Study on Damage Identification Method for Chuan-Dou Timber Frame Structures Based on Evolution of Dynamic Characteristic Parameters

Abstract

To explore the evolution of dynamic characteristics of Chuan-Dou timber structures under different damage states, this study takes a typical Chuan-Dou timber structure in Southwest China as the research object. A 1:7 scaled model of a two-story timber frame with five main columns and four secondary columns, three bays, and two rooms was designed and fabricated, and combined pseudo-static and dynamic tests were carried out. When the specimen was in three typical states, namely intact, moderate damage, and severe damage, the sudden release method was adopted to obtain structural vibration responses. The natural frequencies and damping ratios in the X- and Y-directions under each state were identified, and the damage sensitivity differences among stiffness, frequency, and damping ratio were compared and analyzed. The test results show that with the aggravation of damage degree, structural stiffness degrades continuously, and the natural frequency shows a monotonic decreasing trend. The X-direction frequency decreases from 11.178 Hz to 7.8 Hz, and the Y-direction frequency decreases from 6.2 Hz to 5.156 Hz. The damping ratio increases significantly. The X-direction damping ratio increases from 3.552% to 8.951% (an increase of 152.0%), and the Y-direction damping ratio increases from 4.391% to 11.94% (an increase of 171.9%). Comparative analysis shows that the change amplitude of the damping ratio is about 5 to 10 times that of the natural frequency, and it has higher identification sensitivity to structural non-linear damage behavior. This paper innovatively applies the frequency-damping ratio dual-index collaborative determination strategy to Chuan-Dou timber structures, establishes a damage identification method based on the evolution of dynamic characteristic parameters, and discusses the engineering application paths of sensor optimal layout strategy, structural health archive establishment, and post-earthquake rapid screening. The research results can provide experimental basis and technical reference for daily health monitoring, post-earthquake rapid identification, and seismic performance evaluation of traditional timber structures of Chuan-Dou timber structures.

1. Introduction

The Chuan-Dou timber frame is one of the typical forms of traditional Chinese timber structures. Due to its excellent climate adaptability and regional suitability, it is widely used in rural residences in Southwest China. Located in a seismically active zone, Southwest China has experienced multiple destructive earthquakes since the 21st century, with most seismic hazards concentrated in rural areas. For example, during the 2013 Lushan Ms 7.0 earthquake, a large number of Chuan-Dou timber frame residences in the seismic zone suffered varying degrees of damage [1]. As a traditional residential type with huge stock and wide distribution, its seismic safety is directly related to the lives and property safety of the people. Therefore, conducting targeted research holds significant engineering value and practical significance.

In recent years, scholars have carried out extensive research on the seismic performance and safety assessment of Chuan-Dou timber frame buildings.

Regarding the research on the seismic performance of infill walls in Chuan-Dou timber frame buildings, Li et al. [2] addressed the problem of out-of-plane failure of masonry infill walls in Chuan-Dou timber frames by combining experimental and finite element modeling methods, proposing a calculation formula for the out-of-plane peak load, which provides theoretical basis and data support for anti-collapse design. Lv et al. [3] used an experimentally validated ABAQUS model to compare the mechanical properties of different frame forms (TF, TFD, TFVP), clarifying the influence of infill panel geometric parameters and load transfer mechanisms through parametric analysis and providing data support for seismic optimization design. Guo et al. [4] conducted tests on four full-scale specimens to compare the effects of different infills on the in-plane seismic performance of Chuan-Dou timber frames. The results showed that brick masonry infill significantly outperforms vertical wood panel infill in improving lateral bearing capacity and stiffness, and both types of infills can effectively enhance the energy dissipation capacity of the structure, providing references for infill selection and seismic design. Liang et al. [5] studied the influence of infill wall type and column base slip on structural seismic performance through quasi-static cyclic tests. The results indicated that infill walls can significantly improve structural stiffness and bearing capacity (the initial stiffness of wood panel infill is 6.9 times that of the bare frame), and the coordinated deformation capacity of wood panel infill is superior to that of brick wall infill.

In terms of research on the seismic performance of mortise-tenon joints in Chuan-Dou timber frame buildings: Chun et al. [6] selected four typical mortise-tenon joints in traditional timber structures (mantou tenon, through tenon, half tenon, and straight tenon at the bottom of secondary column) as research objects. Through low-cycle reversed loading tests, they systematically investigated their failure modes, hysteretic characteristics, skeleton curves, rotational stiffness, ductility coefficients, and energy dissipation capacity under horizontal loads. Wang [7] conducted quasi-static tests on Chuan-Dou timber frames and found that half tenon heads are prone to shear cracks under cyclic loads, while through tenons are mainly characterized by rotation and tenon pull-out; the overall structure has good ductility and deformation capacity, but significant stiffness degradation occurs when the horizontal load reaches 50% of the peak value, making it prone to inclined collapse. Shi et al. [8] selected five types of joints (straight tenon, stepped through tenon, half tenon, large-head tenon, and silver ingot shoulder tenon), fabricated ten full-scale specimens using pine and cedar, and carried out quasi-static cyclic loading tests. They clarified the dominant influence of joint type on bearing capacity, stiffness degradation, strength degradation, and energy dissipation and revealed the differences in seismic mechanisms among different joints.

Regarding the research on seismic strengthening of Chuan-Dou timber frame buildings, Wang [9] proposed a cable-stayed strengthening method, which was verified by quasi-static tests on a 1:7 scaled model. The herringbone two-story strengthening scheme was found to be optimal, as it can protect both mortise-tenon joints and enclosure walls and has the advantages of strong applicability, low cost, and convenient construction. Li et al. [10] combined full-scale tests and numerical simulations to evaluate the influence of different strengthening schemes on the out-of-plane seismic performance of masonry-infilled Chuan-Dou timber frames, establishing a simplified calculation model for out-of-plane bearing capacity considering arching effect and reinforcement contribution, which is applicable to two-side and three-side support conditions. Huang et al. [11] designed four full-scale specimens and systematically studied the hysteretic performance, stiffness degradation, strength degradation, energy dissipation, and damage characteristics of the structure under different infill forms through low-cycle reversed cyclic loading tests. Ge et al. [12] developed an assembled hoop reinforcement device with adjustable stiffness, and shaking table tests verified that it can effectively restore the seismic performance of damaged structures, reducing displacement response by 27% and improving joint energy dissipation by 22%.

In the research on timber structure health monitoring and damage detection, Chen et al. [13] summarized the principles, characteristics, and applications of various representative non-destructive testing technologies for timber structures, analyzing the practical gaps in technology implementation and future research directions. More et al. [14] systematically reviewed the core technologies and practical application cases of structural health monitoring (SHM) for timber structures, providing a reference framework for SHM research and engineering applications in timber structures. Ren and Meng [15] proposed a damage identification method combining curvature mode and wavelet transform, verifying its effectiveness through numerical simulations and tests on wooden beams and frames. Ren et al. [16] fabricated a 1:4.5 scaled model based on the Guang-Yue Tower and, through two shaking table tests, compared and analyzed indicators such as damage modes, dynamic characteristics, shear capacity, and energy dissipation capacity between intact and damaged models, providing experimental support for post-earthquake assessment and repair of multi-story traditional timber structures. Perković et al. [17] established a non-destructive/semi-destructive inspection framework based on visual inspection, moisture content testing, and ultrasonic methods, drawing upon a century-old timber roof case study in Croatia, thereby providing a standardized pathway for rapid post-earthquake diagnosis of damaged timber structures. Jiang et al. [18] developed a health monitoring system employing FBG fiber-optic sensors, enabling long-term real-time monitoring of timber beam deflection, timber column inclination, temperature and humidity, and fire hazards, which fills the technical gap in inclination monitoring of mortise-tenon timber structures based on FBG. Wang et al. [19] proposed a damage identification method integrating ARMA time-series models, PCA dimensionality reduction, and Mahalanobis distance for large-span timber grid-shell structures, effectively addressing the misidentification issues associated with high modal density and high-degree-of-freedom spatial timber structures, enabling precise localization of damage positions and quantification of damage severity.

Regarding post-earthquake damage surveys of timber structures, Ye et al. [20] conducted a comprehensive investigation of timber structures in the affected area of the 2025 Myanmar Mw 7.9 earthquake, systematically summarizing the structural characteristics of double-story and single-story timber buildings and revealing the controlling effects of joint connection types, member corrosion, site liquefaction, and surface rupture on seismic damage. Yan et al. [21] focused on brick–timber structures in the meizoseismal area of the 2023 Jishishan Mw 6.2 earthquake, classifying local dwellings into two categories based on double-slope and single-slope high-wall configurations and categorizing the damage into four types, overall/partial collapse, roof damage, wall damage, and ancillary member damage, thereby elucidating that unreasonable structural systems, low mortar strength, poor integrity, and construction deficiencies constitute the primary causes of disaster. Zhang et al. [22] concentrated on minority settlement areas in Qinghai Province affected by the Jishishan earthquake, conducting surveys on over 1800 traditional timber and brick (adobe)–timber residential buildings, identifying key damage characteristics including foundation settlement, column foot slip, and mortise-tenon joint pull-out and tearing and revealing the impact of stiffness mismatch, weak connections, and material deterioration on the seismic performance of structures. Zhong et al. [23], through post-earthquake damage surveys of the 2023 Jishishan Mw 6.2 earthquake, identified weak connections, construction deficiencies, and site amplification effects as the core causes of timber structure damage, providing empirical support for seismic retrofitting and post-disaster reconstruction of timber structures.

