An Embedded Wireless Sensor Network with Wireless Power ... - MDPI

1 downloads 0 Views 6MB Size Report
Nov 7, 2017 - The sensors, consisting of a load cell (LC) with a full scale range equal to 1 ton. (. ∼. =10 kN) (type 3141-1T, supplied by Phidgets Inc.) and an ...
sensors Article

An Embedded Wireless Sensor Network with Wireless Power Transmission Capability for the Structural Health Monitoring of Reinforced Concrete Structures Luca Gallucci 1 , Costantino Menna 2, *, Leopoldo Angrisani 3 , Domenico Asprone 2 , Rosario Schiano Lo Moriello 4 ID , Francesco Bonavolontà 3 ID and Francesco Fabbrocino 5 1 2 3 4 5

*

ID

TME s.r.l.—Test and Manufacturing Engineering, via C. A. Dalla Chiesa—81050, Portico di Caserta (CE), Italy; [email protected] Dipartimento di Strutture per l’Ingegneria e l’Architettura, University of Naples Federico II, Via Claudio 21, 80125 Naples, Italy; [email protected] Dipartimento di Ingegneria Elettrica e delle Tecnologie dell’Informazione, University of Naples Federico II, Via Claudio 21, 80125 Naples, Italy; [email protected] (L.A.); [email protected] (F.B.) Dipartimento di Ingegneria Industriale, University of Naples Federico II, Piazzale Tecchio 80, 80125 Naples, Italy; [email protected] Dipartimento di Ingegneria, Università Telematica Pegaso-Piazza Trieste e Trento 48, 80132 Naples, Italy; [email protected] Correspondence: [email protected]; Tel.: +39-081-7685959

Received: 2 September 2017; Accepted: 3 November 2017; Published: 7 November 2017

Abstract: Maintenance strategies based on structural health monitoring can provide effective support in the optimization of scheduled repair of existing structures, thus enabling their lifetime to be extended. With specific regard to reinforced concrete (RC) structures, the state of the art seems to still be lacking an efficient and cost-effective technique capable of monitoring material properties continuously over the lifetime of a structure. Current solutions can typically only measure the required mechanical variables in an indirect, but economic, manner, or directly, but expensively. Moreover, most of the proposed solutions can only be implemented by means of manual activation, making the monitoring very inefficient and then poorly supported. This paper proposes a structural health monitoring system based on a wireless sensor network (WSN) that enables the automatic monitoring of a complete structure. The network includes wireless distributed sensors embedded in the structure itself, and follows the monitoring-based maintenance (MBM) approach, with its ABCDE paradigm, namely: accuracy, benefit, compactness, durability, and easiness of operations. The system is structured in a node level and has a network architecture that enables all the node data to converge in a central unit. Human control is completely unnecessary until the periodic evaluation of the collected data. Several tests are conducted in order to characterize the system from a metrological point of view and assess its performance and effectiveness in real RC conditions. Keywords: structural health monitoring; wireless sensor network; embedded sensors; wireless power transmission

1. Introduction Since buildings and infrastructures are potentially subjected to environmental degradation or hazard-induced damages, it is necessary that safety, functionality and durability requirements be met throughout their service life. To guarantee efficiency and safety, an effective method consists in monitoring some structure-related properties that can be, in turn, associated with the local or global

Sensors 2017, 17, 2566; doi:10.3390/s17112566

www.mdpi.com/journal/sensors

Sensors 2017, 17, 2566

2 of 21

integrity of the structure itself. The range of possible measurements is wide and depends on the type of structure being investigated; as an example, health monitoring techniques can be based on measures of: local stresses or strain field; inclinometric measures; locations, amplitude and pattern of cracks; local rotations and displacements; temperature or corrosion of metal components. Based on the measured parameters, health monitoring of civil infrastructures can determine the location and severity of damage according to four levels of damage identification [1], as follows:

• • • •

Level 1: Level 2: Level 3: Level 4:

determination that damage is present in the structure determination of the geometric location of the damage quantification of the severity of the damage prediction of the remaining service life of the structure.

In this context, a continuous real-time monitoring of such properties allows for a reliable and comprehensive knowledge of the integrity of in-service structures, entailing the implementation of optimal solutions for planned maintenance while minimizing downtime or damages due to hazardous events. Indeed, a significant portion of safety evaluations is currently based on periodic physical inspections, sometimes appearing inadequate or costly. The current practice mainly relies on Non-destructive Damage Evaluation (NDE) methods, because only a limited knowledge about structural properties of in-service structures is available in real time. In particular, nowadays, structural health monitoring (SHM) is implemented in reinforced concrete (RC) members through various techniques that are based on different operating principles. These have advantages and disadvantages depending on the specific environment where the structural component is located and the overall cost of the SHM technique itself. Passive sensors have been widely studied in recent decades and mainly consist of metallic elements embedded in a given structure. Their resonance frequency changes through physical and geometrical variations that are associated with alterations in the mechanical properties of the structure itself [2]. Reliable solutions have also been introduced by employing mechanical-deformation-based techniques, which typically enable the analysis of the stress state in rock masses, metals, and plastics through the restoration of the deformation field around a predefined septum created by the user. Recently, the use of strain gauge sensors has been implemented to analyze and monitor the mechanical deformation of structures subjected to static and dynamic loads. Strain gauge sensors guarantee very high accuracy in terms of the measurement of strain which is calculated from the electrical resistance variation associated with the deformation of the support onto which they are bonded. Further evolution has been achieved using piezoceramic sensors. These are applied externally to the structures being tested and are then correlated with the deformations of the outer surface to which they are attached. Piezoceramic sensors represent one of the first attempts to utilize embedded sensors in structures. They are typically used for vibrational analyses that exploit the piezoceramic effect. This effect allows the vibrations to be converted into an electric signal which, with appropriate filtering, tracks the acoustic events generated by the formation of internal cracks. More advanced techniques involve the use of two piezoceramic transducers, with the first acting as an acoustic emitter (by means of suitable electrical signals) and the second as a receiver. By analyzing the received acoustic waves transmitted from the first transducer through the structure, it is possible to conduct an analysis of the variation of the structural properties (from the density up to internal failures) that are related to the distortion of the waves propagated in the medium during the transmission [3]. The use of accelerometers is nowadays referred to as a continuous acoustic analysis, with concrete structures (pre-stressed and reinforced by cables) increasingly popular thanks to the growing interest in MEMS devices [4]. Distributed in key locations along a structure, each sensor is cabled to an acquisition unit, which continuously monitors the system without collecting data until there is a trigger event, e.g., an acoustic wave exceeds a predefined threshold. These sensors are effectively used to quantify the energy released by failure events that affect the overall dynamic behavior of the structure.

Sensors 2017, 17, 2566

3 of 21

Over the past two decades, a large number of applications have made use of fiber optic sensors to monitor both new and existing large infrastructures. With a particular focus on pre-stressed structures, fiber optic sensors allow mechanical stresses to be analyzed during the construction phase (including all the deformations caused by physical shrinkage) and throughout the service life of a structure by monitoring its response to variations in applied loads. One example is the Tsing Ma Bridge where, in 2003, 40 fiber Bragg gratings (FBG) were installed on the hanger cable, rocker bearing, and truss girders to monitor the dynamic strain and temperature [5]. The surveillance d’ouvrages par fibres optiques (SOFO) system, literally “structural monitoring via optical fibers”, is a widely-used approach that employs optical fibers to conduct an analysis of mechanical deformations through the differential length between two fibers, one embedded in the structure and another that is free to move [6]. More recently, effective solutions for measuring structural properties and/or induced damage in RC structures, as proposed by Murthy [7], have made use of battery-less, embedded sensors which, for instance, make a direct analysis of possible steel rebar corrosion. As the sensors are wireless, manufacturing and installation costs are kept low. Furthermore, the absence of wiring makes the system totally maintenance-free and, at the same time, is a safer solution, since the embedding of the sensors is not an issue in terms of the stiffness of the structure [8,9]. However, this solution is unable to provide continuous monitoring of a structure, because the sensors can only be interrogated manually. The work by Ceriotti et al. [10] reports on the implementation of a wireless sensor network (WSN) for monitoring structural parameters. The network was comprised of different interconnected nodes, with each one capable of providing different measurements, e.g., temperature. This solution appeared to be very innovative, being a first step in the path to continuous check-ups, although it lacked the capacity of embedded measurements. Even though the considered approaches face the problem of having a system for the continuous and embedded monitoring of structures over long time periods, they do not provide a comprehensive solution that solves all the related issues. In particular, Murthy’s study presents an interesting approach, but it does not allow continuous monitoring over time; on the contrary, Ceriotti’s approach solves this particular issue but does not permit the direct measurement of the phenomenon under analysis. As a consequence, the definition and implementation of such system is still an open challenge and research efforts are required to overcome typical issues such as effective embedment and continuous monitoring of the RC structures over time. In this context, the present paper intends to evolve the ideas presented in the previous solutions, while limiting some of the disadvantages, thus achieving a system that try to solve the MBM paradigm efficiently. 2. Problem Statement Despite a wide variety of SHM solutions, a cost-effective structural monitoring technique that allows for the direct and continuous measurement of the stresses or other structural parameters within a RC structure throughout its service life is still required. Indeed, many of the solutions mentioned above are characterized by some limiting factors and drawbacks. For instance, low-accuracy and the reduced lifetime of passive sensors have been acknowledged in several studies [11], while stress-based methods typically need physical intervention in the structure, which requires a long execution time. Moreover, these SH solutions are unsuitable for either continuous monitoring or critical installations such as dams, power plants, or other structures at biochemical risk. Despite their accuracy, strain-gauge measurements are typically characterized by issues with regard to their installation, conservation over time, and the limitation of the measurements that are only possible for a portion of the structure that is generally referred to an external surface [12]. On the other hand, piezoceramic sensors represent a significant improvement in this field, but they only perform an indirect measurement of the accumulated stresses, which is carried out through a signal scattering analysis [13]. In the case of accelerometers, drawbacks are related to signal filtering operations required to remove superimposed noise and all other uncorrelated events [14].

