Grid4C Uses AI To Optimize Energy Consumption
Grid4C founder and CEO Dr. Noa Ruschin-Rimini talks very fast, and also acts at the same pace. Just over four years ago, she had no special expertise in energy. In the past two years, however, she has repeatedly been gaining recognition as a pioneer in the sector. In February this year, she was on the cover of Energy CIO Insights, a magazine that has mapped dozens of the leading big data solutions in the energy industry. In January 2016, GTM Research described Grid4C as a global leader in the forecasting and analytics of energy consumption – ahead of companies like General Electric, IBM, and Oracle. Last year, her company was also recognized as a leader in the Internet of Things (IoT), and Ruschin-Rimini herself was awarded two prices for excellence in entrepreneurship.
Grid4C provides software products that make it possible to predict and optimize a smart electricity grid for both electricity producers and suppliers and home customers. The company’s development center is in Herzliya, while Ruschin-Rimini herself moved to Austin, Texas with her family 18 months ago. She says that Texas is one of the places where the future of the global electricity industry is being shaped. Other places are New York and Australia, for example. “Here, it’s the Wild West, a jungle,” she says. “There’s a lot of electricity suppliers here fighting over customers in a completely free market, like in the telecom sector. The cost of switching between suppliers is zero. People switch all the time. There’s 40% churn a year.”
Mathematical proofs before bedtime
Like many other high-tech entrepreneurs, Ruschin-Rimini, who was born in 1973, began in IDF Intelligence unit 8200. In 1993, she began a direct track for an MSc in industrial engineering, with a major in information systems, at the Tel Aviv University engineering faculty, where her father, Prof. Shlomo Ruschin, is a faculty member. She completed a doctorate in artificial intelligence (AI) in 2008-2012. “In my doctorate, I specialized in everything having to do with forecasting and detection of anomalies in big data – what is called time series data,” Ruschin-Rimini says. “Before that, I accumulated more than a decade of experience in software development, business development, and sales at large companies, such as Oracle and IBM, and in startups.
“I have shared my father’s love for physics and mathematics since I was a child. In answer to every scientific question I asked, I got a demonstration or a physics experiment. The light and objects on the dresser in my bedroom were the solar system. My father and I traveled to an astronomy observatory whenever there was an opportunity, and read from a book of mathematical proofs. I even remember that I decided to discover the law of prime numbers, and that’s what I do before I go to sleep.”
Ruschin-Rimini divided her first childhood years between Haifa, where she was born, and Ithaca in New York, where her family spent two years when her father was doing a post-doctorate at Cornell University. “When I arrived there at age five, I didn’t know a word of English, of course, and for six months, I didn’t utter a word. When I finally started talking again, I didn’t stop. They say I’ve been compensating for it ever since,” she says. When she was nine, the family moved to Herzliya, and in high school, she again spent a year in the US. Ruschin-Rimini believes that the many moves and the need to get used to a new place each time have shaped her, for better or for worse. “The good part is that I like changes, like starting over, reinventing myself, and I can’t stand a comfort zone. The part that’s not so good is that I’m unable to maintain connections with people in the long term.”
She met her husband, former Taldor Computer Systems (1986) Ltd. (TASE: TALD) CEO Arie Rimini, when she was business development manager at a startup and he was CFO in a Bezeq Israeli Telecommunication Co. Ltd. (TASE: BEZQ) subsidiary. He was divorced with two daughters and 10 years older than her. She says it was “love at first sight that has only grown stronger constantly.” They have a 12 year-old son and a 10 year-old daughter. “After maternity leave with my daughter, Neta, I decided that it was time to realize another dream by doing a PhD in engineering. Then I met a good friend who studied with me during my BSc and MSc, Prof. Lior Rokach from Ben Gurion University of the Negev, one of the world’s leading experts in artificial intelligence (AI). It was 2007, and he told me that AI was the future. I liked the idea, went back to my mathematics notebooks from my BSc, checked out books from the university library, started studying the subject, fell in love, and starting doing research.
