A case study in the introduction of a digital twin in a large-scale smart manufacturing facility
The paper outlines practical issues that hamper the creation of a digital twin focused on reducing the cost of maintenance. The authors say this cost is 30% of annual operating costs for a facility and 60-75% of a machine's lifecycle cost. The paper breaks the practical issues into four categories - Business, Human, Project, and Vendor. Business refers to issues for the wider organization overseeing multiple projects at an enterprise level. Human relates to issues for workers on the shop-floor. Technical issues are in the finer detail of a job due to technical complexity and new technology introduction. Vendor refers to issues encountered by the companies who have been brought in to the manufacturing facility to install services.
Why it’s relevant to Nextspace
Despite the authors' premise of failure, they do manage to install vibration sensors on ten centrifuge pumps which were about to be replaced (noise produced led the crew to believe they were faulty). The data showed that the machines were fine, and drive belts were replaced instead, saving Euro 191,000.
However, the real benefit of this article for Nextspace Partners can be found on pages 4 - 14, which outline practical issues that should be taken into account during any pilot or digital twin installation.
- To succeed, Digital twins need to be a business priority and the timeframe for ROI understood.
- Data availability and project funding can hamper digital twin projects.
- Failure of past projects create resistance in the business.
- Data silos. Cyber-security can prevent use of data.
- Data quality. Manual data generally has quality issues.
- Staff turnover impairs human expertise useful to configure the system.
- Data completeness. Manual work is generally only partially documented.
- Data quality. Machine data can also have quality issues and require ’cleaning’ before it can be utilised.
- The need to install new technology can cause difficulties.
- Complexity of processes can cause issues.
- Data interoperability. Creating virtual copies of assets with various models from multiple vendors is a challenge.
- Data interoperability. Multiple machine vendors on a single production line.
- Converting data into actionable intelligence requires time, expertise and experience.
- Data availability. Proprietary ownership of data by vendors as well as old equipment that cannot export data easily.
