1 Introduction

The seismic damage induced by the April 2015 Gorkha earthquake in Nepal resulted in nearly 9,000 deaths, 22,000 injuries and about 8 million homeless people (NPC 2015). These casualties, the earthquake-induced financial losses and the interruption of the function of numerous schools and hospitals after the earthquake have elucidated the destructive potential of a strong earthquake event on countries with limited financial resources, defined as developing countries. Within this context, the determination of efficient and holistic design strategies for seismic damage mitigation in these countries is deemed vital for the resilience of the communities located in these countries. Moreover, the urgenc-y of the environmental crisis highlights the necessity for rethinking the current resource-demanding design philosophy for building structures, leading to 40% of the global resource consumption (OECD 2018). Therefore, the sustainable application of these design strategies to developing and developed countries necessitates the minimization of their construction cost and environmental impact through the resource-effective use of natural materials or materials that are locally available in these countries (Tsiavos et al. 2019).

Seismic isolation is an established design strategy for the protection of structures from earthquake-induced damage, aimed at the decoupling of the response of the structure from the motion of the ground during an earthquake ground motion excitation. This strategy is extensively implemented in developed countries through the use of highly engineered devices, defined as seismic isolators. One commonly used highly engineered seismic isolation system based on friction comprises the use of friction pendulum bearings (Constantinou et al. 1992; Castaldo 2014; Castaldo and Amendola 2021; De Domenico and Ricciardi 2018; De Domenico et al. 2020; Furinghetti and Pavese 2020; Furinghetti 2022). However, the cost for the construction, the project-specific testing and the transportation of these devices to developing countries would be unaffordably high, thus impeding their application to these countries.

In an attempt to overcome the aforementioned limitations, Tsang (2008) and Tsang et al. (2012) investigated numerically the use of a sand-rubber foundation layer for the seismic isolation of structures in developing countries. This investigation has paved the way for the use of durable, locally available and recyclable materials as a means of a more economical and sustainable approach for the seismic protection of structures in developing countries. The effect of this low-cost, sand-rubber foundation layer on the seismic isolation of structures has been also investigated analytically by Yaghmaei-Sabegh and Rahmani (2012), Mavronicola et al. (2010) and experimentally by Tsiavos et al. (2019). Banovic et al. extended the application of locally available materials for seismic isolation of structures to the use of a foundation layer consisting of limestone sand (Banovic et al. 2018) and pebbles (Banovic et al. 2019, 2021). Gatto et al. (2020, 2021, 2022) explored the injection of polyurethane foam into the soil below the foundation of a building to create a low-cost seismic isolation system that reduces the seismic acceleration of the building during an earthquake ground motion excitation. Kuvat and Sadoglu (2020) explored the use of sand-bitumen mixtures for the low-cost seismic isolation of structures, while several researchers investigated the use of geosynthetics (Nanda et al. 2012; Yegian and Kadakal 2004).

Common theme across the aforementioned studies is the use of low-cost methods for the Geotechnical Seismic Isolation (GSI) of structures, as defined by Tsang (2009), which comprises a flexible or sliding interface in direct contact with geological sediments and of which the isolation mechanism primarily involves geotechnics. However, the existing Geotechnical Seismic Isolation (GSI) methods have not utilized the beneficial characteristics of the rolling motion of a thin layer of sand grains for the low-cost and sustainable seismic isolation of structures.

Within this frame, the engineering concept of seismic energy dissipation through the inclusion of a thin granular layer between selected structural members was utilized for the first time many centuries ago. Greek and Roman temples, Chinese and Japanese Pagodas (Aicher et al. 2014) and Persian monuments, such as the Tomb of Cyrus (Llunji 2016), were constructed by blocks designed to slide against each other during an earthquake ground motion excitation. Calantarients (1909) proposed the design of a building founded on a layer of talc as a low-cost sliding seismic isolation system. Nevertheless, the scientific background behind the different types of friction between surfaces of varying roughness and circular particles was developed many centuries later. Leonardo Da Vinci, as presented recently by Hutchings (2016), pioneered the experimental investigation of the differences between rolling and sliding friction and elucidated the attractively low friction coefficient which can be achieved through the rolling motion of circular objects against smooth surfaces. The nature of the mechanisms of rolling friction was further examined in detail by Amontons (1699) and Eldredge and Tabor (1955). Cilsalar and Constantinou (2019) and Katsamakas et al. (2022) have investigated the seismic behavior of lightweight structures seismically isolated with a system consisting of high-strength concrete slabs and deformable rolling balls.

