12
IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 1, MARCH 2012 241 Smart “Stick-on” Sensors for the Smart Grid Rohit Moghe, Student Member, IEEE, Frank C. Lambert, Senior Member, IEEE, and Deepak Divan, Fellow, IEEE Abstract—Rapid increase in electric power demand, intro- duction of RPS mandates, and a push towards electrication in the transportation sector is expected to increase power system stresses and disturbances. To tackle these power system issues and maintain high system reliability, it is essential to have information about the condition of assets present on the grid. Presently, due to the absence of low cost exible grid wide monitoring solutions, complete information of the system is not achievable. This paper deals with the development of a new class of sensors called the smart “stick-on” sensors. These are low cost, self-powered, uni- versal sensors that provide a exible monitoring solution for grid assets. These sensors can be mass deployed due to low cost, need low maintenance as they are self-powered, and can be used for monitoring a variety of grid assets. This paper also presents the details on the network architecture, interoperability and integra- tion, and different design aspects of the stick-on sensor, such as novel energy harvesting techniques, power management, wide operating range, and reliability. It is envisioned that the smart stick-on sensors shall be an enabling technology for monitoring a variety of grid assets and prove to be an essential element of the Smart Grid. Index Terms—AC-DC converters, energy harvesting, smart sen- sors, wireless networks. I. INTRODUCTION E LECTRICITY demand, in the United States, has been on the rise since the last few decades; growing at the rate of 3% per year with an increase in peak load of 1.8% per year [1]. However, transmission investments have been almost stagnant [2]. Low investments on the transmission grid have led to in- creased congestion and loading of lines beyond their thermal capacity [3]. If historical load growth of 3% per year continues to be served by remotely located generation, corrective action will be required to avoid degradation of reliability [4]. Further, around 30 states have adopted the RPS mandates which require renewable penetration of 33% by 2020 in Cali- fornia, 25% by 2025 in Illinois, 20% by 2020 in Colorado, to name a few [5]. With these initiatives, the ability of the grid to meet load growth will be exacerbated. Given the poor spa- tial correlation between low-cost renewables and load centers, peak loading on certain utility asset is expected to increase. A study shows that with the increase in penetration of plug-in elec- tric vehicles in the system, utility assets such as distribution Manuscript received April 04, 2011; revised July 05, 2011; accepted August 14, 2011. Date of publication October 14, 2011; date of current version February 23, 2012. This work was supported by The National Electric Energy Testing Research and Applications Center (NEETRAC) at Georgia Institute of Tech- nology. Paper no. TSG-00138-2011. The authors are with the Department of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA 30332 USA (e-mail: rohit- [email protected]; [email protected]; [email protected]). Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TSG.2011.2166280 transformers are expected to experience increased peak loading which would dramatically reduce their life [6]. Another big con- cern for utility engineers is modifying the traditional protection and control systems to allow power ow in both directions. With an ever evolving grid and energy policies, one of the major challenges seen by the utilities today is maintaining relia- bility and high efciency of the assets on the grid. Moreover, the move towards a Smart Grid requires utilities to have smarter and improved asset monitoring infrastructure. Presently, most of the intelligence is provided through human operations, control, and management. However, almost 45%–65% of senior utility en- gineers are at or close to retirement age [7]. Therefore, utilities are motivated to implement smart asset monitoring architectures that require minimal human interference or make the human in- volvement less intrusive. An intelligent monitoring infrastruc- ture that provides the pertinent system information to asset man- agement software which performs portfolio analysis can help managers and planners schedule maintenance routines. Health monitoring, condition monitoring, asset maintenance, incipient fault analysis, and replacement of assets can be prioritized in this manner. It is envisioned that expert systems that interact with a smart monitoring infrastructure and help utilities in deci- sion making would ensure the highest levels of reliability in the evolving electricity grid. However, implementation of a smart monitoring infrastruc- ture is not possible without truly intelligent and low cost monitoring devices. With the development of sensors that utilize low-cost ultra low-power mote processors and com- munication links to transmit data, a new regime of sensors known as the smart sensors have recently come into existence [8]–[10]. Some evident advantages of wireless communication over wired networks are being realized and the technology shift is becoming conspicuous [11]. Previously, due to stringent constraints of range, signal noise and power, wireless com- munication was not considered as a viable option. However, extensive research in the area of wireless communication has given rise to low power protocols. In earlier systems, the communication link was point to point. Now, with the advent of meshed systems and enabling of networking technologies such as multi-hop link networks, coverage over large area has been made possible [12], [13]. Different network topologies and effective wireless signal routing algorithms have further improved range and reduced power consumption giving rise to protocols like ZigBee Pro. A unication of the state-of-the-art wireless networking architectures with intelligent sensors has enabled development of very low cost solutions to em- power wide area monitoring. The low cost associated with this technology has been the most attractive feature for its implementation in habitat and environmental monitoring [14], [15]. However, in present utility applications, only a few sensors can be categorized as smart sensors [16]. Moreover, the ones 1949-3053/$26.00 © 2011 IEEE

Smart “Stick-on” Sensors for the Smart Grid

  • Upload
    d

  • View
    217

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Smart “Stick-on” Sensors for the Smart Grid

IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 1, MARCH 2012 241

Smart “Stick-on” Sensors for the Smart GridRohit Moghe, Student Member, IEEE, Frank C. Lambert, Senior Member, IEEE, and Deepak Divan, Fellow, IEEE

Abstract—Rapid increase in electric power demand, intro-duction of RPS mandates, and a push towards electrification inthe transportation sector is expected to increase power systemstresses and disturbances. To tackle these power system issues andmaintain high system reliability, it is essential to have informationabout the condition of assets present on the grid. Presently, dueto the absence of low cost flexible grid wide monitoring solutions,complete information of the system is not achievable. This paperdeals with the development of a new class of sensors called thesmart “stick-on” sensors. These are low cost, self-powered, uni-versal sensors that provide a flexible monitoring solution for gridassets. These sensors can be mass deployed due to low cost, needlow maintenance as they are self-powered, and can be used formonitoring a variety of grid assets. This paper also presents thedetails on the network architecture, interoperability and integra-tion, and different design aspects of the stick-on sensor, such asnovel energy harvesting techniques, power management, wideoperating range, and reliability. It is envisioned that the smartstick-on sensors shall be an enabling technology for monitoring avariety of grid assets and prove to be an essential element of theSmart Grid.Index Terms—AC-DC converters, energy harvesting, smart sen-

sors, wireless networks.

