Study on Thermal Insulation Performance of Cross-Laminated Bamboo Wall

: In recent years, bamboo, as a green building material, has attracted more and more attention worldwide. Inspired by the investigation of cross-laminated timber in structural systems, a new engineered cross-laminated bamboo (CLB) consisting of the cross lamination of bamboo scrimber plates is proposed in this paper. To evaluate its potential in structural applications, the thermal insulation performances of the CLB walls and CLB walls with the EPS foam plate were studied and evaluated by the temperature-controlled box-heat flow meter method. Test results indicated that the thermal insulation performance improved with the increase of thickness, but different wall configurations had little effect on the thermal insulation performance under the same thickness of the CLB wall. The thermal insulation performance of EPS-CLB composite wall was much better than that of CLB wall. In addition, a relatively acceptable accuracy of the theoretical calculations was proved. Finally, the influence of different locations of the EPS foam plate on heat transfer coefficient can be neglected as it was studied based on the validated numerical models.


Introduction
Bamboo is one of important forest resources, which has more than 75 genera and 1250 species [1]. The total area of bamboo forests has reached 22 million hm 2 , mainly distributed in tropical, temperate and even cold zones [2]. Bamboo is a renewable resource with many advantages, such as high strength, fast growth and high yield and the use of the bamboo as construction material is beneficial to the protection of the environment [3]. The raw bamboo is used as basic components in the traditional bamboo buildings. However, due to the variability in geometric size and mechanical properties [4], the raw bamboo cannot meet requirements of physical and mechanical properties specified in modern buildings.
The modern engineered bamboo provides an effective way to solve the above problems and broadens the application of bamboo in structures. There are many kinds of modern engineered bamboo according to its raw material, processing technology and combination mode. The raw materials are mainly divided into five categories according to their geometric dimensions: bamboo chip, bamboo mat, bamboo curtain, bamboo split and bamboo filament [5]. The combination modes can also be divided into two kinds: One is that all raw materials are combined along the direction of the fibers and the other is that all raw materials are not combined along the direction of the fibers. At present, the two most commonly used engineered bamboos are laminated bamboo lumber and bamboo scrimber [3]. Many scholars in China and other countries that has bamboo resources have carried out researches on various aspects of laminated bamboo lumber, such as mechanical properties [6][7][8][9], material modification [10,11] and flame retardant properties [12].
However, the laminated bamboo lumber has a limitation that requires bamboo with large diameter as raw material. The bamboo scrimber is made of one-way bamboo filaments by dipping in the glue and hot-pressing, which has no requirements for raw material [13]. Guan et al. [14] compared bending resistance of timber scrimber and bamboo scrimber through experiments, which indicated that the longitudinal flexural modulus of elasticity of bamboo scrimber is better than that of timber scrimber when their densities are larger than 0.7 g/cm 3 and the longitudinal flexural strength of bamboo scrimber is better than that of timber scrimber when their densities are larger than 0.8 g/cm 3 . Many scholars have studied different aspects of bamboo scrimber, such as mechanical properties [15][16][17], flame retardant properties [18,19] and material modification [20,21].
There have been many studies on mechanical properties of bamboo and bamboo components, but there are relatively few studies on thermal insulation performance of engineering bamboo composites [22,23], especially on the thermal insulation performance of bamboo scrimber as wall material. Gao et al. [24] prepared three kinds of bamboo-wood composite wall, among which bamboo curtain plywood was used for bamboo board, and evaluated their thermal insulation and sound insulation performance, revealing the influence of wall materials and internal filling materials on the thermal insulation and sound insulation performance of the wall. Wang et al. [25] carried out experiments on four different configurations of composite shear walls, including one classical full wood-based configuration, one hybrid bamboo-wood-based configuration and two full bamboo-based configurations with different studs thickness, to study their thermal insulation performance. The results showed that the thermal insulation performances of engineered bamboo composites are slightly lower compared with wood composites and there is the possibility of using bamboo-based shear walls in light-frame buildings ensuring thermal insulation performances similar to classical wood-based one.
The above bamboo composite walls are all used in the light frame construction. In order to break through the height limitation of light bamboo structure and promote the development of modern heavy bamboo structure, a new engineering cross-laminated bamboo (CLB) is proposed in this study inspired by cross-laminated timber (CLT). So as to explore the application feasibility of walls made of CLB, a series of experiments were carried out to investigate the thermal insulation performance of CLB walls and CLB walls with the EPS foam plate (EPS-CLB composite walls). The objectives of this study are: (1) to study the effect of different wall thickness and different wall configurations on heat transfer of CLB walls and EPS-CLB composite walls; (2) to establish validated numerical models for the deeper research on the effect of different EPS locations on the heat transfer in the wall.