Sensors 2017, 17, 2566 Sensors 2017, 17, 2566

4 of 21 4 of 21

In addition, all the methods discussed above are indirect; only the use of optical fibers allows the direct measurement of the deformation of a structure (andonly so the stresses), but allows with high In addition, all the methods discussed above are indirect; theinternal use of optical fibers the costs terms of both measurement directin measurement ofthe theinstallation deformationand of athe structure (and soequipment the internal[15]. stresses), but with high costs It appears thatinstallation an effective solution should: (i) provide in terms of both the andSHM the measurement equipment [15]. direct measurements of the variations of the mechanical in RC structures; (ii) implement economic of sensor systems It appears that an effectiveproperties SHM solution should: (i) provide direct measurements the variations that would achieveproperties the MBM in in RC a flexible and effective manner; and (iii) sensor allow the deployment of a of the mechanical structures; (ii) implement economic systems that would large number of sensors to guarantee the global structural monitoring in several locations of the achieve the MBM in a flexible and effective manner; and (iii) allow the deployment of a large number structure track and the analyze their evolution over time. this issue is addressed as “A to of sensorsand to guarantee global structural monitoring in Sometimes several locations of the structure and track E” paradigm, which is something the current solutions are still a long way from doing. and analyze their evolution over time. Sometimes this issue is addressed as “A to E” paradigm, which is something the current solutions are still a long way from doing. 3. Proposed SHM System 3. Proposed SHM System The system proposed in this paper—an embedded WSN for continuous structural monitoring—combines some ofinthethis features of the previous implementations and aims to structural overcome The system proposed paper—an embedded WSN for continuous their weak points. A WSN is presented here that is capable directly measuring monitoring—combines some of the features of the previousofimplementations andconcrete aims to properties overcome thanks to the embedding in the node of theofstructure; with their wireless their weak points. A WSNofissensors presented here thatunits is capable directly measuring concrete capability, properties the nodes canembedding be safely integrated in the a building without any safety Thecapability, wireless thanks to the of sensors in node units of the causing structure; with theirissue. wireless configuration also extended to the node recharging, thankscausing to an inductive wireless transfer the nodes canisbe safely integrated in a building without any safety issue.power The wireless module. Furthermore, the analysis is node performed directly on the parameters, tracking their configuration is also extended to the recharging, thanks todesired an inductive wireless power all transfer changes a long period of time is thanks to the directly ultra-low configuration. module. over Furthermore, the analysis performed onpower the desired parameters, tracking all their Once integrated into the structure, node provides local information (e.g., stress values, changes over a long period of time thankseach to the ultra-low power configuration. temperatures etc.) while thestructure, same time allnode nodes, thanks local to the interconnected of the Once integrated intoatthe each provides information (e.g., nature stress values, system and a etc.) datawhile processing giveall thenodes, “health” status of the whole structure. feature is temperatures at the stage, same time thanks to the interconnected natureThis of the system facilitated the absence of give connection cablesstatus and allows for an structure. autonomous continuous control and a data by processing stage, the “health” of the whole Thisand feature is facilitated by of structures. Specific devices used for to an track dynamic and events, e.g., accelerometers, are not the the absence of connection cables and allows autonomous continuous control of the structures. considered in this work due todynamic their triggering andaccelerometers, data recording are characteristics. In particular, in Specific devices used to track events, e.g., not considered in this work the proposed monitoring system, data acquisition frequency is set according to the in-service time of due to their triggering and data recording characteristics. In particular, in the proposed monitoring the structure (e.g., weekly, monthly stress/temperature measurements). the contrary, in the system, data acquisition frequency is for set according to the in-service time of theOn structure (e.g., weekly, case of dynamic properties evaluation, the acquisition periods would be no longer based on defined monthly for stress/temperature measurements). On the contrary, in the case of dynamic properties time intervals rather onperiods random events acquisition start from specific evaluation, thebut acquisition would bewhose no longer based onwould defined time intervals but“triggers”, rather on such as acceleration thresholds. random events whose acquisition would start from specific “triggers”, such as acceleration thresholds. The proposed monitoring strategy overcomes overcomes some some of the limitation of current practice where inspections are areconducted conductedinindifferent differentpoints pointsofofthe thebuilding buildingonly only after damage occurs; physical inspections after thethe damage occurs; in in contrast, interconnected network presented here can predict global consequences on the contrast, the the interconnected network presented here can predict global consequences on the structure structure on real time measurements before achieving a damaged state. Moreover, preventive based on based real time measurements before achieving a damaged state. Moreover, preventive actions actions arepossible made possible by the that multiple measures are available in real In this way, are made by the fact thatfact multiple measures are available in real time. Intime. this way, regular regular time maintenance effective maintenance strategies could be scheduled, resulting in effective and timeand effective strategies could be scheduled, resulting in effective implications for implications for security and economy, therefore for sustainability. A schematic representation security and economy, and therefore for and sustainability. A schematic representation of the whole SHM of the whole SHMin solution solution is shown Figure is 1. shown in Figure 1.

Figure 1. The WSN architecture.

Sensors 2017, 17, 2566

5 of 21

Sensors 2017, 17, 2566

5 of 21

The main element of the SHM is the node, i.e., a compact electronic system installed in the RC structure under test which, as generic application, consists of a reinforced concrete (RC) slab or TheThe main element of the SHM is the i.e., a compact electronic system in the RC column. node is mandated to carry outnode, the required measurements (e.g., stressinstalled and temperature) structure under test which, as generic application, consists of a reinforced concrete (RC) slab or column. thanks to its installed sensors; obtained results are transmitted directly to the hub (referred to as The node is mandated carry that out the required measurements stress and temperature) thanks to concentrator). It’s worthtonoting only stress and temperature(e.g., measurements are considered in this its installed sensors; obtained results are transmitted directly to the hub (referred to as concentrator). study, but other properties might be measured through other devices properly equipped in the It’s worth noting that stress and temperature measurements are considered this study, but other nodes. Moreover, theonly assembly described hereafter has been conceived for itsinimplementation in a properties be measured throughmade otherof devices properly equipped in the Moreover, the pilot small might scale structural component concrete material. However, the nodes. outcomes of the tests assembly described hereafter has been conceived for its implementation in a pilot small scale structural are obtained under specific boundary conditions reproducing common in-service scenarios of whole component made of concrete material. However, the outcomes of the tests are obtained under specific RC structures. boundary conditions reproducing in-service scenarios of whole RC structures. The node includes four unitscommon (Figure 2): the control unit, the transceiver, the inductive charging The node includes four units (Figure 2): the control unit, the transceiver, the inductive system, and the sensors. The sensors, consisting of a load cell (LC) with a full scale range equalcharging to 1 ton system, and(type the sensors. Thesupplied sensors, consisting of a load a full scale range equal 1 ton (≅10 kN) 3141-1T, by Phidgets Inc.) cell and(LC) an with analog temperature sensorto (type (∼ 3141-1T, supplied byTechnology Phidgets Inc.) and analog temperature sensor (type MCP9700, =10 kN) (type MCP9700, supplied by Microchip Inc.), areanintended, respectively, to directly measure supplied by Microchip Technology Inc.), are intended, respectively, directly measure inner stresses inner stresses and temperature of the concrete’s mass in which theyto are installed. The transceiver for and temperature with of thethe concrete’s mass iniswhich they are(supplied installed. The transceiver for Solutions). communicating communicating concentrator a ME70-169 by Telit Wireless The with the concentrator is a ME70-169 by Telit Wireless Solutions). Themandated inductiveto charging inductive charging system, which is (supplied used for recharging the on-board battery supply system, which used for recharging the on-board mandated to supply the node, is a CR18650 the node, is ais CR18650 lithium-polymer (LiPo)battery battery (produced by Panasonic). Finally, the lithium-polymerunit (LiPo) battery (produced Panasonic). Finally, the microcontroller (MCU), microcontroller (MCU), consisting of aby Microchip microcontroller series Pic24, is aunit low-power consisting of a Microchip Pic24,for is a all low-power DSP processor not only DSP processor that not microcontroller only acts as a series controller the modules, but alsothat carries out acts the as a controller for the modules, but also carries outconcentrator, the measurements transmits the written related measurements andall transmits the related results to the with anand ad-hoc protocol results the concentrator, with an ad-hoc protocol written for this purpose. for this to purpose.

Figure 2. The node (uncased).

Figure 2. The node (uncased).

The hub is the natural counterpart of the nodes. It is made up of a PC station and a transceiver, Theconcentrator hub is the natural of the nodes. It is made up of a PC station and a transceiver, and the of the counterpart entire architecture is arranged in a topographically barycentric position and concentrator of thedeployment), entire architecture is arranged topographically barycentric position (withthe respect to the nodes’ taking into account in theatransceiver coverage. As for the nodes, (with respect to the nodes’ deployment), taking into account the transceiver coverage. As for the the transceiver is a Telit ME-70 and is connected to the PC with a USB adapter board (board included nodes, the transceiver is a Telit ME-70 and is connected to the PC with a USB adapter board (board in the Telit Democase kit). The PC is equipped with a LabVIEW application that not only allows the included in of thethe Telit Democase kit). The PC with a LabVIEW application that not only automation system to be supervised viaisa equipped selective wake-up/measure/sleep node sequence, but allows automationofofallthe to be supervised via a selective wake-up/measure/sleep node also thethe management thesystem handshaking routines implemented to avoid collisions between more sequence, but also the management of all the handshaking routines implemented to avoid collisions nodes transmitting simultaneously. between more nodes transmitting simultaneously. The architecture has a star network topology and provides the opportunity of covering not only The architecture has a star network topology and provides the opportunity covering not only the whole structure (internal nodes) from the basement to the roof, but also a of small area covering the whole structure (internal nodes) fromtothe to the roof, butrange also of a small area covering multiple buildings (external nodes), thanks thebasement suitable communication transceiver modules. multiple buildings (external nodes), thanks to the suitable communication range of transceiver Transmission power, low operating frequencies, low bit rate, and intelligent management of the serial modules. Transmission low operating low bit rate, and intelligent communication with an power, ad-hoc protocol allowsfrequencies, a virtually indefinite number of nodes tomanagement be placed in of the serial communication an ad-hoc protocol allows a virtually number of nodes to a range of some hundreds ofwith meters. The system is fully automatic, andindefinite the required interactions with the final user are limited to two situations:

Sensors 2017, 17, 2566

6 of 21

Sensors 2017, 17, 2566

6 of 21

be placed in a range of some hundreds of meters. The system is fully automatic, and the required interactions with the final user are limited to two situations: 1. on the hub, during the analysis of the reports (Figure 3); 1. on the hub, during the analysis of the reports (Figure 3); 2. on the node, during the recharging procedures. 2. on the node, during the recharging procedures. Thanks to the PC application, the user can analyse all the data collected from the hub, both in Thanks to the PC application, the user can analyse all the data collected from the hub, both in graphic and chart form, for both monitored quantities. A data plot on a time base can be made to graphic and chart form, for both monitored quantities. A data plot on a time base can be made to analyse the evolution of the temperature and stress measurements, from seconds up to years, in order analyse the evolution of the temperature and stress measurements, from seconds up to years, in to quickly identify faults and their causes. Information from the nodes can also be plotted, such as the order to quickly identify faults and their causes. Information from the nodes can also be plotted, battery voltage, with all the stored data indexed by the relative node. With the same application, the such as the battery voltage, with all the stored data indexed by the relative node. With the same nodes can be interrogated individually once they are woken up and some configuration parameters application, the nodes can be interrogated individually once they are woken up and some can be set, e.g., the idle/sleep time intervals. In any case, operator participation is all but unnecessary: configuration parameters can be set, e.g., the idle/sleep time intervals. In any case, operator the hub is self-sufficient and operates the network automation. Moreover, a database of collected data participation is all but unnecessary: the hub is self-sufficient and operates the network automation. is retained and its exportation is possible, thus allowing it to be processed at a later stage and/or with Moreover, a database of collected data is retained and its exportation is possible, thus allowing it to third-party software. be processed at a later stage and/or with third-party software. The second interaction between the network and the operator occurs during the maintenance The second interaction between the network and the operator occurs during the maintenance procedures of the nodes, which consist of a recharging session of the on-board battery and take place procedures of the nodes, which consist of a recharging session of the on-board battery and take place in proximity to the node itself. The nodes, at time intervals of about 10–12 months (depending on the in proximity to the node itself. The nodes, at time intervals of about 10–12 months (depending on the measurements frequency), run out the charge accumulated in the battery and need to be recharged. measurements frequency), run out the charge accumulated in the battery and need to be recharged. The recharging session is performed with an induction charging system installed for variable time The recharging session is performed with an induction charging system installed for variable time intervals (two-six hrs in the proposed solution) on the outer surface of the slab in which the node is intervals (two-six hrs in the proposed solution) on the outer surface of the slab in which the node is installed. The same procedure can be used to wake up a node, which, for the majority of time, is kept installed. The same procedure can be used to wake up a node, which, for the majority of time, is kept in a deep sleep state to preserve energy and is therefore not connectable. in a deep sleep state to preserve energy and is therefore not connectable.

Figure 3. The hub terminal. Figure 3. The hub terminal.

