738 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 2, JUNE 2012 Utilizing a Smart Grid Monitoring System to Improve Voltage Quality of Customers Kerry D. McBee, Member, IEEE, and Marcelo G. Simões, Senior Member, IEEE Abstract—The implementation of smart grids will fundamentally change the approach of assessing and mitigating system voltage deficiencies on an electric distribution system. Many distribution companies have historically identified customer level voltage deficiencies utilizing a reactive approach that relies upon customer complaints. The monitoring capabilities of a smart grid will allow utilities to proactively identify events that exceed the voltage threshold limitations set forth by ANSI Std. C84.1 before they become a concern to end-users. This proactive approach can reduce customer complaints and possibly operational costs. This paper describes an approach for determining voltage threshold limits as a function of duration for proactive voltage investigations utilizing smart grid monitoring equipment. The described approach was applied to a smart grid located in Boulder, Colorado. This paper also describes the results of this two-year study. Index Terms—Customer complaints, operational costs, power quality, smart grid, smart meter, voltage quality. I. INTRODUCTION T HE Department of Energy (DOE) envisions a smart grid as a distribution system that consists of an extensive monitoring system [1]. Such a system allows for the monitoring of attributes such as voltage, current, kWh, and kVA at substations transformers, distribution transformers, smart meters, distribution switching devices, and strategically installed power quality monitors. The number of system monitors on a smart grid is considerably more than the number found on existing distribution systems, which may only have several power quality monitors per feeder even if installed in accordance with the EPRI recommendations for monitoring feeder level power quality events [2]. Prior to the implementation of smart grids, EPRI recommended the installation of power quality monitors at the substation, middle of the feeder, end of the feeder, and near customers with sensitive loads. Unfortunately many electric distribution systems in use today fall short of even these EPRI recommendations by focusing most power quality efforts at the substation transformer. This lack of monitoring at the distribution level allows many voltage events that occur on distribution feeders or on the secondary side of distribution transformers Manuscript received May 30, 2011; revised October 08, 2011; accepted December 11, 2011. Date of publication April 27, 2012; date of current version May 21, 2012. Paper no. TSG-00153-2011. K. D. McBee is with the Colorado School of Mines, Golden, CO 80401 USA, and also with Xcel Energy, Denver, CO 80235 USA (e-mail: kerry.d. mcbee@xcelenergy.com). M. G. Simões is with the Department of Engineering at Colorado School of Mines, Golden, CO 80401 (e-mail: msimoes@mined.edu). 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.2012.2185857 to go unnoticed. In most cases the only notification the utility has of a voltage discrepancy comes from customers who are experiencing discomfort and/or equipment malfunction. The implementation of a smart grid monitoring system allows utility personnel to perform remote voltage investigations as soon as voltage events occur and before problems occur for customers. These investigations can be performed at the feeder level and at the customer’s point of common coupling (PCC), which is important when investigating customer power quality issues [1], [3]. All devices that sense voltage, even if they do not possess power quality monitoring accuracy, are available to utilize for identifying voltage fluctuations that are damaging to customer equipment and/or reduce the quality of service for customers. A smart grid monitoring system consists of a main communication system that connects all devices with remote monitoring capabilities, utility and end-user owned, to a central data collection location which simplifies access to all data retrieved [4]. The aggregate number of these devices can account for hundreds of monitoring points per feeder depending upon design and customer penetration. Remote monitoring devices include: • capacitor banks; • distribution transformers; • smart meters; • power quality meters; • phase monitoring units. Remote monitoring will allow for the identification of voltage deviations prior to equipment malfunction or misoperation that are noticeable to customers. Also, system deficiencies that have gone unnoticed by customers and utility personnel prior to the installation of a monitoring system will become evident. Four benefits to implementing a remote voltage investigation approach include the following: • reduced number of customer voltage complaints; • lowered operational costs associated with