Advertisement

Wood Science and Technology

, Volume 52, Issue 3, pp 733–752 | Cite as

Moisture content monitoring in glulam structures by embedded sensors via electrical methods

  • Hang Li
  • Marianne Perrin
  • Florent Eyma
  • Xavier Jacob
  • Vincent Gibiat
Original
  • 180 Downloads

Abstract

In recent years, an increasing number of glulam structures have emerged in civil engineering, but their development is limited by durability issues. Degradations related to excessive moisture content (MC) or wetting/drying cycles were observed and can lead to severe structural damages. As a result, the development of continuous monitoring techniques of wood MC becomes essential. Currently, the mostly used MC control methods are based on electrical measurements (resistive or capacitive). Nevertheless, existing solutions are not practical to conduct measurements inside every lamella of glulam. In the light of these observations, it is proposed to transform glulam into “smart material” by embedding the MC monitoring system either inside or between lamellas, considering the major constraints of fabrication of this material (the small glue line thickness, the important bonding pressure, etc.). To achieve this, 4 measurement configurations using “pin-type” or “surface” electrodes were investigated with the following steps. First, a feasibility study was conducted to make sure of the good functionality of the measurement configurations. Thereafter, the selected configurations were used to monitor the MC variation (10–70%) in glulam specimens. At the same time, the influence of electrode spacing and bonding pressure on the electrical measurements was also investigated. Results show that the selected configurations are operational for the local MC measurement in the lamellas of glulam, regardless of the types of sensors used or the measured physical quantities (electrical resistance or capacitance). This study aims to provide valuable information for the future development of embedded MC monitoring system in glulam structures.

Introduction

Today, more and more timber structures are used in building and civil engineering due to new challenges imposed by sustainable development and thanks to their competitive costs. In recent years, glulam has been gaining popularity in construction due to not only its higher mechanical strength compared to solid wood, but also the possibility it offers to build large-span structures (SETRA 2006). However, the problem of durability is a limiting factor for the development of these structures (D’Ayala et al. 2014). Pathologies such as cracks, delaminations or slots, as well as fungal or insect attacks, have been observed on wooden infrastructures. The majority of them can be attributed either to excessive moisture content (MC) (> 22%) or to the wetting/drying cycles in the material (SETRA 2006; Dietsch et al. 2014a). In order to promote the use of wood in construction, infrastructure supervisors have expressed their need for continuous monitoring techniques of wood MC (Kasal 2013).

Currently, the most used MC control methods for wood material are based on electrical measurements (Skaar 1988). They consist in determining wood MC through resistive or capacitive measurements and are widely used in wood industry in the form of moisture meters (James 1963). Nevertheless, both types of moisture meters have their limits: the resistive moisture meter requires the insertion of metallic electrodes in the material, which can harm the material integrity, whereas the capacitive moisture meter is sensitive to the surface conditions of the material (roughness, MC, etc.), which means the measured results are strongly influenced by the surface condition (Kasal and Lear 2011).

In the literature, several studies have shown the possibilities using resistive and capacitive methods with different types of sensors for the continuous monitoring of wood MC (Stamm 1927; Jazayeri and Ahmet 2000; Brischke et al. 2008a; Moron et al. 2016).

Most of the resistive methods are based on the use of “pin-type” electrodes inserted in the wood specimens (Stamm 1927; Fredriksson et al. 2013). Nevertheless, during long-term monitoring, problems in maintaining electrical contact between the electrodes and wood occurred due to dimensional changes of wood (Skaar 1964; Dai and Ahmet 2001). As a result, several authors proposed improved solutions such as using screws (Hjort 1996; Norberg 2000) or applying conductive glue between the coated metallic cable and wood (Brischke et al. 2008b; Fredriksson et al. 2013). On the other hand, a recently published study (Bjorngrim et al. 2017) has explored the feasibility of MC measurement in the glulam beams. The authors used two stainless steel threads clamped between two glulam beams to conduct resistive measurements at the interface of the two beams. Results showed that it is possible to discriminate three MC levels (12, 20 and 25%). In recent years, the MC monitoring in wooden buildings or bridges is reported to have been initiated in several European countries (Franke et al. 2013; Dietsch et al. 2014b; Pousette et al. 2014; Bjorngrim et al. 2016). The pin-type sensors such as screws or nails were used. The monitoring results show that it is possible to use these types of electrodes for MC monitoring in timber structures since the sensors generally sent data corresponding to the climate changes although malfunction of sensors in sending data happened in several cases.

Regarding the capacitive method, the “surface electrodes” were used for the MC monitoring in wood (James 1977; Kabir et al. 1998; Moron et al. 2016). This type of electrodes consists in the use of two parallel metallic plates placed on the outer surfaces of the specimen. Although the feasibility of surface-type electrodes for the MC measurement in solid wood specimens has been proven, no information exists concerning the application of continuous monitoring to wooden structures.

According to the studies cited above, work that has been done on MC monitoring of glulam structures with embedded sensors consists in placing sensors from the outer surfaces. However, it is necessary to realize internal and local MC measurement either inside or between lamellas to obtain more relevant information and furthermore to establish the relationship between the MC (or wetting/drying cycles) and the durability of the structures. The aim of the MC monitoring is to inform the infrastructure supervisors as early as possible of the potential risk of damage so that appropriate maintenance operations can be planned. To achieve this, MC sensors should be embedded either inside or between lamellas, further to placing the sensors from the outer surface as presented by previous studies. Given the industrial fabrication process of glulam beams according to EN 14080 (2013), the major constraints for sensor integration are a simple and economical implementation, the small glue line thickness (about 0.3 mm) and the bonding pressure (about 10 bars).