Heat Transfer Analysis of CLB Wall and EPS-CLB Composite Wall
In this section, the thermal conductivity coefficient λ of bamboo scrimber plate and EPS foam plate and the heat transfer coefficient K of CLB wall and EPS-CLB composite wall were measured through experiments. According to the experimental results, the thermal insulation performance of 5 CLB walls and 5 EPS-CLB composite walls were analyzed. In addition, the theoretical values of K of walls calculated from λ of bamboo scrimber plate and EPS foam plate were compared with the experimental values of K of walls measured by experiments.

Test Specimens and Methodology
Thermal conductivity coefficients of two types of plates are measured in this paper, including the bamboo scrimber plate and EPS foam plate. The raw material of the bamboo scrimber is selected from the moso bamboo in the Huangshan area of Anhui Province, China. Bamboo scrimber plates are made by Anhui Bamboo Track New Material Science and Technology Co., Ltd. The specific production process is as follows: (1) Firstly, the bamboo is soaked in the hot alkaline solution to be softened; (2) Secondly, the softened bamboo is processed into bamboo filaments through roller press and then dried; (3) Thirdly, the bamboo filaments are soaked with glue and then dried again; (4)Finally, the dried bamboo filaments dipped in glue are placed uniformly and directionally, and then hot-pressed to the bamboo scrimber plate.
The bamboo scrimber plate has an average bulk density of 1200 kg/m 3 and the moisture content of bamboo scrimber plate is about 7% in this test. The 30 mm thick peeled high density EPS foam plates, used as the thermal insulation plate, are produced by Yiwu Zeke Packaging Factory in Jinhua city, Zhejiang province, China. The average bulk density of the EPS foam plate is 25 kg/m 3 , only about 2.08% of the bamboo scrimber plate. The adhesive is 1-K-PUR adhesive produced by Jowat Adhesive Co., Ltd. in Beijing, China. The specific heat capacity of the adhesive is 1150 J/(kg·K) and the thermal conductivity coefficient of the adhesive is about 0.3 to 0.6 W/(m·K), which is provided by the manufacturer.
Three groups of bamboo scrimber plate specimens A1-A3 were used to determine the thermal conductivity coefficient of the bamboo scrimber plate and three groups of bamboo scrimber plates specimens B1-B3 were used to study the effect of the average temperature on the thermal conductivity. Three groups of EPS foam plate specimens C1-C3 were used to determine the thermal conductivity coefficient of the EPS foam plate. Each group has two specimens and all the above specimens are 300 mm × 300 mm × 30 mm (l × b × d) in dimension. In order to improve the accuracy of the test results, the surface of all tested specimens should be machined as flat as possible in the manufacturing process to ensure the good contact between the tested specimens and test apparatus.
The method of determining steady-state thermal conductivity properties of the bamboo scrimber and EPS foam plates by the guarded hot plate apparatus was carried out according to the provision of ISO 8302: 1991 [26]. A DRH-300 thermal conductivity apparatus (a Double-sides Guarded Hot Plate apparatus) is shown in Fig. 1, which can test one group of two specimens at the same time. This apparatus can test materials with thermal conductivity coefficients from 0.01 W/(m·K) to 2 W/(m·K) and its test error can be controlled within 3%.  The thermal conductivity coefficient λ can be evaluated according to ISO 8302:1991 [26] as follows:   According to the test results of the above tables, the average values of the thermal conductivity coefficients of the bamboo scrimber plate λB obtained from groups A1-A3 and EPS foam plate λE obtained from groups C1-C3 are 0.263 W/(m•K) and 0.040 W/(m•K), respectively. As observed from groups B1-B3 in Tab. 1, the value of λB only slightly increases with the average temperature. λB increases from 0.260 W/(m•K) in the group B1 to 0.273 W/(m•K) in the group B3 with the increase of Tave from 15.06 o C to 30.36 o C, and the increase of λB is only 5.0%. Therefore, the influence of the average temperature on the thermal conductivity coefficient of the bamboo scrimber plate can be neglected in the thermal calculation. The slight increase of the thermal conductivity coefficient with the increase of the average temperature can be explained as follows: As the average temperature increases, the thermal motion of the bamboo solid molecules and free water molecules in bamboo increases. The thermal conductivity of the air in the pores and radiation heat transfer between the pore walls are also enhanced. Therefore, the combined action of gas phase and solid phase makes the thermal conductivity of bamboo slightly increase with the increase of the average temperature.
Thermal conductivity coefficients of commonly used building materials are listed in Tab. 3 per Chinese standard GB50176-2016 [27]. Compared with most existing common inorganic materials used in building, the thermal conductivity coefficient of the bamboo scrimber plate is far less than that of these materials, which is about 15% of reinforced concrete and about 50% of hollow brick and light-weight aggregate concrete. The EPS foam plate has the superb thermal insulation performance and λE is only 0.04 W/(m•K), which is about 15.2% of λB. Although the EPS foam plate has excellent thermal insulation performance, the compressive capacity of the EPS foam plate is poor, which makes it cannot be independently used to satisfy requirements of load bearing. The mechanical properties of bamboo scrimber plate are good [3]; and therefore, the bamboo scrimber plate and EPS foam plate can be combined to achieve a good thermal insulation effect and satisfy bearing capacity requirements.