I. INTRODUCTION

E LECTRICITY demand, in the United States, has been onthe rise since the last few decades; growing at the rate of

3% per year with an increase in peak load of 1.8% per year [1].However, transmission investments have been almost stagnant[2]. Low investments on the transmission grid have led to in-creased congestion and loading of lines beyond their thermalcapacity [3]. If historical load growth of 3% per year continuesto be served by remotely located generation, corrective actionwill be required to avoid degradation of reliability [4].Further, around 30 states have adopted the RPS mandates

which require renewable penetration of 33% by 2020 in Cali-fornia, 25% by 2025 in Illinois, 20% by 2020 in Colorado, toname a few [5]. With these initiatives, the ability of the gridto meet load growth will be exacerbated. Given the poor spa-tial correlation between low-cost renewables and load centers,peak loading on certain utility asset is expected to increase. Astudy shows that with the increase in penetration of plug-in elec-tric vehicles in the system, utility assets such as distribution

Manuscript received April 04, 2011; revised July 05, 2011; accepted August14, 2011. Date of publication October 14, 2011; date of current version February23, 2012. This work was supported by The National Electric Energy TestingResearch and Applications Center (NEETRAC) at Georgia Institute of Tech-nology. Paper no. TSG-00138-2011.The authors are with the Department of Electrical and Computer Engineering,

Georgia Institute of Technology, Atlanta, GA 30332 USA (e-mail: [email protected]; [email protected]; [email protected]).Color versions of one or more of the figures in this paper are available online

at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TSG.2011.2166280

transformers are expected to experience increased peak loadingwhich would dramatically reduce their life [6]. Another big con-cern for utility engineers is modifying the traditional protectionand control systems to allow power flow in both directions.With an ever evolving grid and energy policies, one of the

major challenges seen by the utilities today is maintaining relia-bility and high efficiency of the assets on the grid. Moreover, themove towards a Smart Grid requires utilities to have smarter andimproved asset monitoring infrastructure. Presently, most of theintelligence is provided through human operations, control, andmanagement. However, almost 45%–65% of senior utility en-gineers are at or close to retirement age [7]. Therefore, utilitiesaremotivated to implement smart asset monitoring architecturesthat require minimal human interference or make the human in-volvement less intrusive. An intelligent monitoring infrastruc-ture that provides the pertinent system information to asset man-agement software which performs portfolio analysis can helpmanagers and planners schedule maintenance routines. Healthmonitoring, condition monitoring, asset maintenance, incipientfault analysis, and replacement of assets can be prioritized inthis manner. It is envisioned that expert systems that interactwith a smart monitoring infrastructure and help utilities in deci-sion making would ensure the highest levels of reliability in theevolving electricity grid.However, implementation of a smart monitoring infrastruc-

ture is not possible without truly intelligent and low costmonitoring devices. With the development of sensors thatutilize low-cost ultra low-power mote processors and com-munication links to transmit data, a new regime of sensorsknown as the smart sensors have recently come into existence[8]–[10]. Some evident advantages of wireless communicationover wired networks are being realized and the technologyshift is becoming conspicuous [11]. Previously, due to stringentconstraints of range, signal noise and power, wireless com-munication was not considered as a viable option. However,extensive research in the area of wireless communication hasgiven rise to low power protocols. In earlier systems, thecommunication link was point to point. Now, with the adventof meshed systems and enabling of networking technologiessuch as multi-hop link networks, coverage over large area hasbeen made possible [12], [13]. Different network topologiesand effective wireless signal routing algorithms have furtherimproved range and reduced power consumption giving rise toprotocols like ZigBee Pro. A unification of the state-of-the-artwireless networking architectures with intelligent sensorshas enabled development of very low cost solutions to em-power wide area monitoring. The low cost associated withthis technology has been the most attractive feature for itsimplementation in habitat and environmental monitoring [14],[15].However, in present utility applications, only a few sensors

can be categorized as smart sensors [16]. Moreover, the ones

1949-3053/$26.00 © 2011 IEEE

Page 2: Smart “Stick-on” Sensors for the Smart Grid

242 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 1, MARCH 2012

which have functionalities of a smart sensor are very expen-sive, large, and bulky. They measure multiple parameters onthe asset which increases cost and complexity of design. Someexamples of sensors present in the commercial domain are Pro-tura’s Powerline Sensor, USi’s Power Donut, ABBs Grid Sync,etc. [17]–[19]. Other sensors recently developed and introducedin the market include Tollgrade’s Lighthouse MV sensors, GridSense’s Line IQ, Sentient monitor, and Grid Sentry’s Line sentryto name a few. These sensors are clamp-on type sensors andtheir application is focused at overhead conductors [20]–[23].Moreover, research in the area of utility asset monitoring fo-cuses on monitoring only select assets, like power lines, trans-former, etc. [24]–[26].All the sensors identified above have one or many constraints

of cost, size, flexibility in implementation, weight, etc., that dis-courages utilities from implementing these on a massive scale.Therefore, it was considered worthwhile to determine utility re-quirements and specification for these smart sensors. The in-formation of interest to utility for an improved monitoring in-frastructure was captured through a survey performed with 14utility partners who also served as advisors in this research.The results of the survey are briefly highlighted: More than

15 monitoring applications for a variety of assets ranging fromconductors, cables, transformers, disconnect switches, shunt ca-pacitors, etc., were identified. Utilities showed interest in im-plementing wireless as opposed to wired communication forasset monitoring. Further, among various parameters to be mon-itored, current, ambient temperature, and asset surface tempera-ture were identified to be important. Utilities also showed desireof having sensors with low maintenance (no batteries), low cost( ), small size ( cm ), low-weight ( kg), flex-ible (clamp-less design) and high reliability. Furthermore, theyalso recommended the sensor to operate in outage conditions atleast once to notify about the outage to the operator.The results of this survey clearly show that there is a strong

interest among utilities for improving the sensing and moni-toring infrastructure. Furthermore, they desire monitoring solu-tions for a variety of applications where traditional sensors ei-ther cannot operate, are not designed for, do not exist and even ifthey exist the design constraints make the conventional sensingapproaches expensive. Therefore, as is evident from this survey,there is a strong need for a sensor that is low cost, small in size,and self-powered.This paper presents a concept of a new class of sensors called

the stick-on sensors. These are smart, low-cost, low-mainte-nance, and self-powered sensors that can be used to monitor avariety of utility assets. The paper has been organized in thefollowing manner. Section II discusses the envisioned sensornetwork architecture, along with the advantages of the wirelesssensing scheme. Section III presents the challenges of makingthe sensor self-powered and provides a solution for the same.Section IV deals with optimal energy harvester design, wherecomparisons of various energy harvester configurations havebeen presented and an optimal design region is investigated.Section V presents a novel power management circuit whichis robust and has a wide operating range. Sections VI andVII present the current and temperature sensing, and wirelesscommunication piece of the developed sensor. Section VIIIpresents experimental testing accompanied with results which

highlight the various advantageous features of the proposedstick-on sensor. Finally, Section IX concludes the paper.

