Abstract
The outriggers are widely adopted in tall and super-tall buildings. Their energy dissipation capacity can significantly influence the nonlinear seismic responses of the entire building structure. Based on an actual tall building project, the structural responses and energy dissipation capacities of three different outriggers were studied through experiments and finite element analyses. The test results of conventional outrigger specimen showed a steep deterioration after peak strength and an unfavorable energy dissipation capacity due to the global buckling of the braces and the local buckling of the chords after flexural yielding. Using buckling-restrained braces and reduced beam sections in a new design of the outriggers, the energy dissipation capacity and the ductility of the outriggers were significantly improved. The yield and peak strengths were further improved with the use of high-strength steel in chords on a third specimen. The finite element simulation of the three specimens indicated that the initial imperfection of the specimens shall be considered, and the developed finite element models yielded good agreements with the test results. The outcome of this work can provide additional references for the application of the outriggers in tall buildings.
Introduction
In recent years, a number of tall and super-tall buildings have been constructed worldwide. How to effectively control the lateral displacement of tall buildings has become a key challenge of engineering design. The lateral displacement of tall buildings can be efficiently reduced by installing outriggers that connect the core tubes and the perimetric frames (Deng et al., 2014; Lu et al., 2015). Therefore, the outriggers were widely used in recently built tall and super-tall buildings. Meanwhile, many corresponding research projects have been performed on the seismic performance of the outriggers.
Taranath (1975), McNabb and Muvdi (1975), Smith and Coull (1991), Hoenderkamp (2008), and Nanduri et al. (2013) analyzed the elastic responses of buildings with different numbers of the outriggers at various positions. Their research outcomes indicated that the lateral stiffness of buildings would be improved effectively if the outriggers were installed in the middle or upper part of the buildings (e.g. the roof or the 2/3 height).
Further research work studied the seismic performance of the outriggers at the nonlinear stage of the structures. Li and Wu (2004), Poon et al. (2011), Jiang et al. (2014), and Lu et al. (2011, 2013, 2014, 2016) found that the outriggers could dissipate a great deal of the seismic energy after yielding, which would significantly influence the nonlinear seismic responses of buildings. Moehle (2015) indicated that the outriggers could be designed to yield and fail as “fuses” of the building, to dissipate the seismic energy, and to protect the primary structure. As a result, the nonlinear performance of the outriggers is also very important in determining the seismic behavior of the entire building structure. Chen et al. (2013) and Nie and Ding (2013) experimentally studied the nonlinear behavior of the outriggers. Their experiments found that global or local buckling occurred in the conventional outriggers, which significantly limited the energy dissipation capacity. Therefore, how to improve the energy dissipation capacity of the outriggers has attracted the focus of many researchers.
In order to improve the energy dissipation capacity of conventional outriggers, some researchers added viscous dampers to the outriggers. Smith and Willford (2007) proposed to add viscous dampers between the outriggers and the perimetric frames to reduce the structural response. Chang et al. (2013) and Asai et al. (2013) proposed and tested a smart damped outrigger system using magnetorheological dampers. Zhou and Li (2014) suggested that the viscous dampers were installed to the outriggers not only in the boundary areas connecting to the perimetric frames but also within the outriggers.
Although viscous dampers have satisfactory seismic performance, the design and analysis procedure of viscous dampers are relatively complicated. In contrast, the buckling-restrained brace (BRB) is much more familiar to structural engineers and much easier to design (Black et al., 2004). Therefore, some researchers proposed to use BRBs to improve the energy dissipation capacity of the outriggers (O’Neill, 2006; Youssef et al., 2010; Zhou et al., 2012, 2014). However, most of the previous research efforts were theoretically works and experimental validations were lacking.
The research presented here used an actual tall building project as the reference and studied a typical outrigger system in this building with the braces replaced by BRBs. The seismic performances of three different outriggers were analyzed through experiments and finite element (FE) simulations. The outcome of this work can provide additional references for the application of the outriggers in tall buildings.
Overview of the background project and the specimen design
The referred tall building project is shown in Figure 1(a) with a total height (H) of 249.6 m. It is located in an 8° seismic design region of China (Ministry of Housing and Urban-Rural Development (MOHURD), 2010; that is the corresponding peak ground acceleration (PGA) value of the design earthquake (i.e. exceedance probability of 10% in 50 years) is 200 cm/s2). The widely used frame-core tube-outrigger system is adopted for this building. There are three sets of steel outriggers in the tall building, which are located at the height of 0.3 H, 0.54 H, and 0.83 H, respectively.

