Mathematical Sciences Research Challenges for the Next-Generation Electric Grid Summary of a Workshop (2015) / Chapter Skim
Currently Skimming:
3 Data and Data Analytics
3 Data and Data Analytics
Pages 18-30
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 18... ...
, Louis Wehenkel (University of Liege, Belgium) , and Matthew Gardner (Dominion Virginia Power)
|
From page 19... ...
In 2013, market optimization benefited from AIMMS 3.13 (Advanced Interactive Multidimensional Modeling System, a software package designed to model and solve large-scale optimization and scheduling problems) and CPLEX 12.5 (an optimization software package accessible through AIMMS)
|
From page 20... ...
This $178 million DOE funded smart grid project is led by the Pacific Northwest National Laboratory and includes collaborators such as Alstom, IBM, the Bonneville Power Administration, 11 utilities, the University of Washington, and Washington State University. Sun explained that the project goals are to quantify the costs and benefits, develop communications protocols, develop standards, and facilitate integration of wind and other renewables.
|
From page 21... ...
Sun proposed the following research and development directions: • Methods to help evaluate expanded business requirements -- Risk-based decision making -- Multilevel and distributed decision making, such as coordination and aggregation -- Analysis of cross-domain interdependence of gas-electric coordina tion, water, etc. • Improved analytical solution technology -- Mixed-integer programming: hot-start, heuristics -- Stochastic/robust optimization -- Post-solution assessment and recommendations • Data and data management -- Methods suitable for heterogeneous data -- Methods for data transformation -- Visual analytics -- High-performance computing FIGURE 3.1 Pathway from research to deployment of practical optimization.
|
From page 22... ...
How Can Data and Data Analytics Help? Models can be split into two types, according to Wehenkel: statistical models based on observational data and physical models created from first principles.
|
From page 23... ...
The physical models are often representations of deterministic constraints among physical quantities that describe a system, such as algebraic and differential equations obtained from first principles. Most electric power system models are a combination of these two types, according to Wehenkel.
|
From page 24... ...
Weather conditions are one of the main influencing factors he described, and they yield correlations among load, generation, and outage rates, both in space and in time. Given that there may be thousands of variables, Wehenkel said, it can be dif ficult to build tractable models from available data with multiple time steps and correlations.
|
From page 25... ...
Wehenkel posited that solving this problem requires a combination of physi cal models of degradation processes along with additional experimental data and ad hoc statistical estimation. He encouraged more data- and experience-sharing among transmission service operators.
|
From page 26... ...
Lastly, he reiterated that machine learning might be used to build tractable proxies of sub systems and of subtasks, with the latter possibly reused in many different contexts. Wehenkel suggested Pearl (2009)
|
From page 27... ...
He said additional PMU deployment then became standard business for the company, which deployed additional PMUs in more than 35 locations over the first year. These PMUs capture three-phase voltages, three-phase currents, frequencies, and breaker status for each relay/PMU deployed and all transmission voltage levels covered (500 kV, 230 kV, 115 kV)
|
From page 28... ...
Gardner explained that a variety of data-quality issues develop from many conditions, including dropouts or packet loss, latency, repeated values, measure ment bias, bad or missing time stamps, loss of GPS synchronization, incorrect signal meta data, planned or unplanned outages, poor server performance, and improper device configurations. He said that many of these data-quality issues could be solved or mitigated by basic steps such as checking data on frequency, voltage, and current to see if the reported values are near what is expected for the system.
|
From page 29... ...
5 show voltages, line flows, system frequency, and angular separation. Other examples include one-line switching diagrams that mirror energy management system navigation and assist human interpretation while improving accessibility by leveraging data connection and providing flex ibility when needed.
|
From page 30... ...
Gardner summarized the following key problem areas: data silos, lack of a semantics layer on top of the data, lack of cross-system integration, difficulty shar ing data and models, excessive time used to validate data/models, data inaccuracy and inconsistency, common data not being in sync and up-to-date, and difficulty propagating data changes to all pertinent data destinations. Gardner concluded by noting that workforce turnover is increasing and there is a need to ingrain knowledge in data so that important information is not d ependent on the experience base of individual staff members.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.