Einsatz von FBGS im Holzbau. Institut für Holzbau, Tragwerke und Architektur Institute for Human Centered Engineering Optolab

Transcription

1 Einsatz von FBGS im Holzbau Development and use of optimized Fibre Bragg Grating sensors for long term monitoring of timber components Forschungsbericht Autoren: Steffen Franke Noëlie Magnière Marcus Schiere Thomas Ferrazzini Christoph Meier Anke Bossen Institut für Holzbau, Tragwerke und Architektur Institute for Human Centered Engineering Optolab

2

3 Einsatz von FBGS im Holzbau Development and use of optimized Fibre Bragg Grating sensors for long term monitoring of timber components Final Report Report Nr. Contract Nr. Classification 2838-PB DHB Public Date Client Adress of the research unit Author Bern University of Applied Sciences Marc-André Gonin Solothurnstrasse Biel Bern University of Applied Sciences Architecture, Wood and Civil Engineering Institute for Timber Construction, Structure and Architecture Area of competence Timber Construction Solothurnstrasse 102, CH-2504 Biel Tel / Fax +41 (0) / Steffen Franke, Noëlie Magnière, Marcus Schiere, Thomas Ferrazzini, Christoph Meier, Anke Bossen Project leader Prof. Dr. Steffen Franke Head of Institute Prof. Andreas Müller Bern University of Applied Sciences Institute for Timber Construction, Structure and Architecture

4 ISBN by Bern University of Applied Sciences 2016 All rights reserved. No part of this publication may be reproduced in any form or by any means, electronical, mechanical, photocopying, recording, scanning or otherwise, without permission of the publisher. Published by Bern University of Applied Sciences Institute for Timber Construction, Structures and Architecture Solothurnstrasse Biel 6 Switzerland

5 Abstract The limits of engineers and wood are pushed nowadays through the increased use of and demand for timber as a raw material in civil structures. Measurements and monitoring play an important role in this process and help to improve the current state of the art of materials, building techniques, and regulations. In conventional monitoring methods, separate sensors and measuring equipment are used to measure climate, moisture content, temperature, and strain. Different systems have to be bought and known in handling Furthermore, a synchronised data recording is very difficult. Through fiber-optics, e.g. Fibre Bragg Grating sensing, these measuring values could all done with one unit and even combined in one sensor. An additional advantage of optical measurement techniques is that data can be transferred through one cable only by multiplexing. The application of fiber-optics in experiments and in monitoring in other sectors than the wood industry has been proven and is widely applied in aerospace, civil, health and energy industries nowadays. In fiber-optic measurements, a broad spectrum of light is emitted into a cable with a sensor at the end. The sensor reflects certain frequencies, which shift due to surrounding conditions. Each individual sensor reflects light at a specific frequency. This allows multiplexing and up to 12 sensors can be connected to one single cable. Correction to measurements needs to be done to compensate for thermal effects though. Measurement of moisture content, can directly be done by measuring electrical resistance between two metal pins, is not possible through fiber-optic measurements directly. Temperature and humidity are though and these can be measured in a small void in the wood. These parameters can be used to calculate the moisture content afterwards through a method that has been researched extensively and is applied in the wood industry already. A second difficulty is the measurement of high strain. Maximum allowable strain levels in the state of the art fiber-optic sensing techniques are up to 2 %. But up to 14 % can develop in timber perpendicular to the grain due to swelling and shrinkage. By the development of a special aluminium portal frame to which the strain sensor was attached, measurements could be made with a wide range of the multiple of the current range without changing the fibre sensor itself. The report presents solutions to overcome the difficulties in applications of the fiber-optic technique in the wood industry. It is shown, that it is possible to measure temperature, moisture content, strain and even climate by using one device. Further steps that should be taken are to be focused on improvement, e.g. down scaling of the sensors. An industrial application should be pursued as well. Keywords: Fibre Bragg Grating, Timber, Moisture content, Strain, Temperature, Monitoring Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 3 / 70

6 Table of Content 1 Introduction Objectives Frame and organisation 8 2 State of the art The monitoring of timber Timber, wood moisture content and related strains The Fibre Bragg Grating sensors FBG sensors and timber monitoring 16 3 Specifications of the sensing units Overview of climates surrounding Swiss timber structures Specifications regarding temperature Specifications regarding moisture content Moisture content related strains for different wood species Other requirements and synthesis of the requirements and specification of the sensors 22 4 Strain sensing unit Development of the strain sensing unit Testing of the strain sensing unit, prototype Development of the calculation sheet to analyse of the portal behaviour Synthesis regarding the strain sensing unit 33 5 Moisture content of wood sensing unit Measuring wood moisture content Air relative humidity sensor Temperature sensor Synthesis regarding the moisture content sensing unit 40 6 Combination of the three sensors Choice of the layout and recommendations Description of the final layout 41 7 Conclusions and outlook Conclusion regarding the project Outlooks 42 8 Regulations of the present report 43 9 Indexes Abbreviations Index of Tables Index of Figures Bibliography 46 4 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

7 Appendixes 51 Appendix A Wood hygroexpansion coefficients 53 Appendix B Development of the strain sensor 55 B.1 Approach 2 Development of a calculation sheet 55 B.2 Investigations over the stiffness of the strain sensor 61 Appendix C Specifications of the FBG relative humidity and temperature sensors 66 C.1 The relative humidity sensor (company: O/E-Land Inc.) 66 C.2 Specifications of the temperature sensor (company: FiberSensing) 67 Appendix D Test C Testing of the temperature FBG sensor controlled climate 69 D.1 Test preparation 69 D.2 Test analysis and results 69 D.3 Conclusion of the test 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 5 / 70

8 1 Introduction Through engineering and innovation, wood is nowadays used as the main load-bearing material in audacious structures. The true potential of this material is revealed in its central position in project which largely exceeds the frame of individual two-story houses to which wood is sometimes thought to be restricted. Timber elements are nowadays the core of multi-story buildings across the world such as the H8 building in Bad Aibling, Germany (Figure 1a). Similarly, three dimensional bar structures or plate structures such as the ones of the Elephant house in Zürich (Switzerland) or of the Pyramidenkogel in Austria (see Figure 1b and Figure 1d) are currently made out of wood. Only the use of the most recent advances in computer technologies combined with exceptional production quality enabled the achievement of such structures. Moreover, the use of the latest innovations from the wood industry such as the cross laminated timber or the combination of hardwoods and softwoods (see Figure 1c) lead to production structures always more efficient in their use of the material itself. This constant innovation of the timber industry enables the development of building techniques which leads to structures always more slender and ambitious. However, all these structures push timber engineers and designers to the very edge of the building standards. The achievement of such structures nowadays, even though possible, still requires a strong personal commitment of the planners. They indeed usually require exceptional regulations which are accompanied by large safety margins measures including strict controlling procedure. (a) (b) (c) Figure 1: (a) the H8 tower in Bad Aibling [1], (b) the Elephant house in Zürich [2], (c) close view of hybrid timber beams from the ski school in Arosa [3] and (d) the Pyramiden Kogel tower in Austria [4] (d) 6 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

9 In other sectors dealing with cutting edge materials, the field of monitoring has been developed. The monitoring is the systematic and continuous recording and control of a structure through a long term measurement system. It has been used to insure in-situ the well behaviour of structures as diverse as nuclear plants, aeronautics or space industry. Nowadays, with the advances of the technology, the Fibre Bragg Grating sensors technology is also available for the use of a larger number of industries. This technology has been used in concrete or steel and is, since a few years, a major topic of discussion for the wood industry. However, due to its relatively recent applications to the timber industry, Fibre Bragg Grating sensors available on the market are not dedicated to the use on wood but rather on steel or on concrete. This non optimization of the Fibre Bragg Grating technology to the timber material is an issue when considering the specificities of the wood material, especially regarding its behaviour as an answer to moisture content changes. The project presented here reports the feasibility study conducted to develop a Fibre Bragg Grating sensor optimized for the study of wood under moisture content changes. Especially, this sensor is designed to be used in two fields: - The investigation of the material behaviour of wood under moisture changes - The monitoring of timber structures 1.1 Objectives The concept of the hybrid sensor developed in this project was using the current Fibre Bragg Grating technology. This development was conducted aiming at creating a sensor optimized for monitoring the moisture content of timber and its related effects. Thus, the project was carried out bearing in mind the following objectives: - The final sensor should be able to monitor three quantities: moisture content, temperature and moisture content-related strains in wood - The sensor should be able to sustain the large moisture content related strains (more than 5%) that wood can induce In order to develop this sensor optimized for the monitoring of timber structures, the following steps have been followed. First a study of the state of the art of fundamental topics related to this project was conducted. This research dealt with, on the one hand the monitoring of timber structure, the moisture content of wood and related quantities, and, on the other hand with the Fibre Bragg grating technology and its application in the field of timber construction. This state of the art, presented in chapter 2, provides a comprehensive overview of the current state of both research and practice regarding the topics detailed above. It also identified the key elements to account for in the progress of the project. Then, based on the study of existing tests and reports regarding the monitoring of timber structures, detailed specifications were obtained. These specifications quantify the performances the Fibre Bragg Grating sensor to be developed should have in order to be qualified as optimized for the monitoring of timber structures. These specifications are detailed in the chapter 3 of this report. The definition of these specifications was a milestone in this project as they were used as a basis in the development of the strain (chapter 4), moisture content and temperature (chapter 5) sensing units. Finally, the investigation of how to combine the three sensing units obtained into a single complex Fibre Bragg Grating sensor optimized for the monitoring of wood was carried out. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 7 / 70

10 1.2 Frame and organisation Project team The project Development and use of optimized Fibre Bragg Grating sensors for long term monitoring of timber components was conducted in cooperation between both the department Architecture, Wood and Civil Engineering and the department Engineering and Information Technology Optolab from Bern University of Applied Sciences. All the persons involved in the achievement of this project are listed in Table 1. Table 1: The project team Name Steffen Franke, Noëlie Magnière, Marcus Schiere Christoph Meier, Anke Bossen, Thomas Ferrazzini Organisation Bern University of Applied Sciences; Institute for Timber Construction, Structures and Architecture Bern University of Applied Sciences; Engineering and Information Technology Optolab Project boundaries and terminology This project was aiming as investigating the potential use and proving the feasibility of the development of a Fibre Bragg Grating sensor optimized for the monitoring of timber. Therefore, the results of this project are a combination of prototypes and concepts. However, further development would be necessary to obtain a marketable product. Finally, a particular terminology was used in this report to qualify the different stages of Fibre Bragg Grating products considered in this investigation. Thus, the expressions sensor or Fibre Bragg grating sensor are used to designate a finished product. It this report it was used either for a Fibre Bragg Grating product already available on the market or for the sensor to be developed. However, the expressions Fibre Bragg Grating sensing unit, or in short, sensing unit or unit was used to designated a system dedicated to the measure of a particular quantity: either moisture content of wood, temperature or strains. A sensing unit can be composed of one or several sensors, either directly use as provided by the producer or adapted for a particular need. In summary, the aim of this project was to develop a sensor as a whole. However, this sensor is constituted of three sensing units: a moisture content unit, a temperature unit and a strain unit. 8 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

11 2 State of the art 2.1 The monitoring of timber Context and history In the early years 2000s and, especially, during the winter , numerous failures of timber structures occurred in Scandinavian countries, Germany, Austria or Poland. These failures, often illustrated by the striking example of the collapse of the Bad Reichenhall ice arena in 2006 (see Figure 2), could not solely be explained by exceptional climate conditions. They triggered a global awareness of public authorities, of the timber construction community but also, on a broader level, of the general public regarding safety of timber but also steel and concrete structures. Figure 2: The Bad Reichenhall ice arena before collapse (left) and partial view of the collapsed roof structure (right) [5] Consequently, several investigations were appointed by the authorities in order to assess, analyse and draw the necessary consequences of these events. The objective was to prevent, by all means possible, a repetition of this tragedy. The scientific community of timber construction was, obviously, highly involved in these investigations. Their actions were coordinated at a European level by the Action COST E55 Modelling the Performance of Timber Structures (Dec Dec. 2011). The aims of this European action were, amongst others, to improve design methods, assessment techniques and maintenance policies for timber structures as well as providing the engineering community with a modern probability based foundation for the efficient performance-based life-cycle design and assessment of timber structure [6]. Major results of this action were several research reports such as the ones of Frühwald et al., Blass and Frese or Dietsch ([7] [9]). All these reports provided statistics of assessment of timber structures, identified the key elements which lead to the failure and draw recommendations for the design of future structures. These recommendations were summarized by Winter and Kreuzinger in [5]. They urged for: - A more reliable quality control of the element produced - A better control of the design of the structures, involving its checking by several persons - A regular inspection and maintenance of the building to account for its natural ageing - The ban of risky structural details which did not allow their monitoring after erection of the building This COST Action, which ended in 2011, pointed the way to timber structure assessment and monitoring. In was followed in 2011 by the launch of the Action COST FP1101 Assessment, Reinforcement and Monitoring of Timber Structures which is still running nowadays. This action aims at disseminating knowledge regarding the assessment, reinforcement and monitoring of timber structures, improving maintenance methods for existing timber structures and disseminating up-to-date results to the industry, code writers, policy maker and society [10]. These two COST actions, through their research results, largely contributed in building the current field of monitoring of timber structures. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 9 / 70

12 2.1.2 Why is it interesting to monitor wood and timber structures? The monitoring of a timber structure can, and has, been used for several functions. Five main functions were identified and are summarized in Table 2. Table 2: Summary of the functions for which monitoring system have been and could be used for timber structures The monitoring of the structure can participate in Example of case Sources Reducing the level of risk associated with the design Ensuring the quality of the construction Increasing the safety of use of the building Reducing the cost of the construction Optimizing the use of timber Verify the adequacy between complex computer models and the structure Calibrate computer models on existing structures Validate a service class used in the design Enable the use of innovative wood products which behaviour (especially on the long term) is poorly known Check unplanned constructive or structural issues (detection and monitoring of evolution of cracks, deformations and displacements) Assisting tools in sensitive constructions with high requirements Control the behaviour of ageing structure (detect abnormal behaviour at an early stage) Provide information on the health of metallic connectors and waterproofing details Trigger alarm in case of decay detection thus evaluating the long term performance of rehabilitations or refurbishments (especially for historic buildings) Detect early signs of decay reducing the amount of maintenance required and the number of in-situ inspections Prove the lasting performances of timber structures in general which may lead to more favourable life expectancy regulations By collecting valuable information on the wood material and its behaviour in real condition of use [11], [12] and [13] [14] [15] [16], [17] [16] [18], [19] Thus, the monitoring of timber structures offers the possibility to control and record virtually an infinite number of quantities in a large number of locations. Consequently, the quantities monitored are highly dependent of the aim of the monitoring itself. Usually, when the aim of the monitoring is the evaluation of the adequacy with a computer model, the quantities monitored are strain and displacement of principal elements (see [13], [20]). In the case of timber bridges, such as the one of Älvsbacka [11], the dynamic behaviour of the bridge is also tested using monitoring systems. In his presentation, Jorge [21] summarized the key facts which should be considered for monitoring in a timber structure. They are: - The bracing of load bearing members and structures - The consideration of performance of connections and how it is affected by variable humidity conditions - The swelling and shrinking of timber members - The cracks caused by moisture-related shrinkage - The orthotropic strength of wood 10 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

