Low Frequency Power Line Carrier Communication Technology

For AMI and Smart Grid Applications

DR. SIOE T. MAK Ph. D. EE

[ IEEE Life Fellow ]

1

TABLE OF CONTENTS

1. GENERAL INTRODUCTION a. The Electric Energy Delivery Infrastructure……………………………….……...………. ..…...3

1. Its hierarchical structure……………………………………………………………… ……..32. Circuit Configurations and Unique Properties of the 3-phase Network………… ………..33. The Switching Transient Phenomenon……………………………………………... ……… 44. Energy conversion devices………………………………………………………...… ………4

b. Advanced Metering Infrastructure (AMI) and communication issues…..………..……… .. .…..42. FUNDAMENTAL ISSUES

a. Introduction..………………………………..………………………………..… ………… ……5b. System Architecture and Infrastructure…...……………………... …………………………… 5c. Operational and reliability issues……………………………………………….………… ..… ..7d. The cost of reliability and added utilization costs………………………………... … ..….….… .9e. The issue of obsolescence……………………….…………………………….………………… 10f. Organization, addressing methods and communication control……………………………………11g. Data security, system recovery and loss of data…………………… …12h.Types of data, data warehousing and accessibility… ……………...…12

3. THE TWACS COMMUNICATION TECHNOLOG 1. Introduction………………………………………………………………………………………….13 Issues pertaining Non Power-line types of Communication Technologies……………………………15 4. ADVANCED METERING AND ADDED VALUE FUNCTION

I. Customer Services and Demand Side Management.………......…. 16 II. Improving of Service Reliability and Optimization of Energy Delivery (Assets Management) 16

III. Supporting Functions…………………………,,,……..…………………………………….. 16 Customer Services and Demand Side Management………………… ………………….…… ..17 Advanced Metering Applications……………………………………….…… ………….....17 Retail wheeling………………………………………………... . 17 Pre-pay metering…………… ... ……….....……….. .. 18 Remotely Operated Service Disconnect Devices . . ……..….... .. 18 Gas and water metering………… ……….………..… ,,… ,,,.19 Load management……………………………………….……….……..…………,….19

Time of Use…………….………………………...…………….….……………….… 19 Demand Response to Improve the Load Factor to Minimize Network Losses… . ..19 Averting Rolling Blackouts and Reducing the Impact of Cold Load Pickup …… .. .20 Monitoring of Distribution Transformer Overloading…… . ……….20 Detection of Unauthorized Use of Electric Energy… ……………………….……. ..21 Alarm…..……………………………..…………………………………………. .. ..21

Improving of Service Reliability and Optimization of Energy Delivery. …..21 Electric Utility Network Outage Management………………………..…………...…..21

Feeder Load Balancing……………………………………...………………….……. .23 Integrated Voltage and VAR control……………………………………… ….……..24 Power Quality Monitoring…… ……………………………...……..25 Causes of Inaccuracies…………………………………………………………28 Examples of extreme waveform distortion……………………… …………..28 Standard issues… …………………….……………………………... 29 Distributed generation………………… ……………………….… ..30 Hybrid Electric Vehicles…………………………………………………… . ……30 Supporting Functions……..…………………………… ………… ...…… 31

Communication Network Monitoring and Control…... ………………… ……....… 31 AM/FM systems ……………………………………………. .……... ..31 Data Warehousing………………………..………………………………… …..……. 31

5. CONCLUSIONS………………………………………………………………………………... 326. REFERENCES………...….…………………………………………...................................... ...32

2

1.GENERAL INTRODUCTION

a. The Electric Energy Delivery Infrastructure and its physical characteristics. 1. Its hierarchical structure.

Electric energy is the life-blood of the nation and is here to stay. One is basically dealing with energy transport when one talks about electric energy. Electric energy is the most versatile transitional form of energy with the following desirable basic characteristic features: It can be transported along narrow corridors in large bulk quantities at practically the speed of light. These narrow corridors for transport are the transmission lines, the distribution network, etc. Also, to serve the multitude of electric energy users, multiple independent parallel simultaneous energy transport channels are used. These channels form an energy delivery infrastructure. For an electric distribution network, the major components of the delivery infrastructure are the distribution substations in conjunction with the distribution buses, feeders and low voltage network. The infrastructure works very reliable at a frequency of 50 Hz or 60 Hz. It can be split into minute quantities or bulk quantities, whatever the need calls for. Conversion into other forms of energy that meet the user’s need is a well-known technology. Heating elements uses

electric resistance wires to convert electric energy into thermal energy. Electro-mechanical conversion devices, such as electric motors convert the electric power into useful mechanical power. Numerous other applications can be listed.

Bulk electric power can be generated at sites away from populated areas lending it more amenable for control of pollution, reducing safety and health hazards, etc.

Distributed Generation (Wind Power, Low Head Hydro, etc. ) are increasingly installed as their prices dropped and tax incentives are given by the government to encourage the generation of renewable green power. Most utility networks can still be viewed as having the following major components.

Major power conversion plants. Plant capacities range between a few hundred MW to a few thousand MW. Transmission network. Transmission voltages range between 34.5 KV to 765 KV. The distribution network. Distribution voltages range between 4.0 KV to 34.5 KV and at the service level

voltages range between 120 V to 480 V.

2. Circuit Configurations and Unique Properties of the 3-Phase Network.Three phase networks operating at 50 Hz or 60 Hz still dominate the existing the delivery infrastructure and the

circuit configurations can be 4-wire, 3-wire, 2-wire and single wire with earth return. The transitions to the various voltage levels are accomplished through the use of three phase Y-Y, grounded or ungrounded neutrals, D-Y or single phase transformers. Under steady state operating conditions, the phase to neutral voltages V an, Vbn and Vcn are quite often transformed into line to line voltages at the secondary side due to the 3-phase transformer windings configuration. However, when one views each of the phase to neutral voltages at the medium voltage as base phasors and the line to line voltages as linear combinations of the base phasors, a unique picture emerges showing some basic characteristic features of the network. For any phasor generated at the substation bus there is a remote corresponding phasor at part of the network served by the substation that is slightly phase shifted with respect to the medium voltage substation bus phasor at a magnitude that depends on the circuit voltage drop and the intervening transformer winding ratio. Fig. 1 shows a simple example of it.

Also, since the load current phasor can be referred to the voltage phasor that served as voltage source to the load current, similar behavior can be expected for the current phasor as the voltage phasor. If there are no intervening power conditioning devices, one can make this phasor connection all the way from remote customer end to generation. Any current drawn at the 120 V side has its corresponding phasor at the generation side. Hence, if for any reason a massive cumulative imbalance of load at the distribution side occurs, it will reflect itself also as massive imbalance at the generation side

For distribution networks in particular, there is also a unique property that is true. Since there are no distribution circuits that have lengths close to a quarter wavelength of a 50 Hz or 60 Hz wave, no long line or Ferranti effects are expected to occur at the distribution network. This unique behavior of the phasors mentioned above has important implications for the design of control algorithm and applications which will be discussed in the next sections.

Another aspect is the time element which ties in intimately with captured data, transient behavior of the network and speed of response of a control system. When a switching action takes place, whether due to man-made control actions or nature induced actions such as lightning flashover across two conductors or a line fault, the electric network reacts according to some specific patterns.

3

Fig. 1

3.The switching transient phenomenon,Another aspect is the time element which ties in intimately with captured data, transient behavior of the network and

speed of response of a control system. When a switching action takes place, whether due to man-made control actions or nature induced actions such as lightning flashover across two conductors or a line fault, the electric network reacts according to some specific patterns.

Energy storage devices such as capacitors (stray capacitances or capacitor banks), inductors, etc. cause a reaction to a sudden change similar to the reaction due to a perturbation function. In most cases, the reaction is transient oscillatory and decays within a very short time and in general in less time than the duration of a quarter wave-length of the power frequency. The magnitude also depends on at what part of the waveform of the power frequency voltage the switching action is initiated and the attenuation can be attributed to the energy absorbing parts of the network. Fig. 2 and 3 show some transient oscillatory behavior of switching responses extracted from the voltage at the service voltage and the current response extracted from the neutral current transformer at the medium voltage substation bus.

Fig.2 Fig. 34. Energy conversion devices

Energy conversion devices such as electric motors, heaters, electric welders, etc. have fairly long time constants to reach steady state operating conditions. The time constants typically last multiple numbers of cycles of the 50 Hz or 60 Hz. Examples are locked rotor starting currents of motors as their output torques increase.

In this manuscript, the primary text material deals with the electric utility distribution network and how to use the available tools to optimize the network operation and the use of energy by the customers. b. Advanced Metering Infrastructure (AMI) and communication issues.

Recently many utilities are aggressively implementing large scale automatic meter reading (AMR) capability in their systems. With the recent advances in the computer, communication technologies and miniaturization of digital electronics, automation of remote meter reading is made possible. Many of the utilities’ concerns are initially with the ability to read consumption of electric energy, gas and water on a scheduled basis and possibly to perform some type of demand side management function.

However, several other reasons are also given to justify investing in a communication system for automated remote meter reading. Replacing meter readers is definitely not one of the main reasons. The annual cost of meter reading is very low in many areas. With a communication system in place, it is expected that other functional capabilities can be added that will benefit the utility company. Concepts of ADVANCED METERING INFRASTRUCTURE (AMI), SMART METERS, SMART GRID, DISTRIBUTION AUTOMATION, UTILITY ASSETS MANAGEMENT, HOME AUTOMATION NETWORKS, etc. start to appear in the technical literature, implying the evolution of futuristic systems.

4

In essence, one does not start making a decision based on which communication technology to choose, but rather the choice has to be driven by the functions one expect to implement in the future and which communication technology serves as an enabling tool to support the intended applications.

The trend in many countries including the USA to deregulate the electric utility industry through unbundling of the utilities’ monopoly results in the creation of generation, transmission and distribution companies. This sets the stage for introducing numerous innovative concepts that are purported to allow competition and which will greatly benefit the energy users. The concept of retail wheeling at the distribution level generates new requirements for meter reading, customer billing, pricing strategies, etc. The electric energy starts losing its social function characteristics and is slowly being considered as a commodity. Non-electric retailers explore the possibilities to become energy retailers, and several non-utility billing companies will eventually come into existence. Technologies are now available to make this possible.

Large scale remote meter reading capability implies the following possibilities: A reliable two-way communication system is available. The communication system can reach practically every customer connected to the utilities’ network. Metering and other types of data can be brought back reliably and accurately in a timely fashion. Large number of data transfers per unit time can be handled by the communication network and processed for storage. Storage of very large amounts of data organized in a way for easy access by multiple interested parties is now available. Installation or replacing of remote devices, which interface with the existing or new electric metering devices should be

easy.For the reasons mentioned above, electric utilities actively start looking what added-value capabilities can be

implemented. The question that has been raised repeatedly is “What else is possible beyond simple meter reading applications? “. Some applications which have been proposed are advanced metering, new types of billing systems, providing meter reading services to other utilities such as gas and water utilities, load control, time of use rates application, demand response management, remote service connect and disconnect, capacitor bank control, real time voltage monitoring, outage management and other distribution automation functions. Expressions, such as SMART METERS, SMART GRID, ASSETS MANAGEMENT, DATA WAREHOUSING and METERING DATA MINING are becoming popular buzz words.

It is also obvious that the different parties within a utility organization have different applications in mind for utilization of the communication system. The customer services department wants timely retrieval of energy metering data for billing purposes. Besides the regularly scheduled operation of the system for billing purposes, numerous unscheduled uses can occur. Satisfying individual customer billing complaints, disconnecting services, pre-pay metering, customer’s profile of energy use, etc. are just a few of them. The load management group primarily uses the one-way global outbound capability of the system. Some preliminary load research requires load data, which can be obtained from time stamped synchronized interval metering data. Outage management is a typical unscheduled application. These are only few of the many possible uses of the communication system being considered. With the recent introduction of Smart Meters, more useful information can be obtained which can be used to design control algorithms that will benefit the electric utilities’ operation. When added-value applications are equally important as the meter reading function to the prospective user of the communication system, it is key that that all the important questions are asked and are answered properly to every party’s satisfaction. The dilemma one faces has always been what questions to ask and what answers are good answers.

The topic that has to be addressed is the following. “ Can a system that is designed for utility automatic meter reading, be economically expanded or upgraded to implement future added-value capabilities without requiring a major overhaul, large increase in capital expenditures and future added utilization costs “.

