Kaiser effect in historic timber elements

The Prosperi-Sacrati palace is one of the buildings located at the famous intersection of the principal streets of the so-called “Herculean Addition” built in Ferrara, Italy, by Ercole I d’Este at the end of the 15th C. (see Fig. 1); the most famous Palazzo dei Diamanti is also located at the same intersection. The Prosperi-Sacrati building was completed in 1513–1514 but its façade was altered in 1857 and the introduction of new roof beams can be traced back to this date. A limited number of these were removed in 2006, due to a significant decay which had occurred at the support ends in contact with the masonry walls. Some of the discharged beams were recovered by the University of Bologna. A dismounted beam, identified as spruce on the basis of visual inspection and existing historic records, with a bevelled-edges rectangular cross-section of about 25 × 29 cm² and length of approximately 6 m, has been used in this laboratory experimental work.

The visual inspection of the beam’s cross-section revealed also the age of the fallen tree to be more than 90 years and its original arrangement on the roof could be clearly deduced from the degradation state of the surface in contact with the masonry supports. From the central part of the beam, four smaller specimens, about 2.6 m in length, were obtained by cutting the original element along the two symmetry axes of the cross-section, see Fig. 2. Bending tests coupled with AE monitoring were performed for three of the four specimens (1, 3 and 4) in 2010 and for two specimens (3 and 4) in 2011.

The laboratory study was carried out in Bologna University by monitoring AE during cyclic loading of specimens 1, 3 and 4 (Fig. 2) in 4-point bending tests in the elastic range. The testing was done in two separate sessions: February 2010 and March 2011. During the tests, a well-defined and repeatable load pattern in bending mode—the type of deformation commonly present in roof constructions—was selected. The load was applied using a single hydraulic jack, a stiff steel beam and two hinge devices transferring in two points the total force from the actuator. The total length of each beam was 258 cm but the clear span was only120 cm and the distance between the loading points was 40 cm (see Fig. 3a). The relatively short span facilitated the AE measurements in which a single analyser collected the AE signal from two sensors located in the middle and at the end of the clear span. A short distance between the sensors allowed the length of cables and therefore possible noise generated in the measuring system to be reduced. The displacement at mid-span was monitored near the lateral faces by two LVDT sensors with 50 mm range (see Fig. 3b). The test speed in terms of the imposed displacement, monitored in real time, was kept almost constant in all test sessions at values close to 1 mm/min. The data acquisition was set at 2 data/sec.

The patterns of loading used in years 2010 and 2011 for beams no. 1, 3 and 4 are summarized in Table 1. The stress acting on the beam mid-span that is subjected to the bending moment is evaluated as flexural stress: where F is the total load applied, a is the length of the loading span, b and h—width and height of the beam, respectively (Fig. 3). According to European standards strength class system (for example EN 338), the characteristic strength value in bending for high quality structural timber is 40 MPa. Values of maximum stresses reached in the applied 4-point bending tests (Table 1) clearly indicate that all loading cycles were performed in the elastic region of wood’s deformation.

In the measuring campaigns summarised in Table 1, the effect of a load exceeding the previous level was analysed in two fast loading cycles (5 cycles lasted less than 0.5 h) applied at a time interval of 1 year (i.e. in February 2010 and March 2011). To make the study consistent and reliable, care was taken to ensure that the experimental procedure during both measurements was exactly the same. It is worth adding that the specimens were kept unloaded between the two measurements in non-air-conditioned laboratory space with temperature and relative humidity varying during the year in the range of 18–30 °C and 20–60 %, respectively.

During all tests (in 2010 and 2011), the settings of the AE monitoring system, that is gain, frequency filters, sampling rate and threshold level, were the same.

AE signals were measured by two (FD26 and FD27) wideband multiresonant piezoelectric sensors (Physical Acoustic Corporation) located, in each experiment, in the middle of the beam and close to the position where load was applied (Fig. 3b). This second sensor was used to double check the operation of the measuring system and to evaluate possible emissions resulting from the pressure of the supporting hinge on the wood during the bending test. The emission from both the sensors was highly correlated proving that noise resulting from processes different than the fracturing of the wood structure did not disturb the measurements. The AE experimental setup consisted further of an acoustic amplifier (EA System) with variable gain and selectable high-pass frequency filters (5, 20 and 100 kHz). The AE signals were digitized by a simultaneous-sampling analogue input card PCI-9812 (Adlink Technology Inc.) and stored in a personal computer used also for post-processing the acquired data. The recorded signals were sampled with the rate of 1 MHz. A single data set consisted of 128 000 points corresponding to 128 ms time-windows, while the duration of a typical AE event was of the order of hundreds of μs. The dead time between two consecutive data sets was found to be around 100 ms. The storage was triggered when the electrical signal produced by the AE event first crossed the threshold voltage of 0.6 V (56 dB) which was twofold that of the continuous background noise (0.3 V corresponding to 50 dB). It was also verified in supplementary tests that the testing machine did not generate AE signals when operating without any specimen installed.

The analysing software was designed to search for individual AE events, extract them and calculate the most important AE features, that is, amplitude, energy, duration and frequency distribution.

The aim of the study performed in February 2010 was to investigate a short-term Kaiser effect in construction elements made of historic wood by analysing the load curves and the AE data in a 4-point bending experiment with consecutively increased load. The second measuring campaign in March 2011, carried out on the same specimens, sought to establish if historic timber retained memory of the loads imposed during the February 2010 tests, i.e. if a long-term Kaiser effect (the memory of loading exerted 1 year before the tests) was detected. As the stress experienced by the specimen is highest at the mid-span, emissions recorded at this location were taken as the indicator of progressive damage.