II. SENSOR NETWORK ARCHITECTURE

As the name suggests, the smart stick-on sensors are of lickand stick type, and can be directly stuck on to an asset andbegin autonomous monitoring. These sensors do not even re-quire physical contact with the utility asset for some applica-tions and can be kept close to the utility assets for monitoringvarious parameters of interest.Consider a substation comprising bus-bars, disconnect

switches, cables, shunt-capacitors, and transformers. Moreover,consider that there are distribution lines and transmission linesgoing out/coming into the substation. Further, suppose that tomonitor all the assets 150 stick-on sensors are required in thesubstation. The status of the all these assets can be directlygiven to operators through these intelligent sensors that areplaced on or in the vicinity of the asset. Each sensor moduleoperates as a communication node, and exchanges informa-tion through formation of smaller networks between adjacentworking sensor nodes.In our scenario, the ZigBee communication protocol is used

by the network. ZigBee is a reliable, low-power and low-cost,open standard for wireless personal area networks (WPAN) de-veloped by the ZigBee alliance. It has found applications inhome energy management, automated metering, habitat mon-itoring, etc. It is built on top of IEEE 802.15.4 Media access(MAC) and Physical (PHY) layers and utilizes 2.4 GHz (250kbps), 915MHz (40 kbps), and 868MHz (20 kbps) radio bands.One of the most advantageous features of ZigBee is its self-con-figuring and multi-hop network nature. To enable these features,the network consists of three different types of networking de-vices, namely, coordinator node, router, and end device. In ourscenario, the stick-on sensors act as the end devices.The end devices (stick-on sensors) can be pre-programmed

to transmit data at regular intervals. The sensor remains in thesleep mode for most of the time. During the active mode, thesensor wakes up, performs default assessment routines, sensingoperations, sensed information processing, and transmission ofthe processed information to a nearby node. After receiving ac-knowledgement about reception of transmitted information tothe nearest node (or router), the sensor goes back to sleep again.In the sleep mode, the power requirement of the sensor reducesto a few Ws, while in the active mode it is on the order ofmWs. Therefore, if these sensors are operated with a very lowduty rate, the average power reduces considerably. In our sce-nario, the current and temperature in the utility assets presentin the substation needs to be collected every 10–15 min. Thesesensors can, in theory, operate maintenance free on ambient en-ergy for nearly 20–30 years (another feature highly desired bythe utilities). The operation schema of the sensor is shown inFig. 1.Although, the duty rates for information transmission are rel-

atively low, these sensors have the smarts to respond to emer-gency situations. Under faulted conditions, an asset failure oran event that crosses predetermined thresholds of the sensor,the sensor wakes up and transmits emergency information to acoordinator node such that corrective actions can be taken in an

Page 3: Smart “Stick-on” Sensors for the Smart Grid

MOGHE et al.: SMART “STICK-ON” SENSORS FOR THE SMART GRID 243

Fig. 1. Operation regime of a smart wireless sensor.

expedited manner. Therefore, these sensors prove to be tremen-dously valuable in critical situations.The data transmitted by the end device is time stamped and

is sent over a multi-hopping communication scheme to a coor-dinator through the shorted path. The shortest path algorithmensures low overall energy consumption of the network, andlow latency. Routing loops are avoided through intelligentaddressing algorithms utilizing minimum multi-hops. Thesubstation environment can have electromagnetic interference(EMI) and corona discharge due to high voltage and currentsin the assets. Therefore, to ensure high signal fidelity, receivedsignal strength (RSSI) and percentage signal recovery (PSR),and range, routers are used. The number of routers depends onthe physical architecture of the substation (depends on obstruc-tion and line-of-sight). In our scenario, assume on an averagefor every 10 sensor nodes there is one router. This gives a totalof 15 routers to support the sensor network operation. Therouter can communicate directly to the coordinator or direct thesignal via other routers to the coordinator depending on whichoption provides the shortest path.Finally, a master node serves as the ZigBee coordinator

(data-collector). The coordinator receives data from all the enddevices through an indirect or a direct link. In theory, con-necting a wireless sensor network (WSN) with Internet protocol(IP) devices can be performed in three ways: through a proxyserver, delay tolerant network (DTN) gateway, or directly usingTransmission control protocol (TCP/IP) suite [27]. The ZigBeecoordinator can be connected directly to Supervisory controland data acquisition (SCADA) system or to internet protocol(IP) devices through a ZigBee gateway device (ZGD) or aZigBee bridge (or expansion) device (ZED). On the one hand,the gateway acts as a mediator between the ZigBee networkand the IP device through an abstraction interface. While onthe other hand, a ZED extends a ZigBee network over the IPnetwork. Although there are certain fundamental differencesbetween the two approaches, they have their own pros and cons[28]. The stack diagram of ZigBee gateway as outlined by theZigBee alliance is given in Fig. 2. A link between the ZigBeenetwork with the IP devices through a gateway (or a bridge)allows interoperability of the wireless sensor network withother standards. Furthermore, it allows direct binding with theSCADA system which has tremendous value for utility. ZGDare already available in the commercial domain and are used tointerface to existing utility SCADA systems [29]. In our sce-nario, the coordinator has a direct link with the gateway. This

Fig. 2. Stack diagram of ZigBee gateway device [28].

Fig. 3. Network architecture and integration to SCADA.

way, all the intelligent information collected by the coordinatoris directly sent to the SCADA system over the internet.The discussion till now has defined the network architecture

of an intelligent WSN based monitoring regime in a substationenvironment. The network architecture is shown in Fig. 3. It can,however, be implemented even on a bigger network (such as atransmission grid).It was mentioned that one of the major advantages of ZigBee

protocols was that these networks have a self-organizing andself-healing nature. When the coordinator initiates commandsfor formation of the network in the beginning, the network or-ganizes itself to form a star, mesh, or cluster tree architecture.Moreover, if a particular router or an end device (or devices)fails, the signal is rerouted through another efficient path whichminimizes latency time, and further improves reliability of theoverall network.Although at a system level, network characteristics such as

low maintenance, low implementation cost, self-organization,self-healing, and wide coverage area prove to be advantageousand makes the wireless sensing architecture a lucrative solution[11]. However, consider a typical commercial utility sensor thatcosts around $5000 (this number is based on industry interac-tions and market research). To install 150 sensors, 15 routers,and 1 coordinator along with a ZGD (connecting the WSNto SCADA through the internet), one would require nearly$800 000 as cost of only the sensors, excluding installation,

Page 4: Smart “Stick-on” Sensors for the Smart Grid

244 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 1, MARCH 2012

calibration, and other O&M expenditure. An investment closeto a million dollars on installing sensors on all utility assetsin only one substation has (and will) discourage utilities fromtheir massive deployment.Hence, the main challenge is to drive down the cost of sensor

without compromising its attractive features. Fundamentally,there are numerous design challenges that have to be solvedto make these sensors an effective and attractive technology.The rest of the paper focuses on design and development of thesmart stick-on sensors. For this study, a current and tempera-ture sensor is considered and the monitoring asset considered isa conductor. However, the design challenges identified and thesolutions provided are generic and can be applied to a variety ofsensing solutions. Moreover, the designed sensor can be usedfor monitoring many different assets also.