A communication system that is used to perform simple meter reading for monthly billing purposes may have severe limitations in terms of expandability to perform other added-value applications. The assessment of whether a technology will be able to meet the future needs not only requires a good understanding about the strengths and limitations of the communication technology being considered, but also requires an intimate knowledge about the operational issues and expected system performance to handle the added-value functions. All these issues will be discussed in great detail in subsequent chapters.

Many communication systems that are used for utilities’ applications are hybrids. The technology used for communication between a central computer and certain major gateway nodal points can be totally different from the one used for communication between the major nodes to the remote devices. A typical system has a Central Net Server computer, which handles all communications to and from the remote intelligent devices, maintains the necessary data-base of communication paths and nodes, nodes and remote devices addresses and routing algorithms for efficient operation to access the remote devices. This computer is linked through a high-speed data link network to primary intelligent nodes. These nodes are the gateways to links with the remote devices. The net server computer is also linked to other parties’ service computers, such as the Customer Service Billing computer, the Load Management Dispatch center computer, the Service and Maintenance Dispatch center, etc. The high-speed data link backhaul network can be the public telephone system, cellular radio network, satellite communication network, microwave radio, fiber optic network, etc. The link from the primary gateway

5

nodes to the remote devices can be a different communication network. It is normally owned by the utility and is solely dedicated for utility applications.. For power line carrier types of communication technologies, the communication infrastructure is identical to the utilities’ distribution power delivery infrastructure. Extensions of this network using hard-wired or radio technologies for reading water and gas meters are already available. There may or may not be sub-nodes required between the primary nodes and the remote metering, monitoring and control devices. These additional nodes can be repeaters or data concentrators.

To simplify the description of the remote devices we coin the generic name “transponder” to all intelligent remote devices, which can be energy metering and other data collecting devices, intelligent monitoring and control devices, alarm and switching devices, etc.

Chapter 2 is devoted to general system architecture and communication infrastructure issues. The concept of baud rate, physical channels, communication channels, data throughput and operational issues, such as system performance, communication retries, reliability, added utilization costs, data flow control and communication operational issues are discussed in great detail.

Chapter 3 describes the TWACS communication technology. The strengths of the technology are explored from a physical and design standpoint. Chapter 4 provides a bird’s eye view of some of the most popular added-value applications. The synergism between the various supporting functions and applications are discussed and explained.

A list of reference materials and the conclusions are in chapter 6 and 7.

2. FUNDAMENTAL ISSUES

a. Introduction To obtain a better understanding and insight what the requirements are for a system to be purchased and installed

that best meet the utilities’ objectives, several basic issues will be discussed in this chapter. A typical system, even to be used for simple automatic meter reading only, has to be able to bring back meter reading information from points scattered over a wide geographical area. If time is not a critical element for the system application – for example all meters have to be read only once a month – and the metering data is only needed for scheduled billing, then the requirements imposed on the communication system are not very stringent. Multiple repeated efforts to try to get a meter reading because of poor communication performance can be tolerated. All the data are sent to one interested party only, which is in all the cases the Customer Service and Billing Department.

The more challenging requirements imposed on the system will be if time interval metering data are required and even more stringent requirements are imposed if interval reading of all the meters have to be synchronized in time and time-stamped. The energy dispatch center, the rate designers, the demand side management group, the distribution control department and the billing department may have interest in the 15 minutes, half hourly or hourly synchronized data retrieved from all the electric energy metering devices. The volume of data, which has to be collected, transmitted from the remote sites and stored is very large. Different parties may also require different amounts and types of data.

This seemingly simple addition of the time element and time synchronization as a requirement triggers several levels of operational complexities. The economy of time utilization for data transfer, data throughput, communication reliability, data storage, quickly identifying changes in data transfer path or link failure and their restoration are critical items to consider for the system.

In addition to the metering devices, there are load control devices, distribution control and monitoring devices, etc. and they are linked to a control or dispatch center through a communication network. From an operational standpoint there can be pre-programmed scheduled operations requiring information retrieval or simple switching type functions. Simple meter reading functions and load control belong to this category of operation.

There are operations that are triggered by unusual events at the remote locations that are unpredictable and practically require almost real time responses with minimum time delay. Outage management is one such function.

Other functions are triggered by needs for optimization. An example is voltage and reactive power control to improve the voltage profile on a line and simultaneously reduce the line losses. In some cases the operation is predictable but there will be occasions where required control actions will be determined by the most recently available information.

b. System Architecture and InfrastructureBased on all the considerations mentioned before, it is obvious that the key to the successful implementation of

functions, is to have a system design which meets the following requirements. It has to provide the ability to communicate reliably in a speedy manner to a large population of devices

scattered over a large geographical area. It has to allow multiple users access to data provided by the communication system.

6

Whether one is dealing with an electric, gas or water utility, the issues of accessibility and communication, the ability to store information for future retrieval, the monitoring and recording of events, data storage and data base management of the communication network, information and transponder addresses, data security and recovery after a system crash, etc., require careful considerations. Each of those can be considered as an essential operational module of the overall system. They form the essential building-blocks that are linked together to form a total operating system. This is a communication system architecture issue.

A general system architecture diagram is shown in Fig.4. The communication control center or the Network Server Computer is linked to a number of main nodes through a high-speed data link network. The high-speed data link network can be a public high-speed communication network, a dedicated telephone network, a fiber optic network or high-speed radio frequency network. The main nodes at the utility communication network maintain the link between the service computer and the sub-nodes and from the sub-nodes to the transponders.. The main node decodes messages received from the net server computer, encodes and transmits the messages through the appropriate physical paths to the sub-nodes and subsequently to the transponders. It also receives information from the sub-nodes and subsequently transmits them to the net server computer through the high-speed data link. It is basically an intelligent trans-receiver with memory and intelligence for communication and data flow control.

Fig. 4

For the TWACS system that utilizes the distribution network for communication medium, the sub-nodes can be remote transponders that are coupled directly to metering or other devices. These transponders collect information, store and transmit that information to a sub-node or main node, which subsequently transmits the received information to the Net Server computer. This action can be prompted by a command from the main node or triggered by a schedule that has been downloaded into its memory. There are quite a few cases where it is necessary for the sub-node to monitor or control several different devices. The sub-node can be resident at one of the devices it monitors or controls and is linked to the other devices through short – hop low power radio or other type of hardwired communication links. The main nodes are linked to the sub-nodes by means of a communication network that is owned or leased by the user.

Interested users of the system can be initially the Customer Service Department where its primary interest is billing and other services that are directly customer related. When added value capabilities are added, the Maintenance and Repair Services and possibly other groups such as the Energy Management group, etc. are interested in using the system for their own need. Their computers will be linked to the Net Server computer through the high-speed data link.

Since these new applications are expected to be available, there might be different types of ownership of certain types of data and control, required speed of responses, volume of data for transmission, priorities of functions, etc. In such a case, the monitoring and control of the physical path for communication, utilization and control of the communication traffic by the Net Server computer also become a communication system infrastructure control issue.

Electric power utilities, which use the distribution network for communication, do not need to capitalize on a communication infrastructure. The electric power utilities already own the distribution network.

c. Operational and reliability issues

7

A communication system is intended to transfer information from one point to another point in the most reliable, optimal and timely fashion. There is a tendency to define this statement in terms of “baud rate”, “be able to receive every bit reliably”, a “guarantee to get all information 100% of the time”, etc. Sometimes these statements can be quite misleading. To get an idea what these statements mean, let us resort to a simple example.

Fig. 5Fig. 5 is used for reference. Suppose a system can communicate at a reasonably high baud rate of 10,000 bits per

second. Assume that each message is 100 bits long. At 100% communication performance, 100 messages can be transmitted per second when operating in sequential fashion. If there are 1,000,000 messages to be retrieved, the time required will be (1,000,000 / 100) = 10,000 seconds which is approximately 2 hours and 48 minutes.

If half of the messages can use a particular physical communication path and the other half can use a different physical communication path, and the two physical paths are non- interfering, then all the messages can be transmitted in 1 hour and 24 minutes. If two independent communication channels can use the same path as shown in the right most part of Fig. 5, then the transmission time can be reduced to 42 minutes.

The example shows how system architecture and infrastructure are intertwined. In the first design (left most) the “hourly interval meter reading” requirement cannot be met. By the time the first meter is read again, 2 hours and 48 minutes have elapsed due to communication delay time. The rightmost example has two physical paths and also allows double the flow of messages at the same time per physical path. The architecture and infrastructure of the system almost meet the hourly meter reading requirement. These simple examples quickly show that baud rate when not coupled with system architecture and infrastructure design, do not provide all the information that the prospective user needs.

Let us now add one level of complexity. Assume only 90% of the data are consistently perfectly received and 10% of the messages are lost because of communication performance problems. During the first communication try, 2 hours and 48 minutes are used. During the second retry about 17 minutes are needed to retrieve data from the remaining 10% of the transponders. During the third retry an additional 1.7 minutes are needed

A method for computing the first try communication performance will be given later. It is obvious that all meter readings can be obtained after numerous retries if the problem is due to communication performance only. This satisfies the requirement “we guarantee that all the meters can be read”. From an operational standpoint the numerous retries are less desirable because they are time consuming and constitute an inefficiency of the system. A high communication performance level during the first communication try is the most desirable.

Now assume that this communication system is to be used to read meters on an hourly basis. If there is only one path and one available channel, then this functional requirement cannot be met. With two paths and one channel per path, the functional requirements can still not be met. Using two paths and two channels per path, it barely meets the requirements if the performance is only 90%, and most likely it will fail to meet the requirement. High communication performance at the first try, coupled with multiple non-interfering physical communication paths, and the ability to use simultaneous multi-channel communication are the most desirable features to look for.

Another issue to consider is about how many communication retries can be tolerated. If a two-way command is issued to a remote device for data retrieval, the total time allotted for this transaction is approximately equal to the length of transmission time needed for the outgoing message plus the transmission time of the incoming message. If the inbound response is rejected because an error is detected in the message and a retry gives a correct response, then the previous message is probably contaminated by noise. However, if there is no response to a query, this could be caused by several other possibilities.

The outbound command never really goes out. The outbound command is contaminated by noise and the receiver cannot decode the message. The remote device is damaged or disconnected The power is out at the device or there is an outage at the network. The physical path for communication is temporarily out of service or rerouting is not available. A communication node is not functioning The inbound transmitter at the transponder does not send the inbound message.

8

The inbound receiver is not functional.

Without any communication network monitoring capability, the operator of the system will have no clue what causes a “no response” to an outbound command that he thinks was issued by the net server computer. If the system is designed to detect any of the possible causes mentioned above, the number of wasteful retries can be drastically reduced. Also the likelihood that data is lost forever during transit from a remote device to the receiver is minimized.

Excluding equipment damage, what are the usual causes of reduction in communication performance? There are several possible causes for degradation of communication performance. The most common one is high level of noise, which contaminates the transmitted data. Applying good filtering techniques, increasing signal strength and applying error detection and correction coding, in addition to using communication retries, can dramatically improve the communication performance. For a quick assessment of the communication performance, the following method can be used.

Assume there are N0 perfect transponders. Assume an outbound first try communication performance is hout %. The number of transponders that receives the outbound message is N0*(hout /100 ) Assume an inbound first try communication performance is hinb% The total successful two-way communication based on single try performance is N0*(hinb/100)* (hout /100)% Hence, the two-way communication first try performance is

htwo-way % = [N0*(hinb/100)* (hout /100)*100] / N0 = [hinb* hout /100]%

The performance numbers can be obtained by running a pilot system specifically set up for communication testing for a sufficient period of time. The number hout can be determined if the transponder is provided with an outbound communication counter. Every time an outbound message is successfully received, the counter is incremented by one. Upon retrieval of this number, it can be compared to the number of outbound commands issued by the Net Server computer and hout can be calculated. Once htwo-way has been determined after a certain trial period, then the magnitude of hinb can be obtained by dividing htwo-way by hout

A more serious issue is when the problem is related to some inherent characteristics of the physical communication paths. Consider the two – paths design in Fig. 5. If communication signals on one path interfere with communication signals on the other path, then both communications mutually destroy each other. This can happen if the physical communication paths are not mutually de-coupled or a spill over occurs from one path to the other. Then the two paths cannot be used simultaneously and one has to resort to sequential path utilization. This will drastically reduce the data throughput per unit time.