Currently, research on Chuan-Dou timber frames mainly focuses on the seismic performance of mortise-tenon joints and infill walls; although there have been abundant research results in the field of structural damage identification in recent years (for example, Nerilli and Ahmadi [24] conducted a study on the degradation of bond performance at the GFRP bar–concrete interface under alkali–thermal coupling environments, revealing the quantitative relationship among bond strength attenuation, interfacial stiffness degradation, and interfacial damage evolution, thereby providing experimental basis for characterizing local structural damage through interfacial mechanical indicators), most of the relevant work has centered on widely distributed typical structural types such as reinforced concrete and masonry structures. As a highly representative traditional residence in Southwest China, research on damage identification of Chuan-Dou timber frames is relatively weak, especially lacking systematic damage identification research with dynamic characteristic parameters such as natural frequency and damping ratio as core indicators. Newmark and Hall [25], Hart and Ibáñez [26], Kenichi Den Suda [27], and other scholars proposed early that structural damping ratios vary significantly with structure type, number of stories, component internal force state, and stress level, and the damping ratio of structures entering the elastoplastic stage can reach more than ten times that of the elastic stage. Elmenshawi et al. [28] conducted in-plane vibration tests on an unreinforced stone wall and found that the viscous damping ratio is closely related to specimen damage. Wang et al. [29] discovered that the damping ratio of masonry structures is affected by both damage state and excitation spectral characteristics; Sun et al. [30] and Zheng [31] quantified the variation law of equivalent damping ratio at different damage stages through masonry wall tests and constructed damage classification models based on time-frequency domain features, providing new ideas for post-earthquake safety assessment.

Based on existing research, this study primarily focuses on two innovative aspects: first, the targeted expansion of the research object. Existing damage identification studies based on dynamic characteristics mainly revolve around modern structural types such as reinforced concrete frames and masonry walls, where damage modes are dominated by material cracking and steel yielding; consequently, the relevant findings cannot be directly transferred to traditional timber structures. As a highly representative form of traditional folk dwellings in Southwest China, Chuan-Dou timber structures exhibit damage concentrated in mortise-tenon joint loosening, tenon pull-out, timber splitting, and column foot slip, with stiffness degradation and energy dissipation mechanisms possessing significant particularities, thus urgently necessitating the establishment of damage identification methods applicable to this structural type. Second, the synergistic construction of identification indicators. Existing studies predominantly employ single indicators such as frequency or damping ratio for damage determination, yet each indicator possesses inherent limitations: frequency is sensitive to early-stage damage but susceptible to environmental fluctuations, while damping ratio reflects energy dissipation capacity yet exhibits considerable scatter. To this end, this paper innovatively applies the frequency-damping ratio dual-index collaborative determination strategy systematically to Chuan-Dou timber structures, aiming to provide a novel technical pathway for damage identification of traditional timber structures.

Therefore, this study integrates existing testing and identification methods for natural frequency and damping ratio, conducting experimental research on typical Chuan-Dou timber frame specimens. Multiple excitation methods are adopted to compare and analyze the variation laws of natural frequency and damping ratio under typical states (intact, slightly damaged, moderately damaged, etc.) and their identification sensitivity at different damage stages; meanwhile, the application path of the proposed method in practical engineering is discussed. The research can provide references for improving the damage identification method of Chuan-Dou timber frames and enhancing the safety assessment level of traditional timber structures, as well as clarify the optimization direction and technical route for subsequent related research, offering data references and technical support for post-earthquake on-site assessment and regular damage identification of Chuan-Dou timber frames.

2. Experimental Program

2.1. Specimen Overview

Based on the field investigation results, measured dimension data, and structural characteristics of Chuan-Dou timber frame residences in Southwest China, a two-story Chuan-Dou timber frame building in a certain area was selected as the prototype. The prototype specimen was selected from a typical Chuan-Dou timber-framed residential building in Southwest China. Based on the field investigations conducted by the research group on the seismic capacity of rural buildings in this region, timber structures represent one of the primary traditional structural forms, with the Chuan-Dou framing system being the most extensively employed type. The selected prototype building possesses significant representativeness among existing Chuan-Dou timber structures in the locality: its characteristics, including the number of stories, construction features, structural configuration, construction era, and building materials, all fall within the typical scope that is both quantitatively substantial and geographically widespread in this region. Furthermore, the research team obtained critical field measurement data at the building site, encompassing member dimensions and joint construction details, thereby providing a reliable basis for the subsequent design and fabrication of the scaled model. Therefore, the selection of this building as the prototype structure can effectively reflect the universal dynamic characteristics and seismic performance levels of rural Chuan-Dou timber structures in Southwest China.

Considering test requirements such as spatial constraints of the loading device and adaptability of data acquisition equipment, a two-story timber frame model with five main columns, four secondary columns, and three bays (two rooms) was designed and fabricated at a 1:7 scale. Scotch pine was selected as the wood material. The total height of the model is 839 mm, and the planar dimensions are 686 mm × 1200 mm; five through-columns are arranged along the depth direction (column spacing: 171 mm), with four secondary columns placed between the through-columns. Columns are connected by multiple Chuan-fang (tie beams) to form a single-frame timber structure, and purlins are directly placed on the top of the columns, with rafters configured according to the tile dimensions; the spacing along the bay direction is 600 mm, and each single-frame timber structure is connected by Dou-fang (corbel beams) and other components to form a stable integral frame. The roof of the model is paved with flat tiles (bottom tiles) and cylindrical tiles (covering tiles) in combination. After the tiles were cut according to the full-scale specifications, the paving was carried out in accordance with local traditional construction techniques. The front elevation and side elevation of the wooden frame model are shown in Figure 1 and Figure 2, respectively.

Wooden columns and beams (Fang members) are connected using typical mortise-tenon joints such as through tenons and hook tenons to ensure the longitudinal and transverse stability of the timber structure; in the column base area, a closed-loop connection is achieved through the base tie beam (Di-jiao Fang) formed by the intersection of Chuan-fang (tie beams) and Dou-fang (corbel beams), which effectively enhances the integrity and stability of the bottom of the timber structure. The design drawings, 3D schematic diagrams, and physical model diagrams of the core mortise-tenon joints and column base beam connection joints are shown in Figure 3, Figure 4, and Figure 5, respectively.

Purlins are processed according to the mortise-tenon form of the prototype structure and directly placed on the top of the columns; the rafters and purlins in the model are connected by nailing. Due to the steep slope of the model’s roof and the limited number of tiles, it is impossible to fully adopt the overlapping laying process of the original structure. Therefore, an appropriate amount of cement mortar is used to fix the roof tiles, and a suitable vertical load is applied to the structure simultaneously. To restrict the displacement of the column bases, rough wood flooring is used in the test instead of the plinths of the prototype, and wooden strips are fastened around the column bases for limiting positions. The cross-sectional dimensions of the primary members of the model structure are summarized in

Table 1. Cross-sectional dimensions of primary members of the model structure.

Member Name	Dimensions/mm	Cross-Sectional Form
Main column (corresponding mortise-tenon joint depth)	D = 35	Circular
Secondary column (corresponding mortise-tenon joint depth)	D = 35
Purlin	D = 30
Chuan-fang (corresponding mortise-tenon joint width and height)	43 × 10 (middle two stories)	Rectangular
	30 × 10 (bottom and top stories)
Dou-fang (corresponding mortise-tenon joint width and height)	30 × 10
Rafter	37 × 7

, and the overall schematic diagram is presented in Figure 6.

2.2. Wood Material Property Test

Wood mechanical performance tests typically encompass compression, tension, shear, and bending types; through tension–compression tests, key mechanical parameters including elastic modulus, shear modulus, tensile strength, compressive strength, and shear strength parallel to grain can be obtained. Furthermore, wood mechanical properties exhibit significant anisotropic characteristics, which can be classified into three categories according to grain direction: parallel to grain, perpendicular to grain in radial direction, and perpendicular to grain in tangential direction, as shown in Figure 7.

Appearance and material grade description: Upon inspection, all defect indicators of the test specimens satisfy the requirements of Grade A (top grade) material specified in GB/T 153-2019 [32]. The wood surface exhibits a natural light yellow-brown color, with straight grain and uniform coloration, showing no obvious knots, decay, insect damage, or cracking defects; the overall appearance remains intact. The structure consists of main columns, secondary columns, Chuan-fang, purlins, and mortise-tenon joints, with regularly arranged column grids; the through tenons and half-tenons at the joints fit tightly, with no signs of loosening or tenon pull-out observed.

To ensure the reliability of test results and reduce uncertainties arising from material scatter, all Pinus sylvestris test specimens in this study were sourced from the same batch. The sampling method and specimen fabrication strictly followed GB/T 1927.1-2021 “Test Methods for Physical and Mechanical Properties of Small Clear Wood Specimens—Part 1: Collection of Specimens” [33] and GB/T 1927.2-2021 “Test Methods for Physical and Mechanical Properties of Small Clear Wood Specimens—Part 2: Sampling Methods and General Requirements” [34].

In accordance with the series of standards [35,36,37,38,39,40,41,42,43] for “Test Methods for Physical and Mechanical Properties of Small Clear Wood Specimens,” basic physical and mechanical performance tests of wood were conducted. The test items included density, moisture content, compressive strength parallel to grain, compressive strength perpendicular to grain in radial direction, compressive strength perpendicular to grain in tangential direction, bending strength parallel to grain, bending elastic modulus parallel to grain, tensile strength parallel to grain, shear strength parallel to grain, and hardness. The results are presented in

Table 2. Wood property indicators.

