|NDT.net Apr 2006 Vol. 11 No.4|
Monitoring Cumulative Damage in Shear Wall Testing with Acoustic EmissionFrank C. Beall
University of California, Berkeley email@example.com Jian Li
firstname.lastname@example.org and Thomas A. Breiner
14th International Symposium on Nondestructive Testing of Wood
AbstractMonotonic testing is performed as the initial step for cyclic loading to evaluate potential seismic performance of shear walls. We have been testing a wide range of configuration and aging effects on 1.2- by 1.2-m assemblies. One of the major challenges is that the assessment of damage must be done by measuring energy dissipation or some less quantitative technique. To provide an additional input of damage development, we attached four AE transducers, a pair at each upper corner, on both the frame and panel. The protocol that we used was a modified "felicity ratio," in which the actuator pushes the assembly through a sequence of predetermined limits and releases the force, with about four cycles per monotonic test. The AE showed a definite change in the felicity ratio with load, providing an independent measure of damage. Also, the pattern of AE was quite different for both framing and panels (plywood and OSB) and for materials in tension and compression. Based on this information, we feel that the AE technique can be a useful complimentary measure for either open- or closed-loop control in both monotonic and cyclical testing of shear wall assemblies.
BackgroundThe conventional approach in simulated cyclic seismic testing is to run monotonic tests to obtain the load-deformation curve and from that curve obtain a reference deflection (־r) that is used to establish a baseline deformation (Figure 1). From the maximum load (Pmax), the deformation at 0.8 Pmax (deformation capacity, ־m ) is obtained (the point of maximum deformation can also be determined by consensus). The reference deformation (־r) is calculated as 0.6 ־m . Both factors (0.8 and 0.6) are completely arbitrary, being largely based on seismic testing of other materials, such as concrete. Typically, the shape of the curves for matched assemblies can be similar up to Pmax, and then very divergent beyond that point. Where damage occurs and how damage develops in the assembly is unclear.
One means of assessing damage development in materials is to monitor acoustic emission (AE). AE is produced in materials under stress and is usually sensed with piezoelectric transducers coupled physically to the surface of the material, and has been used with wood-based materials since the 1960s. The application of AE to wood-based materials has been substantially aided by developments in assessing the integrity of fiber-reinforced plastics (FRP), which have many characteristics similar to those of wood. FRP, like wood, requires special techniques to detect internal flaws, in contrast to metals where x-ray techniques are particularly useful. AE gives indications of weaker (or weakened) material or structures by emissions at low stress levels, and increased numbers and/or rates of events at higher stress levels. Another indicator of the state of a material can be determined from a sequence of load/unload cycles during which AE is monitored. In undamaged material, no AE occurs during reloading until the previous stress level is reached, giving a load ratio of unity, referred to as the Kaiser Effect. However, many materials, such as composites, have inherent flaws, producing a ratio of less than unity, generally decreasing with increasing stress levels, which is termed the Felicity Effect. Composite materials, including wood, typically exhibits the Felicity Effect . The Felicity Effect can be expressed by the Felicity Ratio, the fraction of the applied load at which the acoustic emission reappears during the next application of loading.
Experimental ApproachThe assemblies included plywood (nominal 12 mm Structural-1 Douglas-fir) and OSB (nominal 12 mm Exposure-1 sheathing grade). Each assembly had 400 mm stud separation, and 150 mm edge-nail (8d coated) and 300 mm field-nail (8d coated) spacing. The framing material was 2x4 kiln-dried Douglas-fir Select Structural grade to minimize interference from defects and moisture content changes. Single framing members were used for all edges with 16d bright framing nails.
The assemblies were tested using an MTS 407 controller with a 25-kN actuator mounted on a 0.9- x 3.6-m 10-t table (Figure 2). The test system was designed to be self-reacting since the floor at the Forest Products Laboratory was inadequately reinforced. The actuator drives an upper loading head that is stabilized laterally with guide rollers, and has a special fixture that permits horizontal and vertical movement, but prevents rotation in the plane of the assembly. The system is programmed for both monotonic and dynamic testing, and for automatic data acquisition. Monotonic and Felicity tests were run under deformation control at 0.5 mm/s, the standard rate used for previous tests on these assemblies .
Two panel transducers were attached with 27 N springs at a point 150 mm diagonally from each upper corner using a special spring holder. For the OSB, the panels were oriented to permit attachment to the smooth side. Two framing transducers were mounted similarly on the vertical framing member at 25 mm from the upper frame end. A high viscosity grease was used as a couplant. An AET5500 AE system was operated at 0.5 V automatic threshold and 80 dB total gain, which included 60 dB preamplification, and a preamp filter of 125-250 kHz. The transducers were 175 kHz. The AET5500 data file was synchronized with the load output from the MTS controller. AE events with only one count were discarded as noise, according to accepted practice.
