India’s power grid is notoriously unstable. Outages happen suddenly and can last for hours. IEEE Member Tanuja Ganu knows this problem all too well. She often studied for school tests by candlelight.
With India’s burgeoning middle class able to afford modern conveniences like washing machines and water heaters, the grid will become even more burdened, causing more outages for its utility customers. Although India boasts the world’s fifth-largest supply of electricity, it still cannot meet some 4.5 percent of the country’s demand during peak hours. Ganu hopes to change this with the help of a gadget that fits between an appliance and a wall outlet.
Called an nPlug, the device monitors the voltage and frequency of the power supply coming in from the socket. It uses that information to infer when the grid will face the highest power demand. It can then schedule appliances to run during off-peak hours, which could help reduce the pressure on the grid and prevent blackouts, says Ganu, an engineer at IBM Research-India, in Bangalore. She was named one of MIT Technology Review’s “35 Innovators Under 35” this year.
The normal voltage available from India’s grids is 230 volts. But at peak times, the voltage drops to about 215 V and can rise to around 250 V during off-peak times, Ganu points out. “If we want outages to be reduced, then some of the demand needs to be shifted,” she says. Completely averting blackouts would require a large number of nPlugs, she adds, but even relatively few would help reduce the voltage problem.
ANALYZING THE GRID
When she was a grad student, Ganu interned for six months in 2010 at IBM Research-India, in its new Smarter Energy Group. The group was experimenting with a plug-in device to collect data on the electricity consumption of various appliances in 10 Indian households, seeking useful patterns, such as the days, times, and durations that consumers preferred to use certain appliances. Specializing in data mining and machine learning in graduate school, Ganu immediately took on the role of analyzing the data coming in to those plug-ins. She also developed an algorithm that can schedule appliances to run when demand for electricity is low. The user can have input, requesting a load of laundry, say, to be done by 2 p.m., Ganu explains. However, instead of running the washing machine between 8 and 10 a.m., when grid load is at its peak, the algorithm would schedule it for before or after, she says.
There was just one hitch. The idea wasn’t practical for developing countries. The plug-in sends data over the Internet to an IBM server where the data crunching takes place. Commands are then sent back to the device. Such two-way communication between appliances and utilities would be infeasible in India where just over 11 percent of households have access to the Internet. “Developing countries like India just don’t have the computation and communication infrastructure,” she says.
So when Ganu joined the company as an engineer in 2011, she and her colleagues decided to develop a device that could measure and analyze electricity data at the wall socket and make decisions locally on when to schedule appliances to turn on. That concept became the nPlug.
The nPlug contains frequency- and voltage-sensing circuits. A microcontroller runs software that learns peak and off-peak demand times based on a variety of factors, such as data on power usage and overall demand for electricity. It then utilizes appliance-scheduling software to turn appliances on and off depending on the line-voltage situation. However, using arrow buttons and a 32-character LCD screen, users can enter their own scheduling preferences and override the device’s decisions.
“The major design challenge was to make the plug-in very simple and inexpensive,” she says. Her team came up with a prototype that measures 10 by 10 by 7 centimeters and should cost US $15 if produced in volume.
Ganu and her team tested nPlug units with washing machines and water heaters in two homes, one in Bangalore and another in Chennai. After just one day, the data showed that the line voltage drops during the peak-use period of 6 to 8 p.m., while it is highest in early-morning hours. Ganu designed the nPlug algorithm to work around these hours.
Ganu plans to conduct a bigger pilot study, and will also run computer simulations of potential scenarios in which thousands of households are using nPlugs.
ON A PERSONAL LEVEL
While the nPlug looks at the energy problem from the grid or utility side, Ganu is also working on a device called the Socketwatch that considers the problem of electricity wastage on the user’s side. “The device senses an appliance’s energy consumption and detects its usage pattern,” she explains. Socketwatch can then compare the machine's daily operation with its optimal behavior, and put it in sleep mode, turn it off, or detect unusually high energy consumption due to a malfunction. Appliances do not have to be “smart” for the nPlug to work.
Ganu always had an interest in engineering. Growing up in a small town 400 kilometers south of Mumbai, Ganu was influenced by her father, a mechanical engineer, who had a keen eye for electrical repairs. She would often help him.
In 2004, after getting her bachelor’s degree in computer science from Walchand College of Engineering, a small engineering school close to her hometown, she joined Tata Consultancy Services in Pune, India, as a software engineer, working on an artificial intelligence–powered system that handled online customer complaints and queries. Four years later, she went back to school, pursuing graduate studies at the Indian Institute of Science, in Bangalore.
During those years, she says, “I looked for areas where I could address real societal problems using data insights and technological change.” The Smarter Energy Group at IBM was the perfect fit, she says.