Study of Sensor Network Applications in Building Construction

(1)

Study of Sensor Network Applications in

Building Construction

XIAOTAO WANG

(2)

(3)

Study of Sensor Network Applications

in Building Construction

XIAOTAO WANG

Supervisor: George Athanasiou and Carlo Fischione

Examiner: Carlo Fischione

Stockholm December 20, 2013

Automatic Control Department

School of Electrical Engineering

(4)

(5)

Abstract

This thesis presents a study of wireless sensor networks applied in monitoring the curing process of concrete building structure. To replace the massive user interven-tion of the tradiinterven-tional monitoring methods with automated data acquisition, various types of sensors are developed for different concrete tests, including compressive strength measurement, maturity test, and relative humidity test. The main purpose of the thesis is to provide an analy-sis of existing commercial sensors as well as research pro-totypes, comparing different test methods and communi-cation models. Distribution of sensors, future potential in millimeter-wave, and simulation environments are also discussed.

(6)

The author would like to express gratitude to Dr. George Athanasiou and Professor Carlo Fischione, for their intro-duction, thorough guidance and continuous encourage-ment to this project. They have helped generously in pro-viding adequate resource and patience, without them the project could not have been accomplished.

(7)

List of Figures

1.1 Chapter 2 Outline . . . 3

1.2 Chapter 3 Outline . . . 3

2.1 Range of typical strength to water-cement ratio relationships of Port-land Cement concrete based on over 100 concrete mixtures tested be-tween 1985 and 1999 [1]. . . 5

2.2 Pullout test [2] . . . 6

2.3 Schematic of the pulse velocity circuit [2]. . . 7

2.4 Schematic for temperature history and TTF computed with Eq. 2.1 [3]. . 9

2.5 Schematic of procedures for maturity method including lab test and filed measurement [3]. . . 10

3.1 Typical functional blocks of a wireless sensor system. . . 16

3.2 Piezoelectric patch embedded into concrete blocked smart aggregate. . . 17

3.3 Double piezoelectric transducers: experimental set-up . . . 17

3.4 Compressive strength versus age [4]. . . 18

3.5 The average value of the amplitude of different harmonic excitations [4]. 19 3.6 Comparison of the experimental training data with the fuzzy mapping data [4]. . . 20

3.7 Comparison of the trial compressive strength with the estimated com-pressive strength [4]. . . 20

3.8 Protection package of the PZT sensor [5]. . . 21

3.9 Basic experiment set-up of EMI method using PZT. . . 22

3.10 Comparison of compressive strength development with anti-resonant frequency development using AD5933 and HP4192A [5]. . . 23

3.11 Development of a antiresonance, b peak resistance as load is applied [5]. 23 3.12 SHT11 [6]. . . 24

3.13 SHT11: Change of accuracy under different RH [6]. . . 25

3.14 SHT11: Change of accuracy under different temperature [6]. . . 25

3.15 Pin configuration and size dimension of iButton. . . 26

3.16 Safe operation range of iButton [7]. . . 26

3.17 CC2451 SensorTag Schematic [8]. . . 27

3.18 CC2451 SensorTag content and package [8]. . . 28

(10)

3.22 Location of sensors inside concrete [9] . . . 31

3.23 Results of RH over time [9] . . . 31

3.24 Results of PDR over time [9] . . . 32

3.25 COMMAND Center maturity meter [10]. . . 33

3.26 Maturity Data Loggers [11]. . . 35

3.27 IntelliRock Wireless . . . 36

3.28 Functional block diagram [12]. . . 38

3.29 RF power transmission set-up. . . 39

3.30 Schematic of piezoelectric-based battery charing circuit [13]. . . 40

3.31 Functional block of MASSpatch [14]. . . 41

3.32 State Sensor . . . 42

3.33 Sketch of crack-detector [15]. . . 43

3.34 Inductive coupling from external reader. . . 43

3.35 Experimental setup for measuring the permittivity of concrete [16]. . . . 44

3.36 Diagram of the large coaxial closed cell [16]. . . 44

3.37 Real part of the permittivity over frequency [16]. . . 45

3.38 Imaginary part of the permittivity over frequency [16]. . . 45

3.39 Cement composition. . . 49

3.40 Mixture proportions. . . 50

3.41 Evaluation of the properties and characteristics of the virtual microstruc-ture during and after hydration. . . 50

3.42 Hydration process control. . . 51

3.43 Temperature development over time (hour). . . 51

3.44 Hydration degree over time (hour). . . 52

3.45 Virtual 3D framework . . . 52

3.46 Work flow of CEMHY3D. . . 53

(11)

Chapter 1

Introduction

1.1 Background: Wireless Sensor Network and Concrete

Monitoring

A wireless sensor network (WSN) is a network consists of devices connected re-motely. These devices can communicate with each other and with a base station, making it possible to transfer their data to remote computing systems. These de-vices are equipped with reduced communication, computation and sensing ca-pabilities, and often applied in communication, control and monitoring purposes [17].

An important application of WSN that attracts the interest of the construction companies is the automatic non-destructive measurement in the curing process of concrete slabs or pre-cast production. The curing of the concrete is one of the most critical elements within the building process. A system reliably delivers the status of the curing can greatly optimise the work-flows and reduce the failure rates. It is worth to investigate new WSN technologies to ease the monitoring process by signalling the status of the curing [18]. Critical test parameters include compressive strength, maturity and relative humidity [2].

1.2 Problems in Traditional Measurements

(12)

1.3 Target

By discussion over different research prototypes and commercial products, this thesis will target at the solutions of the following problems:

• Parameters to be measured: strength, maturity, or relative humidity;

• Performance features of sensors: active or passive sensing, power, size, cost, etc.;

• Communication models of sensor network within curing concrete, the work-ing frequency and distribution;

• Protection package of sensors against hazard environment inside concrete.

1.4 Research Method

The main research methods of this thesis work consist of literature review and product assessment. The author performed a deductive investigation and an in-terpretive analysis of existed sensor products and prototypes based on qualita-tive classification of different measuring targets, sensing types, power sources and communication models, etc. A qualitative comparison of critical parameters of the products will also be summarized in the end.

1.5 Delimitations

The initial project plan consisted of building an innovative experiment of wireless sensor network to monitor curing concrete. However, the author had not esti-mated the manufacture time of ad-hoc sensing units like Tyndall Mote [19] could be as long as two months and the price of available commercial maturity meters could be as high as 2500 USD. In addition, it takes at least 28 days for concrete to cure [1]. The time and budget limitations have led to a major alternation in the outcome: instead of an experiment, an analytical review of existed approaches will be presented.

Due to the complexity of concrete, only a handful of academic prototypes have been tested in monitoring the curing process. In order to provide a comprehensive overview, measurements related to structure health monitoring (SHM) of cured concrete are also presented as references, which have the potential of future adap-tation into monitoring the curing process.

1.6 Outline

(13)

1.6. OUTLINE

conclusion and a vision to the future research. The structures of Chapter 2 and 3 are shown in Figures 1.1 and 1.2.