Monotonic runs were performed without interruption to obtain overall AE data. All of the runs were ended when the load fell to 0.8 Pmax, which has been the operating criterion for previous monotonic runs. After this point (and often prior to this), the load-displacement behavior becomes erratic from cumulative damage. A series of three monotonic tests were done (in duplicate) with both plywood and OSB, using a semi-rigid holddown , and an essentially pinned holddown, which was a modified version of that reported by Herbert and King . Since the rigid holddown caused some bending of the vertical studs, it was believed that the pinned holddown would cause a more uniform distribution of stresses in both the frames and panels. In addition, the surface roughness was reduced by coating the transducer contact area of the panels with a wood filler (DAP cellulose fiber filler) and sanding it to provide a surface without voids. In both coupling types, a viscous grease was used for coupling together with a 27-N spring.
Felicity runs were made by choosing nominal load fractions of previous monotonic runs for the plywood and OSB assemblies, beginning at about 35% of Pmax. For each cycle, the load was reduced to about 0 to 10% of the previous peak value before increasing the load at the same rate. The actuator yoke attachment was disconnected to permit the assembly to recover deformation without the actuator pulling it back.
Results and Discussion
Monotonic testsFigures 3 and 4 show the AE events and cumulative AE as the load was applied monotonically without interruption. Fig 3a shows the output of a semi-rigid holddown plywood assembly with no surface treatment of the panel; Fig 3b is a similar assembly, but with OSB. Fig 3c is a plywood panel with a pinned holddown and cellulosic fiber surface filling. The large AE peak in Fig 3a was a result of a split in the upper portion of the left frame member, which is a common occurrence in such testing. Since this was near Pmax, it did not greatly affect the usefulness of the AE data. The OSB matched specimen (Fig 3b and 4b), had relatively similar behavior of AE from each of the vertical frames, but substantially lower AE from the panels. This latter effect was unexpected since OSB was anticipated to be a much "noisier" material based on its structure and creep testing . Also, it was found that the right frame (tension member) was consistently higher in AE output. The assembly with the pinned holddown and surface filling (Fig 3c and 4c) showed a higher output of AE from the panel transducers and more clustering of panel and frame outputs. There was little effect of pinned vs rigid holddowns. The reasons for the differences between plywood and OSB could not be determined from these tests, but OSB is less stiff than plywood and is much more susceptible to creep. OSB panel creep could be assessed by using different rates of deformation.
There were several consistent features:
In all runs, there were few or no AE events up to 0.4 Pmax, which is regarded as the endpoint for determining initial stiffness in the monotonic tests. Although this portion of the curve is not truly linear, the absence of AE indicates that the deformation is reversible and therefore can be considered to validate the practice.
Up to about Pmax, the shape of the AE event curves was the reverse of the load-deflection curve (concave vs convex). The net effect is that about one-half of the AE events occurred up to Pmax.
The Right Frame transducer (at the load application corner) in general had the highest level of AE events, and the Right Panel transducer the lowest level. This was unexpected, since it was anticipated that there would be more damage associated with the fractures in panels from movement of the nails. However, it is possible that the nails acted as waveguides to couple the AE events into the framing, which would have had lower attenuation than the panels.
OSB panels typically had lower AE events than plywood. This was attributed to the much rougher surface of OSB compared with plywood, but it could have also been caused by a greater scattering of waves in the OSB or more brittleness of the plywood.
The most consistent observation is that the AE event rates increase rather uniformly up to Pmax, and then become fairly erratic as damage develops in the assemblies.
One feature that did not occur in the AE outputs was an exponential increase in AE as the system approached failure, which is characteristic of much AE testing.
Felicity testsBased on the monotonic results, the decision was made to use the right frame transducer output for determining the Felicity Ratios for the Felicity Effect tests. Fig 5a and 5b show the outputs for a typical OSB panel assembly. Although it is difficult to see in the figures, the right frame transducer was typically the first to sense AE in the reapplication of load. this was visually determined by expanding the time scale for the load reapplication periods.
Figure 6 is a plot of the calculated Felicity Ratios for the 6 runs There were two general patterns, a somewhat linear decrease for assemblies with higher Pmax, and very abrupt decreases for those with lower Pmax. The patterns of FR values over the first three cycles were not good predictors of the subsequent FR values or Pmax. However, the three higher Pmax assemblies (16.1 to 18.0 kN) were consistently above 0.7 over the complete deformation cycle, whereas the lower three (14.0 to 15.4 kN) had terminal values below 0.6. for these two groups, the terminal FR averaged 0.77 and 0.50, respectively. These values were obtained from the "tension" frame member prior to the improvement of coupling with wood filler on the surface of the panels. It is possible that better information could be obtained with the improved coupling, and that the choice of the specific point for obtaining Felicity information could be affected.
A number of these plots show increases in FR after the initial decreases, and occur largely in the middle range of the load application. This was not expected, since typically FR values show a down trend from the extension of fractures that occur earlier. However, the increases could reflect the influence of fractures that terminate and no longer produce AE, at least within that particular cycle of load-unload.
Conclusions and Recommendations