The rapid technological development and the discovery of new materials during the last decades enable an efficient combination of the attractive characteristics of rolling friction with the engineering and environmental benefits associated with the use of durable, low-friction materials.

The favorable role of a rolling-sliding mechanism initiated between sand particles and a smooth polymer surface, such as PVC (PolyVinyl Chloride), on the decrease of the interference and interlocking between the sand particles was experimentally demonstrated by O’Rourke et al. (1990). This polymer-sand interface friction reduction mechanism manifested by hard, smooth polymer surfaces was attributed to the minimization of plowing: the plastic indentation, which is induced by sand grains in the polymer surface.

Notwithstanding the significant contribution of the aforementioned studies to the illumination of the low-friction, rolling-sliding mechanisms that can be initiated at a polymer-sand interface, they do not integrate these low-friction mechanisms in the design of a novel and sustainable seismic isolation strategy.

In light of these challenges, Tsiavos et al. (2020, 2021a) have experimentally confirmed the efficiency of a novel low-cost and sustainable seismic isolation strategy based on the encapsulation of a thin layer of sand between two PVC surfaces, defined as PVC sand-wich (PVC-s) seismic isolation through the conduction of large-scale shaking table tests. Evidently, the earthquake excitation of structures founded on this PVC-s seismic isolation above a critical amplitude of 0.2 g triggers a rolling-sliding response of the upper PVC surface relative to the bottom PVC surface. The energy dissipation associated with this response reduces the seismic acceleration (and force) demand to the isolated structure compared to the corresponding fixed-based structure and protects the structure from earthquake-induced damage.

However, the investigation of the PVC-s seismic isolation by Tsiavos et al. (2020, 2021a) did not include any design recommendation for the recentering of the structure at its original position after the end of the seismic excitation. Moreover, this study was limited to the use of only one material (PVC) as the sliding layer and did not investigate the use of timber as a means of sustainable seismic protection of structures (Yenidogan et al. 2020, 2022).

The aim of this study is to overcome the limitations of the feasibility study of the PVC-s seismic isolation proposed by Tsiavos et al. (2020, 2021a) through a large-scale experimental investigation. This large-scale investigation entails the conduction of shaking table tests that demonstrate the efficiency of a novel energy dissipation device with recentering ability. The energy dissipation device, defined as Dovetail with Springs (Dove-SP), comprises two timber slabs that are designed to slide against each other and recenter back to their original position after sliding. This configuration utilizes the low friction attributed to the rolling motion of sand grains which are sand-wiched between two rigid surfaces to facilitate sliding between the surfaces and seismic energy dissipation. Therefore, the seismic isolation proposed in this study belongs to the family of Geotechnical Seismic Isolation, as proposed by Tsang (2009). Two different sliding configurations are explored to facilitate sliding between the two rigid surfaces: PVC-sand-PVC and timber-sand-timber. A dovetail connection is designed to restrain the relative displacement between the upper and the lower part of the energy dissipation device. A set of low-cost steel springs is designed and installed on both sides of the dovetail connection to recenter the structure back to its original position after the end of sliding. Last but not least, a novel deformable wood material is used to reduce the accelerations attributed to the impact of the upper slab against the lower slab of the energy dissipation device.

Along these lines, this study aims to illustrate the favourable role of a low-cost and sustainable timber-based energy dissipation system equipped with a sand-wich sliding interface, a dovetail connection and low-cost springs. This system can be used as a seismic isolation strategy or a tuned mass damper for the seismic protection and the recentering of structures located in developing or developed countries.

