Displacement-based seismic design often requires nonlinear time history simulation of building responses which is computationally intensive. This technical note presents an artificial neural network (ANN) designed to generate max inter-story drift for buildings with post-tensioned mass timber (MT) rocking walls systems. This particular lateral system was selected because it is an innovative system with limited physical design parameters, making it an ideal candidate for ANN. The proposed model achieved significantly higher computational efficiency than time history simulation, while maintaining similar level of accuracy. The proposed method could potentially be used for automated design of MT rocking wall lateral systems.

The post-tensioned cross-laminated timber (CLT) rocking wall is a recently developed resilient CLT lateral force-resisting system with a self-centering feature. The structural responses of the systems with different designs need to be determined and evaluated efficiently to promote the development and standardization of industrial applications. This study developed a computationally efficient, component-assembled numerical model for post-tensioned cross-laminated timber (PT CLT) rocking walls that captures decompression, post-tension self-centering, and energy dissipation within a framework. The single wall model was assembled using nonlinear zero-length springs for the compression at the CLT bottom, truss bar element for the PT tendon, and elastic shell element for the CLT panel deformation. The energy dissipation device, the UFP, was modeled with nonlinear one-dimensional springs between the wall panels in the coupled wall model. The wall models were separately calibrated considering the wall designs of single-panel walls and coupled walls. Both single and coupled wall models predicted the initial stiffness, decompression, yielding, post-yield stiffness, and reloading/unloading stiffness. The residual drift and nonlinear unloading captured with the PT model were also validated with the test data. A two-story platform structure model was established based on the NHERI Tallwood project, assembled with the coupled wall model and CLT slab in shell elements and columns in Euler beam elements. With recorded ground acceleration signals from the test, the platform structure’s peak story displacement and inter-story drift were simulated with less than 30% differences for most cases. Unlike existing detailed contact-based models, the proposed approach balances local damage fidelity and computational efficiency. The validated model provides a framework for evaluating PT CLT wall design parameters considering their influence on full structures.