Earthquake response of heavily damaged historical masonry mosques after restoration

Restoration works have been accelerated substantially in Turkey in the last decade. Many historical buildings, mosques, minaret, bridges, towers and structures have been restored. With these restorations an important issue arises, namely how restoration work affects the structure. For this reason, we aimed to investigate the restoration effect on the earthquake response of a historical masonry mosque considering the openings on the masonry dome. For this purpose, we used the Hüsrev Pasha Mosque, which is located in the Ortakapı district in the old city of Van, Turkey. The region of Van is in an active seismic zone; therefore, earthquake analyses were performed in this study. Firstly a finite element model of the mosque was constructed considering the restoration drawings and 16 window openings on the dome. Then model was constructed with eight window openings. Structural analyses were performed under dead load and earthquake load, and the mode superposition method was used in analyses. Maximum displacements, maximum– minimum principal stresses and shear stresses are given with contours diagrams. The results are analyzed according to Turkish Earthquake Code (TEC, 2007) and compared between 8 and 16 window openings cases. The results show that reduction of the window openings affected the structural behavior of the mosque positively.

Mosques which have beautiful appearance and high religious value are very important for Muslims. Many mosques have been built in Muslims area and most of them were built with masonry technics. There are lots of masonry mosques in Turkey and most of these mosques are especially on seismic zones. The presence of earthquakes in Turkey caused damage and destructions on masonry mosques. Also Hüsrev Pasha Mosque which was located in active seismic zone was damaged from some seismic events. In the restoration of Hüsrev Pasha Mosque there is being applied an application which is about 5 reduction of window openings on dome for to improve the structural performance. So it is required to investigate this case.
For this purpose the restoration effect also reduction of window openings effect on the earthquake response of masonry mosques are investigated in this paper.
Today there are some cases which affect the structural performance of the mosque detrimentally. These cases are structural cracks, material deformations, destroyed parts of carrier system, environmental issues-algae and humidity. Some views of these cases and damaged views of the mosque are shown in Fig. 3.

3 Structural Analysis of the Mosque
Finite element analyses were made for investigate the restoration effect on the earthquake behaviour of the mosque considering different opening ratios on dome. The model was constituted with restoration drawings considering 8 and 16 openings on dome. Required information for modelling and analysing obtained from contractor firm, restoration drawings and other sources. Finite element models of the mosque were created with SAP2000 structural analysis program (SAP2000, 10 2015. Earthquake analyses were affected x, y, z directions for better comprehend the restoration effect with different openings on dome. After analyses maximum displacements, maximum-minimum principal stresses and shear stresses were obtained and results are given with contour diagrams.
Structural analyses of the mosque are listed below: 15 Structural analyses of the mosque with 16 window openings Nat. Hazards Earth Syst. Sci. Discuss., doi:10.5194/nhess-2017-141, 2017 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Finite element models of the mosque were constituted with using bar, area and 3D solid elements in SAP2000 program. The 10 bar, area and 3D solid elements have 2, 4 and 12 nodes respectively and each node has three degrees of freedom.
Linear elastic methods cannot give exact solutions in the structural analyses of historical masonry structures which are built with brick and stone masonry. Despite that, nonlinear analyses can give exact results if material properties of masonry are defined in analyses correctly. But if the analysed structures are very big and complex there could be some problems at iterations in nonlinear analyses. Also defining the material properties of masonry structures are fairly difficult because of 15 their non-homogeneity. For these reasons all analyses were made as linear elastic. Material properties which were used in the analyses are given with Table 1.  Mode Superposition Method was used in earthquake analyses and elastic behaviour of the mosque under vertical and earthquake loads were obtained with using the full square consolidation method. Spectral acceleration coefficient was taken as S(T)=2.5 and effective earthquake coefficient was chosen as A o =0.3 for related the region which is in second seismic zone. In masonry structures when the tension stresses reached the tension strength of material, cracks and weakness occur. 10 This situation is taken into consideration in the analyses with using the earthquake load reduction coefficient (R a ) in each period. There is no reduction made in this study (R a =1) but safety stresses are enlarged with 3.
Hüsrev Pasha Mosque wasn't built with considering any building code; it was also built with experience. There are some rules about openings for masonry in the Turkish Earthquake Code (2007). When the mosque is assessed with considering these rules, there has no discrepancy with code. Some rules which are exist in the code about openings for masonry related to 15 this study given below (TEC, 2007):  Plan length of the solid masonry wall which is between the corner of a building and the nearest window or door opening shall not be less than 1.50 m for the first and second seismic zones and 1.0 m for the third and fourth seismic zones.
 Plan length of the solid masonry wall which is between window and door openings shall not be less than 1.0 m for 20 the first and second seismic zones and 0.8 m for the third and fourth seismic zones.
 Except for the corners of buildings, plan length of a solid masonry wall which is between intersection of the walls and the nearest window or door opening to the intersection of the orthogonal walls shall not be less than 0.50 m in the all seismic zones.

