**4. Experimental results of a pilot project**

#### **4.1. Project overview**

6 Will-be-set-by-IN-TECH

the multiplications that are constantly executed during the energy metering process can be executed very quickly, and measuring energy with these general purpose microcontroller ICs

This technique is similar to the one implemented in hardware level, using the same acquisition methods and digital processing principles. However, using this approach, the system can be customized to operate in ultra-low power, as in the technique developed in [15] to design a battery powered energy meter using the microcontroler MSP430AFE2xx [16]. The system was configured to operate in a 60 Hz AC line and during every period of one second, make the measurements only during 3 cycles and stay sleeping for the others 57 cycles. The system calculates the RMS value of the product *V* × *I* and adds this value a register. Assuming that there is no significant change in the current and voltage during each period of less then one

1s

Awake period < 66,67 ms Sleeping period > 933,33 ms

Establishing mesh networks using wireless and Power Line Communications modems (PLC) concentrators/routers together with smart metering sensors which are connected to the home appliances can provide both accumulated and real-time information of the energy spent in every appliance in a house. Adding actuators to these smart sensors, it is possible to manage remotely the status (on-off) of a few important loads (like clothes dryer, air conditioning) in order to avoid that these loads be turned-on during a peak period. In this case the household has to buy the system, since the information has to be continuously monitored and the energy

Another possibility, which does not involve the costs associated with the ownership of the measurement system, are offered as a service by the utilities or an ESCO (Energy Service Company). The energy monitoring is performed through the installation of low-cost smart metering sensors to the home appliances, and after a sampling period (typically 2 weeks), the meters are read and a diagnosis report of the energy use is prepared by the company that is offering the service. When there is no need for getting real time information, the installation

An example of energy monitoring system of the typical appliances in a home was presented by [19]. This proposed monitoring system presents a limited performance because only electrical appliances which were connected to the outlets could be measured, since the monitoring system could not access the lighting equipment, which represents an important part of the

is extremely easy, since the devices do not need to form a mesh network.

t

can be very competitive when compared to the energy meter ICs.

second, the accumulated value in the register is equal to the energy.

i

**Figure 4.** Awake and sleeping periods

**3. Decentralized measurement methods**

savings will be dependent on this monitoring.

energy spent in many households.

Recently a pilot project, supported by ANEEL (the Brazilian Electrical Energy Regulating Agency) and AES Eletropaulo (the major power electric utility in the State of São Paulo) was developed by the Department of Electronics and Microelectronics of the School of Electrical and Computer Engineering, at the University of Campinas [7]. The objective was to develop and test a prototype of a hybrid (wireless and PLC) intelligent sensor network system which should be able to perform the breakdown of the electricity bill of the customers. The premise was that with detailed information in hands, consumers could understand better how much they spend the electrical energy in every single electrical device in a house and, this would lead to changes in habits and substantial savings in electrical energy.

The developed system was planned to be offered as a service by the ESCOs, so that both customers who want to know better how they use the electrical energy and energy-efficiency programs were potential candidates for using the service. In order to be practical, the developed system had to be simple to install and easy to be deployed in a residence or small business, without requiring any changes in the original electrical wiring. Although the electrical appliances which are connected to the mains outlet could be read using a small module containing an energy meter IC, it was mandatory that the energy spent on lighting should be monitored, and special sensors had to be designed for this application.

The core of the system was designed around a ZigBee wireless sensor network, and is composed by five types of modules: a coordinator, a displaying-processing unit (DPU) and three types of smart energy meters. A PLC module was also implemented, for special cases where the mesh wireless network is difficult to be implemented.

(a) Coordinator (b) DPU (c) Power outlet adaptor smart meter

(d) Clamp smart meter (e) Light smart meter

#### **Figure 5.** System modules

The coordinator is the main element, and only one module is required per each mesh network. The coordinator is responsible for storing the information sent by the smart energy

#### 8 Will-be-set-by-IN-TECH 200 Energy Effi ciency – The Innovative Ways for Smart Energy, the Future Towards Modern Utilities Energy Measurement Techniques for Energy Efficiency Programs <sup>9</sup>

meters, and also to sustain the ZigBee mesh network. It also measures and stores the AC voltage to accurately calculate the energy from clamp smart meters. Sometimes, due to the characteristics of the local being monitored or due to the building structure/construction, it is difficult to establish the communications between some end devices and the coordinator. If the devices in the network are too far away from each other, a second coordinator can be added to create and maintain a second ZigBee mesh network. The coordinators will communicate via Power Line Communications (PLC) and a master coordinator will concatenate the information of every smart energy meter.

