The energy transition depends on decentralization and relies on the increased electrification of heating and mobility. This puts a tremendous strain on distribution grids. So far, only a few local grid companies are tapping into the potential of digitalization for monitoring and stabilizing their grids.
Not everything was better back in the olden days, but the operation of power grids was certainly less complicated. The energy used to flow from large-scale power stations to consumers, whose load profile was more or less predictable. Transmission systems transported the electricity across long distances, and in the distribution grids, the voltage was reduced level by level until reaching the consumer’s home.
Today, many small PV systems feed electricity into the distribution grids while electric cars and heat pumps create peak loads. On top of that, energy management systems and residential storage systems enable consumers to shift their consumption to other times of the day. The majority are programmed to promote consumer independence from the power grid, often influenced by electricity market incentives. In the future, they might supply operating reserves to stabilize the frequency of the higher-level transmission system. As a result of all this, it has become hard to estimate how much electricity flows in which direction when, using only standardized profiles.
A number of studies have shown that flexible loads are certainly useful for the energy transition. Ideally, green power generation should match consumption as closely as possible. However, these increasingly volatile and unpredictable load flows pose quite a challenge for the lower grid levels, i.e. the distribution grids, which are not equipped to handle them. According to the “high automation of low and medium voltage grids” task force established by the Power Engineering Society (VDE ETG) within the Association of German Electrical Engineers (VDE), automation at medium and low-voltage levels is generally limited to pilot projects.
The results were published in a paper with the same title. And although the operating conditions of overarching transformer stations have been measured for decades, little is known about what exactly happens in secondary substations.
Local grid operators mainly base their knowledge on calculations rather than measurement data. Using specialized software for distribution grid simulation has been common practice for years. DIgSILENT’s PowerFactory is one such grid analysis software application. The software shows where grid congestion is likely, whether the grid can handle additional producers and consumers, and whether the grid needs to be upgraded. “PowerFactory also provides dedicated models for electric vehicles, batteries and PV systems that even take solar irradiation and other factors into account,” explains Dr.-Ing. Georg Stöckl from DIgSILENT.
The VDE paper suggests that, moving forward, when bottlenecks are imminent, grid operators should take more active measures. However, the possibilities within the distribution grid are limited. Several grid companies rely on regulated distribution transformers (RDTs) to control the voltage. PowerFactory is capable of displaying these RDTs. Mesh networks have more connections between the different grid cords, ensuring more wiggle room for the energy flow. The software can display this, too, though this function is generally not used during operation, the VDE study shows.
In the distribution grid, active grid operation primarily means: Controllable producers and consumers – such as PV systems with storage systems, electric cars or heat pumps – must adjust or delay their output/consumption on command. Dutch grid operator Alliander has already examined the impact of this in Amsterdam, where transport is expected to be fully electric by 2025. Without load management, one in three roads would have to be dug up to reinforce the power grid, Alliander explains. Instead, the company plans to guarantee only a reduced minimum output for each charging point around the clock.
Furthermore, when additional grid capacity is available, vehicles will only be allowed to charge according to a profile determined by the grid operator. The batteries of the vehicles are the key, because they add more flexibility to the system. “This way, we can connect three times as many charging points to the same power line without overloading our grid or losing charging comfort,” says Roy Crooijmans from the Grid Operation department.
Heat pumps are a case in point that active control does not necessarily reduce the level of comfort: Thermal storage systems ensure that houses stay warm even when grid operators interrupt operation for a few hours. This principle has been applied for many decades to use the electricity from non-curtailable coal-fired and nuclear power plants for storage heaters.
Today, customers in many network areas are already offered a reduced tariff for heat pump electricity. In Germany, the newly added Section 14 a of the Energy Industry Act (EnWG) will regulate when and how grid operators may reduce controllable loads. However, if curtailment occurs frequently, grid companies will be obligated to remove the bottleneck and upgrade their grid.
For real-time monitoring, traditional grid calculations require formatted measurement data to show the current grid status. This measurement data may be supplemented, if needed, by smart algorithms such as status estimations and load scaling. Based on this information, action can be taken to detect grid congestion, for example.
PSInsight aims to manage this and other applications with its GridCal system solution. The young company is a spin-off of the University of Applied Sciences Düsseldorf, and its actual aim was to develop software for data analysis in distribution grids. “But we soon realized that we did not have enough data from the local grids to do this. Once we found partners for the hardware, we had to figure out how to install it,” remembers Philipp Huppertz about PSInsight’s early days, when the company struggled with a shortage of skilled workers.
Various companies have now come together to form the GridCal Alliance, so they could jointly provide an all-in-one solution for distribution grid digitalization. The alliance includes infrastructure service provider Omexom and specialist manufacturer for measurement technology, PQ Plus.
To prevent dependence on a proprietary solution (vendor lock-in), the GridCal Alliance only uses standardized components that have been tried and tested across the industry. Since the equipment is mostly installed in decades-old secondary substations, PQ Plus has gone to great lengths to make it compatible, compact and easy to install.
For operators of critical infrastructure, security is paramount. When it comes to traditional grid calculations, it is still relatively easy to ensure security, because they run on centralized computers in the control center. But with real-time monitoring a lot of information has to be processed continuously. To reduce the data sprawl and thus vulnerability, as much of the data processing using GridCal as possible is done on site at the secondary substations. “We don’t bring the data to the algorithm, we bring the algorithm to the data,” explains Huppertz.
Not all raw data is sent to the control center, but only what is absolutely necessary and proactively retrieved. PSInsight relies on edge computing on site at the stations rather than external clouds. It is not only about preventing cyberattacks, but about ensuring that nobody else can govern the data. “If you process your data in the cloud, you depend on another company’s proprietary infrastructure,” Huppertz warns.
While most experts agree that digitalization, artificial intelligence and automation are essential to the energy transition, these technologies are not an end in themselves, the VDE study emphasizes. The goal of building and maintaining a robust power grid must always be paramount. While data is important, not all of it must be available everywhere at all times and updated every second. Other data may be needed in the future, but not today. For this reason, the VDE study cautions against excessive automation overhead.
The more complex a system becomes, the more important it is to remember this: Keep it simple, stupid!
