Repositorio Institucional de Documentos

SMART GRIDS SMART METERS AND NON INTRUSIVE LOAD MONITORING FINAL PROJECT REPORT Date: 28-06-2013 Author: Pablo Rodes Sanagustín SMART GRIDS: Smart Meters and Non Intrusive Load Monitoring AUTHOR OF THE PROJECT: Name: Pablo Rodes Sanagustín E-mail: [email protected] Telephone number: 0644229276 SUPERVISORS OF THE PROJECT: 1) Name: Joost P. Rey Email: [email protected] 2)Name: Sytze Verbeek Email: s.verbee[email protected]l PARTNER UNIVERSITIES: 1) NHL HOGESCHOOL Rengerlaan 10 8917 DD Leeuwarden Tel: 058 251 2345 Fax: 058 251 1950 Email: info[email protected] 2) ESCUELA DE INGENIERÍA Y ARQUITECTURA. UNIVERSIDAD DE ZARAGOZA Calle María de Luna 3 50018 Zaragoza Tel : (+34) 976 76 18 68 Fax:(+34) 976 762 031 Email: [email protected] ABSTRACT The aim of this project is to explain the general concepts of Smart Grids,as well as the expected changes in the power supply and the main technologies to support the development of the same. A Smart Grid is a system that allows two-way communication between consumers and utilities, in a way that the information provided by the consumer can be used by electricity companies for a more efficient operation of the electric network, as well as offer new services to customers. The development of the Smart Grids is essential if the global community wants to achieve common goals of energy security, economic development and climate change mitigation. To do so, they are developing and implementing new technologies such as Smart Meters and new techniques for measurement of power consumption as Non Intrusive Load Monitoring. The Smart Meters are electricity, gas, or water meters that automatically collect measurement data and send them to the electricity companies to allow these to have a better view of the electrical distribution and provide their customers a greater understanding of their own consumption. Non Intrusive Load Monitoring is a technique which detects the events of electrical appliances by analyzing total load demand. This is possible since devices are special features in the moments of connection and disconnection in changes both positive and negative in active and reactive power. As these characteristics are unique for each device, it is possible to recognize the profile of each one of them and know which devices are turning on or off, as well as the consumption of each one of them. This is what offers the Plugwise technology, that through the use of their devices allows us to monitor and control the power consumption of a home, office or company and be able to see the results in our own Smartphone or PC. The use of Plugwise technology in combination with a Smart Meter allows both customers and electricity companies be aware of how much, how and where electricity is consumed. RESUMEN El objetivo de este proyecto consiste en sintetizar los conceptos generales de las redes inteligentes (Smart Grids), los cambios que se prevén en la red eléctrica y las principales tecnologías que apoyaran el desarrollo de las mismas. Una Smart Grid es una sistema que permite la comunicación bidireccional entre el consumidor final y las compañías eléctricas, de forma que la información proporcionada por los consumidores pueda ser utilizada por las compañías eléctricas para permitir una operación mas eficiente de las red eléctrica, así como ofrecer nuevos servicios a los clientes. El desarrollo de las Smart Grids es esencial si la comunidad global quiere alcanzar objetivos comunes de seguridad energética, desarrollo económico y mitigación del cambio climático. Para ello, se están desarrollando e implementando nuevas tecnologías como los medidores inteligentes (Smart Meters) y nuevas técnicas de medida de consumo eléctrico como la monitorización no intrusiva (Non Intrusive Load Monitoring). Los Smart Meters son medidores de electricidad, agua o gas que recopilan de forma automática los datos de medida y los envían a las compañías eléctricas permitiendo a estas tener una mejor visión de la distribución eléctrica y proporcionan a sus clientes un mayor conocimiento de su propio consumo. La monitorización no intrusiva es una técnica que detecta los eventos de aparatos eléctricos analizando la demanda total de la carga. Esto es posible debido a que los aparatos presentan características especiales en los momentos de conexión y desconexión consistentes en cambios tanto positivos como negativos en las potencias activa y reactiva. Como dichas características son únicas en cada dispositivo, es posible reconocer el perfil de cada uno de ellos pudiendo saber que dispositivos se están encendiendo o apagando, así como el consumo eléctrico de cada uno de ellos. Esto es lo que ofrece la tecnología Plugwise, que mediante el uso de sus dispositivos permite monitorizar y controlar el consumo eléctrico de una vivienda, oficina o empresa y poder ver los resultados en nuestro propio Smartphone o PC. El uso de tecnología Plugwise en combinación con un Smart Meter permite que tanto clientes como compañías eléctricas sean conscientes de cuanto, como y donde se consume la electricidad. ACKNOWLEDGEMENTS 'Let us be grateful to people who make us happy, they are the charming gardeners who make our soul flower.' Marcel Proust The culmination of this work represents the end of a big stage of my life, which has been marked by traces of all the people who I've been finding along the way. So I want to thank all them its support, compression, effort and patience. My teachers at the University of Zaragoza for teaching me everything I know. Joost Rey and Sytze Verbeek for allowing me to work with them and being so patient with me. Jacob Hut and Hendrik Bijlsma for his help on the technical part of the project. Marga Zeilstra and Moniek Dijkema from the International Office for their assistance in the first months. To my childhood friends to always be there and be like my second family. To my friends from the University by the good times shared. My friends from the Erasmus for sharing with me a few memorable months. To my family for always being there, supporting me and helping me. You are the greatest. THANK YOU Pablo Rodes AGRADECIMIENTOS ‘Seamos agradecidos con las personas que nos hacen felices, ellos son los encantadores jardineros que hacen florecer nuestra alma.’ Marcel Proust La culminación de este trabajo representa el final de una gran etapa de mi vida, la cual ha quedado marcada por las huellas de todas las personas que me he ido encontrando a lo largo del camino. Por eso quiero agradecerles a todos ellos su apoyo, compresión, esfuerzo y paciencia. A mis profesores de la Universidad de Zaragoza por enseñarme todo lo que sé. A Joost rey y Sytze Verbeek por permitirme trabajar con ellos y ser tan pacientes conmigo. A Jacob Hut y Hendrik Bijlsma por su ayuda en la parte técnica del proyecto. A Marga Zeilstra y Moniek Dijkema de la Oficina Internacional por su ayuda en los primeros meses. A mis amigos de la infancia por estar siempre allí y ser como mi segunda familia. A mis amigos de la Universidad por los buenos momentos compartidos. A mis amigos del Erasmus por compartir conmigo unos meses inolvidables. A mi familia por estar siempre allí, apoyarme y ayudarme. Sois los más grandes. GRACIAS Pablo Rodes Contents 1. INTRODUCTION......................................................................................................................... 