Performance Indicator	Value	Coefficient of Variation (%)
Density	0.495 g/cm3	7.2
Moisture content	7.464%	6.8
Compressive strength parallel to grain	42.29 MPa	2.67
Compressive strength perpendicular to grain in radial direction	7.23 MPa	12.7
Compressive strength perpendicular to grain in tangential direction	4.03 MPa	20.8
Bending strength parallel to grain	60.8 MPa	4.22
Bending elastic modulus parallel to grain	9933.67 MPa	3.86
Tensile strength parallel to grain	111.69 MPa	12.9
Shear strength parallel to grain	3.58 MPa	10.7
Hardness	4016.5 N	7.99

.

It should be noted that changes in wood moisture content can significantly alter its dynamic characteristics. Qiu et al. [44] demonstrated that when moisture content increased from 2% to 12%, the dynamic elastic modulus of spruce decreased by 4.4%, the resonant frequency was significantly reduced, and the loss factor markedly increased. Considering that the lateral resistance and energy dissipation capacity of Chuan-Dou timber structures primarily depend on the friction and embedment characteristics of mortise-tenon joints, and that joint loosening caused by wood shrinkage further exacerbates frequency variations, this study implemented strict control over environmental conditions and specimen moisture content throughout the entire test process. The tests were conducted in a constant temperature and humidity laboratory, with the environmental temperature controlled at 20 ± 2 °C and relative humidity controlled at 60 ± 5%, ensuring a dry environment. Compared with the timescale of wood moisture content changes, which is on the order of days, the single measurement duration in this study was only several minutes, and the overall test period was on the order of hours—far shorter than the timescale required for significant moisture content changes. To further reduce the impact of environmental fluctuations, specimens were wrapped with plastic film at non-loading locations during fabrication completion and test intervals, effectively isolating moisture exchange between specimens and the environment. The specimen moisture content was measured before and after testing, with the measured moisture content variation controlled within ±0.5%.

2.3. Loading Protocol

To investigate the spatial combination effects of structural members under seismic action and account for the influence of multi-directional loading on the structure as comprehensively possible, while overcoming the limitations of unidirectional loading effects, low-cycle reversed loads can be applied simultaneously in both the X- and Y-directions of the specimen. Accordingly, two-dimensional quasi-static loading protocols can be broadly classified into bidirectional synchronous loading protocols and bidirectional asynchronous loading protocols [45]. A bidirectional asynchronous alternating loading protocol was adopted in the test (as shown in Figure 8); that is, loads were alternately applied along the structural depth direction (X-direction) and bay direction (Y-direction).

Given the absence of a distinct yield stage in wooden members, the loading was conducted under a displacement-controlled mode. Prior to the formal loading, preloading was applied to the timber frame in both the X- and Y-directions with a preloading displacement of 2 mm (the minimum value of the loading protocol) to verify the operational status of all testing instruments and the stability of the loading device. Formal loading cycles were initiated only after all inspections were passed.

A graded displacement loading protocol with equal differences (tolerance d = 2 mm, 3 mm) was adopted for the formal loading, and the loading rate was controlled at 1 mm/5 s throughout the test: the first stage (2 mm, 4 mm) employed single-cycle loading; the second stage (6 mm, 8 mm) adopted double-cycle loading; the third stage (10 mm, 12 mm, 15 mm, 18 mm) used triple-cycle loading; and the fourth stage (21 mm, 24 mm, 27 mm, 30 mm, 33 mm) applied double-cycle loading. The detailed loading protocol is shown in Figure 9.

A hydraulic servo actuator with a maximum load capacity of 3 t and a maximum stroke of 1000 mm was adopted to apply bidirectional loading to the timber frame, with the requirement that the two loading planes remain relatively independent. For this purpose, the horizontal loading device was arranged in two layers: the first layer was positioned at an elevation of 450 mm (X-direction, along the structural depth direction), and the second layer at an elevation of 520 mm (Y-direction, along the structural bay direction).

The first-layer loading device was composed of four rectangular steel tubes with a cross-section of 40 mm × 40 mm × 3 mm, which were tightened and fixed relative to the model by turnbuckles. The device and the actuator end were connected by four high-strength bolts with a diameter of 16 mm, enabling the uniform transmission of horizontal load to the six wooden columns of the model along the bay direction. To avoid mutual interference with the second-layer loading device, universal wheels were installed on the six wooden columns in contact with this layer of the device so as not to restrict the translational movement in the other direction.

The second-layer loading device was a spatial steel frame, which was fixed to the test model by externally attached steel tubes with a cross-section of 30 mm × 30 mm × 3 mm and connected by high-strength bolts with a diameter of 10 mm. The loading device and the actuator were fastened by eight high-strength bolts with a diameter of 16 mm. Similarly, to eliminate the mutual influence between the two layers of devices, universal wheels were installed on the wooden columns in contact with this layer of the device to ensure that the translational degree of freedom in the other direction was not constrained.

The detailed layout of the loading device is shown in Figure 10.

2.4. Loading Protocol

According to the research objectives of this study, the contents to be measured in the quasi-static test include actuator output force, overall lateral displacement of the timber frame, interstory drift, column foot slip, column foot uplift, relative rotation angles and slip amounts of partial mortise-tenon joints, and strain variations of the Fang members at mortise-tenon joints, among others. Displacement measurement was conducted by combining displacement transducers with digital image correlation (DIC) technology. Specifically, wire displacement transducers were arranged at the second-story and first-story elevations along the depth direction (X-direction) and bay direction (Y-direction) of the timber frame to measure overall lateral displacement and interstory drift; wire displacement transducers were installed at column foot locations to measure column foot slip amounts; rod displacement transducers were placed at the four corner column feet of the timber frame to measure column foot uplift amounts and compare the differences between central columns and edge columns. In addition, DIC technology was synchronously employed in the test for full-field displacement measurement to obtain the displacement field distribution and deformation patterns on the timber frame surface. Due to the stringent requirements of DIC technology regarding occlusion, camera viewing angles, and lighting conditions, coupled with issues such as overlapping occlusion by members and loading devices in joint regions and surface feature loss caused by large rotation angles during testing, DIC measurement results served only as reference verification, with primary displacement data still based on measured results from wire displacement transducers and rod displacement transducers.

In addition to necessary measuring devices such as load cells, cable displacement transducers, push-rod displacement transducers, and strain gauges arranged on the test specimens, MEMS (Micro-Electro-Mechanical System) acceleration sensors were additionally installed to meet the requirements for calculation and analysis of dynamic characteristic parameters such as natural frequency and damping ratio in the later stage. Six unidirectional acceleration sensors were adopted in the test and placed on the upper parts of three timber frames, respectively, to collect the acceleration responses of each frame in the X-direction and Y-direction. The sensors were firmly fixed to the structure with hot melt adhesive. The detailed layout is shown in Figure 11.

Based on the pre-judgment of damage status before the test, after the completion of each cyclic loading group, it was determined whether to conduct dynamic parameter measurement according to the actual damage condition of the structure. During measurement, all loading devices were unloaded first, and the tension sudden release method was adopted for excitation (i.e., manually pulling the top of the structure to a set position and then suddenly releasing it to generate free decay vibration). The free decay waveforms in both the X- and Y-directions were obtained, respectively. The duration of each measurement was no less than 3 min, and the number of excitations in each direction was no less than 3 times. The waveform quality was monitored in real time during the measurement process, and the number of excitations was supplemented or the measurement time was extended as needed. After the measurement was completed, the loading devices were reset before proceeding to the next set of cyclic loading.

3. Theoretical Basis

3.1. Frequency

The natural frequency can be obtained by the peak-picking method. The principle is to perform Fast Fourier Transform (FFT) on the recorded time-domain vibration data of the specimen to obtain the corresponding amplitude–frequency diagram and auto-power spectrum, thereby determining the natural period of the specimen. The above principle is shown in Figure 12.

3.2. Damping Ratio

3.2.1. Free Decay Method

The damping ratio of free vibration is calculated according to Formula (1) of the free decay method. The schematic diagram of the principle is shown in Figure 13. Three adjacent peaks in the time-history decay segment are selected to obtain two adjacent periods, and then the average damping ratio is obtained by averaging.

$$\zeta ={\displaystyle \frac{1}{2\pi }}\ln {\displaystyle \frac{A_{\mathrm{m}}(t)}{A_{\mathrm{m}}(t+T)}}={\displaystyle \frac{1}{2\pi \mathrm{m}}}\ln {\displaystyle \frac{A_{\mathrm{n}}}{A_{\mathrm{n}+\mathrm{m}}}}$$

(1)

where A_m(t) and A_m(t + T) are two peaks in one apart vibration period T, while A_n and A_n+m denote two peaks in m apart vibration periods.

3.2.2. Half-Power Point Method

A simple harmonic load near to the specimen’s natural frequency is applied to the specimen during a forced vibration test. The damping ratio is calculated by the half-power points method, which is demonstrated in Equation (2) and Figure 14 as a formula and schematic diagram.

$$\xi ={\displaystyle \frac{f_2-f_1}{2f_r}}$$

(2)

where f_r is the natural frequency of the structure, which corresponds to the amplitude A_m on the frequency spectrum; f₁ and f₂ are the frequency values corresponding to A_m/$\sqrt{2}$, respectively.

3.2.3. Power Spectrum Decrement Method

Feng et al. [46] proposed a modal damping ratio identification method based on the frequency-domain characteristics of structural free vibration responses, termed the Power Spectrum Decrement Method (PSDM). It aims to address the limitations of traditional methods in scenarios such as multi-modal vibration and noise interference. Equation (3) is the formula for calculating the damping ratio. During calculation, the entire free decay signal only needs to be split into two segments: the power spectra of the two segments are computed using the Welch method and then substituted into the formula. No modal decomposition is required, making the process simple and efficient.

$$\xi ={\displaystyle \frac{ln\left(S_{y_s(t)}(f)/S_{y_s(t+\Delta t)}(f)\right)}{\sqrt{{\left(4\pi f_d\Delta t\right)}^2+l{n}^2\left(S_{y_s(t)}(f)/S_{y_s(t+\Delta t)}(f)\right)}}}$$

(3)

where $S_{y_s(t)}(f)$ and $S_{y_s(t+\Delta t)}(f)$ are the power spectrum amplitudes of the free vibration signal $y_s\left(t\right)$ of the single-degree-of-freedom (SDOF) system in the initial period and the period after a time increment Δt, respectively; f_d is the low-damping modal frequency of the SDOF system; Δt is the time increment between the two analysis periods, i.e., the duration of N vibration cycles.