The report has a filter that can show only one type of smart meter, two, or all tree of them. A horizontal graphic bar shows the energy consumption of each electrical device. This graphic shows the energy spent (in KWh) or the energy cost (in this software version in R\$ - Brazilian

Three types of smart energy meters were developed to measure and monitor the energy consumption of all electric devices in a house, without requiring any changes in the original electrical wiring installation. The first one, the power outlet adapter smart meter, is employed to measure and monitor the energy consumption of every device that can be plugged to a power outlet. The home appliance power cord is simply plugged to the smart energy meter

> Power outlet adapter smart meter type A

The outlet adapter meter measures the amount of energy consumed by the appliance it is monitoring, and send this information to the coordinator every five minutes. Because it is connected to the AC mains, it does not need batteries, even in the presence of a long power supply shutdown, since the data is stored in an internal flash memory. It also performs the function of a router in the wireless network, forwarding the information from the others smart

The second smart meter is a clamp smart meter, and it runs on batteries. It was designed to monitor resistive loads that demand currents over 15 A, as it is the case of electrical showers, heaters and other high current loads. The current clamp transforms the ac current in ac voltage proportional to the amplitude of the current, and the circuit of the meter calculates the time integral of the electrical current (in Ah). It is important to remember that only the current is measured and, therefore, the energy has to be calculated. This energy calculation is performed by the network coordinator, that multiplies the data (in Ah) sent by the smart clamp meter by

The lighting smart meter developed also runs on batteries. It is employed to measure and monitor the energy consumption of lighting devices which are not accessible by the clamp meter (for example, fluorescent lights in ceiling mounting fixtures). It uses a photodiode to detect the light status (on or off). When the photodiode detects that a light (or a set of lighting fixtures) was turned on, the microcontroller starts to integrate the amount of time that the light is kept on. The system uses one light smart meter for each lighting set that is controlled by one switch. During the system installation, the nominal power of all lights that are monitored by one photodiode is stored in the microcontroller's memory, and the energy consumed by

meters to the coordinator, if the smart meters cannot talk to the coordinator directly.

the average value of voltage measured by the coordinator's circuit.

the set of lights is then calculated by the software in the DPU.

Power cord

Energy Measurement Techniques for Energy Effi ciency Programs 201

module and the module is connected to the AC power outlet, as shown in Figure 8.

Real, but it can easily be converted to any currency) for each appliance.

Power outlet

**Figure 8.** Power outlet adapter smart meter.

The hybrid communication technology used (ZigBee + PLC) made the system very robust in terms of coverage. However, the PLC is used only to transfer the information from a coordinator to another and since reading a coordinator is a simple and fast process, two (or more) coordinators can be read with the same DPU, eliminating the PLC module and resulting in a lower cost system that can be easily installed.

**Figure 6.** PLC communication connecting two wireless networks.

The DPU is basically a PC that connects to the wireless network to retrieve the information stored in the coordinator. It is also responsible for processing the information according to the configuration file created during the set-up phase(by the person who is installing the system), linking each of the appliances to the IDs of the smart energy meters. After processing, the DPU displays the report in an user-friendly screen, showing the energy spent of items in a user-friendly way, like John's notebook, lighting of dining room, air conditioning of Rose's bedroom, etc.


(a) First tab - Configuration file (b) Second tab - Report

**Figure 7.** DPU running the reporting application.

The report has a filter that can show only one type of smart meter, two, or all tree of them. A horizontal graphic bar shows the energy consumption of each electrical device. This graphic shows the energy spent (in KWh) or the energy cost (in this software version in R\$ - Brazilian Real, but it can easily be converted to any currency) for each appliance.

Three types of smart energy meters were developed to measure and monitor the energy consumption of all electric devices in a house, without requiring any changes in the original electrical wiring installation. The first one, the power outlet adapter smart meter, is employed to measure and monitor the energy consumption of every device that can be plugged to a power outlet. The home appliance power cord is simply plugged to the smart energy meter module and the module is connected to the AC power outlet, as shown in Figure 8.

