Analysing grid flow over time as a cornerstone of the Smart Grid strategy

Transcription

1 Analysing grid flow over time as a cornerstone of the Smart Grid strategy SMARTUTILITIES SCANDINAVIA 2013 Stockholm April 2013 Einar Hoffmann, Managing IT Architect, DONG Energy, Denmark eihof@dongenergy.dk

2 Agenda Introduction to DONG Energy Future business challenge (distribution perspective) The primary drivers Modeling the grid and related issues The value gained from the initial Smart Grid prototype project Beyond the prototype, the direction is toward Smart Grid Data and information management perspectives Smart Grid landscape (distribution area) Summing up 2

3 DONG Energy Distribution Key Highlights Geographical footprint Key facts Power distribution 968,000 customers 20,000 kilometres pipelines and cables Gas distribution 125,000 customers 6,600 kilometres pipelines Storage of natural gas in Stenlille 1.5 bn. m 3 gas Oil pipe 330 km. oil pipe Value of transported oil is app. DKK 40 bn. annually Danish power distribution market: 78 companies; 8 largest represent 76 % DONG Energy has highest market share of 26 % Danish gas market: 3 companies: DONG Energy, HMN, and Naturgas Fyn DONG Energy market share 28 % Danish gas storage market: Two gas storage facilities (ENDK and DONG) DONG Energy market share 60 % Treatment of gas in Nybro Owned by Energy Markets. Distribution is in charge of operations Financial highlights 2011 mdkk NPV EBITDA ROCE Gas distribution ,9% Power distribution ,3% Gas Storage 1, ,8% Oil Pipe ,8% 3

4 DONG Energy distribution foresees a number of challenges the distribution grid perspective Increased consumption and congestion on the distribution capacity: A considerable part of the distribution grid is constructed in a time with much lower demand for electricity. The load of the MV grid is almost unknown. In parts of the grid, the original spare capacity is almost consumed now. A high growth in demand is foreseen (electrical vehicles, heat pumps etc.) The solution is a combination of: Utilise the existing capacity in a more efficient way. This requires a much better insight in how the capacity actually is utilised. Reinforcement of the grid to provide the demanded capacity. This requires a huge investment. 4

5 The grid capacity challenge High utilisation of the grid requires good insights to prevent overloading. Good insights is a demanding task but the benefits pays off. Acc. grid investments A main driver behind the Smart Grid strategy is to reduce/ postpone capacity investments by a smarter grid operation. Explore grid capacity and flexible grid operation Support Flexible consumption and production solutions Time 5

6 Better insights releases capacity for operation and pinpoints where reinforcement is necessary Example of how a typical daily consumption profile affects the grid loading AS IS TO BE Future consumption scenarios (e.g. rollout of EVs and heat pumps) might result in congestion of the capacity in the distribution grid. It is not easy to pinpoint the real bottlenecks. A better insight in how the grid is loaded over time enables operation of the grid closer to the capacity limit and provides the foundation for controlling flexible consumption 6

7 Insights in how the grid is loaded over time is the key it is easier said than done, though It can either be measured Requires a huge investment in real-time sensors Or it can be calculated Requires fewer sensors combined with a good grid model. A good grid model raises high demands to the information managements of the grid assets it pays off, though. To verify the possibilities and benefit of base insight in the grid loading on calculation (continuously load flow estimation), DONG Energy decided to proof the concept by a prototype project. These models are described further on next slide 7

8 Even though insights can be archived through different models, it is not the same as they provides the same capabilities Measuring is a static method providing stand-a-lone values from the sensors independent of the grid context. Calculation is a dynamic method to calculate/estimate theoretical measurements at all points in the grid. Therefore, this method is very well suited to support what-if scenarios (eg. forecasting analyses), as you can adjust the model (remove measurements, change grid borders, add or introduce load curves, etc.) and calculate the expected effect of the new grid state. Consequently: Static (measuring) provides you grid insights at the measurement points. This way of achieve insight supports only the AS IS version of the grid, though. Dynamic (calculation based on a grid model) is able to support both AS IS and TO BE grid insight scenarios. If forecasting is part of the goal, calculation based on a theoretical grid model seems to be the obvious choice. 8

9 Calculation requires a good grid model The figure illustrates how the distribution grid is divided into three main levels: High, middle and low voltages levels. The focus for DONG Energy has been to focus load flow calculation to inside the MV grid due to huge economical benefits. The reduced model requires that impact from the missing parts are modeled as network equivalents to enable load flow calculation. 9

10 An integrated work flow process is necessary to drive continuously calculations The Prototype grid model was build by an automated work flow from a number of source systems and components. It have been challenging, as these systems and components originally were not intended for this purpose. 10

11 Calculation enables a variety of analyses Curve representing a line's aggregated load ratio over time: In 80% of time, the line was loaded with at least 85 Amps. In 20% of time, the line was loaded with at least 150 Amps. Curve representing a line's loading over time 11