Theoretic Background Concrete Test Targets Compressive Strength Electromechanical Impedance Method Maturity Method

Others Relative Humidity

Communication Model Electromagnetic Models Sensor Network Frequency Sensor Network Distribution

Figure 1.1.Chapter 2 Outline

Analysis of Existed Solutions Lab Test Sensor Categories Piezoelectric Component Humidity Sensor Maturity Meter Power Active/Passive Sensing & Power

(14)

Theoretic Background

2.1 Testing Methodologies for Curing Concrete

Concrete is a highly heterogeneous material. After the cement is mixed with wa-ter and various aggregates, the process of hydration and strength gain takes place. Generally it takes 28 days to achieve the maximum resistance of a concrete spec-imen and 90 days to achieve the long-term steady state [1]. Different water to cement ratios and types of aggregates will lead to complex changes in final gained strength. In addition, the durability of concrete will be degraded from environ-mental factors. Therefore other than destructive lab tests on concrete specimens, it is essential to monitor the in-place status with non-destructive tests. Traditionally it requires excessive experience and labour hours to monitor the in-place curing process and decide when is appropriate to proceed the next stage in construction.

Non-destructive concrete tests can roughly be divided into two groups. The first consists of tests measuring compressive strength, whether directly or indi-rectly. Those measuring other parameters, such as humidity, density and thick-ness, belong to the second group [2]. The products and prototypes in this thesis will mainly be grouped into different measurements accordingly.

2.1.1 Compressive Strength Measurement

Compressive strength of concrete is determined by tests on cylinder specimens. Axial loading is added to a specimen of a certain size. The reaction is evaluated in megapascals (MPa) at the age of 28 days. According to a test conducted by Port-land Cement Association from 1985 to 1999, if the specimens were cured in moist with the water-cement ratio ranging from 0.25 to 0.85, their 28-day compressive strength would be between 20 to 40 MPa. Figure 2.1 illustrates the result [1].

Meanwhile, compressive strength can be indirectly estimated via other mea-surements, such as the electromechanical characteristics, the maturity factor and resonant frequency, etc.

(15)

2.1. TESTING METHODOLOGIES FOR CURING CONCRETE

28-days strength of moist cured cylinders

Co mpressive Stre ngth / MPa 80 70 60 50 40 30 20 10 0 0.25 0.35 0.45 0.55 0.65 0.75 0.85 Water-cement Ratio

Figure 2.1. Range of typical strength to water-cement ratio relationships of Portland Cement concrete based on over 100 concrete mixtures tested between 1985 and 1999 [1].

magnetic / electrical methods, and last but not the least, the maturity method. Surface Hardness Methods Mainly two types of surface hardness methods are used to measure the increasing hardness of concrete. One is the indentation type. By using a given mass with a kinetic energy, the width and depth of the resulting indentation is measured. The other type is based on the rebound principle. After impacting the concrete with a spring-driven hammer mass, the rebound of the hammer is tested.

A major limitation of the rebound method lies in that the peak rebound value often takes place around the 7th day of curing, while the compressive strength gain usually grows beyond 7 days. Besides, the smoothness of surface has a significant impact. On the same piece of concrete sample, a rough surface reduces the rebound value by 5% to 25% [2].

(16)

mate-rial. During the construction, such relationship can be used to estimate the in-place compressive strength based on the sampled pull-out force [20]. Figure 2.2 gives out an schematic explanation [2].

Pullout test damages the concrete surface, which must be repaired later.

h Reaction D _Reaction Pullout Force Idealized Fracture Surface Insert Head d α Insert Shaft

Figure 2.2.Pullout test [2]

Ultrasonic Pulse Velocity Test This method is literally non-destructive as using mechanical waves brings no physical damage to concrete. A wave generator and transmitter introduces mechanical wave pulses into the concrete. At the receiver end, the arrival time of penetrating pulses is measured, which leads to the pulse velocity.

(17)

Transmiing

Transducer TransducerReceiving

Pulse Generator Time-Display Unit Time Measuring Circuit Receiver Ampliﬁer Optional Display

Figure 2.3.Schematic of the pulse velocity circuit [2].

(18)

2.1.2 Maturity Method

The maturity method is a technique that estimates in-place strength-gain of con-crete based on the measured temperature history of the same material during cur-ing. The maturity function expresses the combined effects of time and temperature. 2.1.2.1 History

The maturity method first came to recognition in 1949, when McIntosh introduced a new approach using the product of time and temperature above a datum temper-ature (-1.1◦C or 30◦F) to estimate the strength development of in place curing con-crete. Soon after this Nurse proposed the first evidence showing that the product of time and temperature could be used to present their effects on strength devel-opment, although his calculation did not involve a datum temperature or in-place temperature, but temperature in lab chambers. In 1951, Saul concluded his re-search in England and for the first time linked the term maturity to the product of time and temperature. He presented the notion later known as the "maturity rule" [22]:

Concrete of the same mix at the same maturity (reckoned in temperature-time) has approximately the same strength whatever combination of temperature and time go to make up that maturity.

2.1.2.2 Maturity Function

Saul came up with the following Equation 2.1 that computes maturity from the temperature history. Later known as the Nurse-Saul function, this is still in wide usage today [2]: M = t X 0 (T − T0)∆t (2.1) where

M =maturity (temperature-time factor, TTF) at age t

T =average temperature of the concrete during time interval ∆t

T0=datum temperature, generally -10◦C (14◦F)

As shown in Figure 2.4, the value M stands for the temperature history at a certain point of time t, which is represented by the shadowed area [3].

The Nurse-Saul function can be turned into equivalent age (te). According to the definition in ASTM C 1074, testands for the time period at a specific reference temperature (Tr) required to produce the maturity equal to that achieved by a cur-ing period at temperatures different from Tr. It is expressed as the following [2]:

te =

P

(T − T0)

Tr− T0

(19)

2.1. TESTING METHODOLOGIES FOR CURING CONCRETE Concrete Temperature T₀ Time,t t* ,T

Figure 2.4.Schematic for temperature history and TTF computed with Eq. 2.1 [3].

Later, Freiesleben Hansen and Pedersen altered the equivalent age equation using the Arrhenius equation [23]:

te = t X 0 e −E R [ 1 T− 1 Tr]_∆t (2.3) where

te=equivalent age at the reference temperature

E =apparent activation energy, J/mol

R =universal gas constant, 8.314 J/mol-K

T =average absolute temperature of the concrete during interval ∆t, Kelvin

Tr=absolute reference temperature, Kelvin And the value of E is defined as the following:

E =

(

33500J/mol for T ≥ 293◦F,

33500 + 1470(20 − T )J/mol for T < 293◦F

2.1.2.3 Strength-Maturity Relationships

The general theory of strength maturity relationship was explained by Carino [2] and is paraphrased as the following.

The initial rate of strength development is dS

(20)

S =concrete strength

S∞=limiting strength at infinite age

k(T ) =rate constant that affects the initial rate of strength development The general form of maturity function can be denoted as M (t, T ):

M (t, T ) =

Z t

t0

k(T ) dt (2.5) A general strength-maturity relationship is expressed as below:

S = S∞

M (t, T )

1 + M (t, T ) (2.6)

2.1.2.4 Practice

The practice of maturity method consists of two steps which can be depicted as Figure 2.5: [3]

1. Lab test

2. Field measurement of in-place temperature history

M*

Lab Test Field Measurement Cube Tests Cylinder Tests T1 T2 T3 Strength S* M* 5.6 k T Maturity Meter Structures Maturity Index

Figure 2.5. Schematic of procedures for maturity method including lab test and filed measurement [3].