Expansion of functional capabilities using the same system that was initially adequate to do simple monthly revenue metering, may not necessarily meet future applications requirements. It is pertinent that one has a good understanding of what will be required by future Added – Value Applications.

In order to put functional capability in the proper relationship to communication system design, it is helpful to understand the requirements of some of the most commonly considered added value functions. The future chapters will treat some of the added value applications in detail.

d. The cost of reliability and added utilization costsRemote devices dispersed over a wide geographical area are exposed to man made and environmental hazards. The

environmental hazards are for example damaging transient voltages due to switching or lightning, high temperatures, ultraviolet light from the sun, humidity, fungus, oxidation, etc. and they all contribute to the reduction of life of a device. Several methods of testing have been designed to determine The Mean Time Between Failure (MTBF) or device life at the manufacturing level. The MTBF is quite often used as indicators in what to budget for replacement of failed units and operating costs.

The classical bathtub curve in Fig. 6 is quite often used as a measure of device reliability. For a large number of the same devices, this bathtub curve gives the probability of rate of failure as a function of time.

At the flat portion of the bathtub curve ( the time between T1 and T2 ) , the failure rate is assumed to be constant. This means that beyond the infant mortality region, the ratio of the number of devices that fail per unit time over the number of devices that survive from the original batch is a constant number.

If No is the number of units initially installed and Nf(t) denotes the number of failed units as a function of time t, then the number of survivors at the time t is Ns(t) = No – Nf(t)

9

Fig. 6Mathematically speaking, the constant failure rate is defined by the following expression:

The failure rate is =

Inserting the expression Ns(t) = No – Nf(t) into the previous equation and integrating it with respect to time yields the following :

The constant l = 1/T has the dimension of inverse time and T is normally defined as the Mean Time Between Failure (MTBF). A simple example illustrates the use of the formula. Suppose at time T 1 the number of installed transponders is N0 = 100,000. The manufacturer of the transponders claims that the MTBF = T is 15 years. The probability of the number of failed units after 1 year is,

Despite tight quality control during the manufacturing of the devices and a controlled burn-in process is used to weed out the devices that are weak and faulty, some of them may still pass the quality control tests. After installation they exhibit early failures and the beginning part of the bathtub curve, indicated by the infant mortality region, shows the higher failure rate. The magnitude for the MTBF is derived from the curve between T1 and T2.

If an estimate can be obtained about the annual failure rate, then the cost of replacing the failed units and the cost of reliability can be factored into the operational cost.

The cost of reliability is the cost incurred when a device does not meet the operating performance requirement. As an example, a load control device is instructed to turn off an air conditioner. However, for whatever reason – lost communication or device malfunction – the load control device does not turn off the air conditioner. There is a cost incurred because of non-performance or “not knowing” and is also part of the cost of reliability. If the communication performance can be predicted and each transponder can be checked for damages on a regular basis, then the cost of “non-performance” or “not-knowing” can be minimized. Its practical implementation requires a full two-way communication system. It is easy to see that such cost is directly related to device failure rate or life and the communication reliability.

After a complete system is delivered to the customer, it is also important to consider the following: What is needed to provide support in case of trouble with the system operation? Is the system user-friendly

enough to be handled by the owner without having to resort to expensive contractual support by the vendor or consultants?

Is the system sensitive to expansion? As an example, does one need more nodes and special tools and measuring equipment to install more remote devices? For power line carrier systems, will the addition of a capacitor bank on the line or an expansion of the network cause an impairment of the communication?

Are expensive consultation fees required to handle the cases previously mentioned? As an example, is it necessary to hire a team of experts who use very complex techniques to determine the type of changes required for a system expansion? These are all items that fall under the category of added utilization cost.

e. The issue of obsolescence.Electric utilities are used to buy power equipments, which have ten to twenty years of useful life. However, recent

advances in the micro-electronic devices and digital technology render many electronic and digital devices obsolete at a much shorter period. Digital electronic chips that were manufactured a few years ago may not be available anymore. In many

10

cases, the patents that pertain to those devices have not expired yet. It will be difficult for other manufacturers, which are willing to manufacture the obsolete chips that are still used in the present devices, because of patents issues. If not carefully planned ahead of time, replacement of damaged units, slowly phasing in new units with advanced capabilities, etc. will not be possible or very expensive to implement.

During initial contract negotiations, downwards compatibility is a requirement in addition to the more advanced capabilities that the new digital electronics can provide.

f. Organization, addressing methods and communication control With such a large number of transponders from which to retrieve data from, to control groups of units or single unit

and scattered over large area, tight organization of schedules, data flow control, actions, etc. are very important. Organizing certain transponders into functional groups, which perform the same type of task and can be reached by the same communication path using one outbound group command, will simplify the communication to these transponders. This requirement implies that one has to be able to determine ahead of time or during installation of the transponders that they are accessible through the same communication path and should not change in the course of time. If the communication path changes, then the change has to be identified quickly. A rerouting algorithm determines the new path. Transponders, intermediate nodes, main communication nodes require power to maintain their operational capabilities. Any outages in the electrical distribution network will affect the integrity of the communication network. The ability to identify the impact of an electric outage to the communication infrastructure is not a luxury, but is actually a basic requirement. Wasted communication efforts to de-energized transponders, nodes, etc., not knowing where to send the maintenance and repair crew etc. have to be considered in the communication network design.

The TWACS technology has a search capability. After a transponder is installed, a search command is issued to locate the transponder by using its permanent identification number. In most cases this identification number is the transponder’s factory serial number, which is burned into a permanent memory. Once a response is obtained, the central control computer knows how to reach that transponder and where the response can be detected and received. This path information is stored into the database as part of the total information pertaining to that transponder. Hence transponders, which have the same physical path, can be grouped and the group address downloaded at the transponders. If the communication technology allows the transponders to respond using several simultaneous parallel channels on the same path, then each transponder is not only assigned a group address, but also assigned the communication channel to use when responding to a group command.

Single group commands can be used to address such groups. Since they communicate using the same physical path, the location where the inbound responses can be received is known ahead of time. Meter reading transponders are excellent candidates for such functional groupings. Transactions can be scheduled, queued and the group responses, after they are received and decoded can subsequently be directed to the appropriate customer service computer. Two-way load control transponders can also be assigned into functional group addresses, although there is no immediate need for inbound group responses.

There is also a need for communication retry commands. They are needed to complement group commands in case some transponder responses from the same group are rejected because of errors in the inbound message. The retry is used on single transponders to retrieve transponder data that are previously impaired by noise during transmission.

However there is also a need to address individual meter reading transponders. This is necessary for accommodating those customers, who complain about their energy bills. Also those who are moving out of their residences want to have their electric meters read before checking out. Pre-paid metering customers are another type of customers that do not lend themselves to be conveniently assigned into groups that can respond to group commands.

There are other types of transponders that do not lend themselves to group addressing due to the unique individual tasks they have to accomplish. Capacitor bank switching transponders, service-disconnect transponders, transponders for alarms and information display are examples of such types of transponders. Their activation schedules are variable and they are not necessarily in coincidence with the schedules of meter reading. Some applications require an immediate sequence of actions and they are usually triggered by an alert or alarm signal.

Based on the discussions above it is clear how the communication system is going to be used. There will be a large number of scheduled operations, there will be numerous unscheduled single operations and there will be emergency cases where other operations have to be aborted in favor of a sequence of single operations in response to an emergency alert. Hence, a method has to be designed to allow some emergency conditions to receive priority access to the system and if necessary to override any prior set schedules without causing loss of data and error free resumption of scheduled operation after a priority interruption. It is important to understand how the communication system is going to be used and shared by various parties with different interests and priorities, as well as how to monitor and access the system.

11

g. Data security, system recovery and loss of dataRetrieved data from remote sites can change upon arrival at its final destination. The collected information can be

incorrectly registered and stored by the transponder. This is essentially a hardware problem and should be tested for existing flaws at the inter-phase between transponder and the device it monitors. Another problem is the contamination of data during transmission. This is normally due to noise. For binary encoded data for transmission, techniques are available to minimize the hazard of data contamination. For error detection, the CRC-16 coding is quite often used. To correct some errors in a message, the Hamming code and/or the BCH code or other error detection and correction codes are sometimes used. Any of these codes for error control are appended to the transmitted data and will increase the length of a message. Quite often a balance has to be considered between implementing an error detection and correction code, which reduces the data transmission efficiency and using a simpler and less powerful error detection and correction code coupled with a communication retry.

Loss of power to the transponders can happen. The loss can be of short duration due to a re-closing breaker operation. When a prolonged outage occurs, the loss of power can last for hours and days. Some of the information that was downloaded to the transponder from the central computer, such as a group address, a schedule, real-time clock and also monitored data, etc. will be lost unless provisions are available at the transponder to store all the pertinent information during loss of power. Not knowing which transponders have lost their remotely down-loaded addresses, spells for disaster. Capabilities, such as “power ride through” are intended to protect from loss of information during short duration loss of power. “Non-volatile” memory devices are usually used for long duration power outages.

There is one more important item to consider. Time stamped data monitoring that are real time clock activated, requires that an internal clock keeps the time, even when power is out.

The consequences of the loss of a group address can be disastrous. One normally does not know when it happens and which transponder is suffering from this loss. If during a group response, one or several transponders from the group are not responding, it is impossible to know ahead of time what happened to the transponders. One might guess that there is a communication problem, the transponders are bad, etc. If ultimately one guesses correctly after several retries to communicate, then the group address has to be reloaded to each of the transponder. To load this group address, the path parameters of the transponder in relation to its original serial number have to be extracted from the database.

Without “non-volatile” memory, time stamped collected data will also be lost forever. Some of these same types of problems can also happen with data concentrators. Suggestions are made to use batteries to maintain the real time clock and the memory. For a small number of devices, using batteries in the remote devices will pose no problems. However, when one deals with very large numbers of devices containing batteries, there may be problems that can be very difficult to manage. Batteries have life and the batteries may not have been installed at the same time in all the transponders. One has to keep track of replacement schedules, procure the replacement batteries and prepare the schedule for the maintenance crew to replace the batteries. In many instances, the cost of one trip to replace the battery in a transponder can be a large fraction of the cost of a new transponder.

Battery technology also has made great strides in terms of energy stored, longevity and miniaturization. New digital electronic devices in the transponder also require less power. All this tend to reduce the problems of battery replacements.

h. Types of data, data warehousing and accessibility.The need to identify the communication path parameters of the transponders is already mentioned. The path

parameter of a transponder is used for communication purposes only and is not required to be related to its physical location on the distribution network, the phase, feeder number, bus number, substation name, etc. although some communication technologies (power line carrier ) already have this information in the communication path parameters. Radio frequency communication infrastructure does not require feeder, phase, bus, etc. information to reach the transponders. Ultimately this may prove to be a disadvantage for the development of many applications that will be very useful for utility applications.

What attributes do the monitored data have to have to render them useful for advanced applications beyond simple billing of energy usage. Time stamped interval energy metering which can be synchronized for the whole or part of the distribution network render the data more useful for demand response. If better stratification of the information can be used to optimize demand response by substation or groups of substations, then clustering the meter reading transponders by phase, feeder number, distribution substation bus, in addition to time will be required.

In the later chapters of this document, it will be obvious why all those attributes tied to a data point are needed for comprehensive implementation of many of the utility functions.

To set the stage for a discussion about the size of a data warehouse, we use the following illustration. Assume an automatic meter reading system is installed with the ability to collect time stamped synchronous interval energy data. The

12

interval duration is half an hour and there are one million customers. A typical metering data might contain the following pieces of information:

KWHRs consumed during the interval Time stamped data pertaining to the interval Transponder address Other possible information which modern digital electronics Smart Meters can provide, such as local

voltage, KVAHr, Total Harmonic Distortion (THD), etc.For each customer there will be 48 data points per 24 hours to be stored in the data warehouse data base. For the total

population there will be 48,000,000 data points generated per every 24 hours. It does not require much of an imagination to realize the massive amount of data that has to be warehoused. It is true that large capacity data storage devices are now available.

Another important issue is how the data is stored in an organized fashion for easy and fast retrieval. One should also understand that the ability to collect interval time-stamped metering data does not necessary imply that one should collect the same type data from all customers in a continuous manner. Algorithms for statistical assessments and categorizing classes of customers, special applications, etc. will dictate the need for the types of data to collect and to store.