To complete the description of the system, some details about the main improvements achieved To complete the description of the system, some details about the main improvements achieved are presented below. are presented below. 3.1. Inductive WPT Charging System 3.1. Inductive WPT Charging System When defining the power source for the nodes, battery-less solutions have been discarded, as When defining the power source for the nodes, battery-less solutions have been discarded, as they they would prevent a WSN-architecture with adequate coverage; the choice for battery powered would prevent a WSN-architecture with adequate coverage; the choice for battery powered solution solution made the design of the recharging system a key factor. Moreover, the embedded made the design of the recharging system a key factor. Moreover, the embedded implementation of implementation of the nodes has highlighted the need for a wireless charging system. RF power the nodes has highlighted the need for a wireless charging system. RF power harvesting systems harvesting systems were discarded, since the low efficiencies evidenced in the state of the art became were discarded, since the low efficiencies evidenced in the state of the art became even worse due to even worse due to the filtering action of the concrete in the presence of high frequency signals. the filtering action of the concrete in the presence of high frequency signals. Approaches based Approaches based on thermoelectric effect, solar power, vibrations, and harvesting proved also on thermoelectric effect, solar power, vibrations, and harvesting proved also unfit, because of unfit, because of very low temperature gradients, need for cables, and insufficient energy provided by the low efficiency, respectively.

Sensors 2017, 17, 2566

7 of 21

very low temperature gradients, need for cables, and insufficient energy provided by the low efficiency, respectively. A wireless, high-efficiency, low-frequency charging system is therefore required to transfer the energy required by the nodes. All the literature about wireless power transfer (WPT) methods, in particular the work by Angrisani et al. [16–20], proposes an inductive WPT method based on very low frequency signals to transfer energy within a range up to 30 cm, suitably outperforming commercial inductive chargers, whose operating range is limited to 5 cm. The system proposed in Figure 4 is a push-pull inverter in the primary circuit and a full bridge rectifier in 17, the2566 secondary circuit. The transmitter circuit (primary) core is an inverter based on7 of the Sensors 2017, 21 principle of a relaxed oscillator. In particular, the MOSFET works in a push-pull configuration, turning on and the twohigh-efficiency, branches of the low-frequency circuit upon thecharging polarity of the signal in the capacitor is installed A off wireless, system is therefore requiredthat to transfer the parallel the coil.by Voltages and currents in the resonant circuit are characterized energy to required the nodes. All the literature aboutinductive/capacitive wireless power transfer (WPT) methods, in by a naturalthe resonant frequency which the capacity of the cellWPT and method mutual inductances of particular work by Angrisani et al.depends [16–20],on proposes an inductive based on very the This frequency been set in thewithin design stage to 100 by cm, suitably setting the values of lowsystem. frequency signals to has transfer energy a range upkHz to 30 suitably outperforming inductance and capacitance. commercial inductive chargers, whose operating range is limited to 5 cm.

Figure Figure4. 4. Inductive Inductive WPT WPT charging chargingsystem. system.

The transmitting system proposed in is Figure is a push-pull inverter the primary circuit a full bridge The device a coil4comprised of four turnsinmade with a 2.5 mm2and stranded cable rectifier in the secondary circuit. The transmitter circuit (primary) core is an inverter based on the wound up to a 14-cm diameter. This is connected to the two branches of the inverter, while the central principle of a relaxed oscillator. tap is connected to the power line. In particular, the MOSFET works in a push-pull configuration, turning on and off the two branches the circuit upon the polarity of the signal intuning the capacitor that The main advantage associatedofwith relaxed oscillators is the automatic capability, is installed parallel to the coil. Voltages and currents in the resonant inductive/capacitive circuit are which allows the automatic following of the maximum power point (MPP), thus guaranteeing characterized by aofnatural resonant frequency whichofdepends on the capacity of the cell andwithout mutual the transmission maximum power. The choice a fixed frequency DC/AC inverter, inductances of the system. This frequency has been set in the design stage to 100 kHz by suitably a feedback auxiliary circuit for the frequency control, would lead to power losses. These are caused by setting thedetuning values ofeffects inductance and capacitance. inevitable (involving a frequency mismatch) linked to: (i) parametric variations of the The transmitting device is a coil comprised of four variations turns made 2.5system mm2 stranded components; (ii) intrinsic tolerances; and (iii) geometrical (inwith bothathe itself andcable the wound up to a 14-cm diameter. This is connected to the two branches of the inverter, while the mutual position of the primary and secondary circuits). central is connected the power line. Thetap receiver circuit istomade symmetrical to the transmitting circuit (in order to have the same The frequency). main advantage associated with relaxed oscillators is the automatic tuning resonant It includes an identical coil (four turns, 2.5 mmq, and 14 cm) andcapability, the same which allows the automatic following of the maximum power point (MPP), thus guaranteeing the capacitor pack. A full-bridge rectifier made with high-frequency diodes completes the proposed circuit, transmission maximum power. choice of a fixed frequency DC/AC inverter, without a thus providingofa very compact form, The which is important for the successive embedment. feedback auxiliary circuit for the frequency control, would lead to power losses. These are caused by inevitable detuning effects (involving a frequency mismatch) linked to: (i) parametric variations of 3.2. Low Power Architecture the components; (ii) intrinsic tolerances; and (iii) geometrical variations (in both the system itself and In order to guarantee a battery life that is long enough to avoid the need for frequent recharging the mutual position of the primary and secondary circuits). sessions, an adequate power design (PCB level) has been developed by working on the power The receiver circuit is made symmetrical to the transmitting circuit (in order to have the same consumption of the components first and, then, on the selective enabling. A great effort has been resonant frequency). It includes an identical coil (four turns, 2.5 mmq, and 14 cm) and the same made in this second stage [21]. From a high-level point of view, the circuitry involved in the supply capacitor pack. A full-bridge rectifier made with high-frequency diodes completes the proposed management can be divided into the following main blocks: circuit, thus providing a very compact form, which is important for the successive embedment. • wireless recharging system, including receiving coil and conditioning circuit; 3.2. Low Power Architecture In order to guarantee a battery life that is long enough to avoid the need for frequent recharging sessions, an adequate power design (PCB level) has been developed by working on the power consumption of the components first and, then, on the selective enabling. A great effort has been made in this second stage [21]. From a high-level point of view, the circuitry involved in the supply

Sensors 2017, 17, 2566

• •

8 of 21

accumulator circuit, composed of a three-stage system made up of input voltage regulator, battery charge regulator, and LiPo accumulator; device voltage regulating circuit composed of the MCU voltage regulator (very low quiescent current to enable ultra-low power standby) and a peripheral voltage regulator (to disable unused external peripherals during the sleep condition).

Joining this low-power circuitry with the deep sleep MCU capability and the Li-Po battery assures very low discharge rates thus allowing very long lifespan to be achieved. The proposed system in the deep-sleep condition achieves measured consumptions lower than 1 µA; in particular, average current values as low as 415 nA have been experienced, if compared to the 150 mA peak values measured during the working state). When considering a realistic scenario with a measurement rate of one measure per week (corresponding to a duty cycle of 0.001% if 5 s average operating time is considered), the whole power consumption is equal to 2.437 mWh/week. This value represents only 1% of the battery self-discharge rate (8–10% capacity/month, i.e., 277.5 mWh/week). In a long-term monitoring configuration, the expected operating cycle can be calculated by making an acceptable approximation in considering the electric consumptions as zero and concerning only the battery shelf life [22,23]. 3.3. MCU Firmware and Communication Protocol The microcontroller unit has two different working states, sleep and wake-up. In the sleep mode, the unit is put in a low consumption state where both external and internal peripherals are electrically disabled (cutting off their supplies), but the real-time clock calendar (RTCC) which stores the wake-up date, and the watchdog timer, which wakes up the unit if a programmed timeout has elapsed (to prevent a misconfiguration that leads to infinite sleep sessions). As stated above, this status is needed to allow a very long battery life and reduce the number of charging sessions [24,25]. When the preset date is reached, the unit exits from the sleep condition and enters the wake-up state. In this condition, all the peripherals are turned on and a three-phase handshake/measureme nt/data transmission session occurs. The first phase is a handshake session where the unit introduces itself in the network, sending a request signal to the concentrator. The message is then repeated (with a random time delay to avoid conflict with other nodes) for a finite number of iterations until an acknowledgement message is received from the concentrator, which contains the information on the number of measurements to carry out and their accuracy. The RTCC is then updated with the current time stamp and the wake-up date. Once the message is received from the node, it carries out the measurement routine, conducting multiple acquisitions to reduce acquisition noise, and then goes back into the sleep mode. If other nodes are asking to log in to the network, each of which having a univocal ID that enables the concentrator to differentiate the requests, the concentrator registers them in a queue and processes their requests with a time priority criterion. An additional condition can take place, even if it does not belong to the regular activity condition, and is represented by a battery recharge session. Once the inductive charging system is turned on and installed in proximity to the node, the start of a recharge leads to a forced node wake-up, which automatically sends a message on the state of the battery charge and performs a quick measurement session. This information is then processed from the concentrator, which notifies the user about the end of the recharging session (and the possibility of removing the charging system). 3.4. Load Cell Sensor and Low Powering Considerations Load cell sensors are widely used in force measurements, and are de-facto utilized most when measuring weights. Their use in the SHM field is relatively new, and this paper moves this idea forward, proposing their utilization in an embedded configuration with a very simple load cell type as a first attempt of this application. The use of LCs is preferred when dealing with large forces, thanks to the wide range of scales available nowadays, even in commercial sensors. There is an increased interest on these sensors, since the cost of even a high range unit, make them an interesting solution for inner stress measurement of

Sensors 2017, 17, 2566

9 of 21

Sensors 2017, 17, 2566

9 of 21

measurement of concrete masses. In this study, a 1000 kg range load cell sensor has been employed. The high required power supply is the main problem associated with the integration of LCs in the concrete masses. Inthis this way, study,aapreliminary 1000 kg range load cell sensor has beeninemployed. Theof high considered nodes; performance assessment the presence lowrequired power power supply is the main problem associated with the integration of LCs in the considered nodes; this supply is mandatory. way, As a preliminary the presence of low power supply mandatory.of the shown in performance detail in theassessment successiveinSection, investigations on the lowis powering As shown in detail in the successive Section, investigations on the low powering ofathe proposed LC (a Phidget 3141-1T) have been conducted by powering the sensor from 3 Vproposed supply. LC (a Phidget 3141-1T) have been conducted by powering the sensor from a 3 V supply. The lower The lower voltage, compared with a 5–18 V minimum supply and a 9–12 V recommended supply, voltage, compared with a 5–18 V minimum andrange. a 9–12Worse V recommended supply, produced produced only a limited reduction of the supply operating uncertainty and noise wereonly not a limited reduction of the operating range. Worse uncertainty and noise were not observed. observed. Further studies enable thethe direct embedding of the in theinconcrete mass. Further studieswere werecarried carriedout outtoto enable direct embedding of sensor the sensor the concrete A metal casing was designed to improve the sensor protective structure in order to avoid unwanted mass. A metal casing was designed to improve the sensor protective structure in order to avoid compression on the sensor’s would have led tohave misreading. As shownAs in unwanted compression on thesensitive sensor’ssurfaces, sensitivewhich surfaces, which would led to misreading. Figure 5, cover consisted two plates, theplates, first one theone bottom to bottom improvetothe structural shown in the Figure 5, the cover of consisted of two theon first on the improve the strength of the bottom lid, and the second one, which is a cap to distribute the force impressed the structural strength of the bottom lid, and the second one, which is a cap to distribute theon force top face, only on the sensor’s button (the sensitive area) and not the upper face. impressed on the top face, only on the sensor’s button (the sensitive area) and not the upper face.

(a)

(b)

Figure Figure 5. 5. Load Loadcell cell sensor, sensor, cased: cased: (a) (a) schematic schematic representation, representation, (b) (b) final final assembly. assembly.

