IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 2, JUNE 2012 761

Online Monitoring of Substation Grounding GridConditions Using Touch and Step Voltage SensorsXun Long, Student Member, IEEE, Ming Dong, Student Member, IEEE, Wilsun Xu, Fellow, IEEE, and

Yun Wei Li, Member, IEEE

AbstractA grounding grid of a substation is essential for re-ducing the ground potential rises inside and outside the substationduring a short-circuit event. The performance of a grounding gridis affected by a number of factors, such as the soil conductivityand grounding rod corrosion. Industry always has a strong desirefor a reliable and cost-effective method to monitor the conditionof a grounding grid to ensure personnel safety and prevent equip-ments damage. In view of the increased adoption of telecom andsensor technologies in power industry through the smart grid ini-tiative, this paper proposes an online condition monitoring schemefor grounding grids. The scheme monitors touch and step voltagesin a substation through a sensor network. The voltages are createdby a continuously-injected, controllable test current. The resultsare transmitted to a database through wireless telecommunication.The database evaluates the grid performance continuously by com-paring the newly measured results with the historical data. Manyof the limitations of the offline measurement techniques are over-come. Computer simulation studies have shown that the proposedscheme is highly feasible and technically attractive.

Index TermsOnline monitoring, step voltage, substationgrounding, thyristor, touch voltage.

I. INTRODUCTION

P ROPER grounding is the first line of defense against light-ning or other system contingency to ensure the safety ofoperators and power apparatus. A poor grounding system notonly results in unnecessary transient damages, but also causesdata and equipment loss, plant shutdown, as well as increasesfire and personnel risk. As a result, Utility companies are ac-tively seeking techniques that can effectively and reliably eval-uate the grounding grid conditions to ensure personnel safetyand prevent equipments damage.The performance of grounding grid is affected by various fac-

tors such as unqualified jointing while building, electromotiveforce of grounding current, soil erosion and theft of groundingrods [1]. Thus, monitoring and diagnosing the conditions ofgrounding grid has been an active research field for many years.However, almost all techniques implemented or proposed forgrounding monitoring are offline types where special instru-ments are installed for grounding condition check on a regular

Manuscript received May 11, 2011; revised August 28, 2011; accepted Oc-tober 14, 2011. Date of publication February 13, 2012; date of current versionMay 21, 2012. This workwas supported by iCORE. Paper no. TSG-00175-2011.The authors are with the Department of Electrical and Computer Engineering,

University of Alberta, Edmonton, AB T6G 2V4, Canada (e-mail: xlong@ual-berta.ca; mdong@ualbeta.ca; wxu@ualberta.ca; yunwei.li@ece.ualberta.ca).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.2011.2175456

or as-needed basis. These existing methods can generally be cat-egorized into two types: measurement of grounding impedanceand detection of grounding integrity.Fall-of-Potential (FOP) method is the basic scheme for

grounding impedance measurement and it has been imple-mented for many years [2]. Its key point is to correctly locatethe potential probe, which is quite time-consuming. A lot ofvariations have been proposed to improve this scheme, such asby using variable frequency source [3] or implementing mul-tiple electrodes [4]. The methods taking account of current splitin transmission and distribution grounding system are furtherdeveloped in [5], [6] for accurately measuring the impedanceof in-service substations. However, the potential probes are stillindispensable in these FOP-based schemes. Several enhancedgrounding grid computer models are developed recently withconsidering soil layer depth in [7], [8] or based on electro-magnetic field methods [9], [10]. But, the accuracy of thesemodels relies on the soil resistivity measurement. Once thesoil condition is changed [11], potential electrode needs to berelocated and it obviously increases the labor.Monitoring the integrity of grounding grid is another way to

evaluate the performance of grounding grid [12], [13]. How-ever, the computation of this method depends on many uncer-tain factors such as soil conductivity, humidity and climate [14].A device based on measuring magnetic induction intensity isdesigned to diagnose the grounding grid corrosion in [15]. It re-quires the current injection between all possible grounding leadson the ground surface to increase accuracy, which is not prac-tical in a large scale substation.All of the aforementioned methods are offline-based, which

at best give one-shot measurement results. If another set ofresults is needed, the measurement system must be redeployed.The offline-based methods have significant disadvantages.Firstly, the results are largely dependent on the soil conditionat the time of measurement. Secondly, sudden changes ofthe grounding grid such as those caused by theft cannot beidentified timely. To solve these problems, the methods that canmonitor the grounding condition on a continuous, i.e., online,basis is highly desired.This paper proposes an online substation touch and step

voltage monitoring scheme, which can continuously injecttesting current into a grounding grid and then measure thecorresponding touch and step voltages. The testing current iscreated by a thyristor-based signal generator which is con-nected between single energized phase conductor and groundto stage a temporary and controllable fault. There is no extracable needed for current flow as power line is utilized as a path