13 2.1.3 Why monitoring moisture content and related impacts? Moisture content (MC) is the main quantity to look for when monitoring or assessing timber structures. Already, four of the five key facts synthetized by Jorge [21] (see the previous paragraph) are directly related, either to the level, or to the variations of MC. Similarly, the various impacts of MC levels and variations are identified in the presentation of Cavalli [22]. They are shown in Table 3. Accordingly, MC content is usually a necessary information without which the analysis of a timber structure is almost impossible. In addition to MC itself, temperature and MC-related strains are also quantities to be considerd in order to monitor the behaviour of a timber structure. Their possible impacts on the behaviour of a timber structure and its analysis are summarized in Table 3. Especially, cracking of timber elements as a result of MC-related strains was shown by Blass and Frese, in [9], to be one of the major factors leading to recent failures in timber structures. In conclusion, MC, temperature and MC-related strains are three quantities which, monitored together provide a reliable picture of the structure s state and can efficiently identify risks which would require maintenance actions. Table 3: Summary of the impacts that moisture content (MC), MC-related strains and temperature can have on a timber structure Quantity Impact on the timber structure Moisture content (source [22]) Moisture content related strains Temperature Affect physical and mechanical properties of wood Can contributes to decay burst and propagation (for a certain range of moisture content) Affects wood shape and dimension Affect the viscoelastic behaviour of wood Affect most non-destructive techniques used in assessment of timber structures (therefore requiring moisture content corrections) Can lead to additional stresses in the section and eventually cracking Can harm the proper behaviour of connection elements Can favour decay (for a certain range of temperatures) Can harm the proper behaviour of some engineered wood products (such as shear properties of glue lines) Affects the measure of most assessment and monitoring devices (therefore requiring temperature corrections) 2.2 Timber, wood moisture content and related strains Wood is a material synthetized by nature and then processed by men to be used for construction, amongst other uses. It is to be distinguished from other construction materials such as steel or concrete which are produced directly by men. Indeed wood is a much more complex material. Its structure can be analysed at several different levels: from the complete tree to the chemical composition of its cells. Its fibrous composition makes it strongly anisotropic. Moreover, void spaces are naturally present in the wood structure. Wood interacts with the water present in the surrounding air by filling or emptying these voids. The amount of water present in the wood structure is represented by w, the MC of the wood. This quantity has a major influence on most properties of wood including dimensions and mechanical properties The moisture content of wood Wood is always tending to equilibrium with the water present in its surrounding environment. This equilibrium is obtained by filling (adsorption) or emptying (desorption) the voids present in its cells walls and lumens with water through complex processes which include diffusion and pressure driven flows. The amount of water in the wood is then represented by its MC, w [%], defined in the standard EN :2002 [23] through the following formula: Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 11 / 70

14 With 100 (1) % percentage of MC of the wood mass of the wet wood which MC is to determine oven-dried mass of the same wood sample In the case of timber elements surrounded by air, the following rule, formulated by Simpson [24] applies: The moisture content of wood depends on the relative humidity and temperature of the air surrounding it. If the wood remains long enough in air where the relative humidity and temperature remain constant, the moisture content will also become constant at a value known as the equilibrium moisture content (EMC). This EMC is provided for one particular air temperature and Relative Humidity (RH) by isotherms available in the literature. These isotherms were obtained experimentally for one particular wood species (such as for Sitka spruce in Figure 3). However, they are usually applicable for wood regardless of the species as shown in Figure 4. Figure 3: EMC as a function of the air temperature and relative humidity for Sitka Spruce (right) based on R. Keylwerth (1949, see [19]) Figure 4: EMC as a function of the air temperature and relative humidity for wood in general (adapted from [25]). In red, the standard testing conditions for wood: 20 C and 65% relative humidity resulting in an EMC of 12% in the wood 12 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

15 In addition, several mathematical models were built to describe these isotherms. One of the most famous and widely accepted model was established by Simpson [24]. This model provides wood s EMC through formula:, (2) With And , % % The equilibrium MC of the wood The temperature of the air The RH of the air The moisture related strains In the voids of the wood cells, water can exists under two forms: - Bound water: water present in the cells wall - Free water : the water present in the cells lumen Under increasing MC, the water first fills the cells walls and then the cells lumens, providing the evolution summarized in Figure 5 on a microscopic scale. As shown in these two figures, a critical point appears when all voids in the cells walls are filled but no free water (lumen water) is present. This point is called the Fibre Saturation Point (FSP). This FSP varies for species to species but is usually approximated to 30% of MC. Below the FSB, the amount of water in the wood has a major influence on its behaviour, especially on its dimensions and mechanical properties. Above the FSP however, the amount of water does not have a lot of influence. As a general rule, the MC of a timber element which is not directly in contact with liquid water will be below the FSP. Figure 5: Evolution of water content in wood, w = moisture content (adapted from [26]) Moisture related strains are directly linked to the MC history of the wood. Indeed, below the FSP, the wooden element will swell when increasing its MC (see Figure 6, left) and shrink when decreasing it, thus changing dimensions. The amount of swelling and shrinkage are defined with respect to the dimensions of the oven dry and green (saturated with water) element respectively. They can be calculated either for one of the three principal directions of wood: longitudinal (L), radial (R) or tangential (T) or for a volume of wood using the following formulas:. 100 or. 100 (3) Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 13 / 70

16 The swelling/shrinking behaviour depends strongly on the wood species considered (see Figure 6, right). Amongst common European wood species, beech is considered to have the highest MC related dimensional changes. Between oven dry and green states; a linear behaviour, identical in adsorption and desorption, is often assumed even through it is not entirely true. Therefore, the strains generated by changes in MC can be calculated using a single coefficient α called the hygroexpansion coefficient: (4) With [-] The strain in L, R or T direction or for the complete wood volume [%/% of MC] The hygroexpansion coefficient in L, R or T direction or for the complete volume [%] The change of MC Hygroexpansion coefficients have been defined experimentally for each wood species. An example of average and extreme values of hygroexpansion coefficients for common wood species is provided in Table 4. αr R L T α T α L Figure 6: Left, swelling of a timber element (source: [27]) and right, volume changes as a function of the moisture content w (adapted from [26]) Table 4: Average and extreme values of hygroexpansion coefficients for various wood species (adapted from [26]) Wood species Spruce Fir Pine Larch Oak Beech Ash Chestnut Black locust Oven dry density ρ 0 [kg/m 3 ] α T [%/%] (tangential) α R [%/%] (radial) α L [%/%] (longitudinal) Consequences of moisture related strains in timber elements MC changes are accompanied by the apparition of internal stresses in the cross section of timber elements. Indeed, MC changes create gradients of MC between the outer part of the element s cross section and its inner part. From these gradients, differential swelling/shrinkage occurs within the cross section resulting in tension and compression stresses. These stresses can exceed the low strength in tension perpendicular to the grain of timber leading to the formation of cracks in the element. This process of crack formation in timber due to moisture changes is illustrated in Figure 7 for a decrease of MC and is similar in the case of an increase of MC. 14 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

17 Figure 7: Apparition of internal stresses in a glued laminated beam subjected to drying and corresponding location of crack formation (source: [27]) 2.3 The Fibre Bragg Grating sensors Principle of the Fibre Bragg Grating (FBG) technology A Fibre Bragg Grating (FBG) is a type of distributed Bragg reflector constructed on a short segment of optical fibre in order to reflect particular wavelengths of light and transmit all others. The distributed Bragg reflector is a multi-layered structure of material in which each layer has alternating refractive indices. Inscription with an Excimer UV-Laser is used to obtain this short segment of optical fibre which shows alternating refractive indices, or in other words: a Fiber Bragg Grating (see Figure 8). If a broadband light source is coupled into an optical fibre containing an FBG, the specific wavelength which meets the Bragg condition is reflected at each grating plane while the rest of the light beam propagates through the grating without obvious optical loss. This Bragg condition can be expressed as such: 2 (5) With [-] The Bragg wavelength or the central wavelength of the Bragg grating The effective refractive index of the fibre core The grating period The period of the refractive structure is a function of its strain and thermal expansion. Therefore, the FBG can intrinsically be used as a sensing element to measure either strain or temperature. To obtain these measures, the reflected light spectrum is analysed using a spectrometer. Figure 8: Structure and principle of a FBG (source: [28]) Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 15 / 70

18 2.3.2 Advantages of FBG sensors FBG based sensors are particularly suitable for long-term and permanent health monitoring of structures. They are already widely used in extremely challenging fields such as the aerospace, the energy or the chemical industry. Moreover, FBG systems are also used in the concrete or steel construction industry. There, their main advantages are the following. - FBG sensors do not need calibration and are easy to install - They are small and lightweight, they can also be embedded in materials - They easily support multiplexing (a single fibre can carry more than 50 sensors) - They have low transmission loss enabling remote sensing and long line transmission (up to several kilometres) - They are durable and robust - FBG sensors are immune to aggressive or extreme environments (extreme temperatures, corrosive environments, electric or electromagnetic fields, nuclear environment etc.) and therefore ensure a high reliability level even in these environments - They are wavelength encoded which means they do not require electricity, their reliability is independent of power fluctuations, they are not interfering with their environment and can safely be used in sensitive environments (such as inflammable environments) - FBG sensors are mass producible at a reasonable cost Nevertheless, the FBG technology has also some disadvantages, amongst others: - They are sensitive to temperature variations, thus most FBG sensors measures have to be corrected with regards to temperature - They are sensitive not only to longitudinal strains but also to transverse ones - The FBG technology is still quite recent therefore few standards are currently available with regards to their use and applications - Compared to more standard technologies, only few suppliers provide FBG sensors 2.4 FBG sensors and timber monitoring Current situation of the FBG technology with regards to the field of timber structures FBG sensors are nowadays commonly used to monitor the behaviour of all types of steel and concrete structures. However, they are barely known in the field of timber structure monitoring. As a striking example, the website of the Swiss company Smartec [29], specialised in structural health monitoring through optical systems, provides case studies of the structures which were monitored with their products. Within these case studies (almost 100), no timber structures are to be found. This fact clearly shows the delay of the timber construction with regards to the use of FBG technology in comparison to its traditional competitors: steel and concrete. According to the literature, it appears that the FBG technology is still at the early research stage for the timber industry. Few publications describe the use of FBG sensors on timber. For instance, a first trial of monitoring timber through FBG sensors was attempted in [18] within the frame of laboratory measurements. The quantities measured through FBG sensors were MC-related strains and temperature for a glued laminated timber block. In his paper, Gustafsson et al. [12], name optical sensors such as FBG sensors as an option for the monitoring of a complex timber bridge in Sweden. Similarly, Morris et al. [30], also mentioned FBG sensors as a possibility considered for the monitoring of the new timber building of the Nelson Marlborough Institute of Technology (New Zealand). However, it appears that the cost of the FBG system was too high for their particular case. The most advanced research projects involving FBG sensors in the monitoring of timber structures are certainly the project presented by Lanata [20] where the deformations of a massive cantilever structure in timber are monitored using FBG sensors. In addition, the five years research project Development of a smart timber bridge, started in 2011 in the USA [31] [35], provides through various thesis and scientific publications both laboratory and field-oriented recommendations with the aim of developing a comprehensive implementation of a monitoring system for timber bridges based on the FBG technology. 16 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

19 2.4.2 Potential use of FBG sensors for timber monitoring Even though FBG sensors are nowadays not the rule in the field of monitoring of timber structures, their specificities could be in their favour in several cases. As an example, two cases where FBG sensors could have been duly used for monitoring timber structures are presented in the following paragraphs. The first advantage of FBG sensors is their immunity to aggressive environment. This quality could be an advantage for monitoring timber structures which are subjected to corrosive climates. Indeed, in the report of Gamper et al. [15], classic electrical monitoring systems showed issues in monitoring the climate and moisture content of timber elements in swimming pools due to the high levels of chlorine in the air. This principle could be further extended to timber structures in contact with salted humid air or immerged in water where electrical monitoring could be an issue due to the tendency of the system to corrode. Another, and probably the main advantage of FBG sensors is their possibility of multiplexing combined with the versatility of its technology. These qualities mean that only one single monitoring system can be used to monitor several quantities at several locations of a structure. In Table 5, two examples of systems used in the monitoring of timber bridges are provided based on the information available in [11], [36], [37]. Both systems monitor four different phenomenon, which correspond to Table 5: Two example of monitoring systems for timber bridges case 1 (left) based on [11] and case 2 (right) based on [36], [37] Monitoring case 1 Monitoring case 2 Phenomenon Sensor/Equipment No. Phenomenon Sensor/Equipment No. Displacements GNSS 3 Climate changes Temperature and relative humidity sensors 3 Cable tensions Force transducers 5 Strain Adapted potentiometric 2 displacement sensor Climate changes Weather station 1 Tension of the Load cell 1 strap Vibrations Wireless accelerometers 18 Displacement Video surveillance 1 Data acquisition for the cable tensions Data logger 1 Data acquisition Computer 1 General data acquisition Computer 2 Data transfer Network 1 four different data types and possibly four different recording clocks which have to be synchronized. Moreover, all these systems require the permanent use of at least one computer in-situ in order to record the data. For most of the quantities monitored in these two cases, suitable FBG sensors are already available on the market. Thus, the same monitoring plans could have been carried out in both cases resorting only to the FBG technology. This would have led to: - A single data type (shifts in central wavelengths) - A single measurement unit with a unique clock to record the different quantities (no synchronisation necessary) - No need of computers on site, as spectrometers which store data are available on the market Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 17 / 70