Smart Grid applications also require information about the physical locations of the transponders at the distribution network and in some cases also their geographical locations. Hence, for completeness data monitored and collected by each transponder require what may be defined by a set of parameters such as the time the data is taken, the electrical network or distribution grid parameters and geographical coordinates. Suppose the physical data that is measured is defined by the letter Q. Then it is possible to describe Q as follows. Q ( t, s, f, Φ, lx , g, …….)Where t ………..time stamped information Φ……….. feeder phase

s ………..substation name lx………...location or line segment on the electrical circuit f ………….feeder number g………....geographical location In the later chapters of this document, it will be obvious why all those attributes tied to a data point are needed for comprehensive implementation of many of the utility functions.

The transponder addresses should also be easily related to parts of the electric network and maybe also to the physical addresses of the electric utility customers. One cannot sufficiently stress the importance of the last one mentioned. Its importance will be clear once we get into the discussion of the “SMART GRID applications”.

3. THE TWACS COMMUNICATION

1. Introduction

A “power-line communication technology” with a proven track record and designed specifically for utility functional applications, is already known by the electric utility industry by its trade name TWACS. TWACS stands for Two-Way Automatic Communication System. This technology uses the electric utility distribution network for its physical communication medium. The propagation characteristics are excellent because the physical behavior of the signal is similar to the transient response of a network to a short perturbation. The controlled perturbation technique is used to generate outbound signals at the medium voltage substation bus and the signals appear at every part of the network served by the medium voltage bus.

The TWACS communication technology generates a small controlled distortion at specified regions of the power frequency voltage or load current waveform.

The voltage modulation is accomplished by drawing a narrow controlled current pulse near the bus voltage zero crossing as shown in Fig. 7. Distribution substation transformers serving a medium voltage network typically have impedance levels higher than distribution step-down transformers. Typical values range between 8.00% and 12.00%. TWACS communication technology takes advantage of this higher impedance feature to generate outbound signals. This incremental pulse type load causes a voltage drop in the distribution substation leakage impedance at the distribution bus and is transient oscillatory.This voltage perturbation is the distribution network transient oscillatory response and serves as a voltage modulation for outbound signaling. The transient oscillatory frequency ranges between 200 Hz to 500 Hz at 60 Hz and as well at 50 Hz systems. Typical outbound signals extracted at the service voltage levels are shown in Fig. 8. The

13

modulation of the system voltage is binary encoded for transmission of information. Two full cycles of the 60 or 50 Hz are needed to define a binary outbound bit.

Fig.7

Fig.8. (Actual Outbound Signals)

To generate an inbound signal, the transponder draws a controlled current pulse at the service voltage level which is

superimposed on the bus or feeder load current. Extraction of the inbound signal is from the current transformers at the distribution substations. Generation of inbound signals

Fig. 9 Actual inbound signals extracted from the current transformers at the substation are shown in Fig. 10. The first plot is extracted from a bus current transformer. The second one is extracted from a neutral current transformer.

14

Fig. 10 The low frequency signal propagation does not suffer from standing wave problems, attenuation by capacitor banks and underground cables. Field studies show that distances up to 100 miles or more pose no problem for communication. These long distribution feeders are commonly encountered in rural areas.

An important feature of this outbound signal is its phasor specific nature. This signal is present near the zero-crossing of all voltages at the distribution network whose phasors correspond to the phasor of the modulated bus voltage.

For a three-phase system, one can identify 3 basic phase voltage phasors,`VAN,`VBN and`VCN for the line to neutral voltages. The line-to-line voltage phasors`VAB,`VBC and`VCA are linear combinations of the 3 base phase-to-neutral voltage phasors. An outbound signal injected at the distribution substation bus on a specific phase voltage follows that phasor everywhere in the system. It goes across different 3-phase step-down transformers of all possible different winding configurations or single phase transformers without seeing appreciable signal attenuation. The inbound signal generated at the transponder follows the load current all the way to the distribution bus.

This unique feature allows one to use the communication path concept. As an illustration, we use the following example:

A search algorithm is used to “find” a transponder. For example, the transponder has a serial number NKNKNK. By sending a query to this transponder on all possible phases to respond back, it is found that the transponder responds to the query when the outbound message is send on phase VAN. The inbound message is extracted from the current transformer of phase A.

The communication path parameter is now defined as follows:i. To reach transponder NKNKNK, the outbound path is phase VAN.ii. The inbound message can be extracted from the feeder or bus current transformer of

phase A. For all subsequent transactions to transponder number NKNKNK, one uses the defined path. In the database, each transponder number is coupled to a unique path parameter that is found by the search

algorithm.The use of low frequency signals practically implies that the baud rate is low. The future requirements for gathering large

amounts of information from remote sites can only be handled by using simultaneous multi-channel communications, reducing all unnecessary information in the message structure and a high degree of communication performance. The TWACS communication can use up to six simultaneous inbound communication channels per physical path per feeder per distribution substation bus. This means that 6 remote devices can respond simultaneously as a response to a single group outbound command per feeder per distribution substation bus.

To increase the data throughput from the remote devices, designs are currently being tested to allow all phases at all the feeders served by a substation to operate independently and simultaneously.

Issues Pertaining Non Power-line Types of Communications Technologies.The communication infrastructure can also cause unpredictable communication delays. Some are caused by unreliable

communication requiring many retries, some may be caused by undetected malfunction of the nodes and some are due to inherent technical and physical limitations of the communication technology. Unless the system is designed to overcome these problems, then obviously the system cannot support hourly meter reading.

In the case of systems requiring large number of nodes, the messages have to hop from node to node in a sequential fashion before reaching their final destination causing arrival delays or loss of data. A node will keep sending the message to a next node and will only stop retrying and flush the message from its own memory until the other node has acknowledged that it has received the message correctly. If no acknowledgment is received because of communication problems or problems at the receiving node, massive delays occur because of the numerous automatic retries. The node becomes a

15

communication bottleneck. This problem can be reduced in severity by automating alternate path selection, in other words “a good routing algorithm” is required. Downloading real time to all units for this system is difficult.

To optimize the operation of hourly meter reading, detection of faulty parts of the electric and communication network is necessary. This helps to prevent wasting time for communication retries and guessing where the problem is. When part of the distribution network is repaired and re-energized for normal operation, reactivation of the communication to the previously de-energized electric meters has to be done quickly in a systematic fashion. Knowing the faulty part of the network is only useful if it can be related to its location and identification of the de-energized meters.

For a communication network that is independent from the electric distribution network, it is necessary to know which communication nodes serve those de-energized meters and which communication nodes are also de-energized because of the power outage. For communication systems that use the power line for communication mediums, the loss of power delivery path to an area also means a loss of communication to that area.

To support one aspect of advanced AMR – in this case synchronized hourly meter reading – outage management of the distribution network and knowing the meter locations at the distribution network in relation with the overlaying communication network are critical for optimization of its implementation.

Here is one example, where a successful implementation of an automatic meter reading function, requires that an outage management and system restoration function and other supporting functions such as network management functions and the communication routing control be implemented at the same time. Associated with this is the avoidance of the “cost of not knowing”.

Electronic meters are coming down in price and becoming more reliable. They can gather and store other types of useful information. When the stored information is constantly updated, timely retrieval of the information from the metering device is needed. Otherwise it will be lost forever.

4.Advanced Metering and Value Smart Grid Applications

In this section, only fixed communication networks are considered. The most popular applications that are nowadays considered by many electric utilities can be grouped under several main categories. Many of these applications can be considered as part of Smart Grid operation. Remote monitoring of local nuisance type events is also discussed in this manuscript. A possible method of categorizing the applications is shown below.

I. Customer Services and Demand Response Electric Energy Consumption Metering Gas and Water Consumption Metering Remote Service Connect and Disconnect Prepay Metering Load Management/Demand Response Time Of Use Alarm

II. Improvement of Service Reliability and Optimization of Energy Delivery (Assets Management). Outage Management and System Restoration Integrated Voltage and VAR control Remote Breaker or Switch control Feeder Load Balancing and Loss Management Power Quality Monitoring High Impedance Faults Distributed Generation Hybrid Electric Vehicles.

III. Supporting Functions Communication Network Monitoring and Control Extension of SCADA Capability into the Distribution Network AM/FM Systems Data Management Multi-party Users Etc.

16

The list above is by no means exhaustive and to discuss in detail the requirements of each possible application and function is beyond the scope of this manuscript. However it is important to note that many of these functions and applications cannot be viewed and discussed independently without noticing the existence of synergism amongst them.

The ability to collect synchronized time stamped interval metering data opens the door to other numerous applications that are previously unheard of or deemed very difficult to implement. If the word “metering” is interpreted in a more generic sense, then one can include all energy related metering, voltage monitoring, power quality monitoring, etc. under collected synchronized time stamped interval metering data. The time duration of an interval depends on the application of the collected information.

A word of caution is necessary about the meaning of interval. For a total energy-metering device such as a KWHr meter, when the meter is read at the time T1, the data is the total energy consumption registration up to the time T1. No time varying fluctuations of energy used can be extracted from the reading. If the reading at T1 is P1 and at T2 the reading is P2, then the magnitude of (P2 – P1) is the energy consumed during the interval D, where D = (T2 – T1). The smaller the interval D is chosen, the more accurate one can characterize the load behavior. For load survey and demand metering, the value of D is typically 15 minutes. For advanced metering applications D is typically 30 or 60 minutes. Normally this degree of resolution is adequate for load profiling.

Interval voltage monitoring may require another level of sub-processing before it is read and collected. Voltages are instantaneous entities. If a voltage V1 is read at time T1, then this is the instantaneous RMS voltage magnitude at T1. The type of information one needs about the behavior of the voltage will determine what is collected, processed and stored within the time interval D = (T2 – T1). The duration of the interval may also be another variable to be predetermined. For voltage profiling of a feeder, the average value for the interval D and time stamped is more important than short time excursions of the peak values of the voltage within the time interval D. If one is interested in intermittent burst transients that may affect digital electronic devices, then some degree of preprocessing of the collected voltage data for a certain time interval D will be required before retrieval. Some bad looking voltages obtained from the field using a data acquisition unit, are shown in Fig. 20 and Fig. 21. Time stamping the retrieved data is essential for correlating the gathered data with other events that may occur at the distribution network. The same arguments as for the time stamped time interval voltage monitoring apply to power quality monitoring.

Discussions in the industry have shown the benefits of interval meter reading. At the moment hourly or half hourly meter reading interval is deemed adequate. The concept was initially galvanized by the idea of retail wheeling. A whole new frontier of retail business is created by the fact that electric energy is now considered a commodity. Energy retailers can sell it in the open market at competitive prices. The hourly time-stamped readings provide sufficient accuracy for apportioning energy usage billing by the different retailers. A data warehouse can store all the time stamped interval metered data that can be made available to different parties.

There is also another important aspect of hourly meter reading application. Hourly time- stamped meter reading data, which are time synchronized can be used for other applications. The cumulative customer time interval energy usage profiles can be used to determine coincident peak demand, the shoulders and valleys for different classes of customers, grouped by distribution substation, by distribution transformer, etc. More accurate load factor prediction for specific areas, coincident hourly load versus time, etc. can be used to develop comprehensive rate structures, for planning and developing load control strategies and also to provide design data for future network expansion. The concept of time stamping and synchronizing the hourly reading interval of all the meters, require a two-way communication capability that can accommodate high data throughput. Data throughput will be quite large and the communication time to obtain the data from all the remote metering transponders should not cause loss of synchronization and loss of data.

The capability to collect synchronized time stamped interval metering data and identifying the grid parameters of the location where the data is collected opens the door for new types of energy pricing and billing systems, but also to other applications, such as Time-of-Use rates, Prepay System, Assets Management, Outage Management, etc. Historical data can be used for planning, assessing system losses, energy theft, etc.

Energy utilization profiling can be used to cluster similar types of customers by phase, feeder and bus. Peak coincident load of customers served by a distribution transformer gives a good indication of the transformer loading as a function of time. Duration of overloads can be monitored accurately.

When a power outage occurs, all the remote units that are de-energized have to be put back into the proper synchronization when the power comes back. Proper synchronization can be accomplished by broadcasting real time to all the transponders as often as possible. The re-energized transponders then re-synchronize themselves. Another method can be accomplished by providing each metering transponder with its own battery operated real time clock. However, this method is not recommended unless good maintenance and well-defined battery replacement schedules are adhered to. The labor cost to replace a battery can be high in comparison to the price of the transponder.