This study adopts the sudden release method to obtain structural free decay vibration responses. Based on this experimental characteristic, the applicability of three damping identification methods—namely the free decay method, half-power point method, and power spectrum decrement method—is analyzed as follows. The half-power point method identifies damping based on the bandwidth characteristics of the frequency response function (FRF), with the prerequisite of obtaining the FRF under steady-state harmonic excitation. However, this study obtains free decay vibration signals rather than steady-state forced vibration responses, which fails to satisfy the fundamental assumptions of this method; therefore, it is not applicable. The power spectrum decrement method identifies damping by analyzing the decay characteristics of the response power spectrum, with its primary advantage lying in effectively handling multi-modal coupling and strong noise interference scenarios, making it suitable for field in situ testing and other working conditions with low signal-to-noise ratios. In contrast, the tests in this study were conducted in a laboratory environment, where high-precision accelerometers were employed and environmental noise interference was effectively controlled. The free decay method directly calculates the damping ratio using the amplitude ratio of time-domain decay waveforms, featuring explicit physical significance and a simplified calculation procedure, and is particularly suitable for scenarios requiring batch processing of large quantities of data. Therefore, the free decay method was adopted in subsequent calculations in this study, which can significantly improve data processing efficiency while ensuring calculation accuracy.

3.3. Stiffness of the Specimen

The stiffness K_i of the specimen during the loading process is based on Equation (4).

$$K_i={\displaystyle \frac{\left|+F_i\right|+\left|-F_i\right|}{\left|+X_i\right|+\left|-X_i\right|}}$$

(4)

where: +F_i and −F_i stands for the peak at the ith positive and negative loads, +X_i and −X_i is the displacement at the ith positive and negative loads. K_i represents the stiffness of the specimen at the ith loading.

4. Calculation and Analysis of Test Data

This section will systematically judge and scientifically define the damage degree of the structure at different stress stages by combining the actual stress response of the specimen during the whole loading process and the macroscopic damage characteristics after loading. As shown in Figure 15, the skeleton curves of the structure in the X- and Y-directions were constructed based on the test data in accordance with JGJ/T 101-2015 “Specification for Seismic Test of Buildings” [47].

On this basis, three representative stress states of the structure, namely intact, moderately damaged, and severely damaged, are selected to describe and summarize the typical failure phenomena and damage characteristics at each stage in detail. Meanwhile, the dynamic characteristic indexes such as the overall stiffness in X- and Y-directions, the natural frequency, and damping ratio of single frame (denoted as ①, ②, and ③, respectively, in the following text) and the whole structure under each typical state are calculated and analyzed, respectively, and the systematic statistical and comparative analysis of the calculation results is carried out to reveal the evolution law of the mechanical properties of the structure at different damage stages.

4.1. Intact Specimen

4.1.1. Damage Phenomena and Stiffness

The specimen under the unloaded condition is defined as the intact state, in which the specimen structure is complete without obvious visible damage. The initial stiffness of the specimen is 54 kg/mm in the X-direction and 13.5 kg/mm in the Y-direction.

4.1.2. Natural Frequency

According to the theoretical calculation method described above, the natural frequencies of each frame and the whole structure of the specimen in X- and Y-directions under the intact state are obtained. Three different waveforms (No. 01, 02, 03) are selected for calculation for each frame, and the average value, variance, and 95% confidence interval are counted. The results are shown in

Table 3. Calculation of natural frequency in X-direction of intact specimen.

Frame No.	Waveform No.	Natural Frequency (Hz)	Average Value (Hz)	Variance	95% Confidence Interval (Hz)
①	01	11.6	12.067	0.1156	(11.03, 13.10)
	02	12.4
	03	12.2
②	01	10.6	10.067	0.2489	(8.55, 11.58)
	02	10.2
	03	9.4
③	01	12.2	11.4	0.7467	(8.77, 14.03)
	02	11.8
	03	10.2
Overall average natural frequency (Hz)			11.178

and

Table 4. Calculation of natural frequency in Y-direction of intact specimen.

Frame No.	Waveform No.	Natural Frequency (Hz)	Average Value (Hz)	Variance	95% Confidence Interval (Hz)
①	01	6.2	6.133	0.0622	(5.37, 6.89)
	02	6.4
	03	5.8
②	01	6.4	6.2	0.08	(5.34, 7.06)
	02	5.8
	03	6.4
③	01	6.4	6.267	0.1156	(5.23, 7.30)
	02	5.8
	03	6.6
Overall average natural frequency (Hz)			6.2

and Figure 16. The subsequent numerical processing methods are the same, and only the average values will be given directly.

Calculated from 3 groups of valid test waveforms, the average natural frequencies of Frame ①, ②, and ③ in the X-direction are 12.067 Hz, 10.067 Hz, and 11.4 Hz, respectively, and the overall average natural frequency of the structure in the X-direction is 11.178 Hz; the average natural frequencies of the three frames in the Y-direction are 6.133 Hz, 6.200 Hz, and 6.267 Hz in turn, and the overall average natural frequency of the structure in the Y-direction is 6.200 Hz.

It can be seen from the statistical results that there are certain differences in the discreteness of the test data in the X- and Y-directions. In the X-direction, the variance and the width of the 95% confidence interval of Frame ③ are significantly larger than those of Frame ① and ②, with a relatively high degree of data dispersion; in the Y-direction, the variance and the width of the confidence interval of each frame are small, and the overall data consistency is better than that in the X-direction. Considering that the statistics in this analysis are based on 3 samples in each group, the limited number of samples is one of the possible reasons for the large dispersion of some frames. However, on the whole, the 95% confidence intervals of the frequencies of each frame reasonably cover the mean values, and the dispersion degrees are within the acceptable range of the test. The data are stable and reliable on the whole and can meet the requirements of dynamic characteristic analysis. At the same time, the coefficient of variation of the mean frequencies of the three frames is small, indicating that there is good consistency among the three frames, and the average value of the three can be used to represent the overall natural frequency of the structure.

The natural frequencies of the structure in the X- and Y-directions are significantly different, reflecting the typical directional stress characteristics. The overall natural frequency in the X-direction is significantly higher than that in the Y-direction, which is consistent with the law that the initial stiffness of the structure is larger in the X-direction and relatively smaller in the Y-direction, and conforms to the structural characteristics of the Chuan-Dou timber frame with large in-plane stiffness and weak out-of-plane stiffness. The small differences in the frequencies of each frame in the X-direction reflect the slight differences in the processing, installation, and boundary conditions of the single frame; meanwhile, the average frequencies of each frame in the Y-direction are very close, indicating that the stress perpendicular to the frame direction is more uniform and the influence of individual differences is small. The overall frequency obtained by using the average value of the three frames can effectively reduce the uncertainty caused by the dispersion of a single frame, can better characterize the initial dynamic characteristics of the Chuan-Dou timber frame and provide a benchmark for the performance comparison at different damage stages in the follow-up.

4.1.3. Damping Ratio

According to the theoretical calculation method described above, the damping ratios of each frame and the whole structure of the specimen in X- and Y-directions under the intact state are obtained. Three different waveforms (No. 01, 02, 03) are selected for calculation for each frame, and the average value, variance, and 95% confidence interval are counted. The results are shown in

Table 5. Calculation of damping ratio in X-direction of intact specimen.

Frame No.	Waveform No.	Damping Ratio (%)	Average Value (%)	Variance	95% Confidence Interval (%)
①	01	3.548	3.678	3.29 × 10−6	(3.125, 4.230)
	02	3.934
	03	3.550
②	01	3.385	3.365	6.03 × 10−7	(3.128, 3.601)
	02	3.261
	03	3.448
③	01	3.873	3.615	3.34 × 10−6	(3.059, 4.171)
	02	3.470
	03	3.502
Overall average damping ratio (%)			3.552

and

Table 6. Calculation of damping ratio in Y-direction of intact specimen.

Frame No.	Waveform No.	Damping Ratio (%)	Average Value (%)	Variance	95% Confidence Interval (%)
①	01	4.143	4.136	2.2 × 10−6	(3.684, 4.587)
	02	4.313
	03	3.950
②	01	4.277	4.481	5.04 × 10−7	(3.799, 5.164)
	02	4.373
	03	4.794
③	01	4.441	4.555	1.1 × 10−6	(3.547, 5.563)
	02	4.218
	03	5.005
Overall average damping ratio (%)			4.391

and Figure 17.

The average damping ratios of Frame ①, ②, and ③ in the X-direction are 3.678%, 3.365%, and 3.615%, respectively, and the overall average damping ratio of the structure in the X-direction is 3.552%; the average damping ratios of the three frames in the Y-direction are 4.136%, 4.481%, and 4.555%, in turn, and the overall average damping ratio of the structure in the Y-direction is 4.391%.

The discreteness of the test data in the two directions is at an extremely low level, the variances of the three frames are all at the level of 10⁻⁶~10⁻⁷, and the 95% confidence intervals closely cover the mean values, with excellent data stability. Although the statistics in this paper are based on 3 samples in each group with a limited number of samples, the dispersion degree of all frames is far lower than the conventional limit of structural dynamic testing and is completely within the acceptable range. The coefficients of variation of the mean damping ratios of the three frames are 4.7% (X-direction) and 5.1% (Y-direction), respectively, with good consistency, and the average value of the three frames can be used to characterize the overall damping ratio of the structure.

