Performance of Timber-Concrete Composite (TCC) Systems Connected with Inclined Screws: A Literature Review

Abstract

Timber–concrete composite (TCC) systems present a viable alternative to conventional timber or reinforced concrete systems. TCC leverages the advantages of both materials, resulting in an enhanced composite structure. Historically, traditional mechanical connectors such as nails, bolts, and dowels have been used in TCC systems to join timber and concrete components. However, these connectors often fall short in providing sufficient load transfer efficiency. Therefore, the use of screws and, more recently, inclined screws in TCC systems has increased due to their enhanced load transfer efficiency and greater stiffness compared to traditional connections. This review paper consolidates findings from contemporary experimental studies and analytical models, examining the influence of factors such as screw type and inclination angle on the performance of TCC systems for both connection and beam specimens in ultimate and serviceability limit states. Key issues addressed include the shear strength, stiffness, and long-term behaviour of the connection type. By offering a comprehensive synthesis of existing knowledge, this paper aims to inform design practices and contribute to the development of more resilient and efficient TCC systems, supporting their increased adoption in sustainable construction.

1. Introduction

Timber–concrete composite systems (hereafter referred to as TCC) are typically one-way slab or beam elements that combine timber and concrete through a connection. TCC leverages the advantages of both materials, resulting in an enhanced composite structure with concrete part primarily in compression and timber part in tension. TCC structures were introduced in the U.S. for bridge construction in the early 20th century, with increased use after WWII due to steel shortages [1,2]. Since the early 1990s, TCC floor systems have been used in numerous mass timber structures worldwide [3], both in new construction projects and for the refurbishment of existing timber structures, especially older buildings in Europe [4]. TCC systems reduce dead load, lowering seismic and foundation demands [5,6]. Additionally, this system enhances acoustic performance while eliminating the need for separate formwork by casting concrete over timber panels such as cross-laminated timber (CLT) [7,8].

The main purpose of connections in composite systems is to transfer the shear forces between materials while minimizing slip. Various connection types have been used in TCC systems, including connections with mechanical fasteners, notched connections, combined notch–dowel connections and glued connections [5]. Based on their installation, connectors can be categorized as discrete/continuous or vertical/inclined. Mechanical fasteners positioned discretely along the length of a beam or slab create a partially composite system with a non-linear relationship between shear force and relative slip [5,9]. Conversely, glued connections typically form a stiff, almost fully composite system with essentially a linear–elastic response until brittle failure [10]. Over the past five decades, nails, bolts, and dowels have been widely used as mechanical fasteners. However, in recent years, experimental studies have increasingly focused on self-tapping screws, as they offer quick installation and improved mechanical behaviour [11,12]. Additionally, research has shown that the orientation of mechanical fasteners considerably influences connection strength and stiffness [13,14].

Currently, few design standards address the design of TCC systems; some examples are part 2 of Eurocode 5 (EC5)- Design of Timber Structures [15], the Canadian Highway Bridge Design Code (CSA S6) [16] for TCC bridge deck design, and the recently published 2024 edition of the Canadian Timber Design Standard (CSA O86) [17]. As with any structural system, TCC system design must meet both Serviceability Limit States (SLS) and Ultimate Limit States (ULS). ULS design focusses on system strength, while SLS primarily addresses deflection in the short- and long-term as well as vibration [9]; however, available studies have shown that, due to the rheological characteristics of timber, concrete, and the connection between them, TCC design is usually governed by SLS requirements, which must be thoroughly checked to ensure acceptable performance under service conditions [18,19,20]. Yeoh et al. [3] proposed that TCC design must consider two major considerations: (1) partial composite action arising from the flexibility of shear connectors and (2) time-dependent properties of the component materials.

This article reviews publications that focus on TCC systems connected by inclined self-tapping screws. It begins by summarizing and comparing short-term experimental works and the analytical models proposed for the prediction of the mechanical behaviour of inclined screws in TCC systems. The article then explores long-term experiments conducted on inclined screws in TCC systems and compares the influences of key determinants, such as screw angle, component properties, and environmental conditions, on their long-term performance along with the prediction models. This study only focuses on static loading to isolate key factors influencing the short- and long-term behaviour of inclined self-tapping screws. Dynamic loading, while relevant, introduces complexities like mass, damping, and resonance effects, which were beyond this study’s scope.

