To improve asset management, planning, and grid operation, we test real-time and historical AMR and grid data technologies in real environments.

Automatic meter reading (AMR) will give the power grid companies far more insight in the state of the distribution grid.

Distribution Management Systems (DMS) are expected to play an important role in real time operation and decision support. DMS tools should be able to utilize and manage new data to increase control room and field personnel system situational awareness.


Smart Cable Guard

Probabilistic planning methodology

Development area Molobyen

Elvia is testing a new technology for faster fault location in medium-voltage cables (11-22kV). Sensors detecting partial discharge (PD) make it possible to identify weak points in the grid (e.g. weakened isolation material). This allows for replacing cable segments before actual faults occur. Additionally, the system indicates the fault location right after a fault has occurred, which saves a lot of time in the fault handling process.

The technology’s ability to identify fault locations has been verified several times, with an accuracy of 1%. Identification of weak points in the grid that have not yet caused a fault has not been verified to this point, although the principle was demonstrated earlier in a Dutch study. The reason for this is still unclear and will be investigated.

We will also analyse if the PD-measurements can be used for other purposes, e.g. dynamic line rating (dynamic calculations of the cable’s current carrying capacity), based on the signal speed between the sensors.

ontact person at Elvia:

Improved planning methodologies can increase security of supply and up-time for customers, and decreased cost for grid companies. As AMS is installed throughout the distribution grid, grid companies suddenly have a lot of data with untapped potential. The question is, can we use this data to improve grid planning methodologies?

Together with CINELDI, Norgesnett is testing a new probabilistic planning methodology based on existing AMS data for a part of their network, located in the Øra Industrial area.

If successful, the methodology will allow Norgesnett to do

  • better socio-economic evaluations of grid investments
  • accurate calculations of the KILE-cost for industry customers with unpredictable loads
  • maintenance at the time that's the least intrusive for the customer

Contact personat Norgesnett:

A zero-emission society cannot use an electricity distribution grid design for a high emission world. We must plan, build, and operate the grid in completely new ways. Therefore, CINELDI and FME ZEN (Zero Emission Neighbourhoods) has joined forces in a pilot project together with Nordlandsnett, testing new distribution grid planning methodology. 

It is being tested on a neighbourhood called Molobyen (formerly known as Breivika Vest) currently under development in Bodø. The entrepreneur and the municipality aim at achieving a zero emission neighbourhood, where energy may largely be supplied by solar PVs and district heating (DH).

At this point, CINELDI has extended an analysis of the (thermal) energy system generated by FME ZEN. The energy system analysis showed that using PVs and DH dramatically reduces the maximum need for electricity grid capacity, from 3.93 MW to 0.61 MW.

Moving forward, the implications of this energy system analysis on the design and dimensioning of the electric distribution grid for the area will be investigated. The dimensioning and design of the electric and the thermal energy system are dependent on each other so that different solutions for both should be considered during the design phase.

Contact person at SINTEF Energi:

Contact person Nordlandsnett:


Machine learning in grid inspection

Risk-based distribution network planning

Predicting peak load in secondary substations

Faster and more precise grid inspection can save both time and money. Today it is common to use drones or helicopters to take pictures of the grid, and then analyse the pictures manually to look for (potential) faults.

Together with Agder Energi Nett, we are testing if machine learning (ML) analysis of the pictures can improve the inspection process: Increase efficiency (less resource and time demanding) while improving output quality (faults are identified with acceptable degree of accuracy).

In the pilot, data has been prepared, the ML model (Custom Vision from Microsoft) has been trained on tagged data and the model has been tested with good results.

Contact person Agder Energi Nett:

A risk-based approach to distribution network planning is desirable in the future. That implies making decisions to minimize or maintain the probability that the power demand will exceed the capacity, and to decide if the potential consequences are acceptable or not.

To improve the planning of the network, we are therefore increasing the knowledge about the dynamics and variation in power in the distribution network, together with CINELDI-partner Agder Energi Nett.

Today, planning uses empirical calculation bases for power, i.e. Velanders formula or time-of-use. These methods work reasonably well but can in some cases lead to off-target conclusions.

A stochastic methodology has been investigated and tested. Results this far show that a stochastic approach gives better insight in the nature of power variations and is better suited for modelling the actual load caused by the individual consumers. The model also makes it possible for managers and network planers to make a conscious choice of the level of confidence to be used for dimensioning.

Further, variants of the method will be investigated and compared to Velander/time-of use to check if a stochastic approach can provide improved information for network planning.

Contact person Agder Energi Nett:

Secondary substations can get overloaded in peak load situations. This may lead to need for maintenance, reduced lifetime and outages. Load forecasting can be used to detect overload, and then possibly to take actions to avoid overload or mitigate the consequences. This may be done e.g. by using demand response, operational readiness, repair, or upgrade the secondary substation.

Together with Agder Energi Nett, CINELDI is looking to use machine learning to forecast loads on secondary substations.

In the pilot we have used hourly load from 26 substations and weather data from Norwegian Meteorological Institute as input data. Different machine learning methods were used and compared to baseline methods using MAPE (Mean Absolute Percentage Error).

There was no «one model best in all cases», but LGBM with individual models (i.e. model trained individually for each substation load profile) emerges as the best option. SARIMAX and the naive models performed well in cases where the load is repetitive.

A fully automated machine learning pipeline is in production at Agder Energi Nett. It is currently predicting load for 123 secondary substations, with a model based on XGBoost. The model is retrained regularly.

Contact person Agder Energi Nett:

Smart meters in the Smart Grid Lab

A lab environment for piloting is currently being established in the Smart Grid Laboratory, containing smart meters of Aidon. The combination of real-time simulation, grid emulation and a real smart meter infrastructure gives vast possibilities to develop and test new functionality involving the smart meter infrastructure. So far, tests concerning harmonics have been conducted, but further tests are now being planned as the lab capabilities are expanded.

Contact person SINTEF Energi:
Contact person AIdon: