Case Study | exxonmobil

Background

ExxonMobil sought to deploy an IoT-based sensor platform to manage and process telemetry data from critical assets, with an emphasis on the GOSCHY sensor. The GOSCHY sensor is a cutting-edge multi-sensor device capable of monitoring various types of data, such as acoustic signals, accelerometry, and environmental conditions like temperature. These sensors play a vital role in helping identify early signs of equipment failures, such as valve leaks or mechanical wear, enabling proactive maintenance.

The existing software infrastructure was not capable of scaling to handle the large volume of sensor data, nor did it offer the level of usability required for efficient sensor management and data interpretation. In addition, there were challenges around optimizing the integration of GOSCHY sensor data into ExxonMobil’s broader infrastructure, including their SCADA systems and IoT Hub services.

A team was formed to help ExxonMobil address these challenges by improving the usability and scalability of the platform to accommodate tens of thousands of GOSCHY sensors. This included designing a user-friendly interface for managing the installation, configuration, and monitoring of sensors, as well as improving the system's architecture for handling real-time data streaming and processing.

Challenges

• Scalability: The platform needed to support the management of 5-50K GOSCHY sensors, requiring a scalable architecture to handle large volumes of real-time data without performance issues.

• Usability: Different user groups faced challenges:

  • Technicians struggled with a complicated and time-consuming setup process.

  • Asset Managers needed a clearer, more intuitive platform to monitor sensor data and receive real-time alerts.

  • Engineers required detailed data analysis tools to interpret sensor readings and predict maintenance needs.

• Integration: Seamless integration with existing systems like SCADA, IoT Hub, and Azure cloud was crucial for efficient data flow and accurate reporting.

• Data Visualization: Users needed better visualization tools and insights from sensor data, which were lacking in the previous system.

• Machine Learning: The system needed enhanced infrastructure to support machine learning models for predictive maintenance and automated decision-making.

Workshop Agenda and Takeaways

Approach

1. Stakeholder Interviews & Workshops: The team conducted a series of stakeholder interviews and workshops to gather insights from ExxonMobil's team. This included assessing the usability challenges and identifying key data visualization needs. The information was used to align the platform’s capabilities with user requirements.

2. User-Centered Design: A human-centered approach was applied to redesign the user interface, focusing on intuitive navigation, real-time data monitoring, and simplifying sensor setup and management. Key features included:

   • Custom roles and user permissions to manage sensor data access.

   • A setup wizard for new sensor installations, enabling field technicians to quickly get sensors operational.

   • Mobile and web applications for on-site and remote sensor management.

   • Real-time notifications and automated reporting to keep teams informed of critical asset conditions.

3. Machine Learning & Data Processing: The system was enhanced with ML capabilities to classify and analyze sensor data in real-time. The platform supports multiple ML models for varied use cases, such as detecting anomalies in sensor data and predicting asset failures before they happen.

4. Agile Development & Iterative Feedback: The project followed an agile development process, allowing ExxonMobil to provide ongoing feedback and adjust priorities. Sprint planning, working sessions, and backlog creation helped keep the project on track and aligned with business objectives.

Interview Matrix

Design

The design process for the IoT Sensor Platform focused on creating an intuitive and efficient experience for technicians, asset managers, and engineers. Here are the key design elements:

Wireframes
Initial wireframes were created to establish the platform’s structure, showcasing tables, search/filter features, user profiles, and actionable buttons.

Flows
User flows were designed for key tasks, for example:

• Sensor Onboarding: A streamlined process for installing and configuring new sensors, including a setup wizard and prompts for required actions.

• Data Monitoring: Flows for viewing real-time sensor data and setting up alerts or notifications based on specific thresholds.

Prototypes
High-fidelity prototypes were created to simulate the final experience. These prototypes were tested with users, allowing for quick iterations and feedback on usability, navigation, and design.

Results

• Improved UX: The updated UI provided a cleaner, more intuitive user experience, making it easier for technicians to install and manage sensors, resulting in quicker deployments.

  • UX/UI Metrics:

    • Task Completion Time: Reduced by 30% for sensor setup.

    • User Satisfaction: 85% of users reported improved ease of use in post-launch surveys.

    • Error Rate: Reduced by 40% in sensor configuration errors after UI updates.

• Scalability: The platform was successfully scaled to handle thousands of sensors, with the backend architecture designed to handle vast amounts of real-time data.

  • KPI:

    • System Downtime: Reduced downtime by 15% due to improved system stability.

    • Sensor Deployment: Increased by 20% in the number of sensors deployed per month.

• Efficiency Gains: The system automated many of the manual processes, reducing time spent on sensor setup, troubleshooting, and maintenance.

  • OKR:

    • Objective: Automate 70% of the sensor setup and maintenance tasks.

    • Key Results: Achieved 65% automation in sensor onboarding and management tasks.

• Enhanced Decision-Making: By incorporating real-time data analysis and ML-driven insights, ExxonMobil's teams were able to make faster, data-driven decisions, optimizing asset management and reducing downtime.

  • KPI:

    • Maintenance Response Time: Reduced by 25% with predictive analytics.

    • Cost Reduction: Lowered maintenance costs by 10% due to more timely interventions.

Future Steps

• Continued Enhancements: Future phases of the project will focus on expanding the machine learning models, integrating more data sources, and refining the platform's ability to predict maintenance needs.

• Field Deployment: The goal is to transition the solution to large-scale field deployments, providing global visibility into sensor performance and asset conditions.