III. POWER REQUIREMENT OF SMART SENSORS

A study of power requirements of some of the existing lowcost, low power sensors available in the commercial domainperformed in [30] showed that most of the wireless motes re-quire batteries for operation. Consider a typical scenario wherethe active mode power requirement of the sensor is 25 mA andthe sleep mode requirement is 100 A. The sensor stays in theactive mode for 1 s and operates after every 10 min. If the sensoris sourced with a 3.7 V, 600 mAh battery, then it can be shownthat the battery will not last for more than 180 days (nearly halfa year). Therefore, relying on batteries for powering these sen-sors would not be a feasible solution for implementing a meshnetwork on the power grid.Considering that utility assets are surrounded by different am-

bient sources of energy, e.g., magnetic and electric fields, solarenergy, vibrations, etc., conceptually the sensors can be poweredfrom the various ambient sources and be self-sufficient. It wasshown in [30] that of all the possible sources magnetic fields andin some cases solar energy prove to be the best candidate sourcefor harvesting energy in the utility space.Although magnetic fields are a candidate source of power, at

lower asset currents, the energy in the fields is not sufficient topower these sensors. An experiment was conducted to test theeffect of decrease in current on the energy that can be harvestedusing an open core Energy harvester (EH). The EH was builtusing Cold rolled grain oriented (CRGO) silicon steel lamina-tions with 300 turns of a 28 AWG magnetic wire wound aroundit. The current carrying asset considered for this experiment wasan ACSR drake conductor. The dimensions of the EH are shownin Fig. 4. The variation in maximum power that can be harvestedfrom the magnetic field near a current carrying conductor usingthe EH is shown in Fig. 5. Another experiment was conductedto test the effect of increase in distance of the EH from the assetand the results are shown in Fig. 6.Fig. 5 shows that the EH, when stuck on a utility conductor,

has the ability to scavenge nearly 116mW at 900 A of conductorcurrent. However, when the current in the conductor is low, say100 A, the maximum power harvested is only 1.25 mW. In ad-dition, even at higher currents, the maximum scavenged powerreduces dramatically with an increase in distance of the flux con-centrator from the conductor. For instance, in Fig. 6, when theflux concentrator is 9 inches from the conductor, even if the con-ductor carries 900 A, the maximum scavenged power is only 3

Fig. 4. Magnetic field energy harvester. (a) Magnetic field energy harvester(EH). (b) EH stuck-on to a conductor.

Fig. 5. Variation of maximum harvestable power with distance of the EH fromthe conductor.

mW. Power on this order is clearly not sufficient for poweringsensor nodes. Therefore, optimal design of the EH is essentialto ensure enough power to operate a smart sensor node reliably.

IV. DESIGN OF ENERGY HARVESTER

Flexibility of use and installation on various assets is one ofthe major requirements for the sensors as outlined by the util-ities. Thus, to ensure this flexibility in design, it is essential tohave a small and open core EH, such that the EH does not needto clamp around the utility asset. Since size and weight is a con-straint, the EH needs to be designed optimally to ensure max-imum energy density.

A. A Systematic Core Design Technique

The EH does not form a loop around the asset and it is ex-pected to have effects such as fringing and field distortion at theedges. Therefore, a finite element analysis software (Maxwell3D) was used to analyze the maximum energy that can be har-vested using different open core structuresAn exhaustive search comprising of different core geome-

tries shaped as bold I (as shown in Fig. 7(a)) were tested. Theasset considered for all the simulations was an ACSR drakeconductor. The bold I structure for the core was selected as itconcentrates the magnetic flux produced by the current carryingconductor and directs it through a channel to maximize the har-vested energy. A subset of the different configurations that weretested is outlined in Table I. The results of the simulations forthese configurations are depicted in Fig. 8. (Note:When the coreis rotated by 90 in its own plane, the misaligned configurationis obtained.)

Page 5: Smart “Stick-on” Sensors for the Smart Grid

MOGHE et al.: SMART “STICK-ON” SENSORS FOR THE SMART GRID 245

Fig. 6. Variation of maximum harvestable power with conductor current.

Fig. 7. EH core, winding diagram, and the simulation setup. (a) Flux concen-trator; material: electrical steel, lamination stacking factor: 0.95. (b) Winding;material: copper, number of turns: 300. (c) Maxwell 3D simulation model; con-ductor: ACSR Drake. (d) Simulation screenshot showing magnetic field inten-sity through the EH.

Fig. 8. Power density of different flux concentrator configurations.

In Fig. 8, it can be observed that all the I-shaped cores havemore power density than an x-shaped core. Moreover, the Com-pact I-core provides the maximum power density than all theother structures. Therefore, this I-structure was selected for fur-ther analysis.Further, it was observed that the voltage induced at the EH

terminal is on the order of a few mVs, for instance, at 50 A itwas 125 mV . Therefore, it is required to connect the EH

TABLE IDIMENSIONS FOR THE DIFFERENT EH CORES

Fig. 9. Equivalent circuit of the flux concentrator and transformer combination.

Fig. 10. Winding geometry.

with a step-up transformer to convert the voltage to levels are atleast above the threshold voltages of diodes and semiconductordevices. The equivalent circuit of the flux concentrator and EHis shown in Fig. 9.To obtain a close to optimal design, it is necessary to calcu-

late the turns on the flux concentrator and transformer that givethe maximum power transferred from the magnetic field to thesensor.

B. Winding Design – Resistance Computation

In this section modeling of resistances and inductances of theEH and the transformer has been performed.Consider a cross section of the EH winding as shown in

Fig. 10.Let the turns along length be , and turns along length

be . Then, total turns are given by (1).

(1)

Given that radius of the wire is , and can be expressed as(2).

(2)

Using (1) and (2), values of and in terms of , andcan be obtained, and are given in (3).

(3)

Page 6: Smart “Stick-on” Sensors for the Smart Grid

246 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 1, MARCH 2012

TABLE IICROSS SECTION AREAS

The areas of cross section of various elements in the windingare shown in Table II.Also,The fill factor can be simplified using (2) and Table III, and a

fixed value 0.78. Further, Resistance of the winding is given by(4).

(4)

In (4), needs to be computed to calculate the value of re-sistance of the winding. Consider the winding to be cuboidal, asshown in Fig. 7(a), with filleted edges. The length of the windingis given by (5).

(5)

In (5) and correspond to length and width of the cuboidalslot of the core. On computation of the summations in (5), thevalue of is given by (6).