The structural damping ratio shows a significant directional difference: the overall damping ratio in the Y-direction (perpendicular to the frame, out-of-plane) (4.391%) is significantly higher than that in the X-direction (along the frame, in-plane) (3.552%), which is highly consistent with the structural characteristics of the Chuan-Dou timber frame with weak out-of-plane constraints and sufficient frictional energy dissipation of mortise and tenon joints. The difference in damping ratios of the three frames in the X-direction is small, and the damping ratios of the three frames in the Y-direction show a slight gradient increase, with limited influence of individual differences. The overall damping ratio obtained in this paper can effectively characterize the initial damping characteristics of the structure and provide a reliable benchmark for the evaluation of the energy dissipation capacity at different damage stages in the follow-up.

4.2. Moderate Damage

4.2.1. Damage Phenomena and Stiffness

The moderate damage state was achieved by controlling the loading displacement, and the specimen was determined to enter this state when the loading displacement reached 18 mm. The induction and evolution process of damage at this stage is described as follows:

Loading mechanism: Displacement-controlled low-cycle reversed loading was adopted. At the initial loading stage, initial gaps existed between the mortise-tenon joints of the timber frame; these gaps gradually closed during the loading process, accompanied by slight sounds. As loading continued, the limiting grooves gradually became effective, restricting column foot slip to a certain extent, and the timber frame emitted obvious force-bearing sounds, indicating that the structure had entered a fully stressed state.

Damage distribution and characteristics: Damage was primarily concentrated at the mortise-tenon joints and column foot locations, with no obvious damage observed in the column bodies or infill regions. The specific manifestations are as follows:

a.

Mortise-tenon joints: Significant joint rotation and tenon pull-out were observed, with multiple splits occurring at the half-tenon tenon heads; the first-story Chuan-fang of Frame No. ② exhibited obvious splitting within the loading plane, with crack lengths of approximately 3–5 cm.

b.

Column feet: Slip occurred universally at the column feet, with partial secondary columns showing inclination; the column foot slip along the loading direction approached the limiting groove threshold, with an obvious uplift of approximately 5 mm.

c.

Inter-frame differences: Frames No. ① and No. ③ exhibited relatively light damage, while multiple mortise-tenon joints of Frame No. ② showed splitting at the mortise openings. This is attributed to the fact that during timber structure fabrication and installation, it is difficult to ensure that all columns are positioned at exactly the same horizontal level; Frame No. ② was relatively prominent in its vertical position, causing it to bear load first and sustain more concentrated damage during the loading process.

Typical damage phenomena are shown in Figure 18.

After calculation, the stiffness in the X-direction is 15.9 kg/mm and the stiffness in the Y-direction is 8.2 kg/mm in the moderately damaged state.

4.2.2. Natural Frequency

The calculation results of the average value, variance, and 95% confidence interval of the natural frequency of each frame and the whole structure of the specimen in X- and Y-directions under the moderately damaged state are shown in

Table 7. Calculation of natural frequency in X-direction of moderately damaged specimen.

Frame No.	Average Value (Hz)	Variance	95% Confidence Interval (Hz)
①	8.733	0.1689	(7.48, 9.98)
②	9	0.0267	(8.5, 9.49)
③	9.067	0.0622	(8.31, 9.83)
Overall average natural frequency (Hz)		8.933

and

Table 8. Calculation of natural frequency in Y-direction of moderately damaged specimen.

Frame No.	Average Value (Hz)	Variance	95% Confidence Interval (Hz)
①	6.133	0.0622	(5.37, 6.89)
②	5.933	0.0356	(5.36, 6.51)
③	5.6	0.1867	(4.29, 6.91)
Overall average natural frequency (Hz)		5.889

and Figure 19.

In the moderately damaged state, the average frequencies of Frame ①, ②, and ③ in the X-direction are 8.733 Hz, 9.0 Hz, and 9.067 Hz, respectively, and the overall average frequency of the structure in the X-direction is 8.933 Hz; the average frequencies of the three frames in the Y-direction are 6.133 Hz, 5.933 Hz, and 5.6 Hz, in turn, and the overall average frequency of the structure in the Y-direction is 5.889 Hz. The discreteness of the test data in the two directions is within a reasonable range: in the X-direction, the variance of Frame ① is relatively the largest (0.1689), and the discreteness of Frame ② is the smallest (0.0267); in the Y-direction, the variance of Frame ③ is the largest (0.1867), and the discreteness of Frame ② is the best. The 95% confidence intervals of each frame effectively cover the mean values, the data stability meets the analysis requirements, the mean frequencies of the three frames have good consistency, and the overall mean value can represent the natural vibration characteristics of the structure in the current state. Compared with the intact state, the natural frequencies of the structure in the X- and Y-directions decrease significantly after moderate damage, and the weakening of the in-plane (X-direction) stiffness by structural damage is more significant, which conforms to the stress damage law of the Chuan-Dou timber frame.

4.2.3. Damping Ratio

The results of the average value, variance, and 95% confidence interval of the damping ratio of each frame and the whole structure of the specimen in X- and Y-directions under the moderately damaged state are shown in

Table 9. Calculation of damping ratio in X-direction of moderately damaged specimen.

Frame No.	Average Value (%)	Variance	95% Confidence Interval (%)
①	4.679	3.18 × 10−6	(4.137, 5.221)
②	4.947	6.59 × 10−6	(4.166, 5.728)
③	4.995	6.27× 10−7	(4.754, 5.236)
Overall average damping ratio (%)		4.874

and

Table 10. Calculation of damping ratio in Y-direction of moderately damaged specimen.

Frame No.	Average Value (%)	Variance	95% Confidence Interval (%)
①	5.503	6.89 × 10−6	(4.704, 6.302)
②	5.149	1.47 × 10−5	(3.983, 6.316)
③	5.591	1.65 × 10−6	(5.201, 5.982)
Overall average damping ratio (%)		5.415

and Figure 20.

Under moderate damage, the average damping ratios of Frame ①, ②, and ③ in the X-direction are 4.679%, 4.947%, and 4.995%, respectively, and the overall average damping ratio of the structure in the X-direction is 4.874%; the average damping ratios of the three frames in the Y-direction are 5.503%, 5.149%, and 5.591%, in turn, and the overall average damping ratio of the structure in the Y-direction is 5.415%. The discreteness of the test data in the two directions is at an extremely low level, the variances of the three frames are all at the level of 10⁻⁵~10⁻⁷, and the 95% confidence intervals closely cover the mean values, with excellent data stability and fully meeting the analysis requirements. The mean damping ratios of the three frames have good consistency, and the overall mean value can effectively represent the damping characteristics of the structure in the current state. Compared with the intact state, the damping ratios of the structure in the X- and Y-directions increase significantly after moderate damage, which is consistent with the law of enhanced frictional energy dissipation of mortise and tenon joints and member deformation energy dissipation after structural damage; in addition, the damping ratio in the Y-direction is still significantly higher than that in the X-direction, continuing the structural characteristic of sufficient out-of-plane energy dissipation.

4.3. Severe Damage

4.3.1. Damage Phenomena and Stiffness

The severe damage state was achieved by progressively increasing the loading displacement, and the specimen was determined to enter this state when the loading displacement reached 33 mm. At this stage, damage further expanded upon the existing basis:

Loading mechanism: As the loading process advanced, the mutual extrusion sounds between columns and Fangs became more frequent and prolonged in duration, reflecting the step-by-step force-bearing mechanism of the structure: at the initial loading stage, the timber column feet first underwent slip within the allowable range of the limiting grooves; as displacement increased, the timber frame gradually assumed the primary load; after the timber frame became fully stressed, the overall column feet of the model exhibited uplift to dissipate the input energy.

Damage distribution and characteristics: Damage remained concentrated at the mortise-tenon joints and column feet, but the severity increased significantly and extended to the column bodies:

a.

Mortise-tenon joints: Tenon pull-out and rotation became more widespread, with the pull-out amount at the first-story elevation reaching approximately 5 mm, and cracking appearing at the mortise openings of multiple joints.

b.

Column feet: The uplift amount further developed to 13 mm, with slip amounts exceeding the limiting groove threshold.

c.

Column bodies: The existing cracks at Column A of Frame No. ② at the intersection with the loading plane continued to develop, and the crack widths at the junction between the Chuan-Fang and Dou-Fang expanded significantly, ultimately resulting in fracture of the timber column.

d.

Overall deformation: The model structure underwent significant overall deformation, with obvious overall inclination.

Typical damage phenomena are shown in Figure 21.

After calculation, the stiffness in the X-direction is 11.03 kg/mm and the stiffness in the Y-direction is 4.9 kg/mm in the severely damaged state.

4.3.2. Natural Frequency

The calculation results of the average value, variance, and 95% confidence interval of the natural frequency of each frame and the whole structure of the specimen in X- and Y-directions under the severely damaged state are shown in

Table 11. Calculation of natural frequency in X-direction of severely damaged specimen.

Frame No.	Average Value (Hz)	Variance	95% Confidence Interval (Hz)
①	8	0.1867	(6.69, 9.31)
②	7.67	0.0356	(7.09, 8.24)
③	7.73	0.1689	(6.48, 8.98)
Overall average natural frequency (Hz)		7.8

and

Table 12. Calculation of natural frequency in Y-direction of severely damaged specimen.

Frame No.	Average Value (Hz)	Variance	95% Confidence Interval (Hz)
①	5.2	0.0267	(4.70, 5.69)
②	5.27	0.0622	(4.51, 6.03)
③	5	0.0267	(4.50, 5.49)
Overall average natural frequency (Hz)		5.156

and Figure 22.