“After my PhD, it was clear to me that it was my turn to found a startup. My two PhD advisors, Prof. Oded Maimon and Prof. Irad Ben-Gal, both had successful AI startups. The question was in which industry and what problem we would solve. Completely by coincidence, I got a call from an IT solutions consultant company that gives advice about information systems to all the private power stations in Israel. They knew me from when I worked at Oracle, and in late 2012, they contacted me and told me, ‘Listen, everyone here has the same problem, and we know that you did a doctorate in this field. Maybe you can help us.’ They said that they needed a very accurate consumption forecast 24 hours and a week ahead at 30-minute resolution that could be given to the systems manager, and every overestimation or underestimation, every deviation in either direction, was penalized very heavily.
“I said, ‘Great, what’s the input? What data do I get for this model?’ Then they told me that there was a history of 15-minute readings going back years of thousands of meters in the industrial-commercial sector. I said, ‘Wait a second, there are so many granular data at the meter level, and they’re giving only a forecast for the total? Let’s build a model that that will predict consumption and anomalies at the level of every meter for thousands and millions of customers.’ It was clear to me that there was a great deal of added value, and I went to study the subject and understand it. I realized that the electricity market was being revolutionized, and that data from sensors and smart meters around the world contained the seeds of solutions to many challenges facing this market.”
A year later, Ruschin-Rimini developed an algorithm and an initial product. In later 2013, she raised $1.5 million from the ICV venture capital fund, which specializes in energy, and started recruiting employees. “Fortunately, Eitan Meir, an inspiring man of vision and CEO of Dalia Power Energies, the largest private electricity producer in Israel, and the other board members believed in the product and the vision, and bought our solutions in early 2014. Grid4C grew and developed, and the company now analyzes billions of sensor and meter readings from countries on four continents, and produces more than 500 million forecasts daily for its customers in less than an hour. Our customers include the major electricity producers in the US, Australia, Europe, and Israel.”
Why don’t the electricity producers in Israel settle for the solutions used by other producers around the world?
Ruschin-Rimini: “Everyone does forecasts for the total, but no one has reached a situation in which he has daily forecasts for all of his customers’ smart meters. With our customers, if there’s a difference between the forecast and what actually happens, they know right away who the customer is, and can call him. Otherwise, a problem can last for weeks without anyone even knowing about it, and that also costs the customer a lot of money. On the other hand, it can be a shutdown about which the customers should have issued a warning in advance, and didn’t do so. In extreme cases, there could be days on which tens of thousands of shekels are saved just by understanding what’s going on.”
“Texas is in the forefront “
The ability to gather data and insights about the customers themselves brought Ruschin-Rimini to Texas. She set up the company’s marketing and sales offices there in surroundings of little regulation – a test field for the future of the energy industry. Up until now, she was almost alone in the US, while an absolute majority of the company’s 20 employees were in its Herzliya development center. By 2018, however, the number of employees is projected to double as a result of large-scale recruiting of sales and marketing personnel in the US. Another doubling of the company staff is slated for 2018.
“The reason why I’m in Texas is that the forefront is here. This is an exciting market here, with many players, and many of our customers are in Houston and Dallas. On paper, there are 106 electricity producers in Texas, but it changes every day, because the large ones are buying out the small ones in order to increase their market share. They’re swallowing each other. There’s a real war here for the customer’s allegiance, and you feel it every day. It’s astonishing how good this is for everyone: the customer, the consumer, the supplier, and the environment. Prices for customers here are half what they are in areas of the US where there’s no competition.”
In Texas, Ruschin-Rimini launched very large-scale cooperation with Direct Energy, one of the state’s three largest electricity suppliers. The US company uses Grid4C’s products to supply 1.5 million of its customers with value added services. “They have a very famous project called Direct Your Energy,” Ruschin-Rimini says. “You travel around here, and you see the screens of our products on billboards and in their television ads. It’s leading their campaign.”