(6)

Using (2) and and Table III, can be expressed in terms of ,and as (7).

(7)

Finally, using (4), (6), and (7) the resistance can be expressed as(8).

(8)

It can be seen in (8) that the resistance is dependent on thenumber of turns for a given volume of winding. Equation (8)was used to compute the resistance of the EH and transformerwindings for the different configurations that were analyzed.

C. Winding Design – Inductance Computation

Since the cores have an open structure, the values of induc-tances cannot be accurately modeled. Thus, Maxwell 3D wasused for this purpose. The value of inductance ( ) for the Com-pact I-core is given in (9).An E-core transformer with inner and outer winding as the

primary and secondary was considered for analysis. The trans-former dimensions are given in Fig. 11. The inductances ( ,and ) of the transformer for inner and outer windings are

given in (9) and (10).

nH nH (9)

nH nH (10)

Fig. 11. Transformer core and winding geometries (all dimensions in mm). (a)E-core dimension. (b) Inner winding dimensions. (c) Outer winding dimensions.

D. Winding Design—Maximum Power Transfer

Finally, using the concept of maximum power transfer, max-imum power was computed for different EH and transformerwinding configurations. The Thevenin equivalent of the circuitshown in Fig. 9 was used for computation. Thevenin equivalentvoltage and impedance are given in (11), (12), and (13).

(11)

(12)

(13)

In (11) and (12), , and are computed using (8),while, , , and are computed using (9) and (10). If theload is purely resistive, the maximum power that is transferredto the load is given as (14). While, if the load is resistive andinductive, the maximum power is given as (15).

(14)

(15)

A locus of the maximum power for all the configurationshas been plotted in Fig. 12. An interesting feature about thisfigure is the presence of a region of maxima among all the max-imums. This maxima region shows that there are many solutionsto choose from, thereby giving enough design flexibility. For in-stance the design corresponding to 1 turn would not be feasibleas it would give a very low magnitude of voltage induced on theEH windings, so a design with more number of turns which alsolies on the maxima region can be chosen.The above analysis was performed at a fixed current. How-

ever, with varying currents the harvested power for any config-uration increases monotonically as shown in Fig. 13. Thus, adesign that is optimal at amps is also optimal at amps forall .These design curves can be used to obtain an optimal design

of the EH and transformer combination and would ensure max-imum power transfer which is an essential requirement for pow-ering the smart stick-on sensors.

V. POWER MANAGEMENT CIRCUIT DESIGN

The power that is harvested by the EH is ac power whilethe sensor uses dc for its signal conditioning circuits, micro-

Page 7: Smart “Stick-on” Sensors for the Smart Grid

MOGHE et al.: SMART “STICK-ON” SENSORS FOR THE SMART GRID 247

Fig. 12. Design curve for maximum harvestable power at 100 A primary cur-rent. (a) Maximum harvestable power plots for different designs in the case of aresistive and inductive load. (b) Maximum harvestable power plots for differentdesigns in the case of a resistive load.

Fig. 13. Monotonous increase in maximum harvested power with an increasein current at , and .

processor, and radio. Thus, the ac power has to be transformedefficiently into dc. As mentioned previously, that the Open cir-cuit voltage (OCV) of the EH at low currents was very low. ThisOCV is not even sufficient to overcome the forward thresholdof any diode or switch. One way to deal with this problem is toincrease the number of turns on the winding of the EH, whichwould essentially increase the OCV. However, this would leadto increase in the size of the EH and increase in the losses dueto additional resistance introduced due to extra turns. As men-tioned in the previous section, a transformer has been used tostep – up the voltage. However, due to practical voltage stressconstraints and the fact that the transformer winding selectionis itself a part of the EH design process, the step up function ofthe transformer is constrained. Therefore, there is a requirementof a direct ac-dc boost converter.

Fig. 14. Conceptual circuit of the 0.2 Vac to 3.3 Vdc boost converter.

A simple way to achieve this is with the use of ac-dc bridgerectifiers followed by dc-dc boost converter. However, this ap-proach leads to multiple energy conversion stages which in-crease losses in the system. Recently, some direct ac-dc boostconverters have been proposed [31], [32]. However, these cir-cuits require batteries for start-up and therefore are not in con-formance with the utility requirements. Due to all these con-straints it was deemed preferable to formulate a new power elec-tronics topology for these smart sensors such that the powermanagement systems can be made highly reliable and efficient.The authors proposed a novel patent pending approach for

a direct ac to dc boost converter which has the capability ofboosting voltages from 0.2 to 3.3 V in [33]. The con-ceptual circuit of the converter is shown in Fig. 14.One of the novelties of the circuit lies in the use of the EH

inductance and transformer leakage inductance as an energytransfer element. This eliminates the need for a separate in-ductor in the system thereby reducing the size of the powermanagement system. Another piece of novelty lies in the con-trol, in that the switches are controlled in such a manner thatthe boost functionality is realized for both positive and negativeinput voltages. Therefore, the circuit operates directly with anac source. In addition, a unique characteristic of the circuit isits high voltage boost capability. There are some other uniquefunctionalities of the proposed circuit which are highlightedin Section VIII. The operation of the converter is depicted inFig. 15.

VI. CURRENT AND TEMPERATURE SENSING

Different utility assets have different monitoring require-ments. However, it was seen from the utility survey that currentand temperature are one of the important parameters and areuseful in many different assets. Therefore, the smart stick-onsensor developed in this research focused at temperature andcurrent monitoring. However, other parameters specific to anyother asset can also be monitored using the proposed sensor.

A. Temperature Sensing

Temperature measurement can be performed by using either,RTDs, thermistors, thermocouples, or temperature transducerICs. Every technique has its own pros and cons. The smartstick-on sensor module uses a temperature transducer IC havinga temperature range of to 155 . The fabricated sensormeasures the asset and ambient temperature.

B. Current Sensing

Since, the EH is an open core, it was found through experi-mental tests that the OCV induced in the EH is highly linear with

Page 8: Smart “Stick-on” Sensors for the Smart Grid

248 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 1, MARCH 2012

Fig. 15. Modes of operation of the ac-dc boost converter (arrows indicate current flow, dotted paths carry no currents).

the primary current. This feature is highly desirable for utilitycurrent sensors, and is unique when compared to most of theclosed core sensors which saturate at high currents.However, when using OCV as an indicator of the current, the

values of OCV can be very high at higher currents. This highvoltage appears across the MOSFET switches and diodes of thepower circuit and may damage them. A simple solution to thisproblem is the use of short circuit current (SCC) for currentmeasurement. The way this is achieved is by shorting the EHwindings through a current measuring precision resistance andmeasuring the voltage across the resistor. It should also be notedthat, at the time of sensing, the power circuit has to be isolatedfrom the sensor load otherwise the measurements will be incor-rect.A unique feature about the smart stick-on sensor is that it uses

the EH for both current measurement and energy harvesting,thereby ensuring a compact design.