(21)

the established strength-maturity relationship can be use to estimate the strength. Nowadays a convenient approach is to use the commercial maturity meters, which will be described in detail in Chapter 3. Maturity meters monitor the in-place tem-perature history automatically and analyze the data with strength-maturity rela-tionship and maturity functions as Eq. 2.1 and/or 2.3.

2.1.3 Relative Humidity Test

The water vapor in air or gas is humidity. The mass of water vapor versus the vol-ume of air or gas is referred to as Absolute Humidity, often expressed in grams per cubic meter, while at a given temperature and pressure, the ratio of partial pres-sure of water vapor in an water-air mixture to the saturated water vapor prespres-sure is called Relative Humidity (RH), often stated in percentage.

The relative humidity of concrete often serves as an important empirical mea-surement in building structures. For instance, the figure of 75% relative humidity is commonly considered as a safe point to undertake the flooring of a concrete slab [24].

According to studies of Parrot and reviewed by Ye et. al [25], the evolution of normalized relative humidity in concrete can be described as the model below.

f (r) = r − ra

100 − ra

= e−kT (2.7)

where

r =relative humidity

ra=ambient relative humidity

k =0.8-0.14T+0.01T2

T = t/t1/2, normalized drying time

t =drying time

t1/2 =the time to reach 79% RH, i.e. the half potential of RH change

The t1/2 is changed by the cement binder materials, such as Pulverized fuel ash(PFA), Ground granulated blast-furnace slag (GGBS) and only Portland cement (OPC). It can be represented as the following:

t1/2=

(

10jd for t1/2 < 414days, 3jd + 290 for t1/2 ≥ 414 days where

(22)

2.2 Communication model

2.2.1 Electromagnetic Models of Concrete

To evaluate concrete conditions with the discussed applications, it demands under-standing of electrical properties of concrete, such as permittivity and conductivity, for that they have essential impacts on the propagation of electromagnetic waves. Take the EMI method for instance, the change of electric permittivity will affect the result in impedance calculation [26].

Concrete and other building materials are dielectric and behave as insulators as well as conductors of electromagnetic waves. Such material has complex dielec-tric permittivity consists of a real part and an imaginary part. The real part deals with the phase velocity, and the imaginary part relates to the conductivity or at-tenuation. The complex dielectric permittivity with both real and imaginary parts is usually expressed in a dimensionless form ε (relative complex dielectric permit-tivity), divided by the dielectric permittivity of vacuum ε0which consists the real part only (= 8.854 × 10−12F/m) [27].

At a given temperature, salinity, porosity and saturation of the pores, the com-plex dielectric permittivity of concrete is frequency dependent. Robert et al. [16] compared five theoretical models of the concrete permittivity with their experi-mental results and hereby summarized as the following. In the first three models, concrete is considered as a homogeneous material, while in the last two models it is treated as a heterogeneous material.

For homogeneous materials, the complex dielectric permittivity varies as a func-tion of frequency:

Cole-Cole This mechanism is explained with a Maxwell-Wagner effect enhanced by platy insulating particles.

ε(ω) = ε∞+ εS− ε∞ 1 + (iωτ )1−α − i σ(0) ωε0 (2.8) Double Layer Polarization This mechanism is depicted as a polarization of dou-ble layer in the water near charged particles.

ε(ω) = ε∞+ εS− ε∞ 1 + (2iωτ )β_{+ iωτ} − i σ(0) ωε0 (2.9) Debye Effect This model deals with a binding effect of water to solid surfaces so as to lower its dipolar relaxation frequency.

(23)

2.2. COMMUNICATION MODEL

ε(ω) = ε0(ω) − iε00(ω): relative complex permittivity,

εS: permittivity at low frequency,

ε∞: permittivity at high frequency,

ω = 2πf : pulsation [Hz],

τ : relaxation time [s],

i =√−1, 0 ≤ α, β ≤ 1,

σ(0): DC conductivity [S/m].

For heterogeneous materials, the mixed constituents shall be taken into consid-eration :

Complex Refractive Index Method (CRIM) Refractive index (RI) ηtfor a medium is defined as

ηt= c0 √

µtε0ε (2.11)

where

c0 = 3 × 108m/sec: velocity of electromagnetic waves in vacuum,

µt=permeability of medium = µ0 = 4π × 10−7henry/mfor vacuum and most dielectric materials.

CRIM asserts that for a certain mixture, the effective complex RI is given by the volume average of the complex RIs of its constituents:

√ εmix= (1 − Φ) √ εmat+ SΦ √ εw+ (1 − S)Φ √ εa (2.12) where

εmix: permittivity of the mixture,

Φ =(volume of voids)/(total volume of concrete) = porosity,

S =(volume of water)/(volume of voids): degree of saturation,

εmat≈ 5.0 : relative dielectric permittivity of concrete solids,

εw : permittivity of water, calculated using Debye effect and considering the DC conductivity of the electrolyte[28].

To simplify the computation, assume the sample material is fully saturated, therefore S = 1, the permittivity is given by:

√ εmix= (1 − Φ) √ εmat+ Φ √ εw (2.13)

Self Similar Model (SSC) The last model was proposed by Sen et al. [29] to avoid the percolation threshold inherent to the effective medium theory and ignored on experimental data. The SSC model is expressed as the following:

(24)

2.2.2 Transmission Frequency

Popular ISM Band For RFID-based sensors, typical frequencies of 433 MHz, 915 MHz and 2.4 GHz are most commonly used in commercial products. The trade-off between antenna efficiency and power consumption plays an important part in frequency selection. The ISM band of 433 MHz and 915 MHz are sometimes pre-ferred because they are not bound to specific standards like 2.4 GHz, leaving more room to power-saving implementation [30].

Millimetre Wave It is worth mentioning that the millimeter-wave band, espe-cially the spectrum at 60 GHz, is attracting heated investigation in high-bandwidth commercial wireless communication systems [31]. This spectrum is unlicensed and plentiful. In addition, the wide bandwidth of more than 5 GHz, the high data rate of 1 Gbps and more freedom in terms of power limits [32] contribute to its po-tential. Yet challenges exist. Severe attenuation, difficulties in compact antenna design and the lack of support from CMOS foundries are among the obstacles at the moment.

2.2.3 Distribution of Sensor Network

(25)

Chapter 3

Analysis of Existed Solutions

A typical wireless sensing system consists of the sensing interface, the computing core, the actuation interface and the layer of wireless radio. Figure 3.1 reveals the functional blocks based on the summary of Lynch and Loh [33]. For wired systems, the radio block can be omitted.

The sensing interface is responsible for converting analogue input from sensor into digital data that can be processed by the computing core. The sample rate, res-olution and number of channels available on its analog-to-digital converter (ADC) have great impact on the performances of the sensing interface. After that, the digitalized data is processed at the computation core to be stored, analyzed and transferred. The wireless radio consists of components like wireless transceivers to communicate with the base station or other nodes within the WSN. The actu-ation interface converts the digital output of the computactu-ation core into analogue signals, helping the sensor network to interact directly with the surrounding phys-ical system [33]. Last but not the least, omitted in this figure is the power source energizing the whole system, with batteries as a popular option, and a few power harvesting techniques have been studied

(26)

ADC to convert analog sensor signals to digital

formats

Transmit and receive data with

other wireless sensors and data

servers Microcontrollers and memory for

sensor data DAC to command active sensors and actuators with analog signals Computing Core

Sensing Interface Actuation Interface

Wireless Radio

Figure 3.1.Typical functional blocks of a wireless sensor system.