4. Metrological Characterization and Performance Assessment 4. Metrological Characterization and Performance Assessment Different experimental tests were conducted to metrologically characterize the LC sensor and Different experimental tests were conducted to metrologically characterize the LC sensor and assess the performance and effectiveness of the inductive charging system. Three sets of tests were assess the performance and effectiveness of the inductive charging system. Three sets of tests were conducted: one set on the LC in free air; two sets on the inductive charging system, one in free air conducted: one set on the LC in free air; two sets on the inductive charging system, one in free air and and the other in a concrete environment. the other in a concrete environment. 4.1. 4.1. Free Free Air Air Tests Tests 4.1.1. LC LC Characterization LC characterization characterizationincluded included theoffirst of aimed whichat aimed at the assessing the LC twotwo tests,tests, the first which assessing LC effectiveness effectiveness under loads applied in off-specification, low-power As under reference loadsreference applied in off-specification, low-power configuration. As configuration. previously stated, previously stated, major modifications were made to simplify the circuit, including power major modifications were made to simplify the circuit, including power consumption improvements. consumption improvements. A firstof change was the reduction the LC voltage This helped A first change was the reduction the LC voltage supply.ofThis helped to supply. avoid the insertion to avoid the insertion of inefficient DC/AC circuits to step up the inadequate battery voltage (3.7the V, of inefficient DC/AC circuits to step up the inadequate battery voltage (3.7 V, unsuitable for unsuitable for the 12–18 V LC required supply). Consequently, a more efficient system was 12–18 V LC required supply). Consequently, a more efficient system was achieved, reducing the LC achieved, reducing the LC power hypothesis consumption. non-working hypothesis power consumption. A non-working wasAimmediately discarded, sincewas LCs immediately are based on discarded, since LCs are based on Wheatstone bridges, involving resistive sensing elements. On the Wheatstone bridges, involving resistive sensing elements. On the contrary, reduced values both of contrary, botheffects of sensitivity to saturation effectsand close to thevoltages input range values) sensitivityreduced (due to values saturation close to(due the input range values) output (because of and output voltages the ratiometric nature of following the LC sensor) the ratiometric nature (because of the LC of sensor) were expected in the tests. were expected in the following tests.test focused on the eventual artefacts due to the structural redesign of the sensor, i.e., A second A second test focused onLC. theIn eventual artefacts due to structural redesign of the sensor, the metal casing added to the particular, the effects of the redistribution of the acting forces and i.e., the the metal casing added to the LC. In particular, the effects of redistribution of the acting forces and elastic behaviour of the aluminium casing have been assessed. the elastic behaviour of the aluminium have been assessed. machine with 500 kN maximum The test was carried out by means casing of universal servo-hydraulic force capability (Figure 6). The instrumentation included an E3648A PSU (produced by Agilent,

Sensors 2017, 17, 2566 Sensors 2017, 17, 2566 Sensors 2017, 17, 2566

10 of 21 10 of 21 10 of 21

The test was carried out by means of universal servo-hydraulic machine with 500 kN maximum test was(Figure carried 6). outThe by means of universalincluded servo-hydraulic machine 500 kNby maximum forceThe capability instrumentation an E3648A PSU with (produced Agilent, force capability (Figure 6). The instrumentation included an E3648A PSU (produced by Agilent, Santa Clara, CA, USA), an Agilent 34401 DMM, and a breakout board of the same instrument amplifier Santa Clara , CA, USA), an Agilent 34401 DMM, and a breakout board of the same instrument Santa Clara , CA, USA), annamely Agilentan 34401 DMM, and breakout board of the same instrument used on the node, INA826 with a gain of 500 toa maximize the output swing. amplifier used on namely the node, INA826 with a gain of 500 to maximize the output swing. amplifier used on the node, namely an INA826 with a gain of 500 to maximize the output swing.test Four Four different different experimental experimental configurations configurations have have been been considered, considered, including: including: (i) (i) aa reference reference test Four different experimental configurations have been considered, including: (i) a reference test with an uncased LC and standard supply; (ii) an uncased low-powered LC; (iii) a cased LC standard with an uncased LC and standard supply; (ii) an uncased low-powered LC; (iii) a cased LC standard with an uncased and standard supply; anwere uncased low-powered LC; (iii) LC sweep standard supply; and (iv) (iv) aLC a low-powered low-powered cased LC.(ii) Data wereacquired acquired threesteps: steps: a cased continuous sweep supply; and cased LC. Data ininthree a continuous 0– supply; and (iv) a low-powered cased LC. Data were acquired in three steps: a continuous sweep 0– 0–10 kN load (elastic response of the DUT); an ascending 20 steps 0–10.5 kN load; and a descending 10 kN load (elastic response of the DUT); an ascending 20 steps 0–10.5 kN load; and a descending 20 10 kN0–10.5 load (elastic response of the DUT); an ascending 20 steps 0–10.5 kN load; and a descending 20 steps 0–10.5 load (alternate step directions used highlight eventual hysteresis effects). 20 steps kNkN load (alternate step directions areare used to to highlight eventual hysteresis effects). steps 0–10.5 kN load (alternate step directions are used to highlight eventual hysteresis effects).

(a) (a)

(b) (b)

Figure 6. 6. Load Loadcell celltest testset-up, set-up,free free air: experimental setup, including hydraulic press, Figure air: (a)(a) experimental setup, including the the MTSMTS hydraulic press, and Figure 6. Load cell test set-up, free air: (a) experimental setup, including the MTS hydraulic press, and (b) detail of the LC under test. (b) detail of the LC under test. and (b) detail of the LC under test.

Figure 7 shows the evolution of the differences, expressed in relative percentage terms, between Figure 77 shows shows the the evolution evolution of of the the differences, differences, expressed expressed in relative relative percentage percentage terms, terms, between between Figure the LC voltage nominal value (calculated as theoretical value in from the sensor specifications at the the LC voltage nominal value (calculated as theoretical value from the sensor specifications at the the LC voltage nominal value (calculated as theoretical value from the sensor specifications at corresponding output) and the measured one ∆VLC = (VNOMINAL − VMEASURED)/VNOMINAL versus the the corresponding output) and one ∆V∆V = (VNOMINAL )/V versus LC LC NOMINAL corresponding andthe themeasured measured one NOMINAL − − V VtoMEASURED MEASURED)/V NOMINAL versus the applied force. Inoutput) particular, four curves are shown: “10= V(V Ori” refers the bare LC sensor operated the applied force. In particular, four curves are shown:“10 “10V VOri” Ori”refers refers tothe thebare bareLC LC sensor sensor operated operated applied force. In“10 particular, shown: at 10 V Supply, V Mod” four is thecurves cased are sensor @ 10 V supply; similartoconsiderations hold with “3 V at 10 V Supply, “10 V Mod” is the cased sensor @ 10 V supply; similar considerations hold with at 10 and V Supply, “10 Vfor Mod” the cased3sensor @ 10 V supply; similar considerations hold with “3 “3 V V Ori” “3 V Mod” testsisinvolving V supply. Ori” Ori” and and “3 “3 V V Mod” Mod” for for tests tests involving involving 33 V V supply. supply.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

9.5

10.0 10.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

9.5

10.0 10.5

∆V∆V LC [%] LC [%]

∆V∆V LC [%] LC [%]

30% 30% 20% 20% 10% 10% 0% 0% -10% -10% -20% -20% -30% -30% -40% -40% -50% -50% -60% -60%

30% 25% 30% 20% 25% 15% 20% 10% 15% 5% 10% 0% 5%

0% 0.0 0.0

10 V Ori 10 V Ori

0.5

1.0

1.5

0.5

1.0

1.5

Applied Force [kN] Applied Force [kN] Applied Force [kN] Applied Force [kN] 10 V Mod 3 V Ori 10 V Mod

3 V Ori

3 V Mod 3 V Mod

Figure 7. Load cell measured voltage by different applied forces and supplies. “Ori” is the LC sensor Figure 7. Load cell measured measured voltage bycased different applied forces and supplies. “Ori” the sensor Figure 7. Load cell voltage by different applied forces supplies. “Ori” is is the LCLC sensor in in a bare configuration, “Mod” is the one. “10 V” and and “3 V” represent the different sensor in a bare configuration, “Mod” is the cased one. “10 V” and “3 V” represent the different sensor asupplies. bare configuration, “Mod” is the cased one. “10 V” and “3 V” represent the different sensor supplies. supplies.

Sensors 2017, 17, 2566 Sensors 2017, 17, 2566

11 of 21 11 of 21

With regard to mechanical properties, the tests confirmed the negligible effect of the aluminum With regard to mechanical properties, the tests confirmed the negligible effect of the aluminum case (top and bottom cover), as highlighted by the remarkable concurrence of the 10 V and 3 V “Ori” case (top and bottom cover), as highlighted by the remarkable concurrence of the 10 V and 3 V “Ori” and “Mod” curves in Figure 7. Considered results were achieved as average value of 30 repeated and “Mod” curves in Figure 7. Considered results were achieved as average value of 30 repeated measurements for each load condition; to better appreciate the agreement among the different measurements for each load condition; to better appreciate the agreement among the different configurations, an inset showing the magnitude of error bar (equal to 1 experimental standard configurations, an inset showing the magnitude of error bar (equal to 1 experimental standard deviation) associated with each measure is given in Figure 7. deviation) associated with each measure is given in Figure 7. Differently from what expected, tests revealed that reducing the power supply does not affect Differently from what expected, tests revealed that reducing the power supply does not affect the the accuracy of the LC only in a specific range of applied force. A reduction of the sensor’s output accuracy of the LC only in a specific range of applied force. A reduction of the sensor’s output range range has indeed been experienced from 1000 kg to 460 kg; in particular, the nominal full-scale range has indeed been experienced from 1000 kg to 460 kg; in particular, the nominal full-scale range (1000 kg (1000 kg @ 12 ÷ 18 V) is reduced, respectively, to 815 kg (8 kN) @ 10 V and 460 kg (4.5 kN) @ 3 V @ 12 ÷ 18 V) is reduced, respectively, to 815 kg (8 kN) @ 10 V and 460 kg (4.5 kN) @ 3 V (Figure 7). (Figure 7). Anyway, this effect can be easily compensated by choosing an LC with a greater range. Anyway, this effect can be easily compensated by choosing an LC with a greater range. An error on the An error on the lowest part of the scale is also noticed because of the superimposed noise which lowest part of the scale is also noticed because of the superimposed noise which became comparable became comparable with the useful signal, in low output conditions. with the useful signal, in low output conditions. Free air tests aimed also to determine the calibration curve of the LC sensor in a low-power Free air tests aimed also to determine the calibration curve of the LC sensor in a low-power configuration by means of a polynomial interpolation, which has been then transferred and saved in configuration by means of a polynomial interpolation, which has been then transferred and saved the MCU. By using the Matlab curve fitting tool, the calibration curve has been split into three in the MCU. By using the Matlab curve fitting tool, the calibration curve has been split into three regions: (i) a region I belonging to the 0–4.5 kN range, represented with an 8th order polynomial regions: (i) a region I belonging to the 0–4.5 kN range, represented with an 8th order polynomial curve; curve; (ii) a region II which is a linear knee region in the range 4.5–5 kN used to join the region I and (ii) a region II which is a linear knee region in the range 4.5–5 kN used to join the region I and III; III; (iii) a region III with an exponential function, in the range 4.5–10 kN (the saturation region, (iii) a region III with an exponential function, in the range 4.5–10 kN (the saturation region, beyond the beyond the limit of the reduced range). limit of the reduced range). In Figure 8 the three regions of the calibration curve are shown; the black circles indicate the In Figure 8 the three regions of the calibration curve are shown; the black circles indicate measurements from the characterization stage while the red curve is the polynomial curve the measurements from the characterization stage while the red curve is the polynomial curve representative of the region I, the blue one is the exponential curve of the region III and the representative of the region I, the blue one is the exponential curve of the region III and the light-blue light-blue curve represents the linear knee of region II. curve represents the linear knee of region II.

Figure 8. LC calibration curve in the low-power configuration showing the three region curve. Figure 8. LC calibration curve in the low-power configuration showing the three region curve.

The ofof thethe calibration curve fitting (based on both a Matlab evaluation of theofresults and Thegoodness goodness calibration curve fitting (based on both a Matlab evaluation the results a and conclusive cross-check of the calibration curve upon the input data) has shown very good results a conclusive cross-check of the calibration curve upon the input data) has shown very good with a maximum deviationdeviation expressedexpressed in relativeinpercentage terms equal to 0.5% into the region I, which results with a maximum relative percentage terms equal 0.5% in the region represents the proposed working range, and a maximum of 3.5% in the region III. I, which represents the proposed working range, and a maximum of 3.5% in the region III.