VII. WIRELESS COMMUNICATION

The stick-on sensor uses the ZigBee network for wirelesscommunication. An experiment was performed to test the fi-delity of signal transmission in the presence of high currents.The test showed high fidelity and signal strength even in thepresence of high currents (up to 1000 A). Results are shown inTable III.Another test was performed to test the range of transmission

in an open area. Tests showed a range of nearly 500 m in thepresence of trees and buildingswith an acceptable level of signalstrength. The results of this test are shown in Table IV.Finally, the functional block diagram of the stick-on sensor is

shown in Fig. 16.

VIII. EXPERIMENTAL TESTING AND RESULTS

A smart stick-on sensor was developed to show functionali-ties such as autonomous power up, reliable operation over widerange, data rates, low power consumption, low-cost, etc. Thesehave been highlighted in this section.

TABLE IIIFIDELITY OF SIGNAL TRANSMISSION IN THE PRESENCE OF HIGH CURRENTS

TABLE IVRANGE OF COMMUNICATION

Fig. 16. Functional block diagram of the stick-on sensor.

A. Power Circuit Fabrication

The fabricated power circuit and signal conditioning boardis shown in Fig. 17(a). TI’s CC2530 SOC solution for ZigBee,

Page 9: Smart “Stick-on” Sensors for the Smart Grid

MOGHE et al.: SMART “STICK-ON” SENSORS FOR THE SMART GRID 249

Fig. 17. Stick-on sensor prototype. (a) Top view of the prototype. (b) Bottomview. (c) Smart stick-on sensor module.

TABLE VFLUX CONCENTRATOR AND TRANSFORMER PARAMETERS

TABLE VIPOWER CIRCUIT PARAMETERS

which comprises a high performance and low power 8051 mi-crocontroller core, 256 kB flash, 8 kB RAM, 12 bit ADC with8 channels and a ZigBee transceiver, was used. The power andsignal conditioning board was built with headers to directly con-nect the CC2530 as shown in Fig. 17(b). The prototype of thesmart stick-on sensor is shown in Fig. 17(c).The EH shown in Fig. 4 was used for testing the prototype.

The equivalent circuit parameters of the EH and the transformercombination are shown in Table V. The components used tobuild the power converter are presented in Table VI.

B. 0.2 V-3.3 V Boost Functionality of the Power Circuit

As discussed previously, the proposed converter can boostvoltages as low as 0.2 V ac to 3.3 V . The boost operationof the converter at different duty cycles is shown in Table VII.The voltage and current waveforms at different points on thecircuit are presented in Fig. 18. Further, the efficiency of the

Fig. 18. Screenshot showing –flux concentrator voltage (0.2 V/Div.),–transformer secondary voltage (2 V/Div.), –flux concentrator current

(10 mA/Div.), –transformer secondary current (2 mA/Div.), and–diode currents (1 mA/Div.).

TABLE VIICONVERTER OPERATION AT DIFFERENT DUTY CYCLE,

, , kHz

power circuit was calculated and was found to be close to 75%at 200 A primary current.

C. Black-Start Capability

One of the major functionalities required by the utility forsmart sensors is to have the sensors start automatically after anoutage condition. Consider a situation when the utility asset car-ries no current and therefore has no magnetic field around it. Inthis situation there would be no output dc supply on the powercircuit. Suppose, after some time the utility asset starts carryingcurrent again. Since there was no dc supply on the power circuitto begin with, no gating pulses would be generated. In the ab-sence of gating pulses, the voltage boost functionality will notbe realized and the sensor would not operate. Therefore, gen-eration of gating pulses for the MOSFET switches is a majorchallenge for the power circuit. Most of the power managementcircuits present in the literature require batteries for start-up andtherefore cannot be used for utility applications.We have solved this problem by using a push-pull circuit that

builds the output voltage and gating pulses using a positive feed-back. An astable multivibrator having a wide operating rangewas used to realize this functionality. The astable multivibratoressentially creates a boot strapping system that builds up theoutput voltage with even a small amount of input voltage. Theoverall power circuit was able to autonomously power up at pri-mary currents as low as 60 A, as shown in Fig. 19. It can be seenthat when the primary current is increased from 15 A to 60 A,the dc voltage is built up from 0.4 Vdc to 3.3 Vdc.

D. Reliable Operation Over a Wide Range

Another major requirement by the utilities for the smart sen-sors is high reliability of operation. The proposed power circuithas a zener diode at the output side which clamps the voltageto be within the safe operating area (SOA) of the MOSFETswitches. The sensor has been tested for current measurementfrom 60 A to a 1000 A and the power circuit works reliably over

Page 10: Smart “Stick-on” Sensors for the Smart Grid

250 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 1, MARCH 2012

Fig. 19. Black-start functionality: –output voltage (1 V/Div.), –Fluxconcentrator voltage (0.2 V/Div.), Time base–(2.5 Sec/Div. ).

Fig. 20. Circuit operating at 60 A and 1000 A primary current, –OutputVoltage (2 V/Div.), SW1–MOSFET Switch 1 Voltage (5 V/Div.).

the entire range. Fig. 20 shows experimental results depictingflawless operation of the power circuit at 60 A and 1000 A ofconductor current. It can be observed that in both the cases, thepower circuit maintains a stable 3.3 V dc supply for the sensor.

E. Operation in an Outage

Another requirement, as outlined by utilities, is the operationof the sensor in an outage. Since, utilities need the sensor tooperate at-least once to inform about the outage, the energystorage requirements are dramatically reduced. The smartstick-on sensor uses a 1 F ultra-capacitor as the backup source.Using an ultra-capacitor the sensor can operate at least 13 timesafter an outage. Considering a 10 min interval between twomeasurements, the sensor can keep operating even 2 hours afterthe outage.Due to the outage the ultracapacitor discharges completely

after some time. When power resumes, the ultracapacitor ap-pears as a low impedance load and restricts the circuit to developthe required voltage at the output. Thus, it cannot be connecteddirectly across the dc bus. We tackled this problem by providinga constant current charging to the ultracapacitor. Overchargingwas avoided by clamping it to the bus voltage through diodewhen enough energy is available in the field. The ultracapac-itor charging circuit is shown in Fig. 21. The capacitor chargingequation is given as (16).

(16)

The overall power management circuit is shown in Fig. 22.

