meyer
5514 Shattuck Ave.
Berkeley, CA 94609
patrik@engineering4theworld.org
URL: www.engineering4theworld.org
Bhunga: Indian Traditional Housing
Cylindrical Structure Subassembly Test
ABSTRACT
The purpose of the present research is to investigate the behavior of
cylindrical masonry structures for both traditional dwellings and historical
buildings. Two of the assumed critical regions of such structures are studied in
depth. These include (a) the interface between the foundations and the base of
the cylindrical wall, and (b) the connection been the top of the wall and the
dome-shaped roof. Moreover, the validity of FRP (fiber reinforced polymers)
systems in retrofitting arch-shaped structures, which are common in many
historical buildings, is also investigated. The data obtained from this research
provide a better understanding of the behavior of circular masonry structures
under seismic loads, both as traditionally built and as retrofitted.
Both traditional low cost housing and valuable monumental structures are
expected to benefit from this investigation.
INTRODUCTION
“Bhungas” are cylindrical mud huts that have been traditionally built for thousands of years, (Fig. 1). Their particularity is that they perform exceptionally well under seismic loads. Their performance during the recent Gujarat, India earthquake only confirmed that simple, low-cost housing can be safe, even under extreme conditions. The shear failure of cylindrically shaped masonry structures under seismic loads has not been fully addressed in the scientific literature (Soroushian, 1981). With the purpose of simulating the equivalent shear forces experienced by the cylindrical structure, a series of 40% reduced-scale, 106 degrees three-layered components of the cylindrical walls are constructed and tested. Fig. 2 shows the interior of a Bhunga under construction. The selected experimental subassembly is representative of a “slice” of the cylindrical wall at its junction with the dome-like roof.
|
|
|
| Fig. 1: Bhunga, Budj – India | Fig. 2 Interior view of a Bhunga under construction |
MATERIALS PROPERTIES
The materials used to build the test specimens are solid concrete bricks. The bricks used in the construction of the Bhungas in India are either mud or soil-cement (3% by volume of cement). Due to the fact that the main purpose of the research is to study the shear capacity of cylindrical structures under seismic loads, commercially-available solid 4″×4″×2 ¼″ concrete bricks are used to avoid excessive dependency on the workmanship. The compressive strength of the used bricks is found to be 3500 psi. The mortar used in the construction of the test specimens is designed such that it has a lower compressive strength than that of the bricks. For this purpose a series of mortar probes are made and tested. The average mortar strength obtained is 2800 psi. In addition, a series of small specimens to test the shear strength (Meli, 1985) of the mortar joints are constructed and tested as shown in Fig. 3.
Fig. 3: Testing the shear strength of the mortar joint
CONSTRUCTION
OF TEST SPECIMENS AND EXPERIMENTAL SETUP
The arch-shaped test specimens are constructed using Styrofoam mold to ensure
the uniformity of the geometry of all the specimens. The previously discussed
blocks and mortar materials are used in the construction. After construction the
specimens are allowed to cure for 28 days before testing.
Initially a wooden beam/gluelam is used as supporting element (Fig. 4). During the first test and at a load of about 10 kip, the gluelam experienced shear failure along the glued joints, at which point the test is terminated. Subsequently, a steel beam (Fig. 5) on which two wedges are welded to act as reaction points replaced the wooden beam. The selected boundary conditions of the test specimen are 3D pin-2D roller arrangement as shown in Figs. 6 and 7. Moreover, the exterior masonry layers are the ones restrained at the supports while the interior layer is not supported (refer to Figs. 4 and 5) to force the mode of failure to take place along the two interfaces between the masonry layers. The loading consists of two equal concentrated loads applied at the third points of the arch span as shown in Fig. 5.
|
|
|
|
|
|
TESTING PROCEDURE
|
Loading Cycles The test specimens are loaded in cycles of increasing magnitude as shown in Fig. 8. Three load cycles of the same magnitude are repeated with increasing increment of 10 kip. The loading rate is kept low at approximately 0.5 kip/min. |
|
Test of Specimen 1
The test setup of specimen 1, supported by a 3D pin and a roller allowing the
out-of-plane displacement, failed in shear (Fig. 9) followed immediately by
out-of-plane instability (Fig. 10) at a load approximately 30% lower than
expected. The data acquired during this test is obtained from four wire
potentiometers to measure relative displacements along the specimen and six
strain gauges to measure strains in the concrete blocks and the mortar joints.
In addition, both the test machine head displacement and the applied load are
continuously recorded during the experiment. The shear of the longitudinal
middle layer occurred along the interface between the blocks and the mortar
without any observed damage to the blocks. The load at failure is 20 kip and the
maximum vertical displacement is of 0.07″ as shown from the total load
versus mid-point vertical displacement relationship in Fig. 11.
|
|
|
 |
| Fig. 8: Test setup: pin-roller | Fig. 9: Crack initiation | Fig. 10: Failure of the specimen |
 Fig. 11: Load-displacement relationship of test specimen 1 |
Test of
Specimen 2
The major difference between the test setup between specimens 1 and 2 is the replacement of the 3D pin in specimen 1 by another 2D roller in specimen 2. This second roller hindered the out-of-plane instability upon occurrence of the shear failure. The number and arrangement of gauges used in this second test are slightly different from the first test. A total of eight displacement wire potentiometers are used, which included three to measure the relative displacement along the arch, four to measure the relative displacement of the transverse layers and one to measure the head displacement. Longer strain gauges (8″ long) are in such a manner that they span from block to block (Fig 12). Unfortunately, the readings obtained from these long strain gauges are too noisy. The failure mode observed in the FE model (Fig. 13) is very similar to that of the actual test specimen 2. The FE model failed in shear along the longitudinal interface between the masonry layers at an ultimate load of 23 kip. It should be noted that the FE model does not include the nonlinearities preceding the failure because the properties of the interface are linear-brittle. A future improvement of the model will include more realistic nonlinear behavior of the masonry and interface (Nguyen, 2003).
