EXPERIMENTAL RESEARCH INTO DYNAMIC PROPERTIES OF MASONRY BARREL VAULTS NON-REINFORCED AND REINFORCED WITH CARBON COMPOSITE STRIPS

The study of dynamic behaviour of vaults of historic buildings reveals new knowledge which can be used for the local analysis and stabilisation and rehabilitation designs of damaged vaulted structures. The analysis of the results of dynamic loading brings objective background material for the identification and localisation of failures according to MAC or COMAC criteria [1, 2] and the assessment of serviceability and structural reliability of vaulted structures of historic buildings.


INTRODUCTION
Barrel vaults belong to the basic vaulted structures.They have been used (to a greater or lesser extent) in all architectural styles starting from Romanesque architecture and they are found in nearly every historic and heritage building.Their shape, derived from the basic cylindrical shape, is the point of departure for a whole group of vaulted structures starting from the cross vault.Barrel vaults were used for roofing buildings with square, rectangular as well as irregular plans, for roofing staircases (straight, sloping, spiral, rising) and other premises for both representative and everyday use.
At the onset of Romanesque architecture in the Czech lands, barrel vaults formed a ceiling swollen to a semi-cylindrical shape which fully sits on, "merges" with vertical masonry.In its secondary, rudimentary, purely structural function, the barrel vault survived the whole Middle Ages.Artistically treated barrel vaults can be found in single cases of two-nave and three-nave halls of townspeople's Mazhauses in South Bohemia (Český Krumlov, Zlatá Koruna) and in gothic castles (e.g.Točník, Švihov).From the Renaissance to the end of the 19th century, barrel vaults were commonly used for roofing both small and large spaces.
The restoration of historic buildings is frequently connected with reconstructions and rehabilitations of barrel vaults damaged mostly by the effects of induced deformations.In regions with the occurrence of natural seismicity, the failures of vaults are caused by their inelastic response to dynamic effects (Figure 1).

Vault failure mechanisms
The vaults failure mechanism arising by exceeding the ultimate strength (deformation) of vault masonry in tension differs from the failure mechanism of masonry pillars.The failure mechanism of vaults (curved masonry) is applied in cases where the thrust line describing the position of the origin of the internal compression force R in all vault cross sections (being the resultant curve of the loading acting on the vaults, vault supporting reactions and potential changes in vault embedding) does not pass through the inside third of the vault cross-section's height.
Vaulted structures are characterised by their high sensitivity to deformations of the supporting structure and, as a result, their response to dynamic effects causing displacements of the supporting structure (supporting system) can be the cause of an appearance of the mechanical failures or a complete failure of the vaulting system.Numerical analyses manifested high sensitivity of vaults to the deformation of the supporting structure (supports) where is, above all, the horizontal displacement (deformation) of the supports in low-order of magnitude values (mm) that can be the cause of the appearance of tensile stresses in corresponding vault sections and their successive failure due to tensile cracking.In this context, the deformations of the vault supporting structure induced by subsoil vibrations due to seismic effects may be extremely severe.
The vault failure process is very complex, including two significant mechanismschanges in a shape, both local and of the whole vaulting system, and the vault masonry failure itself due to the action of tensile and compressive normal stresses reaching, in origins of tensile cracks where the plastic hinges appeared, ultimate values of masonry load bearing capacity in compression.The total vault failure (collapse) is, therefore, usually the result of two correlated parallel processes.It is characterised by the vault bucking together with its local failure, followed by mechanical masonry disintegration and the vault collapse.Both of the processes above are simultaneous and inseparable from each other.The vault stress state during its loading and its failure is accurately described by the thrust line pattern in individual phases of the vaulting system's action.The vault failure mechanism and reaching the vault ultimate load-bearing capacity is dramatically affected by the type of loading, particularly its potential asymmetry, its shape and geometric deflections and imperfections and, last but not least, the stiffness and stability of the supports (Figure 2).The use of FRP materials in the reinforcement and stabilisation of historic, mostly masonry structures, brings numerous advantages, particularly in terms of their low weight, high effectiveness and potential reversibility [3].The application of FRP materials in the area of historic and heritage buildings to-date has been primarily focused on the stabilisation of vertical loadbearing and vaulted structures to withstand the effects of horizontal loading induced by technical and natural seismicity [4,5].