2. Short-Term Performance of Inclined Screws in TCC Systems

2.1. Short-Term Experimental Work

Empirical examinations of TCC systems are usually conducted on their connections or assembly (beam or slabs). These tests are designed to evaluate the strength, stiffness, and failure modes of the system. TCC beams are typically tested using a four-point bending test method, while the mechanical behaviour of the connections is commonly evaluated through a shear (push-out) test. This testing approach has been used to investigate connections in composite systems such as TCC as well as in other systems such as steel–concrete composite beams, since the 1950s [21,22]. From the literature, no standardized tests for TCC connection are available. Consequently, researchers have adopted principles outlined in EN 26891 [23] initially developed for timber-to-timber connections with mechanical fasteners [9,19]. CSA O86 [17] also recommends the testing protocols from EN 26891 [23] and ASTM D6815 [24] for TCC connections that have at least two mechanical fasteners in each shear plane. The outcome of a connection test is load-slip curve, and the key mechanical properties are load-carrying capacity (strength), slip modulus (stiffness), and ultimate deformation capacity or ductility [14]. Figure 1 illustrates various methods for conducting connection tests in TCC systems with mechanical fasteners. Connection test specimens can be categorized as three-layer (double-shear) or two-layer tests [7]. Reinforcement is often added to the concrete, mainly for shrinkage and temperature crack control [25]. Additionally, an interlayer is sometimes placed between the timber and concrete to reduce sound transmission between adjacent units. This interlayer can be wood, polyethylene (PE) or foam, and usually has a thickness of 5 to 30 mm [1,26].

Analysis of the connection test results revealed that the fastener insertion angle to the shear plane can play a critical role in mechanical behaviour. Khorsandnia et al. [13] reported that using screws at a 45° angle in a crossed-pair arrangement, compared to 90° screws, increased connection stiffness by about 66%. Appavuravther et al. [27] also noted that varying the angle from 90° to 45° led to a substantial increase in stiffness, surpassing 100%. The results demonstrated the combined effect of withdrawal and bearing resistance in inclined screws, which enhances load distribution within the connection. This improved force distribution increases both stiffness and strength, ultimately boosting the composite action of the system. Additionally, Kenel and Meierhofer [28] showed that when the screw angle exceeds 45°, the connection stiffness decreases by approximately 10% for every additional 5° beyond this threshold.

Inclined screws can be used as single units or in cross pairs. Studies showed that optimal performance is achieved when they are arranged in a crossed-pair orientation [14,28,29]. In this configuration, one screw is subjected to tension and the other to compression under the applied lateral load, creating a truss-like system that maintains the stability of the fasteners and connecting members, Figure 2. According to Marchi et al. [11], coupling two screws with the same inclination and orientation relative to the shear load is the most straightforward method to double the load-bearing capacity per connection point. Also, the two main types of screws investigated in the literature were fully threaded and partially threaded, as illustrated in Figure 2.

Understanding the behaviour of a composite system requires insight into the strength and stiffness of the connections between its components.

Table 1. Summary of experimental research on TCC connections with inclined self-tapping screws.

Reference	Screw Type and Diameter	Insertion Angle	Timber Type	Concrete Type	Interlayer Thickness (mm)	Strength (kN)	Serviceability Stiffness (kN/mm)
Timmermann and Meierhofer [32]	6 mm (Partially threaded)	45° $\pm 45°$	Lumber	NWAC	20	6.0 20	5.0 14
van der Linden [1]	6 mm (Partially threaded)	$\pm 45°$	Glulam	NWAC	0 19 28	22.0 15.3 15.0	29.2 12.9 15.6
Jorge et al. [26]	6 mm (Partially threaded)	±45° ±45° 45° 45°	Glulam	LWAC	25 - 25 -	7.7 8.2 12.4 11.7	10.5 15.9 11.9 15.7
Khorsandnia et al. [13]	6 mm (Partially threaded)	$\pm 45°$	LVL	NWAC	-	32.6	54.9
Fragiacomo [31]	6 mm (Partially threaded)	45°	Lumber	LWAC	15	3.3	3.0
Marchi et al. [11]	8 mm 10 mm 12 mm 8 mm 10 mm 12 mm (Partially threaded)	45° $\pm 45°$	Glulam	NWAC	-	13.1 16.4 20.3 38.1 50.5 57.4	8.7 9.5 10.2 11.2 16.2 21.2
Mirdad and Chui [14]	11 mm (Fully threaded)	$\pm 30°$ $\pm 45°$	GLT	NWAC	0 5 15 0 5 15	32.4 31.1 22.5 28.4 26.8 22.3	128.5 65.6 52.9 74.0 49.8 24.6
Di Nino et al. [33]	8 mm (Fully threaded)	45°	Glulam	NWAC	22	11.9	11.2
Appavuravther et al. [27]	8 mm 10 mm (Fully threaded)	45°	Glulam	LWAC (Low strength)	-	3.5 5.6	4.0 5.9
Adema et al. [34]	6 mm (Partially threaded)	$\pm 45°$	LSL	NWAC	-	19.9	11.8
Tannert et al. [35]	10 mm (Fully threaded)	30° $\pm 45°$	LSL	NWAC	6	32.6 30.0	68.2 61.7