**Figure 8.** Power outlet adapter smart meter.

8 Will-be-set-by-IN-TECH

meters, and also to sustain the ZigBee mesh network. It also measures and stores the AC voltage to accurately calculate the energy from clamp smart meters. Sometimes, due to the characteristics of the local being monitored or due to the building structure/construction, it is difficult to establish the communications between some end devices and the coordinator. If the devices in the network are too far away from each other, a second coordinator can be added to create and maintain a second ZigBee mesh network. The coordinators will communicate via Power Line Communications (PLC) and a master coordinator will concatenate the information

The hybrid communication technology used (ZigBee + PLC) made the system very robust in terms of coverage. However, the PLC is used only to transfer the information from a coordinator to another and since reading a coordinator is a simple and fast process, two (or more) coordinators can be read with the same DPU, eliminating the PLC module and resulting

> Master Coordinator

ZigBee Network 1 ZigBee Network 2

PLC link

The DPU is basically a PC that connects to the wireless network to retrieve the information stored in the coordinator. It is also responsible for processing the information according to the configuration file created during the set-up phase(by the person who is installing the system), linking each of the appliances to the IDs of the smart energy meters. After processing, the DPU displays the report in an user-friendly screen, showing the energy spent of items in a user-friendly way, like John's notebook, lighting of dining room, air conditioning of Rose's

(a) First tab - Configuration file (b) Second tab - Report

Slave Coordinator

of every smart energy meter.

bedroom, etc.

in a lower cost system that can be easily installed.

DPU

**Figure 7.** DPU running the reporting application.

**Figure 6.** PLC communication connecting two wireless networks.

The outlet adapter meter measures the amount of energy consumed by the appliance it is monitoring, and send this information to the coordinator every five minutes. Because it is connected to the AC mains, it does not need batteries, even in the presence of a long power supply shutdown, since the data is stored in an internal flash memory. It also performs the function of a router in the wireless network, forwarding the information from the others smart meters to the coordinator, if the smart meters cannot talk to the coordinator directly.

The second smart meter is a clamp smart meter, and it runs on batteries. It was designed to monitor resistive loads that demand currents over 15 A, as it is the case of electrical showers, heaters and other high current loads. The current clamp transforms the ac current in ac voltage proportional to the amplitude of the current, and the circuit of the meter calculates the time integral of the electrical current (in Ah). It is important to remember that only the current is measured and, therefore, the energy has to be calculated. This energy calculation is performed by the network coordinator, that multiplies the data (in Ah) sent by the smart clamp meter by the average value of voltage measured by the coordinator's circuit.

The lighting smart meter developed also runs on batteries. It is employed to measure and monitor the energy consumption of lighting devices which are not accessible by the clamp meter (for example, fluorescent lights in ceiling mounting fixtures). It uses a photodiode to detect the light status (on or off). When the photodiode detects that a light (or a set of lighting fixtures) was turned on, the microcontroller starts to integrate the amount of time that the light is kept on. The system uses one light smart meter for each lighting set that is controlled by one switch. During the system installation, the nominal power of all lights that are monitored by one photodiode is stored in the microcontroller's memory, and the energy consumed by the set of lights is then calculated by the software in the DPU.

The system is designed to monitor a house for a period of one month, but it can also be used to monitor a fraction of that time and estimate the monthly cost of the electricity bill based on the acquisition period, for households with very stable and repetitive habits. A period of less than two weeks is not accurate, because some important energy hungry events may be missed, like when there is no need for using the cloth drier machine because the weather was very good in that week.