investigating customer voltage problems; • reduced number of outages caused by failed connections; • prevention of damage to utility or customer equipment. Historically customers have been the main source of identifying voltage discrepancies. The goal of a proactive approach is to identify problems before they become a concern to customers, therefore a new notification system is required. A smart grid can employ numerous software applications to interface with, interpret measurements from, and control distribution system devices. This type of software can provide automatic notification to utility personnel of voltage sags, swells, and other voltage events prior to them becoming a nuisance or problem to customers. Because many voltage events are temporary in nature, 1949-3053/$31.00 © 2012 IEEE MCBEE AND SIMÕES: UTILIZING A SMART GRID MONITORING SYSTEM TO IMPROVE VOLTAGE QUALITY OF CUSTOMERS and do not warrant field investigations, threshold limits that include magnitude and time durations should be employed to insure the effectiveness of a proactive voltage investigation program. This paper examines how to establish threshold limits for the duration and magnitude of voltage events. Section II details both the existing reactive approach and the proposed proactive approach. Upper and lower threshold limits are detailed in this section along with an approach for determining duration threshold levels. To determine the effectiveness of the proactive approach and the methods described for determining thresholds limits, the methodology was applied to a smart grid located in Boulder, Colorado and operated by Xcel Energy. The Boulder smart grid, commonly referred to as SmartGridCity, consists of 26 feeders that serve approximately 25 000 customers. At the time of this research, its smart grid monitoring system consisted of over 10 000 smart meters and 5000 sensors installed on distribution transformers, both of which provide real time voltage readings. The results of applying the proactive approach to this smart grid along with an evaluation of the customer and economic benefits are detailed in Section IV. II. REACTIVE VERSUS PROACTIVE APPROACH INVESTIGATIONS TO A. Reactive Approach to Investigations In the past, most voltage quality investigations performed by distribution companies have been the result of customer complaints. This paper categorizes this approach of identifying and mitigating such problems as a “reactive” approach. The general reactive approach is described below. A utility company is notified by a customer that equipment is malfunctioning due to suspected voltage fluctuations, which are identified by customer equipment failure and dimming or flickering lights. Customers may also hire an engineering firm to investigate problems prior to notifying the utility. Although informative, a distribution company never wants to rely exclusively on a report provided by a customer’s investigator due to the unknowns associated with their monitoring techniques, specifically in regards to the IEEE std. 1159 [3]. Utility field personnel, many times in the form of a lineman, are dispatched to obtain instantaneous readings and/or to identify the severity of the voltage fluctuations. If instantaneous readings do not suggest a problem, the utility company may choose to install power quality recording devices at the customer’s PCC, within the customer’s facility, and possibly at several locations along the circuit serving the customer in accordance with [3]. Voltage readings are recorded for a given time period, which is dictated by the frequency of the disturbances [3]. Many times the duration of recordings are limited by the small number of power quality recorders and the high volume of investigations. For events that are infrequent, this situation leads to investigations that are not thorough. This is rarely the case with a smart grid investigation since information is provided for days, weeks, and months. Power quality investigators analyze the recorded data to determine if the voltage fluctuations violate the threshold limitations set forth by ANSI Std. C84.1 and possibly a local regula- 739 tory agency [5]. If it is found that threshold limits are exceeded, the utility company will act to mitigate the problem. If it is found that the deviation is due to customer equipment, it is the customer’s responsibility to mitigate the problem. The action taken by the distribution company is heavily dependent upon two things: 1) exceeding limits set by ANSI Std. C84.1 and/or tariffs set forth by local regulatory agencies; 2) a customer complaint regarding poor service. Utilizing both known standards and customer complaints has proven to be a very reliable approach for mitigating voltage quality problems in the past. However, a smart grid monitoring system can remove customers as the notification system and thereby improve customer service. B. Proactive Approach