In the light of these observations, it is proposed to transform the glulam into “smart material” by embedding the MC monitoring systems in order to perform internal and local MC measurements either inside or between lamellas. To achieve this, 4 measurement configurations were identified from low to medium cost (0.9–3 €, material purchasing price), using “pin-type electrodes” or “surface electrodes” in order to conduct electrical measurements (resistive/capacitive). The objectives of this study are: (1) to make sure of the good functionality of the 4 measurement configurations; (2) to verify the feasibility of the 4 measurement configurations in the measurement of MC variation in glulam structures; (3) to investigate the influence of bonding pressure and electrode spacing on the electrical measurements.

Materials and methods

Specimen preparation

Four measurement configurations using different electrode setups were identified in this study (Fig. 1). Five identical specimens were prepared for each configuration tested. The specimens were prepared from Douglas fir (Pseudotsuga menziesii), because this species is extensively used in construction (SETRA 2006). In the first place, the configurations 1, 2 and 3 were proposed to achieve resistive measurements. The configurations 1 and 3 were proposed since using screws (Norberg 2000) or applying conductive glue between the coated metallic cables and wood (Brischke et al. 2008b; Fredriksson et al. 2013) can help in maintaining the electrical contact between the electrodes and wood. Configuration 2 was proposed to be a contrast/intermediary configuration for the configurations 1 and 3 (with drilled holes compared to configuration 1, and different types of electrode compared to configuration 3). On the other hand, configuration 4 was proposed since the “surface electrodes” have often been used for capacitive measurements (Moron et al. 2016). Moreover, the resistive measurements can also be conducted in this configuration (Stamm 1927).
Fig. 1

Measurement configurations: a configuration 1; b configuration 2; c configuration 3; d configuration 4A; e configuration 4B (the dimensions are in mm)

In order to investigate the influence of bonding pressure on the sensor performance, both solid wood and glulam specimens were prepared for each configuration. The thickness of lamellas (or specimen) is 33 mm because it is recommended for the construction of large-span structures due to its good management of the swelling and shrinking of wood (SETRA 2006). In order to facilitate the water absorption, the width of the specimens is fixed at 90 mm since it is the smallest width for glulam in industry.

Furthermore, to verify the influence of the electrode spacing on MC measurements, specimens were prepared with two different electrode spacings. As for configurations 1, 2 and 3, two spacings (20 and 40 mm) were used on the same specimen (Fig. 1). Regarding configuration 4, the two spacings were realized with two different specimens with respective thicknesses of 33 and 16.5 mm (Fig. 1d, e). In order to distinguish the specimens with different electrode spacings in configuration 4, those with an electrode spacing of 33 mm were named configuration 4A, and those with an electrode spacing of 16.5 mm configuration 4B. The specimen preparation details of each configuration were described as follows:

Configuration 1 uses stainless screws (Ø3*20 mm) as MC sensors. The screws were driven (in the middle of the width of specimen) directly into the wood with a screwdriver. Then, polyurethane-coated copper cables (wire gauge: AWG 36) were glued to the screw head with a conductive glue (EPO-TEK E4110). The conductive glue is a two-component glue with silver-filled epoxy paste. Its conductivity becomes stable after 72 h at ambient temperature.

Configuration 2 also uses stainless screws as MC sensors, but they were placed in the pre-drilled counter-bored holes (in the middle of the width of specimen). Then, the coated copper cables were glued to the screw head with the conductive glue. At last, the space left at the bottom holes (Ø3.2*11 mm) was filled with the conductive glue (0.1 mL) in order to guarantee the electrical conduction, and the space at the top holes (Ø8*10 mm) was filled with the same polyurethane glue as used for glulam bonding (KLEIBERIT PUR 510 FiberBond) to guarantee an electrical insulation.

Configuration 3 uses directly the copper cables as the MC sensors. The cables were stripped off 11 mm at the end. They were then folded and placed in pre-drilled holes with the same dimensions as for configuration 2. At last, the space left at the bottom holes was filled with the conductive glue (0.2 mL), and the space at the top holes was also filled with the polyurethane glue for electrical insulation purposes.

Configuration 4 uses the copper patches (Ø50 mm, total thickness: 0.065 mm) as MC sensors. They were then stuck directly with their own adhesive layers (which are also conductive) on the center of the specimen, to make sure that the measurements are conducted in the wood material itself and not in the glue lines. At last, the coated copper cables were glued to the center of the patches using the conductive glue. The center was chosen in order to have a more uniform current distribution in the specimen. It should be pointed out that preliminary tests were conducted in order to choose the patch diameter. Tests were conducted with 4 different diameters (20, 30, 40 and 50 mm), and results showed that it is possible to discriminate the variation of MC with all the dimensions. At last, it was decided to use 50 mm to reduce the uncertainty brought by the measurement devices, given that at low MC levels, both the capacitance (only several pF) and resistance (more than 10 GΩ) are close to the measurement limits.

Following the previous steps, solid wood specimens were prepared as described below. For the configurations 1 and 2, the polyurethane glue was applied onto and around the screw head to assure electrical insulation. For configuration 4, a thin layer of polyurethane glue was applied on the patches for the same reason and to prevent the patches from peeling off during moistening. Then, the glulam specimens were prepared using a hydraulic press with a bonding pressure of 10 bar applied during 24 h. In order to obtain glulam, one more lamella (33 mm in thickness) was glued for the configurations 1, 2 and 3 on the screwing/drilling side; two more lamellas (16.5 mm in thickness) were glued for configuration 4 on the two sides where patches had been pasted. The characteristic and measurement details of the specimens of each configuration (for both solid wood and glulam specimens) are presented in Table 1.
Table 1

Characteristic and measurement details of the specimens

Configuration

Namea

Dimensions (mm)

Type of electrode

Electrode spacing (mm)

Measurement directionb

Measured quantity

L

R

T

1

C1S

100

33

90

Stainless screws

20; 40

L

Electrical resistance

 

C1GL

100

66

90

Stainless screws

20; 40

L

Electrical resistance

2

C2S

100

33

90

Stainless screws

20; 40

L

Electrical resistance

 

C2GL

100

66

90

Stainless screws

20; 40

L

Electrical resistance

3

C3S

100

33

90

Polyurethane-coated copper cables

20; 40

L

Electrical resistance

 

C3GL

100

66

90

Polyurethane-coated copper cables

20; 40

L

Electrical resistance

4A

C4AS

90

33

90

Copper patches

33

T

Electrical resistance and capacitance

 

C4AGL

90

66

90

Copper patches

33

T

Electrical resistance and capacitance

4B

C4BS

90

16.5

90

Copper patches

16.5

T

Electrical resistance and capacitance

 

C4BGL

90

49.5

90

Copper patches

16.5

T

Electrical resistance and capacitance

aThe “solid wood” and “glulam” specimens are designated, respectively, by the letters “S” and “GL” at the end of their names

bL, R and T indicate, respectively, the 3 anatomical directions of wood (longitudinal, radial and tangential)

Experimental procedures

Two procedures were used to change the MC of the specimens. Concerning the measurements made at the MC below 20%, a climate chamber which allows changing the temperature and the relative air humidity was used. The specimens were considered to have attained the equilibrium MC when the mass was stable. To attain the MC between 20 and 70%, the specimens were directly immersed in water. The measurements were realized following a schedule fixed in accordance with the results of a study of humidification kinetics conducted prior to launching the experiment. According to the results of this study, the MC and the square root of time present an increasing linear relationship between them. This linear relationship was also observed in the literature (Kumaran 1999). Using this relationship, the time of immersion could be estimated for a given MC to establish the schedule.

Periodically, the electrical resistances for all specimens were measured with a Giga-ohmmeter developed at the laboratory.

For the configurations 1, 2 and 3, in order to distinguish the electrical resistances measured between the two pairs of electrodes, R20 was named the resistance measured between the 20-mm spaced electrodes, and R40 the resistance measured between the 40-mm spaced electrodes. Furthermore, in order to verify the influence of electrode spacing, the wood resistance and the contact resistance associated with the wood/electrode interface were also calculated. According to the literature (Tamme et al. 2012), the measured resistance can be considered as an equivalent series resistance of the wood resistance and the contact resistance. As a result, R20 and R40 can be written in the following form:
$$R_{20} = R_{{{\text{w}}1}} + 2R_{{{\text{c}}1}}$$
(1)
$$R_{40} = 2R_{\text{w1}} + 2R_{{{\text{c}}1}}$$
(2)
where Rw1 represents the wood resistance between the 20-mm spaced electrodes, and Rc1 is the contact resistance. R40 is considered to contain twice the Rw1 since the length of the conducting path is doubled (according to Eq. 3):
$$R = \rho \frac{l}{A}$$
(3)
where R is the electrical resistance (Ω), ρ is the resistivity of material (Ω m), l is the length of the material (m), and A is the cross-sectional area (m2). Solving Eqs. 1 and 2 leads to:
$$R_{{{\text{w}}1}} = R_{40} - R_{20}$$
(4)
$$R_{{{\text{c}}1}} = \frac{{2R_{20} - R_{40} }}{2}$$
(5)
As for the configurations 4A and 4B, in order to distinguish the electrical resistances measured on the two types of specimens, R33 is named the resistance measured on the specimens 4A, and R16.5 the resistance measured on the specimens 4B. For the same reasons as above, the wood resistance (Rw2) and the contact resistance (Rc2) can be expressed by:
$$R_{{{\text{w}}2}} = R_{33} - R_{16.5}$$
(6)
$$R_{{{\text{c}}2}} = \frac{{2R_{16.5} - R_{33} }}{2}$$
(7)
Regarding the capacitive measurements for configuration 4, a LCR meter (GW Instek LCR-816) was used. In order to distinguish the capacitances measured on the two types of specimens, C33 is named the capacitance measured on the specimens 4A, and C16.5 the capacitance measured on the specimens 4B. Furthermore, if the measured capacitance is considered as a series connection of wood capacitance and contact capacitance, the C33 and the C16.5 can be written in the following form (Terzic et al. 2012):
$$\frac{1}{{C_{33} }} = \frac{1}{{C_{\text{w}} }} + \frac{2}{{C_{\text{c}} }}$$
(8)
$$\frac{1}{{C_{16.5} }} = \frac{1}{{2C_{\text{w}} }} + \frac{2}{{C_{\text{c}} }}$$
(9)
where Cw represents the wood capacitance of the specimens 4A and Cc is the contact capacitance associated with the wood/electrode interface. C16.5 is considered to contain twice the Cw since the distance between is halved (according to Eq. 10) (Terzic et al. 2012):
$$C = \varepsilon_{0} \varepsilon_{\text{r}} \frac{A}{d}$$
(10)
where C is the capacitance (F), \(\varepsilon_{0}\) is absolute permittivity of vacuum (≈ 8854 * 10−12 F m−1), \(\varepsilon_{\text{r}}\) is the dielectric constant (CD, also called relative permittivity) which determines some kind of ability to store electrical charge, A is the area of the plates (m2), and d is the distance (m) between them. The solutions of Eqs. 8 and 9 are:
$$C_{\text{w}} = \frac{1}{{2*(1/C_{33} - 1/C_{16.5} )}}$$
(11)
$$C_{\text{c}} = \frac{2}{{2/C_{16.5} - 1/C_{33} }}$$
(12)
The frequency range of the LCR meter extends from 100 Hz to 2 kHz. According to Torgovnikov (1993), it has already been confirmed the CD (proportional to the capacitance, cf. Eq. 10) varies as a function of MC at different frequencies from 0.01 Hz up to the range of GHz. James (1986) realized the measurements of CD at 4 different frequencies (0.2, 1, 10 and 100 kHz) and concluded that the sensibility for MC monitoring increases with an increasing frequency. The present preliminary tests at 0.1 and at 2 kHz (the two extreme values of this measurement device) showed that 2 kHz allows to better discriminate the different MCs (Fig. 2), which is in accordance with the conclusion of James (1986). As a consequence, the frequency is fixed at 2 kHz in this study.
Fig. 2

Capacitance at 0.1 and 2 kHz on wood specimens

The specimens were also weighed at each measurement in order to calculate the exact MC by gravimetric method with the help of the oven-dry weight, according to the European standard EN 13183-1 (2002).

Results and discussion

Validation of electrode instrumentation in wood

The implementation of the different measurement configurations was firstly achieved on solid wood specimens. The measurements at initial state (10% in MC) on these specimens have shown the possibility to realize electrical measurements in wood (Fig. 3). As for the resistive measurements, it was found that electrical resistance is in the order of 109 Ω for the configurations 1, 2 and 3 (in the longitudinal direction) and 1010 Ω for configuration 4 (in the tangential direction). These values are in the same order of magnitude as by extrapolating the information available in the literature (James 1963; Skaar 1988) (Fig. 3).
Fig. 3

Electrical resistance and capacitance of solid wood specimens measured at initial state (10% in MC) compared to results in the literature (configurations 1, 2 and 3: sensor distance = 40 mm; configuration 4: sensor distance = 33 mm) (*resistance for pin-type electrodes (sensor distance = 32 mm); **resistance for surface electrodes; ***capacitance at 1 kHz)

Regarding the capacitive measurements, the values of capacitance found in this study were also in the same order of magnitude with information available in the literature using similar frequencies (James 1977) (Fig. 3). In the present study, the capacitance measured on the 33-mm-thick specimen is 2.3 pF on average, while it is 5.0 pF by extrapolating the results of James (1977) at 1 kHz to the same MC and thickness. The difference can be explained by the different frequencies used since the present results were measured at 2 kHz.

MC monitoring in solid wood specimens

Thereafter, the feasibility of MC monitoring in solid wood specimens was investigated using the moistening procedures described previously. The results of resistive and capacitive measurements are presented in Figs. 4 and 5, respectively. The model (Eq. 13) used in this study to correlate MC and electrical resistance was firstly proposed by Stamm (1927) and has been widely used in the literature (Norberg 2000; Tamme et al. 2012; Fredriksson et al. 2013):
$$\log \left( R \right) = a + b*\log \left( {\text{MC}} \right)$$
(13)
where R (Ω) is the electrical resistance, MC (%) is the moisture content, a and b are the constants depending on species.
Fig. 4

Logarithm of electric resistance as a function of logarithm of MC for solid wood specimens (configurations 1, 2 and 3: sensor distance = 40 mm; configuration 4: sensor distance = 33 mm)

Fig. 5

Capacitance as a function of MC for both solid wood and glulam specimens (configuration 4, sensor distance = 33 mm)

Regarding the capacitive measurements, Eq. 14 is used for correlation since it is the most extensively used in the literature according to Torgovnikov (1993):
$$C = c*{\text{MC}}^{d}$$
(14)
where C (F) is the capacitance, MC (%) is the moisture content, c and d are the constants depending on species.

First of all, it can be observed that it is possible to monitor MC with all the 4 measurement configurations in the range of this study. The results before and after 20% MC were analyzed separately since two different trends can be observed. This can be explained by the two different moistening procedures used in this study, which have led to different moistening physical processes. Before 20% in MC, the water sorption is mainly in the form of water vapor, while after 20% in MC, liquid water absorption is involved (Siau 1984). This means that mainly bound water was present in the specimens before 20% in MC, whereas after 20% in MC, free water began to be absorbed in the intercellular cavities of wood. According to the literature, the influence of bound and free water on the resistive/capacitive properties is different (Lin 1967; Torgovnikov 1993), which can explain the two different trends observed.

It is interesting to point out that the evolutions of electrical resistance of configurations 1, 2 and 3 are almost the same, while that of configuration 4 differs from them (Fig. 4). Several explanations can be given to account for this difference. First, according to the literature, when surface electrodes are used, the measured resistance tends to be affected by the surface roughness of the specimen (Holm 1967). Surface roughness can result in a smaller effective contact area than in the ideal case where the electrodes and specimen are 100% in contact. This will lead to a higher resistance measured than the real value. Second, when wood MC is below 20%, only bound water is present, so that the characteristics of contact interface are not significantly changed with the MC variation. This is why only a slight decrease of resistance is observed from 10 to 20% in MC (Fig. 4). However, from 20% in MC, free water begins to enter into the free space of wood, including the space between the electrodes and the specimen. This can lead to an improvement in the electrical contact brought by the conductivity of water. As a consequence, the decrease of resistance becomes more and more significant from 20 to 70% in MC, and finally the resistance ends up at values close to configurations 1, 2 and 3 (Fig. 4).

Through the discussion of results obtained on solid wood specimens, it turns out that
  1. 1.

    The 3 configurations with pin-type electrodes (configurations 1, 2 and 3) are of equal competence for the MC monitoring since they have the same performances, although they are of different costs;

     
  2. 2.

    For configuration 4, it is better to use the capacitive method for the estimation of MC in view of the results of resistive measurements in low MC levels.

     

In the next paragraph, the results obtained on glulam specimens will be discussed.

MC monitoring in glulam specimens

The feasibility of MC monitoring using the proposed measurement configurations was also investigated on glulam specimens. The results of resistive and capacitive measurements are presented in Figs. 5 and 6, respectively. Foremost, it can be observed that it is possible to monitor MC with all the 4 measurement configurations in the range of this study.
Fig. 6

Logarithm of electric resistance as a function of logarithm of MC for glulam specimens (configurations 1, 2 and 3: sensor distance = 40 mm; configuration 4: sensor distance = 33 mm)

Concerning the resistive measurements, the 4 configurations have almost the same performances as shown by the regression coefficients (a, b in Eq. 13, Fig. 6). Nevertheless, this observation is different from that of the solid wood specimens on configuration 4 (cf. Figs. 4, 6). This difference can be explained by the application of the bonding pressure, leading to a better electrical contact between the electrodes and the specimen (Vermaas 1975). Nevertheless, regarding the configurations 1, 2 and 3, similar trends were obtained on solid wood and glulam specimens (cf. Fig. 7). Actually, the bonding pressure was not directly applied to the sensors so that the electrical contact between the electrodes and wood was not changed by the pressure.
Fig. 7

Logarithm of electric resistance as a function of logarithm of MC (configuration 3, sensor distance = 40 mm)

Regarding the capacitive measurements, it was found that the capacitance of the glulam specimens is greater than that of the solid wood specimens (Fig. 5). This can also be explained by the application of the bonding pressure, which has improved the electrical contact between the electrodes and the specimen. As a result, the effective contact area increased after the application of the bonding pressure, leading to a higher capacitance measured (Dervos and Michaelides 1998). On the other hand, the difference in capacitance between glulam and solid wood specimens is observed to increase with increasing MC (Fig. 5). A possible explanation is given as follows. The capacitance of wood is contributed by the permanent dipoles (hydroxyl groups, molecule of free or bound water, etc.) in the material (Skaar 1988). Since the glulam specimens have two more lamellas than the solid wood specimens, even if the back of the electrode is insulated, the presence of water in the two outer lamellas can affect the electric displacement field around the electrodes. As a result, the dipole components in the two outer lamellas might have contributed to the measured capacitance, which can explain why the difference in capacitance increases as a function of MC. In order to confirm this hypothesis, capacitive measurements were firstly conducted on solid wood specimens (immersed in water) without wiping the water off the exterior surfaces before measurement. Then, the water was wiped off the surfaces completely and the same measurements were conducted. It turned out that results of capacitance obtained with water on the exterior surfaces are on average 10% higher, which can support this hypothesis.

As a further step, the measurement uncertainty is also calculated in this study according to the dispersion in resistance/capacitance among the 5 specimens (see vertical error bars in Fig. 8). For most of the measurements, the uncertainty is less than ± 1% (in MC) for MC below 20% and less than ± 5% (in MC) for MC above 20%. Moreover, the measurement uncertainty (see horizontal error bars in Fig. 8) tends to increase with increasing MC (Fig. 8). Fredriksson et al. (2013) made similar observation, and this might be due to the increasing amount of free water in the specimens (Lin 1967).
Fig. 8

Electric resistance and capacitance as a function of MC with MC measurement uncertainty of glulam specimens of configuration 4 (sensor distance = 33 mm)

Influence of electrode spacing

In order to investigate the influence of electrode spacing, the results of electrical resistance/capacitance obtained with different electrode spacings were compared. In order to determine whether there exists a significant difference between them, the Fisher’s LSD (least significant difference) tests (Muth 2014) were conducted at 95% confidence level with the software Minitab 17.

Regarding the resistive measurements, it turned out that no significant difference can be observed in the MC range under study for the configurations 1 (Fig. 9), 2 and 3. This can be explained by the fact that the contact resistance, which can be calculated with Eq. 5, contributes to the majority of the total resistance measured (Fig. 10). Similar observation was also made in the literature (Skaar 1988; Forsén and Tarvainen 2000). It is believed that the great contact resistance when pin-type electrodes are used can be attributed to the small contact area between electrodes and wood (Skaar 1964).
Fig. 9

Comparison between electrical resistances measured with different electrode spacings (glulam specimens, configuration 1)

Fig. 10

Contribution of wood resistance and contact resistance to total resistance of glulam specimens of configuration 1 (sensor distance = 40 mm)

In respect of configuration 4, the difference between results obtained with different electrode spacings is observed to be insignificant from 10 to 35% in MC (Fig. 11). Nevertheless, from 40% in MC, the difference becomes significant. This observation can be explained by the decreasing trend of the contribution of contact resistance with increasing MC (calculated with Eq. 7) (Fig. 12). When MC is small (especially below 20%), the contact resistance is very large compared to the wood resistance (due to the surface roughness, as explained earlier); then, the results obtained in specimens of different thicknesses cannot be differentiated. However, when liquid water begins to occupy the free space in wood, it enters gradually into the space between the electrodes and the specimen. As a result, the electrical contact is improved thanks to the conductivity of water, which can explain why the contribution of contact resistance decreases as a function of MC beyond the fibers saturation point (27% for Douglas fir; Gérard et al. 2011).
Fig. 11

Comparison between electrical resistances measured with different electrode spacings (glulam specimens, configuration 4)

Fig. 12

Contribution of wood resistance and contact resistance to total resistance of glulam specimens of configuration 4 (sensor distance = 33 mm)

With respect to the capacitive measurements, the capacitance measured on the 16.5-mm-thick specimens was observed to be higher than that measured on the 33-mm-thick specimens on average (Fig. 13). This observation can be explained by the theoretical equation (Eq. 10) which shows that the capacitance will be doubled if the material thickness is halved. However, these results do not show twofold relationship. For MC below 20%, the results show a relationship of 1.8 times. The slight difference can be explained by the contribution of contact capacitance (Tereshchenko et al. 2011), which can be calculated with Eq. 12 (Fig. 14). Above 20% in MC, difference between results obtained on specimens with different thicknesses becomes insignificant. Two facts can explain this observation. First, as mentioned earlier in this paper, the dipole components in two exterior lamellas of the glulam specimens contribute to the measured capacitance. Second, the contribution of contact capacitance is observed to increase with increasing MC (Fig. 14). This observation can be attributed to the infiltration of the free water at the wood/electrodes interface. The free water replaces gradually the air gap between electrodes and wood (Dervos and Michaelides 1998). Since the DC (proportional to the capacitance, cf. Eq. 10) of water (DC ≈ 80) is more important than that of air (DC ≈ 1), the contribution of interface increases with increasing MC (Torgovnikov 1993).
Fig. 13

Comparison between capacitances measured with different electrode spacings (glulam specimens, configuration 4)

Fig. 14

Contribution of wood capacitance and contact capacitance to total capacitance of glulam specimens of configuration 4 (sensor distance = 33 mm)

Conclusion

Within the framework to develop embedded MC monitoring systems in the lamellas of glulam structures in order to improve the infrastructure durability, 4 measurement configurations were proposed and tested in this study, considering the major constraints of fabrication of glulam (the small glue line thickness, the important bonding pressure, etc.). Results showed that it is possible to monitor the MC from 10 to 70%, in the core of wood specimens with all 4 configurations (in the longitudinal or tangential direction).

Alternative options, such as using substitute dowels (Brischke et al. 2008a) or relative air humidity (RH) sensors in bore holes (Melin and Bjurman 2017), also exist in the literature. Nevertheless, before considering the integration of such sensors in glulam, a study of the influence of their presence on the mechanical strength should be conducted as for the 4 proposed configurations. Moreover, the RH method can only provide an estimation of MC in the hygroscopic domain, which would not be sufficient enough to conduct durability study with accelerated wetting/drying cycles in the future.

Among the 4 configurations tested, configuration 4 presents the best potential to be applied to glulam structures for in situ monitoring for the reasons below:
  1. 1.

    The electrodes can be easily implanted in the glulam fabrication line, since no extra drilling/machining is needed.

     
  2. 2.

    This configuration is more versatile since it is possible to conduct both resistive and capacitive measurements, which can provide complementary information.

     
  3. 3.

    The cost is lower compared to other configurations (0.9 €, material purchase price, measurement system not included) concerning both the sensors and their implementation.

     
  4. 4.

    The measurements in this configuration were performed in the tangential direction. As a result, it might be interesting to use it to monitor the MC (or moisture gradients) perpendicular to the grain, which are potentially responsible for possible moisture-induced stresses as well as the durability loss (Fragiacomo et al. 2011; Fortino et al. 2013).

     

Nevertheless, it should be noted that the measurements with configuration 4 are affected by the outer lamellas, whereas the measurements with the configurations 1, 2 and 3 are more local, in view of the dimensions of electrodes and measuring area. Within the scope of continuous monitoring of structures, this may not be a disadvantageous of configuration 4 since it can be calibrated. Nevertheless, for absolute measurement of wood properties, the pin-type configuration (configurations 1, 2 and 3) may remain safer. Finally, the electronic setup of resistive measurement is simpler and more economical compared to that of the capacitive measurement (LCR meter), which gives an advantage to the resistive method for in situ implementation.

On the other hand, the influence of bonding pressure was also investigated in this study through results on glulam specimens. It was found that compared to solid wood specimens, the glulam specimens exhibit lower electrical resistance and higher capacitance. All this can be associated with the application of bonding pressure since the latter can improve the electrical contact at wood/electrode interface.

At last, the influence of electrode spacing on resistive/capacitive measurements was also investigated. The influence can only be observed on configuration 4, and it can be accounted for by the contribution of contact resistance/capacitance. Nevertheless, the electrode spacing does not affect the feasibility of MC measurements. Then, in so far as the thickness of 33 mm is normalized, the latter will be used in future work. Further studies should be made to facilitate the industrialization of the measurement configurations in order to conduct in situ measurement in glulam structures. Further studies should also be made by conducting MC monitoring with embedded sensors in order to retrieve more information on moisture gradient in glulam beams: electrodes with different penetrating depths, electrodes in different anatomical directions of wood, etc.

Notes

Acknowledgements

The authors would like to thank the co-financers of this research project: la Région Midi-Pyrénées, le Conseil Général des Hautes-Pyrénées and le Grand Tarbes. Special thanks should also be addressed to Emannuel Laught for his contribution in developing the Giga-ohmmeter and to Tommy Vilella and Frédéric Leroy for their help in experimentation.