summarizes the existing literature on connection test results for inclined screws. In this table, the effects of screw size, insertion angle, material properties for concrete and timber, and insulation layer on connection strength and stiffness are compared. NWAC and LWAC are normal-weight aggregate concrete and lightweight aggregate concrete, respectively. Laminated Veneer Lumber (LVL), Glued Laminated Lumber (GLT), and Laminated Strand Lumber (LSL) were used in the experiments. It is worth mentioning that variations in connection test methods were noted according to Figure 1; for instance, Khorsandnia et al. [13] tested fully confined specimens, which, as reported, may slightly overestimate strength and stiffness due to horizontal reaction forces that limit rotation [1,30]. The ± symbol in this table represents crossed pairs of screws, and the strength and stiffness results for this configuration are for one pair of screws. Also, the study conducted by Fragiacomo [31] involved an upgrade technique for converting a residential floor to an office in the UK, and LWAC was used since the overall weight of the concrete layer could be a concern.

In most cases, the use of crossed-pair screws has been shown to enhance both strength and stiffness, particularly in connection with fully threaded screws. A comparison of screw inclination angles showed that 30° inclined screws were 38% stiffer and 12% stronger than 45° screws [14]. Similarly, Tannert et al. [35] observed that shifting from a 45° to a 30° configuration nearly doubled the connection’s strength and stiffness. This increased strength and stiffness can be attributed to the more efficient load transfer mechanism at the lower angle. The 30° angle allows for a greater contribution of withdrawal resistance than bearing resistance because withdrawal resistance generally plays a bigger role in resisting lateral load in connections with inclined screws. Also, as expected, increasing the screw size increased the strength and stiffness of connections because larger screws offer greater withdrawal and shear resistance [11,27].

In general, changing from NWAC to LWAC did not significantly affect the strength and stiffness of the inclined connections, even in cases where the concrete had lower compressive strength; however, it was suggested that it could alter the failure mode when the concrete has a lower modulus of elasticity [26,27]. This observation shows that the properties of timber and fasteners are a more dominant component than concrete properties [36]. In the study by Jorge et al. [26], changing NWAC to LWAC with similar concrete compressive strengths reduced the mean connection strength by 30–50%, which can be attributed to aggregate quality and strength or the bonding between aggregates with cement paste in this test. No significant influence on the mechanical properties of inclined screw connections was observed based on the types of wood used [14,35].

The presence of an interlayer between timber and concrete has been shown to decrease both the strength and stiffness of the connection, with a more pronounced effect on the stiffness [1,26]. The decrease can depend on the interlayer thickness and on material that can affect the bonding between each of the components and the interlayer. Mirdad and Chui [14] observed that changing the insulation layer thickness from 5 to 15 mm caused a connection strength decrease of 5 to 22% and around 50% for stiffness, and the influence was more pronounced compared to changing the screw insertion angle. Substantial reductions in stiffness and strength were also observed in timber-to-timber connections with inclined screws, as noted by De Santis et al. [12]. Also, some studies have shown that the influence of the interlayer also depends on its mechanical properties. For example, granite chips bonded to timber with epoxy glue have been shown to enhance both strength and stiffness in TCC members by improving load transfer and reducing slip at the interface [37,38]. While conventional interlayers often degrade performance, tailored solutions can improve it, highlighting the importance of interlayer material selection in TCC design.

Failure in TCC systems with mechanical fasteners typically happens due to the formation of plastic hinges in the fasteners or timber failure which may be accompanied by cracks or minor crushing in concrete at the interface [1,20]. He et al. [39] showed that by increasing the screw size, the failure mode can change from plastic hinge forming to concrete cracking, which can be attributed to higher stresses being transferred to the concrete. Sebastian et al. [40] also investigated TCC beams with inclined screws under negative bending and observed that in concrete tension zones, the connections showed more ductility. For screws inserted normal to the shear plane, the most common failure mode was due to the failure of concrete and plastic hinges forming in the screws [27].

In addition to connection tests, bending tests have also been performed on TCC systems with inclined screws. For $\pm 45°$ screws, van der Linden [1] noted that the gap between the interlayer and the timber was caused by screw pull-out, starting with those near the beam ends under high shear forces. As these screws disengaged, the load was re-distributed to the other connectors until they reached capacity, causing the gap to spread towards mid-span while widening at the ends. For the same connection type, Jorge et al. [26] observed that beam failure occurred primarily in the timber element in a brittle manner, which is expected under bending. Despite this, the composite beam demonstrated significant deformability, indicative of ductile behaviour, while cracking in the concrete was not readily apparent. This indicates that fastener yielding likely occurred before ultimate failure in the timber. Figure 3 shows some of the common failure modes in TCC systems with screw and notched connections.