to Investigations The increased monitoring capabilities of a smart grid will allow for the identification of voltage quality problems at the same time as the customer, or before the customer experiences any significant equipment malfunctions or misoperations. This “proactive” approach will eliminate the customer as a means of identifying system deficiencies, which will leave the utility engineer with only the known standards and regulatory tariffs as a guide for determining system deficiencies. The authors have chosen to refer to voltage, current, and frequency fluctuations found through a smart grid monitoring system as deficiencies and not power quality events. Most definitions of power quality include the powering and grounding of equipment that is appropriate for normal operation [3], [6], [7]. Since customers are removed from the investigation, there is no evidence that equipment has malfunctioned or misoperated. Therefore the incentive of the utility company is not to improve power quality for customers, but to reduce and eliminate system deficiencies that are inefficient and may result in customer and/or utility equipment failure if not addressed. The elimination of customers as a means for voltage problem identification also presents a problem to the utility company. Which voltage deficiencies are investigated to minimize customer impact and optimize operational efficiency? Should a utility company investigate every voltage deficiency, even if its duration is only several cycles and only occurs once? The proactive approach to system monitoring should rely heavily upon voltage thresholds set forth by ANSI Std. C84.1 and local regulatory agencies. Std. C84.1 was developed for the purpose of protecting equipment from experiencing damaging voltages; therefore utilizing these thresholds provides protection for customer and utility equipment [5]. The standard defines two voltage ranges that are applicable for utility service and the utilization of end-user equipment above and below 600 V. Range A, which describes acceptable voltage ranges for steady state operations, provides the most conservative approach to investigations since these limits are the most stringent and will thereby identify more system deficiencies. For residential customers at 120 V the upper and lower limit is , which is more stringent than the minimum variation limits that define voltage sags and voltage swells [3]. Range B accounts for voltages that result from abnormal, but necessary operating conditions. Table I illustrates Range A of ANSI Std. C841 [5]. 740 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 2, JUNE 2012 TABLE I ANSI STD. C84.1-2006 RANGE A ACCEPTABLE VOLTAGE LEVELS Voltage threshold limits are applied according to the location of the device that is being monitored. If the reading is taken at a smart meter, the service voltage limits should fall within the limits set by Range A of B. However, if the voltage is monitored at the distribution transformer, the lower limit should reflect at least a 2 –3 V drop on the service conductors, which would result in a lower limit of 116 –117 V on systems below 600 V. Any violation of these limits, regardless of duration, is considered a voltage deficiency, which by definition includes more events than if only voltage sags and swells were considered. The smart grid monitoring system should interface with software that automatically notifies utility engineers when voltage deficiencies occur. Upon notification, the utility must determine if the event was temporary in nature or is the result of system deficiencies, with the latter type of events being investigated further. These system deficiencies are continuous in nature and will not remedy themselves. Evaluation of continuous events should include summing the duration of all voltage events that occur in a given time period. Temporary faults, which can be the cause of voltage events, can comprise as much as 76%–80% of all system faults on system [8]–[10]. Most customer complaints are the result of long-term voltage problems, not those caused by temporary events. The aggregate time below or above the magnitude threshold is utilized to distinguish between temporary events and events that are consistent, but possibly infrequent. Any violation that deviates below or under the set magnitude threshold for an accumulated time of for a given period is investigated in further detail. The concept is illustrates mathematically in (1) and graphically in Fig. 1. (1) where: total time duration above or below threshold limits for time period ; time duration of event in time period ; total number of different events where voltage thresholds are exceeded; investigation duration, typically a day, week, or month. Fig. 1. Investigation envelope graph illustrating different investigation zones for events based on total time below or above threshold. Investigation zones are defined