1949-3053/$31.00 2012 IEEE

762 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 2, JUNE 2012

Fig. 1. The process of the proposed online monitoring scheme.

for current injection. Touch and step voltages, which directlyreflect operational safety of substation, are used as indicatorsof grounding condition. As the measurement of touch/stepvoltages does not require long cables extending outside of asubstation [16], it is very suitable for long-term online moni-toring. Supported by the historical data made available throughthe online database which wirelessly communicate with thevoltage sensors, the variation of the measured data can be usedto infer the change of grounding grid conditions.

II. THE PROPOSED ONLINE MONITORING SCHEME

As shown in Fig. 1, the process of the proposed online mon-itoring scheme includes a) testing current generation and in-jection; b) touch voltage and step voltage measurement; andc) safety assessment based on both data variation and actualvalues. If the variation between the newly measured data andhistorical data is below the preset threshold plus the measureddata does not exceed the safe value defined in the IEEE standard[1], the grounding grid under test is considered to be in a goodcondition and the next test will be made after a preset period.Otherwise, a warning event will be created and then mandatoryinspections in the suspected spots with high touch voltage orstep voltage will be carried out.

A. Testing Current Generation and Injection

The signal generator for testing current generation consistsof a pair of thyristors connected to the supply via a single-phase step-down transformer, which convert high voltage to lowvoltage for the normal operation of the thyristors. When thethyristors are fired under a preset firing angle, a testing currentwill be injected into the system from the primary side of thetransformer [17]. The two thyristors are operated alternately tocreate a sinusoidal waveform. To reliably measure the resultingtouch and step voltages, the duration of injected current cannotto be too short to establish stable potential profiles [18]. Theminimum time for tolerable touch or step voltage calculation is30 ms according to [1]. On the other hand, the injected currentis required not to interrupt the normal operation of groundingfault protection relay, in which the minimum trip time is about100 ms [19]. In this work, the duration of current injection istherefore setup as 50 ms, which is within the range between30 ms to 100 ms. Not like grounding impedance measurement,which needs square waveform to obtain various frequency com-ponents to avoid the fundamental frequency interference frompower system or requires lightning waveforms to measure tran-sient impedance, this paper focuses on safety evaluation at sub-stations and the sinusoidal waveform is used tomimic a shot-cir-cuit fault.

Fig. 2. The remote current injection scheme.

Fig. 3. The local current injection scheme.

This signal generator can be installed either remotely or lo-cally. In the remote source scheme (see Fig. 2), the signal gen-erator is installed at a downstream site far from the substationto minimize the impact of the current injection to the groundpotential profile. As the ground can be utilized for current pathfrom the injected site to the substation grounding grid, the extracurrent cable is not necessary [20].In the local source scheme, the signal generator including a

step-down transformer is installed in the substation as shownin Fig. 3. The current is directly injected from the substationand it returns from the remote power source. Since the deviceis located in a substation, maintenance can be convenientlyachieved, which is important for long-term monitoring. How-ever, a large transformer is needed as the signal generator hasto be installed at the high voltage side in a substation for thelocal source scheme. This signal generator cannot be installedat the grounded secondary side in a substation, since a currentloop is established by the grounded neutral and the test currentwill not pass through the remote earth [21].

B. Touch/Step Voltage Based Sensor NetworkThe current injected into the grounding grid results in rises

of touch voltage and step voltage, which directly indicate thesafety situation in and around the substation under test. Touchvoltage is defined as the potential difference between an exposedmetallic structure within reach of a person and a point where thatperson is standing on the earth, while step voltage is definedas the difference in potential between two points in the earthspaced 1 meter (or a step) apart [22]. The measurement of touchand step voltages can be easily conducted at many points of in-terest in a substation, which is very suitable for long-term onlinemonitoring. Moreover, the interference with potential electrodeand long cable installation when conducting impedance mea-surement is also eliminated.We further proposed to use a wireless sensor network for

touch and step voltages monitoring (see Fig. 4). Typically, a

LONG et al.: ONLINE MONITORING OF SUBSTATION GROUNDING GRID CONDITIONS USING TOUCH AND STEP VOLTAGE SENSORS 763

Fig. 4. The sensors network of touch/step voltages measurement.