2.2. Analytical Models of Strength and Stiffness

Various analytical models have been developed to predict the strength, stiffness, and failure modes of TCC systems with inclined screws which offer valuable insights into the mechanical performance of these connections. In the case of timber-to-timber connections, analytical models were introduced by Bejtka and Blaß [41] and Tomasi et al. [42]. De Santis et al. [12] also proposed an empirical model for insulated inclined screws in timber-to-timber connections that could describe the experimental load–slip curves. In this section, prominent models proposed in the literature and their effectiveness in simulating the performance of TCC systems are discussed.

The models used for calculating the strength of TCC connections generally come from the Johansen or European Yield Theory for timber-to-timber or timber-to-steel connections [43], as no specific model for the strength of TCC connections has been introduced. The primary approaches adopted for the strength models generally assume that concrete behaves as a completely rigid component while also considering different failure modes in both the timber and the fasteners [10,44]. Kavaliauskas et al. [45] extended Johansen’s Yield Model, considering failure modes in wood and screws while assuming the screw in concrete to be rigid. Although the model aligned with experimental results, the failure modes differed, likely due to limited fastener ductility and the influence of screw inclination on timber strength. In the case of using LWAC with a smaller modulus of elasticity, Appavuravther et al. [27] recommended considering concrete failure as well.

The Johansen Yield Theory is applicable only to screws oriented perpendicular to the shear plane and does not account for critical factors such as screw withdrawal strength, the angle between the screw axis and the applied force, and the friction between timber elements at the shear plane interface [42]. Therefore, the models based on Johansen’s Theory were extended to address these conditions for inclined screws. Marchi et al. [11] proposed analytical models for the strength and stiffness of single and crossed-pair screws based on the modification of the Johansen Yield Model and the model proposed by Kavaliauskas et al. [45] which also included failure in the concrete part. Their results indicated that the friction force between timber and concrete should be considered in analytical models. Other analytical models have been developed by extending the theory in earlier models to calculate the strength and stiffness of TCC connections with inclined screws and an insulation layer [33,46,47]. Mirdad and Chui [47] proposed two analytical models of inclined screws in solid and layered timber by characterizing all possible kinematical failure modes. Experimental validation with various material properties showed that the models predicted the mode of failure and strength within 10% of the experimental results. These models consider both bearing and withdrawal actions of the screws, incorporating corrections for embedment stiffness and friction between the concrete and mass timber panel. For the stiffness of inclined screws in TCC connections, Symons et al. [48] introduced an analytical model using the ‘Winkler’ or beam-on-elastic-foundation approach and modeling timber as orthogonal springs with varying stiffness. The results showed reasonable agreement with experimental data, with an overestimation of about 20%, likely due to the assumption of a fully rigid screw in concrete.

EN 26891 [23] recommends distinct equations for the serviceability and ultimate part of the load–slip curves to address the actual non-linear behaviour of the connection. Based on this test protocol, four different connection stiffness properties are defined: initial stiffness ($K_i$), SLS stiffness ($K_{0.4}\mathrm{o}\mathrm{r}{K}_s$) for two loading paths, and ULS stiffness ($K_{0.8}$). EC5- Design of Timber structures- Part 1-1 [49] in its latest draft proposes Equations (1) and (2) for calculating the serviceability stiffness of laterally loaded and axially loaded mechanical fasteners in timber-to-timber connections, respectively. Also, EC5 recommends considering a fraction of the connection stiffness in the ultimate limit state due to its nonlinear behaviour, as indicated in Equation (3). The connection reduction in stiffness is likely due to plastic behaviour in timber and screw above service load levels [32].

$$K_{SLS,v}=60{\left(0.7d\right)}^{1.7}$$

(1)

$$K_{SLS,ax}=160{\left({\displaystyle \frac{{\rho }_{mean}}{420}}\right)}^{0.85}{d}^{0.9}{l}_w^{0.6}$$

(2)

$$K_u={\displaystyle \frac{2}{3}}{K}_s$$

(3)

where ${\rho }_m$ is the mean timber density, $d$ is the screw diameter, and $l_w$ is the withdrawal length.

To consider the combined effect of laterally and axially loaded conditions in inclined screws without any gap between the members, EC5 [49] recommends that the effective slip modulus be calculated according to Equation (4) for inclined screws in tension and Equation (5) for inclined screws in cross-pair.

$$K_{SLS}=K_{SLS,v}\mathit{sin}\epsilon (\mathit{sin}\epsilon -\mu \mathit{cos}\epsilon )+{\displaystyle \frac{1}{2}}{K}_{SLS,ax}\mathit{cos}\epsilon (\mathit{cos}\epsilon +\mu \mathit{sin}\epsilon )$$

(4)

$$K_{SLS}=K_{SLS,v}{\mathit{sin}}^2\epsilon +{\displaystyle \frac{1}{2}}{K}_{SLS,ax}{\mathit{cos}}^2\epsilon$$

