Correlation of the Near-Fault Pulse-like Ground Motion Characteristics with the Vulnerability of Buildings

Abstract

Determining the impact of pulse-type earthquake characteristics on the vulnerability of base-isolated buildings under non-pounding conditions has yielded conflicting results in previous studies. Moreover, this issue has received less attention for pounding conditions, especially floor-to-floor pounding. Therefore, this study aims to investigate the correlation between pulse-type earthquake characteristics and the seismic response of buildings under both pounding and non-pounding conditions. In the first stage, three base-isolated buildings and one fixed-base building are analyzed separately under 40 pulse-type earthquakes using the nonlinear time history method. Three scenarios are then considered to account for pounding with adjacent buildings. In the first pounding scenario, a base-isolated building with an intermediate moment frame (IMF) is placed between two fixed-base buildings. The second scenario involves changing the base-isolated building’s superstructure system to a special moment frame (SMF). Finally, the third scenario increases the base isolation period (T_b) of the base-isolated building used in scenario two. The correlation between earthquake characteristics and the seismic response of buildings is assessed by linear regression and the Pearson correlation coefficient. The results demonstrate that peak ground acceleration (PGA) has a strong correlation with the seismic response of buildings under pounding conditions, while peak ground velocity (PGV) shows a stronger correlation under non-pounding conditions. However, predicting building vulnerability with a single pulse-type earthquake characteristic remains unreliable unless a large number of ground motions are considered. Otherwise, it is crucial to consider the correlation of all earthquake characteristics with seismic responses.

1. Introduction

Pounding is a phenomenon that can cause severe structural damage. The term “pounding” refers to collisions between structures caused by lateral forces, such as earthquakes. When there is sufficient distance between structures to allow for displacement without collision, the seismic responses are categorized as “non-pounding”. The occurrence of an earthquake with an intensity greater than the earthquake hazard level predicted in building codes, such as the Kahramanmaraş, Turkey earthquake on 6 February 2023, is one of the parameters that can cause pounding [1]. This phenomenon is an important concern in base-isolated structures due to the possibility of structural movement. Several factors can affect the severity of ground pounding, with one of the most important being the characteristics of the earthquake. While numerous studies investigated the seismic response of base-isolated buildings under both far-field and near-fault earthquakes [2,3], a limited number explored the influence of specific earthquake characteristics on this response. Nazarnezhad and Naderpour [4] indicated that the base-isolated building has better performance than a fixed-base building under near-fault pulse-type ground motion, even when the pulse period (T_P) is close to the isolator period (T_b). Bhagat et al. [5] demonstrated that near-fault earthquake pulses have a significant effect on the seismic response of base-isolated buildings compared to buildings under earthquakes without pulses. Habib et al. [6] evaluated the effect of PGA/PGV (peak ground acceleration/peak ground velocity) on the seismic responses of an irregularly base-isolated building under two real pulse-type ground motions. They found that the seismic response of the base-isolated building was significantly influenced by PGA/PGV. However, Mazza and Vulcano [7] pointed out that the accuracy of the PGA is often insufficient to represent the structural damage potential under near-fault pulse-type earthquakes. Sadeghi Movahhed et al. [8] proposed effective ranges for base isolation design parameters, considering the influence of earthquake characteristics. Tajammolian et al. [9] showed that increasing PGV leads to a significant increase in the seismic response of base-isolated structures, especially for a longer T_P. Alhan and Öncü-Davas [10] pointed out that large-magnitude near-fault earthquakes with very long pulse periods can cause significant damage to vibration-sensitive equipment even when operating, while this parameter is less critical for structural integrity. Bhandari et al. [11] found that base isolation systems are ineffective against near-fault earthquakes with a fling-step pulse. Sadeghi-Movahhed et al. [12] investigated the vulnerability of tall base-isolated buildings under multi-seismic hazard levels. They showed a possibility of collapse for certain types of base-isolated buildings designed according to ASCE 7-16 [13].