In the severely damaged state, the average natural frequencies of Frame ①, ②, and ③ in the X-direction are 8.0 Hz, 7.67 Hz, and 7.73 Hz, respectively, and the overall average natural frequency of the structure in the X-direction is 7.8 Hz; the average natural frequencies of the three frames in the Y-direction are 5.2 Hz, 5.27 Hz, and 5.0 Hz, in turn, and the overall average natural frequency of the structure in the Y-direction is 5.156 Hz. The discreteness of the test data in the two directions is within a reasonable range, the variances of Frame ① and ③ in the X-direction are relatively large, and the discreteness of Frame ② is the best; the variances of the three frames in the Y-direction are all at a low level with good data stability. The 95% confidence intervals of each frame effectively cover the mean values, the mean frequencies of the three frames have good consistency, and the overall mean value can represent the natural vibration characteristics of the structure in the current state. Compared with the intact and moderately damaged states, the natural frequencies of the structure in the X- and Y-directions continue to decrease after severe damage, and the structural stiffness further attenuates.

4.3.3. Damping Ratio

The results of the average value, variance, and 95% confidence interval of the damping ratio of each frame and the whole structure of the specimen in X- and Y-directions under the severely damaged state are shown in

Table 13. Calculation of damping ratio in X-direction of severely damaged specimen.

Frame No.	Average Value (%)	Variance	95% Confidence Interval (%)
①	8.533	2.75 × 10−5	(6.937, 10.128)
②	8.908	1.22 × 10−5	(7.844, 9.973)
③	9.411	7.35 × 10−6	(8.586, 10.236)
Overall average damping ratio (%)		8.951

and

Table 14. Calculation of damping ratio in Y-direction of severely damaged specimen.

Frame No.	Average Value (%)	Variance	95% Confidence Interval (%)
①	11.35	2.2 × 10−5	(9.923, 12.776)
②	12.438	7.08 × 10−6	(11.629, 13.248)
③	12.033	4.9 × 10−6	(11.36, 12.706)
Overall average damping ratio (%)		11.94

and Figure 23.

In the severely damaged state, the average damping ratios of Frame ①, ②, and ③ in the X-direction are 8.533%, 8.908%, and 9.411%, respectively, and the overall average damping ratio of the structure in the X-direction is 8.951%; the average damping ratios of the three frames in the Y-direction are 11.35%, 12.438%, and 12.033%, in turn, and the overall average damping ratio of the structure in the Y-direction is 11.94%. The discreteness of the test data in the two directions is within a reasonable range, the 95% confidence intervals of the three frames effectively cover the mean values with good data stability, and the overall mean value can represent the damping characteristics of the structure in the current state. Compared with the intact and moderately damaged states, the damping ratios of the structure in the X- and Y-directions increase significantly after severe damage, the energy dissipation capacity of the structure is enhanced remarkably, and the damping ratio in the Y-direction is always higher than that in the X-direction, continuing the directional energy dissipation characteristic of the structure.

5. Discussion and Reflection on Test Results

5.1. Comparative Analysis of Results

The statistics of stiffness, natural frequency, and damping ratio of the specimen in each direction under different damage states are shown in

Table 15. Summary of stiffness, natural frequency, and damping ratio in X-direction at different damage stages.

Damage Stage	Stiffness (kg/mm)	Natural Frequency (Hz)	Damping Ratio (%)
Intact	54	11.178	3.552
Moderately damaged	15.9	8.933	4.874
Severely damaged	11.03	7.8	8.951

and

Table 16. Summary of stiffness, natural frequency, and damping ratio in Y-direction at different damage stages.

Damage Stage	Stiffness (kg/mm)	Natural Frequency (Hz)	Damping Ratio (%)
Intact	13.5	6.2	4.391
Moderately damaged	8.2	5.889	5.415
Severely damaged	4.9	5.156	11.94

, respectively.

The evolution curves of stiffness, natural frequency, and damping ratio of the specimen in each direction under different damage states are shown in Figure 24.

As shown in

Table 17. Equivalent viscous damping coefficients at different damage stages.

Damage Stage	X	Y
Moderately damaged	0.125	0.119
Severely damaged	0.147	0.173

, in accordance with JGJ/T 101-2015 “Specification for Seismic Test of Buildings” [47], the energy dissipation capacity of the specimen was calculated based on the area enclosed by the load–deformation hysteresis curve, expressed by the equivalent viscous damping coefficient. Although the change rate of this coefficient does not completely equate to that of the damping ratio, it can reflect the enhancement trend of the specimen’s energy dissipation capacity to a certain extent, and this trend is consistent with the variation law of the damping ratio.

5.1.1. Variation Law and Physical Significance of Each Index

The stiffness decreases continuously with the aggravation of damage degree, and the decrease rate shows the characteristics of “rapid decline in the intact—moderately damaged stage and slowdown in the moderately—severely damaged stage.” The stiffness in the X-direction decreases from 54 kg/mm to 11.03 kg/mm with a cumulative decrease of 79.6%; the stiffness in the Y-direction decreases from 13.5 kg/mm to 4.9 kg/mm with a cumulative decrease of 63.7%. The continuous decrease of stiffness directly reflects the continuous weakening of the structure’s ability to resist lateral deformation, the weakening of the constraint of mortise and tenon joints, and the degradation of the force transmission performance between members, which is an intuitive embodiment of the accumulation of structural damage and stiffness degradation.

The natural frequency is highly synchronized with the change of stiffness, showing a monotonous decreasing trend as a whole. The natural frequency in the X-direction decreases from 11.178 Hz to 7.8 Hz with a cumulative decrease of 30.2%; the natural frequency in the Y-direction decreases from 6.2 Hz to 5.156 Hz with a cumulative decrease of 16.8%. According to the dynamic characteristic relationship of the structure, the decrease in frequency essentially corresponds to the attenuation of the overall stiffness of the structure, which can be used as a dynamic identification index of the structural damage degree. The lower the frequency, the more significant the degradation of the overall stiffness of the structure.

The damping ratio increases continuously with the development of damage, and the increase range increases significantly in the severely damaged stage. The damping ratio in the X-direction increases from 3.552% to 8.951% with a cumulative increase of 152.0%; the damping ratio in the Y-direction increases from 4.391% to 11.94% with a cumulative increase of 171.9%. The increase of damping ratio means the enhancement of the energy dissipation capacity of the structure, which is mainly due to the repeated slip, extrusion, and friction of mortise and tenon joints, as well as the increase in member crack propagation and interface contact energy dissipation, reflecting the typical seismic mechanism of wooden structures relying on plastic and frictional energy dissipation in the process of damage development.

The reference [48] conducted shaking table tests on traditional timber frames, similarly observing the pattern that the damping ratio continuously increased with damage development, thereby validating the effectiveness of the damping ratio as a damage identification indicator for timber structures. In the reference [30], the damping ratio of masonry walls increased by approximately 900%, while the reference [31] studied different types of masonry walls, with the damping ratio increase ranging from 234% to 888% from the initial state to the failure state.

The damping ratio increase in the timber structure in this study is significantly lower than that in masonry structures. The reasons for this discrepancy are as follows: masonry walls primarily rely on cracking and slip at mortar interfaces for energy dissipation, exhibiting pronounced brittle failure characteristics, with concentrated damage development leading to a steep increase in the damping ratio; whereas the Chuan-Dou timber structure in this study employs friction and slip at mortise-tenon joints as the core energy dissipation mechanism, with damage developing dispersedly among multiple joints, coupled with the stable viscoelasticity of wood itself, resulting in a more gradual increase in the damping ratio. The magnitude of this increase is consistent with the seismic mechanism characteristics of flexible connections and progressive energy dissipation inherent to timber structures.

5.1.2. Sensitivity Comparison Between Natural Frequency and Damping Ratio

Comparing the overall variation rates of natural frequency and damping ratio from the intact state to the severely damaged state, it can be seen that the variation range of damping ratio is much larger than that of natural frequency.

The cumulative decrease in natural frequency in the X-direction is 30.2% and 16.8% in the Y-direction, both showing a gentle and stable attenuation; while the cumulative increase in damping ratio in the X-direction reaches 152.0% and even 171.9% in the Y-direction, and the relative variation rate of damping ratio is close to 5~10 times that of frequency. This law indicates that the natural frequency can stably reflect the degradation of structural stiffness and is suitable as an index for damage identification and state evaluation; while the damping ratio is more sensitive to the later damage development, crack propagation, joint slip, and other non-linear behaviors, and can more significantly reflect the enhancement of energy dissipation capacity of the structure after entering the non-linear stage. The combination of the two can comprehensively characterize the evolution of the whole dynamic characteristics of wooden structures from elasticity to damage and then to severe damage.

5.1.3. Comparison of Dynamic Characteristics in X- and Y-Directions

In the initial state, the stiffness in the X--direction (54 kg/mm) is about 4 times that in the Y-direction (13.5 kg/mm), and the initial natural frequency in the X-direction (11.178 Hz) is also significantly higher than that in the Y-direction (6.2 Hz), indicating that the Chuan-Dou timber frame has a more complete stress system and stronger mortise and tenon constraints in the in-plane (X-direction), which is the main anti-lateral force direction of the structure; while the Y-direction is perpendicular to the frame direction with weak constraints and small stiffness, and the dynamic characteristics are relatively gentle.