1 2. SMART GRIDS ............................................................................................................................ 2 2.1. Purpose and Reach ............................................................................................................ 2 2.2. Electrical network. Toward the Smart Grids. ..................................................................... 2 2.3. Need of Smart Grids ........................................................................................................... 5 2.4. Security problems in the Smart Grids ................................................................................ 6 2.5. Key findings ........................................................................................................................ 6 2.6.In summary.......................................................................................................................... 8 3. SMART METERS ........................................................................................................................ 9 3.1. Classification ....................................................................................................................... 9 3.2. Definition .......................................................................................................................... 11 3.3. Scientific study ................................................................................................................. 11 3.4. Smart Meters P-Ports ....................................................................................................... 12 3.5. Energy market .................................................................................................................. 13 3.6. Smart metering landscape in Europe ............................................................................... 14 3.7. Smart metering in Netherlands ........................................................................................ 16 3.8. Smart Meter Security ....................................................................................................... 18 3.9. Advantages and disadvantages ........................................................................................ 18 3.10. Opposition and concern ................................................................................................. 19 4. NON INTRUSIVE LOAD MONITORING .................................................................................... 20 4.1. Introduction ..................................................................................................................... 20 4.2. Background....................................................................................................................... 21 4.3. Applications ...................................................................................................................... 22 4.4. NILM Basics ...................................................................................................................... 23 4.5. Data acquisition module .................................................................................................. 26 4.6. Feature extraction ............................................................................................................ 27 4.7. Learning and inference in NILM systems ......................................................................... 29 4.7.1. Supervised learning approaches................................................................................ 30 4.7.2. Unsupervised learning approaches ........................................................................... 30 4.8. Non Intrusive Load Shedding Verification (NILSV) ........................................................... 30 4.9. NILM: Advantages and disadvantages ............................................................................. 31 5. PLUGWISE ............................................................................................................................... 33 5.1. What is Plugwise? ............................................................................................................ 33 5.2. The benefits of Plugwise .................................................................................................. 34 5.3. Products ........................................................................................................................... 34 6. EQUIPMENT USED .................................................................................................................. 35 6.1. Plugwise Home Basic ........................................................................................................ 35 6.1.1. Circle .......................................................................................................................... 35 6.1.2. Circle+: ....................................................................................................................... 35 6.1.3. Stick: .......................................................................................................................... 35 6.1.4. Source: ....................................................................................................................... 35 6.1.5. ZigBee ........................................................................................................................ 36 6.2. Kamstrup Smart Meter ..................................................................................................... 37 6.2.1. Overview .................................................................................................................... 37 6.2.2. Construction .............................................................................................................. 38 7. HOW TO CONFIGURE IT / HOW IT WORKS ............................................................................ 39 7.1. Home Basic ....................................................................................................................... 39 7.2. Kamstrup 382 ................................................................................................................... 39 7.2.1. Start-Up the meter .................................................................................................... 39 7.2.2. Registers .................................................................................................................... 39 7.2.3. Display ....................................................................................................................... 40 7.2.4. Loggers ...................................................................................................................... 41 7.2.5. Energy measurement (Shunt measure method) ....................................................... 41 7.2.6. Permanent memory ................................................................................................... 41 7.2.7. Power calculation method ......................................................................................... 42 7.2.8. METERTOOL............................................................................................................... 42 7.2.9. Installation ................................................................................................................. 