F. Power Consumption of the Sensor

The active mode and sleep mode current consumption wererecorded and have been presented in Fig. 23. The sensor re-quires on an average 0.4 mW during the sleep mode and 39 mWduring the active mode (transmission/reception/wakeup). Also,the sensor takes nearly 400 ms to wake up, communicate withthe coordinator, sense temperature and current, transmit sensed

Fig. 21. Ultracapacitor charging circuit.

information to the coordinator, and then sleep again. A sum-mary of power consumption of every component is shown inTable VIII.The sensed data was transmitted over the ZigBee network to

another CC2530 module which acted as the coordinator nodeand was connected to a laptop; this is shown in Fig. 24(a). Ascreenshot of the window showing the sensed current, ambient,and conductor temperature is shown in Fig. 24(b).

G. Practical Requirements and Cost

The smart stick-on sensors operate with a low duty rate, i.e.,operate once every few minutes. This type of an operation is vi-able for some utility applications; however, it is preferable if thesensor can operate with relatively higher duty rates. Thus, a testwas performed to investigate the maximum reporting frequencyand the minimum primary currents at which the sensor can re-main self-powered. The results are shown in Table IX. It can beseen from the table that at 100 Amps, the sensor node can beoperated with a frequency as high as once every minute. Thismonitoring frequency is sufficiently high for most utility assets.The targeted cost of the stick-on sensor based on single quan-

tity Digikey prices was estimated to be around $ 45, much lowerthan presently available utility asset monitoring sensors.At this cost, if 150 stick-on sensors are considered to be

installed on every asset in a substation. The total investment onsensors is around $ 6,750. This is nearly two orders of magni-tude lower than existing solutions (see Section II). Moreover,maintenance expenditure is reduced to a minimal as these areself-powered sensors. Therefore, the smart stick-on sensorsprove to be a lucrative solution for utility asset monitoring(especially for a variety of assets).

IX. CONCLUSIONS

This paper presented the idea of low-cost, low-maintenancesmart stick-on sensors for monitoring utility assets. These sen-sors can be used for monitoring a slew of utility assets like con-ductors, cables, transformers, disconnect switches, and shuntcapacitors. They can be stuck on to these assets and start au-tonomous monitoring. A wireless sensor network architectureof these sensors was depicted, issues related to integration andinteroperability were addressed and binding mechanisms wereformalized. Various advantages of the presented architecturewere presented. It was identified that the major impediment forhaving present-day sensors massively deployed on utility assetsis their high cost and inflexibility of design. Therefore, the focusof the remaining paper was laid on designing a low-cost smartsensor.Challenges of self-powering these smart sensors were pre-

sented. In addition, detailed analysis and methodology for

Page 11: Smart “Stick-on” Sensors for the Smart Grid

MOGHE et al.: SMART “STICK-ON” SENSORS FOR THE SMART GRID 251

Fig. 22. Detailed circuit of the power management circuit with a wide operating range (Patent pending design).

Fig. 23. Screenshot of current consumption during active and sleep mode. (a)Active mode current consumption. (b) Sleep mode current consumption.

TABLE VIIIPOWER CONSUMPTION OF INDIVIDUAL COMPONENTS AT 3.3 V.

selecting a flexible design for the optimal energy harvesterconfiguration was formulated.A novel power management circuit comprising of a 0.2 V

to 3.3 V direct ac-dc boost converter was proposed, analyzed,and experimentally tested. The power management circuit wasshown to operate over a wide range (60 A–1000 A primary assetcurrent). Black-start functionality of the sensor was also demon-strated. Further, an ultracapacitor was used to operate the sensorat least once in an outage condition. In addition, at 100 A thesensor could operate with a reporting frequency of once everyminute, which is sufficient for most utility assets.All the functionalities that are envisioned by utilities

were demonstrated by building and experimentally testing aself-powered current and temperature wireless stick-on sensor.More than 30 utility and industry partners have acknowledgedthe demonstration of the smart stick-on sensor and are eager tocommercialize the concept as these sensors are the need of thehour.

Fig. 24. Stick-on sensor prototype testing. (a) Wireless stick-on sensormonitoring temperature and current of a conductor. (b) Screenshot of the datarecorded by the coordinator node.

TABLE IXMAXIMUM REPORTING FREQUENCY OF THE STICK-ON SENSOR.

ACKNOWLEDGMENT

The authors gratefully acknowledge the contributions andsupport of NEETRAC technical advisors and Dr. Y. Yang fortheir valuable suggestions.

REFERENCES[1] “Annual energy outlook 2010 with projections to 2035,” U.S. Energy

Information Administration, Washington, DC, Rep. DOE/EIA-0383,2010, .

[2] “EEI survey of transmission investment: Historical and planned capitalexpenditures (1999–2008),”. Washington, DC, Edison Electric Insti-tute, 2005.

Page 12: Smart “Stick-on” Sensors for the Smart Grid

252 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 1, MARCH 2012

[3] National Electric Transmission Congestion Study. Washington, DC,U.S. Department of Energy, Aug. 2006.

[4] National Electric Transmission Congestion Study. Washington, DC,U.S. Department of Energy, 2009.

[5] “Renewable power & energy efficiency market: Renewable portfoliostandards,” FERC, Feb. 22, 2010 [Online]. Available: http://www.ferc.gov/market-oversight/othr-mkts/renew/othr-rnw-rps.pdf

[6] R. Moghe, F. Kreikebaum, J. Hernandez, R. P. Kandula, and D. Divan,“Mitigating distribution transformer lifetime degredation caused bygrid-enabled vehicle (GEV) charging,” in Proc. IEEE ECCE, Sep.2011.

[7] R. R. Hoffman and B. Moon, “Knowledge capture for the Utilities,” inProc. 7th Amer. Nucl. Soc. Int. Top. Meet. Nucl. Plant Instrum., Con-trol, Hum.-Mach. Interface Technol., Nov. 2010.

[8] B. Warneke, M. Last, B. Liebowitz, and K. S. J. Pister, “Smart dust:Communicating with a cubic-millimeter computer,” Comput. J., vol.34, pp. 44–51, 2001.

[9] Z. Yong, G. Yikang, V. Vlatkovic, andW. Xiaojuan, “Progress of smartsensor and smart sensor networks,” inProc. Intell. Control Autom., Jun.2004, pp. 3600–3606.

[10] A. Ukil, “Towards networked smart digital sensors: A review,” in Proc.34th Annu. Conf. IEEE Ind. Electron., 2008, pp. 1798–1802.

[11] F. Cleveland, “Use of wireless data communications in power systemoperations,” in Proc. Power Syst. Conf. Expo., Mar. 2006, pp. 631–640.

[12] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci,“Wireless sensor networks: A survey,” Comput. Netw. J., vol. 38, pp.393–422, Mar. 2002.

[13] I. F. Akyildiz, X. Wang, and W. Wang, “Wireless mesh networks: Asurvey,” Comput. Netw. J., vol. 47, no. 4, pp. 445–487, Mar. 2005.

[14] A. Mainwaring, J. Polastre, R. Szewczyk, D. Culler, and J. Anderson,“Wireless sensor networks for habitat monitoring,” in Proc. 1st ACMInt. Workshop Wirel. Sensor Netw. Appl., 2002, pp. 88–97.

