The Right Data from the Right Source Creates the Right Models

Connected factories are now becoming commonplace, as manufacturers recognise and embrace the power that their operational data can unlock. Indeed, the COVID-19 pandemic forced many companies to pivot and adapt to the ‘new normal’, with the primary protagonist in these efforts being the foundation of data capabilities – digitalisation.

By lifting the lid on their technology, people and processes – and extracting this operational data – many have discovered myriad dormant capabilities and connections that give them even greater insights into the operation, capacity and efficiency of their production lines.

But with all this new data on tap, the big question is: Is this data used to its maximum advantage? Are companies wringing out every last byte of useable information, and are they then deploying it where and when it can make a difference – making the right decisions at the right time?

What’s your source?

In many instances, the right decision can be affected by where the decision is being made, which is arguably as important as how the decision is being made. Each use case is different, as there is value in solving problems at device, control, edge and/or cloud.

The cloud’s value lies in specific types of applications, such as data sharing across ecosystems and the supply chain, data aggregation and non-real-time visualisation (energy monitoring), high-compute applications, machine-learning model development and training, just to name a few.

But typically, the further away you are from the source of the data, the less often you see updates to the data – meaning intermediate data values might be missing. As a result, cloud users typically only see a small set of possible data values and internal device data; process, control and device models aren’t visible at all.

In most cases, decisions should be made as close to the source as possible – the edge – to maintain data context and speed. Contextualisation and modelling at the edge enables richer analytic insights, and edge-to-cloud-based solutions can leverage the best of both worlds to deliver increased agility and productivity.

When you gather and analyse data closer to the source, you get real-time feedback. Without this immediacy, you might build a model or simulation with limited capabilities and scope – because you don’t know what you don’t know! You need information that allows you to understand the real capability, not just the typical observed capability. But an edge solution is just part of the picture.

Data analytics: Limits and myths

Analytics, AI and machine learning can be powerful tools when used in conjunction with libraries and domain experts – who can understand and unlock unseen value and opportunities – and can provide opportunities for process optimisation and future predictions based on observed behaviour. But these tools cannot be applied generically, due to the sheer variety of possible applications across the broad industrial spectrum.

Data analytics does not provide a complete picture, either. You will only see data you can observe, so additional capabilities and limitations typically aren’t visible, and a full-picture status may not be available.

Not enough slices

Take for example an artisan pizza truck and what operational data will tell us about it. It operates for X hours, Y days a week, serving Z pizzas, with a time T between order and delivery. What we can’t ‘observe’ are the truck’s capabilities: How many hours a day could it operate? How many hours do staff need to work? Are they trained, and to what level? What is the maximum number of servings? What is its range (and speed) between pitches? Is it properly maintained? What equipment (capabilities) is in the kitchen? Can other food be prepared?

The pizza truck example illustrates how many companies face a mixture of data and models, where some of the details are not readily available. This is particularly true when trying to examine data remotely. Understanding the limits of observable data and considering where additional data could be collected and included, will ultimately give a more robust picture of any situation.

Compounding this are the multiple challenges around data availability, include sampling rate, the way data is expressed, and limitations caused by network communication resources, resulting in huge gaps in data.

What fits me?

You need to understand the problem you are trying to solve and the outcome you are trying to achieve. If you need to know more and do more, it’s better to start with analytics, which will give you insights into where problems exist, so you can create a strategy to solve them. This typically means adding additional sensors or building models to model the solution. These models can also be critical to discovering and exploring solutions you have never tried before.

There is value in solving problems at device, control, edge and cloud, but in all instances, you should try to use tools that best address your needs. Indeed, you may even find yourself deploying a hybrid solution.

It is crucial to find the right balance to unlock the full potential of a factory. The ultimate aim should be to apply domain knowledge and the right modelling to create an accurate simulation using a complete picture of the data – not just what you already know!