Heat Transfer Analysis of CLB Wall and EPS-CLB Composite Wall 2.2.1 Test Specimens
CLB is formed by the cross lamination of bamboo scrimber plates introduced in Section 2.1.1. The adhesives used in CLB walls and EPS-CLB composite walls is 1-K-PUR adhesive produced by Beijing Youward Adhesive Co., Ltd, China. According to Chinese climate and energy-saving design requirements [28][29][30][31], the thicknesses of walls are determined to be 90 mm, 120 mm, 150 mm, 180 mm and 210 mm, respectively, considering the bearing requirements. In order to investigate the effect of different wall thicknesses and different wall configurations on heat transfer, 5 CLB walls and 5 EPS-CLB composite walls were designed in this test. Each specimen has the same length and width of 980 mm and the thicknesses of all tested wall specimens are listed in Tab. 4. Detailed configurations of the tested wall specimens are shown in Fig. 2. The thicknesses of EPS foam plates are all 30 mm.

Test Protocol
The temperature controlled box-heat flow meter method [32] which combines heat flow meter method and hot box method is used in the present study. The instrument selected for this test is JTRG-1 type heat transfer performance detection device produced by Beijing Century Jiantong Technology Co., Ltd., China, as shown in Fig. 3. Specific technical indicators of JTRG-1 type detection apparatus are summarized as follows: (1)   The specific operation process is summarized as follows: (1) Installation of test specimens. The CLB wall is installed in the test holder, and the polyurethane foaming agent is used to fill the voids around the specimens to ensure good sealing so as to prevent air convection in the cold and hot boxes. (2) Layout measuring points. Three heat flow measuring points (recorded as HF1, HF2 and HF3, respectively) and six temperature measuring points (recorded as HT1 to HT6, respectively) are evenly arranged on the hot side of the walls, and the temperature measuring points are close to the heat flow measuring points. Six temperature measuring points (recorded as CT1 to CT6, respectively) are arranged on the cold side of the wall, as shown in The heat transfer coefficient K within one measurement cycle can be calculated based on Eq. (2) per ISO 8990:1994(E) [33] and Chinese standard JGJ/T132-2009 [34]. The average value of the heat transfer coefficient with an error less than 1% between two consecutive measurement cycles is taken as the acceptable test result.