44 8. FINAL RESULTS ........................................................................................................................ 46 8.1. Home Basic ....................................................................................................................... 46 8.2. Kamstrup 382 ................................................................................................................... 52 9. CONCLUSIONS ......................................................................................................................... 53 10. REFERENCES .......................................................................................................................... 54 11. INTERESTING WEB PAGES .................................................................................................... 55 6 · Promote the active participation of consumers, encouraging the local power generation and delivery of excess energy to the network in peak hours. · Have ability to supply power quality appropriate to the Digital era, thanks to a greater number of points of generation that will allow the delivery of different qualities of energy for each type of application. · Accommodate a wide variety of forms of generation and storage, thanks to the micro and power generation distributed. · Facilitate the increase of markets, due to the inclusion of new elements in the network as the electric vehicle, a larger number of renewable energy, etc. · Make a most efficient optimization of their assets and operation, thanks to the automation of all the elements involved. As mentioned previously, one of the main motivations for the change in the energy model is the environmental aspect. In this new model sustainable development, renewable energies, are considered as inexhaustible, and the peculiarity of being clean energy, sources of energy with the following characteristics: assume a zero or low environmental impact, use has no added risks, indirectly represent an enrichment of natural resources and are an alternative to conventional energy sources, being able to replace them gradually. Taking into account these aspects among others, the European Commission met in 2008 to develop the plan known as Plan 20-20-20. Strategy 20-20 - 20 is an initiative launched to combat climate change with an aim to clear: reduce by 20% greenhouse gas emissions; Save 20% on consumption energy; and provide the energy system with at least one 20% renewable energy; all for 2020. 2.4. Security problems in the Smart Grids The security problems that have been detected at the beginning of 2009 in the deployment of Smart Grids in the United States have demonstrated the need of a new architecture of communications. The security that must be added to the IP networks is the most complex in the history of communications (firewalls, IDSs, spam, spoofing, Trojans, virus, phishing,...). The customer's data and privacy-related issues are in the today hot points of contention in the evolution of electric networks Smart. The issue of who has the data and why is a question that concerned legislators. There is growing concern that these data they are used in ways that customers would have never foreseen. The information from these devices can be combined in ways unexpected and revealing information that consumers do not want to know it. The subject of the problems of security in the Smart Grids will be treated more deeply in point 3.8 .Smart Meter Security 2.5. Key findings 1. The development of smart grids is essential if the global community is to achieve shared goals for energy security, economic development and climate change mitigation. Smart grids enable increased demand response and energy efficiency, integration of variable renewable 7 energy resources and electric vehicle recharging services, while reducing peak demand and stabilising the electricity system. 2. The physical and institutional complexity of electricity systems makes it unlikely that the market alone will implement smart grids on the scale that is needed. Governments, the prívate sector, and consumer and environmental advocacy groups must work together to define electricity system needs and determine smart grid solutions. 3. Rapid expansion of smart grids is hindered by a tendency on the part of governments to shy away from taking ownership of and responsibility for actively evolving or developing new electricity system regulations, policy and technology. These trends have led to a diffusion of roles and responsibilities among government and industry actors, and have reduced overall expenditure on technology development and demonstration, and policy development. The result has been slow progress on a number of regional smart grid pilot projects that are needed. 4. The “smartening” of grids is already happening; it is not a one-time event. However, largescale, system-wide demonstrations are urgently needed to determine solutions that can be deployed at full scale, integrating the full set of smart grid technologies with existing electricity infrastructure. 5. Large-scale pilot projects are urgently needed in all world regions to test various business models and then adapt them to the local circumstances. Countries and regions will use smart grids for different purposes; emerging economies may leapfrog directly to smart electricity infrastructure, while OECD countries* are already investing in incremental improvements to existing grids and small-scale pilot projects. 6. Current regulatory and market systems can hinder demonstration and deployment of smart grids. Regulatory and market models – such as those addressing system investment, prices and customer participation – must evolve as technologies offer new options over the course of long-term, incremental smart grid deployment. 7. Regulators and consumer advocates need to engage in system demonstration and deployment to ensure that customers benefit from smart grids. Building awareness and seeking consensus on the value of smart grids must be a priority, with energy utilities and regulators having a key role in justifying investments. 8. Greater international collaboration is needed to share experiences with pilot programmes, to leverage national investments in technology development, and to develop common smart grid technology standards that optimise and accelerate technology development and deployment while reducing costs for all stakeholders. 9. Peak demand will increase between 2010 and 2050 in all regions. Smart grids deployment could reduce projected peak demand increases by 13% to 24% over this frame for the four regions analysed in this roadmap. 10. Smart grids can provide significant benefits to developing countries. Capacity building, targeted analysis and roadmaps – created collaboratively with developed and developing countries – are required to determine specific needs and solutions in technology and regulation. 8 * The Organisation for Economic Co-operation and Development (OECD) is an international economic organisation of 34 countries founded in 1961 to stimulate economic progress and world trade. It is a forum of countries committed to democracy and the market economy, providing a platform to compare policy experiences, seek answers to common problems, identify good practices and co-ordinate domestic and international policies of its members. 