Fig. 5. (a) The voltage measurement (1-step voltage, 2-touch voltage). (b) Thestructure of the voltage sensor.

grounding grid is buried 0.5 m1 m under ground, which re-sults in potential difference at the surface of ground. The touch/step voltage sensors are distributed at corners of a groundinggrid and some other frequently visited spots with special con-cern of human safety. All of these sensors can transmit signalswirelessly to a computer which could be located indoors. Thecomputer is responsible to collect, classify and update the datarecorded in the database.According to the IEEE Standard 81.2 [22], the simulated-

personnel method is recommended for touch and step voltagesmeasurement. This method utilizes a resistor with 1000 re-sistance represents human body resistance and is connected be-tween two feet. Each foot is made by a metallic plate with200 cm surface area and 20 kg weight. A voltage meter is in-stalled across the resistor with high internal impedance so as notto influence the measurements. A device is designed to measureeither touch voltage or step voltage as seen in Fig. 5. Note that,the distance between two feet is adjustable, which is 0.5 mfor measurement and 1 m for measurement, respec-tively. Moreover, an extra probe is used to contact the exposedconductive surface for touch voltage measurement.It uses a voltage transducer to measure the voltage across ,

and then the measured value is converted to the digital format byan ADC module. A MCU processes the data and the results arefinally transmitted to a central computer through a RF module.In the project, Zigbee 2.4 GHz wireless signal transmission isrecommended and its range can be reach up to 300 ft, whichis adequate for a small or medium size distribution substation.Moreover, it can be easily configured to handle wireless sensornetworking application at a low cost.From the Thevenin equivalent circuit of the touch/step

voltage measurement as shown in Fig. 6, it is found that themeasured or is not the same as the potential difference onthe ground, and the touch or step voltage can be expressed as

(1)

Fig. 6. The Thevenin equivalent circuits of: (a) touch voltage measurement;(b) step voltage measurement.

(2)

where is the potential difference between the feet and thetouch point, is the touch voltage, is the foot resistancewhen two feet are in parallel, is the human body resistance(1 k ), is the potential difference between two feet, isthe step voltage, is the foot resistance when two feet are inseries.However, the metallic plates installed on the surface of the

ground are likely to be corroded due to humidity or other fac-tors, which results in the increase of their resistance accordingly.Equation (1) and (2) indicate that the measured voltage ( or) decreases with the increase of the feet resistance ( or) under the same potential difference ( or ). In this

case, the measured touch/step voltage will be lower than thenormal value and it may mislead the assessment.To eliminate the effects of the feet resistance variation, the

voltages are measured twice, one in a close circuit during onesignaling period and the other in an open circuit during the nextsignaling period. As the switching is operated after current in-jection, it would not cause arcing and it also has no require-ment on the switching speed. As shown in Fig. 6, an electricalcontactor is utilized for the switching purpose. Apparently, thevoltage or in an open circuit is equal to the potentialdifference or . Resolve (1) and (2), the resistance ofor can be obtained.

(3)

(4)

If the variation between the estimated (or ) and itsnominal value is larger than a predetermined value, the mea-sured (or ) cannot be directly used. In this case, the metallicfeet need to be replaced by a new pair of plates. Alternatively,these voltages ( and ) can be adjusted according to the fol-lowing equations:

(5)

(6)

where is the nominal value of is the nominalvalue of .From the study of the potential profile of a substation, it

is found that there are several suspected spots in or around asubstation, particularly in a substations corners or around thefences. Therefore, before installing the measurement tools, it

764 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 2, JUNE 2012

is necessary to inspect those suspected points. One solutionis pretest. A workman walks across a substation with the testdevice to record the locations where the measured voltagesexceed the preset threshold. Further measurements can thenbe made by online monitoring in the suspected locations, thusreducing the time and cost of measurements. Another solutionis to find the suspected spots in the potential profile froma computer simulation results. In this case, the accuracy oflocating the danger points highly depends on the accuracy ofthe simulation model.According to [1], the limit of touch/step voltage is a function

of a) shock duration (i.e., fault clearing time); b) system charac-teristics; c) body weight; and d) foot contact resistance as shownin (7) and (8). The constant is 0.116 for a person with 50 kgbody weight, while it is 0.157 for 70 kg.