17

Customer Services and Demand Side Response

a.Advanced Metering Applications Retail wheeling is becoming a hot concept in many countries and may slowly penetrate into the USA in the near future. Generating, delivery and selling of electrical energy are normally monopolized by one utility company. Retail wheeling allows different companies to sell electrical energy to customers in a competitive fashion. These companies do not necessarily own the energy delivery infrastructure, but they are allowed to buy bulk energy for resale as a commodity at competitive prices. Metering data collection and warehousing can be handled by another company, which provides services to the energy retailers. The following example illustrates the billing scenario.

Retailer ABC sells energy to customer A at a rate of Y cents per kilowatt-hour. The sales contract is for a period of X days starting and ending at noon beginning at a certain date. For a fee the retailer ABC obtains the metering data for billing from the company, which collects and warehouses all interval time stamped interval energy metering data.

The telephone industry has done something similar for quite some time already.Retailer ABC sells energy to customer A at a rate of Y cents per kilowatt-hour. The sales contract is for a period of

X days starting and ending at noon beginning at a certain date. For a fee the retailer ABC obtains the metering data for billing from the company, which collects and warehouses all interval time stamped interval energy metering data.

The telephone industry has done something similar for quite some time already. Pre-pay metering has been touted for many years by several utilities and is also gaining ground in the USA. Pre-pay metering has been practiced in Europe for many years. A simplified pre-pay metering scenario is the following.

A customer tells the electric utility or the energy retailer the intention to pre-pay for a certain amount of KWHr at a mutually agreed rate. The money is deposited by the customer directly to the customer service department of the utility or through a financial institution. At the end of the service contract, when the customer has used up the pre-paid amount of KWHr, then the customer deposits more money for a new service agreement.

This simple illustration of a possible transaction may not be as simple from an operating standpoint as it seems. On the date and time the pre-payment agreement is in effect, the customer meter has to be read. Before the customer runs out of the pre-paid KWHr, should the customer be alerted to deposit more money? How soon should the customer be alerted and when the customer does not respond, should the customer be given a grace period ?

Can the customer services be disconnected for non-payment? Disconnecting a customer without warning may cause grave repercussions to the energy provider as well as to the customer. There may be life savings equipment that needs power. Business transaction through the internet system at that moment will be interrupted. Disconnecting the power also implies that a service disconnect switch is installed at the customer’s premises.

The other question is how to alert the customer that it is time to deposit more money? If the service disconnect occurs on a holiday when all financial institutions or the utilities’ customer services are closed, what provisions are available to the customer to avoid service disconnect.

Another important aspect is customer education. Customers have to learn what a KWHr means and what you can do with it. To help the prospective customer, his normal energy utilization profile can be used as an indicator about his expected energy usage. This will allow the customer to make intelligent decisions on how much energy to buy. Customer profiling will also help the energy retailer how soon to alert the customer to deposit more money.

These items posed as questions have to be addressed ahead of time and to determine to what extent the ability of an AMR system and its communication capability can support the pre-payment application. Remotely Operated Service Disconnect.

Remotely service disconnect devices are now available. The service disconnect switches use the existing communication system to disconnect or to connect power to a residence when a present the tenant is moving out or a new tenant is moving in. It is normally coupled with reading the electric meter prior to connecting or disconnecting the power to the premises. Typical applications are in areas with high turnover rate, such as campuses, time-share apartments, etc.

Some attention has been paid to the concept of using the service disconnect switch to disconnect power to customers who do not pay their electric bills or to pre-paying customers whose total energy use has reached the prepaid amount of the energy consumption. In many states, total disconnect of power to a customer for whatever reason it may be, is prohibited during certain periods of the year. This is true if severe hardship is imposed on the customer as penalty of non-payment of the electric bill.

Some of the basic considerations in this case are:1. The contact ratings and the thermal ratings of the switch2. Short circuit handling capability3. Switch status checking4. Manual operation

18

5. Device address and its link to the metering transponder address6. On site operational testing and activation of the switch7. Some type of customer alert before total power disconnect

One of the safety features being considered is not to allow connecting power to the customer premises remotely when no one is there. The suggested method is to arm the service disconnect switch remotely. The new tenant hence activates the switch upon entry of the premises. This avoids the possibility of causing fire hazard when no one is presence. There were cases where the previous tenant left plastic or papers on top of stoves which was not turned off, but left the apartment because power was shut off by the service disconnect switch.

Gas and water metering.Gas and water metering are quite often coupled to electric meter reading transponders through short hop radio links or

hard-wired links. The transmitters at the gas and water metering devices require their own power supplies. For hard-wired communication links to the gas meters, special precautions are required to insure safety at the interface. Electric safety barriers are now available.

Interval gas and water meter reading has yet to take off in a meaningful fashion for the gas and water utilities. Most of the present applications are for billing only.

Quite often several gas or water meters are coupled to one particular customer’s electric meter through a radio frequency link. The RF transmitters at the water or gas meter have their own battery power supplies. The electric meter hence serves as a data concentrator for the gas or water meters. Water meters sometimes pose problems for installing a communication link to them. In many areas where winters can be severe, water meters are mounted inside manholes with metal covers or in the basements. RF signals have difficulties penetrating the metallic cover or getting access to the basement.

One very important issue to consider is the need to maintain a database of the batteries that provide power to the RF transmitters. The average operating life has to be established first even though claims are made that many of the batteries will last 20 years. The date when they are installed, the operating mode of the transmitters, how long is the shelf life of the batteries. If the RF modules constantly transmit metering information from the water or gas meter to the data concentrator at the electric meter, then obviously this mode of operation reduces the battery life faster than when the transmitters only transmit when queried by the data concentrator. When the RF modules are not transmitting, the power consumptions are merely to maintain the electronics alive and to read and store the metering data. Batteries when not in use have a shelf life. Usually the expiration of life data is available and can be provided by the manufacturer. Hence, once the required battery information is available in the database for each RF module, then a battery replacement schedule can be set up.

Load ManagementThe word Load Management was at one time coined with the concept of deferring or avoiding the need to add

peaking-generation units. As a matter of fact Load Management can be considered as dispatched negative generation. To develop the control algorithm needed to accomplish the load peak shaving task, load survey data on sample loads at many customer sites are gathered and statistically processed and compared with generation data and also whether the coincident peak demands occur at the same time as the system peak demand. Only specific classes of customers who qualify for load management are surveyed. A load cycling switching strategy is then used to accomplish the task.

Time synchronized interval energy metering information provides a much more accurate assessment of coincident peak demand, coincident demand, etc. which can be used for reconciliation of energy usage and generation. Energy retailers or electric distribution companies, which purchase energy from the generating companies, will in all likelihood be penalized if they allow the load to exceed a contractually agreed peak demand. Electric Cooperatives in the USA understand the meaning of demand ratchet very well. Time synchronized interval energy metering of all the customers served by a distribution substation can provide accurate total load and highly selective load trend data. The electric power distribution company can use this information to activate selective load control on non-essential loads and thereby prevent the total load to exceed the demand ratchet level.

Time-of-UseTime-of-Use rates are used as one form of voluntary load control. Different rate structures apply for peak, shoulder

and during base load periods. This type of rate structure may be seasonal, weekly or daily and may not apply to the whole population. Time synchronized interval energy meter reading can be used to obtain a profile of how energy is used by the customers. These profiles can be used to select candidates for Time-of-Use rates. Small customers whose loads are constant

19

most of the time are of course excluded from this rate structure. Time synchronized interval reading of the electric meters will also generate information how effective Time-of-Use rate is to reduce total system peak demand. Utilities use load forecasting that may be tied to weather forecast, available energy for wheeling, etc. to determine when to activate Time-of-Use rates. It is then at the discretion of the customer to curtail demand during hours of high system peak demand.

Demand Response Strategies to Improve the Load Factor to Minimize Network Losses.Load control in a classical sense has been used for deferring the peak load during periods of high demand. This

eliminates the need to build a peaking generating unit to cope with peak demand. Normally load control is accomplished by turning on and turning off load types that are cyclic in nature. Examples of such loads are central air conditioners, water heaters and electric heating systems. In a normal situation, these appliances operate in random cyclic fashion at random times. This load diversity generates an average load that is less than the sum of individual peak demands.

To implement a control strategy, a population of similar appliances, such as air conditioners is divided into groups. The first group is turned on for DT1 minutes and turned off for a period of DT2 minutes. The second group uses the same on-off cycle but starts later than group one. The same time shift is applied to the other consecutive groups. After repeating this process several times, the air conditioners start to follow the control algorithm load pattern. The amount of peak load shifted depends on the number of groups and group size and also on the choice of DT1, DT2 and the time delay between consecutive groups. This technique when properly applied will improve the system load factor. In case of emergency, a scram command is used to disconnect all controllable loads.

An intuitive understanding can be developed using the figure shown above. In the first case a source supplies two

identical loads through a conductor, which has a resistance R. The load duration is T. Calculations show that the loss in the conductor is PLOSS 1= 4I2RT. Also, the source has to be able to handle the double load for a duration of time T. In the second case, the load second load comes in after the first one has ended. The loss in the conductor is 2I2RT. The conductor loss is cut 50% and the source only has to be able to carry one load only. Hence peak demand shaving improves the load factor, which reduces the losses in the supply network, but also reduces the peak demand requirement for the generation.

In order to design the correct control algorithm, load survey data are taken from groups of loads that are assumed to be similar in their usage profile and are candidates to participate in the load control program. For one-way systems, special monitoring devices are installed at the premises. For two-way systems, special AMR transponders can be used, which can do time synchronized interval meter reading set at 15 minutes interval.

Two-way systems also have the advantage to monitor the health of each transponder and identify those units that require to be replaced. The cost of reliability is hence reduced. If during load control, the time synchronized interval meter reading is also operational, then one has immediate assessment of the effectiveness of the load control strategy.

Lately the term demand response keeps appearing in the technical literature. This is a generic term for a variety of novel load control strategies that is now made possible because of the availability of a communication system that can be used for time synchronized interval meter reading. The ability to do load usage profiling of every energy user, helps to categorize customers by different energy usage classes and what best control or pricing strategy to use for those different classes of customers.

Averting Rolling Blackouts and Reducing the Impact of Cold Load PickupWhen the spinning reserve of an electric power system reaches a critical point, utilities quite often practice brown outs to

shave demand. Brown outs are essentially disconnecting power to customers for a specified period of time on a rotational basis. This practice continues until total system load drops sufficiently to insure that a safe margin for the spinning reserve is assured.

Sometimes rolling blackouts can cause problems due to cold load pickup. If a feeder has a large number of customers whose loads are cyclic, because of the load diversity, most of the time the turn on and turn off of the different loads do not coincide. The statistical spread of the cyclic loads with time causes a net load that is a percentage of the total sum of all the

20

cyclic loads when they coincide. Typical cyclic loads are central air conditioners, electric central heating systems, electric water heaters, etc.

If power is disconnected to all these loads - for instance all central air conditioners – for quite some time, then all these houses internal temperatures will increase. Upon power restoration, all the central air conditioners start to turn on simultaneously and cause a large inrush current. This large inrush current is called cold load pickup and may be large enough to cause feeder circuit breaker tripping.

If load control is available, prior to disconnecting power to the feeder, all the large cyclic load appliances can be disconnected. Then the next step is to open the feeder breaker. Upon power restoration, all cyclic loads can be restored in a staggered fashion in order to avoid large inrush currents.

Monitoring of Overloading of Distribution Transformers.Coincident demand monitoring of loads served by a transformer provides valuable information about the transformer

overloading, when it occurs and how long it lasts. This will give the electric utilities a method to decide whether a distribution transformer should be replaced by another one with a bigger capacity. Detection of Unauthorized Use of Electric Energy.

In many cases, unauthorized by-passing of the electric meter is performed by a customer. The meter itself is never unplugged. By installing another meter at the transformer and perform hourly synchronized interval meter reading together with all the electric meters served by the same transformer will quickly tell that an energy theft has occurred. The net consumption reading at the transformer cannot be reconciled by the sum total of the customers’ meter readings. The time of occurrence and its duration can easily be determined. In some countries where unauthorized use of electricity is rampant and out of control, all individual meters are put into one single pole mounted box next to the distribution transformer serving the customers. All service wires to the customer premises come from the metering box. Remote metering capability is used to get billing data from these meters.

AlarmThe ability to send alarm signals globally to certain areas or to send an alarm to a single location, will find many

possible applications. Some of the possible applications are the following:o Civil Defense alerto Traffic re-routingo Peak load alert to energy users on Time of Use rates to curtail load.o Alert to pre-pay metering customers that they need to deposit more funds.o Down-load peak load alert to home automation systems to control demand.o Alarm signals indicating that power is going to be disconnected by the service disconnect switch.o Etc.

Improvement of Service Reliability and Optimization of Energy Delivery.

Electric Utility Network Outage ManagementOutages do occur in the electric distribution network. The effects of an outage are felt not only by the users of electric

power, but also by the communication infrastructure. Outages can occur at an individual customer premise, to a group of customers served by a distribution feeder or to a large area due to a transmission line fault. When an outage occurs, not only are remote communication devices out of power, but some communication nodes become de-energized too if no provision is made for a standby power source.

Single customer outages and total system blackouts will not be discussed here. Most of the outages occur due to line faults at the distribution feeder, which cause protective devices to disconnect certain segments of the feeder network. They can last for a fraction of a second due to a recloser operation, and some can last for many hours or days if damage occurs on the line or pole. Occurrence of an outage is unpredictable, and we define outage management as the ability to detect an outage and its extent, alert the system operator of the outage and notify the necessary department about this event. Fault isolation and system recovery are part of outage management.

There are attempts to create circuit models of a feeder with its many laterals which hopefully provide information about the location of the fault by monitoring the magnitude of the fault current at the distribution substation. Distant relays provide similar capability for transmission circuits. However, due to the complexity of the distribution network with a mix of single phase, two phase and three phase laterals and circuit configurations which do not lend itself to analysis using symmetrical components, it will be very difficult to determine the physical location of the fault from the fault current data.

21

Many distribution networks are designed and protected by protective devices. By properly coordinating the protective devices (selective coordination), it is possible to isolate the faulted segment of the network to the nearest fault location. Hence, all transponders connected to the part of the circuit that is de-energized will also be de-energized. Systems that use the distribution network for communication cannot communicate to the de-energized transponders. The same thing applies to other fixed network systems, unless the transponders have batteries.

The ability to monitor transponders that are de-energized, and which can be related to the part of the network that has lost power, will be defined as outage mapping. For a properly coordinated protection system, when the fault occurs beyond a certain protective device, it will isolate the faulty section of the network from the source. This automatic isolation of part of the network is only the first step of the fault isolation process. The fault occurs only at a small segment of the de-energized part of the network. Disconnecting the faulty segment completes the process of fault isolation. Restoration of power means to re-energize the healthy remaining portion of the network from a different source.

A very useful application, which takes advantage of the ability to communicate with the transponders, is the outage mapping function. The model as shown in Fig. 11 is used for illustration. A main distribution feeder has 7 laterals protected by fuses F1, F2, ……., F7. Only 2 transponders are shown per lateral, indicated by the letters T with the appropriate subscripts. Assume a fault occurs at a location as indicated in the figure and the breaker S1 opens.

Fig. 11. Transponders T5, T6, ………, T13 and T14 are all de-energized. If a problem occurs on the feeder and an alarm signal is

available from a SCADA RTU or relay to trigger a polling sequence of all 14 transponders, then the responses quickly show that the first 4 transponders T1, T2, T3 and T4 are still energized and the remaining 10 are out of power. If the transponder locations can be related to a circuit map, one can easily infer that circuit breaker S1 is open. It is even better if the de-energized circuit can be related to the topological map of the area.

Several methods have been proposed to accomplish this outage mapping task. Continuous polling of all the transponders. This method is fine if a two-way very high speed communication technology is available and the polling action does not take away the ability to perform other equally important functions.

For slower communication technologies, the best method is the use of selective polling. If an alarm can activate the polling sequence, then the communication for outage mapping is only active during the polling sequence. The more precise the information content of the alarm is, the lesser the time is needed for polling. The information content of the alarm can be the distribution substation bus number, feeder and phase number, line to neutral or line-to-line fault, etc. which are typically available from a SCADA type device or an Intelligent Fault Detector at the substation.

For power line communications, the method of selective polling is simpler because a problem with the power delivery infrastructure is the same problem for the communication infrastructure. For RF technologies or other hard-wired technologies, the problem is different. An outage not only affects the transponders but may also de-energize some of the main communication nodes. Either a special communication routing algorithm has to be designed, or the main nodes have to be equipped with standby power. In addition, it is also necessary to know what feeder, which bus and which distribution substation serves that feeder segment.

A display of the geographical location of the de-energized network segment on the dispatch computer screen will help the repair crew to find the fault location. Hence, a good database of the physical locations of the protective devices, circuit segments, and other circuit information related to transponders addresses used for detecting outages is necessary.

This outage information also helps to avoid unnecessary communications to the remote devices that are de-energized, and the same information is also used to identify sub-nodes that are de-energized.

22

Fault isolation is accomplished by opening breaker or switch S2 and restoration is accomplished by connecting the part of the network beyond S2 to another source. A word of caution is needed for in case that switch S2 is a manually operated switch without any fault current sensing device. The one is not quite sure that the fault location is between S1 and S2. In the case there is a distributed generator or large induction motors with capacitor banks providing the necessary magnetizing current, it may be necessary to consider the use of reverse power sensors at S2.

As simple as it may sound, restoration can be a very complex function if one tries to automate and optimize the transfer. If several options are available regarding where to connect the network beyond S2, what is the basis for considering a particular option if the expected repair will take a long time. Some of the reasons to consider are listed below:

1. Overloading of the other source after a certain time.2. Difficulties adjusting voltage regulation and capacitor bank control to minimize losses while maintaining a good voltage profile. For communication technologies that use the distribution network for communication medium, another problem may

exist. In the addressing mode, a group of transponders can be addressed by a single group command. Assume a group of N transponders belongs to one group, out of which N1 are located before switch or breaker S1 and the remaining ones (N - N1) are connected to part of the circuit between S1 and S2. If S1 is open due to the fault and S2 is opened to isolate the faulty segment, and one does not have a clue how large N1 is, then the response to a group command might result in multiple retries to try to retrieve data from the (N - N1) transponders. Organizing group addresses of transponders has to take into account their locations in relationship to the switching or protective devices of the circuit. No group should be separated into two or more subgroups of unknown numbers of transponders due to circuit disconnects by the protective devices.

Feeder Load BalancingEven though a 3-phase source can be perfectly balanced, due to uneven load distribution amongst the phases,

especially along the distribution feeders, the unequal voltage drops on the different phases cause voltage unbalances at the various locations on a feeder.

Under steady state conditions, unbalanced voltages can be decomposed into positive, negative and zero-sequence voltages. In 3-phase electric machines, the negative-sequence voltages generate negative-sequence magnetizing currents. These negative-sequence magnetizing currents generate a rotating magnetic field, which rotates in reverse direction with respect to the rotor. This negatively rotating magnetic field creates a retardation torque. In addition, due to the reverse rotation of the negative sequence torque with respect to the rotating rotor, the additional induced eddy current losses by the reverse rotational field in the magnetic material of the machine will increase dramatically. Operating engineers quite often prefer to express the unbalance in terms of dissymmetry, which is defined as the ratio of the negative sequence voltage to the positive sequence voltage.

FIG.12

To express this in terms of the actual voltages, if the three phase line to line voltages form a triangle and the absolute voltages can be expressed as a =÷Vab÷, b = ÷Vbc÷ and the third one as c = ÷Vca÷ then the Voltage Unbalance Factor is :

In this expression, assuming that ÷Vab÷ has the largest magnitude, then x and y have the following values :

x = b/a and y = c/a

23

In many European countries the standards limit the unbalance to 2% at the medium voltage level. In the USA ANSI C84.1 Annex 1 and NEMA MG1, for voltage unbalances in excess of 1%, a de-rating of motors is needed. Unbalanced loading of a distribution feeder also dramatically increases the copper losses.

In grounded WYE 3-phase networks, the sum of the zero- sequence currents in the phase wires flows in the neutral. The neutral current also quite often generate stray currents in the soil in the vicinity of the neutral wire. This is due to the standard practice of grounding the neutral wire at the poles and also at the step-down transformers. In practically all instances, the center tap at the 240 V service voltage side of the transformer is also grounded and tied to the ground at the distribution voltage side. In addition, the safety ground at the customer premises is tied to the incoming neutral wire. As a result, at each distribution transformer high side grounded neutral, other remote grounds provide additional grounding shunts. This can lead to stray currents in the soil and metallic water pipes. If the stray current is large enough, step-voltages can cause problems to animals and human beings.

Conditions of unbalance are not constant and may start at certain hours of the day lasting for a few hours. Time stamped interval meter reading of all the energy users of each feeder can be used to determine how the loads are distributed amongst the phases at specific times of the day. The coincident hourly demands can be used to determine how to redistribute the loads over the 3-phases of the feeder in order to reduce unbalanced loading.

FIG.13

Some of the necessary steps to accomplish this are the following. Also, the assumption is made that load transfers across the feeder phases are permanent. No remotely controlled load transfer switches are used.

1. Obtain several days of synchronized time stamped interval meter reading data of all the customers served by the feeder one attempts to balance.

2. Check whether the condition of unbalance repeats itself on a daily basis.3. Obtain the coincident load of each distribution transformer by summing the interval meter readings of

all the customers served by the transformer.4. Obtain the coincident loads of each phase of the 3-phase feeder.5. For each interval, determine the main contributors (distribution transformer loads) to the unbalance.6. Transfer some of the distribution transformer to another phase to achieve some improvement.7. For a period of 24 hours check if this load transfer can sustain a certain degree of minimum unbalance.8. Several iterations may be required to accomplish this task.9. Once a degree of balance is achieved, continue to take synchronized time stamped interval meter

reading to verify the load balancing implementation that was based on historical interval metering data. It is not possible to achieve complete balanced loading at all times. However this proposed method will have some

merit by allowing one to come close to being able to proceed more accurately in the attempt to improve the balancing of loads.

Integrated Voltage and VAR control.

24

One of the useful applications for optimizing distribution feeder operation is the Integrated Voltage and VAR control. Capacitor banks and voltage regulators operate in synergistic fashion to maintain a good voltage profile along the feeder and simultaneously control the line losses in an optimal fashion. The basic function of the capacitor is to relieve the source from having to supply the load reactive power. If the capacitor is placed at the correct strategic location on the feeder and switched in at the appropriate time, then the substation current flowing into the feeder is reduced because the capacitors provide the locally required inductive currents of the load. The voltage drop at the feeder and the conductor losses are reduced. Many times, several capacitor banks strategically positioned and switched in and out at the appropriate times are required for the feeder.

One of the difficulties is to locate the optimal placement, to determine the right switching schedules and to correctly size the capacitor banks. The main problem is the lack of accurate information on the load and voltage distribution along the feeder as a function of time and location. Switching in and switching out of capacitor banks are based on the estimation of anticipated line loading and voltage profiles from past studies, weather data, etc.

Synchronized interval meter reading of all the loads on a feeder and coupled with voltage reading along the feeder can provide a correlation between loading profile and voltage profile along a feeder in real time . This type of information will help to design better switching schedules of capacitor banks and voltage regulator setting adjustments to optimize feeder wire losses and feeder voltage profile.

Once a switching schedule is implemented, continuous synchronized meter reading and voltage data can be used to verify how well the switching schedule is designed.

Power Quality MonitoringRecent advances in communication, digital electronics and computer technology have made remote monitoring of the

electric utility distribution network possible. Numerous utilities’ applications, such has automatic meter reading, demand response projects, some distribution control functions are already implemented at many utilities and more added value functions are being looked into or have already reached a state of a pilot project. One of the most recent issues promulgated by the electric power industry is the issue of power quality monitoring. Power Quality is probably one of the very important issues that fall under the broad category of power delivery reliability.

The nature of power quality problems can be defined in terms of its detrimental impact on the customers’ appliances and devices and also on the supply network and the energy metering devices.

In this manuscript, the discussion concentrates on the ability to locate power quality problems that are affecting customers and revenue metering as well as the ability to locate the culprits of harmonic pollution, which may or may not affect a broader segment of the network.

