• Print
  • Decrease text size
  • Reset text size
  • Larger text size

Distribution grid planning for a sustainable future

The dynamics in the distribution grid are about to change due to technological developments such as solar photovoltaic (PV) systems, electric vehicles (EVs) with potential vehicle to grid (V2G) functionalities, residential battery storage techniques, and the electrification of heat demand. Distribution system operators (DSOs) face the challenging task to facilitate the required infrastructure for the energy transition in the most economically efficient manner. Grid planning is key, since it is expected that the current grid lacks the capacity to deal with the changing patterns of supply and demand. In order to make timely and well-informed investment decisions, a way to assess the impact of future technologies and trends on the load on the distribution grid is required. This article presents a method based on our experience gained at Enexis, a large DSO in The Netherlands. In five steps, high level scenarios for the energy transition are translated to insight into when and where system bottlenecks may occur (see figure 1). The purpose and implementation of each step will now be explained. 

Figure 1: The five-step methodology for assessing the impact of energy transition scenarios

Scenario generation

Various market parties, policy makers and scientists have a rough idea about the future energy system and the technologies being used. However, the speed and extent to which technology is adopted remains still highly uncertain. This uncertainty in how the different technologies might materialize could be addressed by making use of scenario analysis.

The adoption of energy technologies is influenced by national and international factors, such as policy measures, the availability of resources and the development of the technology itself. By making different assumptions about these factors, different high-level scenarios can be created. However, various local (sociodemographic) factors determine that the adopted technologies differ from one place to another. For example, the adoption of PV in urban areas differs from rural areas. It is therefore important to ‘translate’ national scenarios into local scenarios using a distribution model. The three main inputs required for this model are:

  • A national scenario describing the adoption per technology over time
  • A set of local influence factors per technology, e.g. about the composition of the built environment and the local population.
  • A set of parameters that describe how these factors influence technology adoption, e.g. a 40% increase in average income doubles the adoption probability for PV.

By making use of historical data and a basic linear regression model, both the influencing local factors can be determined, as well as the parameters that describe their effects on technology adoption.

Bottom-up profile generation and aggregation 

Once the effects of high-level future developments are translated into the level of the low voltage grid, we know where, when, and to what extent the energy technologies could potentially materialize.  Yet on their own they do not tell us the shape of load patterns for individual grid points in the distribution system, which are required to determine the load on the cables and transformers higher up in the grid. The next step in our methodology is therefore load profile generation and aggregation on low (LV) and medium voltage (MV) grid levels.

Load profile aggregation

The load pattern of an individual connection to the grid, a house for example, is created by aggregating the load patterns of all adopted energy technologies and the base consumption pattern as described by the standard load profiles or by making use of smart meter data. Regarding the former, the load patterns of each energy technology differ, partly because they are dependent on different external factors. Therefore a separate load pattern model has to be created per technology. To get to the load pattern of a PV system for example the solar irradiation is required amongst others, which follows an annual cycle. To get to the load pattern of an electric vehicle the average moment of charging is required, which is more likely to follow a weekly cycle.

Detail vs. runtime

An important choice to consider for each model is the granularity or resolution with which the load profiles are to be created. High resolution (e.g. 15-minute values) increases detail, revealing short-term peak loads that could cause outages, which would remain invisible when selecting lower resolution (e.g. daily values). However, more detail comes with a higher simulation runtime. Selecting daily values over 15-minute values decreases the runtime with nearly a factor 100. Finally the load profiles of individual connections can be aggregated per substation, while taking into account whether these loads actually occur simultaneously and assuming a radial grid structure.

Load flow simulation

The next step of the process is the calculation of the load flows in the grid. In its simplest form, the load on a cable or grid station is merely an addition of the load on subordinate components. However, more complex simulation techniques include the laws of nature on electricity, taking into account Ohm’s law describing the relation between resistance, current and voltage and Kirchoff’s circuit laws. Modern load flow calculation packages are generally capable of calculating the load on grid components during both normal and failure situations. In addition, they provide insight in voltage quality, grid losses and various safety parameters. Whether these additional KPIs should be part of the analysis surely depend on the scope of the project, but may elicit grid investments as well. A detailed load flow simulation requires the load profiles, the topology of the grid, and asset data on components such as cables, connection sleeves and transformers as main inputs.

Analysis and visualization 

The data resulting from the load flow simulation has to be visualized and analyzed in order to create valuable input for investment planning. Ideally the data is accessible through an interactive tool, allowing the user to explore the data based on his specific needs. By comparing the nominal capacity of grid components to the expected load, system bottlenecks such as overloaded cables and transformers can be identified. Displaying the impact of the different scenarios next to each other, it becomes possible to make comparisons and determine which future scenarios pose the biggest problem.

Next to knowing where and when system bottlenecks are likely to pop up, insight in the action and associated costs to remove the bottlenecks is required too. In traditional grid planning, the corrective action entails no more than replacing the overloaded component by one with a larger capacity. However, recently more options to deal with grid congestion are coming up, such as the use of batteries and demand-side flexibility. Determining what is the best corrective action to take given the new options has now become a complex task and requires a decision model on its own. A framework for this model has been presented on this blog last January [1].

Implementation challenges 

The method presented in this article allows DSOs to perform grid planning activities in a rapidly changing environment. While the individual steps may seem straightforward, there are a number of aspects that could determine the success of the project:

  • As the analysis is highly data-intensive, data availability, data accessibility, and data quality should be assessed in small scale proof-of-concepts before scaling up.
  • The number of calculations and the data streams involved in the generation of load profiles and load flow simulation for a complete distribution grid with its millions of connections and components is very high and sets requirements for the IT infrastructure used.  The use of a scalable cloud platform may be necessary here.
  • A lot of detail could be put in the models that describe the power consumption and generation patterns that emerge from the use of the aforementioned technologies. To quickly come to usable results, it is key to make assumptions early and work iteratively, improving models over time.
  • With the digitization of the distribution grid, more and more sensor data becomes available. Including this data in the profile generation may significantly improve output quality.

For any DSO, scenario analysis in order to facilitate its own ‘energy transition process’ is of strategic importance. Learning how renewable energy technologies may impact grid load provides valuable insights, required to perform next generation Asset Management. To our view, it will become a DSO core competency as the energy transition kicks off.    

About the authors 

Ruben Moorlag - Consultant Energy & Utilities 

Ruben likes to solve complex problems in the field of energy and infrastructure. In the past years he completed challenging projects at various energy suppliers in the field of e.g. forecasting, process optimization and market entry, His current focus is the development and implementation of a scenario analysis tool at Dutch distribution grid operator Enexis, with the aim to support the transition to a renewable energy future.

Robin Brouwer - Jr. Consultant Energy & Utilities

Robin gained experience through various projects in the energy industry at the Eneco Group, as well as internships at a production company and shipyard in the Qatar Ras Laffan. During his most recent project, he became even more passionate about solving tomorrow’s energy challenges. While researching the value of aggregated flexibility services provided by residential batteries.


[1] http://energy.sia-partners.com/20180319/how-determine-business-case-behi...

0 comment
Post a comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Enter the characters shown in the image.
Back to Top