3.1 Sensing Interfaces for Different Measurements

In this section, a variety of applications are discussed in groups of sensing inter-faces for different measurements, namely the compressive strength test, the rela-tive humidity test and the maturity method.

3.1.1 Compressive Strength Test

To monitor compressive strength of concrete, piezoelectric-based systems are often in use. Piezoelectric materials can generate electric fields in response to applied mechanical stress, therefore they can be used as sensors. Such process is reversible, so an internal mechanical strain can be generated as a result of an applied electrical field. Thus the piezoelectric material can act as an actuator. Most piezoelectric devices can act in the dual capacity of sensor and actuator, and often the term transducer is preferred. Lead zirconate titanate (PZT) is one of the most commonly used piezoelectric ceramic materials.

(27)

3.1. SENSING INTERFACES FOR DIFFERENT MEASUREMENTS

3.1.1.1 Double PZT Transducers

Gu et. al. [4] formed a smart aggregate with a PZT patch protected inside a small cubic concrete block before casting into curing concrete, as depicted in Figure 3.2. To protect the PZT transducer from liquid and vibrating during concrete casting, it is first covered with an insulation water-proof coating, and then put into a small concrete block, with the same concrete material as the product under test. Such aggregates can be embedded into desired position within concrete. The whole system is given in Figure 3.3 .

Concrete Blocked Smart Aggregate Piezoelectric Patch Waterproof Coating Electric Wire Castedinto

Figure 3.2.Piezoelectric patch embedded into concrete blocked smart aggregate.

Quickpack EL1224 Power Amplifier Agilent 33120A Function Generator LeCroy LT342 Digital Oscilloscope Concrete Cylinder Specimen Piezoelectric

actuator Piezoelectric sensor

(28)

The harmonic signal is generated by Agilent 33120A function/arbitrary wave-form generator. A Quickpack EL1224 power amplifier boots it before the signal excites the piezoelectric actuator. At some distance within the same concrete spec-imen, the harmonic response signal of the piezoelectric sensor is recorded through a LeCroy Waverunner LT342 digital oscilloscope. An HP 4275A multi-frequency LCR meter (not visible in the figure) is used to record the capacitance values and the impedance values of the piezoelectric transducers. Three concrete cylinder specimens (I, II, and III) are tested by monitoring the harmonic response ampli-tude from the piezoelectric sensors.

The stress wave propagation within the concrete cylinder is considered as one-dimensional longitude. The harmonic amplitude of the wave can be expressed as a function of Young’s modulus E, which is a major factor of compressive strength. The authors further built a fuzzy correlation system between the harmonic ampli-tude and compressive strength, in order to evaluate the strength based on sampled harmonic amplitude.

During the experiment, twenty-four concrete cylinders were used, among which three were made with PZT aggregates. A universal compression testing machine was used to obtain the compressive strength value. In each test, three cylinders were applied gradually increased load on the same day until they broke. Their average compressive strength was obtained as the value of that day, as shown in Figure 3.4.

Figure 3.4.Compressive strength versus age [4].

(29)

developed at a very high rate. After the first week, the harmonic amplitude curves decreased at reducing speed, and the compressive strength increased at a much slower rate than before. After the 28th day, the characteristic concrete strength was obtained and stable, and the harmonic response amplitude turned flat, as indicated in Figure 3.5.

(30)

Figure 3.6.Comparison of the experimental training data with the fuzzy mapping data [4].

Figure 3.7.Comparison of the trial compressive strength with the estimated compres-sive strength [4].

(31)

3.1.1.2 Single PZT Sensor with Impedance Chip

According to an overview of EMI methods by Park et al. [35], assuming the me-chanical properties of PZT does not change over the testing period of the host structure, the electrical impedance of PZT is directly related to the mechanical impedance of the host structure. Further studies indicated that the real part of the PZT resistance is used. As the real part varies over a range of frequencies, the impedance response is analogous to that of the frequency response functions of the host structure.

A typical impedance measurement is carried out with an impedance analyzer (for instance HP4192A, HP4194A). PZT sensors are interrogated after setting up the impedance analyzer. Collected data are transferred to a desktop computer and analyzed afterwards. Such equipments are bulky and expensive (HP4192A alone costs more than 30,000 SEK). Quinn et al. proposed a light weight system replacing the impedance analyzer and the desktop computer with AD5933 impedance chip. In their article,they compared these two sets of system and concluded that they can realize similar functions in EMI method. Furthermore, the authors suggested that the new system integrated with MCU, RF transceiver and an embedded antenna in the Tyndall 25-mm sensing node [19] can form a wireless sensor network for remote control.

To protect the PZT sensor from aggressive conditions, a package of Robnor Resin PX314ZG epoxy was applied. The epoxy was coated onto the sensor using a specialized mold. The thickness of the coat was less than 2mm and its compressive strength was reported as 79-86 MPa, much higher than that of the tested concrete (around 35 MPa) [34]. Figure 3.8 shows the package of the sensor, where a is the mold and b is the final sensor [5] .

Figure 3.8.Protection package of the PZT sensor [5].

Figure 3.9 shows their proposed experiment set-up. Two CR2430 cell batteries provided the power supply.

(32)

PC with LabView PZT Sensor AD5933 Freq. Sweep Impedance

Figure 3.9.Basic experiment set-up of EMI method using PZT.

anti-resonant frequency development using both approaches [5]. The compres-sive strength increased rapidly over the first 10 days, then the rate of strength gain reduced significantly after 15 days, until a 28-day characteristic strength of approx-imately 34 MPa was achieved. The anti-resonant frequencies of both approaches followed the strength development curve, which verified that the AD5933 system is as accurate as the original HP4192A system.

The authors also discussed compressive testing of cured concrete specimen with antiresonance frequency of reactance as well as maximum impedance. Figure 3.11 indicates the result [5]. An Impact GD10 hydraulic compressive tester was used to find out the effects of loading and failure responding to the sensor. Results showed that the sensor could be used effectively in structure health monitoring and did not introduce much weakness into the test structure.

(33)

Figure 3.10.Comparison of compressive strength development with anti-resonant fre-quency development using AD5933 and HP4192A [5].

(34)

3.1.2 Relative humidity Test

Different commercial humidity sensors are used generically or formed into embed-ded sensor systems with other components. Here the standalone sensors will be introduced first, then the integrated systems.

3.1.2.1 Sensirion SHT Humidity Sensors

SHT1X, SHT2X are two generations of humidity/temperature sensors from Sen-sirion AG. SHT11 (as in Figure 3.12) will be introduced as a detailed example. It is a surface mountable relative humidity and temperature sensor, providing a dig-ital output. A capacitive sensor element is used for measuring relative humidity while temperature is measured by a band-gap sensor. Both sensors are seamlessly coupled to a 14bit ADC and a serial interface circuit, leading to clear signal qual-ity, short response time and tolerance to external disturbances (EMC). Each SHT1X is individually calibrated in a precision humidity chamber. The calibration coef-ficients are programmed into an one-time programmable (OTP) memory on the chip, to internally calibrate the signals from the sensors. The 2-wire serial interface and internal voltage regulation allows for system integration. A filter cap SF1 is designed specifically to protect SHT1X sensors against water, dust, soot and other contaminants. The body of the filter cap is Polypropylene and the filter membrane is made of polytetrafluoroethylene (PTFE) with polyester scrim.

Figure 3.12.SHT11 [6].

Summarized from the datasheet [6], the performance of SHT11 can be featured as in Table 3.1 :

(35)

Energy consumption 80uW (at 12bit, 3V, 1 measurement / s) RH operating range 0 - 100% RH

Temperature operating range -40 - +125◦C RH response time 8 sec (τ 63%)

Output Digital (2-wire interface)

Dimensions 7.5mm*4.9mm*2.5mm

Price 200 SEK

Table 3.1.Performances of SHT11

Figure 3.13.SHT11: Change of accuracy under different RH [6].

(36)

3.1.2.2 Maxim Integrated iButton Hygrochron Temperature/Humidity Logger This is a system that measures temperature/humidity and records the result in a memory section protected by password. The recording is done at a user-defined rate. A total of 8192 8-bit readings or 4096 16-bit readings taken at equidistant in-tervals ranging from 1s to 273hrs can be stored. In addition, there are 512 bytes of SRAM for storing application-specific information and 64 bytes for calibration data. Data collection can be programmed to begin immediately, after a user-defined delay or after a temperature alarm. The DS1923 is configured and communicates with a host-computing device through the serial 1-Wire protocol, which requires only a single data lead and a ground return. The stainless-steel package is resistant to environmental hazards such as dirt, moisture, and shock. The pin configuration and size dimension are illustrated in Figure 3.15 [7].

Figure 3.15.Pin configuration and size dimension of iButton.

The safe operating rage is revealed in Figure 3.16. With software correction via 1-Wire protocol, the typical value of RH accuracy variation is ±5%RH.

Figure 3.16.Safe operation range of iButton [7].

(37)

Memory 512 bytes of SRAM

Input Size 8192 byte

Sampling Rate 1s to 273hrs

Input Accuracy Correctable to ±0.5◦C

Resolution 8-bit (0.5◦C) or 11-bit (0.0625◦C) Operation Range -20◦C to +85◦C, 0 to 100%RH Power Consumption ca. 1.8µW

Price ca. 56 USD

Table 3.2.Key features of iButton

Figure 3.17.CC2451 SensorTag Schematic [8].

3.1.2.3 Texas Instruments CC2451 Bluetooth SensorTag

This is a highly integrated system with 6 sensors, including a Sensirion SHT21 sensor similar to SHT11. The computation block is the CC2451 Bluetooth System-on-Chip, consisting RF layer and an 8051 MCU. A SensorTag app features wire-less communication with mobile devices. Figure 3.17 shows the schematic [8]. Figure 3.18 shows the content and package of the SensorTag. As a generic auto-matic sensing product, it is user-friendly. Still further investigations shall be made to verify its performance inside the curing concrete, and whether bluetooth can propagate through such heterogeneous material.

(38)

Figure 3.18.CC2451 SensorTag content and package [8].

SENSORS

Humidity sensor Sensirion SHT21

Temperature sensor Texas Instruments TMP006 IR Pressure sensor Epcos T5400

Accelerometer Kionix KXTJ9

Gyroscope InvenSense IMU-3000

Magnetometer Freescale MAG3110

Package Size 71*36*15.5 mm

PCB Size 57*25*1.5 mm

Power A single cell coin cell battery (CR2032), quiescent current consumption of 8µA, allowing years of battery life. See Figure 3.18.

Price 25 USD

(39)

3.1.2.4 A Wireless System with SHT11 and Tyndall Sensing Node

Besides the EMI method, Quinn et al. presented another approach using SHT11 humidity sensor and Tyndall sensing node to integrate a wireless sensing node for relative humidity test in curing concrete. The functional blocks can be depicted as in Figure 3.19 [5]. ATmega 128L SHT11 nRF905 TRX SPLATCH 433 SP2 Antenna Control Output Sensor Layer Communication & Computing Layer

Figure 3.19.Wireless sensing system for RH tests.

System Development The 433 MHz ISM band was chosen as the transmission frequency band of this system. An ATmega 128L microcontroller, an nRF905 multi-band transceiver, and a SPLATCH 433 SP2 50Ω patch antenna formed the comput-ing and communication interface. This grounded line patch antenna would allow the minimization of the sensor and package system. The Tyndall 25 mm sensing node is a miniaturized, modular 25 mm * 25 mm, stackable development platform for wireless sensor networks [19]. The developed sensor mote is given in Figure 3.20.

(40)

Sensi

ng

Transmi

ssi

on

Power

Transmi

ssi

on Sensi

ng

SHT15Antenna Microcontroller

& Transceiver

Figure 3.20.Developed layers using Tyndall mote [9] .

Figure 3.21.Protection of WSN by Quinn et. al. [9] .

Feasibility of 433 MHz ISM Band In actual test with concrete block and steel reinforcement, the results showed that concrete and rebar would reduce the trans-mission distance of 150 m in open air to over 5 m inside the concrete structure. Within this 5 m range, wet concrete and the curing process had little effect on transmission of data in 433 MHz ISM band, over the first 3-day period.

In-situ Conditions The next experiment was to determine whether the sensors can perform in a construction site environment. A concrete mold with the dimen-sions of 600 mm * 600 mm * 900 mm was designed using steel backed framework that would be removed after 10 days. Figure 3.22 shows the location of sensors in-side the structure. One sensor at the bottom, four in the center of the four faces with a distance of 40-50 mm to the edge. The last sensor was placed 40-50 mm below the top panel of the concrete. A monopole antenna transmitter was placed in the center to compare transmission success with that of the patch antenna. Thermo-couples were also placed inside the concrete as a reference of temperature sensor. The receiver was placed 1.5m from the concrete.

(41)

Figure 3.22.Location of sensors inside concrete [9] .

3.23. The RH was expected to reach a maximum value at first, and then decrease over time as the concrete hydrated. In the figure, only Sensor 4 showed similar results as expected. The authors believed that condensation occurred within the sensing region and the Gore package did not provide suitable protection.

(42)

The result showed the possibility to transmit data from over 500 mm inside the concrete. A maximum transmission distance of 3.5 m outside the concrete when the steel framework was present and 5 m without the framework. The packet deliver rate (PDR) was also influenced significantly by the steel framework, as presented in Figure 3.24. Compared with RH results, the authors found that the lowest PDR occured arond the first 100h, when the RH was over 80%. It could be indicated that the early moist conditions within the concrete would have an important impact on the wireless transmission of data.

Figure 3.24.Results of PDR over time [9] .

(43)

3.1.3 Maturity Method

Two popular commercial maturity meters adapting this method will be introduced in this section.

3.1.3.1 COMMAND Center Maturity Meter

This product was developed by by The Transtec Group, Inc. A complete system of COMMAND Center maturity meter [10] consist of data loggers and computing devices. The computing core can be either a COMMAND Center pocket reader, as shown in Table 3.25, or a PC running the dedicated software.

Figure 3.25.COMMAND Center maturity meter [10].

The data loggers use the iButton sensors, therefore inherit the main features. The loggers are wired to the reading devices. The system details are available in Table 3.4.