Without a good and reliable communication network, it is practically impossible to monitor each customer site for problems related to power quality because of the sheer large number of sites to monitor in a timely fashion. As an example, the very stringent requirements imposed on the communication system for automatic meter reading opens the door to other applications. Especially, the recent trend to require synchronized time stamped interval revenue meter reading renders the system to be more powerful for time correlation of events or conditions occurring in the power delivery network. Devices capable to monitor power quality problems, installed at strategic locations throughout the network and linked to the communication infrastructure will provide invaluable amount information that can be used to remedy situations that affect the quality of the power delivery. Voltage swells and sags data at many sites, not merely evaluated statistically on an individual site basis, but also correlated by time for all the sites being monitored, will be more valuable to the electric utility for setting up strategies to minimize their detrimental effects.

Other types of power quality problems are related to distortions of the 60 Hz voltage and current wave form. Numerous papers have been written on the effects of harmonics on losses in the electric power network and major appliances. Many sensitive digital-electronic devices are also affected by distorted voltages. Digital clocks, relying on the 60 Hz voltage zero-crossings for their proper operation, are also affected by recurring transient spikes on the voltage waveform. Other malfunctions are caused by memory contamination of these devices due to burst transient phenomena.

The degree of severity of the voltage distortions can be such that many customers’ devices are affected simultaneously due to a main culprit. Monitoring the problems at the point of common coupling quite often produces results, which point to the source of the problem. A more commonly occurring power quality problem, caused by locally generated harmonics and only affecting devices in its immediate vicinity, are harder to locate.

Revenue meters are also prone to be affected by locally generated distorted voltages and currents. It is hard to quantify the cumulative loss of revenues due to harmonic pollution. This paper is an attempt to explain and discuss the main issues related to the detrimental effects of severely distorted voltages and currents on revenue meters and how the existing communication system already used for remote metering can be used to quickly determine the sites affected by harmonic pollution.

25

In addition, there is the question what metric can be used to quantify distortion and what data should a power quality monitoring device generate, time stamp and made available for data gathering using the communication network.

When power quality problems arise at the metering sites, the meter reading accuracy may be compromised. Loads can cause fluctuating voltages at the point of connection to the network that will be different than the source voltage. This voltage can also be distorted by some types of loads at the point of connection. The energy consumption is generally also monitored at the same point of connection.

One of the categories pertaining to voltages under the “power quality” issues is voltage stability. Large steady state loads causing network overloading and subjected to frequent switching in and out during short intervals, loss of voltage due to recloser operations, etc. cause voltage dips or swells that may have undesirable effects to the electric users. For the causes of voltage problems mentioned above, one assumes that the voltage at the point of use maintains its sinusoidal wave shape and frequency except during a short period after a switching action. Present standards and recommendations already address these situations.

Switching generates transient voltages that may cause problems to equipments and devices. Recurrent transients, burst transients caused by power electronic switching devices, arc furnaces, arc welders, etc. occur quite often in the electric network and cause voltage and current distortions. Numerous studies about the hazards and detrimental effects of harmonics on the electric network, equipments, etc. have been published. Another source of power quality problems are high impedance faults, surface discharges on polluted insulators, bad contacts, etc.

Besides affecting appliances and metering devices, there is also the possibility that the quality of power can cause communications problems. Power line communication technologies are especially vulnerable. Communication signals are contaminated by burst harmonics. Recurring transients sometimes cause damage to memory or the electronic hardware of the communication devices. Much effort and money have been spent to make the communication system reliable and less sensitive to power quality problems. Severe communication impairments can be used to alert that power quality problems occur on the communication channel being used. New generation electric meters - digital electronic based meters – are purported to have the ability to measure KVAHr, KWHr, Power Factor, Demand, etc. This leads to the possibility to more comprehensive and novel billing strategies. The existing standards require that these meters guarantee a certain degree of accuracy when calibrated with clean sinusoidal voltage and current waveforms. Intuitively it is clear that a “clean” alternating current electric power source comprises a sinusoidal voltage that maintains a constant amplitude and frequency under various loading conditions. However, this assertion deviates from reality since the source and the power delivery network have impedances, which may produce phase differences and decreased voltage amplitudes at different locations on the network.

Studies are available on the effects of distorted voltage and currents on electric revenue meters and KVAHr metering. However, the lack of a good definition of what distortion means can become a problem. An industry accepted form of standardized distortion recognized by the various testing laboratories has yet to be defined.

In this paper, an attempt is presented to clarify and describe the main issues and to recommend certain steps needed to consolidate future efforts in order to arrive at certain testing standards that are logical and that it is possible to implement.

DESCRIPTION OF SOME TYPES OF WAVEFORM DISTORTIONS

We will attempt to describe some of the most commonly occurring waveform distortions.Transients: A transient is a sub-cycle over-voltage wave in an electric circuit generating a disturbance of the power voltage waveform See Fig. 14 (a).Impulsive: The duration can range between less than 50 nanoseconds to over 1 millisecond and can reach magnitudes of several per unit depending on the disturbing source and on the voltage withstand characteristics of the system.Oscillatory: The duration is between 0.3 and 50 milliseconds in a 60 Hz system. Its frequency ranges between less than 5 kHz to 5 MHz and the magnitude can reach up to 8.0 per unit.Short duration variations (Sag and Swell): Sag is a momentary under-voltage at fundamental frequency lasting from half a cycle to one minute, as seen in Fig 14 (b). Swell is a momentary over-voltage at fundamental frequency of similar duration to a sag. See Fig. 14(c).Instantaneous swell or sag : The duration is between 0.5 and 30 cycles with magnitudes ranging between 0.1 and 1.8 per unit.Momentary swell or sag: The duration is between 30 cycles and 3.0 seconds with magnitudes ranging between 0.1 to 1.4 per unit.Temporary swell or sag: The duration is between 3.0 seconds and 1.0 minutes with magnitudes ranging between 0.1 to 1.2 per unit.Long duration variations:Overvoltage : Duration is longer than 1.0 minute with magnitudes ranging between 1.0 and 1.2 minutes.

26

Undervoltage : Duration is longer than 1.0 minute with magnitudes ranging between 0.80 to 0.9 per unit, as illustrated in Fig. 14(d).Interruptions : Interruption is the complete loss of voltage for a period of time.Instantaneous : The duration is between 0.5 and 30 cycles.Momentary : The duration is between 30 cycles and 3.0 seconds. See Fig. 14(e)Temporary : The duration is between 3.0 seconds and 1.0 minute.Long term : The duration is longer than 1.0 minute. See Fig. 14(f)Distortion: A waveform distortion is any deviation from the nominal sine wave of the AC line voltage or current. The spectrum ranges between 0.0 Hz to 100th harmonic and the harmonic strength ranges between 0.0% and 20.0% for the voltage and for the current the strength ranges between 0.0% and 100.0%. An example of a severe voltage distortion produced by a non-linear load with high 5th harmonic current is shown in Fig. 14(g).Flicker (Voltage fluctuations): Flicker is a variation of the input voltage as illustrated in Fig. 14(h), sufficient in duration to

allow visual observation of a change in electric light source intensity. It is intermittent and its magnitude ranges between 0.1% and 7.0 %.

Noise: Electric noise is unwanted electric signal (see Fig. 14i), which may produce undesirable effects in circuits of control systems in which they occur. They are random and can be very short and intermittent or continuous, with magnitudes reaching up to 1% per unit of the fundamental voltage.

These definitions are nice and well. The question is how to deal with them. A fundamental need is to understand the sources of the distortions, to characterize them and to decide how to use the information. In order to identify these distortions, a method for processing the distorted waveform is needed by a remote device. The result of the processing is the piece of information that will be communicated to the central computer. Would the Total Harmonic Distortion (THD) meet the criterion?

There are cases where a large non-linear load is served by a sinusoidal voltage source. The distorted current creates non-sinusoidal voltage drops. Adjacent neighboring electric customers suffer from the distorted voltage. Sometimes the non-linear load is small enough without causing spillover to adjacent customers and yet is severe enough to cause local problems. This last case is the most difficult to detect.

When the load varies with time, the distortion is not constant with time. What one measures today may not be a reflection of tomorrow’s measurements.

Fig. 14 Some types of common waveform distortions

Many of the published results of investigations on the effects of voltage distortions on electric metering devices have been primarily concerned with the difficulties in defining and measuring an energy entity “KVA” under harmonic distortion conditions. Distinctions have to be made between distorted voltages serving linear loads and voltage distortion caused by non-linear loads. Issues of unbalanced loading on 3-phase systems are also problematic.

All the distortions used to check metering calibrations are due to steady state harmonics. Measurement results indicate that under conditions of distortion due to steady state harmonics, KWHr measurements do not cause appreciable loss of metering accuracy. The problem is with KVAHr measurements, because it is not clear how one should define KVA under distorted conditions of the voltage and current waveforms.

27

One major aspect that has been absent in all previous investigations is the case when the distortion contains non-integer harmonics. The complexity increases dramatically when the relationship between the source voltage and the load current is non-linear. A non steady state direct current component in the load current adds to the complexity of the problem.

Several important questions can be raised in this case: What are the effects on electromechanical meters that use the Ferraris disk? What are the effects on electronic meters? Which components or parts of the meter are involved in metering inaccuracies? Do concepts like KVA, Power Factor and Reactive Power still make sense if these items are used for electric energy metering and billing? Does filtering affect actual energy consumption? Can the distortion be categorized in several classes that can be used for calibration purposes? What types of remedies can be applied to the sources of distortion?

Causes of Inaccuracies :To appreciate the problem that we mentioned above, let us take a fairly simple example for illustration. A simple

induction disk electric meter operates on the following principle. An alternating current passing through the current coil of the meter generates an alternating magnetic field that crosses the rotating disk perpendicularly (see Fig. 15) The induced electromotive force (emf) in the metallic disk causes an alternating current to flow in the disk, which reacts with the magnetic flux generated by the voltage coil, Fig. 15 (a) and Fig. 15 (b). The sum of all the elemental Lorentz forces generates a net mechanical torque, which rotates the disk.

a)Retardation Torques. If in addition to the alternating current a direct current is also present in the current coil, a DC flux is also generated and superimposed on the ac magnetic flux. The disk rotation generates an emf due to motion in the disk. This emf causes current flow in the disk and causes retardation torques.

b)Harmonic Torques. Additional torques can also be produced by steady state harmonic currents from non-linear loads at the customer side and these can be in the same or in opposite direction of fundamental frequency torque, depending on whether the harmonic currents are positive or negative sequence. Slowing down the disk rotation essentially means a loss of revenue. Likewise, continuing positive sequence currents can impose accelerating torques on the disk that may affect customer’s metering. These findings have been identified by several investigators and reported in the literature. Moreover, studies conducted by the Louisiana State University indicated that loss of accuracy by as much as 8% is possible when both voltage and currents contain harmonics.

c) Transducers. Electronic meters are coupled to the power line through transducers. The most common types of coupling devices are the Potential and Current Transformers. These transformers have certain degree of accuracy guarantee when subjected to 60/50 Hz voltages and currents. When subjected to distorted voltages or currents, it is not clear whether the voltage and current ratios and inherent transformer phase shifts are still within the certified range of accuracy for pure sinusoidal conditions. A DC component in the voltage or current may cause core saturation. High frequency harmonics are transferred through capacitive coupling. Skin effect increases the winding impedances. High voltage spikes may affect the contents of memory devices or display used in the electronic meter.

Figure 15. Induction disk rotation

Examples of extreme waveform distortion:

28

Some examples of extreme distortion on voltage and current waveforms are shown in Fig. 16 through 19. The voltage of Fig. 16 was taken at the service box of a three-phase supply serving a pump motor with a variable speed drive. The line-to-line voltage was 480 V and measurements at the medium voltage level did not indicate any cause for concern of damaging spillover to other customers. Fig. 17 depicts the unfiltered current to a 6 phase power rectifier serving a 100 kW TV transmitter load. The smoothing filter capacitors were never installed for some unknown reason.

Fig. 16. Unfiltered current to a Fig. 17. Significant third harmonic 6-phase power rectifier. distortion in a current waveform

In Fig. 17 a load current rich in 3rd harmonic from a small industrial plant is shown. The first two pictures do not exhibit steady state conditions and are difficult to characterize in terms of Fourier series,

which in essence represent a distorted periodic waveform as the superposition of a series of higher-order frequency components (harmonics) and a fundamental sinusoidal component. The last two pictures are more amenable to this kind of analytical representation.