**Appliance Energy %**

Energy Measurement Techniques for Energy Effi ciency Programs 203

**Appliance Energy %** PC with amplified speakers 10.83% Home Theater 0.00% Electric rice cooker 1.58% Satellite TV receptor 2.57% Treadmill 2.24% PC with wireless dongle 19.40% PC + Wireless router + notebook 18.48% Cordless telephone 0.89% TV LCD 43" (bedroom) 16.71% Microwave 0.56% Lighting 120W (TV room) 0.05% Refrigerator + freezer (side-by-side) 16.36% TV LCD 53" (TV room) 9.00% Lighting 160W (kitchen) 1.30% Lighting 50W (closet) 0.02% **TOTAL 100.00%**

From the results of the field tests it was observed that habits and behavior are responsible for a lot of energy waste. From the results measured in these households it becomes clear that a detailed report with the breakdown of the electricity bill can result in significant electrical

Notebook 0.03% Hair Dryer 0.53% Washer/Dryer Machine 0.72% TV (living room) 11.26% Refrigerator 1 10.37% Microwave 1.04% Alarm clock/radio 12.20% Lighting 50W (over the piano) 0.03% Lighting 50W (bathroom) 0.00% Lighting 50W (kitchen) 22.55% Lighting 50W (office - over the PC) 7.47% Lighting 50W (office) 2.37% Refrigerator 2 15.48% Cordless telephone 12.23% TV (bedroom) 3.72% **TOTAL 100.00%**

**Table 2.** Results measured in a medium size apartment with 3 people

**Table 3.** Results measured in a large house with 3 people

energy savings just by changing habits.

#### **4.2. Field test results**

The system was tested in the field, and the results showed that most of the consumers waste a lot of energy due to habits that could be easily changed without any impact on the comfort or the quality of life.

Table 1 shows the results measured in a very small apartment, with one person who leaves the apartment early in the morning and returns only at 7:00 PM. It can be seen that this household spends 76% of the electricity with only two appliances: a Computer and a VCR. The PC is a high load in this household, with 64% of the monitored energy consumption. After the report on how the energy is used was presented, the resident reported that the PC is a high end device with powerful video cards and was kept on 24/7 just to eliminate the slow boot time. Concerning the VCR (which is responsible for about 12% of the consumption), it was informed that it is a very old model, that is never turned-off and is used only to change the channels in the TV, because the TV's remote control was not working. In this case, the energy savings could reach a level of almost 80% if the computer is turned off when it is not in use and a new TV remote control is bought.


**Table 1.** Results measured in a small apartment with 1 people

The results from another household, a medium size apartment with 3 people, is presented in table 2.

This test found that the resident has an old digital alarm clock/radio (which, obviously, was kept turned on 24/7) that presents a huge energy consumption. If this alarm clock was replace with a new and more economic model, it would pay itself back after a few months and energy savings in the order of 12% would be achieved.

The results measured in another interesting case, a large house with three people, are presented in table 3. This household shows a high energy consumption by computers and televisions. The residents reported that the treadmill was kept in stand by during all the acquisition period, and a TV set on a bedroom is always kept on during the whole night. The PCs are also left on continuously to avoid the slow boot time.



**Table 2.** Results measured in a medium size apartment with 3 people

The system is designed to monitor a house for a period of one month, but it can also be used to monitor a fraction of that time and estimate the monthly cost of the electricity bill based on the acquisition period, for households with very stable and repetitive habits. A period of less than two weeks is not accurate, because some important energy hungry events may be missed, like when there is no need for using the cloth drier machine because the weather was

The system was tested in the field, and the results showed that most of the consumers waste a lot of energy due to habits that could be easily changed without any impact on the comfort

Table 1 shows the results measured in a very small apartment, with one person who leaves the apartment early in the morning and returns only at 7:00 PM. It can be seen that this household spends 76% of the electricity with only two appliances: a Computer and a VCR. The PC is a high load in this household, with 64% of the monitored energy consumption. After the report on how the energy is used was presented, the resident reported that the PC is a high end device with powerful video cards and was kept on 24/7 just to eliminate the slow boot time. Concerning the VCR (which is responsible for about 12% of the consumption), it was informed that it is a very old model, that is never turned-off and is used only to change the channels in the TV, because the TV's remote control was not working. In this case, the energy savings could reach a level of almost 80% if the computer is turned off when it is not in use

> **Appliance Energy %** TV 17.93% Cell Phone + Charger 1.00% VCR 12.71% Refrigerator 3.78% PC 64.00% Light bulb 20W (kitchen) 0.58% **TOTAL** 100.00%

The results from another household, a medium size apartment with 3 people, is presented in

This test found that the resident has an old digital alarm clock/radio (which, obviously, was kept turned on 24/7) that presents a huge energy consumption. If this alarm clock was replace with a new and more economic model, it would pay itself back after a few months and energy

The results measured in another interesting case, a large house with three people, are presented in table 3. This household shows a high energy consumption by computers and televisions. The residents reported that the treadmill was kept in stand by during all the acquisition period, and a TV set on a bedroom is always kept on during the whole night. The

very good in that week.