(7)

(8)

Since the injected current is much smaller than the maximumfault current, the measured touch/step voltage is therefore muchlower than the limits defined in (7) and (8). Thus, the originalmeasured data is intentionally increased to maximum value by(9) and (10) when the data transfer to the database. The decisionis then made by the comparison of the measured data with his-torical data or with the maximum tolerable values provided by(7) and (8).

(9)

(10)

For personnel safety evaluation, touch voltage is more se-vere than step voltage [23]. The current caused by touchingan exposed conductor flows through the heart, whereas the onecaused by step voltage bypass the heart. Therefore, the toleratetouch voltage is much lower than tolerate step voltage. Gener-ally, satisfying the touch voltage safety criteria in a substationautomatically ensures the satisfaction of the step voltages safetycriteria. In this project, most areas in the substation are exam-ined for touch voltage, and only the edges of the grid are exam-ined for step voltage.

C. Intelligent Evaluation Techniques With DatabaseAnother novel feature of the proposed scheme is the imple-

mentation of online database. It is known that the resistivity ofthe surface soil layer would be changed in different seasons,which may results in touch/step voltages moving to the hazardside [14]. For example, if the thickness of the low-resistivitysoil layer in raining season is smaller than the buried depth ofthe grounding grid, the touch voltage increase. In another case,the high resistivity soil layer formed in freezing season wouldcause the increase of the touch/step voltage with the thickness orresistivity of the freezing soil layer. One major defect of the ex-isting offline monitoring method is the inability of tracking sea-sonal influences on the safety of substation grounding system.With the support of database, we can continuously monitor and

record the change of touch/step voltage. Particularly, during thesevere conditions, like continuous raining or freezing seasons,the frequency of online monitoring can be increased in order tofind the potential hazards in time.Corrosion, which can damage the effective connections

among the conductors, is another factor affecting the safety ofthe grounding system. The grounding grid corrosion is causedby acid or alkali in soil and the corrosion rate can reach up to 8.0mm per year according to statistic results [24]. This situationbecomes more serious as the steel-grounding or galvanizedsteel-grounding system is widely used, which is more easilycorroded than copper so that it requires more accurate, timelyassessment of grounding grid.While the corrosion of grounding grids may be detected by

regular off-line measurements as it is a slow process, the theft ofgrounding rods, another major concern to utility companies, cansuddenly change the integrity of the grounding grid. Failing todetect this change in a timely manner will cause serious conse-quences. In the proposed online monitoring scheme, the changeof touch and step voltages at the same point are recorded, sothat synthesized and reliable estimation can be made not onlydepending on the IEEE standard but also on the variation due toseasonal influence, corrosion or theft.Based on the above analysis, an intelligent evaluation (see

Fig. 7) can be made as follows:1) Generate and inject a testing current into a grounding gridperiodically. Measure the resulting touch/step voltageswith the sensor network and transfer the data to the centraldatabase.

2) Scale the measured touch/step voltages to the maximumtouch/step voltages.

3) Compare the maximum touch/step voltages with IEEEstandard under the same parameters, like fault clearingtime and the body weight, etc. If it exceeds the safe value,a warning event is created and the suspected location isreported to substation operators for further analysis.

4) Compare the measured touch/step voltages with the his-torical data at the same location. If the variation is largerthan the preset threshold, a warning event is created eventhough the actual value does not exceed the standard. Amandatory examination will be taken around the suspectedpoint to check if the conductors are stolen or broken due tocorrosion.

5) If no suspected spot is found, the database is updated withthe newmeasured data and meteorological parameter, suchas temperature and humility. Then, after a preset period, goback to 1).

III. STUDY OF CURRENT DISTRIBUTIONA simulation model is built in PSCAD to study current dis-

tribution of both remote and local injection schemes as shownin Fig. 8. The distribution substation under test transfers powerfrom 125 kV to 25 kV via a Delta-Yg connection transformer.An overhead ground wire, so called skywire, accompanies withtransmission lines and the ground resistance of a transmissionline tower is 32 . At the secondary side, the neutral line ofdistribution system is multiple-grounded with 15 at each

LONG et al.: ONLINE MONITORING OF SUBSTATION GROUNDING GRID CONDITIONS USING TOUCH AND STEP VOLTAGE SENSORS 765

Fig. 7. The evaluation process with database.

Fig. 8. Computer simulation for current distribution study.

Fig. 9. The structures of transmission line tower and distribution line pole.