Standards Issues

The existing standard on power quality in the USA [8] related to metering accuracy has been discussed by many professionals [9]. The concerns are primarily based on distortions due to integer harmonics of the 60 Hz waveform. However numerous field investigations indicated that many distorted waveforms are rich in non-integer sub-harmonics and behave like steady waves for certain intervals of time. A few of such waveforms were shown in figures 3 to 6.

There are suggestions on introducing new rate structures based on KVA rather than on KW. These suggest penalizing the customer for bad power factor. For a steady state sinusoidal voltage and current, KVA and KW are coupled by the concept of power factor. The problem becomes extremely complex when both voltage and current are distorted. Under non-sinusoidal conditions and 3-phase unbalanced conditions, engineers, scientist involved with the study of power entities have yet to agree what KVA means and whether there is even a correlation between real power and KVA.

At this moment it is not clear how much revenue is lost due to inaccurate meter registration due to severe waveform distortion within an electric utility. Locations or institutions performing calibration of electric energy meters under distorted voltage and current conditions are scarce and are not generally available or known. It is also not clear what standard distortion means and what a good measure of accuracy is.

The time has probably arrived to develop some generally accepted standards of distorted waveforms. For example, in the area of transient voltage testing, standard test waveforms such as the 1.2x50 sec voltage impulse wave and the 10x20 sec current impulse wave for testing dielectric integrity of electric devices were adopted long time ago. And yet, these test waveforms are not necessarily replicas of transient waveforms measured in the electric networks. However it has been proven repeatedly, that withstand levels to transients as defined by these tests using the indicated wave forms, provide a guaranteed level of survival against normally occurring damaging transients.

The same efforts could be applied to arrive at some standard distortion of both voltage and current waveforms. The voltage and current waveforms are not independent of each other and using modern day data acquisition systems, gathering these types of data is not difficult. References [10,11,12] described some of the techniques how to replicate distorted waveforms in a testing laboratory for meter calibrations.

29

With a communication infrastructure already in place, the question arises on what is the nature of information needed to bring back to the dispatch center computer. One possible method, pending the development of new meaningful metrics to measure power quality, is the use of a contiguous sequence of short time monitoring intervals DT. For each interval of DT the total harmonic distortion (THD) can be computed. For a contiguous number N of such intervals the maximum THDmax , the minimum THDmin and the average THDavg can be computed. This method allows one to identify the degree of bursts occurring per unit time. If THDavg is very close in magnitude to THDmax and THDmin then this simply means that there are no minima or maxima. If THDavg is close to the magnitude of THDmax then this simply means that there aremore maxima then minima.

The interval (N*DT) is time stamped and the THD data is retrieved from the remote monitoring site using the communication system. DT can be as short as one or a few cycles of the 60 Hz wave and (N*DT) can be as long as 15 or 30 minutes. To reduce the amount of communication bandwidth utilization, a number of consecutive intervals (N*DT) can be stored in memory at the monitoring device and retrieval of the stored data is done on a daily schedule.

To take advantage of the communication network it is also possible to simplify the time correlation of the collected data by service territory. The following method can be used. All power quality monitoring devices connected to a distribution feeder or phase of a feeder can be assigned into a functional group address, which can be communicated to by a single group command. The functional group data does not only provide time correlation, but also correlation by location served by the same supply or source.

The same technique can be applied to the monitored voltage. By time correlating the time stamped collected data for a certain service territory, feeder or network section served by a distribution bus, one can decide whether the problem is local or widespread. The magnitudes of the monitored data also provide an indication where the culprit is of the harmonic pollution.

Power line types of communication technologies are affected by severe power quality problems. Unusual degradation of the communication performance on a physical channel will serve as an alert to power quality degradation.

Strategically placed advanced electronic meters capable of monitoring THD and voltages can be used to collect useful information. Also power line communication impairments could be used as an alarm to pinpoint distribution network areas where power quality problems occur.

Another useful power quality monitoring capability is to monitor voltage swell and sag in the distribution network. It will be even more useful if the cause and the culprit can be found and identified. By monitoring the loads and the voltages in real time, similar to what has been described in the section on Integrated Voltage and VAR Control, it is possible to identify the culprit of the voltage sag.

Permanently installed monitoring devices, which measure Total Harmonic Distortion of the voltages along the line, can patrol the line for harmonic polluters.

Fig. 18 Fig. 19Some really bad looking voltages are shown in Fig. 18 and Fig. 19. They may affect power line carrier communications

and may also cause detrimental effects on the accuracy of meter reading devices, increase losses in the network and electric appliances. Digital electronic devices, if not properly filtering the distortions on the input voltages can also adversely affect their operations.

Distributed GenerationLately there is a lot of discussions in the electric industry about distributed generation. These generators vary in sizes

ranging between tens of KW up to several MW. The smaller ones are most likely connected to the network at the low voltage side. The larger sizes are coupled at some point to the medium voltage circuit. The prime movers of these generators can be

30

wind turbines, small low-head hydro turbines, gas engines, etc. The electric generators can be induction generator type machines, synchronous generators.

Many of the issues of concern are power quality concerns, islanding issues, safety requirements, etc. The coupling to the utility network can be using direct AC coupling, or through power electronic AC-DC-AC assemblies. This last method, if no proper filtering is applied, will generate harmonics.

Islanding means that part of the distribution network served by the distribution substation bus is disconnected and its load is picked up by the co-generator. This load could exceed the capacity of the co-generator and to make matters worse, if there is another co-generator at the same part of the disconnected network, phenomenon similar to loss of stability due to loss of synchronization between machines will appear. This phenomenon exhibits itself as swings in the AC frequency and voltage and can be very damaging to appliances. Restoration issues after a fault isolation have to be carefully studied.

How the communication system for AMR and other added value applications can be used to maintain the integrity of the electric distribution network and to provide other supporting functions with the presence of distributed generation connected to the network should be investigated thoroughly.

Hybrid Electric VehiclesIn the quest for clean air in large metropolitan areas, hybrid electric vehicles are purported to reduce exhausts that will

reduce air pollution in the cities. The vehicle battery charging devices will be connected to the customer service voltage network. Several new things require serious considerations.

1. The creation of new peak demands.2. Separate metering may be required if electric energy pricing for electric vehicles is allowed to compete against

gasoline or diesel car fuel prices.3. Introduction of local harmonic distortions.4. Protection against damaged chargers and batteries5. Etc.

Supporting Functions

Communication Network Monitoring and Control. For power line communication technologies, the communication network is also the electric distribution network. Hence, the

Outage Management function is also the communication network monitoring function. For other communication technologies, the communication infrastructure overlays the power delivery infrastructure and

the communication network usually have no resemblance to the power delivery network at all. For this case, a different type of communication network outage management is needed. Also, the possibility exists that an outage at the distribution network may or may not affect the communication network. If the main nodes, sub-nodes or data concentrators are also de-energized because of an outage at the distribution network, then it is necessary to couple the electric network data base to the communication network database. The ability to reroute the communication using other paths has to be implemented if a de-energized node is part of the normal communication path.

AM/FM systemsMany utilities have converted their one-line electric network drawings into digital form. Much of that information does

not have geographical information that an AM/FM (Area Map and Facilities Management) can provide. An Outage Management System, coupled with an AM/FM system can improve maintenance and repair services by reducing the time for patrolling the line and by quickly identifying which protective device has operated. The information generated also can be used to determine which communication nodes are de-energized.

Communication Network Monitoring and Control and Outage Management are not only useful functions, but they are necessary to provide service continuity and reliability of services. When those functions are properly designed and used, unnecessary operations and delays can be avoided.

Data WarehousingCollected data which are stored are useful when they can be cross-correlated with time and physical locations. Energy

metering interval data which are time synchronized can be used to determine coincident demand of the total system. If other parameters are tagged to the metering data, such as the phase, feeder number, the substation bus, one can use this information to implement load balancing. If the location of the loads and the voltage on the feeder are added to the collected data, the information can be used for to improve control of the voltage profile on the feeder through optimal scheduling of the capacitor banks switching.

31

The data collected from the network, such as KWHr, KVAHr, Voltage, etc. if they are time stamped and synchronous and also if they also can be spatially correlated to the AM-FM system, they will open the door to numerous applications that will benefit the electric utility company.

5. CONCLUSIONS

The decision which communication technology to choose for implementing advanced Smart Grid applications should not be based on monthly billing capability, high baud rate, etc. We have seen that words such as “high baud rate”, “retries”, “monitor outages”, “100% communication performance” can be misleading unless they can be tied to infrastructure design, system operation, data throughput per unit time, feasibility to implement advanced utility applications, database design, etc. With the introduction of Smart Meters much more different types of data can be obtained which can be used to develop control algorithms for the various future applications. The essential building blocks, which constitute the system architecture have been described and discussed briefly. It is important to plan ahead and obtain a full understanding what is required to implement future applications to support the Smart Grid.

An attempt is made in this manuscript to show the intimate relationship amongst the various applications and the synergism, which exist amongst the various applications and communication functions in order to make a system operate in an optimal fashion. It is also necessary to implement the Outage Management function to the power delivery network and how to minimize the effects on the communication network. Provisions to implement the links between the communication network control computer and the energy delivery network database are necessary. There are physical links amongst paths and nodes between both networks that requires constant monitoring.

In addition, to support the Maintenance and Repair department for speedy locating, repair and restoration of segments of the networks, availability of information of the topological and geographical locations of the protective devices and major nodes will greatly enhance the system operation.

6. REFERENCES

[1] S. Mak and D. Radford, Design Considerations for Implementation of Large Scale Automatic Meter Reading Systems, IEEE Transactions on Power Delivery, January 1995, Vol. 10, No. 1

[2] S. T. Mak and D. Radford, Added Utilization Costs Associated with Different Communication Architectures for Distribution Automation and Demand Side Management , IEEE Transactions on Power Delivery, January 1995, Vol.10, No. 1

[3] S. T. Mak and D. Radford, Communication System Requirements for Implementation of Large Scale Demand Side Management and Distribution Automation , IEEE Transactions on Power Delivery, April 1996, Vol. 11, No. 2

[4] J. Newbury, Communications Field Trials for Total Utility Metering, IEEE Transactions on Power Delivery, April 1996, Vol. 11, No. 2.

[5] W. M. Lin, M. T. Tsay and S. W. Wu, Application of Geographic Information System to Distribution Information Support, IEEE Transactions on Power Systems, February 1996, Vol. 11, No. 1.

[6] EPRI, Proceedings: International Load Management Conference, EPRI EM-4643 Project 1940-15 Proceedings, June 1986

[7] James R. McDonald, Dr. Graeme M. Burt, Stephen D. J. McArthur, Stewart C. Bell, Ian M.Elders, CEPE, University of Strathclyde, Proposed Architecture for Integrated Decision Support Functions Within Second Generation Distribution Management Systems , DA/DSM ASIA 1995, Singapore, Conference Papers.

[8] Terry Devaney, KEMA-ECC, Successful Communication Techniques for Distribution Systems , DA/DSM ASIA 1995, Singapore, Conference Papers.

[9] Dan Ehrenreich, Shlomo Liberman, Motorola Communications Israel, Ltd., Cost Benefits Resulting from Use of Integrated Communication for Distribution Automatio n , DA/DSM ASIA 1995, Singapore, Conference Papers.

[10] Sioe T. Mak, Thomas N. Hilleary, Distribution Control Systems Inc., A Power Frequency Communication System Designed for Utility Application, DA/DSM ASIA, Singapore, Conference Papers.

[11] IEEE PRESS, Load Management Book, 1986

32

[12] Sioe T. Mak, Synergism Between Intelligent Devices and Communication Syst ems for Outage Mapping in Distribution Networks, CIRED, 15th International Conference on Electricity Distribution, Nice, France, 1-4 June, 1999.

[13] Sioe T. Mak, Power Delivery Infrastructure Differences and Their Impacts on Different Types of Power Line Communications for Automatic Meter Reading, CIRED, 16th International Conference & Exhibition on Electricity Distribution, AMSTERDAM, The NETHERLANDS, 2001.[14] Sioe T. Mak, The Random Failure Statistics and Replacement Schedules of Remote Devices for Load Management, IEEE PES Transaction Paper 86 SM 313-1, IEEE/PES , 1986 Summer Meeting, Mexico

33