Sensors 2017, 17, 2566

12 of 21

4.1.2. Inductive WPT Charging System Characterization The inductive charging system has previously been developed in the work by Angrisani et al. [16–20] and largely tested in free air; these tests have been reproduced as a reference for the subsequent tests in concrete. The test set-up showed in Figure 9 includes a DM3058ETM digital multimeter (produced by Rigol, Beaverton, OR, USA) a Rigol DG1022UTM frequency meter, and a PS-305DTM DC power supply (supplied by Long Wei, Dongguan, China). The following measurements have been obtained and were aimed to consider real situation of charging operations in RC structures considering also different supply situations: # #

#

#

# #

Primary electrical absorption and free running frequency: this test has verified that 110 kHz is a project constraint (measured 110.4 kHz ± 0.1% in all the range), total consumption of 263.1–674.6 mA (load off) and a 6 V minimum power supply were also measured. Secondary open circuit (OC) voltage (coaxial coils): the secondary VOC has been tested by loading the primary with the secondary circuit and varying both the distance (0–35 cm) between the coils and the primary power supply (6, 12, 15 V). The interval of interest of coil distance is approximately 10–15 cm (y direction, Figure 10a), representing a maximum plausible distance between an embedded node and an external charging system. Within this distance range, a minimum of 7 V is achieved by a 6 V supply @ 15 cm and a maximum of 31 V by a 15 V supply @ 10 cm. Secondary OC voltage (misaligned coils): with a primary power supply set to 12 V, the secondary VOC is analyzed in this case by displacing the coils in the transverse direction of coil axis (x direction in Figure 10b) to simulate a recharging operation where (once embedded) a perfect alignment is unlikely to be achieved. The test is conducted considering different axial distances between coaxial coil centers (in y direction Figure 10a up to 35 cm) and different coil alignments in x direction (Figure 10b), keeping axis parallel. In particular, coil center alignments of 100–50–25–0% have been considered where the alignment percentage is calculated as (coil diameter − X distance between coil centers)/coil diameter; the value of 100% corresponds to the coaxial case while 0% is the full misalignment. In the 10–15 cm range of axial distance (y direction), we obtained an average 20% power loss @ a 75% alignment, 40% @ a 50% coils displacement, and a 76% loss @ a 0% alignment, thus confirming the goodness of the system if there is an inevitable misalignment during charging operations. Secondary short circuit (SC) current: the output capacity has been tested by supplying the primary section at 12 V and shorting the secondary circuit; current values as high as 125 mA is reached at 15 cm, 365 mA @ 10 cm, and 678 mA @ 7.5 cm, respectively, thus matching the specifications of the required minimum node supply current (defined as 100 mA in the previous project constraints). Secondary SC current (on voltage regulator): with the same primary power supply, the output capacity of the recharging system is tested here, shorting the output of the onboard node regulator: 321 mA @ 10 cm and 128 mA @ 15 cm. Battery recharging test, node recharging test: final tests have been performed to recreate a recharging session of the node, isolating the controller unit first and then, in a further test, testing the whole node system. At a 12 V primary supply, 333 mA @ 10 cm and 131 mA @ 15 cm are achieved.

In Figure 11a the evolution of the secondary output voltage versus different coils distances is plotted for three different primary supply voltage values (6–12–15 V). In Figure 11b a comparison between secondary ISC (“A sec”) and primary input current (“A pri”) is shown in the case of 12 V primary supply, which proved to be the best candidate for the final setup. Figure 12 shows the sensitivity of the system to coils displacement and misalignment: the four curves at different axis alignment, expressed as percentage values as [(coil diameter−distance between axis centers)/coil diameter %], show the secondary output voltage by different coils distance when primary supply is equal to 12 V. Tests outcomes first confirmed the matching of the built system with the one described in the reference papers [16–20]. Moreover, according to a possible scenario of a recharging operation at

Sensors 2017, 17, 2566

12 of 21

Sensors 2017, 17, 2566 4.1.2. Inductive WPT

13 of 21

Charging System Characterization

The inductive charging system has previously been developed in the work by Angrisani et al. 10–15 cmand distance, tests proved of thehave technique rechargingasthe system [16–20] largely tested in the freeeffectiveness air; these tests been for reproduced a node reference foreven the in case of an alignment percentage down to 25%, which represented the worst case. subsequent tests in concrete. Sensors 2017, 17, 2566 13 of 21 Secondary SC current (on voltage regulator): with the same primary power supply, the output capacity of the recharging system is tested here, shorting the output of the onboard node regulator: 321 mA @ 10 cm and 128 mA @ 15 cm. o Battery recharging test, node recharging test: final tests have been performed to recreate a recharging session of the node, isolating the controller unit first and then, in a further test, testing the whole node system. At a 12 V primary supply, 333 mA @ 10 cm and 131 mA @ 15 cm Sensors 2017, 17, 2566 13 of 21 are achieved. o

Secondary SC current (on voltage regulator): with the same primary power supply, the output capacity of the recharging system is tested here, shorting the output of the onboard node regulator: 321 mA @ 10 cm and 128 mA @ 15 cm. Battery recharging test, node recharging test: final tests have been performed to recreate a recharging session of the node, isolating the controller unit first and then, in a further test, testing the whole node system. At a 12 Vcharging primary system supply,test, 333 mA Figure 9. free air. Figure 9. Inductive Inductive charging system test, free @ air.10 cm and 131 mA @ 15 cm are achieved.

o

o

The test set-up showed in Figure 9 includes a DM3058ETM digital multimeter (produced by Rigol, Beaverton, OR, USA) a Rigol DG1022UTM frequency meter, and a PS-305DTM DC power supply (supplied by Long Wei, Dongguan, China). The following measurements have been obtained and were aimed to consider real situation of charging operations in RC structures considering also (b) different supply situations:(a)

6v

Coil distance [cm] 12v

(a)

15v

30

32.5

35

30

32.5

35

25

27.5

25

27.5

20

22.5

20

22.5

15

17.5

15

17.5

10

10

12.5

5

7.5

5

7.5

0

2.5

12.5

35

30

32.5

25

27.5

20

22.5

15

17.5

10

0.0 12.5

5

7.5

0

0

Current [A]

35

30

32.5

25

27.5

20

22.5

15

17.5

10

12.5

5

7.5

0

2.5

primary ISUPPLY.

0 primary ISUPPLY . 2.5

o

Secondary VOC [V]

Secondary VOC [V]

o

2.5

o

Figure 10. Cell displacement test, coaxial (a) and misaligned (b) configurations. Primary electrical absorption and free running frequency: this test has verified that 110 kHz is a project constraint (measured 110.4 kHz ± 0.1% in all the range), total consumption of 263.1– In Figure 11a the evolution of the secondary output voltage versus different coils distances is 674.6 mA and aprimary 6 V minimum powervalues supply(6–12–15 were also plotted for(load three off) different supply voltage V). measured. In Figure 11b a comparison Secondary open circuit voltage secondary VOCishas beenintested byof loading between secondary ISC (OC) (“A sec”) and(coaxial primarycoils): inputthe current (“A pri”) shown the case 12 V the primary with the secondary circuit and varying both the distance (0–35 cm) between the coils primary supply, which proved to be the best candidate for the final setup. 12 showspower the the 12, system misalignment: four and Figure the primary supplyof(6, 15 to V).coils Thedisplacement interval ofand interest of coil the distance is (a)sensitivity (b) curves at different axiscm alignment, expressed percentage values aasmaximum [(coil diameter−distance approximately 10–15 (y direction, Figureas10a), representing plausible distance Figure 10. Cell diameter displacement test, coaxial and misaligned (b) configurations. between axis centers)/coil show the(a) secondary output voltage this by different between an embedded node and %], an external charging system. Within distance coils range, a Figure 10. Cell displacement test, coaxial (a) and misaligned (b) configurations. distance when primary supply is equal to 12 V. minimum of 7 V is achieved by a 6 V supply @ 15 cm and a maximum of 31 V by a 15 V supply @ In Figure 11a the evolution of the secondary output voltage versus different coils distances is 10 cm. plotted for three different primary supply voltage values (6–12–15 V). In Figure 11b a comparison 50 5.0 Secondary OC voltage coils): withinput a primary supply set to theofsecondary between secondary ISC (misaligned (“A sec”) and primary currentpower (“A pri”) is shown in 12 theV, case 12 V 4.5 45 V OC is analyzed in this casetobybedisplacing the coils in the direction of coil axis (x primary supply, which proved the best candidate for the finaltransverse setup. 4.0 40 Figurein12Figure shows 10b) the sensitivity of the system to coils displacement and(once misalignment: the a four direction to simulate a recharging operation where embedded) perfect 3.5 35 curves at different axis alignment, expressed as percentage values as [(coil diameter−distance alignment is unlikely to be achieved. The test is conducted considering different axial distances 3.0 30 between axis centers)/coil diameter %], showFigure the secondary by different coils between coaxial coil centers (in y direction 10a up to output 35 cm) voltage and different coil alignments distance when primary supply is equal to 12 V. 2.5 25 in x direction (Figure 10b), keeping axis parallel. In particular, coil center alignments of 100–50– 20 25–0% have been considered where the alignment2.0 percentage is calculated as (coil diameter − X 50 5.0 1.5 15 distance between coil centers)/coil diameter; the value of 100% corresponds to the coaxial case 4.5 45 1.0 10 while 0% is the full misalignment. In the 10–15 cm range of axial distance (y direction), we 4.0 40 0.5 5 obtained an average 20% power loss @ a 75% alignment, 40% @ a 50% coils displacement, and a 3.5 35 0.0 76% 0loss @ a 0% alignment, thus confirming the goodness of the system if there is an inevitable 3.0 30 misalignment during charging operations. Coil distance [cm] Coil distance [cm] 2.5 25 Secondary short circuit (SC) by supplying the primary 6v 12v current: 15v the output capacity has been tested A pri A sec 20 section at 12 V and shorting the secondary circuit;2.0current values as high as 125 mA is reached (a) (b) 1.5 at 1515cm, 365 mA @ 10 cm, and 678 mA @ 7.5 cm, respectively, thus matching the specifications Figure 11. Inductive charging system characterization, 1.0 air and coaxial coils (coil distance is in Y 10 of the required minimum node supply current free (defined ascoaxial 100 mA thedistance previous Figure 11. Inductive charging system characterization, free air and coilsin (coil is inproject Y direction of Figure 10a): (a) the secondary V OC at different primary supplies, (b) the ISC compared to 0.5 5 constraints). direction of Figure 10a): (a) the secondary VOC at different primary supplies, (b) the ISC compared to Current [A]

o

Coil distance [cm] A pri A sec

(b)

Figure 11. Inductive charging system characterization, free air and coaxial coils (coil distance is in Y direction of Figure 10a): (a) the secondary VOC at different primary supplies, (b) the ISC compared to primary ISUPPLY.