2 Experimental setup

The investigated experimental configuration shown in Fig. 1 comprises the energy dissipation system, defined as Dovetail with SPrings (Dove-SP), which consists of two 2 m × 1 m beech timber slabs designed to slide against each other using a dovetail joint. Beech timber is recyclable, it has long life-span (300–400 years) and has not yet been used for the seismic isolation of structures. Ehrhart et al. (2016) have shown the enormous tensile strength values that can be obtained through the use of beech timber, leading to a mean value of tensile strength equal to 67 MPa. This high mechanical strength of the material can lead to the design of structural members with low thickness, thus saving a substantial amount of resources. Long periods of use of beech timber allow long-term storage of CO2 in buildings (Ehrhart et al. 2018). Bajraktari et al. (2013) propose the use of film forming and semitransparent penetrating stains as the more adequate painting methods for the protection of beech timber from moisture and weathering (snow, rain, low temperature), in case the material is used outdoors. Two different configurations are used for the construction of the ‘sand-wich’ seismic isolation: A PVC-sand-PVC and a beech timber-sand-beech timber configuration. The thickness of the timber and PVC plates used in these configurations is 6 mm, while the surface density of the sand grains is 750 g/m2 (Tsiavos et al. 2020, 2021a).

Fig. 1
figure 1

a Sand-wich sliding interface for two different materials (PVC, timber), b cross-section and c overview of the shaking table experimental setup of the timber-based energy dissipation system with recentering springs (dimensions in mm)

Full size image

The properties of the sand are also defined by Tsiavos et al. (2020, 2021a). In each of the 58 × 28 cm holes of the dovetail connection of the bottom slab, there is an ensemble of 8 low-cost steel springs (four on each side) that are fixed to two plates designed to recenter the structure in its original position after sliding, as shown in Fig. 1. The attachment of the bottom slab of the dovetail joint on the shaking table is shown in Fig. 2a.

Fig. 2
figure 2

a Bottom timber slab of the dovetail joint fixed on the shaking table and (b) Recentering system comprising four low-cost steel springs placed at each side of the dovetail joint

Full size image

A recentering system consisting of four steel springs (Fig. 1, Fig. 2b) and two plates is installed in each side of each hole in the dovetail joint connection. There is a 1 cm gap between the upper plate of the dovetail joint (pin) and the outer plate of the recentering system to allow the initiation of sliding at the sand-wich interface. This outer plate, which may get in contact with the upper part of the dovetail joint during the excitation if the sliding gap is exceeded (Fig. 1, Fig. 2b) consists of deformable delignified wood, which was obtained by structure-retaining delignification. The process has been recently used to obtain a highly deformable balsa wood sponge to enhance the piezoelectric properties of wood (Sun et al. 2020). For the fabrication of the deformable wood plates for this study, the process was simplified and layers of delignified spruce wood were soaked in a mixture of glycerol and water (mixture ratio 1:4), dried and afterwards stacked to achieve a 1.5 cm thick plate, which was finally wrapped in plastic foil. The 1.5 cm thick deformable wood plates were installed at the outer sides of the recentering system to ameliorate the consequences due to the pounding of the upper part of the dovetail joint (pin) with the lower part of the joint (tail). Each low-cost spring has a diameter of 1.8 cm, length 12 cm, axial stiffness k = 842 N/m and cost of 6 EUR. The vibration period of the energy dissipation device consisting of an upper timber slab with a mass m = 172 kg (Fig. 1) was experimentally determined as T = 0.71 s through a free vibration test.

2.1 Ground motion ensemble

The aforementioned specimen was subjected to an earthquake ground motion ensemble of 12 recorded motions using the 2 × 1 m unidirectional shaking table of ETH Zurich laboratory. The ground motion ensemble is presented in Table 1.

