Troubleshooting describes a type of problem solving, which is mostly used to find the cause for failing of a machine or system and repair it. In contrast to 8D or Six Sigma it's target is focused on short term improvements such as the fast repairment or removal of issues and symptoms in manufacturing. It can therefore be considered as a quick fix to handle the current issue, but does not take touch the root cause of manufacturing issues.
Troubleshooting always starts with the easiest and most probable solution to fix an issue. Following the probability chain affords critical thinking and problem solving skills as well knowledge about processes and operations.
Without any knowledge an operator would randomly start replacing components, substituting one after another until the issue is resolved. Such an approach is highly ineffective, which is why causality of the situation and environment can fasten troubleshooting. By taking into account that a machine was just moved to another location, the cause of failure can probably be found within the surrounding infrastructure or damages from transportation. Although the machine could have broken down directly after its movement, such a coincidence is highly unlikely.
These three steps strongly depend on the operators skills and his ability to solve complex problems. Therefore support structures to improve troubleshooting are needed.
A digital support system for operators is able to recommend multiple solutions to a certain issue. By following a more and more granular questionnaire, an algorithm is able to determine the most probable fix. Based on the selection, digital support in the form of a description and a picture of the fix is provided by the system. Following this approach, a faster and easier problem solving process to shorten repair times, is introduced.
Before productively using the troubleshooting system, it needs to be filled with the shop-floor related data. There are two ways on how to generate this structure. First, the upfront generation of the structure by interviewing experts about possible problems and the best fix. The answers are then clustered to form a decision tree (see Figure 1). This solution is strongly based on the individual's experience and only reflects the problems and their fix at a certain point in time.
The second option is to let people actively extend the database by using the system in daily operations. If a problem occurs, operators insert the problem, its localisation and its fix into the system. If a problem already exists, the operator just chooses the existing branch of the system. The system is now able to learn and is extended over time. At the same time probabilities about the occurrence of an error and the most probable solution are chosen.
Combining both options helps to boost productivity right from the start and shortens the duration until operators can benefit from it. To allow this kind of problem transparency all operators need direct access to the system, which is implemented the easiest by using mobile devices in manufacturing.
The Dynamic Process Execution platform provides among others, a machine alarm app for rapid troubleshooting. Machine alarms are sent in realtime to all operators leading to reduced downtime of equipment.