Maia Schweizer

Chief Development Officer, Origin Energy

Supernova Award Category: 

Internet of Things

The Company: 

Origin Energy (ASX: ORG) is the leading Australian integrated energy company with market leading positions in energy retailing (approximately 4.2 million customer accounts), power generation (approximately 6,000 MW of capacity owned and contracted) and natural gas production (1,093 PJ of 2P reserves and annual production of 82 PJe). Through Australia Pacific LNG, its incorporated joint venture with ConocoPhillips and Sinopec, Origin is developing Australia’s biggest CSG to LNG project based on the country’s largest 2P CSG reserves base.

The Problem: 

Origin’s upstream division oversees a portfolio of more than 2,000 coal seam gas wells distributed across Australia. With plans to bring a significant number of additional wells online each year, the company recognized that extending well life and reducing workover frequency could significantly reduce operating costs and maximize production uptime. Changes in the economics of gas production and the increasing scale and complexity of gas field development have created pressure on companies to raise efficiency through improving the development cost per gigajoule extracted and lowering the maintenance dollars spent per well per year. The challenge for this project was the identification of well failure predictive characteristics and optimal equipment configuration options, based on the rich historical operating data available for each well, including thousands of time-series and static data points describing the lifetime of each well from design to commissioning and production.

The Solution: 

Based on proven, large-scale deployments at 25 global enterprises, Origin selected C3 IoT to develop two AI/machine learning applications targeted to achieve specific operational goals:

  1. Well Equipment Health: predict and optimize the run-life of 600+ installed Progressive Cavity Pumps based on machine learning models to detect failure parameters and optimize equipment design choices for each well
  2. C3 Well Output Forecasting: predict output of individual wells before drilling commences, optimize well placement by accurately detecting low-producing wells, and identify parameters to maximize well output.

The engagement spanned four phases:

  1. Discovery - defined analytical approach and data sources required
  2. Data - data from 12 separate sources were loaded into a unified federated cloud image
  3. Analytics - build machine learning models to solve the business challenges
  4. Visualization – build application to showcase insights in a web-based, hosted application
The Results: 

All wells in Origin's portfolio are equipped with standard Progressive Cavity Pump (PCP) systems, yet each asset varies significantly in terms of the specific installed configuration of components as well as the operating context (e.g., geology, maintenance and production history). Converting rich historical operating data for each well to a unified, federated, cloud-based, multiterabyte-scale image was a complex challenge that Origin had tried to address many times previously. Results:

  • Single unified data image w/ all relevant data: hourly & daily sensor measurements from each well, drilling logs, geology estimates, permeability assessments, well work logs, equipment asset records, and more were stored in unified, federated, cloud based image
  • Single master data schema: overcame data complexity with creation of a single, consistent data schema. Previously each source system codified individual wells differently or used different terms to refer to similar entities
  • Codification of insights from experienced development and sub-surface engineers: single application captures insights from within the organization, resulting in optimal response to issues
  • Converting data & insights into a usable enterprise application with predictive models: a set of integrated analytic modelling, application logic, and visualizations with integration to other enterprise systems, enabling rapid development & deployment of insights to the field
Metrics: 

Origin was targeting a 20 percent reduction in maintenance jobs per well per year and a 15 percent reduction in development spend per gigajoule. Together, Origin and C3 IoT identified predictive maintenance opportunities through data analytics and intelligence to help achieve these operational goals.

Highlights:

  • 12 weeks to implement two production-ready applications
  • C3 Well Output Forecasting Application
    • 300+ wells analyzed
    • Production data over 365-day period assessed
    • Data integrated from 12 disparate source systems
    • 340 machine learning features tested
    • 75% of low-producing wells identified, for a 3-4x increase in predictability over baseline
    • A potential $15M+ AUD annual economic value due to improved field development
  • C3 Well Equipment Health Application
    • Analyzed data from 600+ wells over three-year period
    • Data integrated from 10 disparate source systems
    • Structured and unstructured data (e.g., field work logs) processed
    • 200 machine learning features and 80 design features tested
    • Impending failures detected >4-week timescale
    • Separate prediction scores calculated for rod and tubing failure
    • >300-day improvement in pump run life
    • 20+% decrease in workover frequency
    • A potential $20M AUD reduction in annual maintenance spend
The Technology: 

C3 IoT delivers a comprehensive platform for rapidly developing and operating big data, predictive analytics, and IoT applications, as well as a family of configurable and extensible SaaS products developed with and operating on its platform. The C3 Type System, a data object-centric abstraction layer that binds all C3 IoT Platform components, significantly increases productivity by enabling app developers, data scientists, and business analysts to all work on the same framework.

Disruptive Factor: 

Traditionally this kind of project may have been an IT initiative. However the sponsorship for Origin’s digital transformation is very firmly by the business and for the business, with clear metrics for improvement defined for every discrete project. When Origin constructs wells, it adds many sensors that provide a rich data set of real-time data from the wells (dozens of sensors on thousands of wells) but getting value from those data is a major challenge. “When I came to Silicon Valley about a year ago and met with C3 IoT, it suddenly clicked for me…I could see how those incredible volumes of data could be turned in to really valuable insights and C3 IoT was willing to work with us in relatively unexplored territory.” Maia Schweizer, Origin

Following the delivery of the two production-ready applications and the assessment of the capabilities of the C3 IoT Platform, the Origin/C3 IoT team quickly got to work brainstorming a series of other potential use cases that would require a new generation of software applications and intelligent systems capable of extracting and acting on AI-based insights from big data.

Origin has now expanded the engagement with C3 IoT to drive data to decisions across all aspects of the business, establishing a Center of Excellence to design, develop, deploy, and operate a further set of big data, predictive analytics, AI, and IoT applications for use cases across all facets of the business.

Shining Moment: 

“C3 IoT has enabled us to think differently about what’s possible. C3 IoT has awakened our curiosity and I’m excited to see the problems we solve when we unleash this at scale. This is a cultural transformation. There is unlimited possibility and I think we will surprise each other with what we discover.” - Maia Schweizer, Chief Development Officer, Origin Energy

About Origin Energy

Origin Energy is the leading Australian integrated energy company with market leading positions in energy retailing (approximately 4.2 million customer accounts), power generation (approximately 6,000 MW of capacity owned and contracted) and natural gas production (1,093 PJ of 2P reserves and annual production of 82 PJe).