Plant Operation Advisor

Predictive analytics for offshore platform health monitoring

GE Digital (Oil & Gas) • Design Sprint

Plant Operation Advisor Dashboard

Background

Offshore oil and gas platforms are complex industrial environments with thousands of interconnected assets. Equipment failures can cost millions in lost production and pose significant safety risks. GE Digital's Oil & Gas division needed to transform how platform operators monitor and maintain their critical assets.

The traditional approach to maintenance was reactive—fix things when they break—or calendar-based—perform maintenance on a fixed schedule regardless of actual condition. Neither approach optimized the balance between operational costs, safety, and production uptime.

"We needed to give operators the ability to see problems coming before they happen, not just respond to alarms when it's already too late."

The Challenge

GE Digital assembled a cross-functional team to envision a Plant Operation Advisor that would leverage IoT sensor data and predictive analytics to transform asset management.

Data Overload

Thousands of sensors generating millions of data points, but operators lacked meaningful insights

Alert Fatigue

Existing systems generated too many false alarms, causing operators to ignore or disable alerts

Siloed Information

Asset data, maintenance history, and operational context lived in separate systems

Complex Decisions

Maintenance decisions required balancing production schedules, resource availability, and risk


My Role

UX Designer - GE Fastworks Design Sprint

  • Participated in intensive GE Fastworks design sprint (based on Lean Startup methodology)
  • Conducted rapid user research with offshore platform operators and maintenance engineers
  • Created user journey maps and personas for different operator roles
  • Designed predictive analytics dashboard concepts and alert prioritization systems
  • Prototyped asset health visualization and maintenance planning interfaces
  • Facilitated validation sessions with domain experts and potential customers

Process

We applied GE's Fastworks methodology—a lean startup approach adapted for industrial environments—to rapidly iterate from concept to validated prototype.

Design Process
Immersion & Discovery
Spent time with platform operators and maintenance engineers to understand their daily workflows, pain points, and decision-making processes. Shadowed control room operators to observe how they monitored assets and responded to alerts.
Persona & Journey Mapping
Developed detailed personas for platform operators, maintenance planners, and reliability engineers. Mapped the journey from anomaly detection through maintenance completion, identifying key decision points and information needs.
Concept Development
Explored multiple approaches to surfacing predictive insights, from real-time dashboards to proactive notification systems. Designed hierarchical views allowing operators to drill from plant-wide health to individual asset details.

Dashboard Sketches

Dashboard Sketches 1 Dashboard Sketches 2
Rapid Prototyping & Testing
Created interactive prototypes and tested with operators in simulated scenarios. Iterated on information hierarchy, alert presentation, and action workflows based on feedback.

Solution

BP POA Dashboard Overview

Key Design Features

Asset Health Dashboard

A unified view showing health scores for all critical assets, color-coded by risk level. Operators can immediately identify which equipment needs attention and prioritize their response.

Asset Health Dashboard Detail

Predictive Alerts with Context

Smart alerts that go beyond threshold breaches to provide predicted time-to-failure, confidence levels, and recommended actions. Reduced alert noise while surfacing truly critical issues.

Excursions Tab

Root Cause Analysis Support

When anomalies are detected, the system provides relevant historical data, similar past incidents, and potential contributing factors to help operators investigate quickly.

Excursion Analysis Excursion Filters

Process Surveillance Dashboard

Real-time monitoring of process variables with intelligent anomaly detection, allowing operators to spot deviations before they become critical issues.

Process Surveillance Dashboard

Analysis & Insights

Deep dive analytics tools allow reliability engineers to investigate trends, correlations, and patterns across multiple assets and time periods.

Analysis Dashboard

Outcome

The design sprint successfully validated the Plant Operation Advisor concept, providing the foundation for product development investment.

75%
Reduction in false alerts
30%
Faster incident response
$M+
Projected savings

Validated Insights

  • Trust Through Transparency: Operators needed to understand why the system was predicting failures, not just that it was. Explainable AI became a key design requirement.
  • Integration Over Innovation: The most valuable features connected predictive insights to existing maintenance workflows rather than creating entirely new processes.
  • Tiered Complexity: Different user roles needed different levels of detail—operators wanted quick insights while reliability engineers wanted deep analysis tools.
  • Mobile is Critical: Field technicians spent significant time away from desks, making mobile access essential for practical adoption.

Reflections

Working in the industrial IoT space reinforced how critical domain expertise is when designing for specialized environments. The operators and engineers we worked with had decades of experience that couldn't be replicated through secondary research alone.

The GE Fastworks approach—combining lean startup principles with industrial product development—proved effective at de-risking the concept before significant engineering investment. By validating assumptions early, we avoided building features that looked impressive but didn't fit actual workflows.

Perhaps the most important lesson was that predictive analytics systems are only valuable if operators trust them. That trust comes not from accuracy alone, but from transparency, reliability over time, and integration with the tools and processes operators already rely on.

Due to confidentiality requirements, specific metrics and detailed visuals have been generalized for this case study.