Describing Data Collection Functions

SAP Digital Manufacturing provides Data Collection features that workers can use to collect assembly data and assembly process data. You can use the collected information in the following example scenarios:

• Monitoring the quality of the production process

• Documenting quality-related or other manufacturing data for individual SFCs that you can analyze in case of deviations and/or warranty claims

• Documenting the manufacturing process in a regulated environment, such as the manufacturing of medical devices

• Providing input data for statistical analysis of the production process, such as statistical process control (SPC)

The following image shows an example of how you can collect assembly data and assembly process data:

In our bike manufacturing company, we produce finished bikes through several manufacturing operations, such as assembly, quality inspection, and packaging. The first and second manufacturing operations require the worker to gather various data points:

• During assembly, the worker records the serial number of the assembled frame, front wheel, and rear wheel. In addition, they also note the torque used when attaching the wheels to the frame.

• During quality inspection, the system provides a checklist where the worker gathers results from the in-process quality inspection. They can collect numerical results (for example, the tire pressure), binary values (for example, whether the brakes are working or not), or coded values (for example, the presence of no scratches, superficial scratches, deep scratches, and so on).

What is the difference between these two scenarios? During the assembly operation, the data the worker records are linked to the specific component being assembled. For instance, the serial number of the assembled frame uniquely identifies the frame used in a particular bike. On the other hand, during the quality inspection, the data the worker records are tied to the bike currently being manufactured and may not necessarily be traced back to an individual assembled component. Therefore, the first example corresponds to a manufacturing scenario where you require data to be recorded at the component level, while the second scenario exemplifies a manufacturing situation that requires data to be recorded for the assembly process itself.

So far, you've learned that the worker can record data when assembling a component. Let us now dive a little deeper and explain why the worker should do this. The following image shows a genealogy record for a manufactured product, for example, a bike:

In our example Bike Manufacturing company, we manage the bike as a lot-size one material, so that the SFC number the system creates during order release corresponds to a uniquely identifiable item, for example, one bike with the serial number, 4711.

The manufacturing process consists of two steps: In the first step, the worker assembles semi-finished components, for example, wheels, using different raw materials, such as tires and rims. In the second step, the worker uses the semi-finished components (wheels) and other raw materials (frames and light sets) to assemble the finished product. For traceability reasons, you need to know which components were used at which manufacturing step to assemble the finished product. This process is called discrete component traceability, and you can collect all information required for this process in SAP Digital Manufacturing. How does this work?

The process starts at the raw material level. When warehouse personnel receive the components from the suppliers and create inventory in SAP Digital Manufacturing (either directly or indirectly via integration with SAP S/4HANA), they record information provided by the suppliers, such as vendor, vendor lot, vendor serial number, manufacturing date, and so on. In the next step, the system collects, for each manufactured SFC (for example, for the semi-finished or finished product), which component was assembled at which operation, and offers the possibility to collect additional data, such as inventory ID, serial number, assembly-related information, and so on, for each component. After the worker records this information in the POD, the system adds it to the as-built record of the SFC. If you require further information, such as process- or quality-related information, you can additionally record assembly process data for the SFC to enable full traceability.

Collecting genealogy data provides visibility and traceability of the components, for example, in the case of product claims. A typical business use-case follows the top-down approach from product to components: If a customer files a complaint for a purchased product, such as a defect, the as-built record of the unit is available to the Quality Department and helps to quickly analyze the issue by providing information about the manufacturing process and the assembled components. Assuming the root-cause of the product defect is a defective purchased component, you can identify and involve the responsible supplier into the process.

After identifying a defective component, the Quality Department can trigger a follow-up process using the bottom-up strategy. They can track the path from the defective component to all finished products where this component was used. Assuming that some finished products are still in the warehouse, the Quality Department prevents them from being shipped to customers before the defect is fixed via rework. In the worst case scenario, the respective finished goods have already been shipped to customers. In this case, the Quality Department proactively informs the respective customers about the defect. The latter can then initiate containment measures to prevent further issues.

The following figure shows how you can use assembly process data to document your manufacturing process and to perform in-process quality inspection:

The bike manufacturing process consists of several manufacturing operations that your workers perform. In our example, they first undertake the assembly operation, followed by a quality inspection. If the quality inspection is passed, they package the bike into a transport box.