The occurrence and negative impact of pounding on structures depend on several parameters [14,15,16]. While research on base-isolated structures has focused on pounding at the isolation level, the impact of floor-to-floor pounding has received less attention. Khatami et al. [17] investigated the effectiveness of rubber bumpers in mitigating damage to base-isolated structures caused by pounding with adjacent buildings. They found that increasing the number and specific characteristics of the bumpers, such as thickness and stiffness, led to a reduction in pounding forces. Polycarpou and Komodromos [18] investigated pounding between base-isolated structures and adjacent fixed-base structures. They demonstrated that resonance in the fixed-base building during an earthquake can significantly amplify the destructive effects of pounding in adjacent base-isolated buildings. Yaghmaei-Sabegh and Panjehbashi-Aghdam [19] investigated pounding between two base-isolated structures and found that pounding impacted all floors of the structures. Pant and Wijeyewickrema [20] found that the base-isolated structure exhibited acceptable resistance to shear failure under pounding conditions. Agarwal et al. [21] highlighted that the probability of pounding between base-isolated structures may be influenced by the buildings’ stiffness and the location of the earthquake source. Mavronicola et al. [22] demonstrated that both mass eccentricity and stiffness eccentricity within a structure can negatively impact the seismic performance of base-isolated structures when experiencing pounding with adjacent fixed-base structures. Naderpour et al. [23] showed that pounding with a fixed-base structure amplified the accelerations experienced by the upper floors of the base-isolated structure. Moustafa and Mahmoud [24] compared the seismic response of base-isolated structures under pounding with that of fixed-base structures. They indicated that base-isolated structures, due to their high flexibility, dissipate less energy through hysteretic mechanisms compared to fixed-base structures. Karakozova et al. [25] found that the use of isolation systems amplified signals in the vicinity of zero frequency.

Based on the aforementioned studies, two key ambiguities remain regarding the impact of earthquake characteristics on base-isolated structures:

• Contradictions in past studies on the significance of PGA in predicting vulnerability under near-fault pulse-type earthquakes, specifically under non-pounding conditions;
• Insufficient research on the correlation between near-fault earthquake characteristics and the seismic response of buildings under floor-to-floor pounding.

Therefore, this study aims to investigate the knowledge gaps mentioned above. To achieve this, three base-isolated buildings with increasing T_b and varying superstructure force-resisting systems are considered, along with a fixed-base structure. In the first step, the buildings are analyzed individually under 40 near-fault pulse-type earthquakes. In the second step, each base-isolated building is placed between two fixed-base buildings and analyzed again to evaluate the pounding effect under three scenarios:

• Scenario 1: base-isolated building with IMF superstructure and T_b = 2.7 s;
• Scenario 2: base-isolated building with SMF superstructure and T_b = 2.7 s;
• Scenario 3: base-isolated building with SMF superstructure and T_b = 3.3 s.

Finally, the correlation between the seismic response of the buildings (such as column shear force, interstory drift ratio (IDR), and pounding force) and four characteristics of the considered earthquakes (PGA, PGV, TP, and significant duration (SD) of records) is assessed. To evaluate these correlations, linear regression and the Pearson correlation coefficient are employed.

2. Designed Buildings

For this study, three base-isolated steel structures [26] with six stories and four bays in each plan direction (Figure 1) are considered alongside a fixed-base steel structure. All stories have a height of 3.2 m, and each bay spans 5 m. All columns, except those at the four corners, are arranged in an H shape. Corner columns are rotated 90 degrees relative to the others. This pattern is observed in all building models. The superstructures use special moment frames (SMFs) and intermediate moment frames (IMFs) as their force-resisting systems, designed based on the AISC 360 [27] and AISC 341 [28]. The seismic design of the buildings followed the equivalent lateral force procedure outlined in ASCE 7-16 [13]. It is important to note that, to allow for potential pounding, the design did not enforce the minimum displacement capacity requirement. Member cross-sections are presented in

Table 1. Cross-sections of the structures.

