Experimental observation on the seismic performance of reinforced concrete columns with a well-confined plastic hinge region

Damage states

In general, similar damage behavior was observed in all column specimens. Once initial cracks formed, those cracks were propagated as drift ratio increased. Then, yielding of longitudinal reinforcement occurred as the number of cracks increased, leading to concrete cover spalling. Subsequently, fracture of longitudinal reinforcements occurred, and test was terminated. No failure was observed prior to the fracture of longitudinal reinforcements in all specimens. However, drift ratio corresponding to the fracture varied slightly in column specimens. Notably, UNIT6—with the widest vertical spacing of transverse reinforcement—experienced the earliest fracture at a drift ratio of 5%. Drift ratio and lateral displacement corresponding to the fracture are summarized in Table 2.

As observed in UNIT7, the use of larger-diameter reinforcement delayed reinforcement fracture to a higher drift level (7%) compared to UNIT1 (6%). When the total steel area is constant, using fewer, larger reinforcements reduces susceptibility to local buckling and low-cycle fatigue, both key mechanisms of reinforcement fracture in plastic hinge regions. Similarly, in UNIT8, the use of higher-yield-strength steel also delayed fracture to a 7% drift level. Higher yield strength postpones the onset of yielding and plastic strain accumulation, thereby extending the fatigue life of the reinforcement. It also allows the steel to sustain greater load before yielding, resulting in slower damage progression under cyclic loading. The enhanced lateral capacity observed further suggests that the post-yield strength of the reinforcement contributed to resisting additional drift without failure.These findings advocate a transition from conventional strength-based design to a performance-based approach that prioritizes inelastic behavior and fracture resistance in RC columns under seismic loading. By incorporating factors such as reinforcement size, reinforcement yield strength, and drift capacity, designers can enhance energy dissipation, delay reinforcement fracture, and ensure stable performance in plastic hinge regions. This comprehensive strategy promotes greater structural resilience and durability in earthquake-prone areas.

Crack pattern and initial yield

The progression of flexural cracking in the tested column specimens followed a typical pattern observed in reinforced concrete members subjected to lateral cyclic loading. After the formation of initial horizontal flexural cracks near the base of the columns during the first loading cycle, these cracks propagated further in subsequent cycles, with the development of additional new cracks. As the drift ratio increased, the spacing between these new cracks gradually decreased, indicating a more distributed cracking pattern. This behavior is consistent with the classical flexural cracking mechanism, where increased curvature leads to tighter crack spacing.

Interestingly, the formation of new cracks during the second and third cycles at the same drift ratio was limited, suggesting a stabilization of the crack pattern at a given drift level once the tensile strain demand was redistributed across the existing cracks. A notable influence of aspect ratio was observed in the crack evolution process. Specifically, specimens with lower aspect ratios exhibited denser horizontal crack spacing, implying a more localized curvature and strain demand at the base, which facilitates the development of multiple cracks in a confined zone. Fig. 4 shows representative crack patterns of specimens, UNIT2 and UNIT4 at the drift ratio of 1.5% (36 mm) before initial cover concrete spalled.

An increase in aspect ratio leads to a higher initial yielding displacement and a reduction in lateral force capacity. This trend is evident when comparing UNIT3, UNIT7, and UNIT4, which have aspect ratios of 3, 4, and 6, respectively. Their corresponding initial yield displacements were 11 mm, 14 mm, and 26 mm, while the peak lateral forces decreased from 236 kN to 173 kN and 124 kN. Increasing the aspect ratio of reinforced concrete columns leads to a shift from shear-dominant to flexure-dominant behavior (from shorter column to slender column), resulting in higher initial yield displacements and reduced lateral force capacity. As columns become more slender, their lateral stiffness decreases, requiring greater displacement to reach yielding. At the same time, the contribution of shear resistance diminishes, and the columns rely more heavily on flexural deformation to resist lateral loads. This combination of reduced stiffness, increased curvature demand, and lower shear contribution explains the observed increase in yielding displacement and decrease in lateral strength with higher aspect ratios.