utilizing high and low threshold limits and . Voltage events that have aggregate time less than and fall into Zone 1 of Fig. 1 are considered temporary in nature and are not investigated any further. The time duration of this zone, which is defined by , should be based upon customer type. Customers more susceptible to voltage sags and swells, such hospitals and commercial or industrial facilities with critical processes could arrange an agreement with the Utility to have a that is satisfactory to capture repeated voltage sags and swells. Zone 2 consists of voltage levels that are acceptable. Therefore, the only voltages deficiencies investigated are those that fall into Zones 3 and 4, which are undervoltage and overvoltage zones. can range from a day, week, to a month, depending upon the desired level of investigation. III. APPLICATION OF PROACTIVE APPROACH The proactive investigation approach was applied to a smart grid located in Boulder, Colorado and operated by Xcel Energy, an electric utility company that generates, transmits, and distributes energy in eight states. At the time of the analysis, the smart grid consisted of 26 circuits served by 4 substations. The smart grid served over 25 000 customers with its smart meters and 5000 voltage sensors installed on distribution transformers, both of which provided real time voltage and kWh readings. The system, which also utilizes self healing automation, went online in August of 2008, which coincided with the start of proactive investigations. The software utilized for interpreting smart meter and distribution transformer data was OpenGrid Distribution (OGD), which is manufactured by the Current Group. The authors utilized the software and the guidelines described in the previous section to proactively investigate voltage discrepancies that exceeded the voltage investigation envelope illustrated in Fig. 1. Upon the proactive identification of voltage problems, utility field personnel were dispatched to verify problems and to evaluate if mitigation actions were required. The magnitude threshold limits utilized for the study were based on Std. C84.1 Range A, but did not follow them exactly. A lower threshold limit of 117 V was utilized for sensors at distribution transformers, which can have a voltage that is 2–3 V higher than the voltage at the customers’ meter. The upper limit was not adjusted to account for secondary voltage drop so as MCBEE AND SIMÕES: UTILIZING A SMART GRID MONITORING SYSTEM TO IMPROVE VOLTAGE QUALITY OF CUSTOMERS Fig. 2. Voltage profile of 50 kVA transformer with a loose neutral connection as provided by Open Grid Distribution software. not to exceed 126 V for situations where the secondary voltage drop is less than 2 V. The total duration for events was set at 30 min (0.5 h). The period of 96 h was utilized as . Investigations included researching any system event that fell into Zones 3 and 4. Problems found included: • low voltage due to overloaded transformers; • high and low continuous voltage due to loose neutral connections; • intermittent high and low voltage due to loose conductor connectors; • low voltage as a result of system reconfiguration due to outage restoration and new construction; • high and low voltage due to failed distribution transformer windings. The software provided the authors with visual graphs of voltage events, which simplified the identification of violations of the voltage investigation envelope. A loose neutral connection was identified in Fig. 2, which illustrates the separation of voltage between the two transformer secondary conductors (LEG 1 and LEG2). A distribution transformer overload, which is illustrated in Fig. 3, was verified as being caused by a new customer connecting to the distribution transformer. Fig. 4 illustrates the identification of high voltage that was the result of a malfunctioning substation transformer load tap changer. Because the voltage problem found in Fig. 4 was caused by a substation device, the majority of the transformer sensors and smart meters recorded the same excessive voltage, which simplified the identification of the cause. Events such as overloaded transformers, loose connections, and failing transformers are not self-correcting over time; in fact they would most likely become progressively worse due to the nature of electrical equipment failures. Therefore, the utility company can pay for repairs before it affects customers or after. Waiting to mitigate problems can possibly be more expensive if the system deficiency results in utility or customer equipment damage, or a catastrophic failure occurs outside of normal business hours, thereby resulting in utility personnel working overtime and possibly requiring the emergency shipment of equipment. The results of the study revealed that investigation zones accurately captured problems that would eventually lead to cus- 741 Fig. 3. Voltage profile