Discussion and evaluation of a pilot project for computer-based design of low-rise buildings for wind loads

Recent research has pursued the development of a new generation of computer -based standard provisions for wind loading, using databases of aerodynamic time histories along with modern computer technologies to improve significantly the description of wind-induced design loads. A prototype computer application for the specification of wind loads on low -rise buildings has been created as part of a pilot project to assess the advantages of this approach, as well as identify areas for further research. Discussion of the procedures and programming of this application is presented. Preliminary research on the effectiveness of this application in improving the design of main wind-force resisting systems of low-rise buildings shows that this approach can identify risk-inconsistencies in designs, non-Gaussian load distributions, and other effects that impact system performance. cies with respect to internal bending moments in the frames (Whalen et al. 1998), and it has been employed to obtain calculation procedures consistent with the use of a performance criterion for structural collapse, based on the probability of exceeding a collapse wind speed (Simiu & Whalen 1998). Recently, the authors have extended this procedure into a more advanced application called WiLDE-LRS (Wind Load Design Environment for Low-Rise Structures). This application, developed using the software package MATLAB, adopts interactive graphic user interfaces to give a visual, user-friendly design environment. MATLAB scripts are used to perform the required calculations, reducing the time and potential for error associated with traditional code calculations. The application represents a prototype of a computer -based standard provision for wind loading that will ensure realistic, consistent, economical, and safe design in an easy to use system. To demonstrate the potential such a provision has for improving designs, this application was utilized to analyze the behavior of rigid portal frames in low-rise metal buildings under the action of high winds. Attention was paid to the windinduced load effects (bending moment, shear force, and axial thrust) in the frames. Several features of the induced load effects were studied, including the spatial distribution of peak load effects (both within a frame and across the building), their temporal correlation, and the non-Gaussian nature of their distribution in time. All of these effects can influence significantly the design of the frames, yet little of this information can be obtained using conventional standard provisions. This prototype thus illustrates some of the advantages that may be gained from a computer-based design standard. 2 DISCUSSION OF WILDE To better understand the analysis procedure employed in this study, a brief introduction to the application WiLDE-LRS is given here. This is not intended to be a comprehensive discussion of features and algorithms used in the application; such details can be obtained from other sources (Yang 1998, Whalen & Yang 1998). Instead, a general overview of the application, with an emphasis on notable features, is provided. The application is based upon a previously developed MATLAB program frameload used to study wind-induced load effects in single frames. WiLDELRS (or, more simply, WiLDE) is an extension of frameload which expands the analysis capabilities , increases its practicality for low-rise building design, and provides a simpler user environment. Three important features characterize WiLDE. The first is its database of wind pressure time histories acting on low-rise buildings. These were obtained from wind tunnel tests performed at the University of Western Ontario’s Boundary Layer Wind Tunnel Laboratory on scale models of typical buildings (Lin & Surry 1997). Two different scale models (1:100 and 1:200 scale) were utilized, and various wind exposures, wind speeds, and building geometries were simulated. For any given simulation, time-dependent pressure data were measured on the scale building over a series of wind directions moving around the building in increments of five degrees. Thus, a catalogue of realistic aer odynamic wind pressures is available, allowing for a detailed study of the effects of wind direction on frame behavior. It should be noted that a similar catalogue of wind pressures was used in the development of the pseudo-static pressure coefficients specified in current provisions (Davenport et al. 1987); what distinguishes the computer-based approach is the direct employment of the time-dependent pressure histories in the load specification. Another significant feature incorporated into WiLDE is the use of graphical user interface (GUI) windows for data entry and task selection. Such windows provide a direct, user -friendly method for specifying relevant building design information, choosing execution options for the application, viewing the results, and modifying inputs. Data entry is facilitated by the use of graphical displays and on-line help menus, while simple data formats improve the accessibility of the application to novice users. The GUI windows also allow for interactive editing of design information, permitting the user to explore the consequences of various design choices quickly and easily. While such capabilities are now commonplace in most engineering software, standard provisions applications have only recently started to make use of them to improve the design calculation process (Schechter et al. 1995). Ease of use is a crucial factor in the acceptance of any design standard. Thus, the simplifications offered by GUI-based applications make them an important part of