HT1 HT2 HT3
HT4 HT5 HT6 where R' is the total heat transfer resistance, (m 2 •K)/W; Ri and Re are internal and external surface heat transfer resistances of the wall, respectively. According to Chinese standard 'Code for thermal design of civil building GB50176-2016' [27], Ri, Re are respectively taken as 0.11 (m 2 •K)/W and 0.04 (m 2 •K)/W in this study; R is the heat transfer resistance of the wall, (m 2 •K)/W. The thermal resistance R can be calculated as follows: ( ) where θ1j, θ2j are the jth measurement cycle of the inside and outside surface temperature of the wall, respectively, ℃; qj is the jth measurement cycle of the heat flow density of the wall, W/m 2 .

Description of Test Results
The test results are shown in Fig. 5. The value of CTave is the average value of CT1 to CT6. The value of HTave is the average value of HT1 to HT6. The value of HFave is the average value of HF1 to HF3. The average value of the heat flow densities (HFave) of different walls are compared, which is shown in Fig. 6. As can be seen from the Figs. 6(a), 6(b) and 6(c), the HFave of the wall with one EPS foam plate is about 37% to 51% lower than that of the same wall with no EPS foam plate. However, the HFave of the wall with two-layer EPS foam plates is comparable to the same wall with one-layer EPS foam plate, reducing less than 16%. The comparison of the HFave of different walls with the similar thickness is shown in the Fig. 6(d), which shows that the HFave of B3(30)E1 is the lowest and HFave is affected by the different configuration of the wall. In the condition of the same thickness of the bamboo scrimber plates, the HFave of B6(20)E0 is the lowest and the HFave of B3(40)E0 and B4(30)E0 are similar, which can be explained that when heat flow passes through cross layers, obstacles will appear to reduce the heat flow density. (c) (d) Figure 6: Comparison of the heat flow density

Analysis of Heat Transfer Coefficient
Based on the measured values and Eqs. (2), (3), the heat transfer coefficient in a single measurement cycle (generally 3 hours) is calculated. The average value Kave of two consecutive cycles with an error less than 1% is adopted, as listed in Tab. 5.  According to Chinese standard [28][29][30][31], CLB walls with thickness over 120 mm can meet the energy-saving design requirements in Hot summer and warm winter zone. EPS-CLB composite walls can meet the energy-saving design requirements in Moderate climate (A) zone and Hot summer and cold winter zone. The specimen B5(30)E2 whose K is 0.470 W/(m 2 •K) can even meet the energy-saving design requirements in Cold zone, showing excellent thermal performance.
(1) Effect of different thicknesses and configurations on heat transfer coefficients of CLB walls Heat transfer coefficients of CLB walls specimens without EPS foam plate are shown as Fig. 7. It is found that the heat transfer coefficient of the CLB wall decreases with the increase of the wall thickness. When the wall thickness increases from 88 mm in specimen B3(30)E0 to 145 mm in specimen B5(30)E0, the heat transfer coefficient decreases from 1.871 W/(m 2 ·K) to 1.096 W/(m 2 ·K). Even in specimen B5(30)E0, the heat transfer coefficient is still relatively high and larger than 1.0 W/(m 2 •K). According to Eq. (2), the thermal resistance R is inversely proportional to heat transfer coefficient K. With the increase of the wall thickness, R increases linearly and K decreases at a lower rate. It is uneconomical and impractical to realize the requirements of energy-saving design only depending on thickening the CLB wall.
The heat transfer coefficients of specimens B4(30)E0, B6(20)E0 and B3(40)E0 are similar and the maximum difference of the heat transfer coefficient between specimens B6(20)E0 and B3(40)E0 is 9.07%. The configuration has only little effect on the heat transfer performance. Therefore, the influence of configuration on heat transfer performance can be neglected.