Computer Requirement Laptop or Pocket PC

Maturity Method Standard ASTM C 1074 (Temperature-Time Factor, TTF) Datum Temperature User-defined

Sensor Accuracy ±1◦_C

Temperature Range -10 to 85◦C Sensor Data Storage 2048 readings

Measurement Interval User-defined: 1-255 minutes Typical Measurement Intervals (min) 5, 10, 20, 240

Monitoring Period Accordingly (days) 7, 14, 28, 340

Sensor Size 6.35 mm×19.05 mm

Cable Length (feet) Standard 8 or 15, custom lengths ≤ 100

Sensor Battery Life Minimum shelf life of 2 years at room temperature

(44)

To use the maturity meters, three steps shall be conducted. It is important to batch and test concrete under the process as close as possible to that during con-struction in overall processes. Any deviations in concrete preparation including batch size, mixing equipment, mix proportions, material sources and manufactur-ers may reduce the reliability of the maturity method.

Step 1: Establish the maturity-strength relationship Before field construction, maturity sensors are inserted in test concrete cubes and cylinders, using a mix that is as close as possible to the concrete to be used in later construction. The maturity data is collected using the Pocket PC and fitted to a curve using COMMAND Cen-ter software version or another program like Microsoft Excel.The data can be fitted and plotted using a logarithmic or hyperbolic function, a linear fit on a semi-log plot, or a more complex sigmoidal function.

Step 2: Estimate In-Place Strength First, the same concrete used to develop the maturity-strength relationship are batched. Routine quality control and assurance tests are performed to ensure the concrete is in unity. Maturity data loggers are then into the construction. Proper delivery, consolidation and curing practices shall be performed to fulfil the assumptions of maturity theory. Finally, COM-MAND Center’s measured maturity value and established maturity-strength rela-tionship are used to estimate the in-place strength.

Step 3: Verify the Maturity-strength Relationship Materials, mixing equipment performance, and construction conditions may vary over time. Therefore, the maturity-strength relationship shall be revised and verified periodically or before safety-critical operations, to ensure it is up-to-date. Verification can either be done by monitoring the maturity of strength specimens cast during construction and comparing sampled strength data to the maturity curve, or estimating the in-place strength using other methods, such as Surface Hardness Methods or Pull-out Tests.

Table 3.5 shows the relatively high price of this product.

Component Price (USD)

Data Logger 35

Software 550

Pocket Reader 950

Complete Kit: 50 data loggers and software 2150 Complete Kit: 50 data loggers and pocket reader 2550

(45)

3.1.3.2 IntelliRock Maturity System

IntelliRock Maturity System was developed by Engius, FLIR in 2002 [11]. Similar with the COMMAND Center product, it consists of data loggers (sensing inter-faces) and readers. In addition to traditional TTF method based on Nurse-Saul function, the Arrhenius function can also be adapted to this product. The features of loggers in Figure 3.26 are shown in Table 3.6.

Figure 3.26.Maturity Data Loggers [11].

Item Performance

Operating Temperature -5 to 85◦C

Storage Time & Temperature 0 to 35◦C for 2 years. Max Temperature measurement Range -18 to 99◦_C

Temperature Accuracy ±1◦_{C, -5 to 85}◦_C

Temperature Resolution 1◦_C

Time accuracy 1 minute per month

Maximum wire length (feet) 50 (in concrete or water), 200 ( in air) Temperature measurement interval 1 minute

Maturity integration period 1 minute

Data Storage 999 reads

Table 3.6.IntelliRock Loggers: Feature of Measurement Performance

(46)

Recently, intelliRock Wireless has been developed, allowing remote monitoring of intelliRock Loggers. This system is made of four main components in addition to the temperature or maturity loggers: Radio Box, Wireless Base Station, software and intelliRock Wireless Repeaters. Figure 3.27 shows the system.

Radio Box Wireless base station

Maturity Loggers

Figure 3.27.IntelliRock Wireless

The radio box can connect to up to 1-8 loggers at one time.Each box also acts as a radio repeater. The wireless base station connects to a PC via USB and directs any quantity of radio boxes remotely. The working range can be extended using an external power supply, or an enhanced directional antenna, and the wireless base station can communicate with radio boxes up to 20 miles away within direct line of sight. Therefore sensing units around the whole construction site can be controlled by one PC without any hand-held readers.

(47)

3.2. ACTIVE OR PASSIVE SENSING AND POWER HARVESTING

3.2 Active or Passive Sensing and Power Harvesting

Most products and prototypes discussed here are passive sensing systems that only respond to the change of the environment under test, such as humidity sen-sors and maturity meters (which are based on temperature sensen-sors). They are often passively powered by batteries. This is contrast to active sensors that interact with or excite the environment, such as piezoelectric sensors. Because of their duality as sensors and actuators, piezoelectric components can sometimes contribute to power harvesting of the system. In this section, two research prototypes related to active sensing and power harvesting will be studied in detail. Other options of sensing and power will also be discussed.

Note that the two following designs are dedicated to structure health monitor-ing in cured concrete, rather than monitormonitor-ing the curmonitor-ing process.

3.2.1 Wireless Power Transmission

This prototype was proposed by Mascrenas et al. [12]. Figures in this section are selected from their article. The sensing layer and computing layer, consisted of a PZT patch, an AD5934 impedance chip and an ATmega 128L microcontroller, were similar to that of Quinn et al [5] discussed in Section 3.1.1.2, as depicted in Figure 3.28. The differences lied in wireless communication and power layers. For the communication layer, XBee Radio transmitter and receiver working in 2.4 GHz were chosen. The power supply came from not batteries, but microwave wireless energy transmission.

The RF power transmission set-up is described in Figure 3.29. The X band at 10 GHz was chosen to transmit RF energy in order to facilitate small antennas with high gain. 1 Watt was transmitted from one horn antenna to another in the distance of 0.61 m. The second antenna and an eight-stage voltage multiplication circuit acted as an rectenna at the receiver end. The output DC voltage charged up a 0.1 F supercapacitor. Once sufficient energy was built up in the supercapacitor, i.e. both the voltage and the power met the requirement, it would perform as the power supply to sensing, computing and communication interfaces.

The testing process can be described as the following:

• The microcontroller first sends an initial signal, and then frequency param-eters including start frequency, frequency increment, number of frequency points and settling time between frequency points to the impedance chip. • Once the impedance chip is programmed, the microcontroller sends a start

command and the impedance chip excites the PZT sensor with the initial frequency value.

(48)

Figure 3.28.Functional block diagram [12].

• The range of frequency sweep was designed as 70-100 kHz for AD5933, and the response antiresonance frequency of reactance along with impedance peak were collected for further analysis.

(49)

3.2. ACTIVE OR PASSIVE SENSING AND POWER HARVESTING TX RX Voltage Rectifier Capacitor 0.61 m 10 GHz AC Power DC 3.3 V

(50)

3.2.2 Power Harvesting through Piezoelectric Components

Sodano et al. [13] investigated the use of piezoelectric energy harvesting devices for charging batteries. They used an electromagnetic shaker to induce vibration of random signal in 0-500 Hz. An example of their battery charging circuit is shown in Figure 3.30. Results infer that, inducing a random vibration of 0-500 Hz to a 2.5 in * 2.375 in * 0.0025 in (thickness) PZT patch, within 1.2 h, a 200 mAh capaci-tive battery can be charged up to 90% capacity. According to the British Standard Guide to evaluation of human exposure to vibration in buildings [36], the ambient vibration of buildings is normally under 80 Hz for every-day life, and 100 Hz can be induced to buildings when an underground train passes by.