Click
on the picture to see the failure video  |
Click
on the model to run failure |
 |
 |
Fig.
12: Specimen 2 after shear failure without out-of-plane |
Fig.
13: Model failure top view |
The observed
failure for specimen 2 is a pure shear along half of the longitudinal middle
layer. This shear failure is followed by a bending failure (Fig. 12). It is
worth mentioning that this bending failure is a consequence of the selected
geometry and boundary conditions. In the real Bhunga structure, this bending
failure is unlikely. The recorded ultimate load in this case is 30 kip and the
maximum vertical displacement is 0.13″ as shown in the total load versus
the mid-point vertical displacement relationship, which is similar in shape to
that shown in Fig. 11. In this case, the obtained load is more representative of
what is expected in the actual Bhunga structure because the rest of the
structure will restrain the out-of-plane displacement.
Test of Specimen 3
The particularity of test specimen 3 is that it is retrofitted with FRP. In this case a bi-directional carbon fiber mesh with an ultimate strength of 600 ksi is used. After the arch specimen is cured for 28 days, a layer of carbon fibers is applied on each side (top and bottom) of the specimen. Special attention is give to guarantee that the FRP is properly bonded to the concrete blocks to allow the proper load transfer from the blocks into the fibers. The purpose of the retrofit with FRP is to validate that the FRP system is able to prevent the shear failure by providing resistance to the transverse tensile stresses caused by the dilation of the sheared mortar joints. In this way, the full compressive strength of the blocks is expected to be utilized. This type of retrofit technique has not been previously tested on an arched masonry structure.
The obtained failure of test specimen 3 is indeed by crushing of the blocks at the support level (Fig. 14). The initial crushing failure is followed by unsymmetrical bending of the specimen, which caused the fibers to debond from the arch at the bottom of the specimen (Figure 1). The crushing occurred at load of 60 kip (Fig. 15), two times as much as the largest load obtained from the as-built test specimens. The vertical displacement of this third test specimen at the mid-point of the arch is 0.26″ (Fig. 16), again two times higher than the maximum vertical displacement of the as-built specimens. This indicates that strengthening of the arch using the almost mass less FRP system did not affect the stiffness but rather affected the strength. This is a desirable stiffness/strength uncoupling in the design a retrofit system for seismic loading where increase of stiffness may lead to problems in other components of the structure such as the foundation.
|
|
 |
| Fig. 14: Crushing of arch due axial force | Fig. 15: Debonding of the FRP from the arch |
|
|
|
Fig. 16: Load-displacement relationship of test specimen 3 |
FINITE
ELEMENT MODEL
The 3D finite element (FE) model created using ABAQUS is composed of three homogeneous longitudinal layers (Lourenco, 1996) connected to each other using the shear properties of the mortar-bock interface from the shear tests. The compressive strength given to the longitudinal elements is the same as the one found from the brick tests. The reason for not using the lower compressive strength of the mortar joints is that in the arch assembly, the joints are highly confined by the blocks, which increases their actual compressive strength. In the FE mesh, each block-mortar unit is divided into four cubical solid elements (Fig. 17a). The mesh is denser near the crown of the arch to provide better resolution on the onset of the shear failure (Fig. 17b). The middle layer of the FE model is slightly shorter than the outer layers (insert of Fig. 15b) as was the case in the test specimen to allow it to displace in shear by restraining the exterior layers only.
|
Fig. 17a: Meshing of the model |
Fig. 17b: Detail of meshing at the support Note: Observe that the middle layer is |
CONCLUDING REMARKS
The failure mechanism observed in the first two specimens is a direct shear failure along the interface between the bricks and the mortar of the entire middle layer of the test specimens. Although a shear failure is expected, the fact that the failure is expected to occur along the much longer longitudinal interface prior to any compression failure of the blocks or in joints between the blocks, is somewhat surprising.
The FRP retrofitted third specimen shows remarkable improvement in its structural performance. Both the load capacity and the displacements are doubled. The positive effects obtained by retrofitting the specimen with FRP can be observed in Table 5, where the maximum loads and displacements of each of the three specimens are compared. These promising results confirm that FRP is an effective retrofitting system for cylindrical-shaped masonry structures.
 |
|
Table 5: Max. failure load and displacement of the 3 different specimens |
ACKNOWLEDGEMENTS
The authors would like to express their gratitude to The Hellmann Family Faculty Found for providing the major part of the financial support. We are very thankful to Mr. Lev Stepanov of the structural laboratory at UC-Berkeley for his assistance in the construction and testing of the specimens. The material and labor donations of Sigma Composites, LLC, Calstone Company, Sunnyvale, and E&S Masonry are greatly appreciated. Last but not least the technical support of Dr. A. Mosallam of California State University at Fullerton, Dr. V. Mujumdar of CMACM and Dr. M. Serrar of ABAQUS, Inc. was essential for the development of the research.