2.6. In summary A Smart Grid is based on the use of sensors, communications, computing and control, form that is improved in all aspects the features of the power supply. A system becomes intelligent acquiring data, communicating, processing information and exercising control through feedback that allows to adjust to the variations that can be arise in a real operation. Smart grids can play an important role in addressing increasingly untenable economic, environmental, and social trends in the supply and use of energy. By enabling increased awareness of system operation and better informed participation by electricity users, smart grids will increase electricity end-use efficiency while optimising network asset utilisation and increasing grid resiliency. They will also enable efficient integration of variable renewables and electric vehicles, as well as new products and services. Smart grids co-ordinate the needs and capabilities of all generators, grid operators, end-users and electricity market stakeholders. This allows the grid system to operate as efficiently as possible, minimising costs and environmental impacts while maximising system reliability, resilience and stability. Smart grids accomplish this optimisation by using digital and other advanced technologies to monitor and manage the transport of electricity from all generation sources to meet the varying electricity demands of end users. These technologies are essential if the global community is to achieve shared goals for energy security, economic development and climate change mitigation. Fig. 3. Topography of a Smart Grid 9 3. SMART METERS 3.1. Classification A Smart Meter refers to an electricity, gas or water meter that passes its metering data automatically to the utility company on a regular basis. This provides utility companies with new means to monitor their distribution network and gives customers an opportunity to gain insight in their consumption. The Smart Meter offers advantages for both the consumer and the energy supplier. Consumers will receive better service and gain insight in their energy usage. Suppliers can introduce new services enabled by the Smart Meter like providing customers with advice about their usage and energy savings. Grid companies can get a better view in the actual consumer energy usage to improve demand and supply in the energy grid. The equipment for the measurement of the electrical energy consumed is an electrical counter or meter that consists of three main elements, such as the system of measurement, the element of memory and the device information. The equipment of measure of electrical energy can be classified according to their features: · Technological, it can be electronic or electromechanical counters. · Functional as single or three phase. · Energy counters as active and/or reactive counters. · Operative as devices of type recorder or programmable recorder that allow the remote control. The devices of type recorder may be of the two technologies: Electromechanical devices that measure only one type of energy, kWh accumulated or kVAh accumulated, do not possess tariff discrimination being the standard counters electromechanical induction. Electronic, Automatic Meter Reading (AMR), allow only measure energy accumulated, they record the measure of total energy or by monthly time intervals predefined. Provide basic bidirectional communication between the meter and the data server, allowing on the basis of this technology the time measurements of use, Time of Use (ToU). Thanks to replace electromechanical meters counters by solid state electronics counters, it is possible to obtain the information energy in digital form. With this step, it is possible to add capacity to communication to the device, allowing the interested use the AMR technology for remotely access to data through the communication layer. AMR is the technology of automatically collecting diagnostic, consumption, and status data from energy metering devices. Then, transferring these data to a central data base for billing, trouble shooting, and analyzing. Therefore, nearly all of this information is available in real time and on demand, allowing for improved system operations and customer power demand management. 10 An example of this, is the system of reading through driving, thanks to which the company sends a vehicle that is driven by a neighborhood getting very quick action of all the homes thanks to a system of communication wireless. Fig. 4. Meter reader example Programmable measuring equipment, electronic type are: Advanced Meter Infraestructure (AMI), they can be considered an extension of the AMR, these devices allow the reading of the energy consumption accumulated or the instantaneous power, they support different pricing options by type of measure and records demand, or intervals of "load" programming agreed with each client. AMI includes all the components that allow two-way communication between Smart Meters and energy systems to improve demand in reponse. AMI incorporates: *Smart Metering *Telecommunication system and radio system *Meter data management system In total, AMI helps utilities repond in real time to demand for electricity more quickly and efficiently. Smart Meters, these appliances provided through the center of management, information and control of parameters of quality and service programming along with the measurement of form software update telematics. The Smart Meters provide enhanced communication with the network manager and Home Area Network (HAN) with the local teams of consumption. Initially, the introduction of AMR systems and the elimination of manual reading, were carried out to reduce the cost of labor in the reading of the data energy. Currently, however, the industry has realized that the AMR systems allow companies to produce greater benefits and services, such as pricing in real time to promote energy efficiency, immediate detection of faults in the system and more advanced data and accurate user with those who form their consumption profile. 11 3.2. Definition The Smart Meter is basically part of AMI that includes at least the following supplements, energy through programmable PCS (Power Control Switch) to control the limit of consumption, a port HAN (Home Area Network) and services of pricing on request. The general structure of the counter maintains the three main elements such as measurement system, memory and device of main information, which until now were the only communications system. In order to expand its operational capacity, additional elements are added: · Power systems. · Spreadsheet processor. · Communications processor. · Drive / control device. It is expected that the Smart Meters, to comply with the following features: - Determines and stores in real – time or near real – time energy consumption. - Automatically precesses, transfers, management and utilization of metering data. - Automatic management of meters from the electricity companies. - Two – way communication. - Provides meaningful and timely consumption information to the relevant actors and their systems, including the energy consumer. - Supports services that improve the energy efficiency of energy consumption and energy system. 