Managing the risk of wildfires

In the August 25, 2019, issue of the The Washington Post, Steven Pearlstein wrote a column about how San Diego Gas & Electric (SDG&E) is pro-actively managing the risk that their power grid will cause a wildfire.  Some wildfires are caused when trees or branches fall and hit a transmission line and then the energized power line sparks a fire on the ground.  The power company's approach addresses two potential problems: damage to the transmission line, and the transmission line sparking a wildfire.

For the first risk, SDG&E is burying transmission lines in high-risk areas in the mountains and backcountry.  This is a preventive action because buried transmission lines can not be damaged by falling trees and branches.

For the second risk, SDG&E has installed a system that monitors transmission lines in remote areas, detects when a transmission line is damaged, and turns off the power to that transmission line instantly.  This contingency plan (turn off the power if the line is damaged) prevents the lines from sparking a wildfire.

Moreover, SDG&E is monitoring the conditions in the high-risk area and calculating the conditional probability that a spark would lead to a large wildfire in those locations.  If that conditional probability becomes too high, then the risk has increased to an unacceptable level, and the power company cuts the power to that transmission line (even though it is not damaged).  From Pearlstein's column:

“We are doing with wildfires what the National Weather Service has done with hurricanes,” said Scott Drury, the utility’s president. ... The wildfire model gives SDG&E the confidence to shut off power because it can pinpoint precisely where and when the risks are at levels that have resulted in disastrous fires in the past.

Pearlstein's article also discusses how people in the area are managing the risks associated with a power outage: parking outside their garages, storing water, and buying generators.

California's offshore oil and gas platforms

An article by Max Henrion in the February 2015 issue of OR/MS Today described the decision analysis used to determine the best option for decommissioning 27 offshore oil and gas platforms off the coast of Southern California.  A complete report on the analysis can be found here.

The article illustrates two of the three critical perspectives on decision making: (1) the problem-solving perspective (what do with the platforms) and (2) the decision-making process perspective (how to make the decision).

From the problem-solving perspective, the decision is actually 27 decisions, one for each platform.  The article lists multiple options in three categories: complete removal, partial removal, and leave in place for reuse.  Within each category were multiple alternatives. 

The attributes used to evaluate the alternatives were costs, air quality, water quality, impacts on marine mammals, impacts on birds, impacts on the benthic zone, fish production, ocean access, and compliance with lease terms. 

Based on the stakeholders' preferences, the analysts created a multi-attribute model.  For each platform, each decommissioning alternative was given a score (on a 0 to 100 scale) for each attribute, and the scores were combined using a weighted sum.  The alternative with the best total score was identified as the best for that platform.

From the decision-making process perspective, the process was an analytic-deliberative one, and the analysis involved many traditional tools, including influence diagrams, decision trees, sensitivity analysis, and swing weighting.

A multidisciplinary analysis team began by identifying a wide range of options but determined that some were technically or legally infeasible.  They then evaluated the remaining ones in more detail.
This included creating quantitative models to determine how decommissioning would affect fish production and ocean access.  They constructed a computer program that lets a user update the scores and weights.  They conducted sensitivity analysis to determine how uncertainties in costs and the impact of changing the weights on the attributes.  The most influential factor was the weight on compliance; a higher weight on compliance increased the desirability of complete removal.

After completing its analysis, the team then issued its report to its client and released its model to the stakeholders.  The deliberative part of the process included a series of meetings with stakeholders and the public and policy discussions that led to legislation that enables partial removal, the alternative that reduces both environmental impacts and costs.

I recommend the article as a case study of the analytic-deliberative process and an illustration of how decision analysis tools can be used.