a computer-based design standard. The final feature of note for WiLDE is its modular programming approach to the computation of relevant design quantities. In WiLDE, once the building design and wind environment information have been spec ified, a series of functions performs all of the requisite calculations needed to obtain the induced load effects at user-specified points in the frames. Tributary areas are computed, loads transferred to the girts and purlins and through to the frames are evaluated, and influence coefficients are determined within the application. This reduces the need for designers to employ simplifying assumptions in their evaluation procedures, resulting in a more accurate view of the behavior of their structure. The speed of the calculations is also greatly enhanced, allowing the designer to consider the influence of different frame properties quickly and efficiently. The modular programming approach, additionally, permits easy addition to or modification of these functions. This renders them amenable to a consensus approval process similar to that used to validate current standard provisions. These features, as well as those previously discussed, demonstrate the improvements that may be obtained, without loss of validity or consensus approval, when modern computer technologies are utilized in the design load specif ication pr ocess. 3 EVALUATION OF FRAME PERFORMANCE As discussed above , rigid portal frames are typically designed using pseudo-static wind pressures that produce enveloping cases of the maximum windinduced load effects to be resisted. The pressures used are intended to generate conservative values of the frame load effects, independent of the wind direction or exposure category (ASCE, 1993). It is to be expected that frame designs obtained via this procedure will contain excess load carrying capacity due to the conservative nature of the design loads. Moreover, it is not obvious that such large safety margins will actually result in decreased levels of failure risk, since this excess capacity is not necessarily deployed uniformly throughout all locations on all frames. In contrast, a computer-based design standard has the potential to overcome these drawbacks by allowing the designer to investigate the effect of dynamic wind pressures over a variety of directions and exposures, resulting in a more realistic picture of induced load effects in a given frame. In this way, unneeded frame capacity can be identified and eliminated, while still maintaining a uniform level of failure risk. Other phenomena which can be investigated via the use of a computer -based wind load standard include the temporal distribution of load effects at locations throughout the building, as well as the degree of simultaneity between peak values of these effects. Studying these effects allows the designer to better understand the probabilistic nature of the loads acting on the structure. Such information is needed in order to define properly the ultimate limit states of the structure as well as to provide a consistent framework for serviceability and performancebased design criteria. Increasingly, serviceability and performance-based design criteria are recognized as necessary components of a risk-consistent , multihazard approach to the design of high-risk structures and rehabilitation of existing ones (Chock et al., 1998). Thus, availability of probabilistic design information should take on greater importance in the future, and computer-based standard provisions utilizing aerodynamic and climatological databases are ideally suited to provide such information. Figure 1. Maximum values of wind-induced shear forces at selected locations in frames designed for building MO32 according to ASCE 7-93 (o) and WiLDE (+ = Frame #2; * = Frame #5 .) In the sections below, we will study the ability of the application WiLDE to provide the designer with improved information to evaluate the performance of the rigid frames. This will illustrate the potential of the computer-based design standard as well as assist in determin ing areas of needed research. 3.1 Identification of Risk -Inconsistent Designs To investigate the ability of this computer-based procedure to identify inefficient and riskinconsistent designs, peak load effect magnitudes as estimated by this procedure were compared to the values associated with use of the wind pressures specified by the ASCE 7-93 Standard. The application of this standard for comparison purposes is based upon its use by the metal building design firm responsible for providing the original design specifications for the frames used in this study. It is recognized that updated versions of this standard could be used to provide improved information; however, the purpose of this effort is to compare the effect of different design pressure specifications (current pseudo-static methods versus modern computerbased ones) on the resulting load effects. In Figure 1, we show the values of the maximum wind-induced shear force at sixteen positions on frames for building MO32 using the pressures specified by ASCE 7-93 and the aerodynamic databases of WiLDE. Two frames were studied using WiLDE: Frame #2 (near one end of the building) and Frame #5 (in the middle of the building).The shear force values identified by WiLDE are obtained by finding the sample maximum value for each wind direction in the database, then picking the largest value over all of the directional maxima. As expected, the values determined using ASCE 7-93 in most cases bound the sample maxima determined from WiLDE. (The instances where the sample maxima exceed the Figure 2. Ratios of sample maximum shear forces obtained using WiLDE to values obtained using ASCE 7-93. (+ = Frame