In the NEWBuildS tall wood building design project, it was assumed that the 20-storey hypothetical building, CHECKER, is located in North Vancouver, BC, Canada, a location that has high seismic, wind, and rain loads. The total building height is 60 m (3 m per storey), and the standard plane dimensions are 27 m × 27 m with a 9 m frame grid. The importance category of “Normal” and a site class of “D” were used for the seismic design. Various parameters linked to the design of the building are listed in Table 1 and were based on the 2010 National Building Code of Canada (NBCC) [National Research Council (NRC), 2010] for the North Vancouver location.

The proposed LLRS was implemented in the design of the hypothetical 20-storey building, known as CHECKER. According to the study by Chen and Ni (2017), hybrid podium structures with a timber structure above the podium can be designed using a two-step analysis method as long as the first podium floor with a specific weight is stiff enough compared to the bottom timber floor. It indicates that the upper timber structure will behave in a similar way as if it is built on the ground. In this study, CHECKER consists of a 19-storey timber upper structure on top of a stiff concrete podium. Consequently, the structural design and analysis of this building due to wind and seismic loads focuses on the wood portion of the structure only (Chen et al., 2015). Due to symmetry of the building, only the LLRS in the east–west direction is presented.

Design provisions for novel structural system and the innovative connections are unavailable in NBCC [National Research Council (NRC), 2010] and the Canadian Standard for Engineering Design in Wood, CSAO86 [Canadian Standards Association (CSA), 2014]. Therefore, principles of mechanics and numerical simulation were used to design and estimate the structural responses of the assemblies, connections, and the whole building.

At the outset, the structural assemblies and connections were sized based on 1.6 times design wind load, since high-rise building design is usually governed by serviceability limit states of wind action and the design strengths for structural components or connections are approximately equal to their average ultimate load-carrying capacity divided by a load factor of 1.6 [Canadian Wood Council (CWC), 2010]. The analysis and design procedure is summarized below.

Static, dynamic, and experimental procedures are the three methods of determining design wind load on buildings in NBCC. What method to use is dependent on height/width ratio, height, and the lowest natural frequency of the building. Usually, the static procedure is used for buildings with fundamental natural frequency greater than 1 Hz. For cases with fundamental natural frequency between 0.25 and 1.00 Hz, the dynamic procedure shall be used. The experimental procedure is used for buildings with fundamental natural frequency less than 0.25 Hz [National Research Council (NRC), 2010]. The fundamental natural frequency of the building, f_n, can be calculated using Eq. 1 according to Rayleigh’s method.

(1)fn=12π∑i=1nFixixnxn∑i=1nMi(xixn)2 where n is the storey number; F_i is the associated wind force of storey i, which was computed using the static procedure, N; x_i is the horizontal deflection of storey i caused by F_i and computed using FE analysis under static wind load, m; M_i is the associated mass of storey i, N [National Research Council (NRC), 2010]. The fundamental natural frequency of the building, 0.53 Hz, was calculated based on the deformation under static wind load. Since it is in the range of 0.25–1 Hz, the dynamic procedure was used to analyze the structural response of the building to wind action. Both the static and dynamic procedures are discussed in the wind-induced response section below. The results show that the across-wind acceleration is 0.009 g and along-wind acceleration is 0.011 g, which are less than the acceleration limit of 0.015 g for residential occupancy according to the NBCC.

The seismic load applied on the CHECKER building was estimated by using ESFP in the preliminary design. The fundamental natural period, T_a, which was estimated using Eq. 2 according to NBCC [National Research Council (NRC), 2010], is 1.04 s.

