The machine is standing idle and producing nothing but costs, thereby bringing the OEE to its knees. Such downtimes are the most obvious losses in manufacturing, at least when you are standing right in front of the machine.

But have you ever tried to systematically monitor and analyze the downtimes of your manufacturing over a longer period?

Once you start working with the data, you quickly realize that the downtime periods and their reasons are often distorted. This, in turn, leads to the identification of causes and the derivation of actions being difficult to impossible.

Therefore, today we will discuss how downtimes can be accurately and systematically recorded and present proven methods for analyzing and minimizing downtimes. This will make the calculation of the OEE accurate.

You can find more on the topic of OEE software providers and how to compare them here.

The real costs of a machine downtime

What are the real costs of a downtime? In a downtime, not only productive time is lost, during which a machine stands still; additional costs arise from:

• Production losses in upstream or downstream processes

• Scrap production due to reprocessing

• Overtime wages for additional shifts to keep to the schedule

• Increased risk of accidents due to stress and rush during and immediately after a downtime

If these factors are also considered, the real costs of a machine downtime are significantly greater than often assumed.

But how can one systematically reduce downtimes? First, downtimes must be accurately recorded to understand the status quo and optimization levers.

How to accurately and systematically record downtimes?

There are essentially two methods to record downtimes: manually via written records or manual bookings in systems or automated, based on machine data in combination with downtime criteria (these downtime criteria will be explained in more detail below).

In Manual Recording, the following three types of implementation are the most common:

• Pencil and paper-based downtime logs or shift handover logs

• External, purchased MES/BDE systems

• Internally self-built booking systems

In Automatic Recording, we see two types of implementation:

• Purchased shop floor analytics/OEE software such as that from ENLYZE

• Self-built solutions based on various technology components

In a few weeks, a blog post will be published in which we will discuss the advantages and disadvantages of the different implementation types in detail. We will gladly link to it here.

Advantages of automatic downtime recording

In the second article of our OEE series, we have already gone into detail about the advantages of automatic data recording. When collecting data, data quality should always be the focus. The higher the data quality, the more meaningful the later insights, and the better the data can be used for future AI or Machine Learning approaches - Keywords: Predictive Maintenance.

Manual data recording, on the other hand, is highly error-prone and therefore does not meet the requirements for data quality. This applies to both downtime logs with pen and paper as well as digital bookings in an MES or ERP system. Both processes have in common that the human factor causes an inaccurate data foundation due to their bookings or records.

Another advantage of automatic recording is the alleviation of workers, as cumbersome manual bookings are largely eliminated, allowing full concentration to be focused on problem-solving at the time of the downtime.

What information needs to be recorded in the event of a downtime?

The following data points have been identified as relevant based on our experience over the last few years. These should be recorded for each downtime to enable further analyses:

• Start of the downtime (date and time)

• End of the downtime or duration of the downtime

• Product manufactured

• Machine operator during which the downtime occurred

• Reason for the downtime

• Shift (Optional; this can also be determined via the timestamp)

• Additional description, e.g., comments on the measures taken (Optional)

• Location: e.g., machine, line, part of the line (Optional)

Thus, a considerable amount of information must be gathered for a downtime. However, many of these pieces of information can be automatically recorded using digital systems and machine data.

How does an automatic downtime recording work?

For the detection of downtimes, the performance parameter of the machine is typically used in combination with a downtime criterion. In the simplest case, the latter is a threshold value, while in more complex cases, various process parameters are combined. In this blog article, we focus on the simple case.

As soon as performance falls below a predetermined threshold, the start of the downtime is recognized. Once the performance rises above the threshold again, the end of the downtime is recorded. The mechanism is illustrated in the following graphic. The red areas mark a downtime, while the green areas indicate normal production.

In contrast, manual downtime bookings lead to downtimes not being documented correctly. This is exemplified in the following image:

The typical problems of manual recording are:

• Start and end times of bookings are incorrect

• The recorded duration of the downtime does not match the actual duration

• In particular, shorter downtimes are often not recorded at all

What data points can be recorded automatically?

The following data points can be automatically recorded at ENLYZE:

• Start of the downtime (date and time)

• End or duration of the downtime

• Shift (automatically tracked via timestamp)

• Product manufactured (from the job booking of the MES/ERP system)

• Machine operator during which the downtime occurred

The following two data points must be manually booked by the worker:

• Reason for the downtime (ideally from a list of standardized downtime reasons)

• Location (machine, line, part of the line) (Optional for large machines)

The advantage of automatic recording is that the downtime is accurately recorded as such, and only the further context is booked manually. The context booking can also be done later without any problems, without losing valuable information.

Building a downtime catalog

We recommend that every company, whether downtimes are recorded automatically or manually, work with a standardized downtime catalog. Instead of a free text field, a list of downtime reasons categorized into different categories is provided, from which a reason must be selected in the event of a downtime.

Only in exceptional cases, if there is no suitable reason in the catalog for the occurred downtime, can the downtime be described separately through a text field. This structure creates a stable foundation for later analyses.