Floor Level	Beam				Column
	SMF2.7	IMF2.7	TFPI3.3	Fixed-Base	SMF2.7	IMF2.7	SMF3.3	Fixed-Base
6	W12 × 72	W12 × 58	W12 × 72	W16 × 50	W24 × 207	W16 × 67	W24 × 207	W21 × 201
5	W12 × 72	W12 × 58	W12 × 72	W16 × 50	W24 × 207	W18 × 76	W24 × 207	W24 × 207
4	W12 × 72	W12 × 58	W12 × 72	W16 × 50	W24 × 229	W18 × 86	W24 × 229	W24 × 229
3	W12 × 79	W12 × 72	W12 × 79	W16 × 50	W24 × 250	W18 × 97	W24 × 250	W24 × 229
2	W12 × 79	W12 × 72	W12 × 79	W16 × 50	W24 × 250	W18 × 106	W24 × 250	W27 × 258
1	W12 × 79	W12 × 72	W12 × 79	W16 × 50	W24 × 250	W21 × 147	W24 × 250	W27 × 258
Isolated Level	W18 × 86	W18 × 86	W18 × 86	-	-	-	-	-

. In Table 1, the base-isolated buildings are named with both the type of superstructure system (e.g., SMF) and the T_b value (e.g., 2.7 s). For instance, the designation SMF2.7 would represent a base-isolated building with an SMF superstructure and a T_P of 2.7 s. Type C soil conditions were assumed for the design, along with a 2475-year return period earthquake spectrum with S_S = 2.37 g and S₁ = 0.974 g. The steel material properties are assumed to be an elasticity modulus of 20,000 kN/cm², yield stress of 34.5 kN/cm², and ultimate stress of 45 kN/cm². Dead and live loads on floors are 5.5 kN/m² and 2 kN/m², respectively, reducing to 4 kN/m² and 1.5 kN/m² for the roof. The analysis process is carried out using SAP2000 Version 25 software [29]. Pounding is modeled within the software using the gap link. This is a compression-only element that deactivates under tensile forces. The impact of the pounding model type on the maximum seismic response of base-isolated buildings was found to be insignificant [30]. Therefore, a simple gap element is used for modeling pounding. Gap links are modeled at the column-to-beam connection joints along the entire height of the adjacent buildings, as illustrated in Figure 1. The axial stiffness of the story slabs is used to represent the stiffness of the gap links [20]. Triple friction pendulum isolators (TFPIs) are positioned between the building and the ground level. This isolator is used in many studies [31,32,33]. To model these isolators in the software, a triple pendulum isolator link element is employed. The design characteristics of the TFPIs are presented in

Table 2. TFPI characteristics.

Building	${\mathit{\mu }}_1 = {\mathit{\mu }}_4$	${\mathit{\mu }}_2 = {\mathit{\mu }}_3$	${\mathit{R}}_{\mathit{e}\mathit{f}\mathit{f}1} = {\mathit{R}}_{\mathit{e}\mathit{f}\mathit{f}4}$ (cm)	${\mathit{R}}_{\mathit{e}\mathit{f}\mathit{f}2} = {\mathit{R}}_{\mathit{e}\mathit{f}\mathit{f}3}$ (cm)	${\mathit{d}}_1 = {\mathit{d}}_4$ (cm)	${\mathit{d}}_2 = {\mathit{d}}_3$ (cm)
I6R2T2.75	0.12	0.066	213	33	40	5
S6R2T2.75	0.12	0.066	213	33	40	5
S6R2T3.3	0.075	0.033	213	33	40	5

. The determination of these characteristics is based on the following procedure [34]:

(1)

Predicting the required displacement capacity of the isolator (D_b);

(2)

Determining the effective stiffness:

$$K_b={\displaystyle \frac{W}{2R_{eff1}}}+{\displaystyle \frac{\mu W}{D_b}}$$

(1)

where R_eff1 is the effective radius; W is the effective seismic weight; μ is the equivalent friction coefficient;

(3)

Calculating the period:

$$T_b=2\pi \sqrt{{\displaystyle \frac{W}{K_{b}g}}}$$

(2)

where g is the gravity acceleration;

(4)

Determining the damping ratio:

$${\beta }_b={\displaystyle \frac{4\mu W\left(D_b-D_y\right)}{2\pi D_b^2{K}_b}}$$

(3)

where D_y is the yield displacement;

(5)