From the perspective of damage evolution, the decrease ranges of stiffness and natural frequency in the X-direction are both larger than those in the Y-direction: the cumulative decrease in stiffness in the X-direction is 79.6%, higher than 63.7% in the Y-direction; the cumulative decrease in frequency in the X-direction is 30.2%, higher than 16.8% in the Y-direction, indicating that the in-plane stress direction is more sensitive to damage, and damage such as joint tenon pull-out and mortise opening cracking has a more obvious weakening effect on the in-plane stiffness. The damping ratio in the Y-direction is higher than that in the X-direction at all stages, and the final cumulative increase (171.9%) is larger than that in the X-direction (152.0%), indicating that the relative movement between members in the out-of-plane direction is more sufficient, and the slip and friction energy dissipation effect is more prominent. On the whole, the X-direction is characterized by “high stiffness, high sensitivity and fast damage attenuation,” and the Y-direction is characterized by “low stiffness, low sensitivity and stronger energy dissipation capacity.”

5.2. Discussion and Reflections

5.2.1. Discussion on Practical Applicability of the Research

The research results of this paper are intended to provide new ideas and references for the daily use damage assessment of Chuan-Dou timber frame buildings, the post-earthquake residual seismic capacity assessment, and the establishment of a rapid post-earthquake safety identification method based on the change in dynamic characteristics. In terms of practical engineering application, the following discussions and reflections are carried out:

(1)

Optimal Sensor Placement

The optimal layout of sensors is a key issue in structural health monitoring. Scholars at home and abroad have proposed a variety of efficient layout methods from the perspectives of algorithm optimization and step-by-step layout [49,50], which provide an important basis for the sensor layout design in this paper.

In the health monitoring and damage identification of the actual engineering of Chuan-Dou timber frame, the sensor layout should be optimized in combination with its typical structural characteristics and dynamic response laws. It is suggested to arrange measuring points with the key stress parts, easy damage areas, and overall deformation sensitive areas as the core: sensors should be arranged at the mortise and tenon joints such as column–purlin and purlin–purlin first to capture typical damages such as joint tenon pull-out, relative slip, and stiffness degradation; strain or displacement sensors should be arranged at the key sections of column bases, the lower middle part of column bodies, and main flexural members to monitor member bending deformation, crack propagation, and column base lifting and slip; at the same time, acceleration sensors should be reasonably arranged to obtain the overall dynamic characteristic parameters such as natural frequency and damping ratio of the structure. The layout should follow the principles of densifying key parts, combining global monitoring with local monitoring, and simplifying redundant measuring points. On the premise of ensuring the accuracy of damage identification, the number of measuring points is reduced, and sensing equipment with convenient installation, little disturbance to the original structure, and strong durability is preferred, so as to realize the effective monitoring of the whole process of structural stress state and damage evolution and provide a scientific basis for the safety assessment, reinforcement and repair, and intelligent operation and maintenance of traditional Chuan-Dou timber frames.

(2)

Establishment of Structural Health Archives for Dynamic Assessment of Routine Service Conditions

Based on representative building sampling surveys in the target region, structural dynamic parameters such as fundamental frequency and damping ratio are obtained through dynamic testing, combined with detailed records of existing damage, to form a comparable and traceable health database, providing a basis for safety assessment of existing buildings and post-earthquake rapid screening. The specific procedures are as follows:

In the target region (e.g., villages and towns in Southwest China), representative Chuan-Dou timber buildings are selected as samples through stratified sampling by construction era. The sampling principles comprehensively consider the following factors: construction era is classified into three categories—newly built (less than 10 years), aged (30 to 50 years), and severely aged (more than 50 years)—to reflect the effects of natural material aging and long-term load accumulation; structural configurations cover typical constructions that are both quantitatively substantial and geographically widespread in the locality, such as three-bay-two-room and five-bay-four-room layouts; material types include different wood species such as Chinese fir, pine, and cypress to account for variations in material properties; geographical distribution covers different townships or villages to eliminate local environmental biases.

Detailed surveys and tests are conducted for each sample building: existing damage records are obtained through a combination of visual inspection and simple tools, systematically documenting material deterioration (decay, insect damage, cracking), joint loosening (tenon pull-out amount, mortise opening deformation), member deformation (column body inclination, beam frame deflection), and other damage information, which are then classified and archived by damage severity. Dynamic parameter tests are performed by deploying wireless accelerometers at key locations (column tops, ridge purlins, eaves), employing ambient vibration or hammer impact excitation to obtain the first three natural frequencies and damping ratios of the structure. Environmental temperature and humidity are synchronously recorded during testing to eliminate seasonal effects.

Baseline data formation involves correlating damage survey results with dynamic parameters to establish the correspondence between damage states and dynamic fingerprints. Newly constructed buildings are regarded as the baseline for the intact state, while aged buildings of different eras reflect progressive damage accumulation under routine environmental conditions. Periodic retesting of the same building is conducted (at intervals of years), and the measured dynamic parameters are compared with the initial baseline. The frequency change rate is defined as the percentage difference between the initial fundamental frequency and the retested fundamental frequency relative to the initial fundamental frequency; the damping change rate is defined as the percentage difference between the retested damping ratio and the initial damping ratio relative to the initial damping ratio. When the frequency reduction exceeds a certain threshold or the damping increase exceeds a certain threshold, the structure is determined to have undergone significant performance degradation under routine service conditions, and detailed inspection is recommended.

(3)

Post-Earthquake Safety Assessment and Rapid Screening of Buildings

Under the premise of established structural health archives, post-earthquake safety assessment can achieve rapid screening by leveraging the archive baseline data. Following an earthquake, emergency retesting is conducted on buildings within the archive, and the measured dynamic parameters are directly compared with the pre-earthquake baseline, enabling rapid determination of building safety status without awaiting detailed inspection.

During retesting, the existing sensor deployment locations (key positions such as column tops, ridge purlins, and eaves) are utilized. By comparing the spatial distribution characteristics of parameter changes at each measurement point, simultaneous achievement of damage state determination and preliminary localization of damage locations is realized. If the frequency reduction at bottom-story column tops is significantly greater than that at upper stories, column foot slip with tenon pull-out or foundation settlement may be suspected; if the frequency of a particular frame abnormally decreases while adjacent frames show minimal change, loosening of mortise-tenon joints or cracking of the Chuan-Fang in that frame may be suspected; if the frequency difference between the two principal axes in the plane suddenly increases, overall structural torsion or local infill wall collapse may be suspected. Thus, while rapidly classifying buildings, preliminary identification of damage concentration areas can be achieved, providing targeted guidance for subsequent detailed assessment.

Based on the deviation degree of parameter change rates relative to the baseline, buildings are classified into three disposal zones. Buildings with change rates below the threshold and uniform distribution among measurement points are designated as green zones, indicating safe continued use; buildings with change rates approaching the threshold or exhibiting local measurement point anomalies are designated as yellow zones, requiring restricted use and intensified monitoring; buildings with change rates exceeding the threshold or showing significant local anomalies are designated as red zones, requiring immediate evacuation and initiation of detailed assessment. This classification standard relies on individualized baselines rather than uniform thresholds, effectively eliminating misjudgments arising from differences in initial states among different buildings and significantly enhancing the efficiency and reliability of post-earthquake emergency assessment.

5.2.2. Reflections on Uncertainties and Future Research Directions

Based on shaking table tests of a 1:7 scaled Chuan-Dou timber frame model, this study proposes a damage identification method with natural frequency and damping ratio as core indicators. Although the test results validate the effectiveness of this method, cautious consideration of the following uncertainty factors and limitations is still required when extending the research findings to engineering practice, and future research directions should be clarified accordingly.

(1)

Uncertainties Associated with Scale Effects

This study employs a 1:7 scaled model, a proportion commonly adopted in existing timber structure test research; however, scale effects remain a non-negligible source of uncertainty. The similitude relationship between the model and prototype is primarily manifested in geometric dimensions, material density, and elastic modulus, whereas microscopic mechanical behaviors such as joint friction, mortise-tenon gaps, and wood anisotropy are difficult to scale proportionally. Specifically, the contact area and friction force proportions at mortise-tenon joints, the absolute size effects in the wood grain direction, and the scale dependency of crack propagation in the scaled model may all lead to discrepancies between the model and prototype in dynamic characteristics, damping mechanisms, and damage evolution paths. For example, the joint loosening threshold of the scaled model may be lower than that of the prototype, while the initiation and propagation velocities of wood cracks may be influenced by absolute dimensions. Future research should conduct comparative tests with different scale ratios (e.g., 1:3, 1:5, 1:7, 1:10) to systematically analyze the influence of scale effects on frequency identification accuracy, damping ratio variation patterns, and damage thresholds and to establish quantitative conversion relationships between scaled models and prototypes, thereby providing a basis for the scale extension of the damage identification method.

(2)

Influence of Environmental Factors

As a hygroscopic material, wood exhibits extreme sensitivity of moisture content, density, and stiffness to environmental temperature and humidity variations. Studies have shown that when moisture content increases from 12% to 18%, the dynamic elastic modulus of wood may decrease by 5% to 10%, with a corresponding reduction in natural frequency; seasonal fluctuations in environmental temperature also cause thermal expansion and contraction of wood, indirectly altering joint gaps and structural stiffness. The tests in this study were conducted under constant temperature and humidity laboratory conditions, and control measures such as specimen sealing and wrapping were implemented, effectively reducing the interference of environmental fluctuations on short-term testing. However, in actual engineering applications, traditional timber structures are exposed to natural environments where seasonal variations in temperature, humidity, and moisture content are significant. Environmentally induced frequency drift may be comparable to or even greater than damage-induced frequency reduction in magnitude, thereby creating misjudgment risks for damage identification. Future research should specifically design environmental control tests to systematically quantify the variation patterns of dynamic parameters of timber frames under different moisture contents (e.g., 8%, 12%, 16%, 20%), temperatures (e.g., 0 °C, 20 °C, 40 °C), and humidity conditions and establish environmental compensation models. Simultaneously, environmental parameters should be synchronously recorded during long-term monitoring, and “environment-damage decoupling” algorithms (such as normalization methods based on baseline models) should be employed to distinguish environmental fluctuations from true damage effects, enhancing the robustness of the method.

