By Richard Freston, Lloyd’s Register, UK
Digital transformation is proving a challenging journey for many organisations, as they move to create incremental efficiencies and reduce costs by converting traditionally manual time-consuming tasks into digital processes – and in the longer term seek to harness the potential of computer analytics, such as artificial intelligence, and create meaningful insight from numerous data sources. As an example of how industry is advancing new technology applications, LR AllAssets was launched recently at the ARC Forum Analyst Conference in Orlando, Florida – a software platform that provides companies with a full understanding of risk in relation to plant performance and reliability, and a critical component of any company’s Asset Performance Management (APM) strategy.
The product is a cloud-first approach with Open Integration Standards to enable users to easily and securely connect to plant data sources with seamless integration. This allows for more accurate risk mitigation and improved resource utilisation. It houses preconfigured templates and a unique model builder tool enables users to build bespoke models without the need to engage third-party providers, bringing a completely new approach to APM. Users are easily able to configure risk models for their own environments in days or even hours and without any software coding, saving time and money. In a climate of ever increasing scrutiny on performance and risk, software developments such as this provide industry with confidence that operational risk can be managed successfully, and assurance for operators that their plants can be optimised at a technical and engineering level for better performance against regulatory requirements.
But while we may think of digital data as predominantly alpha-numerical, don’t forget there is a more ubiquitous data format which is likely to play a significant role in this transformative environment – visual data.
Visual inspection is the oldest and most natural of techniques for understanding and confirming the nature and condition of anything we are presented with, and the primary means to verify the validity of information obtained through other sources and senses. For that reason alone, it will remain a significant element in any complete inspection solution. Visual information is also an incredibly powerful tool – static and dynamic images are universal and unlike language do not require translation. Its meaning is collective and easily understood.
But unlike manual inspections, using digital techniques for image analysis such as machine learning can unlock solutions at scale, recognising changes in asset condition and identifying defects, across multiple assets, in diverse geographies. Previously unfeasible for a human to achieve at the speed and reduced cost that cloud computing processing and storage power now affords us.
Harnessing the value of this data can and will incrementally change our approach to inspection, reporting and analysis. Data of this type means we can conduct inspections remotely, at scale, with repeatable consistency and quality, and most importantly securely through cyber secure systems. Developments in technologies such as computer vision, artificial intelligence, and wearables are also opening up the opportunity to radically improve how safety risks are identified and managed, from human factors to operational issues.
Continuous data collection
The continuous nature of digital data collection, collation and analysis presents the ability to monitor and benchmark assets throughout their life, not just as individual assets, but across asset fleets and portfolios. The results of such descriptive analytics not only enable us to see patterns and trends, leading to predictive insights and actions, but also allow us to better inform the design phase, bringing the field back into the lab and completing the loop between design and operations. And while it is often assumed that a physical inspection only involves ‘looking’ at the particular items of interest – and that a picture tells a thousand words – in reality a human conducting an inspection will be considering a number of factors in their evaluation of status or condition including sound, smell, touch and, importantly, context. We use all of these, in addition to movement for perspective and other supporting tools (like a torch or a hammer), in order to test and validate our assumptions of what we initially see.
So, as we move to digitalise our visual inspections with a variety of image capture devices, it is important that the strengths and limitations of the approach are fully understood in order to truly move from a qualitative to a quantitative assessment with confidence. This approach should be one that incorporates the natural human need for validation and capitalises on their skill at evaluating complex data to make reasoned decisions. Only then can we fully capitalise on the possibilities of digital data analytics with confidence. That’s why we talk about smart solutions shaped with human intelligence. While in reality the complete digital picture will be built up from a combination of different sources and solutions, the adoption and implementation of these will need to be in stages as each is developed, evaluated and commercialised.