Table 1 Ground motion ensemble used in this study
Full size table

The specimen is excited in the X-direction (Fig. 1).The selected ground motion time histories were taken from the European Strong-Motion Database (Luzi et al. 2020) and the Italian Accelerometric Archive (ITACA).

2.2 Instrumentation

The instrumentation consists of four (4) accelerometers (A) and ten (10) displacement markers (M), placed in the front side of the specimen. The displacements were monitored through an optical tracking measurement system. The instrumentation plan is shown in Fig. 3, while the constructed experimental setup is presented in Fig. 4. Each accelerometer (purple color) measures the acceleration in one direction (X-direction), while each marker (grey color) tracks the displacement in three directions. All measurements are synchronized.

Fig. 3
figure 3

Instrumentation plan (dimensions in mm)

Full size image
Fig. 4
figure 4

Constructed experimental setup

Full size image

2.3 Dimensional analysis

Tsiavos et al. (2020, 2021a) have performed dimensional analysis to demonstrate the fundamental dimensional ratios Π1 and Π2 that quantify the similitude between the sliding behavior of a model structure isolated using a ‘sand-wich’ seismic isolation and its real-scale counterpart. The dimensionless strength ratio Π1 = µg/PGA (µ is the static friction coefficient, g is the acceleration of gravity, PGA is the Peak Ground Acceleration) expresses the strength of the sliding interface relative to the ground motion intensity. The number and dimensions of the pins and tails of the dovetail joint (Fig. 2) can be designed in the real-scale structure to engineer the vertical stress and, therefore, the friction coefficient of each of the sliding interfaces. Moreover, Tsiavos et al. (2020, 2021a) did not observe experimentally any significant dependence of the friction coefficient of a sand-wich sliding interface on the vertical stress, as long as the surface density of the sand grains is 750 g/m2. The increase of the sand surface density above this limit is detrimental, as it leads to a higher friction coefficient of 0.3, determined experimentally by Tsiavos et al. (2020). Within this frame, the acceleration amplitude of the ground motions used in this study represents realistic ground motion records, thus maintaining the dimensionless strength ratio between the model and its real-scale counterpart. The vibration period ratio Π2 = T/Tg represents the relation between the period T of the structural system over the predominant period of the excitation Tg, defined as the period, where the 5% velocity spectrum attains its maximum (Miranda and Bertero 1994). In this case, the vibration period of the structural system emerges from the mass of the upper timber slab and the stiffness of the selected low-cost springs. The frequency content of the motions used in this study was the same with the frequency content of the recorded ground motion ensemble. Therefore, the maintenance of this ratio Π2 for energy dissipation systems of higher scale and higher mass would necessitate the use of more low-cost springs or low-cost springs of higher stiffness for the design of this energy dissipation system in a real-scale prototype structure.

3 Experimental results

3.1 Energy dissipation device with low-cost steel springs and PVC-sand-PVC sliding interface

The model of the energy dissipation device presented in Fig. 4 was subjected to the earthquake ground motion ensemble shown in Table 1. The mass m of the upper timber slab was 172 kg, leading to a vertical stress of 2.5 kPa on each sliding interface of the specimen. The full and magnified acceleration response of the energy dissipation device subjected to the Kobe 1995 ground motion record (No 6 in Table 1) is shown in Fig. 5. The corresponding sliding displacement response of the upper timber slab relative to the bottom slab subjected to the same ground motion is presented in Fig. 6.

Fig. 5
figure 5

a Full and b magnified acceleration time history response of the upper timber slab (mean value of A3 and A4 in Fig. 3) subjected to Kobe 1995 ground motion record (No. 6 in Table 1)

Full size image
Fig. 6
figure 6

Sliding displacement of the upper timber slab relative to the motion of the bottom timber slab due to Kobe 1995 ground motion record (No. 6 in Table 1)

Full size image

As shown in Fig. 6, the upper timber slab manifested significant sliding with a maximum sliding displacement of 23.31 mm. The acceleration at which the sliding of the isolated upper timber slab was triggered (sliding acceleration) was 0.2 g (Fig. 5b), which corresponds to a friction coefficient µ = 0.2 for the PVC-sand-PVC sliding interface. This friction coefficient value confirms the experimental results by Tsiavos et al. (2020, 2021a) for this interface. The presence of low-cost steel springs led to almost complete recentering of the upper timber slab relative to the lower slab, leading to an insignificant residual sliding displacement of 3.87 mm.

