Getting the Power Grid Ready for the Green Future

Kristoffer Rønne Andersen
Destination AARhus-TechBlog
6 min readMay 25, 2020


With the increasing electrification of society caused by e.g. heat pumps and electric vehicles, the distribution system operators (DSO), which deliver power to our homes, offices, schools, etc. are facing a new challenge. Most of the cables that are dug into the ground were laid at a time, where this was not foreseen and despite often having planned with excess capacity in the cables, the utilities are unsure how things will look in just a few years. To make things even more complex many consumers have become prosumers in recent years due to the installation of photovoltaics (PV). They are now both consuming and producing power. Therefore, DSOs are looking into a future with drastic changes and many do not even have a clear picture of their present situation.

The increased electrification of society

Let us dive a little more into the problem before we look at some solutions. Years ago, the power had a fairly simple and predictable path from production to consumption. Large power plants delivered power to the transmission grid with delivered it to the DSOs who had the contact to the consumers. With the increase of renewable energy from PVs and wind turbines, power is now produced throughout the grid in a largely unpredictable manner determined by for example the weather. This power is often fed into the distribution grid at low voltage levels since they are installed on rooftops and e.g. at farms. On the one hand, PVs produce power mainly in the middle of the day, where people are not home and therefore only have a small power consumption. The produced power, therefore, has to be transported away from residential areas. On the other hand, late in the afternoon where the PV production has lowered and people come home from work, they begin to cook and charge their new electrical vehicles. In this case, we have an increased consumption compared to the earlier. In other words, the increased electrification and the increase in distributed power generation add two new challenges: During the middle of the day, there is a total power production which has to be transported to other areas, and in the evening the situation is reversed. Such a situation is shown in Figure 1. The example is from a substation in a small Danish town, where the power flow changes from peak production to peak consumption within just a few hours.

Figure 1: Example of a substation loading on a regular day in April where the load changes from peak production around 1 in the afternoon to peak consumption around 6. The inverted mid-day peak is a clear sign of PVs. The left graph is the active power and the right graph is the apparent power, which also includes a small reactive component.

This challenge has to be handled in many ways. We have to change regulations and make the right incentives for consuming power at the right time and in the right way. The regulations should also take into account the local power grid, where bottlenecks might occur due to limited cable and transformer capacities. The consumers should in a simple way be able to adapt to the situation, so all the electric vehicles are not all charged at the same time but distributed over a longer period of time. Finally, the DSOs who own the distribution grid should be able to make informed decisions about the present load on their assets and estimate future scenarios. One could mention many other ways of attacking this challenge, but for now, we will dig into the last mentioned.

Solutions for the modern distribution system operator

In many countries electrical smart meters have been installed within the last 10 to 20 years. This gives the DSO detailed knowledge about the consumption of all consumers on a detailed level. This could be on a 15 min, 30 min or an hourly basis. However, many DSOs do not have online measurements from the distribution transformers at the lower voltage levels (in a Danish context, we transform the 10kV or 20kV to 230/400V) and therefore, they need to bridge the information known in real-time at the e.g. 60 kV level transformers and the consumers at the 400V level (400V are phase to phase and 230V are phase to neutral). At Kamstrup we think that data measured with smart meters at the consumers have tremendous potential in this challenge.

Utilities with modern and up-to-date geographical information systems (GIS) have detailed knowledge about their assets. That is how consumers are connected to cables, to cable boxes and so on until you reach the secondary substation with the transformer. Furthermore, many DSOs use a tree grid topology in the low voltage grid. In this case, there is a unique path from each consumer to the substation, which makes it possible to calculate the theoretical load on each asset in the grid-based only on consumption and production measurements at the consumers. A grid example is shown in Figure 2 where the unique path from the substation to a consumer is drawn. Knowing the unique path, we can avoid performing a full state estimation, where currents and voltages are estimated throughout the grid but instead only summarise the measured power consumptions. This problem can mathematically be handled in many different ways. In our setup, we see it as a projection from consumers to the nodes and edges in the distribution grid. The edges are the cables, and the nodes are the consumers, the cable boxes and the substations. Based on the grid topology, we populate a sparse matrix with 1 for all nodes and edges that are on the path from the consumer to the substation. I.e. each column in the matrix corresponds to a consumer meter and each row corresponds to a node or an edge. This makes the calculations very simple and fast once the grid topology is determined.

Figure 2: A part of a distribution grid where the unique path from the consumer to the substation is shown in yellow. The red triangle is the substation, the green dots are cable boxes and the yellow dots are consumers.

Knowing the critical assets in your grid

With this simple mapping onto the distribution grid, the DSO knows their present and historical load for all assets. Knowing that a cable is e.g. a 95 mm2 aluminum cable with a rating of 150 A per phase allows the DSO to quickly find the cables that are loaded above their rating and determine if this has just happened once or it happens repeatedly. They might also find that the transformer is overloaded, but the neighboring transformer has free capacity. In many cases, a situation like that would be solved by laying a new cable and exchanging the transformer to a bigger one. However, knowing that the neighboring transformer has spare capacity, and that some of the load could easily be moved to the other transformer by changing a connection in a cable box saves the DSOs a lot of money, while enables them to use their assets better and limits the downtime of the consumer. An advantage for everyone. A small data example is shown in Figure 3, where cables loaded by more than their capacity are marked. To protect privacy these are not real data.

Figure 3: The distribution grid in a small town where the cables loaded by more than their capacity are shown (the data are not actual data to protect privacy).

Knowing the present

In a time, where the DSOs are looking into an uncertain future, the first step is to know the present situation. Using data from smart meters installed at all customers combined with detailed information about the grid topology and simple projections enables the utility to know this. Thereby, they are better positioned to make reasoned and good decisions that make the future a little less challenging and incomprehensible. Importantly, this will also support the transition to a greener future.

By Kristoffer Rønne Andersen
Data Scientist, Kamstrup

About me

I work at Kamstrup and am responsible for the analyses and the algorithms used in our Power Intelligence solutions. I have a background in physics, and we combine a good physical understanding of the power distribution grid with tools from data analysis, time series forecasting and data visualization to bring insights to the power utilities.



Kristoffer Rønne Andersen
Destination AARhus-TechBlog

PhD physics, Data scientist at Kamstrup