Figure 7: Heat transfer coefficients of CLB walls (2) Effect of EPS foam plate on heat transfer coefficient of CLB wall
According to the heat transfer test of walls, the comparison of heat transfer coefficients of CLB walls and EPS-CLB composite walls is shown in Fig. 8. In general, the heat transfer coefficients of CLB walls are much larger than EPS-CLB composite walls. After adding one layer of 30-mm EPS foam plate to the external surface of the CLB wall, the heat transfer coefficient of the CLB wall decreases significantly. Compared with the specimen B3(30)E0 composed of three 30-mm bamboo scrimber plates, the heat transfer coefficient of specimen B3(30)E1 composed of three 30-mm bamboo scrimber plates and one 30-mm EPS foam plate decreases by 51.1%. Compared with the specimen B3(40)E0, the heat transfer coefficient of the B3(40)E1 specimen is reduced by 56.6%. Compared with the specimen B5(30)E0, the heat transfer coefficient of the specimen B5(30)E1 is reduced by 48.3%. Therefore, the use of the external thermal insulation has an important effect on the heat transfer performance of the CLB wall.
The heat transfer coefficient of specimen B3(40)E2 with two layers of 30-mm EPS foam plate decreases by 13.2% compared with the specimen B3(40)E1 which contains one layer of the EPS foam plate. Compared with specimen B5(30)E1, the heat transfer coefficient of specimen B5(30)E2 decreases by 17.1%. Compared with the efficiency of adding one layer of the thermal insulation plate to improve the thermal insulation effect, the efficiency of adding two layers of thermal insulation plates is obviously reduced. The contribution of the EPS thermal insulation plate to the reduction of heat transfer coefficient is becoming small with the increasing layers of EPS foam plates because the magnitude of the thermal resistance R is inversely proportional to heat transfer coefficient K rather than being linear.
The heat transfer coefficient of B3(40)E2 with a thickness of 178 mm is only 2.6% lower than that of B5(30)E1 with a thickness of 175 mm. The total thickness of specimens B3(40)E2 and B5(30)E1 is almost the same, but the load bearing of the specimen B5(30)E1 is obviously larger than that of the specimen B3(40)E2 because the total thickness of bamboo scrimber plates in B5(30)E1 is about 30 mm larger than that in B3(40)E2. Therefore, under the condition of same thickness, the configuration of specimen B5(30)E1 is more reasonable than that of specimen B3(40)E2. The thermal insulation capacity and load-bearing capacity should be considered simultaneously in the design of CLB wall.

Comparison of Theoretical and Experimental Values of Heat Transfer Coefficient
According to the thermal resistance calculation method of the wall composed of homogeneous material layers in ISO 6946:1996 [35], the thermal resistance of each layer is calculated to obtain the thermal resistance of the wall. The equation is proposed as follows: where δB is the total thickness of the bamboo scrimber plates, m; δE is the total thickness of the EPS foam plate, m. The above equation calculates the thermal resistance based on the thickness and thermal conductivity of each ideal homogeneous plate in the wall, while Eq.  As listed in Tab. 6, the theoretical heat transfer coefficient is generally larger than the experimental heat transfer coefficient. The reasons for the errors between the experimental and theoretical heat transfer coefficients are summarized as follows: (1) The actual EPS foam and bamboo scrimber are not completely homogeneous materials.
(2) There are voids between the layers in the wall, which makes the resistance of heat flow is relatively large when passing from one layer to anther adjacent layer.
(3) The water content has an effect on the thermal conductivity. Because the tests in Section 2.1 and Section 2.2 are not conducted at the same time, the water content of the walls in Section 2.2 may be different from that of the specimens in Section 2.1. The theoretical results of the heat transfer coefficient are obtained by the thermal resistance calculated by the thermal conductivity coefficient in Section 2.1. Therefore, there are some deviations between theoretical results and experimental results measured by the controlled box heat flow meter method directly.

Finite Element Analysis
In this section, numerical models were established by finite element software ABAQUS to analyze the steady-state heat transfer of the specimens. Firstly, due to the uncertainty of the thermal conductivity coefficient of adhesive λA provided by the manufacturer, the effect of different λA on the thermal insulation performance of walls was studied. Moreover, numerical simulation results were compared with the representative experimental results, including specimens B3(40)E0 and B5(30)E1, to determine the effectiveness of the established numerical models. Finally, the effect of EPS foam plate location on thermal insulation was explored.

Basic Hypothesis
Because of the complexity of factual situation, some assumptions are adopted in the finite element analysis (FEA) for ease of analyzing the heat transfer of the walls without affecting the correctness. The adopted assumptions are summarized as follows: (1) The layers of CLB wall and EPS-CLB composite wall (bamboo scrimber plates, adhesive layers and EPS foam plate) are all homogeneous and isotropic materials. The thermophysical parameters of these materials are constant and are not affected by temperature, humidity and pressure, etc. (2) The interfaces between layers of the wall fully contact with each other and the influence of the interface thermal resistance [36] is not considered. (3) The atmosphere temperature is assumed as constant.

