Data investment: Start now for future gains

Is the present you looking after the future you? Fundamentally, this is what investment in data now means for your business in the future.
Consider what could happen if one of the sensors on a critical asset fails without anybody noticing. It could result in a catastrophe when an ageing transformer explodes due to overheating, or it could leave you without power to run your other critical assets if your generator runs out of fuel during a power outage.

While there has always been much talk about the collection and the importance of data, it has remained a far-off idea – that thing we’ll get to “One Day”. However, “One Day” has arrived. Some forward-thinking companies have already invested in data, building historical databases using appropriate processes to ensure data credibility. If you want to be one of them, consider this – with a credible, solid historical database you will be able to make accurate decisions about:

• How to position your organisation
• What assets need to be purchased
• How your business should be managed.

Getting there takes some work but don’t be deterred – my aim is to show you the key actions to implement so that investment in data becomes second nature for your business and yields all the desired benefits.

Getting started

Firstly, it’s crucial to have an accurate understanding of all the assets you own. Ensure that your asset register is very reliable and that it reflects what assets you have.
Your asset register should include:

• Asset condition
• Performance type date
• The cost incurred for the asset
• The maintenance performed on each asset
• Who worked on the asset
• Manuals and other relevant information about materials.

Next steps

Once you embark on this journey you’ll be collecting a lot of data, so you’ll need good architecture that can bring all the data elements together and cope with large amounts of data.

Then you’ll need to consider how the data will be maintained. Proper governance processes are necessary to ensure up-to-date and reliable data that can be trusted in the future. For this, you’ll need the right people to maintain the equipment and sensors.
Asset data collection will rely heavily on sensors. You’ll need to know which sensors are becoming redundant and you will need a fallback should one fail.

You will also need a full picture of an asset’s performance. You can’t rely on one type of sensor only.

Let’s take a pump as an example. You could measure the vibration levels alone, but you also need to look at the performance of the pump. For this, you need to know the flowrate generated and the motor current driving the pump to understand if it is operating efficiently.

To realise a credible digital twin, you’ll need sufficient data points to describe the asset accurately and measure things like health, condition and performance.

Also consider technical challenges. Take predictive maintenance as an example and start from the point where the condition of the asset is measured. The very first thing to think about is sensors. For sensor data, you should consider:

• What type of sensors do we need?
• What environment are we going to operate in?
• How will they be powered?
• Will they be battery operated?
• Can we provide them with DC power?
• Will it be necessary to maintain the sensors?
• How will we calibrate them and make sure that they remain accurate?

Data processing, transfer and storage

Sensors are likely to increase data dramatically, which can’t necessarily be processed in the cloud. This creates the need for data compression and analysis at the edge before the aggregated data is transferred to the cloud for higher-end systems to process.

Next, security, data transfer efficiency, the industrial environment and connectivity become concerns. In the industrial environment there might be a lot of electromagnetic fields, ruling out the use of WiFi and 3G data communication and leaving wired communication as the only alternative.

Once the data is in the cloud, provisions must be made for storage, time of storage and further processing. For processing at this level, data volume is minimised before it is passed on to higher systems for decision making. To do this, a lot of different technologies are involved. Knowing that your technology providers are reputable and avoiding issues with obsolescence are pertinent.

Also, interoperability between all your devices is necessary to avoid dealing with different protocols for the devices to communicate with each other. There are many things in the whole data value chain to consider. Think about everything from the sensor right up to the point where the data is processed. It is also essential to know what you want to achieve or do with the data collected.

Then you’ll need to think about how to process and visualise the data for people to use in decision making. Ultimately, the decisions we make must feed back to the asset in some way. In the asset management world, since we’re not looking at controls, the type of feedback will be about replacing the asset when it becomes faulty or deteriorates.
None of this is possible without the right people in your team. Let’s look at the type of skills your business will need. (It’s up to you and your business model whether you bring these skills in-house or outsource them.)

Skills and expertise needed for investing in data

Data science expertise
Volumes of data collected are increasing all the time to the point where there is more data than any person can consume and use for decision making at any one point. We need to find ways to aggregate this data and extract knowledge from it so that users don’t have to analyse fragments of data.

This is where processing becomes pivotal. It can happen at the edge when sensor data is received, and some data processing can also occur in the cloud, where options like machine learning and cognitive processing can take place.

Once in the cloud, the data can be sent further downstream in a more condensed form, highlighting only the main issues and themes that decision makers need. However, they lack the skills that allow them to identify which tools to use at what point and which type of analysis to apply. You’ll need people with the necessary data science expertise to plough through vast volumes of data. For example, look at pure correlations or some other pattern recognition. The need for a skillset around quantitative measures and data is a certainty.

Operational decision makers
You’ll also need operational decision makers to define the basic parameters that they know they will need. The right high-level people are needed to specify the tools and models in which the data is displayed so that the operational staff can use it practically without being overwhelmed. Hence the IT side and the BI component of presenting data is essential. I believe that organisations that have skills at that level can play a significant role in helping clients to make the most of the data that they collect.

Technology maintenance staff
Further down the line, other skill sets will become more critical. In the past, most maintenance was mechanical and electrical. Now, however, we’re starting to introduce more technology into the process, and we’ll need to upskill and focus on people who can maintain this technology.

You’ll also need people with the technical skills required to install and commission the whole digital data chain.

The future starts now

This is an exciting time – disruptive innovation is taking place everywhere. We’re already seeing changes in how services are provided and how we utilise our assets and people. Don’t wait too long to invest in data – you don’t want to find out later that your business is so far behind that you can’t compete. Building your competitive edge starts now and is an investment in your organisation’s future success.