TABLE IPARAMETERS OF COMPUTER SIMULATION FOR CURRENT DISTRIBUTION STUDY

grounded connection. The structure details of the transmissionline tower and the distribution line pole are illustrated in Fig. 9.Other parameters are listed in Table I.In the remote injection scheme, a temporary fault is staged at

downstream of the under-test substation to create a fault currentflowing from the faulted phase to ground and back to the sub-station. However, with the presence of the multiple groundedpoints on the neutral, such as pole grounds and transformergrounds, the current is divided before reaching the substationgrounding grid. As shown in Fig. 10, the current division of theremote injection scheme depends on the distance between thelocation of the staged fault and the substation. When the stagedfault is located 5 km away from the tested substation, only 37%current flows back through substation grounding grid.The local injection scheme requires a pair of thyristors

connected between a single phase of transmission line and the

Fig. 10. Current distribution of the remote scheme with respect to distancefrom subtation.

Fig. 11. Current waveforms of current distribution study.

grounding grid by a step-down transformer. Because of theexistence of overhead ground wires and neutral lines, not allfault current flow through the grounding grid to the remoteearth [21]. The simulation results (see Table II) show that73.58% current across the grounding grid, 26.70% current

in the skywire and 10.59% current in the neutralline. Disconnecting the skywire and the neutral line can largelyincrease the grounding grid current. However, it is impossibleto disconnect these ground wires for long time monitoring inreality. From touch voltage simulation which will be discussedlater, 60 A grounding grid current can result in about 3 V13V touch voltage, which is large enough for effective detection.The current waveforms for the local injection scheme areshown in Fig. 11. Typically, there are relays implemented in thesubstation for ground fault protection. These protective relayshave an inverse current/time characteristic, which suggests theycan tolerate high current with a short duration. As the durationof the injected 60 A current is about 50ms, shorter than 0.1 s, itdoes not interrupt the normal operation of the protective relays[19].The proposed local scheme is also applicable to the substa-

tions with Yg-Yg or Y-Yg connection. As shown in Fig. 12,both the primary side and secondary side of the transformerare Yg connection and the neutral points are connected in thegrounding grid. A staged fault is created at the primary sidewhen the thyristors are turned on for 50 ms. The computer sim-ulation results are listed in Table III. If all the grounded wires

766 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 2, JUNE 2012

TABLE IICURRENT DISTRIBUTION OF THE LOCAL SCHEME WITH A DELTA-YG

CONNECTION TRANSFORMER

Fig. 12. Current distribution study of Yg-Yg connection substation.

TABLE IIICURRENT DISTRIBUTION OF YG-YG AND Y-YG CONNECTION

are connected, the current ratio of is 81.9% for Yg-Ygconnection and it is 76.8% for Y-Yg connection.

IV. COMPUTER SIMULATION OF THE PROPOSED ONLINEMONITORING SCHEME

Computer simulations have also been conducted inCYMGRD [25] to measure touch/step voltages, illustratethe process of intelligent evaluation scheme and analyze the in-fluence of seasons, corrosion and theft. The designed groundinggrid as shown in Fig. 13 is 150 m long and 100 m wide. All con-ductors are buried at a depth of 0.5 m. X-axis has 8 conductorsand Y-axis has 10 conductors. The diameter of all conductorsis 19.1 mm. Plus, 30 grounding rods are vertically connected tothe grounding grid. Each rod is 5 m long with diameter 2.858cm. Moreover, the station surface is with crushed rock of 2500ohm-meter resistivity at a thickness of 0.3 m and the exposureduration is 0.36 s with 4000 A fault current.To begin touch/step voltages simulation, we firstly interpret

the soil resistivity measurements and obtain a soil model for thesubsequent analysis. A two-layer soil model is implemented inthis simulation. From the data provided by IEEE standard (seeTable IV), both the upper and lower layers resistivity can be cal-culated and the depth of the upper layer can be estimated as well.The result of soil model calculation is consistent with the IEEEcalculated values (see Table V), which proves the validity ofthe designed two-layer soil model. Furthermore, the maximumpermissible touch and step voltages in accordance with the sub-station surface and the shock time can be calculated by (7) and(8), which is 1084.2 V and 3551.8 V respectively.The potential profile of the grounding grid diagonal line is

shown in Fig. 14. Apparently, touch voltage at the corner is

Fig. 13. The designed grounding grid in the computer simulation.