Establishment of Numerical Model
According to Section 2.2.1, the length and width of the models B3(40)E0 and B5(30)E1 are all 980 mm and the thicknesses of these two models are 120 mm and 204 mm, respectively. The thickness of adhesive layer was taken as 0.1 mm. Based on above test results, the thermal conductivities of bamboo scrimber and EPS foam is 0.263 W/(m•K) and 0.04 W/(m•K), respectively. The specific heat capacities C of bamboo scrimber and EPS foam are 1570 J/(kg•K) and 1380 J/(kg•K) [27,37]. The thermophysical parameters of adhesive are provided by the manufacturer, which are listed in Tab. 7. 8-node linear heat transfer brick elements (DC3D8) were used. Different layers were tied together, which made two materials at the same temperature on the contact surface. In order to compare with experimental results, the temperatures on both sides of the wall were set to 26 o C and -5 o C to simulate indoor and outdoor environment, respectively. According to Chinese standard GB50176-2016 [27], the surface heat transfer coefficients of internal and external surface are 8.7 W/(m 2 •K) and 23.0 W/(m 2 •K), respectively. The solid surface emissivity εr is 0.9 [38] considering surface radiation between atmosphere and both surfaces of the wall. The absolute zero and Stefan-Boltzmann constant [38] were set to -273.15 o C and 5.67 × 10-8 W/(m 2 •K 4 ) in the numerical model, respectively. In order to improve the accuracy of heat transfer analysis, the mesh was refined locally in the direction of wall thickness.

Effect of Different Thermal Conductivity of the Adhesive λA on the Heat Transfer of Walls
Because of the uncertain range of thermal conductivity of the adhesive listed in Tab. 7, the effect of different values of λA on the heat transfer of the tested walls is analyzed in this section. The specimen B6 (20)E0 is chosen to be analyzed when λA varies from 0.3 W/(m•K) to 0.6 W/(m•K), because the proportion of the total thickness of the adhesive layer to the total thickness of the wall is the largest in all tested walls and the effect caused by the difference of λA on the heat transfer of the wall is largest. The temperature distribution contours and heat flow density distribution contours in Z direction of B6(20)E0 are shown in Fig. 9 and Fig. 10.   (20)E0 increases by only 0.139% from 1.6500 W/(m 2 •K) to 1.6523 W/(m 2 •K), which can be neglected. Therefore, the effect of different λA on the heat transfer coefficients of all tested walls can be neglected and the thermal conductivity coefficient of the adhesive λA is chosen to be 0.3 W/(m•K) in the following finite element analysis.

Comparison of Results of Numerical and Experimental Results
Based on the above modeling method, models B3(40)E0 and B5(30)E1 are established to conduct heat transfer analysis. The temperature distribution contours of models B3(40)E0 and B5(30)E1 are shown in Fig. 11. The heat flow density distribution contours of the two models in the X, Y and Z directions are shown in Fig. 12 and Fig. 13, respectively. According to Fig. 11, the temperature distributes uniformly in X and Y directions and varies in a gradient in Z direction. The temperature decreases gradually along Z direction because of thermal insulation effect of the wall. As seen from Fig. 12 and Fig.  13, the magnitude of heat flow density in X and Y direction is only 10 -12 , which can be ignored. Therefore, the assumption of one-dimension heat transfer can be accepted when the length and width of the specimen are much larger than the thickness.  Fig. 14, the experimental temperature cannot reach the temperature calculated by the finite element method. There are probably two reasons which can explain this phenomenon. The first reason is that there are still some voids between the specimen and specimen holder in the experiment to produce the air convection between hot box and cold box, although the polyurethane foaming agent is used to fill the voids around the specimen. The second reason is that the heat loss exists in the actual experiment, which is not taken into account in the finite element calculation.  15 shows that the simulation values obtained by FEA are higher than the experimental values. This may be due to the second assumption introduced in Section 3.1.1 that the interfaces between layers of the wall are fully contacted and the influence of interface thermal resistance is not considered. In fact, the simulation results are higher than the experimental results because of the voids and interface thermal resistance in the actual specimens. According to Eqs. (2) and (3) in the Section 2.2.2, the simulation value of heat transfer coefficient Ks can be calculated and compared with experiment results Ke, as listed in Tab. 9. The error between Ke and Ks of the two specimens is less than 13%, which demonstrated the accuracy of the proposed finite element model.