In Israel, Grid4C serves the Dalia Power Energies power plan and the Israel Electric Corporation (IEC) (TASE: ELEC.B22) smart city pilot in Binyamina. In the international arena, in addition to several deals now being worked out, the company has made a deal with the connected homes division of UK company Centrica, the parent company of British Gas, which needed AI technologies.
Landis Gyr, a leading manufacturer of smart meters, which has a 20% share of the global market, also selected the Israeli company as its partner, and has begun selling Grid4C’s products in Australia and Asia, while Grid4C is benefiting from access to the enormous quantities of information accumulated by Landis Gyr.
“What we have developed, and which is setting the market on fire, is the ability of our algorithms to break down home consumption into the main devices operated in a home according to signals. We’re able to detect and diagnose malfunctions in the refrigerator, air conditioners, water heaters, washing machines, and dryers – all without any sensors in the home. This is no longer just energy efficiency. There are many cases in which we predicted in May that an air conditioner would collapse and be unable in June to reach the temperature set by the thermostat.
“We can predict how a home should run, and see that there’s a difference between what should happen and what actually happened – an anomaly that we call a behavioral irregularity, for example resulting from devices that weren’t turned off, or were improperly set, and we warn the customer. We have all sorts of algorithms scanning the data and looking for finger prints and signatures of all sorts of malfunctions that we already know how to detect, because we trained our algorithms. We have huge quantities of data, because we’re serving more than 1.5 million homes in the US.”
In order to train the algorithms, Grid4C engineers talk with technicians for air conditioners, refrigerators, and other products, and associate various malfunctions with anomalies that the system detects. Since there are anomalies that can indicate more than one kind of malfunction, the company has developed Bots (AI-operated chats) that come up on the customers’ telephone or computer screens and question them in order to arrive at an accurate diagnosis of the sources of the problem.
The request for value added services for electricity consumers is derived, among other things, from the intense competition between suppliers. “You have to realize that in markets that have been completely deregulated, the suppliers’ profit margins are not large,” Ruschin-Rimini explains. “If you ask them how much profit they make on a home customer in Texas, they’ll tell you that the average is no more than $10-25 a year. Since they can’t cut prices any more, they look for companies like us, because they realize that this is the way they can keep the customer – by giving added value and creating new revenue sources for them through the sale of additional services and products.”
But where’s the money?
“Direct Energy, for example, acquired a home repair company. As soon as they detect a malfunction or an inefficient home appliance, they send the customer a link. The customer clicks on it, and the technician goes there to repair or replace the appliance. These guys also sell, for example, smart thermoses – very efficient water heaters. How are they able to offer these devices to the right customers? By using our algorithms.
“Even if the electricity producers have no repair services to offer, they can sell a lead to such repairs to someone else for $40 or $60. Assume that we have spotted an insulation problem in a home, or a need for a repair or replacement of an electrical appliance. According to what we’re seeing, the customers are very satisfied at being offered this, because they repaired a problem they had at home that greatly inflated their electricity bill. That’s the reason why I decided to found a startup in energy. Everyone makes a profit, and you’re not pushing a product he doesn’t need; you’re offering him something that saves him energy and saves on the entire supply chain. It’s a completely win-win situation.”
Google suddenly became interested in energy
“I had the honor of meeting the president of Direct Energy, who told me, ‘We understand now that we’re no longer energy companies; we’re data companies. We buy and sell electricity, but first of all, we have data.’ That really shows the digitalization process taking place in this market,” Ruschin-Rimini says in pointing out the direction.
“They see that competition is starting to come from Apple Computers, for example (which founded a subsidiary for selling electricity from solar arrays that they own in California and Nevada, N.Y.), and from Comcast (a cable company that is also considering entering the sector, N.Y.). All sorts of new players are trying to enter this field, for one simple reason – this entire sphere of smart appliances in the smart home is giving them all the information coming from within the home, with which innumerable things can be done.”
This raises the question of privacy. It is disturbing to think that even in the home, we are being monitored.
