Few things capture the attention of management as much as high scrap rates. The reason: quality losses are among the most expensive losses overall, as a large part of the value creation that has previously occurred suddenly becomes worthless.

But what exactly falls under the category of "scrap"? And how can active measures be taken to reduce the scrap rate? We explain the concept of quality losses, clarify how the scrap metrics are calculated, and examine strategies to reduce scrap on your shop floor.

Learn more about OEE software and the calculation of OEE in the linked articles.

What are quality losses?

Quality losses indicate the production losses due to defective parts. "Defective" here refers to all manufactured workpieces that do not meet the minimum quality standards for delivery to the customer. 

In the context of “first pass yield”, only parts without defects in the first production run are considered good items. 

This means: Both parts that cannot be reworked (“scrap”, e.g., defective surfaces or incorrect tolerances) and workpieces that can be "saved" with reworking are considered scrap.

In terms of the 6 Big Losses, quality losses are divided into two categories: Start-up scrap and Running scrap. 

Start-up scrap

After setup, many machines require a certain amount of time and material until the manufacturing process runs stably, the previous material has been used up, or the process quality has been checked and approved (example: stamping). The scrap that occurs until the process runs stably is referred to as start-up scrap.

Running scrap

Running scrap, on the other hand, occurs while the process is running stably. Examples include production outside of tolerances or too many defects in the final product.

What information must be recorded in the event of scrap?

To enable meaningful analysis of the scrap afterward, the following information should be recorded:

• Quantity of scrap

• Scrap category: Start-up or Running scrap

• Reason for scrap

• Order and product information

• Employees

• Additional description (e.g., comments on other observations)(Optional)

It is important in this context that the scrap is recorded individually for each scrap event, i.e., for each piece of scrap produced, the quantity and an individual reason should be documented. 

Only this approach allows for a granular analysis of the scraps. If scrap quantities are only recorded at the order/batch level, for example, it is not possible to distinguish between start-up and running scrap. This makes cause analysis impossible.

Benefits of a standardized scrap catalog

As with downtimes, indicating reasons is worthwhile to enable later analyses. Scrap should always at least be divided into the two categories start-up scrap and running scrap. Within these categories, further groups and individual reasons for scrap can then be documented. The level of detail at which this is done is decided individually. However, one should at least be able to recognize the most common causes of scrap to then initiate countermeasures.

To enable a good start, we work with our customers in an onboarding meeting to create an initial scrap catalog. This will then be iteratively refined over time by the production team. To facilitate this, we at ENLYZE have placed great value on ensuring that the catalog can be adjusted without IT support. This allows for quick adjustments without additional costs.

Are you wondering what a scrap catalog would look like for you? Please feel free to contact us by email so that we can discuss your specific case: hello@enlyze.com, subject: Scrap Catalog.

After creating the standstill catalog, scrap quantities can be assigned to individual reasons:

How is the quality factor calculated?

The quality factor relates the quantity of good produced to the total quantity produced:

The calculation of the scrap rate accordingly relates the quantity of scrap produced to the total quantity produced:

What are the challenges in recording quality losses?

To determine quality losses, it is necessary to determine the total quantity of material actually used for each order in addition to the scrap quantity. However, this actual total quantity is today rarely recorded accurately; it often relies on figures from production planning. 

The problem: 

The figures from production planning and reality often diverge. The discrepancy between planned and actual total quantity is sometimes so large that the inventory is exhausted. The lack of transparency regarding actual material consumption makes it impossible to determine a meaningful quality factor. Here too, it helps to accurately capture the manufactured reality based on machine data.

The solution:

Constant performance monitoring of the system using the start and end times of an order provides insight into this manufactured reality: When was produced, how much was produced. When the integral under this performance curve is calculated, the actual material usage is accurately obtained.

Thus, inaccurate data collection currently stands in the way of a clean calculation. With the correct, total amount of raw materials used and the scrap postings, the scrap rate, the quality factor, and the quality losses can be accurately determined.

Another challenge is determining when a potential quality problem and thus scrap is noticed. There is scrap that is only recognized later during quality control, which means that adjustments to scrap figures and other metrics must be possible throughout the lifecycle of the order.

What are the causes of scrap?

The good news: All causes can be comparatively easily addressed. Particularly with the help of machine data, quality losses can be recorded and avoided better than ever before.

Suboptimal commissioning or setup of machines

After setup, many machines require a certain amount of time and material until they can run stably. This scrap at the start of production cannot be avoided 100%, but it can be significantly reduced. Due to non-ideal start scenarios, the startup process often takes longer than necessary and produces unnecessary start-up scrap in the process. An analysis of fluctuations in start-up times and start-up scrap provides a good estimate of potential optimization.