(5)

where $K_{\mathrm{S}\mathrm{L}\mathrm{S},\mathrm{v}}$ and $K_{\mathrm{S}\mathrm{L}\mathrm{S},\mathrm{a}\mathrm{x}}$ are the mean lateral and axial slip modulus per fastener per shear plane, respectively. $\epsilon$ is the angle between the fastener axis and the direction of the grain, and $\mu$ is the friction coefficient between the members ($\mu$ = 0.25). To the best of the authors’ knowledge, no experimental studies have yet been conducted to compare the application of these equations in practice. Such investigations would contribute to a deeper understanding of the parameters influencing the slip modulus under serviceability limit state conditions.

3. Long-Term Performance of TCC Connections and Systems with Inclined Screws

3.1. Experimental Work

As stated, TCC systems must be evaluated for both ULS and SLS design requirements. It is known that SLS requirements, such as vibration or long-term deformations, typically govern the design of TCC systems [9,18,19,20]. The main factors that influence these long-term deformations are sustained loading, concrete shrinkage, and changes in environmental conditions. Concrete exhibits viscoelastic creep under sustained loading and the linear shrinkage coefficient of concrete typically ranging from 400 × 10⁻⁶ to 900 × 10⁻⁶ mm/mm [50]. Timber also experiences viscoelastic creep under sustained load, shrinkage/swelling, and mechano-sorptive creep. Mechano-sorptive creep, in contrast to viscoelastic creep, is a strain increase caused by the combination effect of sustained loading and moisture content changes and is not directly dependent on time. According to Van de Kuilen [51], moisture content variations as small as 1% are enough to trigger mechano-sorptive deformation.

TCC systems are internally statically indeterminate; thus, deformations in one material or component cannot occur independently without inducing stresses in other components. This leads to stress redistribution and gradual increases in deflection over time [8,9,19,52,53,54,55]. Therefore, TCCs are complex systems, and their deflection is expected to gradually increase due to the creep and shrinkage that occur over time in its components [19,20,56,57]. Thus, in addition to the short-term tests, long-term tests have also been conducted on TCC systems with inclined screws, generally to establish the creep coefficient of the system (global creep coefficient) and the connection. The creep coefficient, $\phi \left(t\right)$, for the connection is generally defined as the ratio of slip at a time of interest relative to the instantaneous slip, Equation (6):

$$\phi \left(t\right)={\displaystyle \frac{s\left(t\right)-s\left(t_0\right)}{s\left(t_0\right)}}$$

(6)

where $s$ is the relative slip between timber and concrete, and $t$ and $t_0$ are the time of interest and instantaneous time, respectively. The instantaneous time, $t_0$, used for determining instantaneous slip, $s\left(t_0\right)$, is recommended to be 10 min after load application [10,26,51]. ASTM D6815 [24] considers this time as 1 min after the load application. For the long-term testing of beams, measurements are typically focused on midspan deflection, which will result in the global creep coefficient of the system. According to Equation (6), a larger instantaneous displacement leads to a smaller creep coefficient.

In the long-term TCC connection test, the sustained shear load is often set at 30% of the connection strength, reflecting typical service load levels. For TCC beams, the long-term load is usually between 12 and 15% of the bending strength, excluding the self-weight of the TCC beams, and between 15 and 20% of the bending strength, when the self-weight is included. Considering the components separately, wood and concrete can be considered to behave linear-elastically when the applied stress is below about 40% of their strengths [58,59,60,61].

In long-term creep experiments, various methods have been used to apply the sustained load. For connection tests, the most common method utilizes a lever system to amplify the load applied to samples, which allows for simultaneous loading of multiple test specimens in a series, Figure 4a [8,25,62]. Linear variable differential transformers (LVDT) or dial gauges are usually used to measure the relative slip between timber and concrete. Long-term testing of TCC beams is usually conducted using the same test setup as short-term bending tests, often using a four-point bending method with loads applied at one-third points of the span. Also, in some studies, a uniformly distributed load (UDL) was applied as the long-term load. In most long-term tests, the tests concluded when no significant changes in creep coefficient or mid-span deflection were detected [8,63].

By comparing the results of TCC systems with different connection types, it was observed that in some studies, larger long-term deflections were obtained for stiffer connection types, such as glued and notched connections based on their geometry [54,64]. Stiff connection types restrict relative movement between the layers and can create stress concentrations and residual stresses. When TCC systems with rigid connections experienced environmental changes, the increase in internal stress was more pronounced. Therefore, while different connection types may influence deflection outcomes, current design codes and standards, such as EC5 [49] and CSA O86 [17], consider a creep coefficient that is twice that of timber, regardless of the type of connection used. However, based on the literature, values of creep coefficient varied depending on the connection types.