Sensors 2017, 17, 2566

14 of 21

Tests outcomes first confirmed the matching of the built system with the one described in the reference papers [16–20]. Moreover, according to a possible scenario of a recharging operation at 10– Sensors 2017, 17,2017, 256617, 2566tests proved the effectiveness of the technique for recharging the node system14even Sensors of 21 14 of 21 15 cm distance, in case of an alignment percentage down to 25%, which represented the worst case. Tests outcomes first confirmed the matching of the built system with the one described in the reference papers 35 [16–20]. Moreover, according to a possible scenario of a recharging operation at 10– 15 cm distance, tests proved the effectiveness of the technique for recharging the node system even 30 in case of an alignment percentage down to 25%, which represented the worst case. Secondary VOC [V] Secondary VOC [V]

25 35 20 30 15 25 10 20 5 15 0 10

5

7.5

5

10

12.5 100%

15

17.5

20

22.5

Coils distance [cm] 50% 25%

25

27.5

30

32.5

35

0%

0

Figure 12. Inductive V supply: secondary VOC in the 5 charging 7.5 10system 12.5 characterization, 15 17.5 20 free 22.5air and 25 12 27.5 30 32.5 35

Figure 12. charging free air and 12 V supply: secondary V in coil Inductive displacement test. Coilsystem distancecharacterization, is in Y Coils direction distance of [cm]Figure 10a; displacement is expressed in OC 100% 50% 25% 0% the coil percentage displacement test. Coil distance is in Y direction of Figure 10a; displacement is expressed in values as the X direction of Figure 10b. percentage values as the X direction of Figure 10b. 4.2. Tests in 12. a Concrete Environment Figure Inductive charging system characterization, free air and 12 V supply: secondary VOC in the coil displacement test. Coil distance is in Y direction of Figure 10a; displacement is expressed in percentage values as the X direction of Figure 10b.

4.2. TestsInductive in a Concrete Environment WPT Charging System Characterization

Further tests have been carried out to prove the effectiveness of the inductive charging system Inductive Charging Characterization 4.2.WPT Tests in a ConcreteSystem Environment

in an actual concrete environment. A possible test set-up, which is useful for making multiple

Further tests have carried out to prove the effectiveness of theconcrete inductive charging measurements withbeen the same wireless unit, is represented in one (or more) blocks that are system Inductive WPT Charging System Characterization large enough to avoid fringe effects A on possible the sides oftest the coils of thewhich inductive chargingfor system (Figure in an actual concrete environment. set-up, is useful making multiple Further tests have beenhave carried outimplemented to prove thein effectiveness of set-up: the inductive charging system 13). Multiple configurations been the proposed (i) a set of four concrete measurements with the same wireless unit, is represented in one (or more) concrete blocks that are in an actual concrete environment. A possible test aset-up, which is useful and for making blocks forming a wall of 30 × 30 × 15 cm (to simulate concrete environment); (ii) a 30 ×multiple 30 × 14 large enough to avoidwith fringe effects on the unit, sides of the coils of the inductive charging system (Figure 13). measurements the same wireless is represented in one (or more) concrete blocksofthat are cm concrete block with an embedded 1 mm metal grid (to simulate a worst-case scenario an RC Multipleenvironment); configurations have been implemented in the proposed set-up: (i) a set of four concrete blocks large enough toboth avoid effects on the sides of the coils of the inductive charging system (Figure thefringe VOC and ISC have been measured by varying the misalignment (three step: 0– forming50–100%) a wall of 30 ×power 30 × supply 15 have cm (10 (to simulate a concrete environment); a 30concrete × 30 × 14 cm 13). Multiple configurations been implemented in the proposed set-up: (i)and a set(ii) of four and the step: 6–16 V). forming wall of 30 × 30 ×115mm cm (to simulate environment); and (ii) ascenario 30 × 30 × of 14 an RC concreteblocks block with aan embedded metal grida concrete (to simulate a worst-case cm concrete block with an embedded 1 mm metal grid (to simulate a worst-case scenario of an RC environment); both the VOC and ISC have been measured by varying the misalignment (three step: environment); both the VOC and ISC have been measured by varying the misalignment (three step: 0– 0–50–100%) and the power supply (10 step: 6–16 V). 50–100%) and the power supply (10 step: 6–16 V).

(a)

(b)

(a)

(b)

Figure 13. Inductive charging system test setup, concrete environment: (a) is the test bench set-up, including the required instrumentation; and (b) is a picture of the set-up of the coils.

At 15 cm @ 12 VPRI (coaxial coils), a VOC of 12.722 V is measured in the case of transmission in concrete which, compared to the air case where a 12.726 V is measured, represents a perfect match. Moreover, displacement behavior is analyzed in the same case of 15 cm @ 12 V, with displacements

including required instrumentation; and (b) is a picture of the set-up of the coils. Sensors 2017,the 17, 2566

15 of 21

chargingcoils), systema test concrete (a)in is the the test bench set-up, At 15Figure cm @13.12Inductive VPRI (coaxial VOCsetup, of 12.722 V environment: is measured case of transmission in including the required instrumentation; and (b) is a picture of the set-up of the coils. concrete which, compared to the air case where a 12.726 V is measured, represents a perfect match. Moreover, 17, displacement behavior is analyzed in the same case of 15 cm @ 12 V, with displacements Sensors 2017,At 15 of 21 152566 cm @ 12 VPRI (coaxial coils), a VOC of 12.722 V is measured in the case of transmission in of 0%, 50%, and 100% implicating a reduction of the V OC from 12.722 V to 10.188 V and 3.575 V, concrete which, compared to the air case where a 12.726 V is measured, represents a perfect match. respectively. Asdisplacement stated before, higheris voltages besame achieved primary supply. Moreover, behavior analyzed can in the case ofby15increasing cm @ 12 V,the with displacements of 0%, 50%, and 100% implicating a reduction of the V from 12.722 V to 10.188 V and V, The I SC has also been measured (even if an MPP test should be performed in a future set-up), OC of 0%, 50%, and 100% implicating a reduction of the VOC from 12.722 V to 10.188 V and 3.575 3.575 V, respectively. As stated before, higher voltages can be achieved by increasing the primary supply. confirming the results ofbefore, the previous VOC tests. is abysatisfying concurrence between the respectively. As stated higher voltages can beThere achieved increasing the primary supply. The and IThe also been measured (even anmA MPP test should a future Iair SC has also been measured (even150 if an MPP test should performed ain future set-up), concrete set-up, with respectively and 146 mAbe@be 15performed cm/12 in V/coaxial coils,set-up), and a SC has confirming the results the V OC tests. There is a satisfying concurrence between the confirming of theofprevious tests. There is a satisfying concurrence between the concrete variation inthe theresults misalignment in previous the V three cases of 0%–50%–100%, respectively, of 146 mA, 100 mA OC concrete and air set-up, with respectively 150 mA and 146 mA @ 15 cm/12 V/coaxial coils, and a and 32 air set-up, with respectively 150 mA and 146 mA @ 15 cm/12 V/coaxial coils, and a variation in mA. variation in the misalignment in the three cases of 0%–50%–100%, respectively, of 146 mA, 100 mA the misalignment in the cases of mA,primary 100 mA voltage and 32 mA. Figure 14 shows thethree evolution of 0–50–100%, secondary Vrespectively, OC(a) and ISCof (b)146 versus supply in and 32 mA. Figure 14 shows the evolution of secondary V (a) and I (b) versus primary voltage supplycoil in three different alignment conditions (in X direction) is perfect alignment between OC where 100% SC Figure 14 shows the evolution of secondary VOC(a) and ISC(b) versus primary voltage supply in three different conditions (in X Primary direction) where 100% is perfect alignment between coil centers and 0%alignment is complete current alsoisshown the samebetween conditions. three different alignmentmisalignment. conditions (in X direction) where is 100% perfectin alignment coil centers and 0% is complete misalignment. Primary current is also shown in the same conditions. centers and 0% is complete misalignment. Primary current is also shown in the same conditions. 18

0.20 0.18

16

18

1.25

0.18 0.16

6

12

0.75

0.75

10 8

0.50

0.50

6

4

4

2

2

6

1.00 1.00

0.12

0.12

0.10

0.75

0.10

0.08

0.08

0.50

0.06 0.06 0.04 0.04

0.25 0.25

0.75 0.50

0.25

0.25

0.00 6 6 7 7 8 8 9 910 10 11 11 12 13 12 14 13 1514 1615 16 Primary Voltage [V] Primary Voltage [V] Asec 100% Asec 50% Asec 0%

0.00

0.02 0.02 0.00 0.00

0

0

0.14

0.14

SC

1.00

1.25 1.25

0.16 Secondary ISC I[A] [A] Secondary

8

Secondary VOC [V]

Secondary VOC [V]

10

1.00

14

Primary Current Primary Current [A] [A]

16

14 12

1.25

1.50 1.50

Primary Current [A]

0.20

1.50

Primary Current [A]

1.50 20

20

7 6 87

98

Primary Voltage [V] Primary Voltage [V] Vsec 50% Vsec 50%

Vsec 100%

Vsec 100%

0.00 0.00

9 10 10 11 11 12 1213 1314 141515 1616

Apri 100%

Apri 100%

Vsec 0%

Vsec 0%

Apri 50%

(a)

Asec 100%

Apri 0%

Apri 50%

Asec 50%

Apri 100%

Apri 0%

Apri 50%

Apri 100%

(a)

Apri 50%

Asec 0%

Apri 0%

Apri 0%

(b)

(b)

Figure 14. Inductive charging system characterization, concrete environment: (a) VOC, (b) ISC.

Figure 14. Inductive charging system characterization, concrete environment: (a) VOC ,I(b) I . Figure 14. Inductive charging system characterization, concrete environment: (a) V OC, (b) SC. SC A first goal has been achieved by analyzing the variation of the VOC and the ISC variation of the goal has been beenachieved achievedwhere byanalyzing analyzing thevariation variation the VOC and Ivariation variation of A first has by the ofofthe Vthus OC and thethe ISCthe of the two testgoal set-ups (air-concrete), no differences are evidenced proving SCcomplete transmissivity of the concrete atwhere the used range; as shown in Figure 15.proving the two set-ups (air-concrete), where no differences are evidenced thus proving the the complete two testtest set-ups (air-concrete), nofrequency differences are evidenced thus 20 18 16 14 12 10 8 6 4 2 0

1.50 1.25

1.25

0.75

1.00

0.50

0.75

0.25

0.50

0.00 6

7

8

9

10

11

12

13

14

15

16

0.25

Primary VSUPPLY [V]

0.00 6

7

8

9 Vsec 10 air 11

12

Apri air Primary VSUPPLY [V]

Vsec air Apri air

13 conc 14 Vsec Apri conc

15

Primary ISC [A]

1.50

1.00

Primary ISC [A]

20 18 16 14 12 10 8 6 4 2 0

Secondary VOC [V]

Secondary VOC [V]

transmissivity of the concrete at the used frequency frequency range; range; as as shown shown in in Figure Figure 15. 15.

16

Vsec conc Apri conc

Figure 15. Inductive charging system characterization: secondary VOC and ISC in air and concrete environment.

Even more interesting results have been achieved in the case of the transmission through a sandwich block of 4 cm concrete + 1 mm steel reinforcing grid + 9 cm concrete which represent a realistic scenario where concrete is usually reinforced with embedded steel grids and rebars, whose

Sensors 2017, 17, 2566

16 of 21

Figure 15. Inductive charging system characterization: secondary VOC and ISC in air and concrete environment. SensorsEven 2017, 17, 2566interesting more

16 of 21 results have been achieved in the case of the transmission through a sandwich block of 4 cm concrete + 1 mm steel reinforcing grid + 9 cm concrete which represent a realistic scenario where concrete is usually reinforced with embedded steel grids and rebars, whose at first glance may act as disturbing elements for signal transmission. Normalizing the values by the at first glance may act as disturbing elements for signal transmission. Normalizing the values by the coils distance (Y direction of Figure 10a), the results have shown a great improvement in the power coils distance (Y direction of Figure 10a), the results have shown a great improvement in the power transmission, which can be easily explained considering the model of the system as a transformer. transmission, which can be easily explained considering the model of the system as a transformer. Still referring to the case of coaxial coils and 12 V according to the normalized values, we have Still referring to the case of coaxial coils and 12 V according to the normalized values, we have 0.848 0.848 V/cm in the case of air/concrete, and 1.388 V/cm in the case of reinforced concrete. This behavior V/cm in the case of air/concrete, and 1.388 V/cm in the case of reinforced concrete. This behavior can can be explained by referring to the inductive charging system as an air transformer and not an antenna be explained by referring to the inductive charging system as an air transformer and not an antenna system. By increasing the permittivity of the medium, in this case embedding metallic elements as system. By increasing the permittivity of the medium, in this case embedding metallic elements as meshes or rebars, the power transmission is consequently boosted. meshes or rebars, the power transmission is consequently boosted. Figure 16 shows the secondary VOC and ISC normalized by the distance, as a function of Figure 16 shows the secondary VOC and ISC normalized by the distance, as a function of different different primary supply voltages in the two configurations above discussed: concrete slab (“clc”) and primary supply voltages in the two configurations above discussed: concrete slab (“clc”) and metal + metal + concrete sandwich (“grid”). concrete sandwich (“grid”).