of an overloaded 50 kVA transformer as provided by Open Grid Distribution software. Fig. 4. Voltage profile of a 50 kVA transformer with high primary side voltage, as provided by OpenGrid Distribution software. tomer complaints. A total of 45 voltage discrepancies were identified between August 2008 and December 2010. Fig. 5 illustrates the voltage problems identified, where they fell on the investigation envelope graph, and if the problems required equipment repair or upgrade (Action Required) or if they were due to abnormal operating conditions or were temporary in nature (No Action), which did not warrant mitigation. From these results, a more refined investigation envelope graph was developed, which is also illustrated in Fig. 5. Prior to implementation of the smart grid monitoring system, Xcel Energy regularly experienced voltage complaints in excess of 20 per year in the smart grid territory, with a peak of 50 complaints occurring in 2007. In 2009 and 2010, which were the first two complete years of implementation, 5 and 7 voltage complaints were recorded in the SmartGridCity territory. Fig. 6 illustrates the voltage complaints by year for the smart grid territory. It should be noted that the complaints recorded in 2009 and 2010 where not caused by voltage deficiencies. Customers requested investigations because of equipment misoperations; however, in each case the cause was determined to be due to customer owned equipment. Because of these types of 742 IEEE TRANSACTIONS ON SMART GRID, VOL. 3, NO. 2, JUNE 2012 TABLE II FIVE-YEAR SUMMARY OF VOLTAGE COMPLAINTS Fig. 5. Graph of all voltage deficiencies identified during 28-month analysis period. TABLE III COSTS AFFILIATED WITH VOLTAGE EVENTS 8/1/08–12/31/10 Fig. 6. Voltage complaints between 2004 and 2010 in SmartGridCity footprint. occurrences, voltage complaints may never go to zero in the SmartGridCity territory. Proactive investigations reduced the number of customer complaints, which are illustrated in Fig. 6, but did not reduce the number of voltage investigations. Complaints and proactive investigations were combined to determine the percentage of investigation performed within the SmartGridCity territory to those occurring outside of the smart grid territory. Between 2006 and 2010, the percentage of investigation performed in the smart grid territory, including both reactive and proactive investigations, ranged between 8.72% and 5.08%, with the latter occurring in 2010. Although the percentage decreased with time, as evident from Table II, the authors believe that the slight deviation is possibly the result of annual fluctuations in load, weather, and system constraints. Although the number of investigations did not decrease, the cost associated with investigating voltage deficiencies proactively did decrease. The total cost savings for 2009 and 2010 were calculated based on investigation costs. Specifically, the costs of remote investigations were compared to the costs of performing the same investigations utilizing the reactive approach to voltage investigations, which utilizes lineman and/or field visits by power quality investigators. Between 2006 and 2008, a lineman was initially dispatched for 36% of the investigations. The average cost of a lineman per investigation was $120, which included a single lineman, bucket extension truck, travel time, and investigation time. On-site power quality investigations were assumed to include a power quality investigator, recording equipment, recording equipment installation time, one week of data monitoring, and at least a day of data processing and analysis totaling an average operational cost of $520 per investigation. Smart grid investigations averaged no more than 2 h, which include office and field work. The authors, who are experienced in power quality investigations, performed the remote investigations; therefore the only requirement was 1 h of training on OGD, the software utilized to interpret smart meter and distribution transformer sensor measurements. The proactive investigations consisted of two actions. First, the authors identified a voltage problem with OGD, which took approximately 5 min with the automatic system notification. Once identified, the authors determined if field validation was required. A lineman was dispatched for 84% of the identified voltage events to validate and excess the problem. The average time spent on a remote investigation by the authors was 5–20 min, at an average cost of $14 per investigation. Based on the 45 problems identified during the 16-month period, the total cost of remote investigations was estimated to be approximately $5190. Applying the same 45 voltage problems to the reactive investigation costs described in the previous paragraph would result in a monetary MCBEE AND SIMÕES: UTILIZING A SMART GRID MONITORING SYSTEM TO IMPROVE VOLTAGE QUALITY OF CUSTOMERS impact of $25 320, an