Download Agenda

Interested? Do you feel you will benefit?


All Rights Reserved. marcus evans ® 2019

Outage Prediction and Modeling

Harnessing and Harmonizing Methods to Forecast Outages, Prevent Damage and Expedite Restoration Decisions

19th-20th June 2019
Boston, MA, USA

marcus evans Noth America

What our delegates think of us:

A very well planned well executed conference, providing an excellent networking opportunity and opportunity for sharing best practice

PSEG Long Island

Great opportunity to learn best practices and solutions to challenges facing the utility industry!

Duke Energy

Knowledeable experienced presenters sharing how they are managing common cur- rent challenges in power industry.

Toronto Hydro

An interview with Michael Jensen, Meteorologist, Environmental and Climate Sciences Department at Brookhaven National Laboratory

Ahead of the Outage Prediction and Modeling conference, we spoke with Michael Jensen, Meteorologist, Environmental and Climate Sciences Department at Brookhaven National Laboratory about the development of data-driven models for high-resolution outage prediction.

You have conducted extensive research on the development of data driven models specifically on applying a Bayesian approach to outage prediction models. How has this method helped progress the accuracy of predictions? 

Our work uses historical high-resolution, component-specific, grid outage data and publicly available weather radar observations to develop an outage prediction and estimation scheme through the derivation of failure rate models. The derived failure rate models are used to estimate and predict outages based on current and forecast weather scenarios to be used for planning and deployment of restoration resources. A Bayesian approach is applied that incorporates both the failure rate model estimates and reported outages accounting for each data-stream strengths and uncertainties. The incorporation of the reported outages, even with their inherent uncertainties, results in an improved prediction of component failure rates.       

How does this approach incorporate failure rate models? 

Our work focuses on deriving failure rate models for specific grid components using a combination of weather radar observations for summertime storms and corresponding component-specific grid outage data. Specifically, the unique feature of our work is the use of high-resolution (both temporally and spatially) weather and outage data in the derivation of the failure rate models. This allows us to take into account the time variability and duration of the weather condition that preceded a reported outage.  The failure rate models will be used to calculate outage numbers, which will be further updated using the reported outage numbers based on the Bayesian approach. 

What types of challenges occurred during your research phase of this methodology? What did you learn from those challenges to strengthen the model? 

Challenges during the research phase mostly focused on the methodology for the extraction of the weather data such that it could be best related to the reported outages given the inherent uncertainty in the time reporting of the outages. Evaluating the assumptions made in these stages of the analysis is an ongoing area of research that will continue to strengthen the quantitative output of the failure rate models.   

What do you envision the future of outage prediction modeling holds? How do you plan to continue to enhance and perfect your original studies? 

As computing power increases and new data analysis techniques become more mainstream,  e.g. deep learning, there are new opportunities couple data-driven failure rate models with advanced geophysical forecasting models with an aim towards longer-term forecasting of potential grid infrastructure impacts. Our most recent work has been aimed at bringing these tools to our failure ate model development activities. 

What initially sparked your interest in meteorology? Tell us a little bit about your background.

Growing up with a keen interest in math and science, I did not settle on meteorology until taking a course on the Earth’s energy balance as an undergraduate student. After that course, I was hooked and particularly interested in cloud physics and the role of clouds in the global energy balance. In pursuing my MS and Ph.D., I was fortunate to work for some of the leading researchers in the world and gained great experience in the measurement and analysis of different cloud types. My focus has been split between the study of marine boundary layer clouds and deep convective cloud systems, using millimeter cloud radar as a primary measurement tool. I currently led the Cloud Processes group at Brookhaven National Laboratory where we perform research as part of the Department of Energy’s Atmospheric System Research program using observations collected by the Atmospheric Radiation Measurement facilities fixed and mobile sites. Recent collaborations have brought my expertise in radar observations to the field of grid outage forecasting.

Michael Jensen is a facilitator of a joint discussion

Why you should attend this marcus evans conference?

More than 14 hours of focused end-user driven case studies

  • Utilize the model to spearhead activation plans, determine resource allocation and estimate location of damage to drive decisions

  • Deliberate the benefits and challenges of homegrown modeling tools versus vendor sourced

  • Build damage assessment into models during an event for Estimated Time of Restoration (ETR)

  • Highlight the importance of developing an overall restoration model in conjunction with prediction models

  • Define the impact of an event with OMS and predictiondata to activate the right resources within your Emergency Response Organization (ERO)

  • Refine a predictive storm impact model through re-Training and post-storm assessment

Practical insights from active practitioners in your sector

  • Alex Rojas
    Director, Distributed Technologies

  • Derek Roles
    Director, Emergency Preparedness & Restoration
    Hydro One Networks Inc.

  • Michael Berlinger
    Project Specialist (Meteorologist) Emergency Preparedness
    Con Edison

  • Joseph Muglia, P. Eng.
    Director, Distribution Operations 
    Hydro Ottawa

  • Emmanouil Anagnostou, Ph.D
    Eversource Energy Endowed Chair
    in Environmental Engineering
    Director, Eversource Energy Center
    University of Connecticut

Michael Jensen will be presenting during the second day, 20th of June at 10:15 am.
Presentation topic of the joint discussion: Development of Data Driven Models for High-Resolution Outage Prediction

• Evaluating weather radar observables as predictors of localized outages
• Building failure rate models derived from historical weather radar observations and component outage data
• Assessing impacts of weather forecast uncertainties on outage prediction algorithm
• Acknowledging the weather is extrapolated for areas lacking recorded weather to consider alternate strategies
• Incorporation of reported outages through a Bayesian update algorithm Meng Yue, Electrical Engineer, Sustainable Energy Technologies Department, Brookhaven National Laboratory

For registration pricing and multiple attendee discounts, please contact:

Melini Hadjitheori