Turning to the process data management platform itself, it is important that this should be secure, high-performance, scalable and able to integrate third-party applications that can bring additional functionality to your digital environment. With established players and start-ups all competing for a piece of the pie, it is important to avoid being locked into a limited digital ecosystem that will prevent you from leveraging the full range of solutions.
Ideally, it should be possible to set up your selected process data management platform hierarchically, i.e. at a control system, site and enterprise level. This means the same data structure and data models are shared across data levels, ensuring transfer between levels is straightforward, secure and controllable to the finest degree. This saves significantly on time and implementation costs as the platform expands: for example, if you have to move applications between levels, you don’t need to re-write the interface as it is already harmonized. The right platform should also support a range of open interfaces and application programming interfaces (APIs) for streaming and pulling data, including REST API, .Net SDK, ODATA, ODBC, OPC UA, OPC DA, making the data available to use as new use cases emerge on your digital journey.
Data streaming - both between levels, and to and from the Cloud - should be available in real time and in both directions in order to maximize the advantages of digitalization: if an application calculates the optimal set values for a process, it must be able to communicate those values to the process or there is no benefit! When using the Cloud, it is also important to consider the cost efficiency not only of the data storage but also the cost of data transfer, which can be expensive as data is constantly being transferred back and forth.
Of course, in any process data management platform, strong data modelling features should be present, allowing data to be used and harmonized throughout your organization. For example, an ABB customer wishing to transfer a sensor application between two mills found that they had measured values for the sensors they were following in one mill but lacked set values for the other mill, resulting in no targets for their process data algorithms to optimize. This is where robust data modelling comes in, enabling you to implement process data powered applications on multiple sites to maximize benefits.
Only by considering all the above can a business create a cost-efficient architecture upon which to build their digital future and applications.