increase of $20 130 or 487% from the cost of proactive investigations. Table III summaries the costs associated with both proactive and reactive investigations for the 45 voltage events identified. Even if the proactive approach resulted in more voltage event investigations than if the utility relied exclusively upon customer complaints, a success rate of 20% for proactive investigations would still be less costly than relying exclusive upon a reactive approach. The cost of implementation is not considered in the economic evaluation presented in this paper. Because of the many functionalities of the equipment utilized for proactive voltage investigations, comparing implementation costs to voltage investigation benefits would be inaccurate and misleading. Additional functions of the equipment utilized includes automated meter reading, feeder voltage and var optimization that reduces system losses, transformer loss management, outage notification, and the identification of nested outages during storm outages restoration efforts. IV. CONCLUSION The extensive monitoring system of a smart grid will allow distribution companies to proactively investigate system voltage deficiencies. Without customers acting as a notification system, distribution companies can rely upon ANSI Std. C84.1 and local tariffs in conjunction with predetermined duration thresholds to determine if a potential voltage problem exists. The case study presented in this paper illustrated that utilizing a 30-min duration for a 96-h window of monitoring is sufficient for reducing customer voltage complaints. Although complaints were reduced with a proactive approach to investigations, the total number of investigations or voltage events investigated did not decrease. However, the costs associated with voltage investigations were significantly reduced when compared to utilizing the existing reactive approach. REFERENCES [1] Title XIII, Department of Energy [Online]. Available: http://www.oe. energy.gov/DocumentsandMedia/EISA_Title_XIII_Smart_Grid.pdf [2] M. Baran and K. Scherrer, “Extending power quality monitoring to feeder level,” in Proc. IEEE Power Energy Soc. Gen. Meet.—Convers. Del. Electr. Energy 21st Century, Jul. 2008. 743 [3] IEEE Recommended Practice for Monitoring Electric Power Quality, IEEE Standard 1159-1995. [4] R. E. Brown, “Impact of smart grid on distribution system design,” in Proc. IEEE Power Energy Soc. Gen. Meet.—Convers. Del. Electr. Energy 21st Century, Jul. 2008. [5] American National Standard for Electric Power Systems and Equipment—Voltage Ratings (60 Hertz), ANSI Standard C84.1-2006. [6] IEEE Recommended Practice For Powering and Grounding Electronic Equipment, IEEE Standard 1100-1999. [7] R. Dugan, M. McGranaghan, and H. Beaty, Electrical Power Systems Power Quality. New York: McGraw-Hill, 1996, p. 2. [8] S. F. Tan and S. K. Salman, “Investigation into the implementation of auto reclosing scheme in distribution networks with high penetration of DGs,” in Proc. 43rd Int. Univ. Power Eng. Conf., Sep. 2008. [9] O. Jung-Hwan, Y. Sang-Yun, K. Jae-Chul, and K. Eung-Sang, “Particular characteristics associated with temporary and permanent fault on the mult-shot reclosing scheme,” in Proc. IEEE Power Eng. Soc. Summer Meet., Jul. 2010, vol. 1, pp. 421–424. [10] D. E. Parrish, “Lightning-caused distribution circuit breaker operations,” IEEE Trans. Power Del., no. 4, pp. 1395–1401, Oct. 1998. Kerry D. McBee (BS’99–MS’00) received the B.Sc. degree from the Colorado School of Mines, Golden, in 1999 and the M.Sc. degree in electric power engineering from Rensselaer Polytechnic Institute, Troy, NY, in 2000. He is currently working toward the Ph.D. degree in the Department of Engineering, Colorado School of Mines. During his career he has focused on power quality, reliability, forensic engineering, and distribution design for companies such as NEI Power Engineers, Peak Power Engineering, Knott Laboratory, and Xcel Energy. His field of interest is smart grid implementation effects upon distribution engineering and utility operations. Marcelo G. Simões (S’89–M’95–SM’98) received the Ph.D. degree from the University of Tennessee, Knoxville, in 1995. He is with Department of Electrical Engineering and Computer Science, Colorado School of Mines, Golden, where he is the Director of the Center for Advanced Control of Energy and Power Systems (ACEPS). He has been conducting research and education activities in the development of intelligent control for high-power-electronics applications in renewable- and distributed energy systems and smart-grid technology. Dr. Simões is currently Past Chair for the IEEE IAS IACC and Co-Chair for the IEEE IES Smart Grid Committee. He has been involved in activities related to the control and management of smart-grid applications since 2002 with his NSF CAREER award “Intelligent Based Performance Enhancement Control of Micropower Energy Systems.”