(3)

Influence of Excitation Methods

This study adopts the sudden release method to obtain structural free decay responses. The advantages of this method lie in its controllable excitation, good repeatability, and simplified data processing, making it suitable for laboratory batch testing. However, excitation sources in actual engineering possess diversity and randomness, including ambient vibrations (wind loads, human activities), seismic ground motions, hammer impact excitations, and shaker sweep excitations. Different excitation methods exhibit significant differences in spectral characteristics, energy distribution, and duration of action, which may affect the signal-to-noise ratio of structural responses and the identification accuracy of modal parameters. For example, seismic excitation is a broadband random process that may excite higher-order modes, while ambient vibration energy is relatively low and insufficiently sensitive to minor damage. The method proposed in this study is based on single-degree-of-freedom approximation for fundamental frequency and damping ratio identification, without sufficient validation of its applicability under multi-modal coupling, non-stationary excitation, and strong noise interference scenarios. Future research should consider comparative tests with multiple excitation methods (such as white noise sweep, hammer impact excitation, electromagnetic shaker, and seismic simulation shaking table) to analyze the identification stability of frequency and damping ratio under different excitation conditions, establish correspondence relationships between excitation methods and identification accuracy, and explore adaptive signal processing algorithms applicable to complex field environments.

(4)

Limitations of Structural Types and Configurations

The research object of this study is a typical Chuan-Dou timber frame (three bays and two rooms, five main columns and four secondary columns) without infill walls (wood panel walls, brick walls, etc.), representing a relatively ideal bare frame condition. In actual rural timber structures, the presence of infill walls significantly influences structural stiffness distribution, mass distribution, and energy dissipation mechanisms: wood panel walls can enhance in-plane stiffness and increase damping, brick walls may cause torsional effects due to stiffness non-uniformity, and the interaction between infill walls and timber frames is particularly complex during earthquakes (such as early cracking of infill walls causing sudden stiffness reduction). Furthermore, timber structure types vary across different regions (such as Tai-liang style, Jing-gan style, and Chuan-Dou–Tai-liang hybrid style), joint constructions differ (such as through tenons, half-tenons, and dovetail tenons), and wood species variations (Chinese fir, pine, cypress, etc.) lead to significant material property scatter. The scale effects and damage mechanisms of large-span spatial timber structures and ancient buildings are even more complex. The damage thresholds proposed in this study are based on a scaled model with a single structural form, single wood species, and no infill walls; their universality remains to be verified. Future research should extend to systematic tests of different structural types (with infill walls, different joint forms, different wood species), different spans (from residential buildings to halls), and ancient buildings (with historical damage and repair interventions), establishing categorized damage identification threshold databases. Simultaneously, combined with parametric numerical simulation analysis, the influence patterns of infill wall stiffness, joint types, and span ratios on the dynamic parameter–damage relationship should be investigated, laying foundation for the engineering extension of the method.

(5)

Engineering Calibration of Damage Identification Thresholds and Alignment with Codes

Based on frequency reduction and damping increase across three damage states (intact, moderate damage, severe damage), this study proposes preliminary threshold ranges. However, these thresholds are derived from step-by-step loading tests of a single model with limited sample sizes and do not account for spatial non-uniformity of damage (such as differences between local joint damage and overall member damage). In engineering applications, threshold setting must balance sensitivity and reliability: thresholds that are too low may cause false alarms, while thresholds that are too high may result in missed detections.

The research findings of this study can serve as a supplement and auxiliary means to existing code-based methods for post-earthquake safety assessment. Future work can further clarify the frequency and damping ratio intervals corresponding to different code-based damage grades (such as intact, minor damage, moderate damage, severe damage, collapse), establishing quantitative mapping relationships between “damping ratio-frequency” indicators and code-based damage grades through large-sample test data, enabling dynamic characteristic indicators to directly align with code terminology systems. Simultaneously, this can be incorporated as an auxiliary determination procedure in code-based safety assessment procedures, with “damping ratio-frequency” dynamic indicators serving as supplementary inputs to existing code-based safety assessment workflows. For example, after completing field visual inspection and bearing capacity verification in accordance with codes, the hierarchical determination results obtained from this study can be introduced to further clarify the safety grade of structures in “boundary states” (such as between minor damage and moderate damage) using the quantitative characteristics of frequency and damping ratio, thereby improving overall assessment accuracy and reliability.

Future research can calibrate thresholds through field measurements; simultaneously, probabilistic statistical methods (such as confidence interval analysis and Bayesian inference) can be introduced, considering uncertainties in material scatter, test errors, and environmental fluctuations, to establish statistically meaningful damage probability models, enabling thresholds to evolve from deterministic criteria to reliability-based criteria.

6. Conclusions

This study takes a typical Chuan-Dou timber structure in Southwest China as the research object. A specimen was fabricated at a 1:7 scale and subjected to combined pseudo-static and dynamic tests. The evolution patterns of stiffness, natural frequency, and damping ratio in the X- and Y-directions were systematically investigated under three typical states—intact, moderate damage, and severe damage. The sensitivity differences between frequency and damping ratio as damage identification indicators and the dynamic characteristic distinctions between in-plane and out-of-plane directions of the structure were comparatively analyzed. The main conclusions are drawn as follows:

(1)

The dynamic characteristics of Chuan-Dou timber structures exhibit regular evolution with increasing damage severity. Stiffness degrades continuously, with cumulative reductions of 79.6% in the X-direction (from 54 kg/mm to 11.03 kg/mm) and 63.7% in the Y-direction (from 13.5 kg/mm to 4.9 kg/mm), displaying a characteristic of rapid decrease from intact to moderate damage state followed by gradual attenuation from moderate to severe damage state. Natural frequency decreases monotonically, with reductions of 30.2% in the X-direction (from 11.178 Hz to 7.8 Hz) and 16.8% in the Y-direction (from 6.2 Hz to 5.156 Hz). The damping ratio increases significantly, with increases of 152.0% in the X-direction (from 3.552% to 8.951%) and 171.9% in the Y-direction (from 4.391% to 11.94%). Among the three indicators, stiffness and frequency change are highly synchronized, both intuitively reflecting the cumulative damage degree of the structure, while the damping ratio exhibits a more pronounced increase at later stages.

(2)

Significant differences exist in damage identification sensitivity between natural frequency and damping ratio. From intact to severe damage state, the cumulative reductions of natural frequency in the X- and Y-directions are 30.2% and 16.8%, respectively, showing a gradual attenuation trend; whereas the cumulative increases of damping ratio in the X- and Y-directions reach 152.0% and 171.9%, respectively, with the increase becoming significantly more pronounced at the severe damage stage. The relative change rate of the damping ratio is approximately 5 to 10 times that of the natural frequency, demonstrating higher sensitivity to late-stage non-linear damage (such as crack propagation and joint slip) of the structure, making it suitable as an early warning indicator for damage development. Natural frequency, due to its stable variation and high testing stability, is suitable as a fundamental assessment indicator for damage states.

(3)

The dynamic characteristics of Chuan-Dou timber structures exhibit obvious directional differences, consistent with the structural characteristic of “strong in-plane constraint and weak out-of-plane constraint.” The initial stiffness and frequency in the X-direction (along the frame direction) are 4 times and 1.8 times those in the Y-direction (perpendicular to the frame direction), respectively, representing the primary lateral force-resisting direction of the structure, and are more sensitive to damage with greater attenuation amplitudes in both stiffness and frequency than the Y-direction. However, the damping ratio in the Y-direction exceeds that in the X-direction at all damage stages, with a larger final increase, as the relative motion between out-of-plane members is more sufficient and the slip and friction energy dissipation effects are more prominent. This directional characteristic provides a basis for optimal sensor placement: frequency changes should be prioritized for monitoring in the X-direction to determine overall damage degree, while damping ratio changes should be emphasized in the Y-direction to assess energy dissipation capacity.

(4)

This study innovatively applies the frequency-damping ratio dual-index collaborative determination strategy systematically to Chuan-Dou timber structures, establishing a damage identification method based on the evolution of dynamic characteristic parameters. Through cross-validation of dual indicators, the misjudgment risk of single indicators is effectively reduced, and the engineering application paths are clarified, including optimal sensor placement strategy, establishment of structural health archives, and post-earthquake three-level rapid screening (green zone–yellow zone–red zone). This method advances the damage assessment of traditional timber structures from qualitative empirical judgment to quantitative intelligent evaluation, providing scientific reference and technical support for daily health monitoring, post-earthquake rapid on-site assessment, and residual seismic capacity evaluation of Chuan-Dou timber structures.

(5)

This study also discusses the uncertainties of current research and future research directions from five aspects: scale effects, environmental factors, excitation methods, structural types and configurations, and engineering calibration of damage identification thresholds and alignment with codes, with the aim of providing reference for subsequent related research.

The research findings of this study provide new test data and theoretical support for damage identification of Chuan-Dou timber structures. The revealed evolution patterns of dynamic characteristics lay foundation for establishing rapid post-earthquake safety assessment methods based on natural frequency-damping ratio and also provide scientific reference for daily health monitoring and residual seismic capacity evaluation of this type of structure. It is hoped that the research approach of this study can provide new exploratory perspectives for the industry in conducting damage identification research based on dynamic characteristics, stimulating more in-depth thinking and supplementary improvements and facilitating the development transformation of traditional timber structure safety assessment from qualitative empirical judgment to quantitative intelligent evaluation.