Embedded Files
Key challenges for data-centric manufacturing
Key challenges for data-centric manufacturing
Data-centric manufacturing is a difficult task and many data-driven projects fail to deliver on their expectations. There are a number of well-recognised reasons for this that will be explored here with tools and resources aimed to provide solutions to some of these problems.
Challenges in data-centric manufacturing include:
lack of available quality data,
ill-defined data-driven problems,
staff turnover,
incorrect or insufficient resources,
lack of appropriate skills,
over-hyped and over-promising AI technologies and software.
Lack of available quality data
Lack of available quality data
Collecting data is easy with the right tools. Collecting quality data and making it available to the right people is a much more difficult task and requires a team of people with a variety of expertise.
To collect the most appropriate data requires the digital tools to do so, expertise in the machinery or process from which data is to be collected, a well-defined problem embedded within the domain, knowledge of the available analysis tools and what data is required to use them, and the expertise that can ensure the data is available to the right people.
Assessing the quality of the data is also a difficult task with various standards available to help and guidance on achieving the FAIR principles of Findable, Accessible, Interoperability and Reusable but trusted quantification of data quality is often not available. Until confidence in the data can be achieved, confidence in the entire project cannot be achieved.
Ill-defined data-driven problems
Ill-defined data-driven problems
Many data-driven projects are actually not well-defined. Without a well-defined problem it is very difficult to collect the appropriate data and choose the appropriate analysis techniques. All too often a data-driven project is a result of having some data available and a piece of software that can do analysis. In these cases, results from any data analysis are limited and may not be particularly insightful nor actionable.
To achieve insightful and actionable results it is crucial to have a well-defined problem to solve. It is important that the investments made in data collection and analysis should be motivated by a solid business case as other investments would. For data-driven projects, sometimes the business case can be based around hype and wanting to use the latest technology within an organisation but without a problem to solve this investment will not deliver the expected return.
Staff recruitment and retention
Staff recruitment and retention
Expertise in all digital skills are highly sought-after and many digitally focused organisations pay very well. This can make it hard to recruit and retain staff with the most appropriate skills, especially in smaller organisations.
Where staff recruitment is difficult upskilling and consultation may be more appropriate but a better solution might be to partner with a suitable organisation to provide the required skills.
Where staff have been recruited for data-driven work it can be a real problem if they leave without appropriate data management and project transparency. To complete a project with new personnel requires them to have access to the appropriate data, understand how it has been processed and used and what the initial problem to solve is. This requires good documentation and best practice to be followed which itself requires skill.
Incorrect or insufficient resources
Incorrect or insufficient resources
Preparing the infrastructure and resources required to collect and analyse data can be one of the biggest challenges for smaller organisations.
Even with the availability of tools to help collect data, the data needs to be transported and kept somewhere, perhaps for a long time. Building the infrastructure to achieve this requires expertise and investment that can be a challenge for many.
Once the data is collected, analysing the data can require large computation resources, licences or costs to use resources elsewhere.
Without sufficient resources throughout the whole process a data-driven process can easily fail.
Lack of appropriate skills
Lack of appropriate skills
Collecting, analysing and reporting the appropriate data with the appropriate methods to the right people involves a team of people with a collection of different skills. In many manufacturing settings a sufficient collection of skills is currently unavailable.
The availability of tools to collect and analyse data has grown much quicker than the availability of the skills required to properly use such tools successfully. It's not always obvious what those skills need to be so collecting the right team to cover them is a non-trivial task itself.
Bootcamps and upskilling courses are available for those already working in the manufacturing setting and whilst this is recommended it's important to include different professions within data-driven project teams.
Over-hype and over-promise
Over-hype and over-promise
Big Tech make AI look easy and have provided some great tools to make AI more accessible to the wider public. Unfortunately, most of these tools cannot easily be applied to solve manufacturing-specific problems despite their success elsewhere.
It can also be difficult to know when it is appropriate to use off-the-shelf software for a variety of applications. Marketing strategies, unsurprisingly, promise tools can be applied to solve all your problems but this is often not the case and because of the proprietary nature of the software it is difficult to know what is going on under the hood to make an assessment on their suitability.
Page updated
Report abuse