1. Abstract Representation of Power System Networks as a Function of Regularity Properties ESREL 2014 A.B. Svendsen, T. Tollefsen, R.F. Pedersen & K.P. Petursson Goodtech Power, Bergen, Norway D. Patel, O.D. Lampe Christian Michelsen Research AS, Bergen, Norway

2. Content Part 1: Online risk simulator in operation at Statnett Part 2: Visualization

3. Part 1: Promaps Online Probability method applied to power system, Online calculations

4. Introduction In 2013 the first online risk calculator was put in operation for Norway’s Transmission system operator, Statnett. The model repressing the Norwegian power system, includes a part of the Swedish and Finland power system. Model size: >7000 branches, > 3000 nodes and data update every 10 minutes. Promaps Online gives new insight regarding risk in power system, that is updated every 10 minutes With all this new data, new ways for representing the power systems property regarding the changing risk, is needed.

5. Background It started with a storm called Narve hitting Mid- and Northern Norway in 2006 A need presented itself; to be able to calculate the risk level during operation and for the following hours… – To be aware of the current situation – Planning of measures in areas that cannot handle the consequence of power outage

6. The scope of work In 2009 Statnett R&D started collaborating with Troll Power (now Goodtech) on developing an online risk simulator of power systems The requirements were to be able to calculate: – The risk level in the entire Norwegian power system every 5 minutes – Receive power system data from EMS Scada – Present the risk level with a few colour parameters along with a dynamic colour indicating risk – «a glance at the screen »

7. Status Promaps Norway model

8. Norway model

9. Regions

10. Sub regions

11. Example 1- Analysis The night before October 17th, analysis. We will now study a area of usage for Promaps. In the next figure we have plotted the position on the Contingency list for the line 420KLABU-NEA in Trøndelag. Value 1 equals first place on the list, etc. We see that on October 16th, around 10 pm the load increases from approximately 200 MW to 500 MW, and the line 420KLABU- NEA enters the top 100 list of worst System Continengcies in Norway. When the load is significantly increased, the line 420KLABU-NEA moves to the top of the list (i.e. The component (line/transformer) giving the largest lack of energy if it gets disconnected) When the load decreases at 0600 a.m. on October 17th, 420KLABU- NEA is no longer on the top 100 list.

12. Situation

13. Example 1

14. Example 2

15. Sum load for entire Oslo drops with approximately 100 MW between 0622 a.m. and 0632 a.m.

17. It became obvious that it was needed a better way to get system insight, fast. We need to understand underlying property of the power systems, which are changing every 10 minutes. In the current way it is too much information to process to get a grasp on this property. The challenges

18. The inspiration

19. Part 2 Visualization

20. Motivation To quickly get an overview of risk, reliability and contributing variables in large complex networks. Catch changes in the system in real time

21. Current visualization techniques with geographically fixed nodes and straight lines have issues: – Limited information can be shown – To show the entire grid, multiple large screens must be used, making it hard to get an overview – Visualizations are static and do not show the most relevant information

22. We present three graph visualization techniques – edge bundling, – focus and context visualization and – non-geographic graph layout with clustering.

23. Edge bundling (Cui, 2008) Our result

24. Focus and context visualization Standard visualization of power system. SMS levels on nodes colored with scale green-yellow-red. Percentage of max flow on edges with scale blue-yellow-red Notice red area to the left

25. Focus and context visualization A focus and context map where important area is zoomed in on displacing the nodes and edges around but maintaining an overall correct geographic placement of the nodes. One can imagine a dynamic system automatically creating multiple zoom-ins on areas of importance. Focus area

26. Non-geographic Graph Layout with Clustering

27. Cluster correspondences

28. Supernodes and superbranches

29. reorganized and given quantitaive values

30. Conclusions After installing the new simulation tool for reliability studies at Statnett SF, the need for a customized presentation method has become prominent. The most promising approach has been the graph layout with clustering. This approach gives the opportunity to, by a single look at the screen, identify the most critical cuts, and get an overview of the risk level in each area along with the state of the power system as a whole. By zooming in on a cluster, more information regarding the risk related to each individual branch and bus is shown. The focus and context approach is particularly suitable if the operators want a complete overview of the power system, while still being able to identify individual critical components. It offers a great way of visualizing changes in a network over time, as different parts of the system will be prominent, depending on the current risk level in each area. Edge bundling is suited to use in combination with the focus and context approach, as branches in areas of little interest can be bundled together, in order to simplify the presentation.