20 3 Specifications of the sensing units In order to develop a FBG sensor optimized to monitor timber structures, it is necessary to determine which specifications this sensor should achieve. These specifications were defined for each sensing unit (temperature, MC and MC-related strains) based on both literature and experimental information. According to the literature, temperature, MC and MC-related strains in a timber element are mainly influence by the wood species and the climate (air temperature and air RH) in which the element in situated. The dependencies between these three parameters and the three quantities the FBG sensing units should measure are summarized in Table 6. Regarding the quantity MC-related strains, it is obviously influenced by the MC of the element itself. Therefore, as the MC of wood is influenced both by the temperature and the RH of the air, the MCrelated strains are also influenced by these two parameters. Accordingly, the following paragraphs summarize the information collected to obtain specifications for the temperature, MC and strain sensing units of the FBG sensor to be developed. First, an overview of the climate in which Swiss timber structures are is provided. This analyse lead to average and extreme values of both temperature and MC which can be found in timber structures. Then, the variations of MC obtained are used to provide possible values of strains, especially for beech and spruce wood. Finally, all the specifications of the three sensing units are summarized in paragraph Overview of climates surrounding Swiss timber structures To characterize the climates in which Swiss timber structures are placed, timber structures are classified according to their category of use. This classification enables to have relatively homogeneous climate characteristics within each category of use. The list of the category of use considered in this project, along with the sources from which the information was obtained for this category, is provided in Table 7. Based on the experimental data from [18], [19], [38], the climate in each category of use was characterised with regards to their average values and maximum variations over year cycles. The qualitative repartition of each category of use is provided in Figure 9 for the air RH (left) and for the air temperature (right). These graphs provide an illustration of how representative the climates considered here are for the definition of the specifications of the FBG sensor. In these graphs, the climate surrounding five timber bridges is considered. These climates are assumed to be representative of the range of outdoor climates which could be found in Switzerland. As shown in the Figure 10 and Figure 11, out of the four bridges, one is located in a dry and warm area of Switzerland, one is situated, on the opposite, on a wet and cold area and three other bridges are situated in areas of intermediate climate (mean humidity and mild temperatures). Therefore, the climatic information obtained from both the monitoring of indoor [19] and outdoor [18], [38] timber structures is assumed to be representative of all climates in which timber structures could be placed in Switzerland. Thus the values of MC of wood and temperature obtain from these monitoring are estimated to be relevant for the specifications of the sensors. Table 6: Summary of the main parameters influencing each sensing unit Mainly influenced by Sensing unit Wood species Air temperature Air relative humidity Temperature X Moisture Content X X MC-related strains X X X 18 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

21 Table 7: Summary of the category of use of timber buildings considered to define the FBG sensor specifications Category of use Source Number of buildings Duration of the monitored monitoring Ice rink Opened Partially closed [19] 4 0 (literature) 0 (literature) Closed with AC* Closed without AC* year 1 year Swimming pool [19] 3 1 year Riding hall [19] 3 1 year Sports hall [19] 3 1 year Production hall [19] 2 1 year Agricultural building [19] 3 1 year Warehouse [19] 3 1 year Timber bridge in Switzerland [18], [38] 5 1 or 2 years Total 26 *AC = air conditioning Small Variations Large Ice rink open Ice rink part. open Ice rink close no AC Ice rink close AC Prod. Hall 1 Prod. Hall 2 Swimming pool Sports Hall Riding Hall Agric. Build. Warehouse Bridge Relative humidity Small Variations Large Temperature Ice rink open Swimming pool Ice rink part. open Sports Hall Ice rink close no AC Riding Hall Ice rink close AC Agric. Build. Prod. Hall 2 Warehouse Prod. Hall 1 Bridge Low Mean value High Low Mean value High Figure 9: For different building categories, qualitative repartition of the category depending on average values and variations of relative humidity of the air (left) and of the temperature of the air (right). Analyzed in [38] Figure 10: Average annual precipitation in Switzerland [39] and location of the four timber bridges monitored in [38] Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 19 / 70

22 Figure 11: Average annual temperature in Switzerland [40] and location of the four timber bridges monitored in [38] 3.2 Specifications regarding temperature The temperature of the air surrounding various timber structures was monitored in [19] and [18], [38] for at least a year in each case. The average temperatures, minimal and maximal temperatures as well as the corresponding maximal temperature variations obtained from these monitoring are provided in Table 8 and in Table 9 for different indoor categories of use and Swiss timber bridges. According to it, the temperature unit should be able to operate for temperatures between -11 C and +49 C. Moreover, it should have a maximal accuracy within the range 8 C to 17 C where the average temperature values for the climate of Swiss bridges and other categories of use of timber structures are found. Finally, the sensing unit should be able to support a variation of temperature of 43 C. Table 8: Average temperature and maximal temperature variations for timber structures corresponding to different category of use (adapted from [38]) Measured wood MC [%] Category of use Average value Min. value Max. value Max. variation Swimming pool Ice rink Riding hall Sports hall Production hall Agricultural building Warehouse Summary Table 9: Average temperature and maximal temperature variations for Swiss timber bridges (adapted from [38]) Temperature [ C ] Category of use Structure Average value Min. value Max. value Max. variation Swiss timber bridge Summary / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

23 3.3 Specifications regarding moisture content The MC of timber elements are following the variations of the climate in which the elements are placed. Therefore, it varies significantly during their life cycle (see for example Figure 12). Within the frame of this project, the MC sensing unit is aiming at monitoring the MC of timber elements after their processing. Therefore, it is assumed that relevant MC levels have to be considered only from the production value of 12 % and during the following use and service life where MC levels depend on the structure build. First, experimental values of MC of timber elements in Swiss timber bridges are provided based on [18]. Then, the same type of values are obtained from other categories of use from [19]. Finally the frame of the current European and Swiss standards is provided Swiss timber bridges The MC of elements from four timber bridges was monitored for various depth and locations in each bridge. The average, minimal, maximal and maximum amplitude of variation of MC for each point of measure was calculated. The values obtained for each timber bridge are summarized in Table Moisture content in other timber structures based on [19] In [19], the MC of timber elements from various timber structures was monitored for a depth in the cross section of 15, 25, 40 and 70 mm. Based on these monitoring values, the average, minimum, maximum and the maximum amplitude of variation of MC which can be found in closed timber structures are obtained. They are summarized in Table 11. Figure 12: Sketch of a possible moisture chain i.e. exposure to moisture from the tree to glued-laminated timber elements in the buildings (source [41]) Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 21 / 70

24 Table 10: Moisture content levels and variations measured during the monitoring of four timber bridges (according to [18], [38]) Measured wood MC [%] Bridge Average value Min. value Max. value Max. variation Horen Obermatt Schachenhaus Muothatal Summary Table 11: Moisture content levels and variations measured during the monitoring of timber structures from various categories of use (according to [38]) Measured wood MC [%] Category of use Average value Min. value Max. value Max. variation Swimming pool Ice rink Riding hall Sports hall Production hall Agricultural building Warehouse Summary Moisture content related strains for different wood species According to the previous paragraph, the maximum variation of MC which can be found in timber structures is of 16 %. Moreover, as mentioned in chapter 2.2.1, MC-related deformations are only observed for wood under the Fibre Saturation Point (FSP). This FSP is commonly taken as equivalent to a MC of 30 %. Accordingly, an absolute maximum in order to calculate MC-related strains would be a MC variation of 30 %: from oven dry wood to the FSP. In order to obtain the values of the hygroexpansion coefficients for softwoods species used in timber construction as well as for beech wood, a literature study was conducted. Its results are provided in Appendix A. The extreme values of hygroexpansion coefficients obtained through this literature study are provided in Table 12. In addition, the corresponding strains for variations of 16 % and 30 % of MC are also provided. The strains obtained with a variation of MC of 16 % correspond to the targeted working range of the strain sensing unit. However, in case of direct water leakage on the timber element, it is possible that the MC of the wood reaches the FSP. Therefore, the strain sensing unit should be able to sustain without damages strains corresponding to a variation of MC of 30 %. Table 12: Moisture content related strains for softwoods and beech wood for a variation of MC of 16 % and 30 % Hygroexpansion coefficient Strain ε [%] Strain ε [%] Maximum MC related strains α [%/%of MC] for MC = 16 % for MC = 30 % Radial, softwoods 0.19 (source [45, p. 6]) Radial, beech 0.23 (source [46]) Tangential, softwoods 0.35 (source [45, p. 6]) Tangential, beech 0.44 (source [46]) Hygroexpansion coefficient Strain ε [%] Strain ε [%] Minimum MC related strains α [%/%of MC] for MC = 16 % for MC = 30 % Longitudinal, all wood species Other requirements and synthesis of the requirements and specification of the sensors The four previous paragraphs provided working and operating ranges for the three sensing units of the FBG sensor. In addition to these ranges which provide the minimum and maximum values the 22 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

25 sensing unit should be able to measure, the accuracy of these units was determined. A summary of all these specifications are provided in Table 13. An accuracy of 0.1 C, 0.1 % and 0.01 % have been chosen for the temperature unit, MC unit and strain unit respectively. Indeed, an accuracy of 0.1 C corresponds to the values found for standard temperature sensors used for monitoring structures. 0.1 % accuracy for the MC sensing unit corresponds as well to the values found for standard sensors used for monitoring wood moisture. In addition, due to the large variability of the material wood, a target accuracy lower than 0.1 % for a sensing unit directed for the monitoring of timber s moisture content would be unrealistic. Finally, regarding the MCrelated strains, an accuracy of 0.01 % corresponds to the minimum strains obtained in radial or tangential directions for a change of MC of 0.1 %. Table 13: Summary of the specifications for the different sensing units of the FBG sensor Specification Temperature Moisture content Strain Working range -10 to 50 [ C] 3.4 to 25.5 [%] 7 [%] Operating range -10 to 50 [ C] 0 to 30 [%] 14 [%] Accuracy 0.1 [ C] 0.1 [%] 0.01 [%] To complete these specifications, the three FBG sensing units must be compatible. Therefore, the central wavelength and corresponding maximum wavelength shift of each sensing unit must be chosen in order to avoid possible overlapping of the reflected light from the different sensing units. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 23 / 70

26 4 Strain sensing unit Strain measurement is, along with temperature, the initial application of FBG sensors. Indeed, a mechanical strain applied to the fibre induces an elongation of the grating period which can directly be measured by a shift in the reflected wavelength. In this project, the strain sensing unit of the FBG sensor aims at measuring moisture content related strains in wood of up to 7 %. Moreover, this unit should also be able to support without damages strains up to 14 %. However, according to Zhou et al. [47], the maximum strain supported by current FBG strain sensors are between 1 % and 2 %. 4.1 Development of the strain sensing unit Transmission ratio of the sensing unit In order to be able to reach strain of 7 % and 14 %, the strain sensing unit was designed as a portal shown in Figure 13. In this portal, the bare FBG glass fibre is glued to the bottom of the top beam. This top beam is moving freely while the base of the portal is fixed to the timber element. Thus, when the timber element deforms, it also deforms the top beam of the portal. When deforming, the top beam elongates the grating of the glass fibre resulting in a shift in the grating s wavelength. The ratio between the deformation of the top beam and the deformation at the base of the portal is the transmission ratio of the sensing unit. This transmission ratio depends on the portal material and its geometry. To be able to measure strains up to 14 %, with a maximum strain of the bare FBG of 2 %, the transmission ratio of the portal should be 7. The strain sensing unit was first developed by combining two approaches from the structural mechanics illustrated in the Figure 14. Combined together, the two approaches enable to calculate the theoretical transmission ratio of the portal. Then, a calculation sheet developed for the Matlab environment was created in order to enable to calculate easily the transmission ratio of the unit for various geometries of portal. Figure 13: Functional principle of the strain sensing unit Figure 14: Approaches from the theory of structural mechanics describing the displacements of a beam under various solicitations 24 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

27 4.1.2 Prototypes developed During the project, two prototypes were developed for the strain sensing unit. The aim of the first prototype (Figure 15, left) was to test the feasibility of a portal sensing unit. It was designed to achieve a transmission ratio of approximately 5. After testing (see paragraph 4.2) the transmission ratio measured was 3.7, therefore, it must be taken into account that the approach provided in paragraph tends to overestimate the transmission ratio of the final sensing unit. A second prototype was designed with the aim of achieving a transmission ratio between 10 and 15. The adapted geometry of this second prototype led to a calculated transmission ratio of The geometry of this second prototype is shown in Figure 15, right. Since the principle of the portal frame was proven through tests with prototype 1, prototype two was only developed but not produced. Direction Pull Direction Push Figure 15: Prototype 1 of the FBG strain sensing unit (left) and geometry of the prototype 2, in mm (right) Stiffness of the strain sensing unit During the development of the strain sensing unit, the question of the maximum stiffness of the portal was raised. The question was: how stiff can the portal be that it does not influence the swelling/shrinking behaviour on the timber element which is monitored? In this report, a calculation method is proposed and compared with the characteristics of the strain sensing unit s prototype. In order to answer the question: how stiff can the strain sensor be that it does not influence the swelling/shrinking behaviour on the timber element which is monitored; several approaches have been considered which are summarized in Figure 16. Indeed, theoretically, as soon as a portal is fixed on the timber, it will restrain partially the wood from swelling. Thus, ideally, a strain sensing unit should have a stiffness of K = 0 to have no influence on the swelling behaviour of the wood. Up to a stiffness K1, the influence of the sensor s restrain is small. This stiffness K1 should be the target stiffness for our strain sensor. Moreover, we want in all cases to avoid cracking of the wood due to the restrain of the sensing unit. Thus, the upper bound for the stiffness of our strain sensor is K2 which is related to the strength of wood in tension perpendicular to the grain. Finally, the glass fibre we use in our sensor also provides a lower bound K3 for the stiffness of the sensing unit. This lower bound value is related to the failure of the glass fibre due to too much strain. This stiffness K3 can be adjusted by modifying ratios in the portal s shape. During the project, possible values for K1, K2 and K3 have been explored in the case of the prototype 1. They are detailed in the Appendix B.2 and a summary of the values found are shown in Table 14. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 25 / 70

28 Figure 16: Possible stiffness of the sensor Table 14: Possible values of K1, K2 and K3 according to the geometry of the strain sensing unit prototype 1 Target stiffness K1 4.5 N/mm influence of 3 % Maximum stiffness K2 6.7 N/mm (parabolic stress distribution) 10.0 N/mm (linear stress distribution) Minimum stiffness K3 0.9 N/mm 4.2 Testing of the strain sensing unit, prototype Testing the characteristics The prototype 1 of the strain sensing unit was tested in laboratory conditions. First, before production of the prototype, the bare FBG fibre was characterised. This characterisation was carried out by pulling on the bare fibre and recording the wavelength shift observed through the spectrometer. This shift, represented as a function of the displacement applied to the fibre, is shown in Figure 17. As expected, the behaviour of the bare FBG is linear. According to this test, a sensitivity of the bare fibre equal to 8.95 pm/μm was calculated. In a second step, the sensitivity of the prototype 1 was determined. For this, the base of the prototyped was displaced both in the pull and push direction (see Figure 15, left). Meanwhile, the wavelength shift read on the spectrometer was recorded. Figure 18 shows the wavelength of the reflected light of the FBG as a function of the displacement applied to the base of prototype 1. Here again, the wavelength shift was measured both in the case of a pull on the portal s base and in the case of a push. These two tests lead to the determination of a sensitivity of 2.39 pm/μm in the pull direction and a sensitivity of 2.45 pm/μm in the push direction. When comparing the sensitivity of the bare fibre to the ones of prototype 1 of the sensing unit, a transmission ratio of 3.7 is obtained. Figure 17: Displacement of the bare fibre and corresponding shift in the FBG wavelength with linear fit 26 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

29 Figure 18: Displacement of the base of prototype 1 and corresponding shift in the FBG wavelength with linear fit Testing of the behaviour on wood Firstly, it was conducted to investigate the behaviour of the strain sensing unit, prototype 1 when tied to a wood sample. Moreover, during this test, the behaviour of the relative humidity FBG sample provided by O/E-Land Inc. was proofed again. Finally, this test was used to verify the repeatability of the results obtained in the previous tests when using a different spectrometer. Indeed the initial testing of the strain unit, prototype 1 and of the relative humidity sensor were carried out with the equipment of the department Engineering and Information technology, especially with their spectrometer. In the test, the strain sensing unit was attached to a thin spruce sample along its cross direction using screws. Prior to the test, the spruce sample was conditioned in the standard climate of 20 C and 65 % of RH until it reached a constant equilibrium moisture content (verified by weight measures according to the standard. The spruce sample, to which the prototype 1 of the strain unit is attached, is then placed in a climate chamber Weiss WK11. In the climate chamber are also placed the FBG RH sensor and a datalogger of type ElproLog which is monitoring the temperature and the RH of the air inside the chamber. The measures recorded by the datalogger are used as reference. Finally, the climate in the chamber is varied and the behaviour of the strain unit and of the FBG RH sensor is recorded though a spectrometer. The material parameters and parameters used to record the temperature and the RH of the air through the datalogger ElproLog on the one hand and the ones used to record the evolution of the wavelength of both fibre sensors are summarized in Table 15, Table 16. Table 15: Material parameter for test series Material Climate chamber Datalogger ElproLog FBG sensor interrogator Strain sensing unit RH FBG sensor Wood sample Description/Reference Weiss WK11 Prototype 1 of the project From the company O/E-Land Inc. (see appendix C.1 for the specifications) Spruce sample of dimensions 60 x 24 x 120 mm, conditioned at 20 C and 65% of RH Table 16: Parameters for climate logger Quantity Measurement device Rate of recording Number of measure for each record Temperature Datalogger Each 5 minutes 1 Relative humidity Datalogger Each 5 minutes 1 Strain FBG unit prototype 1 Each 5 minutes 6 (1/sec.) and FBG interrogator Relative humidity FBG RH sensor and FBG interrogator Each 5 minutes 6 (1/sec.) Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 27 / 70

30 The evolution of the air temperature and RH during the test is shown in Figure 19 (left). In addition, on the same figure, the EMC calculated through the model of Simpson (see Equation (2) in paragraph 2.2.1) is shown in green. Finally, the evolution of the wavelength from both FBG sensing units is shown in Figure 19 (right). 100 Temperature RH EMC Temperature [ C] or RH [%] or MC [%] FBGS measure for strains [nm] FBGS measure for RH [nm] Time of measurement [hour] Figure 19: Initial test results FBG strain FBG RH Time of measurement [hour] The strain sensing unit was attached to the wood sample which swells and shrinks depending on the climate conditions. Therefore, based on the air temperature and relative humidity recorded by the datalogger, the theoretical moisture content related strains of the wood can be measured based on Equations (2) and (3, 4) from chapter 2.2: θ, 10 Δ 10 0 (6) With [%] The Equilibrium Moisture Content at time t 0 [%] The Equilibrium Moisture Content at time 0, [%] Simpson s model to calculate the Equilibrium Moisture Content based on the temperature and Relative Humidity (see Equation (2)) [-] The theoretical moisture content related strain calculated based on the climate [%/% of MC] = 0.19 The hygroexpansion coefficient of spruce in radial/tangential direction Similarly, the strains measured by the FBG prototype can be calculated from the wavelength shift. The strain sensing unit behaviour is described by the following Equation. With Δ βθ (7) [nm] [nm] [pm/μm] Δ [μm] β [pm/ C] θ [ C] [nm] The wavelength measured through the FBG interrogator The central wavelength of the prototype The sensitivity of the prototype to mechanical strains The displacement of the base of the prototype The thermal sensitivity of the prototype The temperature of the air The internal constant of the prototype At the beginning of the test, the displacement of the base of the prototype is null and the following values are measured. 28 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

31 Table 17: Measured values Quantity Wavelength measured by the FBG interrogator Temperature of the air Displacement of the base Value nm C 0 μm Equation (11) can be written for the beginning of the test. 10 Δ βθ (8) Thus by subtracting Equations (12) to Equation (11) we obtain: Δ Δ βθ θ (9) As the strain applied on the sensing unit is defined by: (10) We can calculate: 10 βθ θ L (11) With [nm] [nm] = β [pm/ C] θ [ C] θ [ C] = [pm/μm] = 2.39 L [μm] = The wavelength measured through the FBG interrogator The initial wavelength measured by the FBG interrogator The thermal sensitivity of the prototype The temperature of the air The initial temperature of the air The sensitivity of the prototype to mechanical strains The initial length of the sensing unit The sensitivity of the prototype was measured to be between 2.39 and 2.45 in the previous tests. Here the value of 2.39 is used. The thermal sensitivity of the prototype, however, was not measured is previous tests. Thus, Figure 20 shows the strains calculated from the FBG unit without temperature correction (black), using a thermal sensitivity of 10 pm/ C (dark grey) and using a thermal sensitivity of 20 pm/ C. These three curves are compared to the theoretical strains calculated based on the climate (in orange). Accordingly, one can see that the strains from the FBG unit effectively follow the variations of strains induced by the climate. However, the variations of the strain from the climate are more extreme than the ones measured by the FBG. This can be explained by the fact that moisture diffusion in wood is a very slow process. Therefore, the variations of the climate were too fast for the wood to reach the equilibrium moisture content. Accordingly, the moisture related strains of the wood sample were lower than the ones one could estimate from the climate. Regarding the thermal sensitivity of the FBG unit, one can see that higher sensitivities seem to lead to more accurate results. However, this topic should be further studied. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 29 / 70

32 Calculated strains [millistrain] ε from climate ε from FBGS (β = 0) 4 ε from FBGS (β = 10) ε from FBGS (β = 20) Time of measurement [hour] Figure 20: Calculated strains due to climate variations The FBG relative humidity sensor was placed together with the wood sample in the climate chamber and the measures recorded by it were compared to the ones recorded by the datalogger. The behaviour of the RH FBG sensor can be described using the following Equation. With hβθ (12) [nm] [nm] [pm/% of RH] [%] β [pm/ C] θ [ C] [nm] The wavelength measured through the FBG interrogator The central wavelength of the prototype The sensitivity of the prototype to changes of relative humidity The relative humidity of the air The thermal sensitivity of the sensor The temperature of the air The internal constant of the sensor Moreover, the producers provided in the specifications sheet of the sensor the following values. Table 18: Specifications from producer Quantity Value Sensitivity of the prototype to changes of relative humidity 3.8 to 4.0 pm/% of RH Central wavelength CWL nm Temperature of the air when the central wavelength was measured 24.5 C Relative humidity of the air when the central wavelength was measured 26.6% Accordingly, the internal constant of the sensor can be calculated. cst 26.6 β 24.5 (13) Thus the relative humidity of the air can be calculated based on the wavelength measured by the FBG sensor using the following equation: 1 10 βθ (14) First, the RH calculated based on the recordings of the FBG sensor when neglecting the thermal sensitivity of the sensor (β=0) and for different values of relative humidity sensitivity. As shown in Figure 21, the range of sensitivity provided by the sensor s producer leads to relative humidity values (curves light to dark grey) which are relatively far from the actual relative humidity of the air (blue curve). A 30 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

33 curve fitting analysis leads to the calculation of a sensitivity of (black curve). However the values obtained using this sensitivity ratio are still apart from the actual relative humidity values, especially on the second part of the test (hours 80 to 170) where the relative humidity of the air was quite low. Thus, a temperature correction was applied to the values measured by the FBG sensor. Through a curve fitting analysis, the thermal sensitivity of the FBG sensor was found to be of pm/ C for a relative humidity sensitivity of 4.53 pm/% of RH. These two sensitivity values lead to a very good fit between the relative humidity measured by the FBG sensor and the actual relative humidity of the air (see Figure 22) Relative Humidity [%] RH from FBGS (α = 3.8) RH from FBGS (α = 4.344) RH from FBGS (α = 3.9) RH from climate RH from FBGS (α = 4.0) Temp. from climate Time of measurement [hour] Figure 21: Test of RH sensor Temperature [ C] Relative Humidity [%] RH from FBGS RH from climate Temp. from climate Time of measurement [hour] Figure 22: Test of RH sensor Temperature [ C] This test showed that the measures conducted with the equipment of the department Engineering and Information technology could be repeated with the equipment of the department Architecture, Wood and Civil Engineering. Indeed, the measure of the relative humidity of the air through the FBG sensor was shown to be very accurate with a thermal sensitivity of the sensor of pm/ c and a sensitivity to relative humidity of 4.53 pm/% of RH. In addition, the measure of the moisture related strain of wood through the FBG strain sensing unit, prototype 1 was shown to be possible. Even though the climate used in this test had too fast variation for enabling the wood sample to reach equilibrium, the strains measured with the FBG unit are agreeing in their general behaviour with the theoretical strains obtain from the climate variation. However further, more precise tests should be conducted to obtain a final conclusion, especially regarding the thermal sensitivity of the unit. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 31 / 70

34 4.3 Development of the calculation sheet to analyse of the portal behaviour A calculation sheet was developed to predict the behaviour of the portal depending on its geometry. Through this sheet, the geometry of the portal can easily be adapted to meet the requirements of the strain sensing unit. This sheet, which script is provided in Appendix B.1, was developed in a Matlab environment. The calculation sheet models the FBG portal sensing unit as a beam system as shown in Figure 23. For this beam system, the material of each beam can be defined through it E-modulus. The geometry of each beam can also be changed by the user. This geometry corresponds to the neutral axis of the real portal. Once this information is provided to the calculation sheet, the displacement of each node of the system is calculated (see for example Figure 24). In addition, the transmission ratio of the portal is also given. This calculation sheet was tested on the geometry of the prototype 1; it resulted in a transmission ratio of 5.7. The results provided by the calculation sheet were also verified using the static software Cubus Statik 6. The difference between the transmission ratio obtained by the calculation sheet and the one of the prototype 1 of the sensing unit could be explained by geometry tolerances and slip of the FBG fibre with regards to the top bar of the portal. Table 19 provides an overview of the transmission ratios obtained and calculated for the two prototypes of the FBG strain sensing unit. Figure 23: Beam model of the portal strain sensing unit Figure 24: Geometry of the beam model representing the strain sensing unit, before (in red) and after deformation (in blue) 32 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

35 Table 19: Comparison of the transmission values as calculated and actually obtained for prototype 1. Transmission ratios Prototype According to the first calc. According to the calc. sheet Experimental Synthesis regarding the strain sensing unit In conclusion, the principle of strain sensing unit developed in this chapter is suitable to monitor MCrelated strains. Indeed, the tests performed on prototype 1 showed that the use of a portal enabled to increase the maximum strain the unit is able to measure. The prototype 1 reached a transmission ratio of 3.7 corresponding to a maximum strain of the sensing unit of 7.4 %. The prototype 2, however not tested, was designed to reach a theoretical transmission ratio of 12.5, corresponding to a theoretical maximum strain of 25 %. This maximum strain of 25 % is larger than the maximum strain of 14% required by the specification of the strain sensing unit. However, while testing the first prototype, it was observed that the theoretical transmission ratios calculated are often overestimating the real transmission ratio of the units. In addition, a calculation sheet was developed to enable a fast calculation of the theoretical transmission ratio of the sensing unit depending on its geometry and material. Finally, the question of the stiffness of the sensing unit was identified as a potential issue in the development of the sensor. Basic calculations were carried out to obtain an overview of it. However, further tests would be required to definitely close this topic. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 33 / 70

36 5 Moisture content of wood sensing unit 5.1 Measuring wood moisture content Current methods to measure wood moisture content The method currently used to measure wood MC and described in the EN :2002 [23] is called the oven-dry method. It is the only direct method existing to determine the MC of a wood sample. When neglecting the weight of wood s volatile extractives, the oven-dry method provides accurate values for the wood MC. However, in addition to be very slow, this method is also destructive, as it requires cutting the wood sample. The oven-dry method is, therefore, mainly restricted to laboratory use and definitely not suitable for in-situ assessments or long-term monitoring of timber elements. Thus, other methods, classified as indirect methods, are commonly used in the timber industry to estimate the MC of wood samples and timber elements (see Figure 25). These indirect methods are classified in two categories. In the first category, the MC is deduced by measuring the evolution of other physical properties, such as electrical resistance or capacitance to name the most common ones. In the second category, the MC of wood is derived from the climate in which the wood piece is in equilibrium, namely the air relative humidity and temperature. This derivation can be based either on experimental sorption isotherms (see for example Figure 3) or on mathematical models describing the sorption behaviour of wood such as the one provided by Simpson [24]. Figure 25: Summary of existing measurement methods for moisture content in timber, [48] Wood moisture content and FBG technology Currently, no FGB sensor is available to measure directly wood MC. Similarly, no FBG sensors were developed to measure physical parameters such as electrical properties. Indeed, the advantages of the FBG technology rely on the exclusive use of FBG components. Thus the introduction of components related to other technologies, such as electronic components, would necessarily result is the loss of advantages such as the immunity to corrosion or low signal loss. However, FBG moisture and water content sensors are available on the market. They rely on a hygroscopic polyamide coating of the fibre s surface at the level of the Bragg grating section (as illustrated in Figure 26). This polyamide resin was selected because of its linear response in volume swelling in function of humidity levels. It expands with increasing humidity. As a result of this expansion or shrinkage, the Bragg grating section of the fibre experiences mechanical strains [28]. These types of sensors have been already successfully employed to measure the moisture in concrete [49]. 34 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

37 Figure 26: Polyamide coated humidity sensor [28] Wood moisture content and climate in a cavity in the wood Usually, the indirect method deriving the MC of wood from the climate is based on the following principle: The moisture content of wood depends on the relative humidity and temperature of the air surrounding it. If the wood remains long enough in air where the relative humidity and temperature remain constant, the moisture content will also become constant at a value known as the equilibrium moisture content (EMC) [24]. While monitoring five timber bridges in Norway, Dyken and Kepp [50] assumed that the phenomenon of the EMC could also be applied in the opposite situation. Namely, if a small volume of air is trapped in wood, it would reach an equilibrium with the MC of the wood. Thus, the measurement of the temperature and relative humidity of this trapped air would enable to calculate the MC of the surrounding wood. The investigations indeed showed that it was possible to estimate the MC of wood based on the temperature and RH of air trapped in this wood. However, Dyken and Kepp also showed that models commonly used to derive the EMC of wood from the climate (such as the model from Simpson [24]) could not be applied in this situation. Indeed, when applying Simpson s model to the climate measured in the cavity, the MC of wood was underestimated and showed very large fluctuations with regards to the cavity s air temperature. Dyken and Kepp therefore concluded that the relation between RH and MC, at a given temperature, is not the same for a small piece of wood in a large volume of ambient air and a small volume of air inside a large volume of wood [50]. In order to still used their cavity method to monitor wood MC, Dyken and Kepp developed a new model describing the evolution of the MC based on the air temperature and RH in a small cavity. This model was also validated experimentally by Pence [32]. It combines two parabolas: one with regards to the temperature and one with regards to the RH of the cavity s air. Equation (6) shows the model obtain with its nine coefficients., (15) With, % %,, The percentage of MC of the wood The temperature of the air in the cavity The RH of the air in the cavity Coefficients of the model The coefficients which were obtained experimentally in [50] and used as well in [32] are provided in Table 20. Table 20: Experimental coefficients of the EMC model of wood when measuring the air in a small cavity [50] Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 35 / 70

38 5.1.4 Synthesis regarding the measurement of wood moisture content using FBG technology Based on the facts presented above, it was decided in this project to develop the MC unit of our hybrid FBG sensor by combining two existing FBG sensors: a RH sensor and a temperature sensor. These both sensors, placed in cavity in the wood not in contact with the outside air, will enable the estimation of the MC of the wood element. For that, the model developed by Dyken and Kepp [50] is used and verified experimentally. In addition, the temperature measured in the MC unit will be used to correct with regards to temperature the strains measured by the strain unit (see chapter 0). 5.2 Air relative humidity sensor The FBG humidity sensor used in this project was manufactured by the company O/E-Land Inc. It is constituted of a glass fibre which is polyamide coated at the level of the Bragg grating section. This section of the fibre is protected by a perforated steel tube while the rest of the fibre is protected by a flexible plastic tube (see Figure 27). The specification of the sensor provided by the company O/E- Land is given in Appendix C.1. According it, a linear sensitivity of the sensor with respect to relative humidity change was first assumed with a sensitivity of 3.9 pm/% of RH. Figure 27 : The FBG relative humidity sensor OEFHS-100A Characterization of the humidity sensor controlled climate The producer s specifications of the RH sensor were validated by testing the behaviour of the RH sensor when increasing and decreasing the air relative humidity in a climate chamber. For these tests, the RH of the air was increased, respectively decreased, by steps of 20% of RH. The temperature was maintained constant at 25 C for the complete duration of the tests. The RH measured through the shift of the FBG reflected wavelength was compared to the values measured by a calibrated data logger of type EcoLog TH1 (ElproLog). The results of these tests are shown in Figure 28 (increasing humidity) and Figure 29 (decreasing humidity). In these graphs, the actual values of relative humidity (in green) and temperature (in red) obtained from the climate logging device ElproLog are shown. They are compared to the values measured by the FBG humidity sensor (in blue). These humidity values were calculated according to the reflected wavelength and the sensitivity of the sensor (3.9 pm/% of RH). In order to reduce the differences between the values measured by the FBG humidity sensor and the one from the climate logging device, an adjustment of the sensor sensitivity was implemented. For each level of RH, the sensitivity of the FBG humidity sensor was adjusted to fit the reference measurement from the ElproLog device (see Table 21). Based on these eight different sensitivity values (four with increasing humidity and four with decreasing humidity), two polynomials of degree 4 were calculated through polynomial fit. The humidity values calculated using these polynomial sensitivities are shown in black in Figure 28 and Figure / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

39 Figure 28: Measured data by FBG humidity sensor and ElproLog with increasing humidity. Figure 29: Measured data by FBG humidity sensor and ElproLog with decreasing humidity. Table 21: Adjusted sensitivity for increasing and decreasing humidity For increasing humidity For decreasing humidity Humidity in % RH Sensitivity in pm/%of RH Humidity in % RH Sensitivity in pm/% of RH Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 37 / 70

40 5.2.2 Characterization of the humidity sensor real climatic conditions In order to validate the measures obtained from the RH FBG sensor in the case of real climatic conditions, the sensor was placed outside and the reference climate was recorded using a climate data logging device ElproLog. The results of this test are provided in Figure 30. The reference relative humidity and temperature, measured by the ElproLog, are shown in green and red respectively. The values originally measured by the RH sensor are shown in black, calculated using a constant sensitivity ratio of 3.9 pm/% of RH without compensation of the thermal expansion. During the measuring period of 60 hours, the temperature dropped by 15v C. As can be seen on the following figure, the thermal expansion induces a relatively large wavelength shift of 10 pm/ C. This shift is to be compared with the shift induced by relative humidity itself of 3.9 pm/% of RH. Therefore, the thermal influence on the measurement of the RH sensor has to be compensated. The blue curve shows the RH measured by the FBG sensor when including a thermal compensation. Figure 30: Measured data of FBG humidity sensor and climate logging device in real conditions Synthesis regarding the relative humidity sensor The tests conducted on the RH FBG sensor proved that it is possible to be used for monitoring the RH of the air. However, temperature compensation has to be included to obtain reliable measures. Thus, the use of a separate temperature sensor is necessary when considering using this RH sensor. In addition, more accurate values of relative humidity can be achieved when using a sensitivity which follows the levels of relative humidity rather than a constant sensitivity of 3.9 pm/% of RH. 5.3 Temperature sensor The FBG temperature sensor used in this project was manufactured by the company FiberSensing. It is originally a weldable temperature sensor, therefore the Bragg grating part of the fibre is protected by a stainless steel casing while the remaining of the fibre in protected by a flexible plastic tubing, see Figure 31. More specifications provided by the company FiberSensing regarding this sensor are available in Appendix C. FBG temperature sensors are essentially the same as FBG strain sensors. However, in this case, the strain is induced by thermal expansion of the fibre itself instead of external mechanical stress. 38 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

41 Figure 31: The FBG temperature sensor FS 6300 from the company FiberSensing Characterization of the temperature sensor In order to validate the specifications provided by the producer, the two temperature sensors that were available were tested in a climate chamber with controlled temperature, the tests are described in Appendix D, test C. The temperature in the climate chamber was increased stepwise from -10 C to 40 C during daytime, held constant overnight at 40 C, and reduced again in steps the next day until - 5 C. The wavelength shift of the FBG temperature sensor was recorded. The measurements were then compared with the reference temperature values which were recorded during the test by a calibrated data logger of type EcoLog TH1 (ElproLog). The results of this test are provided in Figure 32. The figure shows a time trace of the recorded temperature during the experiments on the left and the temperature of the FBG measurements compared to that of the Ecolog TH1 on the right. The time in which the temperature made large jumps was omitted from the data because the sensors did not adjust at the same speed. The sensors and climate chamber needed approximately 5 minutes to adjust. Large jumps in temperature are not expected in monitoring conditions. Figure 32: Testing of the FBG temperature sensor from the company FiberSensing The measurements with the temperature sensors showed a high resolution of values 0.4E-3 C over a 3 minute interval, where a resolution 0.1 C was required in Section 3.5. Theoretically, the resolution should be 1.05E-4 C per second regarding the printed accuracy of the numbers in the FBG-files. These values were found as well. A linear fit was done between the FBG temperature sensor and the Ecolog was done and differences of 0.24 % and % were found for FBSG1 and FBGS2 respectively. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 39 / 70

42 5.3.2 Synthesis of the temperature sensor The temperature sensor provided by FiberSensing showed a good ability to monitor the temperature of air. Compared to the initial requirements, the sensor accurately measured the temperature within the range -10 C to +40 C. Finally, this sensor will be used for multiple functions in the final hybrid FBG sensor developed in this project: - It will provide an accurate temperature level in order to calculate the moisture content of the wood according to the equation from Dyken and Kepp [50]. - It will be used for the temperature compensations of the values measured by the humidity FBG sensor - It will be used for the temperature compensation of the values measured by the strain sensor 5.4 Synthesis regarding the moisture content sensing unit According to the observations of the experiments to verify both the relative humidity sensor and the temperature sensor, both parameters can be measured accurately. Implementation in a moisture content sensing unit would be possible through a combination of both measurements in the equations by Dyken and Kepp [50]. 40 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

43 6 Combination of the three sensors 6.1 Choice of the layout and recommendations The integrated sensor to monitor wooden structures which was developed in this project has to consist of the following three sensing units: a MC-related strain sensing unit, a MC sensing unit and a temperature sensing unit. As the FBG technology enables to multiplex several Bragg gratings along a single fibre, these three sensing units can be read out in a single channel of the spectrometer. However, to be able to analyse the signal of the different sensing units properly, two adjacent peaks of the reflected spectrum corresponding to two sensing units have to be separated enough to allow a reliable identification of the peaks. Therefore, the offset between the centre wavelengths of two sensing units has to be greater than the possible shift of the reflected peaks. Then, to multiplex three sensing units in a single fibre, the particular fibre of each unit can be spliced with the one of the adjacent unit to achieve a single fibre. In this process, the ends of two fibres get stripped, cleaned, cleaved and finally fused together. With this process the connection of several sensing units and furthermore of several complete FBG sensors (including a temperature unit, a MC unit and a MC-related strain unit) can be achieved. The use of the splicing technique enables to achieve a multiplexing without resorting to bulky fibre connectors. 6.2 Description of the final layout The concept for an integrated sensor is shown in Figure 33. It consists of the temperature-, humidityand strain sensing units spliced together and mounted on a baseplate with a dilatation joint. The humidity- and temperature units are fixed on this baseplate, while the strain sensor is able to follow the deformation of the wooden structure. Figure 33: Possible setup for an integrated sensor Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 41 / 70

44 7 Conclusions and outlook 7.1 Conclusion regarding the project This project showed that the development of a hybrid sensor based on the Fibre Bragg Grating technology and dedicated to the monitoring of timber elements is possible. Indeed, according to the specifications of such sensor, identified based on existing values from reports of monitoring of timber structures, a moisture content, temperature and moisture content-related strain sensing units were developed. The concept of each of these units was tested, using prototypes, in laboratory conditions. These tests validated the ability of these prototypes to reach the desired specificities. In addition, a concept was developed enabling to combine these three sensing units into a single hybrid Fibre Bragg Grating sensor which could be used as such either embedded or surface mounted to a timber element. The concept of a hybrid Fibre Bragg Grating sensor developed in this project was optimized for the monitoring of timber structures. Indeed, this use provided a favourable frame for determining the specifications of the sensor itself. However, its use for testing the behaviour of wooden samples, for instance for laboratory testing, is also suitable. Indeed, depending on the experiment conducted, the expected range of variation of the three quantities monitored (temperature, moisture content and strain) can perfectly be within the operational range of the sensor. Finally, if necessary, the concept of the sensor can be slightly modified to match other specifications, especially in the case of the strain sensing unit. 7.2 Outlooks The Fibre Bragg Grating technology is currently at a turning point of its development. Indeed, this technology can be applied to monitor the quantities to which Fibre Bragg Gratings are intrinsically sensitive (strain and temperature changes) in a very reliable and efficient way. In addition, for these applications, feedbacks regarding the use of Fibre Bragg Grating sensors in middle to long term period are starting to be available. These feedbacks confirm the adequate behaviour of strain and temperature measuring Fibre Bragg Grating sensors on the long term. All this makes Fibre Bragg Grating sensors a technology already widely used in the monitoring of constructions in steel and concrete and therefore relatively easily available. However, dedicated applications of the Fibre Bragg Grating technology such as for the measurement of relative humidity are still in the beginning of their development. Indeed, currently, only the company O/E-Land Inc. appeared to be able to provide this type of sensors. Similarly, few Fibre Bragg Grating sensors have been developed to monitor other quantities than temperature and strain. Pressure, displacement or accelerometers Fibre Bragg Grating sensors are existing on the market but scarcely available. When comparing the relatively low number of producers and the quite narrow range of products they currently offer with the actual possibility of the Fibre Bragg Grating technology, this technology should follow a quite drastic development in the near future. This hypothesis tends to be confirmed by the fact that the company FiberSensing which provided some of the Fibre Bragg Grating sensors for this project was recently acquired by the well-known quality proof company HBM. This move shows again the potential of future development of this technology. During a discussion between the company FibreSensing and Ms. N. Magnière, dealing with the characteristics and possibilities of the FiberSensing s products, the company stated its interest in possible cooperation together with the BFH aiming at developing Fibre Bragg Grating sensors. This statement, however not formal, can be again taken in favour of a further project dealing with the Fibre Bragg Grating technology and oriented towards the industry. Finally, as noticed in the chapter 2 (State of the art), the Swiss company Smartec seems to provide very complete and reliable products in the field of building monitoring based on the Fibre Bragg Grating technology. They, however, do not have any case study dealing with timber structures, but exclusively with concrete or steel structures. It appears that the sector of the timber construction should put effort into using and optimizing the Fibre Bragg Grating technology for the timber material in order to be able to compete with steel and concrete in this area. For this, the development of sensors dedicated to the monitoring of timber and its production would be a definite advantage. 42 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

