Frequency Domain Analyses of the Dynamic Response of an Engineered Wood I-Joist Floor System

Open Access
- Author:
- Worley, Stacy Kenneth
- Graduate Program:
- Agricultural and Biological Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- July 02, 2004
- Committee Members:
- Harvey Bright Manbeck, Committee Chair/Co-Chair
John Jack Janowiak, Committee Member
Virendra Puri, Committee Member
Joseph Irudayaraj, Committee Member - Keywords:
- dynamic testing
dynamic analysis
signal processing
wood structures
I-joist
structural vibration - Abstract:
- Engineered wood product (EWP) framed structural systems are a viable alternative to conventional, solid-sawn lumber framed systems. However, a better understanding of the dynamic structural behavior of these products is necessary if optimal utilization of their structural characteristics is to be realized. An effective model of floor system dynamic response would provide an important tool for effective and efficient serviceability criteria. Such criteria are critical tools for engineers to develop floor system designs with acceptable dynamic performance. However, models rely on experimental data, and, as such, rely on the quality and precision of that experimental data. The development of tools to provide high quality data input to these methods, in the form of careful data acquisition system design and signal processing technique development and implementation, is a first step in and essential to the effective and efficient implementation of these model development techniques. The objectives of this study were to: (1) accurately measure and record the vertical and horizontal responses of an I-joist based wood floor system, (2) utilize existing or develop new signal processing techniques to maximize the quality of the experimental data, (3) perform both time and frequency domain modal analyses, and (4) experimentally evaluate the lateral bracing effects of overall system response. An experimental program evaluated the free response of four I-joist floor specimens, each without any lateral bracing applied to the joists and each with three levels of lateral bracing, to impact loading. Impact force and accelerations at the joist bottom flanges were measured in two directions. Measurements were taken at the center point of the floor, the ¼-points of the center joist, and midspan of an outboard joist. Measurements were recorded and processed using a package of tools assembled from existing techniques with the inclusion of a novel data set selection technique. This technique reduced the data set size and reduced noise levels without altering the data set and without prior knowledge of system response. Bias correction via the moving average, normalization, and zero padding were used to optimize the input file for the frequency domain analysis. The frequency domain analysis was based on the frequency response function (FRF) and estimates of natural frequency, corresponding modal magnitude, damping, and response linearity were determined. Inspection of graphic data, simple statistics, and ANOVA were used to discover patterns and correlations in the data. The frequency domain analysis did not rise to the level of a modal analysis and the time domain modal analysis was not implemented. Despite considerable effort, convergence was never achieved with the modal analyses. The lack of convergence may likely be due to the higher than expected levels of floor system damping and non-linearity. The effectiveness of lateral bracing at the joist mid-span was also evaluated. Four bracing conditions: baseline (no bracing added), solid blocking, tension strap, and solid blocking with a tension strap were examined. The average experimental first vertical natural frequency of the floor was 24.8 Hz with a high degree of confidence for all bracing treatments. Of the vertical response, only the magnitude associated with the first natural frequency was dependent on test factors, namely the measurement location. For the horizontal responses, both first natural frequency and the corresponding magnitude were strongly dependent on bracing treatment. No apparent link between the horizontal and vertical responses was found in the frequency and magnitude data. However, the nonlinearity of the vertical response increased as the bracing level increased. Damping was evaluated for the first vertical natural frequency and found to range between 3 and 6 percent. Linearity was tested. In all test cases, the response from ten repetitions of the same test varied from the mean response by more than ±2.5 percent of the mean. Overall, 48 test cases varied from the mean by ±2.5 to ±5.0 percent of the mean and 80 tests varied more than ±5.0 percent from the mean. This level of variability in the response with respect to excitation arguably violates the linearity assumption of normalization, the FFT, and the FRF. The level of damping is also near the limit of what can be classified as lightly damped. Based on the results of this study, the following conclusions were drawn: (1) the test procedures and the experimental apparatus employed in this study were adequate and provided high quality initial data, (2)the signal conditioning modules were effective in reducing errors and otherwise improving the quality of data sets, (3)the novel data set selection module successfully selected appropriate data subsets while reducing the noise content without distorting the frequency or magnitude of the signal and with no resolution limitations, (4) the observed lateral response of the I-joists to impact excitation was significant (p<0.05) in magnitude, (5) There was no discernable pattern in the spacing between the first and second vertical natural frequencies, but the spacing between the first and second horizontal natural frequencies increased with the addition of bracing, (6) there was a weak interaction between the vertical response and lateral bracing treatments as indicated by the increase in non-linearity of the vertical response with the addition of bracing Further research into the impact of lateral responses on overall floor system response is warranted. Future efforts to develop an experimental model of system response must consider non-linear and more heavily damped response.