Experimental study on the quasi-static progressive collapse response of post-and-beam mass timber buildings under corner column removal scenarios

Abstract

This study presents the results from two quasi-static experimental tests performed on a ¼-scale, 2 × 2-bay, post-and-beam mass timber substructure, with Cross Laminated Timber (CLT) floors and Laminated Veneer Lumber (LVL) frames, subjected to corner column removal scenarios. One substructure was constructed and tested twice under different CLT panel configurations. In the first test (Test CM-1), the bay adjacent to the removed corner column consisted of three CLT floor panels, with two panels spanning one bay and one panel extending past the bay and spanning two bays. A constant Uniformly Distributed Pressure (UDP) was first applied to the bays not adjacent to the removed column and the UDP on the bay adjacent to the removed column was increased till failure through a six-point loading tree. As failure did not extend past the vicinity of the removed column in CM-1, the damaged elements were replaced and the specimen was re-tested by removing the diagonally opposite corner column. In this second test (Test CM-2), the bay adjacent to the removed corner column now consisted of one CLT floor panel spanning one bay and two panels spanning two bays. A similar loading protocol to that of Test CM-1 was used in CM-2. The substructure was assembled from a commonly used commercial beam-to-column connection. The failure modes, load redistribution mechanisms (alternative load paths) and overall behaviour are presented in this paper. The two test results showed that the applied load was mainly transferred to the three columns, the closest to the removed column. The ultimate load in Test CM-2 was 1.62 times higher than Test CM-1, demonstrating that the layout of the CLT panels plays a critical role against progressive collapse. A simplified analytical model, consistent with the current industry practice and pre-defined alternative load paths, was developed for the tested specimen to predict the ultimate resistance capacity and was also compared to the experimental tests. On average, the experimental failure load was 2.7 times higher than the predicted one, indicating that simplified current design practice is conservative.