3.3. Scientific study A Smart Meter measurement system exists of an electricity meter and possibly water and gas meter coupled to each other. With the Smart Meter, we don’t have to send our readings to the electricity company and the electricity company doesn’t need to send a technicien once a year to read the meter. The Smart Meter will send the readings every 15 minutes to a Central Access Server (CAS) system. This way the electricity company can send a technicien once every 5 years and save a lot of money. Because the electricity company gets data every 15 minutes it can adjust the billing to our usage and so we don’t have to pay extra at the end of the year when we have used more energy. 12 Fig. 5. How Smart Meters work 3.4. Smart Meters P-Ports Although in our Project we only use the P1 Port, I’ll describe all the ports available in a Smart Meter. These ports can be physical connections or logical relations between the different components that exist in a smart metering environment: *Port P1: is used for communication between the customer’s metering system and one or more modules that can use the information from the meter system. Acess to this port is read only. *Port P2: communicates between the metering system and additional meter sensors, such as a gas or water meter. *Port P3: is used for communication between the metering system and the Central Acess Server (CAS) that collects metering information from the connected metering systems and can send control commands to the connected devices. *Port P4: is the port as the CAS which is located at the grid operator and will also be accesible to independent services providers and suppliers. Besides these four ports, some smart meters are equipped with two more ports: *Port P0: provides a connection between the meter and a external device that allows engineers to perform on-site maintenance of the meter. 13 *Port P5: is basically an extensión to port P4 that enables the costumer to get information about their energy usage from the supplier. Fig. 6. Smart Meter ports 3.5. Energy market A few of the goals of this new market model are to make it easier to switch to a new energy supplier, to provide clear and correct billing of energy usage and a single communication cannel for consumers. In the new market model the following parties in the Dutch energy market can be identified: suppliers, grid operators and metering companies. 1) Supplier Suppliers are responsible for all customer related processes and are the central point of communication for the customer. A supplier is responsable for the checking and gathering of metering data, which they collected through a certified metering company. Based on the metering data from this metering company, the suppliers can bill its customers. 2) Grid Operator Grid operators are the owners of the physical regional electricity grids. They are responsable for the transportation of electricity from various power stations or other grids to the consumer. The grid operators are also responsable for the administration and the installation of the Smart Meters. 3) Metering Company Suppliers hire metering companies to gather raw metering data from their costumers. The metering data is gathered from various grid operators that have customers from the supplier connected to their grid. Once the data is gathered, the metering company will check and validate this data, after which the clean data is send to the supplier. 14 Fig. 7. Dutch metering information flow 3.6. Smart metering landscape in Europe The legislative push by the European Union is currently the main driver for the introduction of intelligent metering systems in Europe. As a consequence, the smart metering landscape is highly dynamic at the moment with many Member States adjusting their energy legislation to comply with the third EU energy market package and the Energy Services Directive. On the other hand, across the European Union, countries are moving towards electronic energy metering as a means to modernise electricity grids and improve the information that is available for grid operators. The modernisation of the electricity grids is key for the integration of highly volatile sources of electricity such as wind. An intelligent grid does not stop at electricity production but includes flexible consumers that help to balance demand and supply. There are various layers of action in and between EU Member States and different EU institutions that are currently working on standardisation, regulatory recommendations, technical functionalities, and other issues of importance. While some Member States await the results of these various working groups and task forces, some actively move towards smart metering and start with a rollout independent of existing barriers to the deployment of smart grids. The overall goal of this, is to promote innovative smart metering services in all Member States that have the potential to achieve energy savings and peak load 15 reduction. That is to say that to accomplish this the matter of importance is the contribution of innovative metering technology and metering services to a sustainable energy system. Due to the regulatory push and the efforts of market actors, the development of legislation and regulation for smart metering in Europe is highly dynamic. The Member states can be arranged into five groups: 1. Dynamic movers are characterised by a clear path towards a full rollout of smart metering. Either the mandatory rollout is already decided, or there are major pilot projects that are paving the way for a subsequent decision. Denmark, Finland, France, Ireland, Italy, Malta, The Netherlands, Norway, Spain, Sweden and the UK are part of this group. 2. Market drivers are countries where there are no legal requirements for a rollout. Some legally responsible metering companies nevertheless go ahead with the installation of electronic meters either because of internal synergetic effects or because of customer demands.Estonia, Germany, Czech Republic, Slovenia, and Romania are in this group. 3. Ambiguous movers like Austria, Belgium and Portugal represent a situation where a legal and/or regulatory framework has been established to some extent and the issue is high on the agenda of the relevant stakeholders. However, due to lack of clarity within the framework, at this point only some of them have decided to install smart meters. 4. Waverers show some interest in smart metering from regulators, the utilities or the ministries. However, corresponding initiatives have either just started, are still in progress or have not yet resulted in a regulatory push towards smart metering implementation. Bulgaria, Cyprus, Greece, Hungary and Poland are ranked in this group. 5. Laggards are countries where smart metering is not yet an issue. This group consists of Latvia, Lithuania, Luxembourg and the Slovak Republic. Fig. 8. Regulation and implementation of smart metering in Europe 22 was originally developed in 1982 by the Massachusets Institute of Technology (USA). The idea started when the professor George W. Hart was collecting and analyzing load data as part of a residential photo-voltaic systems study. Fig. 11. George W. Hart After measuring the electricity consumption of homes, Hart discovered that it was possible to analyze the obtained results to tell what was happening in the monitoring homes. As a result, the professor and the MIT Energy Laboratory Staff realized that this kind of system could have significant value to utilities. The Electric Power Research Institute (EPRI) has sponsored NILM research since its origin. EPRI chose Telog Instruments to commercialize the NILM into a research tool available to electric utilities. First commercial product was developed in the United States at the end of last century. However, several research and development projects have been performed along the years in many countries such as Japan, France, Finland, or Denmark. 