References

  1. Bjorngrim N, Hagman O, Wang X (2016) Moisture content monitoring of a timber footbridge. BioResources 11:3904–3913Google Scholar
  2. Bjorngrim N, Fjellström P, Hagman O (2017) Resistance measurements to find high moisture content inclusions adapted for large timber bridge cross-sections. BioResources 12:3570–3582CrossRefGoogle Scholar
  3. Brischke C, Rapp AO, Bayerbach R, Morsing N, Fynholm P, Welzbacher CR (2008a) Monitoring the “material climate” of wood to predict the potential for decay: results from in situ measurements on buildings. Build Environ 43:1575–1582CrossRefGoogle Scholar
  4. Brischke C, Rapp AO, Bayerbach R (2008b) Measurement system for long-term recording of wood moisture content with internal conductively glued electrodes. Build Environ 43:1566–1574CrossRefGoogle Scholar
  5. D’Ayala D, Branco JM, Riggio M, et al (2014) Assessment, reinforcement and monitoring of timber structures: FPS cost action FP1101. In: 9th international conference on structural analysis of historical constructions, 14–17 October, Mexico City, Mexico, pp 1–12Google Scholar
  6. Dai G, Ahmet K (2001) Long-term monitoring of timber moisture content below the fiber saturation point using wood resistance sensors. For Prod J 51:52–58Google Scholar
  7. Dervos CT, Michaelides JM (1998) The effect of contact capacitance on current-voltage characteristics of stationary metal contacts. In: 43rd IEEE Holm conference on electrical contacts, 20–22 October, Philadelphie, Pennsylvania, pp 530–540Google Scholar
  8. Dietsch P, Franke S, Franke B, Gamper A, Winter S (2014a) Methods to determine wood moisture content and their applicability in monitoring concepts. J Civ Struct Health Monit 5:115–127CrossRefGoogle Scholar
  9. Dietsch P, Gamper A, Merk M, Winter S (2014b) Monitoring building climate and timber moisture gradient in large-span timber structures. J Civ Struct Health Monit 5:153–165CrossRefGoogle Scholar
  10. EN 13183-1 (2002) Moisture content of a piece of sawn timber—part 1: determination by oven dry methodGoogle Scholar
  11. EN 14080 (2013) Timber structures—glued laminated timber and glued solid timber—requirementsGoogle Scholar
  12. Forsén H, Tarvainen V (2000) Accuracy and functionality of hand held wood moisture content meters, vol 420. VTT Publications, EspooGoogle Scholar
  13. Fortino S, Genoese A, Genoese A, Nunes L, Palma P (2013) Numerical modelling of the hygro-thermal response of timber bridges during their service life: a monitoring case-study. Constr Build Mater 47:1225–1234CrossRefGoogle Scholar
  14. Fragiacomo M, Fortino S, Tononi D, Usardi I, Toratti T (2011) Moisture-induced stresses perpendicular to grain in cross-sections of timber members exposed to different climates. Eng Struct 33:3071–3078CrossRefGoogle Scholar
  15. Franke B, Franke S, Mueller A, Vogel M, Scharmacher F, Tannert T (2013) Long term monitoring of timber bridges—assessment and results. Adv Mater Res 778:749–756CrossRefGoogle Scholar
  16. Fredriksson M, Wadso L, Johansson P (2013) Small resistive wood moisture sensors: a method for moisture content determination in wood structures. Eur J Wood Prod 71:515–524CrossRefGoogle Scholar
  17. Gérard J, Guibal D, Paradis S, Vernay M, Beauchêne J et al (2011) Tropix 7. CIRAD.  https://doi.org/10.18167/74726f706978 Google Scholar
  18. Hjort S (1996) Full-scale method for testing moisture conditions in painted wood paneling. J Coat Technol 68:31–39Google Scholar
  19. Holm R (1967) Electric contacts—theory and application. Springer, BerlinGoogle Scholar
  20. James WL (1963) Electric moisture meters for wood. Department of Agriculture, Forest Service, Forest Products Laboratory, Madison, WisconsinGoogle Scholar
  21. James WL (1977) Dielectric behavior of Douglas-fir at various combinations of temperature, frequency, and moisture-content. For Prod J 27:44–48Google Scholar
  22. James W (1986) The interaction of electrode design and moisture gradients in dielectric measurements on wood. Wood Fiber Sci 18:264–275Google Scholar
  23. Jazayeri S, Ahmet K (2000) Detection of transverse moisture gradients in timber by measurements of capacitance using a multiple-electrode arrangement. For Prod J 50:27–32Google Scholar
  24. Kabir MF, Daud WM, Khalid K, Sidek HAA (1998) Dielectric and ultrasonic properties of rubber wood. Effect of moisture content grain direction and frequency. Holz Roh Werkst 56:223–227CrossRefGoogle Scholar
  25. Kasal B (2013) Assessment, reinforcement and monitoring of timber structures–COST FP1101. Adv Mater Res 778:1037–1040CrossRefGoogle Scholar
  26. Kasal B, Lear G (2011) Moisture measurement. In: Kasal B, Tannert T (eds) In situ assessment of structural timber. RILEM state of the art reports, vol 7. Springer, Dordrecht, pp 99–104. ISBN: 9789400705609Google Scholar
  27. Kumaran MK (1999) Moisture diffusivity of building materials from water absorption measurements. J Build Phys 22:349–355Google Scholar
  28. Lin RT (1967) Review of the electrical properties of wood and cellulose. For Prod J 17:54–60Google Scholar
  29. Melin CB, Bjurman J (2017) Moisture gradients in wood subjected to relative humidity and temperatures simulating indoor climate variations as found in museums and historic buildings. J Cult Herit 25:157–162CrossRefGoogle Scholar
  30. Moron C, Garcia-Fuentevilla L, Garcia A, Moron A (2016) Measurement of moisture in wood for application in the restoration of old buildings. Sensors 16:697–705CrossRefPubMedCentralGoogle Scholar
  31. Muth JED (2014) Basic statistics and pharmaceutical statistical applications, 3rd edn. CRC Press, Boca RatonGoogle Scholar
  32. Norberg P (2000) Monitoring wood moisture content using the WETCORR method part 2: calibration and validation. Holz Roh Werkst 58:129–134CrossRefGoogle Scholar
  33. Pousette A, Gustafsson A, Fjellström P (2014) Moisture monitoring of beam and pylon in a timber bridge. In: Proceedings of cost timber bridge conference CTBC2014, 25–26 September, Biel/Bienne, Switzerland, pp 125–131Google Scholar
  34. SETRA (2006) Les ponts en bois: comment assurer leur durabilité. (Timber bridges: how to ensure their durability). SETRA (Service d’études technique des routes et autoroutes), France (in French), ISBN: 9782110946553Google Scholar
  35. Siau JF (1984) Transport processes in wood. Springer, BerlinCrossRefGoogle Scholar
  36. Skaar C (1964) Some factors involved in the electrical determination of moisture gradients in wood. For Prod J 14:239–243Google Scholar
  37. Skaar C (1988) Wood-water relations. Springer, BerlinCrossRefGoogle Scholar
  38. Stamm AJ (1927) The electrical resistance of wood as a measure of its moisture content. Ind Eng Chem 19:1021–1025CrossRefGoogle Scholar
  39. Tamme V, Muiste P, Kask R, Padari A, Tamme H (2012) Experimental study of electrode effects of resistance type electrodes for monitoring wood drying process above fibre saturation point. For Stud 56:42–55Google Scholar
  40. Tereshchenko OV, Buesink FJK, Leferink FBJ (2011) An overview of the techniques for measuring the dielectric properties of materials. In: The 30th general assembly and scientific symposium, 13–20 August. Istanbul, Turkey, pp 14–18Google Scholar
  41. Terzic E, Terzic J, Nagarajah R, Alamgir M (2012) Capacitive sensing technology. In: A neural network approach to fluid quantity measurement in dynamic environments. Springer, London, pp 11–37. ISBN: 9781447140597Google Scholar
  42. Torgovnikov GI (1993) Dielectric properties of wood and wood-based materials. Springer, BerlinCrossRefGoogle Scholar
  43. Vermaas HF (1975) Experimental variables affecting the measurement of the DC resistance of wood. Holzforsch Int J Biol Chem Phys Technol Wood 29:140–144Google Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  • Hang Li
    • 1
  • Marianne Perrin
    • 1
  • Florent Eyma
    • 1
  • Xavier Jacob
    • 2
  • Vincent Gibiat
    • 2
  1. 1.Institut Clément Ader (ICA), CNRS, UMR 5312University of Toulouse, UPSTarbesFrance
  2. 2.PHASE Laboratory, EA 3028University of Toulouse, UPSToulouseFrance

Personalised recommendations