Effect of EPS Foam Plate Location on Heat Transfer Coefficient in the EPS-CLB Composite Wall
The simulation value of heat transfer coefficient calculated by ABAQUS is proved to be efficient, the numerical model can thus be used to analyze the effect of EPS foam plate location on heat transfer coefficient in the EPS-CLB composite wall. Three B4(30)E1 models with different positions of EPS foam plate have been established, namely, the EPS foam plate placed on the cold side of CLB wall, the hot side of CLB wall and in the middle of CLB wall. These three models are named as Cold, Hot and Middle, respectively. The temperature distribution contours and heat flow density distribution contours in Z direction of models were obtained, as shown in Fig. 16 and Fig. 17, respectively.
It can be seen from Fig. 16 and Fig. 17, the temperatures and heat flow densities of these three models are very close after entering into the steady state of heat transfer. According to Eqs. (2) and (3) in the Section 2.2.2, the simulation value of heat transfer coefficients of these three models can be calculated and compared, as shown in Tab. 10. As listed in Tab. 10, the heat transfer coefficients of the cases Cold and Hot are 0.56% and 2.79% higher than that of the case Middle, respectively. The effect of thermal insulation is best when EPS foam plate is placed in the middle of EPS-CLB composite wall. However, the construction quality of the sandwich insulation is difficult to be inspected and the sandwich insulation will produce thermal bridge [39]. Therefore, the EPS-CLB composite wall adopting external thermal insulation (the EPS foam plate placed on the cold side of the wall) is relatively recommended, considering both of the thermal insulation effect and construction quality.

Conclusion
The thermal conductivity coefficients of bamboo scrimber plate and EPS foam plate are tested through guarded hot plate method. The heat transfer coefficients of CLB walls and EPS-CLB composite walls are tested through the temperature-controlled box-heat flow meter method and compared with the theoretical derivations. Furthermore, the influence of different locations of the EPS foam plate on the thermal insulation performance is analyzed by the finite element method. Based on experimental, theoretical and numerical analyses, the following conclusions can be drawn: 1. The thermal conductivity of bamboo scrimber plate λB is about 0.263 W/(m•K), which is far fewer than that of most existing common inorganic materials used in building, about 15% of reinforced concrete and about 50% of hollow brick and light-weight aggregate concrete. λB slightly increases with the increase of the average temperature on both sides of the bamboo scrimber plate. The thermal conductivity of EPS foam plate λE is 0.04 W/(m•K). 2. The heat transfer coefficient K of the CLB wall decreases with the increase of the thickness of the wall. The effect of configuration on K can be ignored. The heat transfer coefficient of EPS-CLB composite wall with one layer of EPS foam plate is significantly lower than that of CLB wall with no EPS foam plate. In terms of improving the thermal insulation effect, the efficiency of adopting two layers of EPS foam plates is obviously reduced when compared with the efficiency of adding one layer of EPS foam plate. The theoretical calculations of the heat transfer coefficient of composite walls are compared with the experimental values. The average error is about 15.36%, which is acceptable.
3. According to Chinese standard [28][29][30][31], most CLB walls can be used in Moderate climate (B) zone and Hot summer and Warm winter zone. However, the EPS-CLB composite walls can be used in Moderate climate (A) zone, Hot summer and cold winter zone and even Cold zone. 4. When thermal conductivity coefficient of adhesive λA ranges from 0.3 W/(m•K) to 0.6 W/(m•K), the influence of the change of λA on the heat transfer coefficients of walls can be neglected. The error between the experimental and numerical results of K is less than 12.75%, which proves the validity of finite element models. The influence caused by different locations of the EPS foam plate in EPS-CLB composite wall can be neglected.