The text was first published on the AWS website. Find it here.
Home energy management company NET2GRID supports renewable energy use by offering an energy monitoring solution that utility companies can provide consumers for home use, but scaling to meet demand proved a challenge. The company’s customer base grew by 10 times in under 2 years.
NET2GRID used Amazon Web Services (AWS) to design, develop, and implement a new serverless architecture, reducing operating costs by 400–500 percent and increasing scalability to enable widespread access to energy metering. Through its new architecture on AWS, NET2GRID can offer a more cost-effective, commercially viable solution to utilities, which in turn can offer real-time home energy management to consumers.
“It’s really about helping consumers change the way they use energy,” says Bert Lutje Berenbroek, CEO of NET2GRID. “To achieve that huge behavioral change, utilities need to do a better job of engaging energy users. And that’s where we help energy companies—by empowering them with the real-time monitoring capabilities they need to move their business from selling commodities to selling services.”
Using Data Science and AWS to Solve a Renewable Energy Challenge
NET2GRID was founded in 2011 to offer added-value services to utilities across Western Europe, Australia, and the United States by analyzing real-time and smart meter data on energy usage. The company’s goal is to support energy providers in driving the transition to renewable energy sources, which are greener and more cost-efficient than carbon-based energy sources. The challenge is that scalable and affordable renewable energy sources have been slow to catch on, and utility providers need ways to educate customers about their options and accurately track their energy usage habits and preferences. NET2GRID uses data science to solve this challenge by helping utilities gain consumer energy insights from smart meter data, forecast future demand, and increase consumer awareness of their home energy usage. Initially, NET2GRID’s services were cloud-agnostic and used containers, relational databases, and Amazon Elastic Compute Cloud (Amazon EC2) instances to stream, store, and analyze smart meter data. Amazon EC2 is a web service that provides secure, resizable compute capacity in the cloud. But NET2GRID needed a solution that would be scalable and cost-effective enough to cover a mass-market business case. The company opted to develop a new solution on AWS that would enable it to go serverless—achieving low latency at scale, reducing infrastructure management and costs, and providing access to real-time data.
NET2GRID decided to implement a serverless architecture using AWS Lambda, a serverless compute service. The company also opted to move from its relational databases to flat storage on Amazon Simple Storage Service (Amazon S3), an object storage service, and to query the data using Amazon Athena, an interactive query service that makes it simple and cost-effective to analyze data in Amazon S3. NET2GRID also chose Amazon Kinesis Data Firehose, a service for loading streaming data into data lakes, data stores, and analytics services. “Amazon S3, working alongside Amazon Kinesis Data Firehose and Amazon Athena, basically acts like a more efficient database,” says Peter Broekroelofs, chief technology officer of NET2GRID. “This provides tremendous value for us in terms of cost reduction and scalability.”
Bringing Energy Data to the Cloud Using Utility Meter Data Analytics on AWS
Along with implementing its serverless solution on AWS, NET2GRID adds value for customers using Utility Meter Data Analytics on AWS, an AWS Quick Start that uses machine learning to enable utility companies to analyze data from smart utility meters and gain insights into customer energy usage—insights that are helpful for marketing and forecasting. “Utility Meter Data Analytics on AWS helps utilities bring data to the cloud so that they can actually monetize it,” says Lutje Berenbroek. “We are interfacing our own algorithms on top of that solution to provide an integrated service that directly delivers value for utilities.” NET2GRID offers three types of data science algorithms to support three use cases that have a radically different granularity of available consumption data. First, it applies general machine learning models that recognize appliance categories in household energy consumption data, derived from smart meters, without the need for additional hardware in the home. Next, machine learning models that have been trained for each household using real-time data are applied to the disaggregate energy consumption of individual appliances, achieving over 90 percent accuracy. The third algorithm uses pretrained edge models to identify appliance events with 98 percent accuracy within 15 seconds. These models process 100 Hz data and run on the latest-generation smart meters. NET2GRID also benchmarks the quality of its models with real-time disaggregation, inputting over 50,000 ground-truth appliance events.
NET2GRID enables energy companies and consumers to access more accurate data on their energy usage in real-time and make adjustments to maintain energy availability. As a result, its utility company customers are better able to help homeowners reduce operational expenses and get the most out of their capital investment in home energy solutions. Energy providers are also better equipped to market to their customers, who can opt-in and give the mandate for utilities to use their data. That enables the companies to customize offers for individual consumers—such as those who use electric car charging stations and solar panels. Finally, this data can be used for electricity load forecasting to help utilities predict future energy consumption. “Utilities in the Netherlands lose €12 million per year because they’re not accurately predicting the next day’s consumption,” says Lutje Berenbroek. “With our artificial intelligence tools and models, we can predict next-day energy consumption with 98 percent accuracy. That can save millions for utilities, and with our service, energy providers could receive a return on investment of more than three times their costs.”
Through its newly architected solution on AWS, NET2GRID has seen gains in customer satisfaction, with its customers achieving a referral rate of over 85 percent. Most utility monitoring apps get customer ratings of no more than two out of five stars, but NET2GRID’s white label app gets four stars. Of German customers, 90 percent visited the app at least once per week and 50 percent visited it daily. “Customers are not used to getting these kinds of services from their utilities,” says Lutje Berenbroek.
Laying the Groundwork for Smarter Energy Choices
Now that it has realized the benefits of going serverless on AWS—including low latency, access to real-time information, and reduced costs for itself and its customers—NET2GRID is working toward a future that runs on renewable energy solutions. “The transition to renewable energy is a global need,” says Broekroelofs. “On AWS, we are equipped to empower growing numbers of utilities and their customers to make smarter choices that will lead to cleaner, greener energy use for a
Comments