0.020

1.80 1.60 0.015 1.40 1.20 1.00

0.010

0.80 0.60 0.005 0.40 0.20 0.00

Normalized Secondary ISC [A/cm]

Normalized Secondary VOC [V/cm]

2.00

0.000 6

7

8

9

10

11

12

13

14

15

16

Primary VSUPPLY [V] Vsec clc

Vsec grid

Asec clc

Asec grid

Figure 16. Inductive charging system characterization, normalized secondary VOC and ISC in concrete Figure 16. Inductive charging system characterization, normalized secondary VOC and ISC in concrete and reinforced concrete environment. and reinforced concrete environment.

This way, the proposed inductive charging system proved to be a valid candidate for wireless This in way, proposed inductiveConducted charging system proved to be a valid candidate wireless charging an the SHM implementation. tests, in fact, highlighted that nominalfor recharging charging in an SHM implementation. Conducted tests, in fact, highlighted that nominal recharging performance was not affected by the installation in concrete environment. Moreover, the presence of performance not affected by(typically the installation in concrete environment. Moreover, the presence steel elementswas as grid or rebars involved in RC structures) results in outperforming the of steel elements as grid or rebars (typically involved in RC structures) results in outperforming the free air conditions, due to advantageous magnetic coupling with ferromagnetic materials. Finally, free air conditions, due to advantageous magnetic coupling with ferromagnetic materials. Finally, the the influence of coils misalignment (X direction of Figure 10b), if limited to 50%, prove to be not influence of showing coils misalignment (X direction of Figure 10b), if limited to 50%, prove to be not significant, significant, only a limited loss in performance that is almost irrelevant. showing only a limited loss in performance that is almost irrelevant. 5. Combined Mechanical–Electrical Test in a Real Scenario 5. Combined Mechanical–Electrical Test in a Real Scenario Final tests have been performed with the aim of assessing the performance and effectiveness of Final tests have been performed with the aim of assessing the performance and effectiveness of the proposed SHM system in a realistic scenario. Specifically, a 3D printed concrete casing has been the proposed SHM system in a realistic scenario. Specifically, a 3D printed concrete casing has been produced. As shown in Figure 17, the casing has a dimension of 20 cm per side and is printed by produced. As shown in Figure 17, the casing has a dimension of 20 cm per side and is printed by nine nine layers of fast-setting concrete. The cube also has a removable top for further inspection. The layers of fast-setting concrete. The cube also has a removable top for further inspection. The system system was then installed in the case, with consideration given to both the orientations of the was then installed in the case, with consideration given to both the orientations of the antennas and the forces impressed on the block; for these reasons, a concrete pillar was installed to support the LC and let the forces distribute on the active sensing area.

Sensors 2017, 17, 2566

17 of 21

Sensors 2017, 17, 2566

17 of 21

antennas and the forces impressed on the block; for these reasons, a concrete pillar was installed to support the LC and let the forces distribute on the active sensing area.

Sensorsantennas 2017, 17, 2566 and the forces impressed on the block; for these reasons, a concrete pillar was installed to17 of 21

support the LC and let the forces distribute on the active sensing area.

(b)

(a)

(b)

(a)

Figure 17. Concrete sensing element: (a) the device under test, (b) open element showing the

Figure 17. Concrete sensing element: (a) the device under test, (b) open element showing the enclosed system. sensing element: (a) the device under test, (b) open element showing the Figure 17. Concrete enclosed system. enclosed system. The system has been tested in the same environments of Section 4 in order to compare the results within thebeen operative conditions. The system in in thethe same environments of Section 4 in order to compare the results The systemhas has beentested tested same environments of Section 4 in order to compare the The inductive charging system has been tested by measuring both the V OC and the I CC of the within the operative conditions. results within the operative conditions. secondary WPT,charging with the system primaryhas current analyzed. Tests have shown a perfect matchthe with theof the The beenalso tested by both VVOC and I CC The inductive inductive charging system testedeffects by measuring measuring both the the OC and the ICC of the previous results, confirming thehas lackbeen of fringe in the previous tests, and a definitive secondary WPT, with the primary current also analyzed. Tests have shown a perfect match with the secondary WPT, of with primary of current also analyzed. have shown perfect match confirmation the the transparency the concrete to the 100 Tests kHz frequency in thea near field. A VOCwith of the previous results, confirming the lack of fringe effects in the previous tests, and a definitive confirmation previous confirming theA lack of measured fringe effects the previous tests,tests and(50% a definitive 27 V @results, 12 V supply and a 0.776 ISC were by the in cube side. Displacement and of the0% transparency of the concrete theconcrete 100 the near field. Aparameters V 27as VAin @V12 V OC of confirmation the distance) transparency oftothe tofrequency the 100 kHz frequency the near field. OC of of the of radial have confirmed thekHz same amount ofindegradation ofinthe supply and 0.776 A ISC the cube side. Displacement tests(100–50–0%), (50% tests and 0% the previous tests. Figure inby three different alignment conditions theof and 27 Vthe @ 12 Vasupply and a were 0.77618measured Ashows, ISC were measured by the cube side. Displacement (50% radial distance) have confirmed the same amount of degradation of the parameters as in the previous measured secondary V OC (a—”Vsec”) and I SC (b—”Asec”) as a function of different primary supply 0% of the radial distance) have confirmed the same amount of degradation of the parameters as in tests. Figure compared 18tests. shows, three alignment conditions (100–50–0%), the measured secondary voltage, toinprimary current absorption the previous Figure 18 different shows, in three (“Apri”). different alignment conditions (100–50–0%), the

V and IV a function of different voltage, compared to OC (a—”Vsec”) SCOC(b—”Asec”) measured secondary (a—”Vsec”)asand ISC (b—”Asec”) as aprimary functionsupply of different primary supply 45 1.50 1.50 1.50 primary current absorption (“Apri”). voltage, compared to primary current absorption (“Apri”). 40

0.75 20

1.00

15

25

10

20

5

15

0

0.50

0.75 0.25 0.00 0.50 6

7

8

10

9

0 6

7

0.25

Vsec 100%

Vsec 50%

Vsec 0%

Apri 100%

Apri 50%

Apri 0%

8

9

0.00

10 11 12 13 14 15 16

(a)

0.75

1.00

1.25 1.00

0.50

0.50

0.75 0.25

0.25

0.75

0.00

0.50

0.00 0.50 6

10 11 12 13 14 15 16

Primary Voltage [V]

5

1.25

0.75

Primary Current [A]

1.25

Primary Current [A]

25

1.50 1.00

1.00

Secondary ICC [A] Secondary ICC [A]

30

1.00

1.50 Primary CurrentPrimary [A] Current [A]

Secondary VOC [V]

35

Secondary VOC [V]

40

1.50

30

1.25

1.25

1.25 35

45

7

8

9

Primary Voltage [V]

0.25 0.00

10 11 12 13 14 15 16

0.25

Asec 100%

Asec 50%

Asec 0%

Apri 100%

Apri 50%

Apri 0%

6

7

8

9

0.00

10 11 12 13 14 15 16

(b)

Primary Voltage [V]

Primary Voltage [V]

Figure 18. WPT analysis in the embedded system: (a) VOCAsec of the WPT system, Vsec 100% Vsec 50% Vsec 0% 100% Asec(b) 50%ICC.

Asec 0%

Apri 100%

Apri 0%

Apri 50%

(a)

Apri 0%

Apri 100%

Apri 50%

(b)

Figure18. 18.WPT WPTanalysis analysisininthe theembedded embeddedsystem: system:(a) (a)VVOC of of the the WPT WPT system, system, (b) (b) IICC.. Figure OC CC

The LC sensor has been tested in the same electrical working conditions of the developed node system: a 3 V supply with an amplifier gain of 500. The specimen has been tested with an applied force of 0–70 kN both in an ascending and descending ramp.

Sensors 2017, 17, 2566

18 of 21

SensorsThe 2017,LC 17,sensor 2566

has been tested in the same electrical working conditions of the developed18node of 21 system: a 3 V supply with an amplifier gain of 500. The specimen has been tested with an applied force of 0–70 kN both in an ascending and descending ramp. Figure the voltage measured from thefrom embedded load cell (“V”) the(“V”) corresponding Figure1919reports reports the voltage measured the embedded loadand cell and the calculated force (“Up” and “Down”) values, as a function of a 0–70 kN ramp of applied force the two corresponding calculated force (“Up” and “Down”) values, as a function of a 0–70 kNinramp of different loading cases: ascending (“up”) and descending (“down”) ramp. The applied force is the applied force in the two different loading cases: ascending (“up”) and descending (“down”) ramp. 2 area of the concrete element. By properly force imposedforce by the machine ontoby thethe 20 testing × 20 cmmachine The applied is testing the force imposed onto the 20 × 20 cm2 area of the designing the applied load protocol (i.e., load steps and thresholds), it is load possible retrieve the forceit concrete element. By properly designing the applied load protocol (i.e., stepstoand thresholds), exerted ontoto theretrieve load cellthe areaforce on the basis of mechanical of the two is possible exerted onto the loadconsiderations cell area on and the areal basisratio of mechanical components (concrete cover and effective load cell area). Therefore, the expected value is also considerations and areal ratio of the two components (concrete cover and effective load cellplotted area). (green curve) and refers to the resistant force associated with the area of influence of the load cell Therefore, the expected value is also plotted (green curve) and refers to the resistant force associated sensor within the concrete specimen. with the area of influence of the load cell sensor within the concrete specimen. 7.0 2.00

5.0 1.50 4.0 1.00

3.0

Sensor Voltage [V]

Force (Measured) [kN]

6.0

2.0 0.50 1.0 0.00

0.0 1.00

11.00

21.00

31.00

41.00

51.00

61.00

Applied Force [kN] Expected Value

Up

Down

Figure 19. 19. Load Load cell cell analysis analysis of of the the embedded embedded system system during during load load and and unloading unloading phases. phases. Figure

The mechanical test conducted on the embedded WSN system has shown a deviation of force The mechanical test conducted on the embedded WSN system has shown a deviation of force readings limited to a range of ±20% with respect to the applied force after a 10 kN threshold. readings limited to a range of ±20% with respect to the applied force after a 10 kN threshold. Loading and unloading steps have shown little difference, due to the occurrence of concrete Loading and unloading steps have shown little difference, due to the occurrence of concrete irreversible irreversible deformations; a sensor saturation effect has also been noted, confirming the previous deformations; a sensor saturation effect has also been noted, confirming the previous analysis in free analysis in free air where a reduced range has been noticed after the reduction of the sensor’s power air where a reduced range has been noticed after the reduction of the sensor’s power supply. supply. A final test has been performed to analyze the transmissivity of the wireless transmitter which, A final test has been performed to analyze the transmissivity of the wireless transmitter which, as as assumed, shows an irrelevant attenuation in concrete. A brief logging session with the developed assumed, shows an irrelevant attenuation in concrete. A brief logging session with the developed software has been executed, confirming what has been assumed above (Figure 20). software has been executed, confirming what has been assumed above (Figure 20).

Sensors 2017, 17, 2566 Sensors 2017, 17, 2566

19 of 21 19 of 21

Figure 20. Short-term monitoring analysis. Figure 20. Short-term monitoring analysis.

