Testing Shear Walls for Cyclone Resistance

Structural engineers design buildings to safely resist gravity loads (vertical loads), such as the self-weight of the building and the weight of people and furniture, as well as loads from extremely rare events, such as earthquakes and cyclones (lateral loads - image below).  Gravity loads are transferred down through the building to the foundations via beams, columns and load-bearing walls.  Lateral loads are transferred through the structure via a series of interconnected horizontal diaphragms (roof and suspended floors) and vertical shear walls (see image above).

Over the next few years I will be developing and testing shear wall systems for tall timber buildings, so I am interested in finding the best methods for laboratory testing of timber shear walls.  My early research on this topic has uncovered some practices which I find odd.

In an ideal world we would construct shear wall prototypes to match practices on building sites and then test them under conditions that mimic, as close as possible, the loads that would occur in an actual earthquake or cyclone.  To properly test the wall, we need to know:

1. How the wall is fixed to the substructure (i.e., timber floor or concrete foundation);
2. How the wall is fixed to the superstructure (i.e., timber floor or roof framing);
3. How the lateral loads are transmitted into the shear wall; and
4. The details of the governing load combination:
• Do earthquake or wind loads govern?
• What are the gravity loads on the wall?
• Are there any tension loads due to wind uplift and / or overturning?
• Are there any out-of-plane loads acting on the wall? 

In my reading of the literature on this topic, I am left wondering whether these questions have been given much thought by researchers.

It might be worthwhile to present a quick overview of the history of timber shear wall testing.  Test methods for shear walls have evolved over the years.  In the early years of testing, hydraulic jacks were pumped by hand up to the design load, sustained at that load for five minutes, released to zero and then tested to failure (see image below).  This method is still accepted in the current Australian Standard for Timber Structures (AS 1720.1 - 2010, Appendix D) and the European Standard (BS EN 594:2011) even though other standards around the world have moved on to using cyclic loading protocols (e.g., International Standard ISO 21581:2010 and American Standards ASTM E2126 and AWC SDPWS 2015).

Curiously, earthquake loads, not wind loads, appear to be the key concern driving the move to cyclic load testing of timber shear walls.  Most of the research on timber shear walls seems to be presented at earthquake engineering conferences and published in earthquake engineering journals.  I find this 'seismic' bias strange because timber buildings are known to be more resilient to earthquake loads than they are to cyclonic loads (Walker, 2009).  Don't get me wrong, earthquake loads can be quite severe and they do cause serious damage to timber homes, but the load profile of an earthquake is relatively short (i.e., seconds) in comparison to that of a cyclone (i.e., hours).  Timber houses have been known to survive earthquakes in good structural condition after being shaken off their foundations (see image below).  Cyclones and tornadoes, on the other hand, have been far more devastating to timber buildings.  I honestly expected to see the focus on timber shear walls coming from a wind engineering perspective.

From the viewpoint of a shear wall, there are perhaps two key differences between earthquake loads and wind loads.  First of all, cyclones and hurricanes last much longer than earthquakes.  A strong earthquake typically lasts between 10 and 30 seconds, whereas the destructive winds of a cyclone can buffet a building for hours.  Secondly, the shaking of an earthquake is cyclic in nature, whereas the wind from a cyclone has a prevailing direction as the cyclone approaches and, after the eye of the storm passes by, the direction of the wind changes.  All of the methods I have reviewed to date for repeated load testing of timber shear walls, except for one notable exception, mimic a seismic load rather than a wind load both in terms of the duration of the test and in terms of the cyclic nature of the applied load (see image below from Kirkham, Gupta & Miller, 2014).  