Since the first yielding occurs at outer-most longitudinal reinforcements in both loading directions for a circular section, initial yielding was determined by the measured value of strain gauge installed at outer-most longitudinal reinforcements in the bottom of column specimens. In addition, rotation corresponding to the initial yielding was also calculated at the bottom regions of the column specimens using horizontal and vertical displacements measure by LVDTs. Table 3 summarizes initial yield displacements, lateral force and rotations corresponding to initial yielding. As observed, initial yield displacement was affected only by aspect ratio.

Representative crack patterns.

Cover concrete spalling and crushing of confined concrete

Cover concrete spalling of a column is one of important parameters for the evaluation of stiffness change and seismic performance. This is due to the fact that spalling is gradually developed along the height of column specimens in the plastic hinge region and leads to a crushing of confined concrete. Accordingly, presence and height of spalling can be a salient measure for determining repair and retrofitting of concrete members. The displacement at the time of initial spalling was visually investigated at each cycle of load steps. In addition, heights of initial cover concrete spalling were also measured along the column length. Cover spalling of unconfined concrete was progressively developed in the plastic hinge regions as the drift ratio increased, leading to crushing of confined concrete. The crushing of confined concrete may lead to an erroneous estimation of displacement measured by LVDT. Thus, applied drift ratio was evaluated when the initial crushing of confined concrete occurred. Table 4 summarizes the drift ratio and height of initial cover concrete spalling, drift ratio corresponding to initial crushing of confined concrete and buckling of longitudinal reinforcement, and final spalled region. Representative initial spalling and final spalled region for UNIT5 and UNIT6, respectively are shown in Fig. 5.

The onset of cover spalling occurred at drift ratios of 2% for UNIT6 and UNIT1, and was delayed to 3% for UNIT5, indicating that reducing vertical spiral spacing to a certain threshold can postpone cover spalling. Spalled heights was reduced significantly when reducing vertical spiral spacing. Core crushing occurred at drift ratios of 4%, 5%, and 6% in UNIT6, UNIT1, and UNIT5, respectively, demonstrating that the impact on delaying core crushing was even more pronounced. Tighter spiral spacing significantly contributed to improving resistance against early core degradation. Using higher yield strength (UNIT 8 vs. UNIT1) or larger diameter (UNIT7 vs. UNIT1) of longitudinal reinforcements also delayed initial spalling and core crushing. However, increasing the yield strength of longitudinal reinforcements seemed to have no impact on spalled height, while increasing the diameter of longitudinal reinforcements slightly increased spalled height. Furthermore, increasing the aspect ratio (comparing UNIT3, 7, and 4) tended to raise the drift levels at which initial spalling and core crushing occurred, but also resulted in greater spalled heights.

Based on this observation, the results suggest that current confinement strategies—typically based on minimum transverse reinforcement limits—may not be sufficient to delay critical damage such as cover spalling and core crushing under seismic loading. While code-prescribed spacing may satisfy strength and ductility requirements, the experimental evidence shows that tighter spiral spacing notably improves damage resistance, especially under large cyclic drifts. This indicates that existing provisions may underestimate the benefits of closely spaced reinforcement in enhancing post-yield performance. Revising confinement guidelines to include more performance-based criteria, particularly in plastic hinge regions, could improve seismic resilience and deformation capacity in reinforced concrete columns.

The drift ratios associated with each damage state, as presented in Table 4, can be regarded as rational criteria for predicting damage levels in the probabilistic seismic performance evaluation of reinforced concrete (RC) columns. The experimental methodology of categorizing damage states based on drift ratios merits careful consideration. This is because structural damage is typically assessed by examining the performance of individual components—such as columns, beams, and other load-bearing elements—each of which exhibits unique damage characteristics. As such, experimental observation of column behavior is essential for accurately identifying damage states linked to mechanisms such as cracking, spalling, yielding, buckling, and reinforcement fracture. This level of detail is critical, given that the field of earthquake engineering has been shaped by thorough observation and documentation of physical damage phenomena.

However, acquiring reliable experimental data—particularly local strains and rotations within the plastic hinge regions of RC columns—poses significant challenges. In light of these difficulties, drift ratio has emerged as a more practical and intuitive metric for classifying damage states. It offers a simplified yet meaningful representation of structural deformation and has thus become a widely accepted parameter in seismic performance assessments. This practicality underlies the use of drift ratio as the primary basis for damage state classification in the present column tests.