4.3. Applications NILM can be used in different fields from residential to industrial. Some of these applications are: • Energy Management: NILM can be used to give feedback on energy usage in order to reduce electricity bills. As an example, the system can be temporarily installed at the customer’s house in order to analyze the characteristics of the appliances and how electricity is used. After some time, it would be possible to suggest how to reduce consumption and costs. • Load Forecast: It is possible to estimate future energy demands by a continuous analysis of the power consumption. This will optimize the energy production implying savings for the power suppliers. In addition, it will help planning transmission and distribution infraestructures. • System Failures: NILM allows to reveal system failures by detecting unusual information from the electrical consumption behaviour. Therefore, it can be used in security control applications such as failure of monitoring devices. 23 4.4. NILM Basics Basically, it’s a device that sits on a home wall socket ( or the electrical panel ), and monitors voltage/current fluctuacions. It’s smart enough to distinguish between fluctuations caused by a cofee pot and a TV set being turned on. By using a learning algorithm and signal processing techniques, it’s able to discern when these devices are turned on/off, and how much power is used by each device. The main goal of NILM is to partition the whole house – building data into its major constituents, like: ( ) ( ) ( ) ( ) 12 ......... n Pt p t p t p t= + ++ The task of NILM is to perform decomposition of P(t) into appliance specific power signals in order to achieve disaggregated energy sensing. Fig. 12. Decomposition of aggregated load power into individual appliances Electrical loads exhibits a unique energy consumption pattern often termed as “load or appliance signatures” that enables the disaggregation algorithms to discern and recognize appliance operations from the aggregated load measurements. NILM: Types of Appliances: 1. ON/OFF Appliances: Only two states of operation. E.g. table lamp, toaster… 2. Multi State Appliances: with a finite number of operating states also referred to as Finite State Machines (FSMs). E.g. washing machine, stove burner… 3. Variable Appliances: their variable power draw characteristics with no fixed number of states. E.g. power drill… It is very challenging for NILM methods to disaggregate these type of appliance from the aggregated load measurements. 4. Permanent Appliances: appliances that remain active throughout weeks or days consuming energy at a constant rate. E.g. telephone sets, smoke detectors… 24 Appliances that should not be taken as target of NILM can be classified as follows: *Appliances with very small power consumption *Appliances which are always on. *Continuosly variable Appliances Small appliances cannot be measured because of noise in the recording equipment. Appliances that are always on can not be recorded because of missing signatures. NILM Approach: Fig. 13. General framework of NILM approach Data acquisition (hardware) and disaggregation algorithms (software). 25 Fig. 14. Power consumption Fig. 15. Examples of different current signatures 26 4.5. Data acquisition module The role of the data acquisition module is to acquire aggregated load measurement at an adequate rate so that distinctive load patterns can be identified. There is a wide variety of power meters designed to measure the aggregated load of the building that can be further classified as follows: Low-Frequency Energy Meters: The sampling rate determines the type of information that can be extracted from the electrical signals. In order to capture the higher order harmonics of the electrical signals, which are integral multiples of fundamental frequency (i.e.50 / 60 Hz), the sampling rate of the energy meter must fulfill the Nyquist–Shannon sampling criteria. In addition, traditional power metrics such as real power, reactive power, Root Mean Square (RMS) voltage and current values can be computed at a low sampling rate (i.e., 120 Hz). High-Frequency Energy Meters: In order to capture the transient events or the electrical noise generated by the electrical signals, the waveforms must be sampled at a much higher frequency in a range of 10 to 100 MHz. The typical NILM system makes use of whole-house data acquired from a single meter. However, one limitation of such an approach is that the identification of low-power and variable appliances in the presence of high-power loads from the whole-house data, which often becomes quite challenging. An alternative approach is to make use of the circuit-level power measurements, as it is often the case that high-power appliances receive a dedicated circuit within homes. The task of power decomposition becomes much easier as there are fewer devices on each circuit in contrast to the whole-house NILM, but at the expense of increased installation complexity and cost. 27 Fig. 16. Data for NILM sampling rate 4.6. Feature extraction A non intrusive signature is one which can be measured by passively observing the normal operation of the load, e.g., a step change in the measured power. Within the non intrusive signatures there is a natural dichotomy according to whether information about the appliance state change is continuously present in the load as it operates (steady-state changes) or only briefly present during times of state transition (transient changes). • Steady-state changes: Steady-state signatures are much easier to detect than transient signatures. The sampling rates and processing requirements necessary to detect a step change in power are far less demanding then those required to capture and analyze a transient current spike. Real power(P) and Reactive power(Q) are two of the most commonly used steady-state signatures in NILM for tracking ON/OFF operations of appliances. Steady-state signatures relate to more sustained changes in power characteristics when an appliance is turned ON/OFF, which can be captured with low – frequency sampling. Fig. 17. Active power changes during appliances application 28 • Transient changes: The transient methods make use of transient signatures that uniquely define appliance state transitions by extracting features like shape, size, duration and harmonics of the transient waveforms. However, distinctive transient signatures can only be extracted if the sampling rate is higher than 1000 samples per second. Transient signatures are more difficult to detect and provide less information than steady-state signatures. However, they can provide useful information to augment that from steady-state signatures. For example, appliances having similar steady-state signatures may have very different transient turn-on currents. Analysis of the transient could provide the deciding information to determine which of the two actually is on in the total load. Fig. 18. Example of an instant-start fluorescent lamp bank transient There has been a debate considering the use of either steady-state or transient based features extraction methods for load disaggregation as both of these approaches have their advantages and disadvantages. Transient signatures are more difficult to detect than steady-state signatures as they measure a switching mode instead of a stabilized state. For this reason, transient analysis require a measurement device to a considerably higher sampling rate. Bearing in mind the cost of the solution, the steady state methods seem to be a more feasible approach because it requires low-cost hardware. In additon, transient signatures provide less information than steady-state signatures because they are available for registration only the switch-on at the time. 29 However, as they provide different information, they might be useful to identify appliances that exhibit similar steady-state behaviour. For this reason, load disaggregation algorithms can incorporate transient features to improve the segregation of appliances with overlapping steady-state features, but at the cost of expensive hardware. Next, I presented in two tables, the steady-state and transient changes: Table 1: Summary of steady-state methods: Table 2: Summary of transient-state methods. 