The background building project and the sketches of three different specimens: (a) the background tall building project (m), (b) the prototype outrigger (mm), and (c) the test outrigger (mm).
A typical outrigger located at 0.54 H of the building is selected as the prototype outrigger, as shown in Figure 1(a). The prototype outrigger is a “V”-shaped truss with a length of 9.2 m and a height of 5.5 m, as shown in Figure 1(b). Numerous researchers (Kim et al., 2004; Kumar et al., 1997) have proven that a scaled model can simulate the behavior of the prototype model well if a suitable scale ratio is adopted. Meanwhile, the scale ratio should meet the requirements of the code such as Chinese Specification for Seismic Test of Buildings (JGJ/T 101-2015; MOHURD, 2015). Based on these literatures and due to the limitation of the space and the loading capacity of the laboratory, a 1/3-scale model was adopted in this research. Many test results (e.g. Dan et al., 2011; Hitaka and Matsui, 2003) have demonstrated that the 1/3-scale model can appropriately represent the properties of the prototype structures.
This study designed and tested three different specimens, as shown in Figure 1(c):
Specimen CO: conventional outrigger specimen;
Specimen RBO: reduced beam section (RBS)-BRB outrigger specimen;
Specimen HRBO: high-strength steel-RBS-BRB outrigger specimen.
Specimen CO was designed directly following the drawing of the prototype outrigger. The test results of Specimen CO indicated that it had several major drawbacks, such as the global buckling of the braces and the local buckling of the chords after flexural yielding.
To prevent the buckling of the braces in Specimen CO, the braces were replaced with BRBs in Specimen RBO. However, the deformation capacity of the chords of Specimen CO was limited by the local buckling of the flange after flexural yielding. Hence, in Specimen RBO, all the sections in the chords that might experience flexural yielding were redesigned to be RBSs (Figure 1(c-2)). The experimental results showed that the RBS could provide adequate deformation capacity so that the BRBs could fully exhibit their energy dissipation capacities. The performance and ductility of Specimen RBO were much better than those of Specimen CO.
However, the yield strength of Specimen RBO was smaller than that of Specimen CO. In order to further improve the yield strength, the material of the chords of Specimen RBO was replaced with high-strength steel. The specimen with high-strength steel chords was named as Specimen HRBO (Figure 1(c-3)). The test results showed that Specimen HRBO had similar energy dissipation capacity to Specimen RBO, but higher yield and ultimate strengths.
Experimental observation and results of Specimen CO
Design of Specimen CO
Specimen CO was designed directly following the drawing of the prototype outrigger in the background building project. The pseudo-static test was carried out in this study. The geometry of scaled model is 1/3 scale of the prototype model (MOHURD, 2015). The upper and lower chords and the inclined braces were all constructed with welded shaped steel members. The chord and brace sections were of H 270 × 200 × 10 × 10 and H 200 × 200 × 10 × 10, respectively. Chords were embedded in two foundation beams with double-fillet welded. They were also connected to the loading beam by double-fillet welds. Three steel gusset plates with a 10-mm thickness were used to connect the diagonal braces to the chords. Square-groove welds and strengthening plates were used to connect the gusset plates to braces. The dimensions and construction details of Specimen CO are shown in Figure 2.

Dimensions and details of Specimen CO (mm): (a) Brace I, (b) Brace II, (c) Chord I, (d) Gusset plate II, (e) Gusset plate I, (f) Chord II, (g) Gusset plate III, (h) 1-1, and (i) 2-2.
All specimens were made of Q345 steel. The actual material properties were obtained from coupon tests: yield stress fy of 388 MPa, tensile strength fu of 479 MPa, and ultimate elongation δd of 34%. The shear walls and frame columns in the building were simplified as the end constraints of the outriggers.
Test setup
The test setup is shown in Figure 3. In order to accommodate the size of test specimen and the loading system in the laboratory, the outrigger specimens were tested in a vertical fashion. The bottom of the specimen was fixed to the strong floor of the laboratory. Because the outrigger is constrained by the shear walls and the frame columns in the building (Figure 1(b)), the deformation of the outrigger can be assumed to experience a pure shear mode. Consequently, a pantograph was installed at the top of the specimen to ensure the pure shear deformation of the outrigger (Figure 3(a)). A cyclic load was horizontally applied to the specimen with two parallel actuators. The outrigger in real buildings is restrained by the surrounding slabs, shear walls, and frame columns, and global out-of-plane buckling is unlikely to occur. Thus, the out-of-plane displacement of the “L”-shaped loading beam above the outrigger was restrained. In addition, because the upper and lower chords of the outriggers are restrained by the floor slabs in actual buildings, the global buckling of these components will not occur either. Hence, special restraints were installed in the test setup to prevent the global buckling of both the upper and lower chords.