Figure 3.30.Schematic of piezoelectric-based battery charing circuit [13].

Grisso et al. [14] proposed a design of wireless active sensing and power har-vesting system both using piezoelectric components. Their prototype was called the MEMS-Augmented Structural Sensor (MASSpatch). Such a system incorpo-rated the use of piezoelectric based Macro-Fiber Composite (MFC) sensors and ac-tuators, computed all impedance-based structure health monitoring on one cheap chip-sized computer, was self powered from piezoelectric-based power harvest-ing, and communicated remotely. The functional blocks are shown in Figure 3.31. Diamond System’s Prometheus PC104 board consisted of a microprocessor (ZFx86 at 100 MHz), 16-bit eight-channel ADC, and 32 MB flash disk was the com-puting core of the system. One Radiometrix RX2M-458-5 radio and one TX2M-458-5 radio, were used for wireless receiving and transmitting, respectively. They are tuned to two distinct frequencies (458.5–491 MHz and 433.05–434.79 MHz), and have communication speed up to 5 kbps and maximum communication ranges of 1 km.

(51)

3.2. ACTIVE OR PASSIVE SENSING AND POWER HARVESTING PC 104 Digital Output A/D TX RX Function Generator Bo lte d Jo int M FC Low Cost Circuit

Figure 3.31.Functional block of MASSpatch [14].

(52)

3.2.3 Passive RFID

Although different research prototypes have been proposed for power harvesting, commercial products are yet to come in use, not to mention the bulkiness of the power harvesting components compared to the need of compact portable testing. At the moment, batteries still dominate the commercial market of wireless sensor. To cope with the challenge of finite battery life, passive RFID sensors have been accepted widely. By near-field inductive coupling, a remote reader can transmit power to a RFID-based wireless sensor tag within the working distance. For vari-ous transmission frequencies, the effective distance between the tag and the reader is usually under 10 m [33].

The electromagnetic waves working in RFID transmission usually are not able to penetrate metal and have difficulties penetrating water [37], resulting in their inefficiency inside curing concrete, where water and metal framework exist. Yet RFID-based sensor system can still be suitable for applications in structure health monitoring of cured concrete structures.

Resonant Circuit State Sensor L1 L1 C2 C2 C1 C1

Figure 3.32.State Sensor

An example is the RFID-based sensor as a crack-detector, designed by Novak et al [15]. It was based on the concept on state-sensors, as in Figure 3.32. The on/off state of the switch result in different resonant frequency of the circuit. In practise, Electronic article surveillance (EAS) stickers were selected as the resonant circuit for the crack detection sensors for their low price and availability. Copper foil tape was used as the switch. The tape is designed to tear when a crack occurs.The conceptual sketch was shown in Figure 3.33.

(53)

3.2. ACTIVE OR PASSIVE SENSING AND POWER HARVESTING

Figure 3.33.Sketch of crack-detector [15].

External Transmitter/Receiver Prototype Sensor M C2 C2 Ltrans Ltrans C1 C1 Rtrans Rtrans L1 L1 V1 V1

Figure 3.34.Inductive coupling from external reader.

(54)

3.3 Communication

3.3.1 Comparison of Electromagnetic Model in Concrete

A. Robert [16, 28] compared the five models mentioned in Section 2.2.1. He set up a series of tests to compare measured permittivity of concrete with modeling results. The measurement was conducted using a HP 8753C Network analyzer and a large coaxial reflection/transmission cell to average the local fluctuations in permittivity inherent to the concrete structure, as shown in Figures 3.35 and 3.36 ,

Figure 3.35.Experimental setup for measuring the permittivity of concrete [16].

(55)

3.3. COMMUNICATION

Between 50 MHz and 1 GHz, his test results revealed that the CRIM model and SSC model are not fit for calculating the complex permittivity of the concrete. See Figures 3.37 3.38.

Figure 3.37.Real part of the permittivity over frequency [16].

Figure 3.38.Imaginary part of the permittivity over frequency [16].

(56)

(57)

3.3. COMMUNICATION

3.3.2 Transmission Frequency

Jiang and Georgakopoulos [38] studied wireless wave propagation using patch antennas through concrete structure at 433 MHz, 915 MHz and 2.45 GHz. Results suggested 915 MHz as the most suitable among the three frequencies, because an-tennas offer larger coupling and are less sensitive to changes of humidity condi-tions and rebar configuration inside the concrete structure.

Alternatively, Quinn et al. [5] selected 433 MHz as more suitable for their ex-periments in wireless RH sensor network in curing concrete. Comparing to the 150 m typical transmission range in unobstructed open air, monople antennas placed inside concrete has a much shorter transmission distance of over 5 m. Within the minimum range, moisture contents have little effect on transmission. When con-ducted in-situ experiments, 550 mm was set as the greatest distance between an-tennas inside the concrete. 3.5 m was found to be the farthest distance between the antenna transmitter inside the conrete and the receiver outside with the presence of steel framework and 5 m without the framework. Detailed distribution of sen-sors are revealed in Figure 3.22. Examination of the package delivery rate during the process also indicates the impact of steel and moist: the PDR never fell below the 90% level after the removal of steel and early hydration.

Shams and Ali [39] explored the feasibility of wireless power transmission around 5.7 GHz. However they did not test the effects of metal rebars.

(58)

3.4 Protection

Sensors inside curing concrete are exposed to hazard environment conditions, in-cluding but not limited to water, alkali and crush from aggregates. Protection package are necessary. Three types of protection shells, plastic, epoxy and cement mortar, have appeared in research designs mentioned in earlier parts.

Plastic Waterproof plastic material is often selected as protection package for hu-midity tests. The Sensirion SHT huhu-midity sensors have specifically fabricated plas-tic filter caps. The body is made of Polypropylene and the filter membrane is made of PTFE with polyester scrim. The filter cap fits the outer dimensions of the SHT humidity sensors and allows for compact system design and fixing on PCB. It is also designed to keep response time low. An epoxy ring can be applied to seal to seal the filter cap to PCB.

Integrating SHT11 humidity with Tyndall mote, Quinn et al. [5] protected the whole sensing node with polyoxymethylene plastic and covered the sensor region with Gore Vent. However, their result suggests that Gore membrane can not pro-vide adequate protection, and the sensor may measure the internal humidity of the package.

Epoxy A thin layer of epoxy made of Robnor Resin PX314ZG was used by Quinn et al. [5] in EMI method using the PZT patch and the impedance chip. Its compres-sive strength is higher than that of the concrete, so as not to induce weakness into the structure.

Cement Protection Some of the studies related to PZT patches used cement pro-tection package. Gu et al. covered the PZT patch with a cubic concrete block after coated it with waterproof material[4].

Barroca et al. [40] provided a similar protection for their sensor node. In their design, the SHT humidity/temperature sensor and the microcontroller unit were soldered on one PCB. After the SHT sensor was shielded by the special plastic filter cap, the PCB was put into a shell made of high porosity cement. Although this package was tested as effective in protecting sensor wire connections, it could not prevent all the damage of alkaline environment.