Structural analyses of the mosque with 16 window openings
The mosque has 16 window openings before the restoration so finite element model of the mosque was constituted with 16 window openings. The model of the mosque has 27297 nodes, 127 bar elements, 25653 area elements and 102460 3D solid elements. Finite element model of the mosque with 16 window openings is shown in Fig. 4. Nat. Hazards Earth Syst. Sci. Discuss., doi:10.5194/nhess-2017-141, 2017 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 20 April 2017 c Author(s) 2017. CC-BY 3.0 License.

Defining the dynamic characteristics and modal analysis of the mosque with 16 window openings
Dynamic characteristics were obtained with modal analysis. %5 damping ratio was used in the analysis. 20 mode shapes were obtained after analyses. First four mode shape and frequency are given with Fig. 5.

Structural analyses of the mosque with 8 window openings
The mosque has 8 window openings before the restoration case so finite element model of the mosque was constituted with 8 window openings. Finite element model of the mosque has 27305 nodes, 127 bar elements, 25685 area elements and 102588 3D solid elements. Finite element model of the mosque with 8 window openings is shown in Fig. 18.

Defining the dynamic characteristics and modal analysis of the mosque with 8 window openings
Dynamic characteristics were obtained with modal analysis. %5 damping ratio was used in the analysis. 20 mode shapes were obtained after analyses. First four mode shape and frequency are given with Fig. 19.    Nat. Hazards Earth Syst. Sci. Discuss., doi:10.5194/nhess-2017-141, 2017 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 20 April 2017 c Author(s) 2017. CC-BY 3.0 License.

Structural response of the mosque with 8 window openings under dead load and horizontal earthquake load (G+EY)
The maximum displacements contour diagram of Hüsrev Pasha mosque with 8 window openings under dead load and earthquake load (G+EY) is shown in Fig. 24. It can be seen from the Figure    Nat. Hazards Earth Syst. Sci. Discuss., doi:10.5194/nhess-2017-141, 2017 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 20 April 2017 c Author(s) 2017. CC-BY 3.0 License.