Test setup and the loading protocol of specimen: (a) test setup of specimen (mm) and (b) the loading protocol.
A pseudo-static loading procedure with displacement control was applied to the specimen. The displacement at the top of the outrigger specimen was monitored. The loading procedure was divided into 16 levels. For each level, the same displacement was repeated twice. The loading protocol was performed according to the provisions of the code of Chinese Specification for Seismic Test of Buildings (JGJ/T 101-2015; MOHURD, 2015), as shown in Figure 3(b). When the specimen failed or could not bear further loads, the test would be stopped.
Experimental results
Observation of the specimen
When the loading displacement reached 15 mm, the stiffness of the specimen began to decrease, implying yielding occurred in the specimen. When the loading displacement reached 20 mm, the out-of-plane global buckling occurred in Brace I (Figure 4). Subsequently, slight global buckling of Brace II was observed during the first cycle of the loading displacement of 25 mm. After that, a significant global buckling of Brace II was observed during the second cycle of loading displacement of 25 mm. Meanwhile, the out-of-plane buckling of the braces also induced the torsion of Chord I. When the loading displacement reached 30 mm, such torsion resulted in the rupture of steel in the web of Chord I at Section B (Figure 1(C-1)).

The hysteretic curve of Specimen CO.
When the loading displacement reached 35 mm, local buckling occurred in the flange of Section F of Chord II and Section A of Chord I, due to the flexural yielding of the chords. The loading procedure was stopped at 40 mm because the rupture length at Section B exceeded half of the height of Chord I. The final failure mode of Specimen CO (Figure 5) was as follows: global buckling of the braces; torsion and rupture of Chord I; and local buckling at Sections A, C, D, E, and F. In addition, local buckling was also observed in the gusset plates of the joint regions.

The finally failure mode of Specimen CO.
Load–displacement hysteretic curve
The load–displacement hysteretic curve of Specimen CO is shown in Figure 4. The peak strength of Specimen CO is 1554 kN and the peak displacement is approximately 20 mm (1/154 of the length of the outrigger). When the displacement reaches 27.7 mm, the strength reduces to 1060 kN, only 66% of the peak strength. The hysteretic curve of Specimen CO deteriorates steeply after the peak point.
The drawbacks of Specimen CO
The test demonstrated the following drawbacks of Specimen CO:
The poor stability of the braces led to the early global buckling under axial force, which further induced the torsion of the chord.
Significant local buckling occurred at the end sections and the joint regions of the chords, due to the flexural yielding of the sections.
The energy dissipation capacity of the outrigger was limited. The ultimate displacement and the ductility of the outrigger were small.
Improvement of the design
Therefore, in order to solve the problems above, the following methods were proposed to improve the design of the outrigger:
In order to increase the energy dissipation capacity of the braces, BRBs were adopted to replace the conventional braces.
RBSs were adopted for the flexural-yielded sections of the chords to improve the deformation capacity (Naserifar and Danesh, 2016).
Experimental observation and results of Specimen RBO
Design of Specimen RBO
In order to improve the performance of Specimen CO, RBSs were adopted to replace the six flexural-yielded sections of Specimen CO. The structural design of RBS followed the requirement of FEMA-350 (Federal Emergency Management Agency (FEMA), 2000). Meanwhile, BRBs were used for the braces. To ensure the comparability of the two outriggers, the loading procedure of Specimen RBO was the same as Specimen CO. The material of the shaped steel and the steel plate of Specimen RBO was the same as that of Specimen CO. Low-yield stress steel LY225 was used for the core part of BRBs, with a yield stress fy and a tensile strength fu of 273 and 328 MPa, respectively. The dimension and construction details of Specimen RBO are shown in Figure 6.

Dimensions and details of Specimen RBO (mm): (a) BRB, (b) 1-1, (c) Gusset plate I, (d) Gusset plate II, (e) Gusset plate III, (f) 2-2, (g) the overhead view of Chord I, (h) the elevation of Chord I, (i) the overhead view of Chord II, and (j) the elevation of Chord II.
Note that the experimental results will indeed be more reliable if the prototype RBO and HRBO are first designed and then scaled to test. However, the structural design of super high-rise building is very complicated and the design method of energy dissipating outriggers has been rarely reported. In view of the fact that many researchers (e.g. Canbolat et al., 2005; Zhao and Astaneh-Asl, 2004) investigated the strategy to improve the seismic performance of key component based on the scaled model and the test results were also considered reliable, a similar method is adopted in this work.
Test of BRB
One BRB was selected randomly from all BRBs to check its performance. The loading procedure followed the requirement of the Chinese Specification for Seismic Test of Buildings (JGJ101-2015; MOHURD, 2015). The load–displacement curve of BRB is shown in Figure 7. The BRB has a stable energy dissipation capacity and suitable ductility, which meets the demand of the outrigger.