The full and magnified acceleration response of the same energy dissipation device subjected to the Umbria 1997 ground motion record is shown in Fig. 7. The corresponding sliding displacement response of the upper timber slab relative to the bottom slab subjected to the same ground motion is presented in Fig. 8.

Fig. 7
figure 7

a Full and b magnified acceleration time history response of the upper timber slab (mean value of A3 and A4 in Fig. 3) subjected to Umbria 1997 ground motion record (No. 10 in Table 1)

Full size image
Fig. 8
figure 8

Sliding displacement of the upper plate relative to the motion of the bottom plate due to Umbria 1997 ground motion record (No. 10 in Table 1)

Full size image

The excitation of the energy dissipation device subjected to this motion with PGA = 0.65 g led to almost the same value of sliding acceleration of the upper timber slab αmax = 0.2 g and friction coefficient µ = 0.2, thus confirming the repeatability of the results determined for the Kobe 1995 ground motion record. The maximum sliding displacement of the upper timber slab relative to the bottom slab was 12.85 mm, while the corresponding residual sliding displacement value was 4.93 mm. This insignificant residual displacement confirms the ability of the low-cost springs to recenter the structure back to its original position after the end of the earthquake ground motion excitation.

3.2 Influence of the sliding interface on the response-Timber-sand-timber sliding interface

The effect of the use of beech timber instead of PVC in the ‘sand-wich’ sliding interface (Fig. 1a) is shown in Fig. 9, presenting the comparison of the acceleration response of the specimen founded on the two different sliding interfaces to the Umbria 1997 ground motion record.

Fig. 9
figure 9

a Full and b magnified acceleration time history response of the upper timber slab (mean value of A3 and A4 in Fig. 3) subjected to Umbria 1997 ground motion record (No. 10 in Table 1) for two different sliding interfaces: PVC-sand-PVC (PVC sand-wich) and timber-sand-timber (timber sand-wich)

Full size image

The use of timber as a sliding layer in a timber ‘sand-wich’ seismic isolation triggered a rolling motion of the sandwiched sand particles and reduced the friction coefficient from 0.2 to 0.16, thus creating even more attractive frictional characteristics in the interface.

The values of the acceleration, at which sliding is initiated, for all the ground motions used in this study and the two sliding interfaces are shown in Fig. 10. These values correspond to the experimental determination of the friction coefficient µ of the two interfaces as µ = Sliding acceleration/g.

Fig. 10
figure 10

Sliding acceleration of the upper timber slab for two different sliding interfaces: PVC-sand-PVC (PVC sand-wich) and timber-sand-timber (timber sand-wich) for the ground motion ensemble used in this study

Full size image

As shown in Fig. 10, the mean value of the sliding acceleration for the PVC-sand-PVC sliding interface (PVC sand-wich) is 0.187 g, corresponding to a friction coefficient µ = 0.187. The mean sliding acceleration for the timber-sand-timber interface is 0.159 g, leading to a friction coefficient µ = 0.159. This comparison shows the mechanically favourable role of the use of a beech timber surface on the sand-wich interface for the initiation of sliding at an attractively low seismic intensity level.

The values of the peak acceleration response of the upper timber slab founded on a timber sand-wich sliding interface for varying values of the Peak Ground Acceleration (PGA) of the ground motion records used in this study are shown in Fig. 11.