Accordingly, the onset of concrete spalling, core crushing, and reinforcement buckling were categorized as indicative of slight, moderate, and extensive damage limit states, respectively. These classifications are intuitive and align with observed damage progression. However, in the absence of standardized damage limit state definitions in South Korea, they should be regarded as exploratory classifications intended to guide further development of performance-based seismic evaluation criteria.

Representative initial spalling and final spalled region.

For evaluation purpose, drift ratios summarized in Table 4 were compared with those proposed by HAZUS¹⁷ and Dutta and Mander¹⁸. Table 5 shows comparison of drift ratio at each damage limit state. It is noteworthy that although classification of damage limit state is different in between HAZUS¹⁷ and Dutta and Mander¹⁸, proposed drift ratio at each damage limit state is identical. As observed, the present evaluation exhibited notably less conservative in comparison with those by HAZUS¹⁷ and Dutta and Mander¹⁸.

The experimentally derived drift ratios indicate that the damage thresholds proposed by both HAZUS and Dutta and Mander are conservative. The drift limits in HAZUS were developed primarily for purposes of seismic risk assessment and loss estimation, and thus adopt conservative assumptions to accommodate uncertainties and ensure broad applicability across various structural systems. In contrast, Dutta and Mander’s drift limits were based on a combination of experimental data and analytical models for specific structural types, particularly bridge columns designed according to U.S. seismic codes. These design standards are typically more conservative than those currently adopted in Korea. As a result, the drift ratios obtained in this study are less conservative than those suggested in prior literature, emphasizing the influence of national design practices and regulatory frameworks on damage limit definitions.

As for slight damage limit state, drift ratio suggested by the present test results exhibited a difference more than twice despite of the damage limit state corresponding to initial spalling. A similar difference was observed in the case of moderate damage limit state. This can be attributed to different provisions of seismic design details among different countries. The drift ratio of a column can be influenced by various variables such as specifications, material properties, cross-section dimension, yield strength and ratio of reinforcement, aspect ratio and so on. When seismic design details for these variables differ from countries, the drift ratio corresponding to damage limit state may also be different. Therefore, it is deemed necessary to conduct a study on the drift ratios for damage limit states based on experimental results of reinforced concrete columns designed in accordance with seismic design details of a country.

Lateral force-displacement hysteretic response

The lateral force–displacement hysteretic responses of the column specimens are presented individually and comparatively in Fig. 6 and 7. Points marked in the response indicate the onset of each damage state, i.e., initial yielding of longitudinal reinforcement, initial cover concrete spalling, crushing of confined concrete, buckling and fracture of longitudinal reinforcement, as described in the figure legend. In addition, while dashed line represent lateral force corresponding to theoretical moment strength, dotted line indicates the lateral force considering second moment effect of P-\(\Delta\). As observed, all columns experienced no strength reduction with increasing displacement until the fracture of longitudinal reinforcement, and thus showed stable hysteretic response. As shown in the response of specimens of UNIT1, UNIT5 and UNIT6, the most stable response was observed in UNIT5 employing the smallest vertical spacing of transverse reinforcement. The hysteretic loops over every loading cycle of specimen UNIT6, 1, 5 were plotted as Fig. 8. As can be seen from the figure, before drift level of 4%, the loops of three specimens are completely identical. There is a slight difference among them when the drift level reaches to 4%, and the difference becomes significant between UNIT 6 and the others when drift level reaches 5%, the drift at steel fracture in UNIT6 (cycle 22). Compared to UNIT1 and 5, the maximum lateral load reduces significantly, the loop of UNIT 6 at this drift level is rotated to be more horizontal and the pinching effect becomes pronounced from this drift level. When the drift level reaches 6%, the steel of UNIT1 fractures in the first cycle (cycle 25) and that of UNIT5 fractures in the third cycle (cycle 27). This indicates that tighter vertical spacing of transverse reinforcements extends the inelastic deformation range prior to longitudinal reinforcements fracture.

