User-centered power network visualizations empower digital twins

Subscribe to ABB Review

ABB’s novel visualization display design will allow engineers to quickly grasp State Estimator digital twin function. Grid ­problems might be solved more efficiently, thereby creating a more reliable power network system for the future.

Antony Hilliard ABB Automation Solutions Västerås, Sweden, antony.hilliard@se.abb.com; Giuseppe Martinelli ABB Enterprise Software Network Management Västerås, Sweden, giuseppe.martinelli@se.abb.com

Electric transmission and power distribution network systems are increasingly smart and vastly complex. These systems must boast real-time predictive monitoring capabilities that support system operators to manage their interconnected power grid in compliance with international regulations. The massive Eastern North American black-out and the Italian black-out, in 2003, brought vulnerabilities to light. Consequently, regulation authorities now require transmission network operators to monitor their neighbor’s networks in addition to their own – a herculean task [1].

Visualizations are designed according to the job.
Visualizations are designed according to the job.

ABB’s Ability™ Network Manager (NM) offers a combined Supervisory and Control Data Acquisition system (SCADA) and Energy Management System (EMS) to provide advanced monitoring and operation support applications to customers.

The Network State Estimator (SE) is at the heart of the EMS and relies on a digital twin, or virtual model, of the customer’s complete electric network – generators, transformers, circuits, etc. The SE predictive model runs in real-time to smooth over noisy or missing data using a weighted least square estimation algorithm with augmented blocked matrices. Furthermore, gaps in electric network visibility can be filled-in, thereby easing recognition and decision-making for both human operators and EMS applications. Hence, SE fulfills critical functions and supplies exactly those features that customers depend upon.

Maintaining the State Estimator
Nevertheless, crucial challenges exist: The State Estimator must operate within varying environments. Over time, grid equipment is replaced and network connections are altered but the quality of the solution must be maintained. This is especially problematic if enough model mismatches combine with corrupted data from failed telemetry or cyber-attacks. In these instances, the State Estimator algorithm might be unable to forge solutions in either part or all of the network resulting in degradation of automatic monitoring, eg, contingency analysis, and risk of violating regulations. Such SE failure eliminates the tools that EMS engineers depend on to maintain situation awareness and solve grid problems.

Additionally, the requirement to monitor neighboring networks can increase exposure to anomalous models or data. For instance, the majority of the 113 SE outages that occurred in eastern North America between 2013 and 2017 resulted from modeling or communications issues [2].

Commissioning and monitoring of the SE to maintain reliability in the face of these challenges is costly and time- and labor-intensive. To rectify this situation, ABB investigated visualization methods to help SE engineers, experts and non-experts alike, to easily and efficiently monitor SE model health and diagnostics, eg, solution residuals and convergence iterations.

Currently, SE monitoring displays show data inputs, estimated results and rely heavily on numbers and tables. Displays are designed for maintenance by advanced network tuning experts who are few in number and high in demand. Such design is challenging for non-experts to use; demands time and labor and impedes the monitoring and diagnostics of model health – critical for utilities to maintain a competitive edge in today’s market.

Project inception and methods
In 2018, ABB initiated a research project to investigate ways to visualize just where the SE model might be encountering problems: mismatches between model and data. The fashioning of displays that intuitively help experts and non-experts make sense of problems would allow staff to efficiently and pro-actively reduce risks to solution reliability by:
• Pin-pointing faulty telemetry or measurements to block faulty data
• Locating aspects of the model that are obsolete
• Assessing model tuning and how to improve it

ABB validated the visualization through design workshops and interviews with ABB experts, and by eliciting external reviews (four Network Mana­ger customers: two in Europe and two in North America reviewed the concept).

Mapping the electric connectivity
In order to create an agreeable visual analytics solution best suited to SE diagnostics, the design team evaluated conventional maps of electric network connectivity, typically employed for different purposes [3]: geographic view and transmission schematic overview →1.

  • 01a Geographic views show where physical assets are located and have been in use since the 1970s.
  • 01b Standard diagrams for transmission networks show the location of connections logically. Here, Nominal voltage is mapped to line thickness after [5].
  • 01c Power balance shows network function based on Cuffe and Keane’s academic work and include ABB’s design modification [4].

Power distribution companies send maintenance crews into the field and troubleshoot physical damage to equipment. Geographic views show the physical location of assets and are easy to interpret and so are well-suited for these tasks (navigation by continuous plan with zoom features) →1a. Nonetheless, physical location is irrelevant to the functionality of the SE model, making geographic views alone unsuitable.

