Future of Plant Construction How Process Mining Transforms Production

Related Vendors

tresmo GmbH

Wuxi InfiMotion Technology Co., Ltd.

Many machines in production are isolated solutions. They only know their process step and generate data about it. A holistic solution for the entire process at the manufacturer and customer increases revenues on both sides. What role can process mining play in this?

Marcel Achner works at Tresmo GmbH in the Connectivity and R&D team, focusing on the areas of modular data integration, data transformation and process mining.

Plant manufacturers aim to make their customers' work easier with innovative devices. The machines are usually technically well developed and offer many functionalities. However, their process efficiency depends on the compatibility of their own data with other data formats. The customer's machines in the production process generate data with various data formats.

Usually, plants from various manufacturers are used in the customers' production processes. Each plant represents an isolated solution - with its own data formats, APIs, and technical specifications. Consequently, the individual plants are involved in the entire production process. However, their individual data is not viewed from a holistic perspective for the entire process.

However, this point is crucial in order to increase the profitability of companies. It is time to end isolated solutions and enable a holistic process view.

The Initial Situation

The previous problem is reflected in production processes. Here, several plants are involved and each plant provides data in different data formats. In addition to the machines providing data, there are also data from products from other manufacturers, e.g., batch data. In addition, further data come from suppliers through external interfaces.

The Solution

For an overarching process view, a holistic solution is needed to aggregate data from different sources and formats. With a structured database, a holistic view of the entire production process can then be successfully implemented in a second step. This process view can be done with process mining, which can lead to a better understanding and optimization of manufacturing processes.

What is Process Mining?

Process Mining refers to the use of process data to obtain information and insights about completed processes. The overarching goal is the analysis of business processes in companies - based on the collected data about the individual operations. Overall, insights about the workflows can be generated from this data and process optimizations can be derived.

Process Mining essentially consists of three steps:

• With Process Discovery, models of workflows are created from the event logs of the process data.

• Subsequently, Conformance Checking is used to verify whether the generated process model can implement the process sequences from the event logs.

• The identified deviations between process flow and process model are used in Process Enhancement to improve the process model.

At the end, there is a meaningful process model that can be used as a basis for improving real business processes.

Why are companies still cautious about process mining?

So far, process mining is not used by the majority of companies, as data and processes have to meet some prerequisites. These preconditions for processes are accurately described by Sebastian Human in his article "What is Process Mining and who benefits from it?": "They [the processes] should be known at all times, be well structured and modeled, and every step must be readable from the IT system."

This means that the process data for process mining must be in the form of event logs in a certain structure.

There are systems from large process mining providers such as IBM and Celonis that facilitate the introduction of process mining for well-known ERP and CRM systems. In this process, the providers' applications receive the raw data from the companies' systems via special interfaces. The data is transformed into event logs and process mining analyses are carried out on them. The main customers of large process mining providers are predominantly firms from the financial sector, insurance area, and consumer goods sector.

However, particularly small and medium-sized companies do not have ERP and CRM systems that cover company processes comprehensively and provide data. In addition, there are various business processes in production companies that cannot be represented in either ERP or CRM systems. Therefore, the hurdle for the introduction of process mining in the context of unstructured data and heterogeneous data sources lies in the preprocessing of the data.

The preprocessing of data for process mining.

The data must be processed and transformed into an event log. This requires several intermediate steps of data integration, aggregation, and transformation.

First, the input data from various data sources is integrated for data processing. The attributes of the data are mapped to the attributes of the target data format in terms of their necessity for data transformation into event logs. The essential attributes for data mapping are usually Case ID, Activity, and Timestamp. Process events are abstracted from the reduced data. Process events are events that occur as part of business processes. These events mark significant changes in the process flow and are assigned to process instances via the Case ID. Process instances represent specific sequences of business processes. An example is a "Order" process instance with the process events "Process" and "Complete". The unneeded data attributes of the instances are removed using data reduction techniques. After that, an event log can be generated in any process mining capable format, such as eXtensible Event Stream (XES). This transforms the input data into event logs and makes them accessible for process mining. Subsequently, process mining can provide insights, for example, on how production processes take place and can be optimized.

And it is precisely here that equipment manufacturers see potential. They can offer their customers a comprehensive view of the entire production process using process mining as a feature.

How equipment manufacturers succeed in becoming process mining providers.

To enable process mining as an equipment manufacturer, it is necessary to manage the machines using a Smart Product Platform. This term is increasingly used in the IoT environment. Many companies are already using a classic IoT platform to map their own service processes, for example.

Smart Product Platforms can do much more: The software solutions transform physical products into smart, connected devices and enable new business models.

In total, large amounts of data are generated on a Smart Product Platform, which have not been used for process analysis so far. The full data potential can be exploited with process mining as a feature on Smart Product platforms, such as the IoT cockpit by Tresmo. This can uncover previously immeasurable and unidentified process sequences and optimize inefficient processes. The additional data used opens up new possibilities for providers and users of the platform.

With Smart Product Platforms and data preprocessing services, smaller companies from the production sector can connect their individually manufactured products and take advantage of Process Mining functions. For this, equipment manufacturers simply offer their customers a Smart Product Platform with Process Mining as an on-demand feature in combination with their physical products.

This results in new monetization opportunities for plant manufacturers. With a process mining feature, the customer gets opportunities for process improvement. This leads to cost savings in the customer's production processes. Plant manufacturers receive a share of the saved costs for the process mining feature. Overall, the customer's expenses decrease, while the plant manufacturers increase their profits.

Making machines and systems smart

To be able to provide data for process mining, plants and machines must be smart. Smart products are typically IoT devices that contain information about their manufacturing process. They are also connected to the internet, collect data, and share it with other devices. In this way, smart products form the basis for data-based services and monetization options. In the introduction of smart services or innovative business models, the business use cases and needs of the end customers should be the focus.

So far not smart systems have to be connected to the internet first. This allows the data to be provided as input for process mining. There are various ways to do this, which won't be discussed in detail here.

Central device networking and new monetization options are important arguments for a paradigm shift. Plant manufacturers should rather integrate overall solutions into their product portfolio instead of standalone solutions. In an overall solution, new analysis options are offered for customers through the process mining feature of the plant manufacturers. Thus, with collected data and process mining, not only is money saved, but new revenue streams are also tapped.