12 Do you have a good model, the result will be likewise good. The "garbage in garbage out" rule applies (unfortunate) here too. Primary substation (inflow) Secondary Primary substation (measurements) Example of correspondence between calculated and measured voltage level during a rather long MV (10 KV) feeder. The model is based on: Measured inflow (current & voltage) at the primary substation. Measured current level at two secondary substations. Customers consumption modelled as load curves to transform yearly consumption to point-in-time load. In this example, deviation was found to be around 1 %, which is very satisfactory Connected customers Deviation between measured and calculated values at the feeder-end point 8 days 12

13 Assembling the grid model is a demanding task Grid model A grid model suitable for load flow calculations requires a much higher level of details as does ordinary grid models used for normal grid operation. Source systems The grid model must be assembled from several source systems, however Source systems has not been intended for being assembled into one consistent grid model. Source systems are often spread across multiple business "silos", supported by uncoordinated information maintenance processes. Basic information structures (eg. primary keys) are often not defined identical across systems. The time window from a change occurs (planned or real) to it is reflected by the different source systems might various depending on the different work processes. This demand changes to the work processes. Issues More problems in data than originally anticipated. Even though the existing domain specific systems (GIS, asset management, SCADA etc.) worked well from the original business context, a lot of information model and process related problems revealed when information from these systems were assembled into a single grid model. 13

14 Several techniques were used to circumventing the model issues Transformation All information had to be transformed to a common model, as it could not be assembled in it's original form. Assembling grid model A detective work Synthetic load curves Tolerant processes Odd load-flow results All information had to be assembled into a (live) grid model to detect problems (grid elements missing information or grid islands indicating insufficient topology). It was a detective work to assemble all needed types of asset information (cable type information, relay information, transformer information etc.). Formulas transforming yearly consumption to instant load was established. These were divided into more than 20 customer categories and seasons (to take weekends and holidays into account). Auto correcting common problems (several element ID problems across the different systems was circumvented this way). Incomplete asset information was circumvented by applying defaults. All identified issues was logged to support an offline (manual) process to address the rootcause in the source systems. Odd load-flow results do often refer back to issues in the grid model, which must be pinpointed and corrected in the source systems. 14

15 Surprising how valuable information assembled in a consistent data model actual is The benefits, of activating the grid information in a consistent and easy accessible model, can be boiled down to: When unveiling and utilising the grid information you actually have, you will be surprised of how valuable and useful this information is and how far it can bring you! However: Do not underestimate the effort required to transform the master data systems from administrate singular asset information to become an integrated part of the food chain of consistent grid model. Notice: These benefits are particular relevant for utilities, which operates distribution grids constrained on available capacity. 15

16 Prototype proofed Short term focus Longer term focus Beyond the prototype project, the direction is toward Smart Grid seen in a distribution perspective Verified the use of loadflow calculation as one of the corner stones behind DONG Energy's Smart Grid strategy. Results has proven to reduce/postpone grid investments. Utilised identification and prioritising of strategic Smart Grid initiatives. Access to an unified grid model enriched with load-flow calculation across the time dimension has revealed positive side effects by enable new types of analyses. Optimising grid operation by utilising load-flow calculation in real time as an key element in operational decision support. Optimising investments in the grid by using loadflow calculation as one element in the analyses to pinpoint the real capacity bottlenecks in the grid. Future operational systems. DONG Energy is under way modernising the suit of Distribution Management Systems and both requirements and vendor selection has been heavily influenced by the focus on load-flow calculation. Establishing more advanced analysis and forecasting support. Utilise models to curtail consumption peaks as an alternative to just reinforce the grid to cope with distribution capacity peaks. More focus on planned asset maintenance based on how the asset actual has been used/loaded over time. 16

17 Grid border change feed back Grid change deployment Smart Grid landscape heavily depending on grid calculation (both operational and analytic perspectives). And last but not least, highly optimised processes are required. Asset Management and Operational need tight coordination to ensure consistency Condition Based Maintenance (statistical based) CBM Feedback Asset Management GIS Asset Master Data Etc. Analytic Ad hoc and automated analyses Expose grid information Condition Based Maintenance (usage based) Field Crew Processes Communication (with operational staff). Manual grid operation in the field. CBM Feedback Statistically based standard probability of failures to support accelerated root-cause analyses and restoration Operational SCADA/DMS OMS Low Voltage grid 17

18 Summing up on the experience of using continued grid flow analyses as a cornerstone in the Smart Grid strategy Agile approach to Smart Grid The prototype has provided a work bench for early verification of potential Smart Grid initiatives. This agile approach has been more beneficial than original anticipated. Measured vs. calculated Decide which approach to focus on a decision on both short and long term. Master data challenges The process of change previous stand-a-lone asset management systems to become part of an integrated process has been very demanding. Focus on eliminate cross business domain ("silo") issues. DMS strategy The prototype project has been an important basis for shaping the scope of future operational system (DMS under implementing), as it provided DONG Energy with a much better understanding of the practical Smart Grid possibilities and how to address the master data challenges. Challenges & benefits Data unification and cross business domain ("silo") issues must be addressed. The challenges might look overwhelming the benefits justifies it all, though! It is all about Turning data into information! It is not enough to just to collect values the associated grid context is likewise important! 18

19 Thank you for your attention! Contact information: Einar Hoffmann, Managing IT Architect, DONG Energy dk.linkedin.com/pub/einar-hoffmann/2/456/790 19