45 However, the main question remaining is the cost of such sensors and sensing systems. There should, therefore, be not only a development regarding the Fibre Gragg Grating technology itself, but also regarding the processing and implementing of the sensing systems for timber structures. As an example, the development process of the SOFOS system could be analysed. The development of the Fibre Bragg Grating technology for timber should, in addition, be integrated to coordinated actions regarding the best-practice dealing with monitoring of timber structures. 8 Regulations of the present report It is not allowed to copy or reproduce this report without the prior agreement of the Bern University of Applied Sciences Architecture, Wood and Civil Engineering. Furthermore, the publication of this report or any extracts from the report requires the prior written agreement of the Bern University of Applied Sciences. The original report will be kept for 5 years. This report is only valid with the signatures of both the head of the institute Timber Construction, Structure and Architecture and the project leader. This report includes 70 pages incl. appendix. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 43 / 70

46 9 Indexes 9.1 Abbreviations Abbreviation FBG MC EMC FSP L R T RH Meaning Fibre Bragg Grating Moisture Content Equilibrium Moisture Content Fibre Saturation Point Longitudinal direction of the wood material Radial direction of the wood material Tangential direction of the wood material Relative Humidity of the air 9.2 Index of Tables Table 1: The project team... 8 Table 2: Summary of the functions for which monitoring system have been and could be used for timber structures Table 3: Summary of the impacts that moisture content (MC), MC-related strains and temperature can have on a timber structure Table 4: Average and extreme values of hygroexpansion coefficients for various wood species (adapted from [26]) Table 5: Two example of monitoring systems for timber bridges case 1 (left) based on [11] and case 2 (right) based on [36], [37] Table 6: Summary of the main parameters influencing each sensing unit Table 7: Summary of the category of use of timber buildings considered to define the FBG sensor specifications Table 8: Average temperature and maximal temperature variations for timber structures corresponding to different category of use (adapted from [38]) Table 9: Average temperature and maximal temperature variations for Swiss timber bridges (adapted from [38]) 20 Table 10: Moisture content levels and variations measured during the monitoring of four timber bridges (according to [18], [38]) Table 11: Moisture content levels and variations measured during the monitoring of timber structures from various categories of use (according to [38]) Table 12: Moisture content related strains for softwoods and beech wood for a variation of MC of 16 % and 30 % Table 13: Summary of the specifications for the different sensing units of the FBG sensor Table 14: Possible values of K1, K2 and K3 according to the geometry of the strain sensing unit prototype Table 15: Comparison of the transmission values as calculated and actually obtained for prototype Table 16: Experimental coefficients of the EMC model of wood when measuring the air in a small cavity [50] Table 17: Adjusted sensitivity for increasing and decreasing humidity Table 18: Characteristics of the beams of the model Table 19: Restrictions applied to nodes 1 and Table 20: Numerical values used in the calculation Table 21: Calculated maximum stiffness for the prototype Table 22: Stiffness of the prototype Table 23: Minimum swelling stresses required Table 24: Measured values Table 25: Specifications from producer / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

47 9.3 Index of Figures Figure 1: (a) the H8 tower in Bad Aibling [1], (b) the Elephant house in Zürich [2], (c) close view of hybrid timber beams from the ski school in Arosa [3] and (d) the Pyramiden Kogel tower in Austria [4]... 6 Figure 2: The Bad Reichenhall ice arena before collapse (left) and partial view of the collapsed roof structure (right) [5]... 9 Figure 3: EMC as a function of the air temperature and relative humidity for Sitka Spruce (right) based on R. Keylwerth (1949, see [19]) Figure 4: EMC as a function of the air temperature and relative humidity for wood in general (adapted from [25]). In red, the standard testing conditions for wood: 20 C and 65% relative humidity resulting in an EMC of 12% in the wood Figure 5: Evolution of water content in wood, w = moisture content (adapted from [26]) Figure 6: Left, swelling of a timber element (source: [27]) and right, volume changes as a function of the moisture content w (adapted from [26]) Figure 7: Apparition of internal stresses in a glued laminated beam subjected to drying and corresponding location of crack formation (source: [27]) Figure 8: Structure and principle of a FBG (source: [28]) Figure 9: For different building categories, qualitative repartition of the category depending on average values and variations of relative humidity of the air (left) and of the temperature of the air (right). Analyzed in [38] Figure 10: Average annual precipitation in Switzerland [39] and location of the four timber bridges monitored in [38] Figure 11: Average annual temperature in Switzerland [40] and location of the four timber bridges monitored in [38] Figure 12: Sketch of a possible moisture chain i.e. exposure to moisture from the tree to gluedlaminated timber elements in the buildings (source [41]) Figure 13: Functional principle of the strain sensing unit Figure 14: Approaches from the theory of structural mechanics describing the displacements of a beam under various solicitations Figure 15: Prototype 1 of the FBG strain sensing unit (left) and geometry of the prototype 2, in mm (right) Figure 16: Possible stiffness of the sensor Figure 17: Displacement of the bare fibre and corresponding shift in the FBG wavelength with linear fit Figure 18: Displacement of the base of prototype 1 and corresponding shift in the FBG wavelength with linear fit Figure 19: Beam model of the portal strain sensing unit Figure 20: Geometry of the beam model representing the strain sensing unit, before (in red) and after deformation (in blue) Figure 21: Summary of existing measurement methods for moisture content in timber, [48] Figure 22: Polyamide coated humidity sensor [28] Figure 23 : The FBG relative humidity sensor OEFHS-100A Figure 24: Measured data by FBG humidity sensor and ElproLog with increasing humidity Figure 25: Measured data by FBG humidity sensor and ElproLog with decreasing humidity Figure 26: Measured data of FBG humidity sensor and climate logging device in real conditions Figure 27: The FBG temperature sensor FS 6300 from the company FiberSensing Figure 28: Testing of the FBG temperature sensor from the company FiberSensing Figure 29: Possible setup for an integrated sensor Figure 30: MC w of solid wood (left) and of glued laminated timber (right) as a function of the air RH ϕ (SN /1:2009, Figures 1a and 1b, 4.2 [51]) Figure 31: Beam model of the strain sensing unit Figure 32: Degrees of freedom of a beam element [58] Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 45 / 70

48 Figure 33 : Code developed in Matlab environment to calculate the behaviour of the strain sensing unit Figure 34: Area where the calculation is conducted Figure 35: Initial and final state of the area under consideration with an increase of moisture content of the wood Figure 36: Initial test results Figure 37: Calculated strains due to climate variations Figure 38: Test of RH sensor Figure 39: Test of RH sensor Figure 40: Time trace of measured temperature (right) and comparison of temperature measured with the FBGS and the Ecolog Bibliography [1] H8 Building in Bad Aibling. Photo: Y. Hilinci, Wikipedia.. [2] Le dôme des éléphants de Zürich. (c) DR.. [3] Miteinander verleimtes Eschen- und Fichtenholz, Skischule Arosa. Photo: Lukas Denzler, office fédéral de l environement OFEV.. [4] The observation tower Pyramidenkogel. Photo: Johann Jaritz, Wikipedia.. [5] S. Winter and H. Kreuzinger, The Bad reichenhall ice-arena collapse and the necessary consequences for wide span timber structures, presented at the WCTE World Conference in Timber Engineering, Miyazaki, Japan, [6] Brochure of the COST action E55.. [7] E. Frühwald, E. Serrano, T. Toratti, A. Emilsson, and S. Thelandersson, Design of Safe Timber Structures - How Can We Learn from Structural Failures in Concrete, Steel and Timber?, Division of Structural Engineering, Lund University, Lund, Sweden, [8] P. Dietsch, Typische Tragwerksmängel im Ingenieurholzbau und Empfehlungen für Planung, Ausführung und Instandhaltung, in 8. Grazer Holzbaufachtagung, [9] H. J. Blass and M. Frese, Schadensanalyse von Hallentragwerken aus Holz, Karlsruher Institut für Technologie (KIT) Lehrstuhl für Ingenieurholzbau und Baukonstruktionen, Karlsruhe, Germany, Karlsruher Berichte zum Ingenieurholzbau Band 16, [10] Borchure of the COSt action FP [11] E. Saracoglu, A. Gustafsson, and P. A. Fjellstroem, Short term monitoring of a cable-stayed timber footbridge, presented at the ICTB International Conference on Timber Bridges, Las Vegas, USA, [12] A. Gustafsson, A. Pousette, and N. Bjoerngrim, Health Monitoring of timber bridges, presented at the ICTB International Conference on Timber Bridges, Lillehammer, Norway, [13] M. Ledesma and J. Rodrigues, Dynamic Monitoring of a large span wood roof, in SHATIS International Conference on Structural Health Assessment of Timber Structures, Lisbon, Portugal, [14] L. Jorge and A. Dias, X-Lam panels in swimming pool buildings - monitoring the environment and the performance, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Trento, Italy, [15] A. Dias, L. Jorge, M. Ferreira, and H. Martins, Monitoring and testing of a timber-concrete bridge, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Lisbon, Portugal, [16] M. Hasan, R. Despot, J. Trajkovic, A. O. Rapp, C. Brischke, and C. R. Welzbacher, The Echo pavilion in forest park Maksimir Zagreb - Reconstruction and health monitoring, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Trento, Italy, [17] I. Kirizsan and B. G. Szabo, Investigation and monitoring of historic roof structures during conservation, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Lisbon, Portugal, / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

49 [18] B. Franke, A. Müller, M. Vogel, and T. Tannert, Langzeitmessung der Holzfeuchte und Dimensionsänderung an Brücken aus blockverleimten Brettschichtholz, Bern University of Applied Sciences, Architecture, Wood and Civil Engineering, Bern, Switzerland, Research report 09.11, [19] A. Gamper, P. Dietsch, and M. Merk, Gebäudeklima - Langzeitmessung zur Bestimmung der Auswirkungen auf Feuchtegradienten in Holzbauteilen, Stuttgart, Germany, Technical report F 2816, [20] F. Lanata, An on-going monitoring project of a new timber structure, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Trento, Italy, [21] L. Jorge, Environmental and Moisture Measurements, presented at the Training School on Monitoring of Timber Structures, Nantes, France, Jun [22] A. Cavalli, Monitoring of timber structures, old ideas for new needs, presented at the Training School on Monitoring of Timber Structures, Nantes, France, Jun [23] SN EN :2002 Teneur en humidité d une pièce de bois scié - Partie 1: Détermination par la méthode par dessiccation. SIA Swiss Society of Engineers and Architects, [24] W. T. Simpson, Equilibrium Moisture Content of Wood in Outdoor Locations in the United States and Worldwide, Forest Products Laboratory, US Department of Agriculture, Research note FPL-RN-0268, [25] Wood handbook - Wood as an engineering material, Forest Products Laboratory, US Department of Agriculture, General technical report FPL-GTR-190, [26] J. Natterer, J.-L. Sandoz, and M. Rey, Construction en bois: Matériau, technologie et dimensionnement, 2nd edition. Presses polytechniques et universitaires romandes, [27] B. Franke, S. Franke, A. Müller, M. Vogel, F. Scharmacher, and T. Tannert, Long term monitoring of timber bridges - Assessment and results, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Trento, Italy, [28] L. Wang, N. Fang, and Z. Huang, Polyimide-Coated Fiber Bragg Grating Sensors for Humidity Measurements, in High Performance Polymers - Polyimides Based - From Chemistry to Applications, M. Abadie, Ed. InTech, [29] Website of the company Smartec. [Online]. Available: [30] H. Morris, M. Worth, and P. Omenzetter, Monitoring modern timber structures and connections, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Lisbon, Portugal, [31] U. M. Deza, Development, Evaluation and Implementation of Sensor Techniques for Bridges Critical to the National Transportation System, Doctoral Thesis, Iowa State University, Ames, Iowa, USA, [32] T. Pence, Development of a smart timber bridge- sensor evaluation and data processing techniques, Master Thesis, Iowa State University, Ames, Iowa, USA, [33] B. M. Phares, T. J. Wipf, U. M. Deza, and J. Wacker, Development of a Smart Timber Bridge A Five- Year Plan, Forest Products Laboratory, US Department of Agriculture, Madison, WI, USA, General Technical Report FPL-GTR-195, [34] J. Wacker, Development of a Smart Timber Bridge Girder with Fiber Optic Sensor, presented at the ICTB International Conference on Timber Bridges, Lillehammer, Norway, [35] J. Wacker, U. Deza, B. M. Phares, and T. J. Wipf, Development of a Smart Timber Bridge Girder with Fiber Optic Sensors, presented at the ICTB International Conference on Timber Bridges, Lillehammer, Norway, [36] Broennimann, R. Widmann, U. Meier, P. Irniger, and A. Winistoerfer, Design, construction and monitoring of a bowstring arch bridge made exclusively of timber, CFRP and GFRP, presented at the WCTE World Conference in Timber Engineering, Riva del Garda, Italy, [37] R. Widmann, R. Broennimann, and U. Meier, Monitoring of a CFIRP-GFRP-timber bowstring arch bridge using a novel sensing systems, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Lisbon, Portugal, [38] B. Franke, A. Müller, S. Franke, N. Magnière, "Langzeituntersuchung zu den Auswirkungen wechselnder Feuchtegradienten in blockverleimten Brettschichtholzträgern", Research Report, Bern University of Applied Sciences, Institute for Timber Construction, Structures and Architecture, [39] Système d information géographique, Carte Suisse des précipitation annuelles médianes. Centre informatique de l UNIL. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 47 / 70

