Horizontal integration – from business integration to digital twin
Posted on: Mar 01, 2018 by Andreas Schreiber and Murli Mohan Srinivas
The horizontal coexistence of data management tools in different domains raises questions regarding data ownership, dependency and traceability that were not addressed so far by industry solutions. The various horizontal integration domains such as requirements management, product design, manufacturing planning and execution as well as supply chain management processes pose high priority technology issues with arguably medium availability of skills or ideas to fix them.
A horizontal integration platform allows organizations to select best in class products tailored for their organizations in a heterogeneous environment. Interoperability needs to be bi-directional, both for internal and external systems, include transformation of key data such as product structure, while incorporating robust business process workflows and rules. Our focus is particularly on requirements specific to discrete manufacturing.
What does horizontal integration mean?
Let’s discuss horizontal integration by picking two representative applications out of product design and manufacturing: PLM and MES. The methods to manage design data and processes (CAD) are today well understood by various industries including auto, aerospace and others. Consequently, this approach known as product data management (PDM) is now being applied for other domains such as requirements management, simulation and test.
Figure 1: Technologies that are open to horizontal integration
MES systems streamline and simplify complex workflows while increasing the degree of integration and collaboration in manufacturing processes. They directly impact KPIs like cycle time, automation, level of work instruction assignments, data accuracy, through-put availability, reliability and quality.
Looking at horizontal integration (downstream integration) we envision the capabilities of major applications from product design, manufacturing and after sales to tightly couple with each other, using exactly the same data sources.
Here’s an example of horizontal integration: In automotive final assembly, a collision is detected. A wire harness is hindering a worker from installing a steering axle. There are two options at hand to solve the problem, either change the design (move the wire harness some millimeters), or by changing the production sequence (install wire harness after steering axle). The right decision needs to be made based on a number of factors: price (for product design, for manufacturing and for supplier parts), quality, supply-chain impact and many other parameters. Such an analysis can only be done based on consistent data out of product design, manufacturing and manufacturing planning. This requirement is carrying us to the topic of enterprise data models.
The use of an enterprise data model
The horizontal integration of data coming from different domains requires a common understanding of the data semantics. Already the interpretation of a single descriptor like “Part” might come to different conclusions: a part to be designed in product design, a CAD-document representing a part, a part already produced on shop floor, or a spare part waiting in a warehouse to be used. This gets even more challenging if the data needs to be shared with external partners, like customers, suppliers, engineering partners or maintenance contractors. Hence common acceptable standards such as ISO STEP, JT, sysML need to be adopted to share master data across applications.
Following semantic data challenges, appropriate infrastructures need to be put in place for making horizontal integration happen.
A Cloud-based integration architecture
Beside their standard advantage of almost unlimited performance coming with a decrease of infrastructure expenses, cloud infrastructures seem to be a good fit for horizontal integration topics spanning multiple enterprises. This can be seen in discrete industries such as Automotive or Aerospace.
Figure 2: Horizontal integration framework based on Cloud technology
The architecture naturally enables the integration of real-time field data with structured product design or manufacturing planning data to create a mash-up to help build analytics on top. This concept is still maturing; however some initiatives have emerged in this area to add to the horizontal integration layer. One of these initiatives is called Digital Twin.
Business integration as digital twin
Performing validation through simulations has been an age old practice in the manufacturing industry. The ongoing practice is to physically build prototypes to help validate the design, whether it is for product development, production planning or manufacturing. The value of a Digital Twin is to reduce or even eliminate costly physical prototype builds and validations. As we have seen horizontal integration across domains is essential to help cross functional teams work together to accomplish business tasks and objectives.
Front loading is made possible not only to test in controlled environment, but to validate the design assumptions in diverse external conditions, through the use of an intelligent combination of horizontal system integration with agile micro-services based architecture design. Moving forward we will see the co-existence of both architecture blocks, namely horizontal integration which will focus on traditional domains such as PLM, MES, SCM etc. and microservice based data integration where time series data is mashed-up.
The future holds an exciting phase where the traditional world meets with the new IoT world in solving real world business challenges through an intelligent coexistence of horizontal domain integration with the agile IoT integration through microservices architecture.