(2)Ta=0.05(hn)3/4 where h_n is the building height. Using frequency analysis discussed below, the period of the building model was 1.97 s. It is almost twice that estimated by Eq. 2, which also usually happens to mid-rise light wood frame structures and fulfills the requirement of Sentence 4.1.8.11. 3).d).iii) in NBCC. This is because Eq. 2 was derived from measurements on concrete structures rather than timber structures, which are lighter and generally more flexible. To account for potential energy-absorbing capacity and the dependable portion of reserve strength in a structure, the concept of seismic force modification factors R_oR_d was adopted in NBCC. As mass timber panel system is novel, specification on the values of R_oR_d specifically for this type of structural system is unavailable in NBCC. An R_o of 1.5 and an R_d of 2.0 for CLT panel system with simple and standard connections are recommended by the CLT Handbook (Gagnon and Pirvu, 2011). Green and Karsh (2012) have suggested a higher R_d value of 3.5, but that suggestion was not based on any specific test evidence. The results of pushover analysis discussed below show that the ductility ratio of this LLRS was 2.55, which justifies the use of R_oR_d = 3.0 (1.5 × 2.0) in seismic design for the CHECKER building. A specified design base shear of about 4,900 kN was obtained using ESFP based on a period of 1.97 s and R_oR_d of 3.0.

The concept of capacity-based design was adopted in this project for seismic design. The mass timber panels were designed with sufficient strength so that yield or failure would occur in the connections which in turn were designed to fail in a specific sequence. Dowel-type connections with low strength and high deformability and HSK connection system with high stiffness and strength were assigned to the vertical joints of shear walls and core subsystem. HSK connection system with superior stiffness, strength, and ductility was used in the shear connectors and hold-downs at the base of the LLRS. Thus, the vertical joints of shear wall subsystem would yield first and generate most of the plastic deformation before the vertical joints of the core subsystem yield, while the shear connectors and hold-downs provide the major stiffness and strength to the whole system and fail after yielding of the vertical joints with plastic deformation.

Based on meeting the requirements for structural loads (gravity, wind, and seismic) and fire performance of the structural members, the structural member details of a typical floor of the CHECKER building is shown in Figure 2. Grade 2.1E “TimberStrand^® LSL” [Canadian Construction Materials Centre (CCMC), 1994] with dimensions of 19 m (length) × 2.44 m (width) × 89 mm (thickness) was used to build the shear wall and core subsystem. Three layers of LSL panels are combined together to build the shear wall and the core to achieve a total wall thickness of 267 mm. Steel beam S5 × 10 of Grade 50 with specified yield strength of 345 MPa (CSA, 2009) was used to connect the shear wall and core subsystem (Figure 3). Dowel-type connection of LSL using 19 mm (3/4″) dowel with a slenderness ratio of four times fastener diameter (Moses, 2000) was used as the vertical joints of shear wall, while the HSK system was used as vertical joints of core, shear connectors, and hold-downs for the whole system. For the dowel-type connection, the stiffness and strength of each dowel are 25.5 kN/mm and 32.5 kN, respectively. The HSK system is a patented product of TiComTec GmbH. In this study, the stiffness and strength of each glued dowel in the HSK system parallel to the grain are 7.4 kN/mm and 0.8 kN; while those in the perpendicular direction are 2.5 kN/mm and 0.8 kN, respectively (Bathon, 2014). The design capacities of the connections were assumed to be the measured strengths divided by 1.6. The column number and row spacing of the dowel connection, and the column number, layer number, and row spacing of glued dowel of the HSK system were determined based on the forces resisted by the corresponding connections. More details were provided by Chen et al. (2015).

Finite element analysis was performed to calculate the lateral deformation of the tall wood building under static wind load plus gravity load. As given in NBCC, the specified external pressure or suction due to wind on a building was calculated using Eq. 3.

(3)p=IWqCeCgCp where p is the specified external pressure acting statically and in a direction normal to the surface, either as a pressure directed toward the surface or as a suction directed away from the surface; I_W is the importance factor for wind load (Table 1); q is the reference velocity pressure, as provided in Table 1; C_e is the exposure factor, 0.7(h/12)^0.3 but not less than 0.7 since the location is classed as rough terrain; C_g is the gust effect factor, 2.0; and C_p is the external pressure coefficient, averaged over the area of the surface considered, 0.8 and −0.5 for windward and leeward sides, respectively [National Research Council (NRC), 2010]. Only one partial loading case where full wind pressure [National Research Council (NRC), 2010] was applied in east–west direction was considered. The calculated wind load, P_i, was applied at the floor or roof level, as illustrated in Figure 5.

