Open Source Hardware and Software to Support Energy Efficiency and Demand Response in LV Installations

Transcription

1 Open Source Hardware and Software to Support Energy Efficiency and Demand Response in LV Installations G.T. Andreou, Member, IEEE, I.M. Chatzigeorgiou Dept. of Electrical and Computer Engineering Aristotle University of Thessaloniki Thessaloniki, Greece Abstract-The design and implementation of Smart Grids takes usually into account the participation of LV electrical energy consumers (through techniques such as Demand Response). However, there is still a lack of cost effective hardware and software solutions that would allow these consumers to monitor their installation, comprehend their electrical energy consumption, take action towards achieving energy efficiency, and effectively participate in Demand Response programs. This paper aims to address this issue, presenting a complete, open source hardware and software solution, which enables measurements of electrical voltages and currents at the installation main distribution panel, the transmission and storage of the respective data in a personal computer, and their subsequent processing by the Cassandra open source software platform. The analysis of the proposed solution components is followed by results obtained by the implementation in a residential installation. Index Terms-Non intrusive Load Monitoring, Home automation, Demand Reponse I. INTRODUCTION On a theoretical basis, the research on Smart Grid paradigms has produced a multitude of useful methodologies and techniques over the last years. The basic aim of these methodologies and techniques is often energy efficiency and sustainability, at individual installation or community level. Naturally, such an aim can only be achieved through the active participation of electrical energy consumers, and thus these consumers have become the target group for many novel paradigms [1 ]-[3]. There is an inconsistency though between theory and practice on this matter, mainly due to the inability of a great portion of electrical energy consumers to implement such paradigms in their installations. More often than not, the consumers lack the required metering equipment in order to be able to monitor their consumption. Even when the electrical utilities undertake the task to provide this equipment (e.g. through the installation of smart meters), the respective information seldom reaches the consumer involved, in a way that is comprehensive for himlher. There are commercial solutions able to provide the required services for a consumer to understand his/her consumption and act on it, but they naturally don't cover the cost for the required metering and communication hardware, and they usually require a fee. Regarding the essential hardware, solutions may be found in Open Source communities, usually also combined with respective software. These solutions may be simpler [4], or more elaborate [5], in terms if the required skill for their implementation. The effective implementation of energy efficiency or demand response techniques however requires a level of analysis on the respective measurement data that lies beyond the capabilities of the average electrical consumer, and usually is not to be found in the simpler open source solutions. On the other hand, the more elaborate ones often exceed the capabilities of an average consumer with DIY capabilities. What is more, the available software solutions are usually combined with specific hardware. In this paper, a complete, open source hardware and software solution is presented, aiming to enable the active involvement of an electrical energy consumer with respect to energy efficiency and the implementation of demand response, at low cost. More specifically, the hardware part is based on the open source hardware platform Arduino. The software part utilizes the Cassandra open source software platform [6]. This platform has been developed within the scope of the EU funded FP7 project "CASSANDRA - A multivariate platform for assessing the impact of strategic decisions in electrical power systems". Among its functional capabilities, it can accept as input per minute measurements of active and reactive power regarding a single LV installation, disaggregate them into specific appliances, and perform data analysis on the results. Moreover, the user can utilize the resulting model of hislher installation in order to simulate specific scenarios (ranging from the replacement of energy consuming