Experimental research into the response of barrel vaults to dynamic loading
Experimental research into segmented barrel vaults was performed on test pieces fabricated in a 1:1 scalesegmented masonry barrel vault sections 0.75 m in width, with a span of 3 m, with a rise of 0.75 m and a vault masonry thickness of 0.15 m, made of bricks of P15 quality and MVC 2 mortar (Table 1, Figure 3).The objective of experimental research was the verification of the response and dynamic characteristics of segmented masonry barrel vaults exposed to repetitive static loading and dynamic loading in the horizontal and vertical direction.Nowadays, the dynamic response is very often used for the identification of potential damage to structures which need not be detectable by other means.The principle of such tests is the comparison of dynamic characteristics of structures exposed to low-level dynamic (i.e.non-destructive) excitation.The investigated characteristics are most often resonant frequencies and corresponding oscillation shapes.A change in frequency, most often its drop, may indicate the appearance of internal cracks in the tested specimen.A change in the oscillation shape, in turn, indicates its global failure. K36reinforced vault, reinforcement at the extrados (upper surface) and on the soffit (lower surface) always with 2 carbon fabric strips (see K35).The strips are 100 mm in width, the composite thickness for the calculation is 1 mm.The strips on the upper surface (extrados) again cover the whole length of the extrados, while the strips on the lower surface (soffit) are placed at the vault crown, the length of strips being 1180 mm.The strips (strip edges) are placed 137 mm from vault edges.
The experimental tests followed the procedure specified in Table 2.

Response of barrel vaults to static loading
The static loading of segmented barrel vaults was carried out with the help of two synchronised hydraulic actuators applying individual, monotonously rising steps of 2x 3kN.Partial results of the behaviour of barrel vaults symmetrically loaded in one third of their span with a pair of vertical forces (see Figure 3) are summarised in Table 3 and graphically displaced in Figures 4 and  5.
Tab. 3  The results of experimental values obtained from the investigation of the response of vaulted structures to static loading imply (comparison of vault deformation values from the second and third static phasesee Table 2) that the effect of dynamic loading results in the "stabilisation" (consolidation, "strengthening") of the vaulted structure.We may presume that some portion of non-elastic deformations occurred during the first dynamic test (additional pushing of bed joints, stabilisation of the supporting system, etc.).These values can be experimentally measured (during dynamic loading) with difficulty.This issue will be the subject of further experimental verification.