Figure 6 shows the deformation and inter-storey drift of the FE model under the design static wind load. The roof drift was 31.2 mm (≈h_n/1,800), and the inter-storey drift of each storey was less than h_i/500 (=6 mm) specified in NBCC. The fundamental natural frequency of the building, 0.53 Hz, was calculated using Eq. 1 based on the deformation under static wind load. Dynamic procedure was performed below because the frequency was in the range of 0.25–1 Hz [National Research Council (NRC), 2010].

A numerical simulation of the FE model under dynamic wind load plus gravity load was also performed (Chen et al., 2015). The equivalent dynamic wind load was also estimated by Eq. 3. The exposure factor, C_e, and external gust effect factor, C_g, are different from the factors used in the static procedure, but the pressure coefficient, C_p, is the same. C_p = 0.8 and −0.5 for windward and leeward sides, respectively [National Research Council (NRC), 2010].

In the dynamic procedure, the exposure factor, C_e, is based on the profile of mean wind speed, which varies considerably with the general roughness of the terrain over which the wind has been blowing before it reaches the building [National Research Council (NRC), 2010]. Since this building is located at exposure A (rough exposure) terrain, C_e can be calculated by Eq. 4 [National Research Council (NRC), 2010].

The gust effect factor, C_g, which is defined as the ratio of the maximum effect of the loading to the mean effect of the loading, and according to NBCC, the gust effect factor of the CHECKER building under dynamic wind load was 2.36. The storey drift and inter-storey drift of the FE model under the design dynamic wind effect are shown in Figure 6. The results show that the response of the building under dynamic wind load, in terms of storey drift and inter-storey drift, was slightly larger than that induced by the static wind load. The roof drift is 33.7 mm (≈h_n/1,700), and the inter-storey drift of each storey is less than the h_i/500 limit (=6 mm) specified in NBCC.

In the dynamic procedure, not only the wind-induced lateral deformation but also the vibration and vortex-shedding effect are checked. While the maximum lateral wind loading and deflection are generally in the direction parallel to the wind (i.e., the along-wind direction), the maximum acceleration of a building leading to possible human perception of motion or even discomfort may occur in the direction perpendicular to the wind (i.e., the across-wind direction). Across-wind accelerations are likely to exceed along-wind accelerations if the building is slender about both axes, that is if wd/H is less than one-third, where w and d are the across-wind effective width and along-wind effective depth, respectively, and H is the height of the building. The across- and along-wind accelerations, a_W and a_D (m/s²), are calculated from Eqs 5 and 6.

(5)aW=fnW2gpwd(arρBgβW) (6)aD=4π2fnD2gPKSFCeHβDΔCg where ar=78.5×10−3[VH/fnWwd]3.3 (in N/m³); ρ_B is average density of the building (in kg/m³); β_W and β_D are fraction of critical damping in across- and along-wind directions, respectively, and are taken as 0.015 according to NBCC; f_nW and f_nD are fundamental natural frequencies in across-wind and along-wind directions, respectively (in Hz); Δ is maximum wind-induced lateral deflection at the top of the building in along-wind direction (in m), which was obtained by conducting FE analysis on the CHECKER building under dynamic wind load; and g is acceleration due to gravity = 9.81 m/s² [National Research Council (NRC), 2010]. Using Eqs 5 and 6, the across- and along-wind accelerations were calculated to be 0.009 and 0.011 g, respectively. They are less than the acceleration limit of 0.015 g for residential occupancy [National Research Council (NRC), 2010].

Response spectrum analysis is typically utilized to evaluate the seismic performance of buildings under design response spectrum acceleration. In an attempt to obtain accurate results from RSA, a sufficient number of vibration modes must be extracted to model the dynamic response of the system. A criterion to determine the required number of modes involved is that at least 90% of the mass must participate in the modal analysis in the direction of interest [National Research Council (NRC), 2010]. For the CHECKER building, it was found via FE analysis that this criterion was met when 20 vibration modes were involved in the RSA.