Structural response of the mosque with 8 window openings under dead load and horizontal earthquake load (G+EZ)
The maximum displacements contour diagram of Hüsrev Pasha Mosque with 8 window openings under dead load and earthquake load (G+EZ) is shown in Fig. 28. It can be seen from the Figure 28 that the maximum displacement occurred at the middle point of the big dome as 9.80 mm. Beside this displacements have a decreasing trend from top of the dome to 5 lower part of the mosque.  Nat. Hazards Earth Syst. Sci. Discuss., doi:10.5194/nhess-2017-141, 2017 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 20 April 2017 c Author(s) 2017. CC-BY 3.0 License.
After the analyses, maximum displacements, maximum compression and tension stresses and maximum shear stresses results for 16 window openings and 8 window openings cases collected in Table 3 and Table 4 respectively.
According to the results the displacements reduce related to reduction of window openings. These reduction ratios are %14, %11 and %50 for dead and earthquake analyses with x, y and z directions respectively. This result shows that reduction of window openings affects the displacements substantially. 5 The results show that the compression stresses decrease when the window openings are reduced. At the outer side of the mosque these reduction ratios are %15, %13 and %15 for dead and earthquake analyses with x, y and z directions respectively. Beside this at the inner side of the mosque the reduction ratios are %18, %13 and %16 for dead and earthquake analyses with x, y and z directions respectively.
The results show that the tension stresses decrease when the window openings are reduced. At the outer side of the mosque 10 these reduction ratios are %9, %16 and %19 for dead and earthquake analyses with x, y and z directions respectively. Beside this at the inner side of the mosque the reduction ratios are %10, %13 and %27 for dead and earthquake analyses with x, y and z directions respectively.
It is seen from the tables the shear stresses decrease with the reduction of window openings. At the outer side of the mosque these reduction ratios are %15, %16 and %40 for dead and earthquake analyses with x, y and z directions respectively. 15 Beside this at the inner side of the mosque the reduction ratios are %16, %18 and %55 for dead and earthquake analyses with x, y and z directions respectively.
Stresses generally occur at near the openings. Therefore openings decrease the structure stability. With the reducing the opening ratios so ensure the integrity on the walls, the stresses and the displacements decrease substantially.

Conclusions
In this study restoration effects on the earthquake behaviour of masonry mosques is investigated with considering different opening ratios on dome. For this purpose Hüsrev Pasha Mosque was selected and finite element model of the mosque constituted with SAP2000 software. Two cases considered in analyses for to determine the restoration effects namely firstly 5 the mosque constituted with 16 window openings then constituted with 8 window openings. In order to better understand the earthquake behaviour of the mosque, earthquake loads were affected from three directions, x, y, z. Mode Superposition Method was used in earthquake analyses. As a result of the study the following observations were made:  To understand the structural behaviour of Hüsrev Pasha Mosque, firstly 16 then 8 window openings on dome taken into consideration in modelling and four different analyses were made for each case, those are modal analysis and 10 earthquake analyses which were effected x, y and z directions.  Compression stresses, tension stresses and shear stresses collects some critic areas which are especially near openings and crossing points because of this, those areas must construct as monolithic in restoration process.
 When the displacement results are compared, %14, %11 and %50 decrease ratios are obtained for dead and 15 earthquake analyses with x, y and z directions respectively.
 According to the comparison between the compression stresses results, decrease ratios are obtained as %15, %13 and %15 for dead and earthquake analyses with x, y and z directions respectively for outer side of the mosque and %18, %13 and %16 for dead and earthquake analyses with x, y and z directions respectively for inner side of the mosque. 20  According to the comparison between the tensile stresses results, decrease ratios are obtained as %9, %16 and %19 for dead and earthquake analyses with x, y and z directions respectively for outer side of the mosque and %10, %13 and %27 for dead and earthquake analyses with x, y and z directions respectively for inner side of the mosque.
 When the shear stresses results are compared, decrease ratios are obtained as %15, %16 and %40 for dead and earthquake analyses with x, y and z directions respectively for outer side of the mosque and %16, %18 and %55 for 25 dead and earthquake analyses with x, y and z directions respectively for inner side of the mosque.
 FRP strengthening can be used for to resist the tension stresses in restoration applications.
Consequently, the reduction of window openings ensures integrity on the walls, so this situation supports the structural performance of the mosque. It is seen from the study that restoration applications, especially reduction of the window openings on dome are improves the earthquake response of the mosque. 30 Nat. Hazards Earth Syst. Sci. Discuss., doi:10.5194/nhess-2017-141, 2017 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 20 April 2017 c Author(s) 2017. CC-BY 3.0 License.