Storm Studio
Context
FPL is a leader in Hurricane Recovery with an average rate of 7 days to 80% power restoration. This is largely due to their ability to mobilize and stage 20,000-50,000 frontline workers prior to hurricane landfall. Workers consist of in and out of state power contractors, and all internal workforce in their storm roles. Recovery costs for the company can be in the hundreds of millions of dollars, and all of the internal day-to-day operations and special projects are halted for the pre-storm, post-landfall and recovery duration.
I was brought in as a Design Lead to support the IBM team.
GOAL: Create a vision of the future that would help decrease the frontline workforce and lower the cost, while retaining the speed of recovery for affected customers.
Problem
There are 3 systems currently being used by the company to prepare for the storm and manage the recovery:
Storm Modeling
-
Storm Force focuses on hiring, managing and dismissing power contractors. It includes staging sites, contractor movement and assignment progress, hotel arrangements, travel expenses and final payments. Currently powered by an internal ADMS (Advanced Distribution Management System).
-
Data comes from heavily manual, paper based processes that upload once every 24h from Staging Sites. This means that teams will at times have 24h idle time (a big financial loss for the company).
-
Storm Modeling focuses on prediction physics to inform the company of the most likely landfall area.
-
Currently these heavily leverage the models produced by the National Hurricane Center with little information on local geography and topography, city layouts, rainfall and FPL specific service areas.
These models are cross referenced with their internal SDM (Storm Damage Mitigation) practice to prepare and reduce weather impact. Much of that process is run by excel spreadsheets and internal infrastructure data from SAP, with little to no automation.
There is no flow of data from these systems to the other two.
Damage Assessment
Storm Force
-
Damage Assessment documents and tracks damage across service areas and deploys service tickets through their multiple ticketing systems (vegetation, lights, power). The damage is assessed by truck and foot patrols consisting of the internal staff in their storm roles.
-
Data comes from a digital form with no ability to upload visual information and relies heavily on accurate assessment of the damage by the patroller; delays there are largely impacted by access to cell service and/or WIFI.
The three systems do not communicate with one another. All information is cross-referenced by personnel. Design was not a part of the creation of these original platforms with IT leading all decisions on usability and development. There was no usability testing done and no research.
Strategy
Knowing that Design was never before prioritized by the client, and that there were teams already involved with the development of the currently used platforms, my strategy for the projects was anchored in three main pillars.
Analyze
Review and understand the current dashboards, technologies, and approach. Learn from the current team about what was done and why, what they knew of the end users, the client, and the progress of the project. Identify gaps, questions and human centered opportunities.
Infer
Based on the analysis, create a draft of the human needs, mental models, and attitudes. Draft an emotional impact model around the storm event. Create an inquiry plan to validate which assumptions should be pursued and which must be adjusted.
Confirm
Using workshops, interviews, visuals and prototypes, test the assumptions and derive more detailed information on needs of the users.
Execution
After an evaluation of the current dashboards I noted issues in the following categories:
Data: data is not timely and there are issues with accuracy (problems validating).
Action: data is not actionable, few is any drill-downs exist. No suggested next steps.
Value: little insight into what the user needs the data for, data not presented to provide value to user based on need.
Context: no context or relationship provided for the data to derive deeper meaning.
Timeliness: no trend prediction or way to highlight data anomalies.
Clarity: no areas of focus, no data relationships created.
Persona Sketches
Out of all the conversations and interviews with the teams and client, I’ve narrowed down the personas that we would consider for the vision prototype:
C-Suite wanted to have transparency into the operations on a high level, understand the money spent, number of hired crews, the recovery progress. They wanted to understand the impacts of their decisions. Currently their questions can’t easily be answered on the fly, as teams are either preparing for recovery or are actively responding to the damage.
Operations want to better understand if the staging sites have the right equipment and the crews with the right skillsets. They want to understand if the crews are working or idle, and their overall performance, so they can make decisions about assigning more tickets or dismissing the crew early.
Damage Assessment Teams send out patrols and compile information after the storm landfall. Only a small team is assigned to review the data gathered and create assignments for Field Staff. They needed faster access to information and more detailed views of the damage without being out in the post-storm conditions themselves.
KEY INSIGHT:
The information needed and decisions made during the Storm Preparation phases are distinct and different than the information and decisions after the Storm made Landfall.
Vision Requirements
Across a series of short term and long term vision workshops, the teams collected a series of technological concepts and requirements. I’ve used these and what I’ve gathered about the users to experiment with the interface.
Use pre and post storm satellite imagery
Crowd source damage images from first responders
Deploy Drones to gather damage images
Enhance Storm Models with AI predictions based on historical, topographic data and new physics models
Use GPS Magnets to track movement of Field Operations
Use AI Agents across the system
Use Computer Vision to analyze gathered imagery and identify damage types
Automate Work Distribution
Automate Ticket Creation
Detail out the Power Grid onto the UI
Advanced Flood Modeling
Itteration
Throughout the discovery process multiple enhancements and versions of the prototype were created and refined with
bi-weekly reviews with multiple stakeholders from IT specialists, data modelers, industry SMEs and client end users.
Prototype
Outcomes
As the prototype versions progressed, the client excitement about this opportunity increased and an additional $4.7M was added to the overall proposal to account for the future design work itself to be spread across 3 years.
Client signed a contract for the first $1.8M for work to be delivered in 2025, which includes early stage design efforts. As a utility company FPL currently does not have an internal UX design team, with Product efforts being run directly by Engineering with little Human Centered innovation. This Contract is an opportunity for further design expansion across other parts of the company.
I was brought in as a Design Lead for this effort in part due to my previous engagement with FPL where I was able to identify an average 5.5h time savings per process for internal personnel allowing the company to save an estimated $2.5M in financial losses per year. This project allowing me the opportunity to gain even further traction of Design within the Utility Giant.
$4.7M
across 3 years added to proposal for Design
$1.8M
signed for 2025 to include
early stage design
Reflection
This project was an opportunity for me to further collaborate with Senior IBM Partners and C-Suite Executives. Working with strictly business end-goals (like financial savings), can be ambiguous for design which anchors itself in human need and experience. This project allowed me to slowly build influence and trust with leadership as well as the other disciplines and transform what was a set of ambiguous technology requests into a cohesive vision that elevated the importance of the Design Function.