Response of barrel vaults to dynamic loading
Dynamic loading in both the vertical and horizontal direction was applied with the TIRAvib electrodynamic exciter, type TV5550/LS with a weight of 550 kg.Its operating frequency range is from 0 up to 3 kHz and the maximum rated travel of mobile mass is 100 mm.The exciter was fixed to the vault and adapted to vertical and horizontal oscillations (Figure 6a).The vertical response was measured with five Wilcoxon accelerometers, model CMMS 793L, with an output sensitivity of 51 mV/ms -2 .One of them was positioned on the mobile part of the exciter to read the mass motion.Other sensors were positioned on the vault (Figure 6b).
Following each (partial) static loading, the vaults were exposed to dynamic loading.It was after the first static test in the vertical direction, after the second static test in the horizontal direction to verify the influence of dynamic effects on the gradual failure of a non-reinforced and reinforced vault (reduced stiffness) and to obtain findings about the effect of vault reinforcement with composites in terms of increased vault's resistance to dynamic loading.A total of 6 vault states were produced: Ainitial undamaged state Bstate after the first static test (including transition into non-linear state) Cstate after the first dynamic loading in the vertical direction Dstate after the second static test Estate after the second dynamic loading in the horizontal direction F -state after the third static test (until structure's failure) Prior to each loading, the identification impact test had been performed and resonant frequencies were identified from it.The identification test had been performed with a hammer with a rubber head.The induced vault oscillations (its response) were transformed into frequency spectra from which frequency peaks were read.Dynamic loading effects were performed both at the first resonant frequency (ca 18Hz) and outside resonant frequencies, namely at frequencies of 5Hz and 50 Hz, which characterise common external loading in buildings or traffic-induced dynamic loading.
Following each loading step, especially after static loading, there was a drop in the effective vault stiffness, which was manifested by a drop in the natural frequencies and, in some cases, by increased internal damping, most likely due to the appearance of microcracks over the whole vault cross section.Changes in stiffness and geometry after each loading step were manifested by the changes in the dynamic characteristics during the monitoring of changes in the vault's state.Frequencies were also detected on a numerical model (Figures 7 -10) comparing theoretical and experimental oscillation shapes corresponding to the 1st to 6th natural frequencies.The obtained patterns show satisfactory agreement of theoretically and experimentally identified oscillation shapes.Successive harmonic loading occurred in the vicinity of such identified resonant frequencies with the objective of identifying also resonant oscillation shapes.Tables 4-6 present frequencies before loading (see Table 2) and after individual loading steps.As was assumed there is a visible drop in frequencies connected with the progressive damage of the vaulted structure.This is due to the decrease in its stiffness, which is evident mainly in non-reinforced vaulted structures.In the case of a reinforced vault, K35, there was a re-growth in frequencies detected after the second static test applying to all oscillation shapesall frequency peaks.This is most likely caused by the cross-section's consolidation, i.e. the closing of cracks where the upper vault reinforcement probably plays a positive role.In the case of vault K34, this phenomenon is not so distinctive, the frequencies only rise in the first resonant shape, which is dominant, i.e. very important for the detection of damage.

CONCLUSION
Experimental results represent very important background material for numerical modelling during a repetitive validation of the computational model, but also a significant source of information for the selection of effective rehabilitation procedures for vaulted structures.The results are of major importance particularly for vaults of historic buildings situated in the seismically active regions or in places with intensive technical and induced seismicity (mining, stone quarrying, traffic).

Fig. 2 .
Fig. 2. -Vault collapse due to the supporting system's failure (horizontal displacement of supports)

Fig. 3 .
Fig. 3. -a) Scheme of vault K34non-reinforced, b) Scheme of vault K35reinforced at the extrados, c) Scheme of vault K36reinforced at the extrados and on the soffit, d) Test setup example Experimental tests dealt with the verification of dynamic properties of masonry barrel vaults for the following cases:  K34non-reinforced vault,  K35reinforced vault, reinforcement at the extrados (upper surface) with 2 strips of the Tyfo SCH 41 carbon fabric glued with the Tyfo S epoxy resin.The strips

Fig. 4 .
Fig. 4. -a) Pattern of vertical deformations δy × L of a non-reinforced vault (K34) under static loading (2x24 k N) in individual loading phases, b) Pattern of vertical deformations δy × L of a vault reinforced on the extrados (K35) under static loading (2x24 k N) in individual loading phases

Fig. 6 .
Fig. 6. -a) Mounting of the TIRAvib electrodynamic exciter on the vault, b) Distribution of sensors on the vault

Fig. 7 .Fig. 8 .Fig. 9 .
Fig. 7. -Oscillation shapes, theoretical and obtained from experiments, corresponding to the first natural frequency (see Tables 1 -3) Tab. 1. -Material characteristics of vault masonry . 2. -Phases of dynamic and static tests of segmented barrel vaults . -Experimentally identified deformation values and ultimate loading of segmented masonry barrel vaults Tab. 4. -Selected frequencies measured after different types of loading for vault K34_NZ Tab. 5. -Selected frequencies measured after different types of loading for vault K35_Z Legend: The test was interrupted for the reason of the electrodynamic exciter's failure.