The production of paper is very capital-intensive. The cost of a new paper machine can quickly amount to 500 million euros. New investments are rare in Europe due to the economic situation in the paper sector and the global market is subject to constant price pressure.

This means that the industry is facing the challenge of having to transform a large number of existing machines into the digital age. At the same time, the paper production process is very complex, meaning that process-related changes can only be made very carefully so as not to destabilize a well-running process. These are reasons why the industry is considered to be very conservative when it comes to change. Nevertheless, digitalization also offers potential for revolutionary progress in the paper industry. Thanks to almost complete automation, the process is equipped with a large number of actuators and sensors that provide qualitative data on the process. This forms a solid basis for the application of new analysis methods, Industry 4.0 concepts and the use of AI-based algorithms for self-optimizing production to transform a paper mill into a smart factory. The multitude of possible technologies and standards often presents companies with the challenge of having to assess the interoperability of existing systems with future technologies.

The following examples show how an Industry 4.0 transformation can be achieved step by step with existing machines.

Step-by-step adaptation

The Industry 4.0-based value creation network

© PTS

In the gradual integration of Industry 4.0 into machines, the sensors, actuators and the control system must be described digitally. These real existing objects are referred to as assets in the I4.0 view. Each asset has an asset administration shell, which is responsible for the digital description and communication with the rest of the Industrie 4.0 system. The architecture of the existing systems must be developed in such a way that I4.0-compliant communication is possible via the asset administration shell. The migration steps towards an Industry 4.0 architecture comprise the following steps:

• Gradual introduction of IP-based communication protocols
• Definition and use of standardized information models
• Provision of additional added value.

Accordingly, the architecture for the cross-company use of data can be designed as follows.

In general, there are three elements, each of which must be networked via an administration shell:

1. the collection and provision of data (see image above: a and b) is achieved by
The integration of sensors, engineering documents, documentation, product and planning data or a software source that provides data. It is important to note that the data can only be transferred in one direction using technical measures. This excludes the repercussions of the analysis platform on the process. The aim here is to bring data into the system via standardized interfaces. This not only involves standardization with regard to the protocol or API, but also the definition of uniform semantics across the submodels of an Industrie 4.0 asset administration shell.

2. the application of algorithms to data (see image above: c) uses the data provided as the basis for a model-based analysis, for learning AI applications or for evaluation by an already trained AI application. The semantics of the data are defined via the submodels of the asset administration shell. In addition to submodels for temperature sensors or control valves, for example, there are also submodels for implementing the interaction between asset administration shells. This defines different types of interaction - from simple queries of values to the implementation of a negotiation (smart contracts).

This means that both conventional business models and future IoT business models are supported.

3. to integrate the data (see image above: d) obtained through the analysis, it must first be decided whether, for security reasons, only read access to the process should take place or whether write access is permitted. Various interesting added values can be created with read-only access: Quality control, optimization or preparatory simulations. The implementation of write access allows the realization of control loops; for example, the use of data as input for a model-predictive controller or online optimization by a neural network.

The architecture allows the data to be used across companies in the same network. For example, the analysis hardware of one manufacturer can be integrated into the system and the data made available for the analysis software of another manufacturer. The aim is to make data exchange as plug & play as possible.

Structure of the administration shell

Initially, static information about the asset can be stored in the asset administration tray, similar to an instruction manual or type plate. The storage of product information in file folders can thus become a relic of the past. Saving and viewing product data sheets, CAD files or other documents is possible with suitable software. Plattform Industrie 4.0 recommends the free open source software AASX-Explorer (Asset Administration Shell) for creating and displaying administration shells.

As part of a research project, the measuring devices of the Paper Technology Foundation were equipped with such an administration shell using the AASX Explorer. The data, which was previously available as a table in Excel, was transformed into an XML representation and can now be viewed clearly using the AASX Explorer. With the standardized XML representation, the information is no longer just human-readable, but also machine-readable and suitable for further processing.

OPC UA offers a way of linking the static information of the sensor with dynamic information - such as continuously generated measured values. Within the project, a database solution was also designed that stores measured values and has an OPC UA interface. Such an asset administration shell, which can exchange dynamic data with other system components, contains the following components A to D.

A: The adapter to the Industrie 4.0 system must be implemented specifically for the communication protocol used and connects the asset administration shell to an I4.0 network based on the same protocol. No specific communication technology is currently prescribed for Industrie 4.0. When implementing an asset administration shell, it is advisable to separate the adapter from the rest of the asset administration shell logic in order to be able to follow further developments flexibly.

If asset administration shells are not only to be a pure data supplier (passive), but also to request further Industrie 4.0 components (active), both the client and server roles or publisher/subscriber roles must be implemented.

B: The asset model contains asset-specific analysis functions and model transformations such as interaction with communication protocols. The asset models are the main component of an asset administration shell. This is usually software that uses the interface to the asset. There are also additional software components such as databases. It therefore makes sense to view the implementation of an asset administration shell in general as a problem of mapping software models to the Industrie 4.0 interfaces. Software models are created by libraries for accessing the communication protocols or data structures that describe engineering data.

C: The interface to the asset: This is used for communication between the asset administration shell and the asset. Depending on the asset, a communication technology such as OPC UA or Profinet is used for this. If the asset is formed by data or software such as simulation models, the corresponding interface is used.

D: The mapping of the asset model to the Industrie 4.0 submodel: This is used to connect the asset-specific models to the standardized submodels of an Industrie 4.0 asset administration shell, for example the state machines for interaction. The asset model describes access to the asset and is a software component. In many cases, these software components can be examined semi-automatically (through reflection or a compiler-supported analysis). In this way, the mapping of asset models to standardized Industrie 4.0 models can be partially automated. The standard-compliant submodels can be easily generated from the official Swagger interface description for the desired platform. Using the software developed by the authors, these submodels can then be automatically connected to the Industrie 4.0 system via OPC UA, REST Services or MQTT.