Efficient processes for a sustainable energy, heating, and mobility transition: SWM relies on Azure IoT, artificial intelligence, and big data analyses

The challenge: Connecting large amounts of data from different sources

Ensuring the city has an entirely carbon-neutral supply of district heating when needed, having plants capable of producing as much green electricity as the city consumes, and providing a public transport fleet of 100 percent electrified vehicles—this sounds like a vision, but in Munich, this plan is already being implemented one step at a time. Stadtwerke München (SWM) already generates enough green electricity at its own plants to cover around 90 percent of Munich’s energy needs. Up to 80 percent of the local public transport fleet currently runs only on electricity. This is because SWM believes that, as a municipal energy provider, it has a responsibility to play an active role in shaping the energy transition and help the city achieve its climate targets in the areas of energy, heating, and mobility.

What are required are maximum-efficiency processes, such as predictive infrastructure maintenance and optimized operations planning. For example, during peak times there are up to 500 buses on the streets of Munich. Each of the fleet’s electric buses must be relied upon to manage at least 280 kilometers on a single charge—whatever the weather. Since charging takes three to four hours depending on the size of bus, this can be done only in the gap between closing down and restarting service. So it’s imperative that operations management run like clockwork. “Ensuring that such complex processes run smoothly calls for data-driven IT solutions based on AI,” says Martin Klaus, IoT and Data Lab Team Leader at SWM. “Automated processes, smart applications, and transparency are built on data and connectivity.”

And Munich has plenty of data to offer: each subway train alone sends values for over 2,000 data points to SWM—and does so every 10 seconds. Geothermic solutions collect around five gigabytes of data every hour. These are just two of about 30 digitized areas of infrastructure. This presents considerable challenges for the IT architecture: “We need a scalable solution with services specifically designed to process and analyze big data,” Klaus says. “At the same time, we wanted to establish a data platform that brings together the various data sources and formats, makes these understandable at a glance, and is set up to handle future integrations. With their managed services, Microsoft Azure, Azure IoT, and Azure IoT Hub really fit the bill. In combination with artificial intelligence, we have the predictive processes we require.”

The solution: Microsoft Azure makes bus charging more efficient, and energy use more sustainable

Today, the first step toward automation and optimization is being taken in procurement. Take an electric bus, for instance: each one SWM buys is fitted with a manufacturer-agnostic IoT edge device, which collects technical data—including battery charge, engine speed, and activation of warning lights—in real time in accordance with all relevant regulations. The data provided by the 120 or so sensors is then sent to the INSIGHT platform, which is based on Microsoft Azure. The result is a digital twin of each bus. “As a result, SWM depots always know the condition of each bus, which means they can resolve malfunctions faster, optimize repair processes, and practice predictive maintenance,” says Dr. Denis Bytschkow, Software Engineer at SWM. “This makes the processes for scheduling and maintaining the vehicles more efficient.”

SWM has also found that insights drawn from historical and current data can help optimize other operational processes as well. For instance, it can use data on battery charge levels to calculate the actual energy consumption and then store this information in Azure Data Lake. The data is then combined with data from other sources—such as weather forecasts, bus master data, and timetables—in Azure Databricks. “Next, we use an AI model based on Azure Machine Learning to determine how much power a bus consumes on a given day, at a given time, on a given route, and during a given weather situation,” says Sarah Frank, Machine Learning Engineer at SWM. “This lets us compile a forecast 14 days in advance for how much power each bus on each route will need.”

The depot management system uses this forecast to compile a charging schedule that it forwards to the charging load management system, which in turn informs the charging infrastructure of the buses’ charging needs by way of a series of prioritized values. All this makes it possible to use the fleet of buses more efficiently and sustainably. “Batteries are charged only to the level required to complete the route in question. This reduces unnecessary trips to the charging station, protects the batteries, and ensures more sustainable use of the energy available,” Frank says. “Buses thus spend more time on the road and can run at an increased frequency. So in the end, we need fewer buses to serve the same route.”

Since 2019, SWM has digitized more than 30 infrastructure areas—and that’s only the beginning. “We can easily transfer the knowledge we gain from one use case to another. This means that we can now tap the potential for optimization and digital transformation in less time, for less money, and with less fuss than before,” Klaus says. 

“Thanks to Azure IoT and the Azure cloud platform with their big data and AI services, we can offer the departments end-to-end support and deliver extremely quickly. This frees up capacity for new ideas and innovations, especially when it comes to sustainability,” Bytschkow adds. “The scalable infrastructure of Azure lets us design our architecture, not according to peak loads, but flexibly to match our actual demand. In the spirit of sustainability, we use only the power that we actually need.”

In this way, the INSIGHT IoT and AI solution developed by Klaus and his team helps SWM to do more than digitally map infrastructure processes and assets, establish data transparency, and optimize processes. It is also playing a major role in Munich’s transformation into a connected and sustainable smart city.