If we want to mimic wind loads instead of earthquake loads, it would be preferable to extend the duration of the test and apply buffeting loads to the prototype shear wall first in one direction and then in the other direction.  I have only been able to find one test protocol which matches these requirements.  The Cyclone Testing Station (CTS) in Townsville developed a repeated load testing protocol in 1980 for shear walls which is shown in the table below (Reardon, 1980).  In this table, 'D' is the unfactored design load of the shear wall.  The method calls for 1021 loading cycles, each lasting at least 3 seconds, for each shear wall that is tested.  The minimum total time for each test in this regime is 51 min plus the time taken for the final sequence up to the ultimate load.  The CTS used this loading regime in subsequent tests (e.g., Boughton, 1984); however, I can only find one other published example of a similar test protocol being used, albeit with just 15 cycles of phase 3A (Songlai, Chengmou & Jinglong, 2010).  I wonder whether the CTS protocol is too time consuming to be a popular choice?

The next strange thing I have noticed is the connection between the actuator and the shear wall.  The most common method, by far, of connecting the actuator to the shear wall is by bolting a spreader bar directly onto the top plates of the frame (see below).  I don't believe this is an accurate way of representing the way that lateral loads are actually transmitted into the wall.  Wind and earthquake loads are transmitted into shear walls by means of connections between structural members.  If the structural members are fixed to the wall frame directly, then the detail below is acceptable.  On the other hand, if the fixings are connected to the frame through the sheathing then it would be more appropriate to ensure that the connections are likewise fixed to the frame through the sheathing.  This type of connection will ensure a more effective transfer of load into the sheathing.

There are surprisingly few studies combining uplift loads (i.e., vertical tension) with lateral loads.  I'm surprised by this because it is common knowledge that wind events cause uplift on the roof, some of which will inevitably be transmitted into shear walls.  For taller buildings, load combinations causing overturning will cause vertical tension in shear walls in addition to the shear load.  Because shear wall failures tend to occur in the tension regions of the wall (either by hold-down failure, failure of the sheathing fasteners or failure of the sheathing) not the compression regions, we should expect that uplift loads will likely reduce the capacity of the wall.  If this reasoning is correct, then test methods without uplift loads might not be sufficiently conservative.

There are, no doubt, many other issues I will have to consider as I work on this project.  For now, I'll wrap this blog post up with a shout out to some great work at the Cyclone Testing Station in the last couple of years focused on using data from wind tunnel tests of scaled building models to develop more realistic loading patterns which are then applied to a full-scale partial structure (Boughton et al., 2017).  Results from these tests have shown that non-structural components, such as gypsum cladding and corniced ceiling, take 40% of the applied lateral load (Satheeskumar et al., 2016).  Experience tells me to be wary of using non-structural components under Ultimate Limit State design, but it seems overly conservative to ignore them under lower loads (say, unfactored loads).  I wonder whether these findings can be used to refine the CTS loading protocol for shear walls?  Hmmm.

Boughton, GN. (1984). Simulated High Winds Tests on a Timber Framed House. Paper presented at The Engineering Conference, 1984. The Institution of Engineers, Australia: Barton ACT.

Boughton, GN, Parackal, K, Satheeskumar, N, & Henderson, DJ. (2017). Development of a Full-Scale Structural Testing Program to Evaluate the Resistance of Australian Houses to Wind Loads. Frontiers in Built Environment, 3, 21.

Kirkham, WJ, Gupta, R, & Miller, TH. (2014). State of the Art: Seismic Behavior of Wood-Frame Residential Structures. Journal of Structural Engineering, 140(4), 04013097.

Reardon, GF. (1980).  Recommendations for the Testing of Roofs and Walls to Resist High Wind Loads: Technical Report No. 5. James Cook University: Townsville, QLD.

Satheeskumar, N, Henderson, DJ, Ginger, JD, Humphreys, MT, & Wang, CH. (2016). Load Sharing and Structural Response of Roof-Wall System in a Timber Framed House. Engineering Structures, 122, 310-322.

Songlai, C, Chengmou, F, & Jinglong, P. (2010). Experimental Study on Full-Scale Light Frame Wood House under Lateral Load. Journal of Structural Engineering, 136(7), 805-812.

Walker, G. (2009). Comparison of the Impacts of Cyclone Tracy and the Newcastle Earthquake on the Australian Building and Insurance Industries. Paper presented at the 2009 Australian Earthquake Engineering Conference. Australian Earthquake Engineering Society: Newcastle, NSW.