Calculating the damping reduction factor:

$$B_b={\left({\displaystyle \frac{{\beta }_b}{0.05}}\right)}^{0.3}$$

(4)

(6)

Determining the required displacement capacity of the isolator and comparing with the estimated value in the first step:

$$D_b={\displaystyle \frac{g{S}_1{T}_b}{4{\pi }^2{B}_b}}$$

(5)

The nonlinear time history method is employed to evaluate the seismic response of the buildings. The analysis considers 40 near-fault pulse-type earthquakes (normal direction component) selected based on the report by Baker et al. [35]. The near-fault ground motion conditions are determined according to ASCE 7-16 [13], which specifies two criteria: (1) The earthquake epicenter must be located within 15 km of the surface projection of a known active fault capable of producing earthquakes with a moment magnitude (Mw) of 7.0 or greater, or (2) the epicenter must be located within 10 km of the surface projection of a known active fault capable of producing earthquakes with an Mw of 6.0 or greater. The 40 near-fault pulse-type earthquakes selected for the analysis all satisfied the mentioned criteria. The list of the selected earthquakes is provided in

Table 3. Near-fault pulse-type earthquakes.

No.	Event	Station	Year	Magnitude	PGA (g)	PGV (cm/s)	TP (s)	SD (s)
1	Imperial Valley-06	EC County Center FF	1979	6.53	0.18	54.47	4.5	14.88
2	Imperial Valley-06	EC Meloland Overpass FF	1979	6.53	0.38	115.02	3.3	6.23
3	Imperial Valley-06	El Centro Array#4	1979	6.53	0.36	77.84	4.6	10.24
4	Imperial Valley-06	El Centro Array#5	1979	6.53	0.38	91.48	4.0	9.42
5	Imperial Valley-06	El Centro Array#6	1979	6.53	0.44	111.82	3.8	8.55
6	Imperial Valley-06	El Centro Array#7	1979	6.53	0.46	108.78	4.2	4.80
7	Imperial Valley-06	El Centro Array#8	1979	6.53	0.47	48.56	5.4	5.80
8	Imperial Valley-06	El Centro Differential Array	1979	6.53	0.42	59.6	5.9	6.89
9	Morgan Hill	Coyote Lake Dam (SW Abut)	1984	6.2	0.81	62.29	1.0	3.08
10	Loma Prieta	Gilroy—Gavilan Coll	1989	6.9	0.29	30.78	1.8	5.20
11	Loma Prieta	LGPC	1989	6.9	0.94	96.96	4.4	9.98
12	Landers	Lucerne	1992	7.3	0.70	140.33	5.1	13.06
13	Northridge-01	Jensen Filter Plant	1994	6.69	0.52	67.4	3.5	8.03
14	Northridge-01	Jensen Filter Plant Generator	1994	6.69	0.52	67.35	3.5	8.03
15	Northridge-01	Newhall—Fire Sta	1994	6.69	0.72	120.08	1.0	5.52
16	Northridge-01	Newhall—W Pico Canyon Rd	1994	6.69	0.43	87.73	2.4	7.07
17	Northridge-01	Rinaldi Receiving Sta	1994	6.69	0.87	167.1	1.2	7.15
18	Northridge-01	Sylmar—Converter Sta	1994	6.69	0.59	130.28	3.5	13.51
19	Northridge-01	Sylmar—Converter Sta East	1994	6.69	0.83	113.56	3.5	7.26
20	Northridge-01	Sylmar—Olive View Med FF	1994	6.69	0.73	122.85	3.1	5.78
21	Kobe, Japan	KJMA	1995	6.9	0.85	95.75	1.0	9.56
22	Kobe, Japan	Takarazuka	1995	6.9	0.65	72.52	1.4	5.07
23	Kocaeli, Turkey	Gebze	1999	7.5	0.24	51.18	5.8	7.45
24	Chi-Chi, Taiwan	CHY028	1999	7.62	0.66	77.65	2.2	8.66
25	Chi-Chi, Taiwan	CHY101	1999	7.62	0.38	75.28	4.6	30.32
26	Chi-Chi, Taiwan	TCU049	1999	7.62	0.29	46.08	11.7	21.52
27	Chi-Chi, Taiwan	TCU052	1999	7.62	0.38	165.54	8.4	16.25
28	Chi-Chi, Taiwan	TCU053	1999	7.62	0.22	40.87	12.8	22.18
29	Chi-Chi, Taiwan	TCU054	1999	7.62	0.16	60.37	10.5	23.34
30	Chi-Chi, Taiwan	TCU068	1999	7.62	0.56	184.6	12.2	12.50
31	Chi-Chi, Taiwan	TCU075	1999	7.62	0.33	88.56	5.2	27.13
32	Chi-Chi, Taiwan	TCU076	1999	7.62	0.31	67.81	4.0	29.04
33	Chi-Chi, Taiwan	TCU082	1999	7.62	0.23	57.8	9.0	22.67
34	Chi-Chi, Taiwan	TCU087	1999	7.62	0.13	43.68	9.4	23.73
35	Chi-Chi, Taiwan	TCU101	1999	7.62	0.21	68.35	10.0	18.82
36	Chi-Chi, Taiwan	TCU102	1999	7.62	0.30	109.04	9.7	16.26
37	Chi-Chi, Taiwan	TCU103	1999	7.62	0.13	62.13	8.2	20.88
38	Chi-Chi, Taiwan	TCU122	1999	7.62	0.22	42.43	10.9	30.82
39	Chi-Chi, Taiwan	WGK	1999	7.62	0.30	67.62	4.4	28.39
40	Imperial Valley-06	Agrarias	1979	6.53	1.33	230.47	2.3	11.47