50 [40] Système d information géographique, Carte Suisse des temperatures annuelles médianes. Centre informatique de l UNIL. [41] P. Dietsch, A. Gamper, M. Merk, and S. Winter, Building climate - Long-term measurements to determine the effect on the moisture gradient in large-span timber structures, in CIB Working Comission W18 - Timber Structures - Meeting 45, Väjö, Sweden. [42] SN :2012 Timber Structures. SIA Swiss Society of Engineers and Architects, [43] SN EN :2004 Eurocode 5 Conception et calcul des structures en bois - Partie 1-1: Généralités - Règles communes et règles pour les bâtiments. SIA Swiss Society of Engineers and Architects, [44] NF EN /NA:2008 Eurocode 5 - Design of Timber Structures Part1-1: General Common Rules and Rules for Buildings National Annex to NF EN :2005. European Committee for Standardization, [45] V. Angst-Nicollier, Moisture Induced Stresses in Glulam - Effect of Cross Section Geometry and Screw Reinforcement, Doctoral Thesis, Norwegian University of Science and Technology, Faculty of Engineering Science and Technology, Department of Structural Engineering, Trondheim, Norway, [46] J. F. Rijsdijk and P. B. Laming, Physical and related properties of 145 timbers: information for practice. Dordrecht ; Boston: Kluwer Academic Publishers, [47] Z. Zhou, C. Lan, and J. Ou, New kind of FBG-based crack (large strain) sensor, in Smart Structures and Materials 2006: Smart Sensor Monitoring Systems and Applications, San Diego, CA, 2006, pp [48] B. Franke and S. Franke, Monitoring of timber structures, presented at the SHATIS International Conference on Structural Health Assessment of Timber Structures, Trento, Italy, [49] T. L. Yeo, D. Eckstein, B. McKinley, L. F. Boswell, T. Sun, and K. T. V. Grattan, Demonstration of a fibreoptic sensing technique for the measurement of moisture absorption in concrete, Smart Mater. Struct., vol. 15, no. 2, pp. N40 N45, Apr [50] T. Dyken and H. Kepp, Monitoring the Moisture Content of Timber Bridges, presented at the ICTB International Conference on Timber Bridges, Lillehammer, Norway, [51] SN /1:2009 Construction en bois - Spécifications complémentaires. SIA Swiss Society of Engineers and Architects, [52] SN EN :2004/A1:2008 Eurocode 5: Conception et calcul des structures en bois - Partie 1-1: Généralités - Règles communes et règles pour les bâtiments - Amendement A1 à la norme EN :2004. SIA Swiss Society of Engineers and Architects, [53] DIN EN /NA:2010 Nationaler Anhang - National festgelegte Parameter - Eurocode 5: Bemessung und Konstruktion von Holzbauten - Teil 1-1: Allegemeines - Allgemeine Regeln und Regeln für den Hochbau. DIN Deutsches Institut für Normung, [54] SN EN 335:2013 Durabilité du bois et des matériaux à base de bois - Classes d emploi: définitions, application au bois massif et aux matériaux à base de bois. SIA Swiss Society of Engineers and Architects, [55] Informations complémentaires - Classes d emploi. Lignum, [56] SIA /1:2003, Timber Structures - Supplementary Specifications.. [57] V. Angst and K. A. Malo, Moisture-induced stresses in glulam cross sections during wetting exposures, Wood Sci. Technol., vol. 47, no. 2, pp , Jul [58] Z. El Maskaoui, Application de la programmation orientée objets à l optimisation discrète sous contraintes des structures metalliques formées de poutres via les algorithmes génétiques - Chapitre 2: Analyse linéaire des structures à poutres. Formulation aux éléments finis, Doctoral Thesis, Faculté polytechnique de Mons, Mons, Belgique, [59] Sciences de l Ingénieur - Exemple de diagnostic d une simulation : déformations d un portique. [Online]. Available: [Accessed: 13-Oct-2014]. 48 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

51 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 49 / 70

52

53 Appendixes Appendix A Wood hygroexpansion coefficients 53 Appendix B Development of the strain sensor 55 B.1 Approach 2 Development of a calculation sheet 55 B.1.1 Principle of the sheet 55 B.1.2 Matlab code of the sheet 59 B.2 Investigations over the stiffness of the strain sensor 61 B.2.1 K 2 Sensor s maximum stiffness to prevent cracks in the wood 61 B.2.2 Comparison with the characteristics of the prototype 63 B.2.3 Minimum swelling stress 64 B.2.4 Discussion and conclusion over K 2 64 B.2.5 K 1 - Target sensor stiffness with little influence on the swelling behaviour 64 B.2.6 Discussion and conclusion over K1 65 B.2.7 K 3 Minimum sensor stiffness related to the strength of the glass fibre 65 Appendix C Specifications of the FBG relative humidity and temperature sensors 66 C.1 The relative humidity sensor (company: O/E-Land Inc.) 66 C.2 Specifications of the temperature sensor (company: FiberSensing) 67 Appendix D Test C Testing of the temperature FBG sensor controlled climate 69 D.1 Test preparation 69 D.1.1 Aims of the test 69 D.1.2 Material used for the test and test set-up 69 D.1.3 Parameters of the test 69 D.2 Test analysis and results 69 D.2.1 Test initial results 69 D.2.2 Conversion of recorded wavelength to temperature 70 D.3 Conclusion of the test 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 51 / 70

54

55 Appendix A Wood hygroexpansion coefficients Literature values of hygroexpansion coefficients for the three principal directions of softwoods were collected. The following table provides the values obtain as well as their originating sources. Wood species Radial [%/% of MC] Tangential [%/% of MC] Source Softwoods [56] Larix decidua [46] Larix decidua [46] Picea Abies [45, p. 6] Picea Abies [57, p. 8] Picea Abies [57, p. 8] Picea Abies [45, p. 6] Picea Abies 5-6 C [46] Picea Abies 5-6 C [46] Picea Abies Central Europe [46] Picea Abies Central Europe [46] Picea Abies Nordic [46] Picea Abies Nordic [46] Picea Abies Norway [3] Picea Abies Norway [3] Mean values Maximum values COV (%) 21% 18% Literature values of hygroexpansion coefficients for the three principal directions of beech wood were collected. The following table provides the values obtain as well as their originating sources. Wood species Radial [%/% of MC] Tangential [%/% of MC] Source Fagus sylvatica [55] Fagus sylvatica European [46] Fagus sylvatica European [46] Fagus sylvatica Presteamed [46] Fagus sylvatica Presteamed [46] Mean values Maximum values COV (%) 25% 10% The standard SN /1:2009 [51] is considering the influence of wood s MC on the dimensions of a structural element. Thus, it provides mean values of swelling/shrinkage coefficients for solid wood. These coefficients are valid for 1% of MC change under the Fibre Saturation Point (FSP) of the wood. However, the standard also says that variations from 10 to 20% of these coefficients are common. Indeed, the average sorption curves (Equilibrium Moisture Content (EMC) of the wood as a function of the air Relative Humidity (RH)) themselves show variations as displayed in Figure 34. Table C4: Swelling/shrinkage coefficients for solid wood in % per % of MC change under the FSP (extract from SN /1:2009, Table 6, 4.1[51]) α T Species α R α 90 α L (Tangential) (Radial) (Transversal)* (Longitudinal) Softwood Oak Beech * Mean value between α T and α R Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 53 / 70

56 Figure 34: MC w of solid wood (left) and of glued laminated timber (right) as a function of the air RH ϕ (SN /1:2009, Figures 1a and 1b, 4.2 [51]) 54 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

57 Appendix B Development of the strain sensor B.1 Approach 2 Development of a calculation sheet B.1.1 Principle of the sheet In order to develop the calculation sheet for the strain sensing unit, the geometry of the portal is modelled as a beam system as shown below. In this model, each beam corresponds to the neutral axis of the actual geometry of the sensing unit. The characteristics of each beam of the model (6 beams in total) can be defined by the user. The calculation sheet therefore unable to change: The geometry of the frame (relative lengths of each beam, cross section of each beam) The material used for the portal (influencing its E-modulus) As support the information provided in [58], [59] was used and adapted to this particular application. Figure 35: Beam model of the strain sensing unit B Description of the bar model used The model used in the calculation sheet is constituted of six beams which are linking seven nodes. Each beam starts and ends at a node. The complete behaviour of the beam is depending on the behaviour of its starting (node i) and ending node (node j). Therefore, each beam shows six degrees of freedom as illustrated in Figure 36. For each beam, the quantities summarized in Table 22 can be chosen freely by the user. The calculation sheet was verified, in the case of the geometry of prototype 1, using the software Cubus Statik 6. To obtain a deformation of the model and thus be able to calculate its transmission ratio, a force F0 is applied at node 7 while node 1 is restrained. The restrictions applied at node 1 (support of the system) and 7 (where the load is applied) are provided in Table 23. According to these restrictions and loads applied to the nodes, the degrees of freedom vector Inc and the loading vector F can be obtained (see Matrix 1). Finally, the displacements of each node of the portal, represented by the 21x1 vector U, are linked to the loads applied on the portal through the global stiffness matrix of the system Ksys. For each beam, the local stiffness matrix is calculated. This stiffness matrix describes the behaviour of the beam taking into account both tension/compression movements and bending movements of the beam. The stiffness matrix of a beam, expressed in its local coordinates, is given in Matrix 2 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 55 / 70

58 Figure 36: Degrees of freedom of a beam element [58] Table 22: Characteristics of the beams of the model Bar E-module [N/mm2] Length [mm] Width [mm] Thickness [mm] Same characteristics as bar L12 E1 a b1 t1 L67 L23 E2 h b2 t2 L56 L34 E3 l b3 t3 L45 Table 23: Restrictions applied to nodes 1 and 7 Node Displacement in x direction Displacement in y direction Rotation around direction z 1 (support) Restrained Restrained Restrained 7 F0 Restrained Restrained Matrix 1: Degrees of freedom vector and external loading vector for the complete portal Node u1 v1 θ1 u2 v2 θ2 u3 v3 θ3 u4 v4 θ4 u5 v5 θ5 u6 v6 θ6 u7 v7 θ7 Inc = F= F0 0 0 In F, the grey numbers correspond to restrained degrees of freedom of the nodes. Therefore these columns as deleted when solving of the system Matrix 2: Stiffness matrix Ki of a beam considering tension/compression and bending behaviour, expressed in the local coordinates of the beam With AE/L 0 0 -AE/L EI/L 3 6EI/L EI/L 3 6EI/L 2 0 6EI/L2 4EI/L 0-6EI/L2 2EI/L -AE/L 0 0 AE/L EI/L 3-6EI/L EI/L 3-6EI/L 2 0 6EI/L 2 2EI/L 0-6EI/L 2 4EI/L A [mm 2 ] E [N/mm 2 ] L [mm] I [mm 4 ] Cross sectional area of the beam E-modulus of the beam Length of the beam Modulus of inertia of the beam 56 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

59 B Calculation of the stiffness matrix of the complete system In order to express the stiffness matrix of each beam in the global coordinates of the model, the following matrices are used. First a 6x6 rotation matrix T is used to transfer the original stiffness matrix of the beam in the global coordinates. Then, the assembly matrix Ce is used. This matrix is placing the 6x6 stiffness matrix of the beams in a 21x21 empty matrix so that its position corresponds to actual place of the beam in the model. Finally, the 21x21 stiffness matrices obtained for each beam are summed and the result is multiplied by the restrain matrix Del in order to obtain Ksys, the stiffness matrix of the portal. Therefore, the Del matrix is the identity matrix (diagonal matrix with ones on the diagonal) where columns are removed. The numbers of the columns removed correspond to the ones where a zero is in the restrain vector Inc. In matrix writing, these calculations can then be written as: With (16x16) (21x16) (6x21) (6x6) (6x6) (-).... (16). Stiffness matrix of the portal, in the global coordinates Restrain matrix of the system Assembly matrix of a beam Rotation matrix of a beam Stiffness matrix of a beam in its local coordinates Transpose of matrix M Matrix 3: Rotation matrix T of a beam Lij c x c y c y c x c x c y c y c x With c x and c y calculated as such The length of the beam L ij Directing cosines of the beam in x direction Directing cosine of the beam in y direction Abscissa of the node (i or j) in the global coordinates Ordinate of the node (i or j) in the global coordinates Matrix 4: Assembly matrix Ce, example of matrix for the beam L Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 57 / 70

60 B Resolution of the system In order to calculate the displacement of each node of the portal, the Equation (16) is inversed leading to the following equation: (17) With (16x1) (16x16) (16x1) Unknown displacements of each node in the global coordinates (degrees of freedom which are restrained are not considered here) The stiffness matrix of the portal, in the global coordinates The external loading vector each node, in the global coordinates According to [58], the strain of each point s of a beam is linked to the displacements of its nodes by the matrix B: B = -1/L 6/L 2-12s/L 3 4/L-6s/L 2 1/L -6/L 2 +12s/L 3 2/L-6s/L 2 With L The length of the beam considered s The relative abscissa of the point considered on the beam (between 0 and 1) To obtain the total strain of one beam, the strains are integrated for s between 0 and 1. Therefore, the strains of beam L34 and L45 can be obtained, added together; they correspond to the strain of the top part of the portal. The strain of the base of the portal corresponds to the displacement of the node 7 in x direction divided by the initial length of the base of the portal. Knowing the strains of the base and of the top port of the portal, the transmission ratio of the portal can be obtained: (18) With Transmission ratio of the portal Strain of the base of the portal Strain of the top of the portal (where the FBG fibre is glued) 58 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

61 B.1.2 Matlab code of the sheet clear; clc; initgraph=1; fingraph=1; coeff=5; %PARAMETERS TO DEFINE E1=3000; E2=3000; E3=3000; a=7.25; h=14.75; l=48/2; b1=8; b2=2; b3=2.5; t1=2; t2=2; t3=2; % Definition of the beams characteristics E=[E1 E2 E3 E3 E2 E1]; L=[a h l l h a]; b=[b1 b2 b3 b3 b2 b1]; t=[t1 t2 t3 t3 t2 t1]; %Loading characteristics %1 shows the final geometry otherwise 0 %1 shows the final geometry otherwise 0 %amplification coefficient for the graphic display %Beams E modulus [N/mm2] %Beams length [mm] %Beams width [mm] %Beams thickness [mm] %Load applied on the horizontal axis on the beam F0=1; % Geometric characteristics x=[0 L(1) L(1) L(1)+L(3) L(1)+L(3)+L(4) L(1)+L(3)+L(4) L(1)+L(3)+L(4)+L(6)]; y=[0 0 L(2) L(2) L(2) 0 0]; %Abscissas and ordinates of the nodes I=t.*b.^3/12; A=t.*b; N=7; %Definition of the no. of unknowns at each node Inc=[ ]; %Definition of the restrain matrix m=sum(sum(inc)); Del=zeros(3*N,m); j=0; for i=1:3*n j=j+inc(i); if Inc(i)==1 Del(i,j)=1; end end %% GRAPHICAL VIEW OF THE INITIAL GEOMETRY if initgraph == 1 plot(x,y,'red'); xlim([-l(1) 3*L(1)+2*L(3)]); ylim([-l(2)/2 1.5*L(2)]); legend('initial geometry of the sensor'); end %Beams inertias [mm4] %Beams sections [mm2] %Number of nodes %% CALCULATION OF THE STIFFNESS MATRIX OF THE SYSTEM K=zeros(3*N,3*N); for i=1:n-1 %Elementary matrices in the local axes Ki=E(i)*I(i)/L(i)^3*[A(i)*L(i)^2/I(i) 0 0 -A(i)*L(i)^2/I(i) 0 0; *L(i) *L(i); 0 6*L(i) 4*L(i)^2 0-6*L(i) 2*L(i)^2; -A(i)*L(i)^2/I(i) 0 0 A(i)*L(i)^2/I(i) 0 0; *L(i) *L(i); 0 6*L(i) 2*L(i)^2 0-6*L(i) 4*L(i)^2]; %transformation into the global coordinates Lb=sqrt((y(i+1)-y(i))^2+(x(i+1)-x(i))^2); %distance between node i and i+1 (bar i) cy=(y(i+1)-y(i))/lb; cx=(x(i+1)-x(i))/lb; T=[cx cy ; -cy cx ; ; cx cy 0; cy cx 0; ]; Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 59 / 70