Comparison of the hysteretic curves of BRB.
Experimental results
When the loading displacement reached 15 mm, elongation deformation was observed in BRB II. Subsequently, when the loading displacement ranged from 20 to 75 mm, more and more deformation was observed in BRBs. The largest deformation of BRBs before the end of the test was 30.02 mm (for BRB I) and 27.65 mm (for BRB II), respectively. Before the end of the test, neither the webs nor the flanges of the RBSs lost their stability. Additionally, no failure occurred in the welding. The loading procedure was stopped at 75 mm because local buckling occurred in the webs and the flanges of Sections A and B of Chord I. Even after such local buckling, the strength of the specimen still kept stable, which demonstrated an outstanding ductility. The failure mode of Specimen RBO is shown in Figure 8.

The finally failure mode of Specimen RBO.
The load–displacement hysteretic curve of Specimen RBO is shown in Figure 9. The peak strength of Specimen RBO is 1554 kN and the peak displacement is approximately 75 mm (1/40 of the length of the outrigger). In addition, no significant strength deterioration was observed after the peak point, which demonstrated good energy dissipation capacity and ductility of Specimen RBO.

The hysteretic curve of Specimen RBO.
Comparison between Specimen CO and Specimen RBO
The yield load Py, the peak load Pp, the yield displacement Δy, the peak displacement Δp, and the ductility µ of Specimen CO and Specimen RBO are shown in Table 1. Table 1 shows that the ductility of Specimen RBO has been increased by 56% compared with that of Specimen CO. However, the initial stiffness of Specimen RBO is slightly smaller than that of Specimen CO. Such smaller stiffness will be discussed in Section FE analysis of Specimen RBO.
Comparison between the backbone curves of three specimens.
CO: conventional outrigger; RBO: RBS-BRB outrigger; HRBO: high-strength steel-RBS-BRB outrigger.
Compared with Specimen CO, Specimen RBO has a much higher energy dissipation capacity and greater ductility (Figure 10(a)). However, the yield strength of Specimen RBO is smaller than Specimen CO, which is also smaller than the design requirement of the prototype outrigger. Thus, in order to increase the yield strength of Specimen RBO and delay the yielding in the sections of the chord, another specimen named Specimen HRBO was designed with high-strength steel for the chords.

The results of Specimen HRBO: (a) comparison of the hysteretic curves of three different specimens and (b) the finally failure mode of Specimen HRBO.
Experimental observation and results of Specimen HRBO
Design of specimen HRBO
Ban et al. (2011) indicated that the rupture elongation of the high-strength steel whose yield stress exceeds 690 MPa commonly fails to meet the requirement of the design code. Thus, Q600 high-strength steel was adopted in Specimen HRBO with a yield stress fy, a tensile strength fu, and an ultimate elongation δd of 670 MPa, 780 MPa, and 19%, respectively, from coupon tests. The properties of the steel met the requirement of the code of Chinese High Strength Low Alloy Structural Steels (GB1591-2008; General Administration of Quality Supervision, Inspection and Quarantine (GAQSIQ), 2008). To ensure the comparability of the two specimens, the design and loading procedure of Specimen HRBO was the same as Specimen RBO.
Experimental results
When the loading displacement reached 15 mm, elongation deformation was observed in BRB II. Subsequently, when the loading displacement ranged from 20 to 60 mm, more deformation of BRBs was observed. The largest deformation of BRBs before the end of the test was 29.5 and 19.86 mm for the two members, respectively. Before the loading displacement reached 60 mm, neither the web nor the flange of RBS lost their stability. The loading procedure was stopped at 60 mm because local buckling occurred in the web and the flange of Section B of Chord I. Even after this point, the strength of Specimen HRBO still kept stable. The failure mode of Specimen HRBO is shown in Figure 10(b).
The load–displacement hysteretic curve of Specimen HRBO is shown in Figure 10(a). Figure 10(a) and Table 1 show that the yield strength and the peak strength of Specimen HRBO were increased by 20% and 13%, respectively, due to the use of high-strength steel for chords. The yield strength is also greater than that of Specimen CO.
The comparison in Figure 10(a) shows that Specimen HRBO has significant ductility and energy dissipation capacity. Even though the peak displacement of Specimen HRBO was smaller than that of Specimen RBO, the ultimate story drift of Specimen HRBO reached 1/51 (60 mm/3070 mm), which was much greater than the drift limit (1/100) specified in the Chinese Code for the Seismic Design of Buildings (GB5001-2010; MOHURD, 2010). Thus, Specimen HRBO is also considered to have adequate deformation capacity for real buildings.
FE analysis of the specimens
Using the general-purpose FE program MSC.Marc (MSC Software Corp., 2007), the three-dimensional FE model of the three specimens was created to reveal the failure mechanism, as shown in Figure 11. The shell elements and von Mises elasto-plastic model were adopted to simulate the steel members and the gusset plate. The BRBs were modeled using the truss elements in conjunction with the Chaboche constitutive model. The detailed discussion of the FE models of BRB is shown in section “Model calibration for BRB.”