A base shear of 4,500 kN was obtained from the RSA. Note that this is similar to the design seismic force derived by ESFP in the preliminary design. Figure 7A shows the lateral deflection of each storey obtained by RSA. According to NBCC, the calculated lateral deflection using RSA should be multiplied by R_dR_o/I_E to account for plasticity of the system [National Research Council (NRC), 2010]. The inter-storey drift shown in Figure 7B is less than the 2.5% h_s limit (=75 mm) specified in NBCC.

Non-linear static pushover analysis was performed on the CHECKER building model using an inverse triangular loading pattern to investigate the failure mechanism of the LLRS. The total force resisted at the bottom of the building is plotted against a reference displacement at the top of the model to define a capacity curve, as shown in Figure 8.

As shown in Figure 8, the load increases linearly with displacement in the initial elastic stage (0–5,500 kN). Then, the relationship between load and displacement becomes non-linear, as the vertical joints of shear wall and core yield sequentially (Figures 9A,B). Once the shear connectors and hold-downs yield and then fail completely (Figures 9C,D), the building loses its lateral resistance, as shown in Figure 9E. Since idealized constitutive models with abrupt unloading mechanism were adopted to model the connections, the load–displacement curve dropped sharply once the modeled connection reached its capacity. This analysis has shown that the structure would fail according to the intended sequence, thereby verifying the adopted capacity-based design approach.

The load–displacement curve of the tall wood building model was analyzed using equivalent energy elastic–perfectly-plastic (EEEP) method in accordance with ASTM Standard E2126 (ASTM, 2011). The EEEP procedure basically ensures that the area under the measured load–slip curve (energy dissipation) is the same as the equivalent EEEP bilinear model shown in red in Figure 8. Using this procedure, the initial stiffness, yield load, and maximum load reached are 23.7 kN/mm, 7,430 kN, and 8,140 kN, respectively. The maximum load is about 1.6 times the design seismic force derived by ESFP. The ductility ratio of this building model is 2.55. Using Eq. 7 for medium and long period buildings (Newmark and Hall, 1982; Chen et al., 2014), a corresponding R_d factor of 2.55 was obtained. A conservative value of 2.0 was used for the seismic design, as was discussed above.

The non-linear dynamic behavior of the CHECKER building model under design hazard level was analyzed with the explicit dynamic analysis method using the central-difference operator (Hibbitt et al., 2011). Ten “Far-Field” earthquake records (NGA #: 169, 721, 752, 829, 900, 1111, 1158, 1244, 1602, and 1787) in the fault normal direction [Applied Technology Council (ATC), 2009] (Table 3 were scaled at the corresponding fundamental period of the building model (1.97 s) to match the spectral acceleration, S_a, of the North Vancouver design spectrum, as shown in Figure 10.

A scaled input acceleration of an earthquake (NGA #: 1602) with a duration of 15 s, and the structural responses, in terms of base shear, lateral drift at top and inter-storey drift ratio of the building are shown in Figure 11. Connections play a significant role in the structural performance of the mass timber panel system. Due to lack of the load–displacement curves of the connections under cyclic loading, idealized elastoplastic models with isotropic hardening and damage behavior were used to model the reversed cyclic load–slip responses of HSK system and LSL dowel connection. Similar idealized hysteresis loops were derived from the lateral resisting system.

Figure 12 shows the statistics for the base shear and inter-storey drift ratio of the tall wood building under the 10 earthquake excitations, along with the design base shear of 4,893 kN estimated by ESFP, the yield load of 7,430 kN calculated using the EEEP procedure of ASTM 2126 (ASTM, 2011), and the maximum capacity of 8,140 kN of the building calculated from pushover analysis. Figure 12A illustrates that most (nine) of the base shears of the building are higher than the design value of 4,893 kN. However, all the base shear values are less than the yield load of 7,430 kN and the maximum capacity of 8,140 kN. In addition, as shown in Figure 12B, all the inter-storey drift ratios are less than the design criterion of 2.5% specified in NBCC. It should be noted that under all earthquakes only the vertical joints between the shear walls of the tall wood building showed yielding behavior.