Your Master Data Specialist can attach one or more data collection groups to each manufacturing operation. For example, during assembly, the worker documents assembly parameters for wheel and fork assembly, respectively. Each data collection group contains one or more parameters. For example, the torque and the screwdriver used when attaching the wheels to the frame (for example, wheel assembly parameters), and the fork pressure (for example, fork assembly parameters). During the quality inspection, the SAP Digital Manufacturing system displays an inspection checklist for the worker to record various inspection parameters. In our example, the worker records numerical results (for example, front and rear tire pressure), binary values (for example, whether the lights a working or not), or coded values (for example, no scratches, superficial scratches, deep scratches, and so on).

What is the difference between the parameters a worker records in the assembly and quality inspection operations, respectively? During assembly, the worker simply records the as-is data to document the details of the manufacturing process. For example, they note that they used a screwdriver with the ID, 57745, and the maximum torque applied when they tightened the wheel's screws to the frame was 15 Nm. The system stores this information in the data collection record for the SFC without further evaluation. From a business perspective, you use this scenario when you require information about the manufacturing process, for example, for further analysis or due to obligatory documentation requirement.

During quality inspection, the worker collects information about the manufactured device (or the manufacturing process), for example, the as-is values of the front and rear tires, respectively. During the design phase, engineering specified an acceptable pressure range (for example, 2 - 3 bar). Depending on the measured value, the inspection characteristic either passes (if the value is within the range) or fails (if the pressure is outside the range). Similarly, if the lights don't work, the system rejects the respective inspection characteristic. Unlike the recording of as-is data, the system now evaluates and reacts to the information the worker entered. As depicted in the following image, there are two possible scenarios, depending on the outcome of the quality inspection:

The worker conducts the quality inspection and records the results as depicted in the figure. In our example, the pressure of the front tire (3.5 bar) falls outside the acceptable range (2 - 3 bar) and the lights are non-functional. Consequently, the quality inspection is unsuccessful, and the system automatically directs the SFC to an analysis and repair station. Here, a worker investigates the quality issue. If they can resolve it, they document the process in the system, then direct the SFC back to its original routing to perform the next manufacturing operation, such as packaging. If the worker cannot fix the defect, they scrap the SFC.

As illustrated in the figure, the system only initiates the analysis and repair sub-process if the worker detects a defect. If quality inspection is passed, the system directs the SFC to the next operation of the routing, for example, packaging.

You've learned in the previous chapter that the worker can perform in-process inspections using SAP Digital Manufacturing. To standardize the inspection process, the system provides a checklist to the worker, outlining the quality-relevant parameters they must check. Also, if an inspection parameter fails, the system can automatically generate a nonconformance record to initiate the analysis and repair process.

The inspection process often involves steps where the worker visually examines the device. If any defects are found, the worker manually creates a nonconformance record to document the details of the defect. By using a machine learning model in SAP Digital Manufacturing, you can further standardize the visual inspection process. By using cameras, you can capture images of manufactured materials, automatically analyze them, and identify potential defects. In the following example, the worker performs a visual inspection task in a Production Operator Dashboard using a camera:

On the left side of the screenshot, you can see a picture of a manufactured device. In our example, the camera took a picture of the identification plate of the bike. The identification plate shows manufacturing data, such as the production order number, serial number, and so on. On the right side of the plate, additional information encrypted in a QR code and an RFID tag is available. Both are automatically printed during the manufacturing process and glued on to the plate. As you can also see, the second sticker isn't correctly attached.

On the right side of the screenshot, the SAP Digital Manufacturing system displays a list of possible defect codes. Without machine learning, the worker must decide which defect code best fits to the observed deviation and manually select the respective code from the list. In our example, the system proposes defect codes by automatically analyzing the picture on the left using a machine learning algorithm. The algorithm was trained beforehand and is able to propose defect codes for visual inspections of the identification plate. As you can see, the algorithm proposes various defect codes (for example, defective sticker, QR code missing, card missing, and wrong sticker position) sorted by descending probability. Indeed, the defect code with the highest probability value (defective sticker, 99.3 %) correctly describes the observed defect.

The worker now chooses the Log NC button on the right, and the system automatically creates a defect record for the SFC. Then, the system directs the SFC to an analysis and repair workstation. As the system has automatically linked the image to the defect record, the worker responsible for analysis and repair promptly understands the nature of the defect and proceeds with the necessary repair tasks (for example, removing the old sticker, printing a new one, and correctly attaching it to the identification plate).

Technically, SAP Digital Manufacturing offers an API for data collection that allows you to upload images from cameras positioned on your shop floor. This enables you to automate the collection of images and inspection results for future review and analysis. Also, you can download the data as analytical data sets to train new machine learning models.