Fig. 11
figure 11

Peak Ground Acceleration (PGA) and peak acceleration response of the upper timber slab for the ground motion ensemble used in this study

Full size image

The mean value of the peak acceleration response of the upper timber slab is 0.286 g, which is 5.24 times lower than the mean value of the Peak Ground Acceleration of the selected ground motion ensemble. This reduction elucidates the beneficial seismic isolation effect of the presented energy dissipation system, which can lay the basis for the low-cost and sustainable seismic protection of structures.

3.3 Influence of the low-cost steel springs on the response

The influence of the use of low-cost steel springs on the seismic response of the energy dissipation system is shown through the comparison of the response of the system equipped with springs with the response of the corresponding system without springs. The response of both models based on a PVC-sand-PVC sliding interface and subjected to the Kobe 1995 ground motion excitation is shown in Fig. 12.

Fig. 12
figure 12

Sliding displacement of the upper timber slab relative to the motion of the bottom slab due to Kobe 1995 ground motion record (No. 6 in Table 1) for two model configurations of the proposed timber-based energy dissipation system: with springs and without springs

Full size image

As presented in Fig. 12, the installation of low-cost steel springs in the bottom slab of the dovetail joint reduced the residual displacement of the specimen from 26 to 4.93 mm.

The values of the residual sliding displacement and the maximum sliding displacement of the upper timber slab for all the ground motions used in this study and the PVC-sand-PVC interface are shown in Figs. 13 and 14. The mean value of the residual sliding displacement of the upper timber slab without the addition of low-cost springs is 28.62 mm. However, the mean value of the residual displacement of the energy dissipation system equipped with low-cost springs is significantly smaller, 8.93 mm. The mean value of the maximum sliding displacement of the system with low-cost springs is 25.59 mm, while the corresponding value for the system without springs is 45.84 mm.

Fig. 13
figure 13

Residual sliding displacement of the upper timber slab relative to the motion of the bottom slab for the ground motion ensemble used in this study (Table 1) for two model configurations of the proposed timber-based energy dissipation system: with springs and without springs

Full size image
Fig. 14
figure 14

Maximum sliding displacement of the upper timber slab relative to the motion of the bottom slab for the ground motion ensemble used in this study (Table 1) for two model configurations of the proposed timber-based energy dissipation system: with springs and without springs

Full size image

The aforementioned comparison shows the ability of the designed low-cost springs to recenter the structure back to its original position after the end of the ground motion excitation, thus substantially improving the sliding behavior of the energy dissipation system.

3.4 Influence of added mass on the response

A 3 cm thick steel plate was fixed to the top of the upper timber slab equipped with low-cost steel springs and timber-sand-timber sliding interface, thus leading to an addition of a mass of 234 kg and a total top mass of the specimen equal to 406 kg (Fig. 15), an increase of almost 240%. This weight corresponds to a vertical stress of 6.25 kPa on each sliding interface of the specimen. The vibration period of the energy dissipation device consisting of an upper timber slab with a mass m = 406 kg was experimentally determined as T = 1.05 s through a free vibration test. Within this frame, the tested device is full-scale for per-device tributary gravity load between 234 and 406 kg, or interface pressure between 3.5 and 6.25 kPa, assuming a sand of the properties and surface density defined by Tsiavos et al. (2020, 2021a).

Fig. 15
figure 15

Longitudinal and transverse cross sections of the shaking table experimental setup of the sliding dovetail connection with recentering springs with mass m = 406 kg (dimensions in mm)

Full size image

As shown in Fig. 16, there were minor differences between the acceleration time history response of the specimen with the larger mass subjected to Kobe 1995 ground motion record compared to the specimen with the smaller mass, leading to only 5% increase in the sliding acceleration and the corresponding friction coefficient of the interface. A friction coefficient value µ = 0.17 was determined for the timber-sand-timber sliding interface and vertical stress of 6.25 kPa. The effect of the added mass on the sliding displacement of the specimen is shown in Fig. 17.

