Automating the capture of asset condition data has never been more affordable. Goldman Sachs calculates that sensor costs have dropped 75 percent in the past 10 years and this is likely to continue.1 Dedicated communications bandwidth and cloud technologies have made installing sensors cheaper both with respect to initial outlay and ongoing support. Drone data capture is practically a commodity now.

However, organisations need to avoid the temptation of simply capturing high volumes of data and focus on the value and insights the data collected can offer. To ensure the effective automation of asset condition information capture there are a few key questions to consider:

What is right for the operational context of the asset?

The operational context should be the prime consideration in deciding an automation approach. Often this will drive or constrain the type of deployment. Remoteness, safety, utilisation and criticality will drive selection. For example, an Internet of Things (IoT) sensor communicating through a 4G network (e.g. LTE-M channels) may be fit for purpose in metro areas though unlikely to be suitable in the outback. Further, assets in confined spaces or in high volume traffic areas may, for the sake of safety, justify the use of technology that more mundane locations will not. When deploying technology on distributed assets a suite of solutions needs to available. In short, one size rarely fits all.

What risks does automated condition capture introduce?

Insight into asset condition protects against and prepares for a failure mode. It enables decision making that minimises risk – downtime, safety, operational or capital expenditure. Yet the sensors themselves can introduce their own risk. Therefore, a risk assessment of the deployment options is needed. The risk could be from hackers accessing information or controlling the asset through a communication channel opened for the data acquisition. Risk could also be introduced from invasive works undertaken to install technology. Poorly thought out use of drones that violate the Civil Aviation Safety Authority rules, or false positives from machine learning could jeopardise any future data collection program within an organisation.

What type of condition information is needed?

Starting with a goal in mind is crucial. Why is the condition data captured in the first place? Just because you can collect the data, doesn’t mean that you should. Is the condition data captured for operational intervention, maintenance optimisation or long-term renewal planning? The answer will drive the type of information and the rate of information required. For example, millisecond temperature readouts from an electric motor is unlikely to be of use when looking at long term performance and trending.

How to capture the condition information?

Fixed IoT sensors are not the only type of automation that can be used. Condition automation can be split into three different levels: embedding, deploying and processing.

Embedding refers to situations where continual collection of data on the asset is required and justified by the cost and criticality of the asset. This could be, for example, strain gauges and accelerometers fitted to a critical bridge on the road network.

Deploying refers to situations where technology is taken to the asset temporarily to collect data. This could be drone photography, LIDAR scanning or temporary displacement measurement. Choosing a deployment option can free up capital investment and enable the capture to be outsourced. As discussed, if the condition sampling regime is optimised against the failure modes you are protecting against, this math becomes clear cut.

Processing refers to when the technology calculates or infers condition but does not collect the data. Technology such as photogrammetry, image categorisation or other types of machine learnt detection fall into this camp. This is the newest area of condition automation. Even if you are not ready to hand the condition assessment over to a machine yet, if your data collection standard is prepared for this eventuality, you stand to realise significant insight into your assets in the years to come. This also opens up the possibility of repurposing existing data that has been collected and using it for condition assessment.

Where in the asset lifecycle are you?

The deployment time is crucial to technology selection. For example, the right solution for a newly built asset, is rarely appropriate for an asset nearing its end of life. This is due to two main considerations; ease of install and payback. Embedding accelerometers in the core of a high-rise during construction is one thing, retrofitting them is another. Wherever possible on greenfield sites, an embedded approach is going to provide the optimum balance of cost and accuracy.

If you are at the beginning of the asset lifecycle or continually improving your digital engineering approach, condition automation can provide a valuable layer of information to the digital twin of the asset base. But it is also important to consider any embedded technology as a part of the twin. Without it, the next intervention on the asset may well disrupt or destroy your collection solution.

By considering these questions, the optimum implementation strategy that balances the lifecycle cost and risk mitigation against the accuracy and efficiency gained will be clear. The automation landscape for asset condition capture is not without risk. With emerging technology, it never is. Yet with a consideration of these points, the time is right to embrace the capability and deliver significant value to the maintenance, operation and lifecycle planning in your organisation.


1. Goldman Sachs, BI Intelligence Estimates, 2014

Further reading