TABLE IVTHE SOIL RESISTIVITY MEASUREMENTS DATA WITH THE FOUR-PIN METHOD

PROVIDED BY IEEE STANDARD

TABLE VCOMPARISON OF THE SIMULATION RESULTS AND IEEE VALUES

much larger than in the middle center. It is due to less conduc-tors buried around corners than around center. The suspectedpoints can be clearly located from this potential profile, whichis very useful for installation of the voltage sensors. This pro-file also confirms that the value of maximum permitted touchpotential has a dominant role in determining the design of thegrounding grid. If a grid satisfies the requirements for safe touchpotentials, it is very unlikely that the maximum permitted steppotential will be exceeded. In Fig. 14, the margin between thecalculated touch voltage and the permissible touch voltage isabout 200 V800 V, while this margin for step voltage is as largeas 3500 V.As the injected current through the grounding grid is actually

about 50100 A, the concern here is if the 50100 A currentis able to result in detectable touch/step voltage. The profile inFig. 15 is obtained with 60 A grounding grid current, whichcauses the touch voltage between 313 V. The voltage in thisrange can be easily detected by the voltage sensors. For safetyevaluation, the actual voltages are scaled up to the maximumvalues in the database according to (9) and (10).With the support of database, synthesized and reliable es-

timation can be made depending on IEEE standard constraintand recorded data variation. To better clarify the concept of the

LONG et al.: ONLINE MONITORING OF SUBSTATION GROUNDING GRID CONDITIONS USING TOUCH AND STEP VOLTAGE SENSORS 767

Fig. 14. The potential profile of the grounding grid diagonal line.

Fig. 15. The potential profile with 60 A grounding grid current.

Fig. 16. The conductors on the left edge are stolen.

safety evaluation with database, we consider two scenarios, oneis theft and the other is the change of soil resistivity due to sea-sonal influence. Theft is a serious threat for the safety of sub-station. As shown in Fig. 16, the conductors on the left edge ofsubstation are stolen so that touch voltage around that area islargely increased as illustrated in Fig. 17.When comparing the profiles of Figs. 14 and 17, it is easy

to detect the difference in the corner area. An alarm is createdimmediately and investigation in the corner should be made assoon as possible. On the other hand, touch voltages at somespots also exceed the limit, a mandatory examine is requiredat these locations.

Fig. 17. The potential profile after theft.

Fig. 18. The soil resistivity in different seasons.

As mentioned above, soil resistivity depends on a number offactors: soil type, chemical composition, moisture, temperature,etc. For an existing grounding grid, it is mainly affected by sea-sonal variations. Especially in North America, the frozen soil inwinter is a hazard for grounding grid safety. Fig. 18 is the fieldtest data of soil resistivity in 12-month study [11]. It is apparentthat during the summer, the resistivity becomes lower due tohigh precipitation, while the resistivity goes higher during thewinter because of the frozen soil. The influence of these vari-ations on the touch/step voltages are analyzed in the followingsimulation.Three locations are picked up from the left-bottom corner to

the middle of the grid as shown in Fig. 19 and the touch voltagesare investigated for 12 months (Fig. 20). When a fault occurs inJune, all touch voltages are under the limit. However, if a faulthappens in December, the increase of soil resistivity cause thetouch voltages increase. It is clear that some of the touch volt-ages exceed the limit. In this case, if an offline test is taken inJune but not in December, these danger spots cannot be founded;whereas, in the proposed online monitoring, these spots can bereported in timewith monthly evaluation. The frequencies of themeasurement can be adjusted according to the requirement ofa utility company. More frequently the measurements are con-ducted, more timely the danger spots can be found.

V. CONCLUSIONIn this paper, an online monitoring scheme for substation

safety assessment is proposed, which periodically measurestouch and step voltages with a preset frequency and effectivelyevaluates the grounding grid conditions with the help of a data-base. The test current is generated by firing a pair of thyristors

768 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 2, JUNE 2012

Fig. 19. Three suspect spots are picked up from the diagonal line.

Fig. 20. Touch voltage of three suspect spots in different seasons.

for 50 ms. The current can be injected remotely or locally.The local injection scheme has a larger portion of the injectedcurrent flowing through the grounding grid, but it costs morethan the remote scheme due to high rating voltage and highcapacity of the step-down transformer.The condition monitoring is achieved with a wireless

touch/step voltage sensors network installed at various lo-cations of a substation. These sensors are connected to acentral database where an evaluation process is carried out bycomparing the newly measured data to the limits from IEEEstandard, or checking if the data variation at the same spotexceeds safety thresholds. Furthermore, current distributionhas been studied with computer simulations, which verified theeffectiveness of the proposed local and remote schemes. Fromthe case studies of conductor theft and seasonal influences, theadvantage of online monitoring is very clear since some dangerspots cannot be found in time without continuous measurement.With further research, the proposed scheme could be used tolocate broken section or missing grounding electrode basedon the step/touch voltage profile obtained from the sensors.Compared to offline methods, which at best gives one-shotassessment, the proposed online grounding grid monitoringscheme is more effective and reliable, and it could become animportant component of a smart substation.The paper has presented an overall concept of the proposed

monitoring scheme. A lot more research works are still needed.