62 End %Transformation matrix from local to global coordinates for the bar i Ke=T'*Ki*T; %Elementary matrix of bar i in the global coordinates %location matrix Ce=zeros(6,3*N); Ce(1,3*(i-1)+1)=1; Ce(2,3*(i-1)+2)=1; Ce(3,3*(i-1)+3)=1; Ce(4,3*i+1)=1; Ce(5,3*i+2)=1; Ce(6,3*i+3)=1; %Final elementary matrix K=K+Ce'*Ke*Ce; %Force vector F=zeros(1,3*N); F(1,19)=F0; F=(F*Del) ; %% RESOLUTION OF THE SYSTEM Ksys=Del'*K*Del; U=inv(Ksys)*F; U %Matrix of all displacements %% GRAPHICAL VIEW OF THE FINAL GEOMETRY if fingraph==1 hold on xf=x+[0 U(1) U(4) U(7) U(10) U(13) U(16)]*coeff; yf=y+[0 U(2) U(5) U(8) U(11) U(14) 0]*coeff; plot(xf,yf,'blue'); xlim([-l(1) L(1)+xf(7)]); ylim([-l(2)/2 1.5*L(2)]); legend('initial geometry', 'final geometry'); str = sprintf('amplitude coefficient : %d',coeff); xlabel(str); hold off end %% CALCULATION OF THE TRANSMISSION RATIO OF THE FRAME u3=[u(4) U(5) U(6) U(7) U(8) U(9)]; %Displacements of nodes 3 and 4 u4=[u(7) U(8) U(9) U(10) U(11) U(12)]; %Displacements of nodes 4 and 5 %Case of beam 3 i=3; B=[-1/L(i) 6/L(i)^2-6/L(i)^3 4/L(i)-3/L(i)^2 1/L(i) -6/L(i)^2+6/L(i)^3 2/L(i)-3/L(i)^2]; eps(i)=b*u3'; %Case of beam 4 i=4; B=[-1/L(i) 6/L(i)^2-6/L(i)^3 4/L(i)-3/L(i)^2 1/L(i) -6/L(i)^2+6/L(i)^3 2/L(i)-3/L(i)^2]; eps(i)=b*u4'; %Calculation of the transmission ratio of the portal epstop=abs(eps(3))+abs(eps(4)); epsbase=abs(u(16))/(l(1)+l(3)+l(4)+l(6)); TR=epsbase/epstop %Transmission ratio Figure 37 : Code developed in Matlab environment to calculate the behaviour of the strain sensing unit 60 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

63 B.2 Investigations over the stiffness of the strain sensor B.2.1 K 2 Sensor s maximum stiffness to prevent cracks in the wood B Calculation This calculation is based on the local behaviour of the timber element as shown in Figure 38. The area under consideration corresponds to the volume of wood present between the two fixings of the strain sensor and up to a depth of t. In this small area, the wood goes from a state 0 to a state 1 with an increase of its moisture content as shown in Figure 39. Figure 38: Area where the calculation is conducted Figure 39: Initial and final state of the area under consideration with an increase of moisture content of the wood Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 61 / 70

64 The following assumptions are made: A 1: The swelling stresses decrease to reach the value 0 over the distance L A 2: The decrease of the stress in the x direction is described by a function f o 2a: f is linear, 1 o 2b: f is parabolic of form ax 2 +b then, 1 A 3: Only linear elastic and swelling strains are considered, all others be neglected In the wood: According to A3, the total strain of the wood volume considered is: At the point P, we have:,0, 0 With ε 0 =0; σ w,0 (0) =0 and σ w,1 (0) = -σ max (compression) Thus In the sensor: The sensor is fixed to the wood, thus it also elongates by the distance L. Then, the stiffness of the sensor K is defined by: 62 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

65 With Eventually L therefore, The detail of the calculation of F and K are provided in Appendix A. A 4: the transverse direction is neglected thus Ka and Kb found correspond to the actual stiffness of the sensor in N/mm Numerical application: The following numerical values are used: Table 24: Numerical values used in the calculation Variable Value Unit Reference L 60 mm Length of the prototype t 60 mm Depth of influence of the sensor s restrain E 370 N/mm 2 Cross sectional modulus of wood according to the EN338:2009 α 0.35 %/% Hygroexpansion coefficient taken in the tangential direction for this application (see report WP2) u 18 % Maximum moisture change expected (see report WP2) σ max = f t,90,m 1.2 N/mm 2 Average tension strength of wood perpendicular to the grain Thus the following maximum stiffness are obtained: Table 25: Calculated maximum stiffness for the prototype Case a, linear Case b, parabolic K 2 [N/mm] B.2.2 Comparison with the characteristics of the prototype B Stiffness of the prototype Using a hand Newton meter (maximum value 5 N), the stiffness of the prototype was calculated. Table 26: Stiffness of the prototype Measure Displacement Force Calculated stiffness [mm] [N] [N/mm] Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 63 / 70

66 As a result, the actual stiffness of the prototype is between 4.0 and 4.5 N/mm. This range is lower than both results from paragraph B.2. Thus the stiffness of the prototype is low enough to ensure the wood will not cracks while swelling. B.2.3 Minimum swelling stress The minimum stress needed to correspond to the stiffness of the prototype is calculated. The following values are found: Table 27: Minimum swelling stresses required Prototype stiffness Minimum stress in the wood [N/mm 2 ] [N/mm] Case a (linear decrease) Case b (parabolic decrease) B.2.4 Discussion and conclusion over K 2 The numerical values used in this report are subject to uncertainties. Indeed, the distribution of the stresses within the depth of the wood as well as its maximum values σ max are unclear in the literature. Here two assumptions have been studied: a linear and a parabolic decrease both over a distance of L. Eventually, we can see that even in the worst case: - Prototype stiffness of 4.5 N/mm - Linear decrease of the stress in the wood s depth - The minimum swelling required to deform the sensor is of 0.55 N/mm 2 which lower than the cracking strength of wood perpendicular to the grain (1.2 N/mm 2 ). - Therefore, there is no risk that the strain sensor s prototype of stiffness 4.0 to 4.5 N/mm 2 produces cracks in the wood when monitoring climate changes up to u = 18%. B.2.5 K 1 - Target sensor stiffness with little influence on the swelling behaviour The aim of this calculation is to provide a target sensor stiffness. This target sensor stiffness should influence little the swelling behaviour of wood. In order to quantify this influence we decide to compare the swelling observed on the wood piece restrained by the sensor with the swelling of the nonrestrained wood piece. 64 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

67 B Calculation B.2.6 Discussion and conclusion over K1 According to the calculation, the stiffness of the prototype (4.0 to 4.5 N/mm) corresponds to an influence of 3% to 4% on the swelling behaviour. Thus with a sensor of such stiffness, the strain measure by the sensor would correspond to 96% to 97% of the free strain obtained on a piece of wood without sensor. Such range of influence is acceptable. B.2.7 K 3 Minimum sensor stiffness related to the strength of the glass fibre The minimum stiffness of the sensor is calculate by using the deign stiffness (4.5N/mm) over the design amplification factor (5), resulting in a minimum stiffness of 0.9 N/mm. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 65 / 70

68 Appendix C Specifications of the FBG relative humidity and temperature sensors C.1 The relative humidity sensor (company: O/E-Land Inc.) The specifications of the relative humidity FBG sensor provided by the company O/E-Land Inc. are shown below. The first table corresponds to the information related to this type of sensors in general while the figure shows the specifications corresponding to the sensor acquired. Parameter Sensor OEFHS-100A Range (RH [%]) Sensitivity [pm/% of RH] 4.5 Central Wavelength [nm] 1550 Size (diameter x length) [mm] 3 x40 Temperature sensor N.A. Fibre output Double Fibre cable 3 mm Operation temperature [ C] 0 ~ / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

69 C.2 Specifications of the temperature sensor (company: FiberSensing) The specifications of the temperature FBG sensor provided by the company FibreSensing are shown below. The first table corresponds to the information related to this type of sensors in general while the figure shows the specifications corresponding to the sensor acquired. Sensor Sensitivity (typical values) 10 pm/ºc Measurement range -20 to 80 ºC Resolution (for 1pm resolution in wavelength measurement) 0.1 ºC Maximum calib. error 0.5 ºC Optical Central wavelength 1500 to 1600 nm Spectral width (FWHM) < 0.2 nm Reflectivity > 65% Side lobe suppression > 10 db Inputs / Outputs Cable type Ø 3 mm indoor (kevlar) Ø 3 mm outdoor (armor) Cable length 2 m each side (±5 cm) Connectors FC/APC SC/APC NC (No Connectors) Environmental Operation temperature -20 to 80 ºC Protection class embedded IP68 Mechanical Materials Dimensions Weight stainless steel weldable 45 x 15 x 0.6 mm weldable 5 g Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 67 / 70

70 68 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

71 Appendix D Test C Testing of the temperature FBG sensor controlled climate D.1 Test preparation D.1.1 Aims of the test Proof the required sensitivities and measurement accuracy for the FBG temperature sensor D.1.2 Material used for the test and test set-up Two FBG temperature sensors were used in the experiments along with Ecolog temperature and relative humidity measurements. A climate chamber was used to impose temperature changes in the range of -10 C to 40 C in 50 %RH. Although the relative humidity was measured by the Ecolog, there is no relevance for the result of the test. D.1.3 Parameters of the test The following material was used for test C. Material Description/Reference Datalogger Ecolog TH FBG sensor interrogator FS2200, , v.1.4, Temperature FBG sensor 1 and 2 From the company FiberSensing Climate chamber Feutron KPK400 The time intervals used to record the temperature and the RH of the air through the datalogger Ecolog TH1 on the one hand and the ones used to record the evolution of the wavelength of both fibre sensors are the following. Quantity Measurement device Rate of recording Number of measurements for each record Temperature Datalogger 3 minutes 1 Temperature FBG temperature sensor and FBG interrogator 1 Hz 1 The tests were conducted over a two day period. On the first day, a stepwise increase in temperature was applied from -10 C to 40 C, with increments of 10 C each hour. Overnight, the temperature in the climate chamber was left at 40 C. The next day, the temperature was decreased stepwise to 30 C and 20 C, after which two other decrease steps were set to 5 C and -5 C. The latter was done due to difficulties of the climate chamber to maintain a correct temperature and 50 % RH at 0 C. It was expected that this could be better if the temperature was kept either just above the freezing point of right under it. The two temperature sensors were hung in the climate chamber, along with the Ecolog sensor. It must be noted that neither of the sensors was placed really close to the other, each at a distance of about 20cm from any other sensor. D.2 Test analysis and results D.2.1 Test initial results The evolution of the air temperature in the climate chamber is shown in Figure 40 (left). The figure shows the temperature measured by the Ecolog, the FBGS and the settings of the climate chamber over the two day testing period. The Figure 40 (right) shows the comparison of the measured temperature by the two FBGS sensors with the temperature measured by the Ecolog. Measurements around the large temperature jumps were removed from this graph as they are not expected to be representative for outside monitoring conditions. Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences 69 / 70

72 Figure 40: Time trace of measured temperature (right) and comparison of temperature measured with the FBGS and the Ecolog. A linear fit of the FBG temperature sensors and the Ecolog data was calculated. The FBG sensor 1 showed an error of 0.24% and the FBG Sensor 2 showed an error of -0.04%. Theoretically, the sensitivity of the sensor is 1.05E-4 C around 30 C. This is calculated by using the precision of the logged values the FBG prints in the output file. The last digit of the printed value is given the smallest increase possible. By using the conversion equation from wavelength shift to temperature, the minimum precision of recorded temperature can be calculated. The theoretical value is also found in the measured values on an interval of 1 second. The minimum value on the three minute interval was 1.05E-4 C. This was observed by calculating the minimum change in measured temperatures from the printed values in the measurement data. Regardless of the time interval, the observed sensitivity meets the requirement of minimum sensitivity of 0.1 C set in section 3.5. D.2.2 Conversion of recorded wavelength to temperature The conversion of the raw data to the temperature values is done through 4 parameters. First the reference wavelength is deduced from the measured wavelength to calculate the wavelength shift. This should result in a zero value at a temperature of 30 C. Then a second order conversion equation is given to convert the wavelength shift to a temperature., (19) With T [ C] [nm] [ C] Temperature Wavelength shift Calibration values manufacturer Reference temperature (30 C) D.3 Conclusion of the test This test showed that the measurements with the FBG temperature sensors can be used to accurately measure the temperature and that the requirements regarding sensitivity are met, see Section 3.5. The measured sensitivity of the sensor on a three minute interval was 0.4E-3 C, meeting the requirement of 0.1 C. The measurement error of the FBG sensors was minimal, within 0.25% over the range of -10 C to 40 C. 70 / 70 Berner Fachhochschule Haute école spécialisée bernoise Bern University of Applied Sciences

73

74 Berner Fachhochschule Architektur, Holz und Bau Solothurnstrasse 102 Postfach 6096 CH-2500 Biel 6 Telefon Telefax fe.ahb@bfh.ch ahb.bfh.ch Fördergeber Berner Fachhochschule Forschungskommision c/o Architektur, Holz und Bau Solothurnstrasse 102 Postfach 6096 CH-2500 Biel 6