2 appliances to the implementation of energy saving practices), and use the results to support energy related decisions. II. HARDWARE The hardware of the proposed solution can be divided into two sub-systems, namely the metering apparatus, and the microcontroller unit that undertakes the basic processing of the input data, and their transmission to a computer. the measurements, whereas the combination of an Xbee Shield and Xbee Antenna are used to handle communications, implementing the IEEE protocol. A second Arduino Uno board is used with the same Xbee add-on in order to receive the data, and store them through a USB connection in a computer. The two boards used for the purposes of this work are depicted in Figure 3. A. Metering Apparatus The metering apparatus consists of the sensors utilized to capture the voltage and current waveforms, as well as the essential circuit in order to transform these waveforms into permissible inputs for the Arduino microcontroller unit. A split-core current transformer is used to capture the current waveform, whereas a V AC/AC voltage transformer is used for the voltage waveform. The interface between these sensors and the Arduino microcontroller is basically a voltage follower, implemented through the utilization of an operational amplifier (Figure 1). Figure 3. Microcontroller units. VOl1AVe I co A.! '0'" I l T' ''' "'3:j'.,,,"'" 1 0 ;o 1 II: _ W e, t o =,", I. G',D 0.,z ::J :::>0 a: «Figure 1. Eagle schematic of the interface circuit between the voltage/current sensors and the Arduino board. In Figure 2, the actual metering interface is depicted. Figure 2. B. Microcontroller unit Metering interface The basic processing of the measurements, as well as the transmission of the results to a computer, are conducted with an Arduino board. More specifically, an Arduino Uno board is used in the proposed solution, in order to gather and process III. SOFTWARE As aforementioned, the metering hardware is a necessity in order to monitor the electrical energy consumption in an installation. However, it is not enough by itself, especially in the case of LV consumers with no understanding of electrical energy properties. Such consumers require the presentation of energy consumption information in a manner that is comprehensible for them, and will help them act on it. To this purpose, the Cassandra software platform was selected within the scope of the solution presented here, as it offers comprehensible data analytics for average consumers, regarding their electrical energy consumption, while at the same time it is a free, open source platform. A, Cassandra software pla(form The Cassandra software platform was designed and developed as a decision support tool, which allows energy market stakeholders to model, simulate and understand energy demand and consumption. Among its features, the platform can: model installations based on single point (home meter) data, through a disaggregation methodology that utilizes per minute measurements of active and reactive power, simulate consumption and correlate it to consumer activities, rather than appliance use, group consumers into consumer social networks based on their load profiles and consumption behavior, easily scale-up simulations from one installation to neighborhoods or even cities, identijy the carbon footprint of the simulated scenario, and implement various Demand-ResponselFeedback Response scenarios and explore new Pricing Schemes on the simulated populations

3 Within the scope of the solution presented here, the Cassandra platform can accept as input the measurements obtained by the hardware analyzed in Section II of the paper, disaggregate them, and build a model of the installation under study (Figure 4). What is more, it can correlate the respective consumption to consumer activities (such as cooking, cleaning, recreation etc.), helping thus the respective consumer comprehend the energy cost of his/her habits. A. Measurements The metering apparatus described in sub-section ILA. was used to obtain the initial measurements for the residential installation under study. Figures 5 and 6 depict two indicative days of measurements. r , Active Power 2 Reactive Power Cha"" ;;; 1 c.. c -- L. J L_ \l--, OO = time Figure 6. Indicative day 2 of measurements. Figure 4. Model ofl V residential installation in the Cassandra platfonn The consumer can also use the Cassandra model of his/her installation in order to run simulation scenarios, that will for example show the feasibility regarding the replacement of an energy consuming appliance, or will guide him/her in the attempt to make the most of a specific time-of-use pricing scheme. IV. SYSTEM IMPLEMENTATION For the purposes of this work, the proposed solution has been installed in a residential installation in the city of Thessaloniki, in Northern Greece, aiming to show the results as presented to the owners, as well as the general functionality of the Cassandra platform. 3000,-----,-----,-----, 1 1 In total, two weeks of measurements were used as input for the Cassandra platform. The initial measurements were stored in a computer, and subsequently processed, in order to result in per minute average values of active and reactive power, in a.csv file, so as to meet the input requirements of the Cassandra platform. B. Installation modelling in the Cassandra platform The disaggregation procedure in the Cassandra platform was used for the automatic modelling of the installation. Based on the two weeks of measurements used as input, 44 appliance instances were identified, along with the installation background consumption (corresponding to the aggregated power demand of all appliances in standby mode). These instances are categorized in Table I. TABLEL IDENTIFIED ApPLIANCE INSTANCES Appliance type Number of Instances Activity type Washing Machine 11 Cleaning Vacuum Cleaner 2 Cleaning Cooking 9 Cooking Cooking Extractor 6 Cooking Microwave Oven 4 Cooking Computer 9 Entertainment Lighting 2 Lighting Refrigerator 1 Refrigeration -5OO0----5OO= OO----15OO lime Figure 5. Indicative day 1 of measurements. It should be noted that the different instances of an individual appliance type may correspond to different operation modes (e.g. in the case of a vacuum cleaner or cooking extractor), to different appliances with similar