Table 2. Comparison of creep coefficients for TCC connections with inclined screw.

Reference	Screw Angle and Diameter	Timber Type & Strength Class	Concrete Type & Strength Class	Specimen Dimensions (mm)	Environment Condition	Load Level (%) a	Time (days)	Creep Coefficient
Kenel and Meierhofer [28]	±45° d = 6 mm	Lumber (C24)	NWAC (C40/50)	120 × 400 × 500 (T-C-T) b	Constant	30	1200	0.68
Jorge et al. [8]	±45° d = 6 mm	Glulam (GL24h)	LWAC (LC20/22)	160 × 300 × 530 (C-T-C) b	Constant	30	365 600	0.51–0.64 0.69–0.91
Fragiacomo [31]	45° d = 6 mm	Lumber (C16)	LWAC (LC25/30)	100 × 375 × 550 (C-T-C)	Constant	25	421	$~$0.8

^a Percentage of the connection’s shear strength. ^b T-C-T: timber–concrete–timber, C-T-C: concrete–timber–concrete.

and

Table 3. Comparison of mid-span deflection for TCC systems with inclined screws.

Reference	Screw Angle and Diameter	Timbe Type & Strength Class	Concrete Type & Strength Class	Connector Spacing (mm)	Span (mm)	Environment Condition	Load Amount	Time (day)	Midspan Deflection (mm)
Kenel and Meierhofer [28]	±45° 45–90° d = 6 mm	Lumber (C24)	NWAC (C40/50)	100–300	4000	Variable	2 × 6 (kN)	1825	13 10.86
van der Linden [1]	±45° d = 6 mm	Glulam (GL24h)	NWAC (C25)	200	4000	Variable	40% a	1230	$~$45
Kanócz and Bajzecerová [65]	±45° d = 5 mm	Lumber (nailed) (C22)	NWAC (C25/30)	150	5000	Variable	Service load	1825	$~$8
Hailu [66]	±45° d = 5 mm	LVL (C40)	NWAC (32 MPa)	300–600	5800	Variable	1.7 kPa	1460	33

^a Percentage of the bending strength.

summarize the long-term connection and beam tests conducted on TCC systems with inclined screws. In these tables, the key parameters for comparison include environmental conditions (constant or variable), fastener diameter and insertion angle, timber and concrete material type and strength classes, beam span, and connector spacing. Long-term testing of TCC systems is typically conducted under variable environmental conditions to investigate rheological phenomena, such as mechano-sorptive creep and concrete shrinkage. Variable environmental conditions are characterized by significant temperature changes and fluctuations in relative humidity, which lead to moisture absorption or desorption, primarily affecting the timber. These environmental conditions are usually referred to as dry/wet according to CSA O86 [17] or Service Class 1 to 3 based on EC5 [49].

Under constant environmental conditions, Kenel and Meierhofer [28] observed that ±45° screw orientations exhibited less slip over time compared to a connection system with a combination of 90° and ±45° screws, which behaved similarly to their short-term performance. On the other hand, a varying environment was observed to influence the TCC system (global) creep coefficient and mid-span deflection of TCC systems, especially in situations where relative humidity and moisture content changes caused mechano-sorptive creep. Kanócz and Bajzecerová [65] found that mechano-sorptive creep increased total deflection by 30%, with the environmental conditions having a more pronounced effect on the ±45° connection compared to the notched connection. When testing TCC beams with wet wood, Kenel and Meierhofer [28] found that the 1-year mid-span deflection had three times the instantaneous deflection. For 90° screws, it was found that the connection creep coefficient under variable environmental conditions was almost double that of systems under constant environmental conditions [19,62].

The tables do not show a significant effect of the screw size on the creep coefficient, as the screws were nearly uniform in size. However, it can be expected that larger diameters can reduce the creep coefficient as studies on different diameters of dowels have shown [62,67]. This can be related to the greater stiffness of the larger fastener which can distribute loads more effectively. Enhanced load transfer reduces localized stress concentrations, which can help mitigate creep deformation over time.

Based on test results, replacing NWAC with LWAC did not significantly change the creep coefficient of the connections, which is attributed to the dominance of timber and connector type in long-term TCC behaviour [8,25]. Jorge et al. [8] also showed that using LWAC with different compression strengths did not influence the creep coefficient. While LWAC reduces the dead load of TCC systems, it exhibits greater shrinkage strain than NWAC, potentially increasing deflections from shrinkage and decreasing concrete stiffness [8,25]. However, replacing low-shrinkage concrete with NWAC reduced the long-term deflection by approximately 14% [53], which shows the effect of concrete shrinkage on long-term deflections.

In terms of the effect of timber type on the long-term performance of TCC systems with inclined screws, there is insufficient information available to draw definitive conclusions. However, in testing TCC systems connected with dowels, Van de Kuilen and Dias [62] observed that while the timber grades varied across the tests, both solid timber and glued laminated timber exhibited similar creep behaviour. This may suggest that the type of timber (solid or laminated) may have a limited influence on creep performance in these systems, though further investigation is needed to confirm whether this finding extends to TCC systems with inclined screws and under varying environmental conditions.

As an overall conclusion, for the mechanical fasteners considered in tests reported in Table 2, a creep coefficient of less than one can be expected for inclined screws, depending on their properties.

3.2. Prediction Models for the Long-Term Behaviour of TCC Systems and Connections

The first series of TCC analytical models was based on the composite beam theory proposed by Newmark et al. [21] for concrete-steel connections and Möhler [68] for composite structures. For the long-term behaviour of TCC systems, various approaches have been used to predict creep response, with a focus on determining the 50-year creep coefficient (system and connection) or midspan deflection. These models usually incorporate rheological models of timber, concrete, and the connection system separately.

Toratti [69], Hanhijarvi [70], Becker [71], and Martensson [72] all developed rheological models for the long-term behaviour of timber. For concrete, widely used models include those from ACI 209.2R-08 [73] and CEB [74]. Natterer and Hoeft [75] and Ceccotti [76] used an effective elastic modulus method to address the time-dependent characteristics of each component in TCC systems under sustained load. This method has been widely used by researchers over the years, often with some modifications. Other methods are finite element models with one-dimensional frame [56,77,78] or analytical models that are based on provisions from existing codes and standards and have been validated with experimental results and are easy to use for design purposes [18,54,57,66,79,80,81]. However, existing codes and standards, such as CSA O86 [17] and EC5 [49], do not differentiate between connection types and instead associate the creep coefficient of connections with that of timber. This approach may not provide accurate approximation of the long-term behaviour of TCC systems, and thus, more precise methods may be required. Further details about these models can be found in the aforementioned references.

Another common method involves fitting mathematical functions to experimental curves. These functions include equations that can model the creep behaviour of materials such as power-type (Equation (7)) and logarithmic forms (Equation (8)). Also, in the case of viscoelastic materials such as concrete or wood, rheological (mechanical) models such as the generalized Kelvin model can be used (Equation (9)), which describes viscoelastic behaviour with a parallel combination of spring (elastic part) and dashpot (viscous part). These equations have been used by researchers to predict the long-term deflection behaviour of TCC systems [1,19,25].

$$\phi \left(t\right)=a{t}^b,$$

(7)

$$\phi \left(t\right)=aln(t+b),$$

(8)

$$\phi \left(t\right)=J_1(1-\mathit{exp}\left(-{\displaystyle \frac{t}{{\tau }_1}}\right)+J_2(1-\mathit{exp}\left(-{\displaystyle \frac{t}{{\tau }_2}}\right),$$

(9)

where $t$ is time, and a and b are constants of power-type and logarithmic functions, respectively; $J_1$, ${\tau }_1$, $J_2$ and ${\tau }_2$ are the constants of Kelvin’s model; $J$ and $\tau$ represent the creep compliance and the retardation time, respectively.

In terms of the inclined screws in TCC systems, Kenel and Meierhofer [28] obtained Equations (10) and (11) for ±45° and a combination of ±45° and 90° screws, respectively, after curve fitting. Based on their evaluations, the power-type function showed the best agreement compared to exponential or hyperbolic equations.

$$\phi \left(t\right)=0.104t^{0.268},$$

(10)

$$\phi \left(t\right)=0.160t^{0.260},$$

(11)

where t is time in days.

Equations derived from fitting models provide a reasonable prediction of the 50-year creep coefficient or midspan deflection of TCC systems and connections, albeit they are unable to predict all the fluctuations in creep curves, especially due to variable environmental conditions with mechano-sorptive effect.

Table 4. Comparison of predicted TCC system deflection and creep coefficient.

Reference	Screw Angle and Diameter	50-Year Connection Creep Coefficient	50-Year Global Creep Coefficient	50-Year Midspan Deflection	Model Used to Obtain 50-Year Deflection
Kenel and Meierhofer [28]	±45° 45–90° d = 6 mm	1.44 2.04	3.87 4.12	-	Power-type
van der Linden [1]	±45° d = 6 mm	6.8	3.4	-	Logarithmic
Jorge et al. [8]	±45° d = 6 mm	1.08–1.54	1.1–1.8	-	Logarithmic
Fragiacomo [31]	45° d = 6 mm	2.55	-	$~$9	Power-type (connection) Numerical (beam)
Hailu [66]	±45° d = 5 mm	-	7.9	51.4	Logarithmic

presents a comparison of the 50-year creep coefficient for connections and midspan deflection in TCC systems with inclined screws. The results from these tests were projected to 50 years using numerical or analytical models discussed previously. The TCC beams studied include both T-section and rectangular section utilizing different types of timber and concrete.