Solution:

Through analyses, best practices for optimal start-up parameters can be derived. With the implementation of Standard Operating Procedures (SOPs), all workers have access to these best practices and can apply them. This leads to less fluctuation in start-up times and a reduction in start-up scrap. 

It is extremely important to involve the workers here and educate them. The following questions must be clarified with shop floor personnel: 

• Why are SOPs being implemented?

• How were these SOPs derived?

• What do you hope to achieve with them, i.e.,
  What goals should be achieved with the SOPs?

In answering all these questions, a good data basis helps to transparently present the necessary information and derived measures.

Poor maintenance of machines and tools

Poor maintenance of machines and tools affects the scrap rate multiple ways. Poor maintenance leads to suboptimal process flows, which increases running scrap. Potential unplanned downtimes when these errors are noticed not only reduce availability (and thus OEE) but also lead to increased start-up scrap when restarting the machine after downtime.

Solution:

Using data and historical comparison values, ideal maintenance intervals can be derived. This approach reduces unplanned downtimes and, thus, idle times and scrap. The foundation for this is a clean documentation of maintenance activities as well as downtimes and the resulting scrap production.

User errors

User errors often lead to non-ideal and fluctuating processes. They promote production outside tolerances and process defects like spots, ultimately leading to scrap, longer start-up times, and unnecessary emergency stops. 

Solution:

The greatest lever lies in continuous training of employees. Data can help to illustrate the actual size of the problem and bring everyone on board. Additionally, there’s again the possibility to work with SOPs and continuously improve them. The co-pilot from ENLYZE also offers a solution here, which points out fluctuating and incorrect process parameters. 

Manual processes

Manual data collection does not allow for continuous monitoring of the production process. This can result in deviations going unnoticed for a long time, leading to unnecessary scrap being produced.

Solution:

By continuously monitoring all relevant process parameters live instead of through manual sampling, problems are recognized more quickly and countermeasures can be initiated sooner. Often, early corrective action can completely prevent scrap.

Poor production planning and control (PPS)

Good pre-planning in PPS has a significant impact on start-up scrap. If the production program is well planned, a product change can be made without significant scrap production. However, if production is poorly planned, for example, due to the wrong sequence of colors of different products, it can have a significant impact on scrap due to purging. 

Color is just one of many relevant parameters that must be monitored. In reality, several dozen parameters must be considered in an optimization to find errors and identify necessary levers.

Solution:

Production planning is a highly complex optimization problem with many different input variables. Today, it is often still done manually, usually by a person without any technical support. 

Yet, we as humans are not suited to solving such high-dimensional optimization problems. At least not as well as modern statistical or AI-based algorithms. However, in order to apply these and train the AI models, the necessary data are currently lacking. Machine data collection can document the necessary training data and provide them to production planning.  The AI can then solve this optimization problem in favor of a lower start-up scrap and thus suggest an optimized production plan.

Poor quality of raw materials

The quality of the raw materials used has a significant impact on the manufacturing process. Poor quality raw materials lead to clearly more unstable or more fluctuating and thus harder-to-manage processes. 

Faulty and low-quality raw materials almost exclusively count towards running scrap. However, due to inferior raw materials, maintenance-related downtimes increase due to increased wear and contamination. The likelihood of a process-related stop and thus scrap during start-up should also not be underestimated.

Solution:

A correlation analysis of used raw materials and scrap can lead to interesting insights for both new and old materials. One example: One of our customers purchased the same raw material from different suppliers. The operations manager informed us in a conversation: 

“The raw material from supplier X always seems to run better than the others.” 

We conducted a correlation analysis between used raw materials, production speeds, and scrap, and lo and behold: The operations manager's hypothesis was clearly confirmed. 

• Production speeds were higher

• processes fluctuated less 

• and the scrap was lower. 

Even before ENLYZE, the operations manager had tried to prefer purchasing raw materials from this supplier. Due to the higher price, there were always obstacles to approval. With the analyses now conducted, the benefits were objectively demonstrated, a ROI analysis was conducted, and ultimately, the high-quality supplier was preferred.

Difficult-to-manage processes

Some manufacturing processes seem inherently difficult to manage, for example, due to strong process fluctuations. The running and start-up scrap can easily soar here.

Solution:

A product portfolio analysis shows which products produce above-average amounts of scrap. Because: High levels of scrap aren't always bad. Some products are harder to produce, high scrap is programmed, and also factored into the sales price. In this case, scrap from a financial perspective is not an issue.

The goal of the product portfolio analysis is to highlight the "problematic products" with high scrap. If there are products here that should run more stably, improvements must be made.