4.7. Learning and inference in NILM systems The supervised disaggregation methods for NILM systems can broadly be divided into optimization or pattern recognition based algorithms. The supervised learning mechanism requires labeled data sets to train the classifier so it would be able to recognize appliance operations from the aggregated load measurement. However, system training requires setting up initial instrumentation, which incurs extra cost and human effort. Therefore, lately researchers are actively looking to devise completely unsupervised or semi-supervised methods that can reduce the effort of acquiring the training data. 30 4.7.1. Supervised learning approaches The supervised disaggregation algorithms need adequate labeled data for learning the model parameters in order to perform the task of appliance recognition. This approach can be divided into: a) Optimisation methods: Optimization based methods deal with the task of load disaggregation as an optimization problem. In the case of single load recognition, it compares the extracted feature vector of an unknown load to that of known loads present in the pool of the appliance database and tries to minimize the error between them to find the closest possible match. b) Pattern reconigtion methods: Pattern matching approaches are the ones most frequently used by the researchers for load disaggregation. The appliance database contains multiple appliance specific features that are used to define the structure and parameters of the recognition algorithm. 4.7.2. Unsupervised learning approaches Recently researchers have started to explore methods to achieve disaggregated energy sensing without a-priori information. It is highly desirable for the NILM systems to be installed in a target environment with a minimal setup cost as the training requirement for the supervised load identification algorithms is expensive and laborious. Hence, unsupervised learning approaches are needed for a wider applicability of NILM techniques. 4.8. Non Intrusive Load Shedding Verification (NILSV) A significant AMI application is demand response, in which meters collect interval readings, transmit signals to appliances, and provide usage data to consumer portals to support poweruse patterns that reduce electricity costs. One strategy gives indirect control to a consumer. The electricity service provider (ESP) assigns a price for electricity in a given time interval; the customer uses this information to make decisions about power use. In direct-control strategies, the ESP sends signals to consumer appliances to alter their use, typically by limiting use during peak demand periods. Each approach has advantages and disadvantages. Demand response, a cornerstone of smart-grid technology, lets consumers participate directly in energy markets by limiting their energy use during periods of emergency or peak demand. In a direct-control strategy,ESP offers consumers discounts or other incentives if they agree to let the ESP send load-shed instructions (LSIs) to specified appliances. Direct control can save consumers money and provide ESPs with valuable tools for controlling energy generation costs and grid stability. But these benefits depend on the LSIs producing the expected response from appliances. Load-shed verification (LSV) can improve reliability and eliminate freeloaders who accept incentives without implementing direct controls. However, this generates many trust challenges because the consumer owns and operates the appliance and because effective demand response depends on the integrity of the appliances’ responses to LSIs. 31 To address these challenges, it has been implemented an algorithm based on a non intrusive load monitoring learning phase that runs during an initialization period at the ESP. The result is a distributed NILM algorithm—a non intrusive load-shed verification (NILSV) algorithm deployed on the residential meter. Ideally, when residential consumers register an appliance for direct control, that appliance can receive LSIs but isn’t required to provide confirmation of compliance. NILM conducted through the residential meter can confirm that the appliance has acted on the LSIs. For instance, suppose an ESP sends a household an LSI to turn off a 1000 W appliance. In a predefined time frame, the meter will respond with an LSV indicating the appliance’s transition. The correct behavior is on to off or off to off. We call this NILSV. NILSV’s key challenge is handling the large amounts of detailed data and complex calculations that aren’t naively suited to typical residential meters’ low bandwidth, computing, and sensor capabilities. Fig. 19. The Non Intrusive Load-Shield Verification (NILSV) process 4.9. NILM: Advantages and disadvantages Unlike ordinary monitoring systems which utilize multiple sensors or meters, NILM collects electrical data by sampling the power consumption at a single point. This means fewer components to install, mantain, and remove. 38 6.2.2. Construction The meter is designed as a three-piece plastic construction, consisting of a housing, verification and top covers, all parts made of fire resistant plastic. The verification cover protects the metrological functions. It is not possible to open the meter without breaking the metrological seal. Fig. 23. Kamstrup Electricity Meter 39 7. HOW TO CONFIGURE IT / HOW IT WORKS 7.1. Home Basic Each Circle must be plugged in between a devide and the socket. The Circle+ is the first Circle to be plugged in, as this is the coordinator of all the Circles. Each Circle measures the energy usage of a plugged in device. After installing the software program Source, we have a real time insight our energy usage. The Stick creates the link between the ZigBee network and Source just plugging it into the USB gate of our computer. Then, the Stick automatically sends measurement data of our plugged in devices to Source, which provides detailed and historical information about energy usage per device. Now we know exactly which devices are using energy when not necessary. The next step is to reduce energy usage by switching these devices off. In Source, we can set up time schedules to prevent energy waste. For example, turn grops of appliances off during the night. 