Schematic views show the logical electric connectivity of network circuits and stations and are therefore standard for transmission companies →1b. The schematic representation style is more abstract than the geographic view: it displays buses as straight lines and navigation occurs by discrete steps along lines. Essential for planning, the schematic view enables grid operators to readily discern all the possible connections that circuit breakers and switches can be used to create. This is invaluable knowledge and yet the SE only solves the present situation: a tiny visual difference on a schematic could greatly underrepresent the impact of an event on the SE modeled solution, eg, closing a circuit breaker.

Because State Estimator model troubleshooting is dissimilar to distribution- or transmission network operation, a new approach is required. Based on previous academic schemes [4], ABB applied standard graph theory with visual formatting, landmarks and navigation aids to create a network visualization that is abstract, intuitive to use and displays the underlying structure of the network model →1c–2 [5]. Ultimately, this visualization, if proven useful for SE, might also be effective for monitoring power grids of the future.

Even though network graphs are successfully employed for analytics in many fields today, eg, social sciences, logistics, etc., ABB’s challenge was to design this demanding application to be user-centered for power network applications.

Design evolution
Because SE model health indicator visualizations must scale up to be useful for large network models, a minimalist graph design approach was chosen. This allows space for data and a ‘global’ visual effect style, one that relays overall network and model properties.

02 ABB designed a Network Topology visualization for Power System Explorer’s State Estimator. The result complements the existing interface for the transmission network digital twin.
02 ABB designed a Network Topology visualization for Power System Explorer’s State Estimator. The result complements the existing interface for the transmission network digital twin.

Accordingly, ABB incorporated features of basic network graph design: node circles as stations, and edge lines representing electric circuits. Three main visual factors were used: node size, line width, and line length. Node sizes encode the difference between station power generation and power load. So, larger circles indicate important stations: either large generators or large loads, eg, cities. Smaller circles represent self-sufficient cities, or transmission stations, and small junctions (circles are hidden). Line widths visualize circuit capacity or voltage level; a thicker line denotes higher power flow rating. This style distinguishes strong from weak circuits and visually separates networks of different voltage levels →2,3.

03 The SE network visualization concept complements existing tabular data structures and can support intui­tive user navigation by pan, zoom, voltage layers, selection in tables, or station/landmark search.
03a Network visuali­zation shows tabular structure.
03b The visualization concept allows users to zoom in on structures of interest.
03 The SE network visualization concept complements existing tabular data structures and can support intui­tive user navigation by pan, zoom, voltage layers, selection in tables, or station/landmark search. 03a Network visuali­zation shows tabular structure. 03b The visualization concept allows users to zoom in on structures of interest.

Although operators often use color for network voltage, color coding is better suited to represent data – the key to comprehending the quality of the SE model solution.

Typically, line length represents distance or km, but, since distance is not directly relevant to the SE, line length was drawn using impedance-derived Net Power Transfer scaling [4]. Hence, line length supports a visual metaphor of “power travels in straight lines” and results in the least distortion of the network structure [4]. Furthermore, this device helps SE engineers visualize the “path of least resistance” and the underlying electric model. Longer lines represent less direct power flow paths →3,4.

04 By using monochrome visual design to show network struc-ture, color visualizes SE health indicators eg, solution residuals or convergence iterations. By designing color scales so that significant deviations are shown in higher contrast, the eye is drawn to the source of the problem.
04 By using monochrome visual design to show network struc-ture, color visualizes SE health indicators eg, solution residuals or convergence iterations. By designing color scales so that significant deviations are shown in higher contrast, the eye is drawn to the source of the problem.

Vitally important major boundaries in the SE model solution (eg, between the customer’s network and neighboring networks, or between what is “observable” and “unobservable”) are visualized by extending line lengths slightly to create blank space, thereby forging separate areas.

The resultant visualization is an insightful view of how the SE sees the current network configuration – how power flows from large generators to large loads, and which aspects of this flow are solved properly by the SE →3,4.

Fostering visualization
Important questions remained, however, once the design concept was complete: How can EMS engineers relate to this visualization design in terms of what they know about their network and their region? Which features foster the usefulness of network-centered visualization?

Many challenges can affect visualization acceptance. The use of consistency and conventions help avoid misunderstandings that can lead to hazards. Consequently, ABB is developing the network-centered design for EMS SE model maintenance staff (either IT or electrical engineers) to complement traditional representations →2.

Another barrier to acceptance is the risk of user-disorientation. To avoid this, conventional design features are included, eg, north-up, west-left. The graph was decluttered by bundling minor junctions and line segments inside the circuit lines and progressively revealing text labels →3. For smaller stations, this is accomplished with a zoom-in feature. Network landmarks, like groups of lines in a System Operating Limit (SOL), are represented by large background visual features, eg, a load “pocket” in a city is denoted as a boundary circle (the SOL describes a limit to the total power flow on electric circuits that the boundary circle crosses). SOLs for regional transfers like an East-West flow are shown as boundary lines. Here the boundary lines are drawn vertical. →3b.