. Moreover, their corresponding response spectra are shown in Figure 2. SD is calculated based on the duration between 5% and 95% of the Arias intensity.

3. Results and Discussion

This section investigates the correlation between earthquake characteristics (PGA, PGV, T_P, and SD) and seismic results (IDR, column shear force, and pounding force) using linear regression and the Pearson correlation coefficient (Equation (6)):

$$corr\left(x,y\right)={\displaystyle \frac{E[\left(x-E\left(x\right)\right)\left(y-E\left(y\right)\right]}{\sqrt{var\left(x\right)}\sqrt{var\left(y\right)}}}$$

(6)

where x and y are the random variables, E is the expectation, and var(x) and var(y) are the variance of x and y, respectively.

3.1. Interstory Drift Ratio (IDR)

Figure 3, Figure 4 and Figure 5 show a positive correlation between IDR and both PGA and PGV. Conversely, a negative correlation is observed between IDR and T_P, as well as SD.

Table 4. Pearson correlation coefficient between IDR and earthquake characteristics.

No.	Building	PGA		PGV		TP		SD
		Pounding	Without Pounding	Pounding	Without Pounding	Pounding	Without Pounding	Pounding	Without Pounding
Scenario 1	Left-side building	0.863	0.447	0.549	0.792	−0.567	−0.136	−0.459	−0.244
	IMF2.7	0.837	0.761	0.524	0.822	−0.564	−0.437	−0.452	−0.474
	Right-side building	0.830	0.447	0.485	0.792	−0.620	−0.136	−0.503	−0.244
Scenario 2	Left-side building	0.834	0.447	0.524	0.792	−0.610	−0.136	−0.491	−0.244
	SMF2.7	0.842	0.727	0.530	0.811	−0.570	−0.382	−0.464	−0.459
	Right-side building	0.836	0.447	0.519	0.792	−0.571	−0.136	−0.446	−0.244
Scenario 3	Left-side building	0.830	0.447	0.522	0.792	−0.569	−0.136	−0.458	−0.244
	SMF3.3	0.842	0.551	0.527	0.866	−0.568	−0.166	−0.463	−0.282
	Right-side building	0.835	0.447	0.528	0.792	−0.573	−0.136	−0.453	−0.244

confirms these correlations through the Pearson correlation coefficient. PGA demonstrates the strongest correlation with IDR for both base-isolated and fixed-base buildings in all pounding scenarios. In contrast, PGV shows the strongest correlation with IDR under non-pounding conditions. For both pounding and non-pounding conditions, SD has the weakest correlation with IDR for base-isolated buildings in all scenarios, while T_P shows the weakest correlation for fixed-base buildings. Changing the base-isolated building’s superstructure system from IMF to SMF or increasing T_b has an insignificant impact on the correlation between earthquake characteristics and the IDR of base-isolated buildings under pounding. However, these scenarios significantly affect the correlations under non-pounding conditions. Changing from IMF to SMF reduces the correlation values of base-isolated buildings’ IDR with earthquake characteristics. In addition, increasing T_b decreases the correlations for PGA, T_P, and SD, while on the other hand increasing the correlation for PGV.

In all pounding scenarios, the increase in the base-isolated buildings’ IDR (compared to the non-pounding case) is much higher than that observed in fixed-base buildings. Pounding increases the IDR of base-isolated buildings under all ground motions, whereas fixed-base buildings experience a decrease in IDR under some records.

When the superstructure system is changed from IMF to SMF or T_b is increased, pounding leads to an increased IDR in base-isolated buildings under most earthquakes. However, the IMF to SMF change results in a decrease in IDR for adjacent fixed-base buildings under pounding. Increasing T_b has complex effects on the IDR of fixed-base buildings, causing reductions under some records and increases under others.

3.2. Column Shear Force

Figure 6, Figure 7 and Figure 8 illustrate the shear force distribution in columns E5 and K5 of the left and right fixed buildings, respectively. Additionally, they show the maximum shear force for columns F5 and J5 in the base-isolated building. The analysis reveals a positive relationship between column shear force and both PGA and PGV. However, column shear force exhibits a negative relationship with both the T_P and SD.

Table 5. Pearson correlation coefficient between column shear force and earthquake characteristics.

No.	Building	PGA		PGV		TP		SD
		Pounding	Without Pounding	Pounding	Without Pounding	Pounding	Without Pounding	Pounding	Without Pounding
Scenario 1	Left-side building	0.857	0.343	0.542	0.758	−0.559	−0.008	−0.452	−0.153
	IMF2.7	0.831	0.716	0.513	0.861	−0.551	−0.321	−0.432	−0.452
	Right-side building	0.829	0.343	0.492	0.758	−0.600	−0.008	−0.490	−0.153
Scenario 2	Left-side building	0.820	0.343	0.502	0.758	−0.601	−0.008	−0.485	−0.153
	SMF2.7	0.838	0.690	0.525	0.878	−0.551	−0.311	−0.448	−0.428
	Right-side building	0.846	0.343	0.526	0.758	−0.568	−0.008	−0.451	−0.153
Scenario 3	Left-side building	0.824	0.343	0.507	0.758	−0.563	−0.008	−0.459	−0.153
	SMF3.3	0.836	0.514	0.503	0.884	−0.551	−0.118	−0.454	−0.271
	Right-side building	0.854	0.343	0.539	0.758	−0.573	−0.008	−0.459	−0.153

presents the Pearson correlation coefficient values, which quantify the strength of these correlations. PGA demonstrates the strongest correlation with column shear force for both base-isolated and fixed-base buildings in all pounding scenarios. In contrast, PGV shows the strongest correlation with column shear force under the non-pounding conditions. For both pounding and non-pounding conditions, SD has the weakest correlation with column shear force for base-isolated buildings in all scenarios, while T_P shows the weakest correlation for fixed-base buildings. For base-isolated buildings under non-pounding conditions, changing the superstructure system from IMF to SMF or increasing T_b weakens the correlation between column shear force and PGA, T_P, and SD, while strengthening the correlation with PGV. It should be noted that these scenarios have a negligible impact on the correlation between column shear force and earthquake characteristics under pounding conditions.

Pounding affects base-isolated and fixed-base buildings differently in terms of column shear force. In scenario 1, the increase for base-isolated buildings due to pounding compared to no pounding is lower than for fixed-base buildings for specific ground motions (numbers 1, 3, 5, 10, 16, 23, 29, 30, 35, and 37). However, in scenarios 2 and 3, this enhancement is higher than that of fixed-base buildings under all ground motions. On the other hand, for fixed-base buildings, pounding does not always lead to an increase. In fact, certain ground motions can lead to a decrease in column shear force.