**4.2. Field test results**

or the quality of life.

table 2.

and a new TV remote control is bought.

**Table 1.** Results measured in a small apartment with 1 people

PCs are also left on continuously to avoid the slow boot time.

savings in the order of 12% would be achieved.


**Table 3.** Results measured in a large house with 3 people

From the results of the field tests it was observed that habits and behavior are responsible for a lot of energy waste. From the results measured in these households it becomes clear that a detailed report with the breakdown of the electricity bill can result in significant electrical energy savings just by changing habits.

#### 12 Will-be-set-by-IN-TECH 204 Energy Effi ciency – The Innovative Ways for Smart Energy, the Future Towards Modern Utilities Energy Measurement Techniques for Energy Efficiency Programs <sup>13</sup>

In the case presented in table 3, the electricity bill was monitored during the next six months after the results were presented. The family reported that, after the report was presented, all three PCs were turned-off during the night and the timer of the bedroom TV was set to turn-off the TV set at midnight.In the case of the The monthly savings in the electricity bills during these six months period were over 20%, when compared to the same month of the previous year. This result is a clear evidence that if a customer is informed of a habit that is wasting energy and this habit can be easily changed, a lot of energy can be saved and and this savings tend to be permanent.

two different-sized step changes are clearly present, providing characteristic signatures of the refrigerator and the microwave oven. Knowing the time of each on and off event, it is possible to determine the total energy consumption of the refrigerator and the microwave oven.

> 16 Microwave cycles

Energy Measurement Techniques for Energy Effi ciency Programs 205

Time

A method to construct taxonomy of electrical appliances based on load signatures is presented in [10]. In this work the authors suggests a 2-dimensional form of load signature denominated voltage-current (V-I) trajectory to characterize typical household appliances. The V-I trajectory load signatures consists in acquiring the steady-state voltage and current in one-cycle long, normalize them to eliminate the effect of the current magnitude in the size of V-I trajectory, and then plot the V-I trajectory. After creating the trajectories for the appliances, the shapes of

The proposed methodology for constructing the load taxonomy is summarized as: (1) the voltage and current waveforms of the household appliances are measured; (2) load signatures in the form of V-I trajectory are constructed; (3) shape features are extracted from the V-I trajectories; (4) hierarchical clustering method is applied to cluster the appliances; (5) the load

In [13] the authors proposed a methodology of using load signatures and Genetic Algorithms (GA) to identify electrical appliances from a composite load signal. They introduced a classification method to group the appliances and how to disaggregate the composite load signals by a GA identification process from a generated random combinations of load

The methodology consists of defining a signature for each appliance by averaging 50 consecutive one-cycle steady-state current waveforms. Then the current waveforms are grouped by the ratio of their fundamental (50Hz) component verses their Root-Mean-Square (RMS) total, after a Fast Fourier Transform (FFT) calculation. That means that the higher the

**Figure 9.** Characteristic signatures of a refrigerator and a microwave oven sensed on the same circuit In [12] the authors highlighted the fact that the complex electrical loads of today have signatures that vary with time, depending on their state and mode of use and that common appliances can have non-linear load characteristics. They propose a conceptual modeling to characterize an appliance based on three sets of signatures that are extracted from the appliance: steady state, transient state and operational pattern and therefore construct a

Refrigerator cycle #2

Refrigerator cycle #3

Power consumption

Refrigerator cycle #1

taxonomy for the appliances.

the trajectories of the appliances can be analyzed.

signatures from the groups of appliances.

taxonomy is constructed according to the clustering results.

The measurement system was designed to get the information from every appliance in the house in real time, but the customers reported that they were interested only in the two weeks final result. Thus, the ZigBee network could be removed from the smart meters and a simple peer-to-peer communication (from each smart meter to the DPU) could be used, reducing the power consumption of the measurement system and increasing the batteries' life.
