Process Mining for Manufacturing

What is Process Mining?

Process mining is an emerging discipline to discover, monitor and improve business processes by analysing the available process traces in information systems. Such process traces or so-called event logs are data points recorded when executing a process.

The three main tasks typically performed by process mining software are process discovery, conformance checking and process model enhancement. Process discovery generates process models based on the analyzed event logs. The resulting model is the as-realized process. Conformance checking allows the comparison of the as-realized process flows with the as-designed process model, which contains the originally defined processes. Process model enhancement uncovers optimization potential of the process model. 

Based on the insights, improvements or changes to the underlying process can be implemented. For example, the identified bottlenecks or unplanned process steps can be eliminated. The target of process model enhancement is the optimization of the process model and thus of the underlying process.

A typical improvement cycle with process mining includes the following steps:

1. Collecting process data with event logs in information systems
2. Process discovery and Conformance checking to identify optimization potential
3. Process model enhancement to define the target process model and how to reach it
4. Implementation of the identified process improvements
5. Continuous improvements: collecting data from the new process implementation

‍

Why Process Mining in Manufacturing?

Manufacturing companies focus on creating automated and lean production processes in order to sustain high product quality and decrease the overall costs of their production processes. Typically, manufacturing companies monitor and update their production processes with different techniques like lean production or tools like Value Stream Mapping. However, an increasingly dynamic production environment with smaller lot sizes and more and more product variants are pushing existing techniques and tools to their limits.

For example, with Value Stream Mapping, manufacturing companies try to identify improvement potentials with a visual representation of the production process. The value stream map displays the required steps in a production process and quantifies the running time and volume taken at each step. As such, value stream maps typically comprise the flow of materials and information as they progress through the process. Due to the rather manual creation process of value stream maps with high human efforts that only provide static snapshots of process flows, not all existing problems and bottlenecks can be found and solved with the value stream method. The use of value stream maps can not be recommended for flexible or complex processes with a high number of process types, variants, and multiple manual interactions within the process.

To address the shortcomings of existing lean tools such as Value Stream Mapping, manufacturing companies have started to implement Process mining in manufacturing. In contrast to manual process mapping, Process Mining allows to dynamically analyze production processes and identify improvement potential such as non-value-adding activities in an automated way.

Process Mining software uses event log data from process execution on the shop floor to automatically generate the as-is process model, including all variants. The resulting process graph includes all the information from the data points of the connected infrastructure like e.g. machines or IT-systems such as ERP or MES. Opposed to the more high-level value stream analysis, process mining can incorporate process information on a detailed level with diverse and often invisible process variants, down to a very fine-granular level of detail. Furthermore, intelligent Process Mining software includes manual tasks data from human operators and can thus recognize as-is processes with human interactions and thus any wasteful activity such as longer processing times or waiting times in between the process flow. By extending intelligent process mining with real-time workflows on mobile devices, the identified process bottlenecks can directly be addressed resulting in improved processes with an optimal flow.

Benefits of Process Mining in Manufacturing

The potential of Process Mining in Manufacturing is immense. To unlock the potential, companies can focus on three areas: Remove bottlenecks, decrease process variations, and eliminate all non-value-adding activities.

1. Most bottlenecks are not detected via traditional process analytic tools due to their limitations in capturing dynamic flows over time. The entire dependency of steps results in a complexity, where symptoms (e.g. longer cycle times, high inventory) are often mistaken as single bottlenecks.
2. Especially in B2B manufacturing, the produced lot size is close to one. Additional variants of processes handling those items result in a decrease of productivity. Standard processes and best practices need to be established into the new process model and continuously monitored and improved. Only then unnecessary variations can be identified and eliminated.
3. Waste, such as non-value-adding events, plays a crucial role in Lean Production environments. Unnecessary long ways or overproduction are examples for activities that are not contributing to the value chain and should be removed to increase efficiency in manufacturing.

For more information about how companies established process mining into their manufacturing and what you can learn from them: Check our Ultimate list of Process Mining case studies in manufacturing or have a look at our solution that extends process mining software with event logs from manual activities.

‍

Authors

Sabrina Joos, Solution Consultant WORKERBASE

Thorsten Krüger, Co-Founder WORKERBASE