The building footprint was constrained to a size that could be supported by the shake table platen, which is $7.6\times 12.2\mathrm{m}$ ($25\times 40\mathrm{ft}$). While there is no height limit, the height of the building was constrained to be within a reasonable proportion to the floor plan dimensions. Based on these physical constraints, the test building floor plan and elevation were sized as shown in Fig. 2 (the $X\text{–}Y$ axis of the shake table facility is also shown). A single floor area was $84{\mathrm{m}}^2$ ($900{\mathrm{ft}}^2$) and was uniform over the building’s height. The total building height was 34.4 m (113 ft). The column grid consisted of 14 gravity columns arranged within the space limitations of the shake table surface. Cantilevered precast concrete foundation blocks were used to expand the surface of the shake table so that the building’s first floor plan included cantilevered patio portions toward the north and south directions beyond gridlines A and D, which supported four different types of exterior nonstructural façades.

During the design process, the gravity frame members (LVL beams and columns) were sized first using a typical loading scenario for office space [i.e., $3.3\mathrm{kN}/{\mathrm{m}}^2$ (70 psf)] for dead and $3.1\mathrm{kN}/{\mathrm{m}}^2$ (65 psf) for live loads. The columns and beams were sized for a 2 h fire rating based on fire design requirements from the National Design Specification for Wood Construction (NDS, AWC 2018), which added about 50 mm (2 in.) of thickness for mass timber members on all exposed sides. Fire design was specifically implemented for the test building to represent realistic column and beam sizes for connection detailing, as many wood beams and columns will be exposed in such buildings. A uniform cross section was used for beams and columns to simplify fabrication and column splice design (member sizes could vary in a real building project for cost savings). Similarly, structural strength or stiffness was not the controlling criteria for the selection of the floor panel thickness in this study. As shown in Fig. 3, multiple types of mass timber products were used for the floor diaphragms to test the feasibility of different products. Five-ply CLT panels were selected for Floors 2 and 3 to provide adequate fire resistance of a diaphragm with an exposed ceiling. On the remaining floors, other mass timber floor systems were selected to have similar thicknesses as the CLT floors for similar fire resistance and detailing. The bounding columns, located on each side of the rocking walls, had larger cross-sections than typical columns to resist axial forces generated from the steel yielding energy dissipation devices [U-shaped flexure plates or UFPs were used in this test for its cost-effectiveness and reliability, but other type of energy dissipation devices could also be used (Skinner et al. 1974)]. The gravity and lateral system materials and dimensions are summarized in Table 1 and illustrated in Fig. 3, including the four mass timber rocking walls shown in exploded view for clarity.

To achieve damage-free performance under repetitive earthquake tests, connections for the gravity system in the test building were specially designed by an industry partner to allow free rotation up to 5% interstory drift while providing adequate axial stiffness and strength for column bracing (Pryor et al. 2024). Examples of these connection details can be found in Fig. 4(d), with detailed structural drawings available in Busch (2023). The column base was detailed as a true pin to allow biaxial rotation [see Fig. 4(a)]. While such a column base would not be required even in a resilient mass timber building, it was necessary for this study to ensure the columns were not damaged for many phases of testing by multiple research teams. The column splices were located at the floor levels and were designed for axial loads from gravity. Bounding column splices were also designed for a net tension force demand generated from a subsequent project to NHERI Tallwood, which reused the 10-story specimen. The beam-to-column connection used a beam hanger with slotted holes that transferred shear but permitted rotation about a top bolt to minimize moment transfer. The beam hanger was embedded within the wood for fire protection of the steel components.

The design of the building’s lateral force-resisting system was intended to limit structural and nonstructural component damage by controlling the drift of the building. An in-depth description of the design of the lateral force resisting system can be found in Wichman (2023) from preliminary design to performance verification using nonlinear response history analysis as well as the calculation of demand-to-capacity ratios for all limit states of the walls and connections. A brief summary is provided in the following.

The building was designed for a location in Seattle, Washington, on Site Class C soil. The ATC hazards by location tool (ATC 2021) was used to generate short period ($S_{MS}$) and 1 s ($S_{M1}$) risk targeted maximum considered earthquake (${\mathrm{MCE}}_{\mathrm{R}}$) spectral acceleration values of 1.65 and 0.72 g, respectively. Uniform hazard spectra (UHS) were also generated for several different hazard levels (i.e., return periods [RP]), as shown in Fig. 5. The seismic hazard in Seattle is unique in that there are significant contributions from crustal faults and the Cascadia Subduction Zone (producing interplate and intraplate earthquakes). The UHS were developed using the USGS 2014 US earthquake source model (USGS 2017a) and generated for a site with a time-averaged shear-wave velocity to 30 m depth ($V_{S30}$) equal to $500\mathrm{m}/\mathrm{s}$ (using USGS 2017b). The site-specific ${\mathrm{MCE}}_{\mathrm{R}}$ spectrum was also developed per ASCE/SEI 7-16 Section 21.2.1.1 (ASCE 2016) and is as shown in Fig. 5. The target vertical acceleration spectra were developed in accordance with Section 11.9 of ASCE/SEI 7-16, which is typically used for defining the ${\mathrm{MCE}}_{\mathrm{R}}$ vertical spectra. For other hazard levels, that same procedure was used but with $S_{MS}$ replaced by the 0.1 s horizontal spectral acceleration from the various UHS given in Fig. 5.