For example, the proposed scheme is focused on touch and stepvoltage indices which are related to personnel safety concerns.Since grounding design has other objectives such as facilitatingequipment protection, the proposed scheme needs to be furtherexpanded to include sensors and indices that address equipmentprotection concerns. It is possible that acceptable touch and stepvoltages at sufficient locations in a substation may imply an ac-ceptable grounding condition from equipment protection per-spectives. But research is needed to verify this postulation.The proposed scheme involves a sensor network and its data

collections. There are many challenges to build and maintainsuch networks. The reliability of the network needs to beconfirmed as well. These are exactly the subjects of interestto ICT- (information and communication technology) orientedsmart grid researchers.

REFERENCES

[1] IEEE Guide for Safety in AC Substation Grounding, IEEE Std.80-2000, 2000.

[2] IEEE Guide for Measuring Earth Resistivity, Ground Impedance, andEarth Surface Potentials of a Ground System, IEEE Std. 81-1983,1983.

[3] I. Lu and R. Shier, Application of a digital signal analyzer to the mea-surement of power system ground impedances, IEEE Trans. PowerApp. Syst., vol. PAS-100, no. 4, pp. 19181922, Apr. 1981.

[4] A. Meliopoulos, G. Cokkinides, H. Abdallah, S. Duong, and S. Patel,A PC based ground impedance measurement instrument, IEEETrans. Power Del., vol. 8, no. 3, pp. 10951106, Jul. 1993.

[5] J. Choi, Y. Ahn, H. Ryu, G. Jung, B. Han, and K. Kim, A newmethod of grounding performance evaluation of multigrounded powersystems by ground current measurement, in Proc. Int. Conf. PowerSyst. Technol., Nov. 2004, vol. 2, pp. 11441146.

[6] L. S. Devarakonda, J. Moskos, and A. Wood, Evaluation of groundgrid resistance for inservice substations, in Proc. 2010 IEEE PowerEng. Soc. Transm. Distrib. Conf., Apr. 2010, pp. 14.

[7] C. Chang and C. Lee, Computation of ground resistances and as-sessment of ground grid safety at 161/23.9-kv indoor-type substation,IEEE Trans. Power Del., vol. 21, no. 3, pp. 12501260, Jul. 2006.

[8] A. Puttarach, N. Chakpitak, T. Kasirawat, and C. Pongsriwat, Sub-station grounding grid analysis with the variation of soil layer depthmethod, in Proc. IEEE Power Tech. Conf., Jul. 2007, pp. 18811886.

[9] L. Qi, X. Cui, Z. Zhao, and H. Li, Grounding performance analysis ofthe substation grounding grids by finite element method in frequencydomain, IEEE Trans. Magn., vol. 43, no. 4, pp. 11811184, Apr. 2007.

[10] J. Ma and F. Dawalibi, Analysis of grounding systems in soils withfinite volumes of different resistivities, IEEE Trans. Power Del., vol.17, no. 2, pp. 596602, Apr. 2002.

[11] R. Gustafson, R. Pursley, and V. Albertson, Seasonal grounding re-sistance variations on distribution systems, IEEE Trans. Power Del.,vol. 5, no. 2, pp. 10131018, Apr. 1990.

[12] B. Zhang, Z. Zhao, X. Cui, and L. Li, Diagnosis of breaks in sub-stations grounding grid by using the electromagnetic method, IEEETrans. Magn., vol. 38, no. 2, pp. 473476, Mar. 2002.

[13] J. Hu, R. Zeng, J. He, W. Sun, J. Yao, and Q. Su, Novel method ofcorrosion diagnosis for grounding grid, in Proc. 2000 Int. Conf. PowerSyst. Technol., pp. 13651370.

[14] J. He, R. Zeng, Y. Gao, Y. Tu, W. Sun, J. Zou, and Z. Guan, Seasonalinfluences on safety of substation grounding system, IEEE Trans.Power Del., vol. 18, no. 3, pp. 788795, Jul. 2003.