4 consumption patterns (e.g. in the case of the Cooking appliances, which correspond to resistive appliances such as stovetops, boilers, etc.), or to a combination of both categories. In any case, all identifications regarding the operation of an appliance within a specific measurement time frame, are categorized into one of the corresponding appliance instances. Thus, an appliance instance may correspond to one or more appliance operation identifications. For every appliance instance, a consumption pattern is identified from the measurements (e.g., in Figure 7 the identified pattern of a washing machine instance is shown). Moreover, three probability density functions are also determined, corresponding to the appliance operation cycle duration, the operation start times within a day, and the possible operation repetitions within a day 'Active Power rn)' - 'Reactive Power eva)' I Distribution Chart x ,1;' 0.7.g 0.5 rr Figure Start Time (Minute of day) Identified start time probability density function regading a washing machine instance Dlstnbution Chart t: Figure Identified active and reactive power patterns for a washing machine instance. 0.9,1;' Ill o o.a Daily Repetitions Figures 8-10 depict the corresponding probability density functions regarding the washing machine instance of Figure 5. Distribution Chart l O Duration (Minutes) t:. Figure 10. Identified repetition probability regading a washing machine instance. The combination of these parameters, as they are extracted by the platform, may be used in order to perform simulations regarding the installation under study. In this context, Figure 11 shows the simulated active power consumption in the installation for a random day, whereas Figure 12 depicts the corresponding simulation consumption per activity type and appliance type.,, I- ",. I- "'" Figure 8. Identified duration probability density function regading a washing machine instance., n 7\-1/ ""1,. n M m _ Time (5 Minutes) Figure 11. Simulated active power consumption for a random day.

5 C. Extractor Standby Consumption Vacuum C l eane r Microw av e Oven Refrigerator Cooking Washing Mach ine Entertainment Lighting Figure 12. Simulated active power consumption per activity type and appliance type. Decision Support The Cassandra platform can also accept as input information regarding the pricing scheme for a given installation. Using this feature with the simulated installation under study, and running a simulation over a whole billing cycle (120 days in Greece), a total electricity cost of results. The pricing scheme considered is the one valid for the installation, as depicted in Table II. TABLE II. PRICING SCHEME FOR THE INSTALLATION UNDER STUDY Billing cycle (days) 120 Total fixed cost per billing cycle ( ) 3.40 Scale (kwh) Scalable energy cost Cost ( /kwh) > The interesting aspect of this pncmg scheme is that the scale price corresponding to the overall consumption of the consumer is imposed on the total of consumed kwhs. However, this is a trivial matter for the installation under study, as the overall consumption in a billing cycle does not exceed the first consumption scale. Moreover, the platform can also act as a decision support tool for energy related interventions. To this purpose, the user may find indicative appliances of different energy classes in the Cassandra Library, and subsequently use them to replace appliances in the simulation under study, aiming to explore the feasibility of a possible replacement. In the example presented here, a replacement of the installation refrigerator with a new one, of Energy Class A, yields a cost difference of 9.04 for one billing cycle, i.e. a saving of over a year. Thus, in this case it is evident that the replacement is not justified in terms of feasibility. Finally, the user can also utilize the platform in order to explore the benefits regarding the implementation of energy saving practices. For example, in the installation under study there is a standby consumption of a minimum of 40W. Simulation in the platform shows that the decrease of this minimum standby consumption to low will save a total of over a billing cycle, corresponding to a 26.85% cost reduction. V. CONCLUSIONS In this work, an affordable, open source solution is presented that can provide data analysis regarding the electrical energy consumption of an installation. The hardware part of the proposed solution, designed to provide the required measurements regarding the installation consumption, is based on the Arduino microcontroller. Moreover, the data analysis is performed by a platform developed within the scope of the EU funded FP7 project "CASSANDRA - A multivariate platform for assessing the impact of strategic decisions in electrical power systems". The platform is able to disaggregate electrical energy measurements of an installation in order to effectively model it, based on the respective appliances, and the corresponding consumer activities. Apart from that, the platform is able to simulate consumption scenarios for a modeled installation, aiming to provide decision support for energy saving related interventions. In this work, this functionality is exhibited by the modeling and simulation of a residential installation in Northern Greece. REFERENCES [I] K. Spees, LB. Lave, "Demand Response and Electricity Market Efficiency," The Electricity Journal, vol. 20, Issue 3, pp , Apr [2] C.-L Su, D. Kirschen, " QuantifYing the Hlect of Demand Response on Electricity Markets." IEEE Trans. Power Systems, vol. 24, no. 3, pp , Aug [3] P Palensky, D. Dietrich, "Demand Side Management: Demand Response, Intelligent Energy Systems, and Smart Loads," IEEE Trans. Industrial InformatiCS, vol. 7, no. 3, pp , Aug [4] The Open Energy Monitor Project: org/emon/ [5] J. Xiaofan, S. Dawson-Haggerty, P. Dutta, D. Culler, "Design and implementation of a high-fidelity AC metering network," in Proc. International Coriference on lriformation Processing in Sensor Networks (IPSN 2009), pp , April [6] GT Andreou, AL Symeonidis, C. Diou, PA Mitkas. D.P. Labridis, "A framework for the implementation of large scale Demand Response," Presented at the 2012 International Coriference on Smart Grid Technology, Economics and Policies (SG-TEP), 3-4 Dec Available at IEEEXplore: org/stamp/stamp.jsp?tp=&arnumber=