Specimen UNIT7 exhibited a similar overall response to UNIT1, as both had the same longitudinal reinforcement ratio, though UNIT7 used fewer bars with larger diameters. However, reinforcement fracture in UNIT7 occurred later than in UNIT1, suggesting that larger-diameter bars are more effective in delaying fracture. Meanwhile, comparison between the hysteretic responses of UNIT1 and UNIT8 showed differences only in the peak lateral force, indicating that the yield strength of the longitudinal reinforcement had minimal impact on the overall hysteretic behavior. For a detailed representation, refer to the hysteretic loops for each loading cycle shown in Figs. 13, 14, 15 in the Appendix.

Lateral force-displacement hysteretic response.

Hysteretic loops of all specimens, plotted in groups for comparison.

Hysteretic loops over every cycle of lateral loading of specimens UNIT1, 2.

Stiffness degradation and energy dissipation capacity

Figure 9 illustrates the stiffness degradation behavior of all specimens over the examined drift ratio range. The degradation curves display symmetrical patterns in both the push and pull directions of lateral loading, attributed to the constant axial compression applied during testing. Specimens with greater aspect ratios (UNIT3, UNIT7, UNIT4) exhibited lower secant stiffness, indicating increased flexibility. In contrast, specimens subjected to higher axial compression (UNIT1 vs. UNIT8) or reinforced with higher yield strength steel (UNIT1 vs. UNIT8) showed enhanced stiffness, reflecting improved confinement and load resistance. Variations in longitudinal reinforcement diameter and transverse reinforcement spacing had minimal influence on stiffness degradation.

Secant stiffnesses of all specimens.

Figure 10 shows the evolution of accumulated dissipated energy over lateral loading cycles. The accumulated dissipated energy was calculated as the summation of areas of hysteretic loops from the beginning to the considered loading cycle. At a particular cycle of lateral load, the specimen with higher aspect ratios had smaller accumulated dissipated energy. Specimens under higher axial compression levels exhibit higher accumulated dissipated energy. Other parameters including yield strength, diameter of longitudinal reinforcement and vertical spacing of transverse reinforcement almost have a minimal influence on energy dissipation prior to the fracture of longitudinal longitudinal reinforcements.

Accumulated dissipated energy of all specimens.

Maximum lateral strength

The theoretical strength calculated using the equivalent rectangular stress block based on Concrete Bridge Design: Limit State Design Method¹⁶ were compared with the maximum lateral forces obtained experimentally in both push and pull directions for each column. Theoretical strength and the maximum lateral forces were summarized in Table 6.

As presented in Table 6, the experimentally measured maximum lateral forces exceeded the theoretical strengths for all column specimens. This discrepancy is primarily attributed to nonlinear mechanisms that are not fully captured in conventional P–M interaction analyses based on simplified strain compatibility and force equilibrium. Specifically, factors such as concrete confinement, steel strain hardening, and compression strut action contribute to enhanced strength under cyclic lateral loading. While theoretical models often assume unconfined or idealized confined concrete, transverse reinforcement in real specimens improves confinement effectiveness—particularly in plastic hinge zones—resulting in increased compressive strength and ductility. Furthermore, strain hardening of longitudinal reinforcement, typically unaccounted for in elastic–perfectly plastic assumptions, contributes to increased flexural capacity. Additional effects such as localized cracking, stress redistribution, tension stiffening, and hysteretic energy dissipation under cyclic loading further raise lateral resistance beyond static predictions.

The relative increase in experimental lateral strength was more pronounced in specimens with higher axial force ratios and reinforcement yield strength. For instance, specimen UNIT2 (15% axial load ratio) exhibited a maximum lateral force approximately 15% greater than that of UNIT1 (10% axial load), indicating that increased axial compression enhances the safety margin relative to theoretical predictions. This improvement arises from several interrelated nonlinear effects: greater axial compression increases confinement through lateral restraint (Poisson effect), enhances flexural capacity by shifting the neutral axis and amplifying internal force couples, delays tensile cracking, reduces crack widths, and improves stiffness and damage resistance. These beneficial mechanisms are often underestimated or neglected in theoretical models, which tend to adopt conservative assumptions regarding degradation, bar slip, and post-yield behavior.

Finally, the maximum lateral force recorded for UNIT7, which had half the number of longitudinal reinforcement bars compared to UNIT1, was nearly identical to that of UNIT1. This suggests that, under the same design strength and detailing, the number of longitudinal bars had a negligible influence on the overall lateral strength within the tested range.