To zero in on modeling problems, different network voltage levels can be navigated as “layers” that can be brought to the forefront, while the user can view the rest of the network as background. Instead of trying to include all details within stations, the design helps users navigate to the existing diagram display once they have located the modeling perturbation →3.

Recognizable visual features allow users to apply their knowledge and experience to steer and search through the network representations; this user-centered design helps SE engineers diagnose SE model health stress factors.

Color aids interpretation
ABB designed an attractive user-centered visualization scheme that employs scalable styles to reveal the structure of the network model. Because the display functions successfully in black and white, color can be used to represent data such as SE model health indicators →4.

Continuous color scales can show ubiquitous analog data, eg, model residuals and model convergence times. By increasing brightness to draw the eye to the model area showing signs of stress, the user can immediately grasp patterns in SE health issues. Concomitantly, discrete indicators, eg, last bus converged or bad topology detected, can be shown with pop-up pins →4,5.

05 Data visualization overlays on the core network can be used in the future to compare different grid states according to customer needs eg, time history, power-flow studies and simulations. In this screenshot, pinned points are shown.
05 Data visualization overlays on the core network can be used in the future to compare different grid states according to customer needs eg, time history, power-flow studies and simulations. In this screenshot, pinned points are shown.

To make the system even more user-centered, ABB embraced customer feedback for extended functionality, eg, animated particle flow trails could show patterns of real or reactive power →5.

Power network future is visual
Once the visualization scheme is approved for SE monitoring and diagnosis, it could be re-used for other critical future applications. For instance, the consequences of the highest risk contingencies could be visualized or the visual representation of a network following an outage could be displayed along with the consequent overloading spread. Changes in a neighbors’ network situation through­out a busy day or large changes in flows or redundancies could be summarized.

The ability to quickly and easily evaluate queries visually will help power network operators maintain reliability and security efficiently. Furthermore, the same features needed to monitor the SE will help ABB commissioning experts set up the model, demonstrate acceptance tests, and train customer staff.

Commonly used, but rarely maintained to a high level, SE requires continuous fine-tuning from experts for full functionality. And yet, the ongoing energy revolution will undoubtedly continue to increase the demand for greater situational awareness of power network systems. Consequently, more effort will be devoted to SE development and expansion. The current growth of renewables and the need to maintain capacity are potential drivers. SE technology has yet to be included in the small distributor market and this market sector is expanding [6]. Improvements in infrastructure and control systems enable expansion of SE in the balanced network market, ie, the high and medium voltage market, and allow entrance to the low voltage market.

The US power transmission and distribution markets are expected to grow significantly by 2023 [6]. By improving State Estimator model solution capability and usability, ABB aims to enhance profitability: ABB’s SE visualization design is a first step.

References
[1] NERC. (2018, June 5). Lessons Learned: External Model Data Causing State Estimator to not Converge. [Online] Available under: www.nerc.com/pa/rrm/ea/Les-sons%20Learned%20Document%20Library/LL20180602_External_Model_Data_Causing_State_Estimato r_to_Not_Converge.pdf

[2] NERC. (2017, December 12). Reference Document: Risks and Mitigations for Losing EMS Function. [Online] Available under: www.nerc.com/comm/OC/ReferenceDocumentsDL/Risks_and _Mitigations_for_Losing_EMS_Functions_­Reference_Document_20171212.pdf
[3] A. Hilliard, F. Tran, G.A. Jamieson and A. Greg, “Work Domain Analysis of Power Grid Operations”. In Cognitive Work Analysis: Applications, Extensions and Future Directions. 2017, pp. 149–170.
[4] P. Cuffe and A. Keane, “Visualizing the Electrical Structure of Power Systems”. IEEE Systems Journal, 11(3), 2017, pp. 1810–1821. Available under: doi.org/10.1109/JSYST.2015.2427994
[5] C. Mikkelssen J. Johansson and M. Cooper. “Visualization of Power System Data on Situation Overview Displays”. 2012, pp. 119–126 Available under: doi.org/10.1109/IV.2012.41
[6] Market Watch. (2019, Feb. 20). Distribution Transformer Global Market Projected to Grow Radiantly by 2023. [Online] Available under: www.marketresearchfuture.com/reports/distribution-transformer-market-2581

Links

Contact us

Downloads

Share this article

Facebook LinkedIn Twitter WhatsApp