Fig. 16
figure 16

a Full and b magnified acceleration time history response of the upper timber slab (mean value of A3 and A4 in Fig. 3) subjected to Kobe 1995 ground motion record (No. 6 in Table 1) for timber-sand-timber sliding interface and two values of top mass: m = 406 kg and m = 172 kg

Full size image
Fig. 17
figure 17

Sliding displacement of the upper timber slab relative to the motion of the bottom slab due to Kobe 1995 ground motion record (No. 6 in Table 1) for two model configurations of the proposed timber-based energy dissipation system: one with top mass m = 172 kg and one with top mass m = 406 kg

Full size image

As shown in Fig. 17, the minor increase of the friction coefficient of the interface in this case led to a smaller sliding displacement for the specimen with more mass in one direction (between t = 10.5 s and t = 11.5 s), which triggered a higher sliding displacement of the specimen with more mass in the other direction. This high sliding displacement of structures subjected to long-period pulses has been experimentally demonstrated by Tsiavos et al. (2013, 2017, 2021b, 2021c) and analytically by Güneş (2022). The insignificant residual sliding displacement of 5 mm in this case demonstrates the efficiency of the low-cost and sustainable energy dissipation device to recenter the structure back to its original position after the end of the excitation for this increased value of the top mass.

The values of the acceleration, at which sliding is initiated, for all the ground motions used in this study and the two mass values are shown in Fig. 18. These values correspond to the experimental determination of the friction coefficient as µ = Sliding acceleration/g.

Fig. 18
figure 18

Sliding acceleration of the upper timber slab for the ground motion ensemble used in this study for two model configurations of the proposed timber-based energy dissipation system: one with top mass m = 172 kg and one with top mass m = 406 kg

Full size image

As shown in Fig. 18, the mean value of the sliding acceleration for the structure with top mass m = 172 kg is 0.159 g, corresponding to a friction coefficient µ = 0.159. The mean sliding acceleration for the structure with top mass m = 406 kg is 0.166 g, leading to a 4.4% higher friction coefficient µ = 0.166 compared to the specimen with the smaller mass. This comparison indicates that the dependence of the frictional characteristics and the sliding behaviour of the timber-sand-timber interface from the vertical load is rather negligible. Tsiavos et al. (2020, 2021a) have investigated experimentally at large scale the seismic behavior of buildings isolated using the 'sand-wich' seismic isolation strategy and have shown the independence of the friction coefficient of the interface from the vertical stress and the efficiency of the 'sand-wich' seismic isolation for higher values of vertical stress than the values presented in this study. However, this effect could be further investigated by future numerical or experimental studies.

The values of the peak acceleration response of the upper timber slab founded on a timber sand-wich sliding interface for the two mass configurations are shown in Fig. 19.

Fig. 19
figure 19

Peak acceleration response of the upper timber slab for the ground motion ensemble used in this study for two model configurations of the proposed timber-based energy dissipation system: one with top mass m = 172 kg and one with top mass m = 406 kg

Full size image

The mean value of the peak acceleration response of the upper timber slab with the larger top mass m = 406 kg is 0.274 g, which is slightly smaller than the corresponding value of the slab with the smaller top mass m = 172 kg, which is equal to 0.286 g. However, this trend is not consistently observed among the different ground motions used in this study and the results vary according to the type of the ground motion excitation. However, the results indicate that the ability of the novel system to decouple the motion of the upper structure from the motion of the ground, thus protecting the structure from seismic damage, does not substantially depend on the mass of the upper structure.

4 Conclusions

This study presents the results of a large-scale shaking table investigation demonstrating the efficiency of a novel, low-cost and sustainable energy dissipation system, which can be used as a seismic isolation system or a tuned mass damper for the seismic protection of structures in developing or developed countries.

The energy dissipation system, defined as Dovetail with Springs (Dove-SP), comprises two timber slabs that are designed to slide against each other in a motion that is restrained by a dovetail sliding joint. Two sliding interfaces are experimentally investigated: A PVC sand-wich (PVC-s) sliding interface, comprising a thin layer of sand that is sandwiched between two PVC layers and a novel, timber sand-wich sliding interface consisting of a thin layer of sand encapsulated between two timber surfaces. A set of low-cost steel springs is designed and installed on both sides of the dovetail joint to recenter the structure back to its original position after the end of an earthquake ground motion excitation. A novel, low-cost and deformable wood material fabricated from delignified balsa wood (Sun et al. 2020) is used to ameliorate the consequences due to pounding occuring at the dovetail joint before the activation of the steel springs.

