One of the most potent areas in the digitisation of industrial processes is undeniably a bit of a mouthful: interoperability. It essentially refers to the ability of systems to communicate, and it plays a key role as better data is becoming available and in greater volumes. To succeed, at least two important skills are needed.
– “I doubt whether many companies today consider this to be easy,” says Marie-Louise Bergholt, Head of the Digitalisation in Industry focus area at RISE.
She describes a typical situation in an industrial chain today: You lead a company that specialises in a processing step. Here the raw material is shaped, welded and undergoes surface finishing to create a component. However, for each raw material and every manufacturing step, you need accurate data to optimally control the next step.
- Do the properties vary between individual materials and batches?
- Is the material circular?
- Do the parameters of my machine need to be adjusted? By how much?
The production chain uses machines from China, the US and another from Italy, from different suppliers with different digital maturity and standardisation. What data structures, formats, and data quality do you need in your specific production line so that the finished component – and accompanying data – can be properly handed over? It is akin a continuously changing digital package leaflet according to an agreed standard. In a best-case scenario.
– “The degree of maturity is not the same everywhere,” says Bergholt. “If it is, it’s a fortunate coincidence.”
Key skills for data-driven work
She sees two key skills that need to be in place to seriously work data-driven and take advantage of the data created in the production chain:
Domain knowledge, i.e. a true understanding and specialist knowledge of your process.
– “Without concrete domain knowledge, you will find it incredibly difficult to judge what is or isn’t reasonable. If you just download and combine the data,your data analysis may be all over the place.
Digital knowledge, referring to data collection, data analysis, data structure, data management, etc. How the different machines in the manufacturing chain can communicate, including what formats and quality requirements are needed.
– “If you have poor control over your measurement accuracy, you cannot assess the usefulness of the data. If 5 out of 78 sensors in the production line are not calibrated and time-synchronised, you won’t get an accurate reading anyway. You need to have good methods to be able to detect what equipment isn’t behaving properly along with domain knowledge to assess the consequences.
Utan verklig domänkunskap får du otroligt svårt att bedöma vad som är rimligt eller inte
Sustainability an important skill
Bergholt says that a third key skill should be sustainability. Based on collected data, it is not difficult to add a raster to analyse and streamline processes in order to save energy or minimise environmental impact.
In the field of healthcare, platforms are in place to interconnect the systems in the so-called patient journey – everything from the first test at a healthcare facility and the referral to a specialist to the operating table and follow-up visits. Here, an interoperability hub works in the background, syncs the different information systems and provides them with relevant patient data.
– “Industry faces a similar challenge in terms of building ecosystems with common standards for interoperability,” says Bergholt. “Combining information from the systems in an industrial process chain is labour-intensive and time-consuming.” She explains that RISE, by means of research funding, has carried out pilot projects with a few companies, which have produced very good results and have clarified the number of challenges that need to be overcome. RISE comes into the picture when the processes or standards are not yet mature.
Where do you start as a company?
– “You can start, for instance, by obtaining good control of a critical process step, something that has a major impact on the quality of the end product. Here you should collect data and apply and control sensors for the different types of measurement data needed to understand and control the process. But there is always a degree of uncertainty – you may not know the precise quality from earlier in the process. It may have been somewhat poorer one step before.”
What pitfalls should be avoided?
– “There is generally a lot of uncertainty about data quality. To what extent is the data good enough for use and for what? And whether we can we truly trust our data is a constant concern in many companies.”