Shifting from an IMF to an SMF system and increasing T_b generally leads to higher column shear forces in the base-isolated building under pounding. However, this change has the opposite effect on adjacent fixed-base buildings, causing a decrease in their column shear forces. Moreover, increasing T_b has a mixed impact on fixed-base buildings. Some ground motions trigger reductions in column shear force, while others induce increases.

3.3. Pounding Force

This section explores the relationship between earthquake characteristics and the maximum pounding forces generated on both the left- and right-side base-isolated building. Figure 9, Figure 10 and Figure 11 reveal a negative correlation between pounding force and both T_P and SD, while PGA and PGV exhibit a positive correlation with pounding force.

The Pearson correlation coefficients (

Table 6. Pearson correlation coefficient between pounding force and earthquake characteristics.

No.	Pounding Location	PGA Pounding	PGV Pounding	TP Pounding	SD Pounding
Scenario 1	Left-side pounding	0.582	0.253	−0.327	−0.269
Scenario 2	Left-side pounding	0.628	0.340	−0.343	−0.328
Scenario 3	Left-side pounding	0.603	0.337	−0.398	−0.272

) reveal that PGA has the strongest correlation with pounding force in all scenarios. Conversely, PGV shows the weakest correlation in scenario 1, while SD exhibits the weakest in scenarios 2 and 3.

Changing the base-isolated building’s superstructure system from IMF to SMF increases the correlation between earthquake characteristics and the pounding force. However, increasing the base period (T_b) only strengthens the correlation between T_P and pounding force, while weakening the correlations with other earthquake characteristics.

It is common practice to select a limited number of earthquakes when assessing building behavior under pounding scenarios. This study demonstrates that predicting building vulnerability under pounding using a single characteristic, even PGA (which shows a strong correlation with a building’s seismic response), can lead to errors, especially when using a limited set of near-fault pulse-type earthquakes. For example, records 3 and 4 come from the same event with the same soil type but have different PGA values. Interestingly, increasing PGA within this set does not show a clear trend with pounding force.

Similar inconsistencies are observed with other earthquake characteristics when using a limited dataset. Records 2, 4, 25, and 27 all share the same PGA (0.38 g), but the correlation between pounding and PGV, TP, and SD varies across different scenarios. As PGV increases in this set, the pounding value can either increase or decrease. This inconsistency is also evident within records from a single earthquake. For instance, records 28 and 38 from the Chi-Chi earthquake have the same PGA (0.22 g), yet PGV shows a positive correlation with pounding in scenario 2 and a negative correlation in scenario 3, where the only difference is T_b.

4. Conclusions

This study investigated the correlation between near-fault pulse-type earthquake characteristics and the seismic response of buildings under floor-to-floor pounding and non-pounding conditions. Three pounding scenarios were considered, each involving a base-isolated building positioned between two fixed-base buildings. Nonlinear time history analysis was conducted using 40 near-fault pulse-type earthquake records. The correlation between earthquake characteristics and building response was then determined using linear regression and the Pearson correlation coefficient. The results are as follows:

• Peak ground acceleration (PGA) and peak ground velocity (PGV) show positive correlations with the seismic response of buildings under both pounding and non-pounding conditions. However, pulse period (T_P) and significant duration (SD) exhibit negative correlations with the seismic responses.
• Under pounding conditions, PGA demonstrates the strongest correlation with the seismic response. Moreover, PGV has the strongest correlation with the seismic response under non-pounding conditions.
• Pounding significantly increases the seismic response of base-isolated buildings compared to fixed-base buildings under most of the ground motions. Furthermore, pounding may show a reduction in the interstory drift ratio (IDR) of fixed-base buildings.
• Predicting building vulnerability based solely on PGA for base-isolated buildings and PGV for fixed-base buildings may be possible when using a large number of earthquake records. However, for smaller datasets, it is recommended to consider the correlation of seismic responses with all earthquake characteristics investigated in this study.
• In this research, there are some limitations, such as the exclusion of the building’s period, response spectrum ordinates, and spectrum intensity scales. These factors could be explored in future studies.