6. Conclusions 6. Conclusions In this paper, a structural health monitoring solution based on a wireless sensor network has In this paper, a structural health monitoring solution based on a wireless sensor network has been proposed aiming to enable the automatic and real-time monitoring of a whole RC structure. been proposed aiming to enable the automatic and real-time monitoring of a whole RC structure. The network is comprised of wireless distributed sensors embedded in the structure by means of The network is comprised of wireless distributed sensors embedded in the structure by means of several several nodes which can be placed in strategic points of the structure itself, such as slabs, beams or nodes which can be placed in strategic points of the structure itself, such as slabs, beams or columns. columns. The monitoring system is comprised of four main units: the control unit, the transceiver, the The monitoring system is comprised of four main units: the control unit, the transceiver, the inductive charging system, and the sensors. In the present study, attention has been focused on a inductive charging system, and the sensors. In the present study, attention has been focused on a force sensor (load cell) and an analog temperature sensor; they are intended to directly measure the force sensor (load cell) and an analog temperature sensor; they are intended to directly measure the concrete’s inner stresses and the temperature of the mass in which they are installed, respectively. concrete’s inner stresses and the temperature of the mass in which they are installed, respectively. Several tests have been conducted on a small-scale sensing concrete elements. The tests concerned Several tests have been conducted on a small-scale sensing concrete elements. The tests load cell sensor and inductive charging system characterization, both in free air and in concrete concerned load cell sensor and inductive charging system characterization, both in free air and in environment, trying to reproduce realistic in-service measurement and charging conditions. The free concrete environment, trying to reproduce realistic in-service measurement and charging air test outcomes have revealed that, for the proposed inductive charging system, the involved values conditions. The free air test outcomes have revealed that, for the proposed inductive charging of voltages and currents are able to guarantee an effective wireless recharging up to 20 cm–15 cm of system, the involved values of voltages and currents are able to guarantee an effective wireless coil distance, even in case of an alignment percentage between coil centers of 25%, which represents recharging up to 20 cm-15 cm of coil distance, even in case of an alignment percentage between coil a very a realistic scenario for in-service recharging operations. In terms of effects of concrete and centers of 25%, which represents a very a realistic scenario for in-service recharging operations. In steel material, it has been found that the proposed inductive charging system is not affected by the terms of effects of concrete and steel material, it has been found that the proposed inductive presence of a concrete medium, but is even more efficient with the presence of a steel element, being charging system is not affected by the presence of a concrete medium, but is even more efficient with that a common situation in case the system is embedded in a RC structure. As final assessment, the the presence of a steel element, being that a common situation in case the system is embedded in a mechanical test conducted on the embedded WSN system revealed maximum of ±20% deviation in RC structure. As final assessment, the mechanical test conducted on the embedded WSN system the force readings compared to the applied force beyond a 10 kN threshold value. revealed maximum of ±20% deviation in the force readings compared to the applied force beyond a The results obtained represent a proof of the concept that should be validated also at the full-scale 10 kN threshold value. of representative RC structures. In particular, other fundamental aspects need to be addressed The results obtained represent a proof of the concept that should be validated also at the for the scale up of the SHM system which are currently under investigation by the authors, such full-scale of representative RC structures. In particular, other fundamental aspects need to be as: sensor miniaturization, structural testing design, RC sample design, data validation over time, addressed for the scale up of the SHM system which are currently under investigation by the mechanical interactions between sensors and concrete medium, dynamic properties assessment, and authors, such as: sensor miniaturization, structural testing design, RC sample design, data validation battery effectiveness. over time, mechanical interactions between sensors and concrete medium, dynamic properties As a future perspective of development, the SHM system presented in this study represents a first assessment, and battery effectiveness. attempt to create an effective interconnected monitoring network based on a large-scale deployment of As a future perspective of development, the SHM system presented in this study represents a the sensors. The real-time analysis of all measurements provided by those sensors has the potential first attempt to create an effective interconnected monitoring network based on a large-scale to make possible a comprehensive assessment of the health of the structure. Thanks to continuous deployment of the sensors. The real-time analysis of all measurements provided by those sensors monitoring, the system would also be able to conduct real-time failure signalling, which was previously has the potential to make possible a comprehensive assessment of the health of the structure. Thanks impossible to achieve with a periodic pre-programmed and time-scheduled analysis/inspections. to continuous monitoring, the system would also be able to conduct real-time failure signalling, which was previously impossible to achieve with a periodic pre-programmed and time-scheduled

Sensors 2017, 17, 2566

20 of 21

An even greater benefit that follows on from this implementation is that automatized monitoring can make early warnings possible thanks to the continuous evaluation of the variations along the time of the parameters that can lead to future failures. In this way, the proposed system contributes to the creation of smart structures according to the requirements of Industry 4.0. Author Contributions: L.A. and D.A. conceived the basic idea of the proposed structural health monitoring solution; L.G. and L.A. designed the electronic components; D.A. and C.M. handled the integration of the sensing elements into structural elements; L.G. and C.M. produced the prototypes, performed the experiments, and wrote the paper; R.S.L.M. and F.B. elaborated the test results and contributed to the paper preparation; F.F. contributed to the mechanical analysis of the tests carried out on the prototype sensing element made of concrete. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4. 5. 6. 7.

8. 9. 10.

11. 12.

13. 14.

15.

16.

Chang, P.C.; Flatau, A.; Liu, S.C. Health monitoring of civil infrastructure. Struct. Health Monit. 2003, 2, 257–267. [CrossRef] Hu, B.; Kundu, T.; Grill, W.; Liu, B.; Toufigh, V. Embedded Piezoelectric Sensors for Health Monitoring of Concrete Structures. ACI Mater. J. 2013, 110, 149–158. Song, G.; Gu, H.; Mo, Y.L.; Hsu, T.T.C.; Dhonde, H. Concrete structural health monitoring using embedded piezoceramic transducers. Smart Mater. Struct. 2007, 16, 959. [CrossRef] Alessandro, S.; Niezrecki, C.; Fortino, G. Wireless MEMS-Based Accelerometer Sensor Boards for Structural Vibration Monitoring: A Review. IEEE Sens. J. 2016, 17, 226–235. Chung, W.H.; Liu, S.Y.; Guan, B.O. Structural monitoring of Tsing Ma Bridge using fiber Bragg grating sensors. Eng. Struct. 2003, 144–146. [CrossRef] Porco, F.; Fiore, A.; Porco, G.; Uva, G. Structural monitoring using fiber optic sensors of a pre-stressed concrete viaduct during construction phases. Case Stud. Nondestruct. Test. Eval. 2014, 2, 27–37. Murthy, S.G.N. Batteryless Wireless RFID based Embedded Sensors for Long Term Monitoring of Reinforced Concrete Structures. In Proceedings of the International Symposium Non-Destructive Testing in Civil Engineering (NDT-CE), Berlin, Germany, 15–17 September 2015. Rainieri, C.; Fabbrocino, G.; Cosenza, E. Integrated seismic early warning and structural health monitoring of critical civil infrastructures in seismically prone areas. Struct. Health Monit. 2011, 10, 291–308. [CrossRef] Rainieri, C.; Fabbrocino, G. Sustainable Design of Smart Health Facilities in Seismically Prone Areas. In Key Engineering Materials; Trans Tech Publications: Zurich, Switzerland, 2013; Volume 569, pp. 278–285. Matteo, C.; Mottola, L.; Picco, G.P. Monitoring heritage buildings with wireless sensor networks: The Torre Aquila deployment. In Proceedings of the 2009 International Conference on Information Processing in Sensor Networks, San Francisco, CA, USA, 13–16 April 2009; pp. 277–288. Sumitro, S.; Hida, K.; Le Diouron, T. Structural Health Monitoring Paradigm for Concrete Structures. In Proceedings of the 28th Conference on Our World in Concrete & Structures, Singapore, 28–29 August 2003. Choi, H.; Choi, S.; Cha, H. Structural health monitoring system based on strain gauge enabled wireless sensor nodes. In Proceedings of the IEEE 2008 5th International Conference on Networked Sensing Systems, Kanazawa, Japan, 17–19 June 2008. Behnia, A.; Chai, H.K.; Shiotani, T. Advanced structural health monitoring of concrete structures with the aid of acoustic emission. Constr. Build. Mater. 2014, 65, 282–302. [CrossRef] Andrés, B.-B.; Zedek, S.F.; Camps, T.; François, R.; Escriba, C.; Fourniols, J.-Y. SHM Based on Modal Analysis: Accelerometer and Piezoelectric Transducers Instrumentation for Civil Engineering in Heterogeneous Structures. In Proceedings of the EWSHM-7th European Workshop on Structural Health Monitoring, Nantes, France, 8–11 July 2014. Branko, G.; Daniele, P.; Daniele, I.; Figini, A. Structural health monitoring method for curved concrete bridge box girders. In Proceedings of the 15th International Symposium on: Smart Structures and Materials & Nondestructive Evaluation and Health Monitoring, International Society for Optics and Photonics, San Diego, CA, USA, 9 March 2008. Angrisani, L.; d’Alessandro, G.; D’Arco, M.; Paciello, V.; Pietrosanto, A. Autonomous recharge of drones through an induction based power transfer system. In Proceedings of the 2015 IEEE International Workshop on Measurements & Networking (M&N), Coimbra, Portugal, 12–13 October 2015; pp. 1–6.

Sensors 2017, 17, 2566

17.

18.

19.

20.

21.

22.

23. 24.

25.

21 of 21

Angrisani, L.; d’Alessandro, G.; D’Apuzzo, M.; D’Arco, M. Enabling induction and wireless power transmission technologies aimed at supplying remote equipment in critical logistic scenarios. In Proceedings of the 2013 IEEE International Workshop on Measurements and Networking Proceedings (M&N), Naples, Italy, 7–8 October 2013; pp. 184–188. Angrisani, L.; Bonavolontà, F.; d’Alessandro, G.; D’Arco, M. Inductive power transmission for wireless sensor networks supply. In Proceedings of the 2014 IEEE Workshop on Environmental Energy and Structural Monitoring Systems (EESMS), Naples, Italy, 17–18 September 2014; pp. 1–5. Angrisani, L.; d’Alessandro, G.; D’Arco, M.; Accardo, D.; Fasano, G. A contactless induction system for battery recharging of autonomous vehicles. In Proceedings of the 2014 IEEE Metrology for Aerospace (MetroAeroSpace), Benevento, Italy, 29–30 May 2014; pp. 494–499. Angrisani, L.; d’Alessandro, G.; D’Arco, M. An experimental energy set-up for wireless battery recharging. In Proceedings of the 2015 IEEE International Instrumentation and Measurement Technology Conference (I2MTC), Pisa, Italy, 11–14 May 2015; pp. 1699–1704. Ramos, A.; Lazaro, A.; Girbau, D. Time-coded chipless sensors to detect quality of materials in civil engineering. In Proceedings of the 2016 9th European Conference on Antennas and Propagation (EuCAP), Lisbon, Portugal, 13–17 April 2016; pp. 1–4. García Diego, F.-J.; Esteban, B.; Merello, P. Design of a hybrid (wired/wireless) acquisition data system for monitoring of cultural heritage physical parameters in smart cities. Sensors 2015, 15, 7246–7266. [CrossRef] [PubMed] Asensio, Á.; Blanco, T.; Blasco, R.; Marco, Á.; Casas, R. Managing emergency situations in the smart city: The smart signal. Sensors 2015, 15, 14370–14396. [CrossRef] [PubMed] Leccese, F.; Cagnetti, M.; Calogero, A.; Trinca, D.; di Pasquale, S.; Giarnetti, S.; Cozzella, L. A new acquisition and imaging system for environmental measurements: An experience on the Italian cultural Heritage. Sensors 2014, 14, 9290–9312. [CrossRef] [PubMed] Leccese, F.; Cagnetti, M.; Sciuto, S.; Scorza, A.; Torokhtii, K.; Silva, E. Analysis, design, realization and test of a sensor network for aerospace applications. In Proceedings of the 2017 IEEE International Instrumentation and Measurement Technology Conference I2MTC, Turin, Italy, 22–25 May 2017. © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).