[15] Y. Liu, X. Cui, and Z. Zhao, May 2010, A magnetic detecting andevaluation method of substations grounding grids with break andcorrosion, [Online]. Available: http://www.springerlink.com/con-tent/f71631x2733n7158/fulltext.pdf

[16] A. P. S. Meliopoulos, S. Patel, and G. J. Cokkinides, A new methodand instrument for touch and step voltage measurements, IEEE Trans.Power Del., vol. 9, no. 4, pp. 18501860, Oct. 1994.

[17] W. Xu, G. Zhang, C. Li, W. Wang, G. Wang, and J. Kliber, Apower line signaling based technique for anti-islanding protection ofdistributed generatorsPart I: Scheme and analysis, IEEE Trans.Power Del., vol. 22, no. 3, pp. 17581766, Jul. 2007.

LONG et al.: ONLINE MONITORING OF SUBSTATION GROUNDING GRID CONDITIONS USING TOUCH AND STEP VOLTAGE SENSORS 769

[18] R. Verma and D. Mukhedkar, Fundamental considerations and im-pulse impedance of grounding grids, IEEE Trans. Power App. Syst.,vol. PAS-100, no. 3, pp. 10231030, Mar. 1981.

[19] P. Rush, Network Protection & Automation Guide, ALSTOM T&DEnergy Automation & Information 2002.

[20] J. Sarmiento, H. G. Fortin, and D. Mukhedkar, Substation groundimpedance: Comparative field measurements with high and low cur-rent injection methods, IEEE Trans. Power App. Syst., vol. PAS-103,no. 7, pp. 16771683, Jul. 1984.

[21] S. Patel, A complete field analysis of substation ground grid by ap-plying continuous low voltage fault, IEEE Trans. Power App. Syst.,vol. PAS-104, no. 8, pp. 22382243, Aug. 1985.

[22] IEEE Guide to Measurement of Impedance and Safety Characteristicsof Large, Extended or Interconnected Grounding Systems, IEEE Std.81.2, 1992.

[23] R. Kosztaluk, R. Mukhedkar, and Y. Gervais, Field measurements oftouch and step voltages, IEEE Trans. Power App. Syst., vol. PAS-103,no. 11, pp. 32863294, Nov. 1984.

[24] P. Sen and N. Mudarres, Corrosion and steel grounding, in Proc.1990 22nd Annu. North Amer. Power Symp., pp. 162170.

[25] Cooper, Dec. 2009, CYMGRDSubstation Grounding Program[Online]. Available: http://www.cyme.com/software/cymgrd/B1100-09079-CYMGRD.pdf

Xun Long (S08) received the B.E. and M.Sc degrees in electrical engineeringfrom Tsinghua University, Beijing, China, in 2004 and 2007, respectively, andis currently working toward the Ph.D. degree in the Electrical and ComputerEngineering Department, University of Alberta, Edmonton, Canada.His main research interests include power line signaling, distributed genera-

tion and fault detection.

Ming Dong (S08) received the B.Eng. degree in electrical engineering fromXian Jiaotong University, China, in 2004. He is currently working toward thePh.D. degree in electrical and computer engineering with the University of Al-berta, Edmonton, Canada.His research covers smart grid, grounding systems, and power quaility.

Wilsun Xu (F05) received the Ph.D. degree from the University of British Co-lumbia, Vancouver, Canada, in 1989.From 1989 to 1996, he was an Electrical Engineer with BCHydro, Vancouver

and Surrey, respectively. Currently, he is with the Department of Electrical andComputer Engineering, University of Alberta, Edmonton, Canada, where he hasbeen since 1996. His research interests are power quality and distributed gener-ation.

YunWei Li (S04M05) received the B.Sc. degree in engineering from TianjinUniversity, China, in 2002 and the Ph.D. degree from Nanyang TechnologicalUniversity, Singapore, in 2006.In 2005, he was a Visiting Scholar with the Institute of Energy Technology,

Aalborg University, Denmark. From 2006 to 2007, he was a Postdoctoral Re-search Fellow in the Department of Electrical and Computer Engineering, Ry-erson University, Canada. After working with Rockwell Automation Canada in2007, he joined the Department of Electrical and Computer Engineering, Uni-versity of Alberta, Edmonton, Canada, as an Assistant Professor. His researchinterests include distributed generation, microgrid, power converters, and elec-tric motor drives.