The structure

A proven structure for downtime recording looks as follows:

The categories are the loss categories of availability: "Unplanned Downtimes", "Planned Downtimes", and "No Production Planned". The distinction into these categories is necessary to be able to filter downtimes in later analyses and to quickly identify problem areas. A summary of the distribution of downtimes across the individual categories provides a quick insight into which areas measures should focus on:

A few rules of thumb for the individual downtime categories:

If there is a large proportion of downtimes in the category "No Production Planned", then problems cannot be resolved in manufacturing. Potential can only be leveraged here if planning is improved, more orders are acquired by sales, or a different shift system with higher utilization is introduced by management.

If there is a large proportion of "Planned Downtimes", a further analysis of the reasons is necessary. Two typical levers here are the reduction of setup times and the reduction of proactive maintenance. Approaches like SMED can help optimize setup processes. If there are too many proactive maintenance measures, they can be continuously reduced while observing the development of unplanned downtimes due to technical defects.

If many "Unplanned Downtimes" are present, either more preventive maintenance measures could be carried out to reduce the number of unplanned technical defects, or the staff could be better trained to reduce process-related shutdowns.

The use of categories thus helps to provide a rough structure for downtimes and to derive relevant measures more quickly in the later analysis step.

Standardized downtime reasons

Within each category, downtime reasons are recorded. If there are many different reasons, groups should be used. In general, to achieve good usability for workers, no more than 6 reasons should exist within a group.

This is illustrated below using the interface of the ENLYZE app. Here you can see the mask from which one can select the downtime reason for a recorded downtime and thus accurately log a downtime with a click:

There is also the possibility to indicate a different, non-existent downtime reason via the option "No suitable reason". This can occur in the case of previously unknown or very rare downtimes.

When we address this point with our customers, concerns often arise that workers will always select the option "No suitable reason" to bypass accurate downtime recording. Here, we recommend making the effort for a booking without a concrete reason greater than for a pre-selected reason. At ENLYZE, a comment must always be provided if the option "No suitable reason" is selected.

The effort to write a comment is comparatively higher than booking a pre-made reason. Both the necessary flexibility and the correct logging remain preserved.

It is also important to continuously review the catalog and to examine all bookings without a reason. If similar descriptions accumulate, it may make sense to create a corresponding downtime reason. The same applies to downtime reasons that are never used. These should be removed from the catalog to increase clarity.

Adjustments are always to be expected. Therefore, care should be taken that the system can also be adjusted by shop floor personnel. In today's systems, this is unfortunately often not the case. Adjustments can only be made with the help of IT or external service providers. This prevents the systems from being further developed and well accepted.

Do you want to implement a downtime catalog for your shop floor? Then feel free to contact us, and we can create a customized downtime catalog for you.

From this point onward, downtimes will be accurately recorded and systematically named. The next step is now to identify measures to reduce downtimes.

What measures can reduce downtimes?

Visibility on the shop and top floor

Once accurate downtime data is recorded, it should be made visible on the shop and top floor. This simple step has the "magical" effect that downtimes automatically decrease.

The reason for this is that a new awareness of downtimes arises. The losses due to downtimes become visible to everyone, creating an internal pressure to act quickly and motivating everyone to reduce downtimes. In addition, real-time information enables a quick response to downtimes, further reducing downtime periods.

To create visibility on the shop floor, we recommend Gemba boards on large screens. Below is an example from our customer, DuoPlast AG:

For more transparency on the top floor, an overview of the production progress at each machine is also recommended so that one can get a quick overview of the downtimes of the last 24-48 hours.

Identifying Low Hanging Fruits

Once data has been collected over several weeks, the most common downtimes can be identified using reports and analyses. It is advisable to work with Pareto analyses here. An analysis of the reasons helps identify measures that can be implemented easily and quickly.

Here, the downtime catalog comes into play again: Since categories and standardized downtime reasons have been used, downtimes can now be easily filtered and grouped.

Software solutions like ENLYZE can save a lot of time with out-of-the-box analyses and reports and allow these insights to be shared directly with all stakeholders.

To ensure a successful implementation of improvement measures on the shop floor, we recommend the following framework:

In a kick-off meeting, the status quo is shared with all shop floor staff, and measures or solution approaches are developed for the top downtimes together. It is important to involve the shop floor employees to build responsibility and engagement and to take everyone along. Only then will there be a change on the shop floor.

After a possible measure has been identified, progress should be shared in a weekly stand-up (5-10 minutes) with the shop floor personnel. Continuous monitoring of progress and successes is important here. Solutions like ENLYZE can also help create the right reports automatically without manual effort. If the hoped-for improvement effects do not occur, possible reasons and potential solutions can be discussed. This meeting should also be used to gather general feedback from the shop floor.

Quick overview & complex analyses

With a system like ENLYZE, much more complex analyses are also easily possible. For example, investigating fluctuations in the same downtime reason, as it becomes evident: where fluctuations exist, there is often significant improvement potential hidden.

If, for instance, the duration of setting up the same product on the same machine fluctuates between 20 and 40 minutes, there is likely an optimization potential of approximately 20 minutes.

In such cases, the differences between quick and slow setups should be examined closely, and insights should be documented as best practices for all shop floor employees.

Furthermore, Lean methods such as SMED or the Total Productive Maintenance (TPM) approach can be utilized to data-driven find the largest levers with ENLYZE.

Digital tools like ENLYZE make it easier to collect all relevant data, conduct analyses, derive measures, and communicate successes. However, this only has an effect when the measures are also implemented by shop floor personnel. Therefore, it is important to take everyone on the shop floor along and convince them of the advantages of such a system.