Democratising data analytics and AI

Democratising data analytics and AI

How Rolls-Royce turns engineers into data scientists

At Rolls-Royce, we rely on data analytics to provide customers with fuel-efficient and safe engines. Our teams collect data across the product lifecycle - from the way engines are designed, tested, manufactured, operated and maintained; to environmental data and flight operations.  

Being the most complex part of an aircraft, issues do surface during manufacturing and operations.  It is of paramount importance for the root causes to be identified as quickly as possible to ensure safety.  In addition, engine reworks and delivery delay penalties can impose higher costs to the company.  Engineers need to analyse the voluminous amount of data for actionable insight.

Empowering subject matter experts with data analytics and AI

In the conventional approach, subject matter experts (SMEs) have to analyse and solve problems by themselves. That is now changing as more and more companies employ a team of data scientists to extract insights from the data and deliver them to SMEs.

This inspired our Technology team to develop Smart Discovery, a suite of technologies enabling SMEs without data science skills to perform data analytics themselves. Terence Hung, Chief of Future Intelligence Technologies at Rolls-Royce in Singapore, explained, “Many of our engineers have to deal with complex issues on a daily basis but we only have a small pool of data scientists available to help them. By empowering our engineers with the right data analytics tools to solve the problems quickly themselves, we will significantly improve our ability to deliver quality products and services rapidly.”

Instead of using automated machine learning to help data scientists expedite analysis, we flipped the approach and gave SMEs an intuitive tool that automates the heavy lifting of data science. With the collaborative effort of the Technology team and the Rolls-Royce@NTU Corporate Laboratory at Singapore’s Nanyang Technological University, we incorporated AI methodologies and best practices to drive data analytics in Smart Discovery. Project advisor Dr Stephen King, Senior Lecturer – Advanced Analytics, Cranfield University, UK (formerly Associate Fellow – Advanced Analytical Methods, Civil Digital Solutions at Rolls-Royce) said, “There are a lot of technical challenges to overcome in developing a robust tool like Smart Discovery. However, the teams were very open to feedback and enthusiastic about engaging with peers, end-users and customers. This has resulted in a useful piece of technology that can solve engineering problems and bring about massive cost savings.”

The Technology team was awarded the Sir Henry Royce Award in 2019 (Digital Category). From left to right: Bicky Bhangu, President – South East Asia, Pacific and South Korea at Rolls-Royce; Warren East, CEO at Rolls-Royce; Gary Lee, Team Lead, Future Intelligence Technologies at Rolls-Royce Singapore; Terence Hung, Chief of Future Intelligence Technologies at Rolls-Royce Singapore; Kurichi Kumar, Head of Technology at Rolls-Royce Singapore; Henry Kasim, Future Intelligence Technologist at Rolls-Royce Singapore.

Achieving significant cost avoidance

We began to apply Smart Discovery to our manufacturing processes in early 2018. It helped us proactively fix issues, reduce engine rework and achieve significant cost avoidance that year. The results helped the Technology team win the Sir Henry Royce Award in 2019 (Digital Category), which recognises excellence in digital innovation within the company.

Chris Wright, Vice President, Quality & HSE - Services Engines at Rolls-Royce, said, “Smart Discovery is the heart of a toolset that helps us deliver safe high-performance engines even more confidently. I believe that this has the potential to change the game fundamentally for all our Rolls-Royce engines.”

Dr. Airil Seet, Functional and Vibration Lead Engineer, Whole Engine Design, speaks of the benefit of Smart Discovery: “As a lead investigator, the heart of any root cause investigation is the evidence and the data being gathered, consolidated, analysed and interrogated. This task can be overwhelming particularly with complex multi-variant problems that has a multitude of drivers creating the problem. This is where I find Smart Discovery extremely useful to provide me and my team insights to the root cause investigation. It helped me to eliminate suspected features and highlight features or areas for the team to focus on in an efficient and easy to use manner without the need to be a data scientist. I am now able to use the tool independently to investigate new issues. In a way, Smart Discovery is not just an analytical tool but a game changer. It changes the way traditional investigators investigate and think. In uncertain and troubled times such as what we are all facing today, Smart Discovery tool is indeed a refreshing change that we need for our industry.”

Sensing opportunities elsewhere in the company, we reached out to colleagues from Engine Health Monitoring (EHM). With a high volume of engine performance data to analyse, backlogs can occur quickly, with disruptive repercussions to customers. All that changed with Smart Discovery, which made the engineers’ work much more efficient. In three separate investigations, Smart Discovery was able to pinpoint the same root causes that were previously identified manually by the experts, shortening weeks of manual effort into minutes. In some cases, Smart Discovery reduces the number of false positives, saving time that would otherwise be spent to investigate irrelevant candidate root causes.

Smart Discovery won high praise from our EHM colleagues who have been using it on engines on active fleets, pre-empting future flight disruptions with a projected cost avoidance in the multi-millions over five years.  Dr. Zhuo Li, Services Analytical Specialist – EHM (Civil Aerospace) at Rolls-Royce UK, said, “We are able to use this tool to automate processes such as picking the correct parameters and grouping the plots to identify key features. Smart Discovery was also designed to help solve complex issues, like high dimensional correlations and interpretable predictive models, and has reduced investigation times by weeks and months.”

We have now established a community of Smart Discovery practitioners across Rolls-Royce. In 2020, more SMEs have familiarised themselves with the solution to a point where they can perform independent investigations. New users that have come onboard include Rolls-Royce Electrical and Defence businesses.  We have also started migrating the tool to Microsoft Azure so that more colleagues across our global network can benefit from its use.