(59)

3.5. SIMULATION TOOLS

3.5 Simulation tools

Cement hydration reactions play a vital part in the curing process of concrete, In nature, the cement hydration is a heat transfer process, and the changes in temper-ature may lead to thermal cracking problem. To predict the tempertemper-ature change process, one can prepare a test mixture of the concrete of interest and monitor the hydration process under adiabatic conditions. Alternatively, computer models have been developed to assist the prediction and simulate the micro structure.

There have been two types of computer models. The first one can be described as continuum. The hydration process as a whole can be computed as a function of the particle size distribution, the chemical composition, water-to-cement ratio and actual reaction temperature. An example approach, the HYMOSTRUC model, was formulated by van Breugel [41, 42]. The second type is called digital-image-based model. Such model operates at the sub-particle level as each cement particle is represented as a collection of elements/pixels. A processed scanning electron microscope (SEM) image and an X-ray image serve as input. An example is the CEMHY3D model developed by Bentz [43]. The remain of this section will intro-duce these two models.

3.5.1 HYMOSTRUC

HYMOSTRUC is a software running on Windows PC [44]. The input profile con-sists of chemical compositions, water-to-cement ratio, aggregate distribution pa-rameters, temperature and other hydration phase parameters. The interface can be seen in Figures 3.39 to 3.42.

(60)

Figure 3.40.Mixture proportions.

Figure 3.41.Evaluation of the properties and characteristics of the virtual microstruc-ture during and after hydration.

(61)

Figure 3.42.Hydration process control.

the calculated results in a 3D framework, as shown in Figure 3.45.

Future implementation can be investigated to adapt results from other sens-ing system, such as the changsens-ing ambient temperature, to the input values of HY-MOSTRUC.

(62)

Figure 3.44.Hydration degree over time (hour).

(63)

3.5.2 CEMHYD3D

The name of CEMHY3D is a combination of cement, hydration and 3D. Unlike the UI-based program HYMOSTRUC, it is written in C language and runs in UNIX/Linux environment [45]. A work flow of CEMHY3D is shown in Figure 3.46 [43].

Cement

Powder Reconstruct 3D Microstructure

3D Cement Hydration

Model

Heat of

Hydration Validation Lab SEM Images Particle Size Distribution Hydration Reaction Rules Volume Stoichiometrics 3D Microstructure Evolution Chemical Shrikage Physical Property Computation

Figure 3.46.Work flow of CEMHY3D.

At first, the cement of interest was first dispersed in a smooth epoxy that was subsequently cured. Then SEM and X-ray both viewed the polished surface and

(64)

the particle size distribution was measured. This led to the reconstruct of 3D mi-crostructure. Combining hydration reaction rules and volume stoichiometries, the 3D cement hydration model could be generated. This can be used to predict the 3D microstructure evolution, thermal reactions of hydration and chemical shrinkage. A example image of simulated 3D microstructure can be seen in Figure 3.47 [46]. Results of predicted heat profile during hydration can be adapted with maturity method to assess compressive strength.

(65)

Chapter 4

Comparison and Summary

This chapter summarises all practise from the aspects of hardware/software de-tails of prototypes, communication models and protection.

4.1 Sensor Systems

In this section, several tables present all prototypes with critical parameters as a guidance for future study. Tables 4.1 and 4.2 are academic prototypes. It can be in-ferred from the examples that piezoelectric sensors are popular for measurement of concrete strength, whether in curing or cured concrete. Except for the use of compact plug-and-play Tyndall Mote [19] by Quinn [5], other researchers did not pay specific attention to size reduction. Comparing to the power consumption in wireless data transfer, computation of MCU costs less power, therefore it may be more energy-efficient if the computation unit is embedded in the concrete with the sensing unit. In addition to commonly usage of batteries, power harvesting meth-ods have been tested, yet their bulkiness diminishes the flexibility of the whole system.

(66)

PROTOTYPE Quinn et al.[5] Quinn et al.[5] Mascarenas et al.[12]

TEST CONDITION Curing Concrete Curing Concrete Existed Structure DATA ACQUISITION

Analog Input RH/Temperature Impedance Impedance

Sensor SHT PZT+AD5933 PZT+AD5933

Sample Rate 1 Hz 1 MHz 1MHz

A/D Channels 1 1 1

A/D Resolutions 12/14 bit 12 bit 12 bit

Digital Inputs 1 1 1

COMPUTATION

Processor Atmega 128L PC Atmega 128L

Bus Size 8 bit 8 bit

Clock Speed 16 MHz 16 MHz

Program Memory 128 kB 128 kB

Data Memory 4 KB 4 KB

Software LabVIEW

COMMUNICATION Wireless Wired Wireless

Radio nRF905 XBee DigiMesh

Frequency Band 433 MHz 2.4 GHz

Wireless Standard DigiMesh

Outdoor Range 150 m 90 m

Enclosed Range 3.5-5 m 5.2 m

Data Rate 50 kbps 250 kbps

FINAL ASSEMBLED UNIT

Physical 65 mm×45 mm

Power On 212.85 mW,

Sleep 51.81 µW Power Source Two CR2430 coin

cells

Microwave trans-mission

Price >7000 SEK 150 USD

Protection Polyoxymethylene, Gore Vent

Epoxy

(67)

4.1. SENSOR SYSTEMS

Prototype Gu et al.[4] Grisso et al.[14] Novak et al.[15]

TEST CONDITION Curing

Con-crete

Existed Structure Existed Struc-ture

DATA ACQUISITION

Sensor PZT PZT State Sensor

Input Harmonic

Volt-age

Impedance Resonant

Fre-quency

Sample Rate >50 kHz

A/D Channels 8

A/D Resolutions 16 bit

Digital Inputs 1

COMPUTATION

Processor None Diamond Systems PC104 None

Bus Size 8 bit/16 bit

Clock Speed 100 MHz

Program Memory 64 KB

Data Memory 32 MB

Software DOS

COMMUNICATION Wireless Wireless Wireless

Radio Radiometrix RX2M-458-5 & TX2M-458-5 Frequency Band 458.5-459.1 MHz or 433.05-434.79 MHz Wireless Standard Outdoor Range Enclosed Range > 1 km Data Rate 5 kpbs

FINAL ASSEMBLED UNIT

Physical 79 g

Power +5 VDC

Power Source PZT vibration power

har-vesting

RFID reader

Price 150 USD

Protection Waterproof+cement

Study of Sensor Network Applications in Building Construction

Study of Sensor Network Applications in

Building Construction

XIAOTAO WANG

Study of Sensor Network Applications

in Building Construction

XIAOTAO WANG

Supervisor: George Athanasiou and Carlo Fischione

Examiner: Carlo Fischione

Stockholm December 20, 2013

Automatic Control Department

School of Electrical Engineering

Abstract

Contents

List of Figures

Chapter 1

Introduction

1.1

Background: Wireless Sensor Network and Concrete

Monitoring

1.2

Problems in Traditional Measurements

1.3

Target

1.4

Research Method

1.5

Delimitations

1.6

Outline

Theoretic Background

2.1

Testing Methodologies for Curing Concrete

2.2

Communication model

Chapter 3

Analysis of Existed Solutions

3.1

Sensing Interfaces for Different Measurements

Sensi

ng

Transmi

ssi

on

Power

Power

Transmi

ssi

on Sensi

ng

3.2

Active or Passive Sensing and Power Harvesting

3.3

Communication

3.4

Protection

3.5

Simulation tools

Chapter 4

Comparison and Summary

4.1

Sensor Systems