[15] G. Barrenetxea, F. Ingelrest, G. Schaefer, and M. Vetterli, “Wirelesssensor networks for environmental monitoring: The sensorscope expe-rience,” in Proc. IEEE Int. Zurich Seminar Commun. (IZS), Mar. 2008,pp. 98–101.

[16] Y. Yang, F. Lambert, and D. Divan, “A survey on technologies forimplementing sensor networks for power delivery systems,” in Proc.IEEE Power Eng. Soc. Gen. Meet., Jun. 2007, pp. 1–8.

[17] Protura Powerline Sensor Protura, Sep. 20th, 2010 [Online]. Available:http://www.protura.no/images/files/PLS.pdf

[18] Power Donut2 Usi, Aug. 18, 2010 [Online]. Available: http://www.usi-power.com/

[19] GridSync Wireless Sensor ABB, Feb. 16, 2011 [Online].Available: http://www.abb.com/product/db0003db004279/39766f868d8c677c8525772f0050ba11.aspx?product-language=us&country=us

[20] Line Sentry Grid Sentry, Apr. 20, 2011 [Online]. Available: http://grid-sentry.us/

[21] Line IQ Grid Sense, Apr. 20th, 2011 [Online]. Available: http://www.gridsense.com/

[22] Sentient Monitor Sentient, Apr. 20th, 2011 [Online]. Available: http://www.sentient-energy.com/product/monitors/

[23] Lighthouse MV Sensor Tollgrade, Apr. 20th, 2011 [Online]. Available:http://www.tollgrade.com/LightHouse_MV_Sensor.aspx

[24] J. M. Major, “Ensuring the health of our power lines,” 2010 [Online].Available: http://www.swri.org/3PUBS/ttoday/Summer06/Powe-Lines.htm

[25] Y. Yang, D. Divan, R. Harley, and T. Habetler, “Design and imple-mentation of powerline sensornet for overhead transmission lines,” inProc. IEEE Power Energy Soc. Gen. Meet., 2009, pp. 1–8.

[26] F. Poza, P. Marino, S. Otero, and F. Machado, “Programmable elec-tronic instrument for condition monitoring of in-service power trans-formers,” IEEE Trans. Instrum. Meas., vol. 55, no. 2, pp. 625–634,2006.

[27] A. Dunkels, J. Alonso, T. Voigt, H. Ritter, and J. Schiller, “Connectingwireless sensornets with TCP/IP Networks,” in Proc. 2nd Int. Conf.Wired/Wirel. Internet Commun., 2004.

[28] P. Kinney, “Gateways: Beyond sensor networks,” 2011 [Online].Available: http://www.zigbee.org/zigbee/en/events/documents/Sen-sorsExpo/7-Sensors-Expo-kinney.pdf

[29] Zigbee Scada Gateway, OEM Technology solutions, Mar. 12th, 2011[Online]. Available: http://www.oem.net.au/index.php?a=31&b=109

[30] R. Moghe, Y. Yang, F. Lambert, and D. Divan, “A scoping study ofelectric and magnetic field energy harvesting for wireless sensor net-works in power system applications,” in Proc. IEEE Energy Convers.Congr. Expo., Sep. 2009, pp. 3550–3557.

[31] P. D. Mitcheson, T. C. Green, and E. M. Yeatman, “Power processingcircuits for electromagnetic, electrostatic and piezoelectric inertial en-ergy scavengers,” J. Microsyst. Technol., vol. 13, pp. 1629–1635, 2007.

[32] S. Dwari and L. Parsa, “An efficient AC-DC step-up converter for lowvoltage energy harvesting,” IEEE Trans. Power Electron., vol. 25, no.8, pp. 2188–2199, Aug. 2010.

[33] R. Moghe, Y. Yang, F. Lambert, and D. Divan, “Design of a lowcost self powered stick-on current and temperature wireless sensorfor utility assets,” in Proc. IEEE Energy Convers. Congr. Expo., Sep.2010.

Rohit Moghe (S’07) received the B.Tech. degreein electrical engineering from the Indian Instituteof Technology, Roorkee, in 2007 where he wasawarded the best undergraduate research award. Hereceived the M.S. degree in electrical and computerengineering from Georgia Institute of Technology(Georgia Tech), Atlanta, in 2010, and is currentlyworking toward the Ph.D. degree at Georgia Techunder the guidance of Dr. Deepak Divan.He is one of the founding members of Energy Club

at Georgia Tech and served as the President of theclub in 2010–11. His research focuses on energy harvesting for wireless sensornetworks in power system applications and power electronics for utility appli-cations. He was a summer intern at ABB US Corporate Research Center andSiemens Energy & Automation during May-August 2008 and 2009, respec-tively.

Frank C. Lambert (S’70-M’73-SM’87) receivedthe B.S. and M.S. degree in electrical engineeringfrom the Georgia Institute of Technology (GeorgiaTech), Atlanta.He serves as the Associate Director of the National

Electric Energy Testing Research and ApplicationsCenter (NEETRAC) at Georgia Tech. He is respon-sible for interfacing with NEETRAC’s members todevelop and conduct research projects dealing withtransmission and distribution issues. He worked atGeorgia Power Company for 22 years in transmis-

sion/distribution system design, construction, operation, maintenance, and au-tomation. He participates in the IEEE PES Distribution Subcommittee and thePES Switchgear Committee and is serving on the PES Governing Board repre-senting Regions 1–7. He also serves as a member of CIGREWG A3.23 on faultcurrent limiters.

Deepak Divan (S’78–M’78–SM’91–F’98) receivedthe B. Tech. degree from the Indian Institute of Tech-nology, Kanpur, in 1975, and the M.Sc. and Ph.D. de-grees from the University of Calgary, Calgary, AB,Canada, in 1979 and 1983, respectively.He is currently a Professor in the School of Elec-

trical and Computer Engineering at the Georgia In-stitute of Technology, Atlanta, where he is also theAssistant Director of the Intelligent Power Infrastruc-ture Consortium (IPIC). He is a serial entreprenuer,founding and serving as Chief Scientist at Innovolt

Inc., Atlanta. From 1995 to 2004, he was the Chairman and the Chief ExecutiveOfficer (CEO)/CTO of Soft Switching Technologies, a company in the indus-trial power quality market. His current research interests include application ofpower electronics for power quality, power reliability, and utility and industrialapplications. He is the author or coauthor of more than 200 papers, and holds28 issued and four pending patents.Dr. Divan was the recipient of the 2006 IEEE William E. Newell Award for

contributions in power electronics. He was President of the IEEE Power Elec-tronics Society for 2009–2010, and was Chair of the IEEE Energy 2030 confer-ence in November 2008.