7.2. Kamstrup 382 7.2.1. Start-Up the meter In the first five seconds after being connected the power, the meter will show two differents numbers, First, a 8 digits number which describes the software versión. In the next five seconds, the 4 or 5 digits number which appears, indicates the ROM checksum number. 7.2.2. Registers Kamstrup’s direct meters are constructed as 4-quadrant meters, which provides safe registration of various measured data such as imported and exported energy both active and reactive energy, tariffed energy, power, voltage and current. 40 Every time you take a measurement, the meter indicates in which quadrant is working. 7.2.3. Display The display makes it possible to read out the meter’s registers. Which register depends on the current configuration. The required configuration can be preprogrammed from the factory or configured by means of the METERTOOL. Fig. 25. Meter Display Fig. 24. Measurement in 4 quadrants 41 7.2.4. Loggers The meter has several different loggers for registration of data and events and a load profile logger, and in meters of generation K and later also an analysis logger. The load profile and analysis loggers share the same logging depth, which means that the depth of the analysis logger depends on the logging depth of the load profile logger and thus the configuration of the meter. It is possible to configure the loggers and check their results throught the METERTOOL. 7.2.5. Energy measurement (Shunt measure method) The power that we want to measure is led through a ‘shunt’ resistor, which is in serie with the power grid. A shunt is a special type resistor whose resistor value is precisely yet. The shunt is capable of handling currents up to 63 amperes, which is much than practically ever occurs. The shunt replaces the “current transformers” from the former Ferraris meters and makes the measures much more accurate. The voltage over the shunt is directly proportional with the measured currents. This is the most common method of current measurement, the internal shunt resistor of the Kamstrup is inserted into the circuit and using ohm’s law it’s possible to calculate the current based on the known resistance value and measured voltage value. Using shunt as measuring principle for the current measurement where a resistance stable metal provokes a given drop of voltage at a given current, makes the energy measurement secure and reliable. To summarize the measuring principle: -Current: Single phased current measurement by current shunt. -Voltage: Single phased voltage measurement by voltage divider. Like voltage drop, energy consumption is calculated as an expression of the current compared to the phase voltage and time. 7.2.6. Permanent memory Measured and calculated data is safely stored in the memory (EEprom). Data is stored by every change in energy register values. Furthermore the values are stored at THE END of a debiting period. Measured and claculated data can be read out using the Kamstrup KWh meter tool METERTOOL. 42 7.2.7. Power calculation method There are different methods for the calculation of the registered total power in 3-phase meters. Our meter uses the next method: Fig. 26. Vector summation The first one measures the incoming on L1 and L2 and adds them together. Meanwhile it measures L3 and substracts L3 from sum L1 and L2. This is also named Ferraris method and is based upon the old meters. This calculation method is sensitive to incorrect installation and manipulation, especially if the meter does not have return flow lock. 7.2.8. METERTOOL METERTOOL FOR Kamstrup kWh meter is a tool intented for techniciens and laboratorios to change configuration of meters. It can be also used as backup in communication networks to read out data. It’s essential to have METERTOOL in order to set the meter up and get access to their lectures of measurements. 43 In the CD enclosed with this report, I attached the manual of METERTOOL, as well as the software to install it on PC’s. In order to connect the Smart Meter to the PC, a special extra module is needed. Fig. 27. Kamstrup - PC connection 44 7.2.9. Installation The valid connection diagram appears from the type label on the front of the meter. In our case, Kamstrup 382J the connection is the following: 3 phases, 4 wires: Fig. 28. 3P - 4W connection This is the general connection when we use the Kamstrup 382J but in this case we only have used a monophasic load, therefore the connection has been just: 1 phase, 2 wires: Fig. 29. 1P - 2W connection 45 Fig. 30. Kamstrup - Bulbs connection For more information and details, the Kamstrup Configuration manual is attached in the CD. 46 8. FINAL RESULTS 8.1. Home Basic I succesfully installed the Plugwise Home Basic in my own home and I measured and monitorized my consumption during almost 2 weeks. Exactly, the Plugwise system was installed since 27/04/13 untill 06/05/13. In this time, I was aware about the consumption at home and I made schedules to automatically swith ON/OFF some devices. The appliances in which I installed the system were: TV, Laptop, Fridge, Microwave, Lamp, Adapter (In which they were connected speakers, router and video games console) and Printer. Next, I show the characteristics of some appliances, as well as graphics on the daily measured consumption during the studied period. 47 a) Samsung CW683CNG 28" TV: 230V, 50Hz, 100W Fig. 31. TV report usage 54 10. REFERENCES Below are those documents that have been consulted or that are referred to throughout this report -Security analysis of Dutch smart metereing systems. Sander Keemink and Bart Roos. Univeseteit van Amsterdam. -Smart Grids y la evolución de la red eléctrica. Observatorio Industrial del sector de la electrónica, tecnologías de la información y telecomunicaciones. -Smart Grids: Smart Meter and Non Intrusive Load Monitoring. Joost P. Rey. HSDE/NHL. -Smart meter applications, benefits and issues. Joost P. Rey. HSDE/NHL. -The Smart Meter: Integrating sustainable techniques into applied sciences. Jasper Westra, Sikke de Jong, Benjamin Feenstra. NHL Hogeschool. -Reducing energy costs with peak shaving in industrial environments. Philip Yeung. Asia Pacific Business Development Power Monitoring & Control. -Non Intrusive Appliance Load Monitoring (NIALM): Review and Outlook. Michael Zeifman, Kurt Roth. Fraunhofer Center for Suitainable Energy Systems. -Non Intrusive Load-Shed Verification. David Bergman, Dong Jin, Joshua Juen, Naoki Tanaka and Carl Gunter. University of Illinois at Urbana-Champaign. -European Smart Metering Landscape Report. Stephan Renner, Mihaela Albu, Henk van Elburg, Christoph Heinemann, Artur Lazicki, Lauri Penttinen, Francisco Puente and Hanne Saele. -Kamstrup 162/382 Technical Description. Kamstrup. -The Smart Grid in 2010: Market segments, applications and industry players. David J. Leeds. GTM Research. -Plugwise Catalogue Products. Plugwise. 55 11. INTERESTING WEB PAGES *Plugwise Web Site. http://www.plugwise.com/nl/idplugtype-f/ *Kamstrup Web Site. http://kamstrup.nl/ *Non Intrusive Appliance Monitoring. http://www.georgehart.com/research/nalm.html *Plugwise Unleashed. http://www.maartendamen.com/wpcontent/uploads/downloads/2010/08/Plugwise-unleashed-0.1.pdf *Domotica Forum Europe: Plugwise. http://www.domoticaforum.eu/search.php *How to configure a router to use DHCP. http://www.wikihow.com/Configure-a-Router-toUse-DHCP *Why Smart Meter might be a dumb idea. http://www.consumersdigest.com/specialreports/why-smart-meters-might-be-a-dumb-idea