The FE model of three different specimens: (a) the FE model of Specimen CO and (b) the FE model of Specimen RBO and Specimen HRBO.
FE analysis of Specimen CO
Figure 12 shows the comparison between the experimental data and the FE results. The strength and the energy dissipation capacity were overestimated by the FE model if the effect of initial imperfection was not considered.

The FE simulation of Specimen CO.
Thus, the initial imperfection should be considered in the FE model. The method proposed by Eurocode 3 (EN 1993-1-1:2005, 2005) to simulate the initial imperfection of members was adopted, as presented in Figure 13 and equation (1)
where e0 is the initial imperfection; l is the member length. For the braces in Specimen CO, e0/l = 1/350 according to Eurocode 3 (EN 1993-1-1:2005, 2005). q0 is the equivalent uniform forces; Nk is the design value of the compressive force. By adding q0 to the braces, the modified FE simulation results are compared to the test results in Figure 12. The FE simulation with initial imperfection agrees well with the test results.

The initial imperfection of components (EN 1993-1-1:2005, 2005).
FE analysis of Specimen RBO
Model calibration for BRB
The existing research (Rossi, 2014; Wigle and Fahnestock, 2010) indicated that the Chaboche constitutive model had a high accuracy to model the cyclic stress–strain responses of the metallic materials. Therefore, the simulation of the BRB was performed first using the truss element in conjunction with the Chaboche constitutive model. The simulation results of BRB agree well with the test results, as shown in Figure 7. Therefore, the truss elements and Chaboche constitutive model are capable of simulating the complicated hysteresis behavior of BRBs.
FE analysis of Specimen RBO
The Chaboche constitutive model and the truss elements validated above were thus used to simulate the BRBs in Specimen RBO. Due to the fact that the truss element is a one-dimensional element, the sectional dimension details cannot be represented using the truss element. Hence, for Specimen RBO, how to consider the installation eccentricity e0 of the BRB with a truss element is a challenge (Figure 14). The FE analysis indicated that the installation eccentricity e0 had minimal influence on the strength. But it would influence the initial stiffness to some extent, as shown in Figure 15. The simulated results had a slightly larger initial stiffness than the test results. Therefore, a rigid arm was added to the end of the truss element to consider the installation eccentricity e0. According to the actual measurement, the installation eccentricity e0 was approximately 10 mm. The hysteretic behavior accounting for the installation eccentricity e0 is shown in Figure 15, which agrees well with the experimental data.

The installation eccentricity.

The FE simulation of Specimen RBO.
FE analysis of specimen HRBO
Specimen HRBO was simulated with the same FE model of Specimen RBO except for the steel strength of the chords. A comparison between the experimental data and the simulation results is presented in Figure 16, and it shows a good agreement.

The FE simulation of Specimen HRBO.
The above comparisons indicate that the simulated hysteretic behaviors agree well with the experimental data; thus, the developed FE models can be used to analyze the behavior and strength of the outriggers.
Conclusion
This study analyzed the seismic performance of three different outriggers through pseudo-static tests and FE simulations. The conclusions are as follows:
The test results of Specimen CO indicated that the failure modes were the global buckling of the braces and the local buckling of the chords after flexural yielding, which resulted in a steep deterioration of strength after peak and an unfavorable energy dissipation capacity.
Using BRBs and RBSs in Specimen RBO, the buckling of braces was prevented, and the energy dissipation capacity and ductility of the outriggers were significantly improved. However, the yield strength of Specimen RBO was smaller than that of Specimen CO.
Specimen HRBO with high-strength steel chords had higher yield and peak strengths, which fully satisfied the strength and deformation demands of the outriggers.
FE models were developed to simulate the three specimens. The FE results showed good agreements with the test results. These models can be used for future analysis of the outriggers with different chord and bracing configurations.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors are grateful for the financial support received from the Beijing Natural Science Foundation (No. 8142024).