The comparison between the proposed models and experimental curves demonstrated that the models predicted the behaviour of TCC systems with inclined screws with good accuracy. Equations derived from fitted models provided a reasonable prediction of the long-term behaviour of TCC systems. However, these models were unable to account for fluctuations caused by environmental changes. As a result, it was proposed to use numerical models that separately simulate the behaviour of timber, concrete, and the connection system for more accurate predictions [13]. Van der Linden [1] used an indirect method to estimate the creep coefficient of ±45° screws based on the global creep coefficient of TCC beams. However, Van de Kuilen and Dias [62] argued that this approach could cause unrealistic results, suggesting that direct long-term testing of connections may be necessary for more accurate predictions. In general, due to the limited data on the impact of screw orientation on the midspan deflection of TCC systems, further research is essential to establish clear correlations.

The final step in evaluating the 50-year midspan deflection in TCC systems is to compare it against the deflection limit to determine if the system meets serviceability requirements. CSA O86 sets the total deflection limit of L/180 span. In contrast, EC5 specifies that the allowable instantaneous and final deflections of beams (or slabs) shall be within L/300 to L/500 and L/150 to L/300, respectively, where L is the span. Based on the other studies with different connection types not listed in this table, it was observed that environmental variations and concrete shrinkage play an important role in the final mid-span deflection of TCC systems [52]. The results also showed that very stiff connection types such as glued and notched-dowel connections showed higher total deflection [53,54,64]. In fact, concrete shrinkage and environmental variations cause stress-free strains that can lead to internal stresses called eigenstresses. The intensity of these eigenstresses is determined by the stiffness of the cross-sections [5]. These factors can have a significant impact on long-term deflections, emphasizing the need for a thorough approach to assess the performance of TCC systems with different connection types over time.

4. Conclusions and Recommendations

In this paper, experimental and analytical studies on the short- and long-term performance of TCC systems connected by inclined screws are reviewed. The key conclusions are presented below.

The performance of TCC systems is profoundly influenced by the strength and stiffness of the connections between their components. Research highlights the importance of the screw insertion angle, revealing that inclined screws significantly enhance both stiffness and strength compared to perpendicular screws to the shear plane. The optimal arrangement for these screws is often a crossed-pair configuration, which effectively improves load distribution through a synergy of withdrawal resistance and bearing capacity. This configuration facilitates a truss-like action under lateral loads, where one screw experiences tension and the other compression, leading to enhanced connection performance.

Experiments showed that inclined screws at 30° provided superior strength and stiffness compared to those at 45°, attributed to more effective load transfer and enhanced withdrawal resistance. While the choice of concrete type has a minimal impact on overall connection performance in inclined configuration, it can influence failure modes. Additionally, the presence of interlayers tends to reduce both strength and stiffness depending on the type of the interlayer used. Although several analytical models exist to predict the behaviour of inclined screws in TCC systems, such as modifications to the Johansen Yield Model and empirical approaches suggested by EC5, they often do not fully account for factors like screw inclination, withdrawal strength, and friction between timber and concrete. Consequently, there is a pressing need for further research to refine these models, particularly in simulating the mechanical performance of crossed-pair inclined screws in various configurations and conditions.

The design of TCC systems should consider both ULS and SLS. Existing literature indicates that SLS requirements often dictate the design of TCC systems. Previous research highlights the crucial role of the rheological behaviour of timber and concrete in influencing long-term deformations in TCC systems. The connection system is similarly impacted by this rheological behaviour, contributing to long-term deformations. Studies showed that the inclined orientation of screws improved the long-term performance by contributing to the load distribution, enabling a more effective transfer of forces between timber and concrete. However, it was observed that fluctuations in environmental conditions and concrete shrinkage had a more substantial impact on long-term deformations which were more pronounced in stiff connections that prevents the free occurrence of differential strain between the concrete slab and timber beam, causing additional deformations in the composite beam.

The creep coefficients given in codes and standards fail to differentiate between connection types, linking the creep coefficient solely to timber. This may lead to inaccurate predictions, highlighting the need for more precise methods.

Future research should prioritize the development of standardized test protocols to evaluate the mechanical properties of TCC systems and their connections. Establishing consistent testing methods will facilitate comparative studies, improve the reliability of experimental data, and enhance the accuracy of analytical models. This effort should encompass various factors, including different screw configurations, concrete types, and environmental conditions. Additionally, to predict the long-term behavior of TCC systems more accurately, further research should focus on refining analytical models that account for rheological behavior and environmental influences. By addressing these areas, researchers can enhance the understanding of TCC performance, ultimately leading to more robust design guidelines and improved structural performance in practical applications.