“There are countries in Europe in which you have to allow access to information on your smart meter. What’s good about it, though is that in contrast to any other sphere in which your privacy is invaded, your clicks are followed, and advertising is thrown at you, here people feel that you’re concerned about them. In our case, in the case of energy, and again, that’s what’s good about it, you’re using this for a good purpose – good for the home consumer, good for the electricity supplier, and good for the environment and a sustainable future.
“For us, the smart meter data are like putting a camera inside the home. With our algorithms, we know everything that goes on inside the home, when you got home, and when you left for work. This information, like information coming from other smart devices, is worth its weight in gold to all the companies, and so you’re suddenly seeing companies like Google buying (smart thermostat manufacturer) Nest, and companies like Apple and IT&T suddenly talking about energy. There’s a war over who will receive the data from the smart home.”
In concern about privacy, Ruschin-Rimini says, people in the US are less worried about an electricity meter outside the home, but are unwilling to install sensors within their home, for example. “The way to bypass this, of course, is to put smart devices inside the home, and then you get more information. Google is the only company – and this is why we, as a startup, don’t like Google and don’t like Nest – that is unwilling to transfer the information gathered from its smart thermostats to the electric companies serving the customer, which means that most of the advantages I mentioned, such as predicting malfunctions and the effect of demand-response events for the purpose of reducing consumption at peak times, are not realized. The competitors are pursuing an open policy for the benefit of the home consumer and electricity savings.”
Electricity as a source of additional revenue for the consumer
The Burger King fast food chain created a storm in the US in early April by broadcasting a television ad that made Google Home and the company’s Android devices start operating in many homes. During the ad, a seller at Burger King said, “OK, Google, what is the Whopper Burger?”, causing the company’s devices to start reading the definition of the chain’s hamburger from Wikipedia. A counter-response followed, in which sentences were inserted into Wikipedia saying that the chain’s hamburger was carcinogenic and made out of fingernails and rats. The tumult culminated in the temporary blocking of the editing value.
“We’re trying to take this in a good direction,” says Ruschin-Rimini. “For example, the vision we’re working on with our customers is that someone knocks on your door with the right part, the right replacement parts, before you know that you have a malfunction at home.
“Another interesting revolution is that electricity consumers are starting to become prosumers (producers-consumers). As a result of solar panels, we’re not only consuming electricity; we’re also generating it, and sending it onto the grid, and are receiving money for it. Furthermore, as a result of home panels, such as Tesla’s, that enable us to store electricity for when it is needed, and as a result of the programs for the home and business consumer for reducing electricity consumption at peak times, as consumers, we have the potential for making money, if we’re smart enough to know when to use the electricity we have generated, when it’s worthwhile storing it, when it’s worthwhile sending it to the grid, and when it’s worthwhile reducing consumption. In other words, home consumers can make money out of proper management of electricity, and electricity can become another source of income.”
Beyond home electricity efficiency, the smart meters indicate a change in the model for handling peak consumption times. While peaker power plants are now being built designed to operate only 600-1,000 hours a year at very high cost, Ruschin-Rimini is describing a situation already being implemented on a large scale in the US that is likely to make the existing model superfluous.
“A sentence we use a lot in our industry is that AI is actually creating a new source of energy, because management of demand-response events is unquestionably a source of energy. Instead of operating another power station, which will usually be polluting and expensive, you can cut down consumption. In order for demand-response events to be considered an energy source for all intents and purposes, you have to be able to predict accurately how much electricity will be reduced in such an event, and that’s where Grid4C enters the picture.
“The recompense received by customers for demand-response events is for the difference between their projected consumption according to Grid4C’s software and their actual consumption. I’ve been at conferences in which people got on the stage and told how much money they were making every month from electricity. They said, ‘I make over $100 a year, several hundred dollars a year, from participating in demand-response events.’ The vision is that eventually, every home will be in effect an energy producer and an energy seller, and will maximize its profits.”