With the hazard levels defined, desired maximum damage states for the structural systems and connections were selected for each hazard level and then the maximum drift limits corresponding to the damage states were assigned. The maximum drifts corresponding to each damage state were determined based on experimental data (Wichman et al. 2022), detailed analyses of specific components (e.g., the gravity frame connections), or engineering judgment from past wood building fragilities. The drift targets also served as a key design constraint (2.5% limit was used in the design stage) for the nonstructural systems, including the façades, interior walls, and stairs, which are described in later sections and detailed in Roser et al. (2024), Wynn et al. (2024), Ji et al. (2024), and Sorosh et al. (2024). While it is the first time most of these nonstructural systems were tested under 3D motions in a full-scale building setting, the research team collaborated with experienced industry partners to meet these target damage states (which were later validated). The design objectives for the test building at different hazard levels are summarized in Table 2. Note the no-damage designation for energy dissipation elements corresponding to no fracture or visible cracking of such elements. These elements will experience yielding and nonlinear response as they deform. It should also be noted that the low-cycle limits for UFPs developed in Skinner et al. (1974) were used to inform the UFP radius design.

The lateral system consisted of two pairs of posttensioned rocking walls: one pair constructed using CLT panels; and the other using mass plywood panels (MPPs). Two materials were used to demonstrate the versatility of the design approach regardless of the mass timber product used. The seismic weight of the building was estimated (see Table 3) as 277 metric tons (611 kips). Rocking wall parameters were first determined using a prescriptive design method developed by the project team and industry collaborators (Busch et al. 2022). This method applied modal response spectrum analysis with an elastic model of the walls, using the site’s seismic hazard and an assumed seismic force reduction factor, $R$, of 6 to obtain initial wall dimensions and thickness, posttensioning sizes and initial stressing, and energy dissipation capacities and distribution. Note this value of $R$ is slightly less than the value of 7 recommended in Sarti et al. (2017) and therefore slightly conservative relative to those recommendations.

Following the initial design, the lateral force-resisting system was checked and refined using the performance-based design methods outlined in the LA Tall Building Guidelines (LATBDC 2023), which are often adopted for tall buildings in urban areas of the West Coast. This process included the development of an OpenSees (Mazzoni et al. 2006) model for nonlinear response history analysis as described in Wichman (2023). The analyses used suites of ground motions selected and scaled to match the intended hazard levels and representative of the source characteristics of the hazard at the building’s design location (i.e., the suites at each hazard level contain records from crustal, interplate, and intraplate earthquakes in relative quantities consistent with the source contribution to each hazard level). The ground motions were selected and scaled over the period range illustrated in Fig. 5, consistent with the requirements of ASCE 7-16. Following the LA Tall Building Guidelines, 1.3 times the suite mean and 1.0 times the suite maximum demands from ${\mathrm{MCE}}_{\mathrm{R}}$ hazard level simulations were checked against force-controlled actions (i.e., component strength limit states) and suite mean and suite maximum component deformations were checked against selected deformation limits for deformation-controlled actions. Additionally, the distributions of the suite mean story drifts at other hazard levels were also computed and compared with the design targets in Table 2. The final wall design parameters are provided in Table 4, and the final configuration is shown in Fig. 6. A detailed engineering drawing set for the as-designed building can be found in Busch (2023).

Detailed design calculations for all components and connections of the rocking wall lateral system can be found in Wichman (2023). The rocking walls were allowed to reach strains above yield but below crushing for the ${\mathrm{MCE}}_{\mathrm{R}}$ demands such that the building could be reused for subsequent research projects. Other components were designed at ${\mathrm{MCE}}_{\mathrm{R}}$ using limit states derived from the NDS (AWC 2018) for timber components and AISC 360-16 (AISC 2016) for steel components.

Lateral load transfer from the diaphragm to the rocking wall was achieved using a shear key detail shown in Fig. 7. This connection detail was adopted from an earlier shake table test (Pei et al. 2019; Blomgren et al. 2019) because it demonstrated resilient performance with no damage to the diaphragm, walls, or steel components. The vertically slotted connection in the rocking wall panel allowed the wall to move independently from the diaphragm in the vertical direction, decoupling the rocking walls from the gravity load-bearing system. The connection between the shear key and the floor diaphragms varied depending on how diaphragm shear was developed for the different mass timber panel materials used. For diaphragms without plywood sheathing (CLT and VLT), a concentrated wing-plate with 45-deg screw connections was used. For floors with plywood sheathing (glulaminated timber, nail laminated timber, and dowel laminated timber floors), a different connection detail was adopted so that the shear would be effectively transferred into the sheathing layer. The rocking walls were braced out-of-plane to the diaphragm at all floor levels using a connection designed to resist tension and compression forces while still allowing vertical wall movement relative to the floor. The demand for the wall lateral braces was estimated using the lateral bracing requirements for point bracing of members in combined bending and axial compression from AISC 360-16 (AISC 2016). Photos for these connection details are presented in Fig. 7.

Due to the relatively small footprint of the building, the floor diaphragms were designed and modeled as rigid. The lateral force demands for the diaphragms were obtained from the nonlinear time history analysis following the LA Tall Building Guidelines. The design objective was to keep the diaphragm damage-free during all tests. Similarly, the diaphragm chords and shear straps were designed to remain elastic. On the sheathed floors, NDS SDPWS (AWC 2021) high-capacity diaphragm design provisions were followed. The diaphragm design demands were largest in the upper levels, which had the largest floor accelerations from the nonlinear analyses.