Stories

KBR remains at the forefront of digital innovation as we continue to deliver solutions for our customers. As the industry becomes more technology-enabled, we are constantly innovating, implementing and leveraging our shared knowledge across our global business to benefit the organizations we work with around the world.

In addition to digital engineering and data analytics and intelligence, we’re using our experience with machine learning in the US and the UK to benefit our customers across Australia. KBR recently developed a machine learning model to predict asset performance and enable data-driven decision-making on a wastewater treatment plant for a major Australian water utility. One of our leading Australian digital professionals, Aaron Gillett, recently presented this at the Australian Water Association’s (AWA) annual OzWater awards – Australia’s most prestigious water industry event.

With the cost of electricity growing and using a significant portion of a plant’s total operating expenses, KBR was engaged to implement various modelling techniques which focused on the prediction of electrical demand under different usage scenarios. Through this, KBR undertook a Proof of Concept (PoC) study on the wastewater treatment plant, where we constructed a transient model to provide ‘what-if’ analysis capabilities to predict plant performance under hypothetical load scenarios for future capital expenditure planning. The PoC study showed that machine learning can be utilized to achieve CAPEX and OPEX savings for our customer, revealing it can reduce the time and cost of the study compared to the traditional approach. This indicates that machine learning techniques have a significant role to play in the automated model-building arena. 

A subset of AI, machine learning is a method of data analysis that automates the construction of predictive mathematical models. With sufficient data, machine learning systems can automatically learn from data, learn patterns and make decisions with minimal human intervention.

The process of using machine learning starts with reviewing the data to identify any quality issues, before training the model using a variety of machine learning modelling techniques. Further adjustments are then identified and improved upon before being tested and validated using a testing data set, after which the validated model is then deployed.

We follow a standardized approach to deliver our machine learning projects at KBR, which is leveraged from our Data Analytics and AI framework used for customers such as NASA and the United States Department of Defence. Built around three basic steps – to discover the customers’ needs, design the model, and deliver – the framework covers all necessary steps and ensures consistent and successful outcomes for our customers.

With proven ability to help our customers meet their objectives, the success of this Proof of Concept study demonstrates that machine learning has an important role to play in the future of the Australian infrastructure market, delivering solutions through the full project lifecycle of an asset.