A large-scale experimental campaign was performed to investigate the seismic response of the novel energy dissipation system subjected to an ensemble of recorded earthquake ground motion excitations using the shaking table of ETH Zurich. The timber sand-wich interface facilitated the sliding of the upper timber slab against the lower timber slab at an attractively low seismic acceleration of 0.16 g and a corresponding friction coefficient µ = 0.16. This value is significantly lower than the corresponding friction coefficient of the PVC sand-wich sliding interface, which was experimentally determined as µ = 0.19. Moreover, the use of beech timber at the timber sand-wich sliding interface increases the sustainability and reduces substantially the environmental impact of the design solution compared to the PVC sand-wich interface.

The novel recentering system comprising low-cost steel springs reduced substantially the residual sliding displacement of the upper slab after the end of the excitation compared to the same specimen without springs. The mean value of the residual sliding displacement of the system with recentering springs was 9 mm, while the corresponding mean sliding displacement value without recentering was 29 mm. This insignificant value of the residual sliding displacement with low-cost springs shows the efficiency of the system to recenter the structure back to its original position after the end of the earthquake excitation, thus overcoming the deficiencies of the existing solutions for low-cost seismic isolation of structures.

The attractive sliding behavior of the novel energy dissipation system was not substantially influenced by the increase of the seismic mass acting on the upper timber slab. The sliding acceleration of the upper slab of the system with the higher mass was only 5% higher compared to the system with the lower mass, while the sliding behavior of the two systems was similar. The attractive sliding behavior of the energy dissipation system for varying values of its seismic mass demonstrates the ability of the system to be used as a low-cost and sustainable seismic isolation system or a tuned mass damper, which can be tuned in different frequencies to protect a wide range of buildings from excessive displacements or damage due to wind or earthquake ground motion excitation.

Further numerical or large-scale experimental studies could be performed to investigate the effect of the increase of the mass and the dimensions of the system on its efficiency as a low-cost seismic isolation system or as a sustainable tuned mass damper. Furthermore, the design of the system could be further investigated by follow-up studies to optimize the behavior of structures subjected to earthquake ground excitation or wind excitation, such as low-rise masonry buildings subjected to earthquakes or high-rise timber buildings subjected to wind. The use of the presented Dove-SP energy dissipation system for low-cost seismic isolation with recentering is not limited to timber slabs, but can also be designed using Reinforced Concrete (RC) slabs, especially if the system is located at the foundation of the building, which can be subjected to moisture or saturated conditions. In this case, the use of timber would require protection and special treatment, while the RC slab design solution would be much less affected by moisture. Within this frame, future studies could explore the implementation of the timber-based system in a moisture-affected environment and the seismic behavior of the same system with RC slabs replacing the presented timber slabs.

The use of deformable wood plates contributed to the reduction of the acceleration at the impact in the dovetail joint. The energy dissipation attributed to the compression of the deformable wood plates was rather small compared to the energy dissipation attributed to the sliding-rolling displacement of sand grains at the sand-wich interface. Within this frame, the use of deformable wood plates of higher thickness as a means of higher energy dissipation and impact reduction could be further numerically or experimentally investigated by future studies.

The low-cost of the presented energy dissipation system and the use of timber as a low-carbon material lead to the development of a sustainable design practice, which enables increased service lifetime, lower environmental impact and applicability to a higher spectrum of countries and buildings worldwide compared to the existing steel-based, highly manufactured solutions for seismic isolation of structures. Along these lines, the integration of low-cost seismic protection and sustainability in a novel design methodology lays the foundation for the rethinking of the design of structures in developing and developed countries.