• Subscribe

Improving operational efficiency through better data visualization

Make easy, consistent, and visually appealing charts for data-driven decisions using SmartFactory Material Control and APF Reporter

In my former roles as AMHS Manager and Automation Technology Development Leader, I’ve sat through fab “Ops Review” meetings where each module owner is responsible for highlighting relevant issues impacting fab operations. In those meetings, while working at three different semiconductor companies, I found a common link between these different fabs—a link pertaining to the consistency of data visualization. Several commonalities exist, including a fab dashboard for 30 days showing moves per hour (MPH), delivery time, alarm Paretos over time, and storage utilization ordered by bay.

In this blog I focus on the 30-day AMHS dashboard report. The idea is on a single chart—30 days’ worth of trends, commonalities, and irregularities are communicated at-a-glance. Figure 1 shows the typical dashboard I’ve observed at most fabs.

Figure 1 30 Day Dashboard
Figure 1. A typical 30-day trend dashboard showing daily delivery time and move counts

The dashboard in Figure 1 consists of x, right, and left axes, explained next.

  • The bottom x-axis shows the time under review, in this case a 30-day trend of 24-hour movement statistics.
  • The right axis shows the average and 95th percentile delivery times achieved by eliminating the outliers (the top 5% times). This is represented in the two-line charts and is the bread-and-butter KPI metric for AMHS optimization. It is a lagging indicator of how efficient the storage optimization and move dispatching capability may be in the factory. Naturally, the focus is drawn to the outliers where delivery time is trending higher. These trends usually require an explanation to management in an Ops Review. Outliers are sometimes tied to a new tool or AMHS bay coming online and usually indicate the need for better organization of up-stream storage locations or improved scheduling logic to ensure material is stored closer to the tool, which leads to shorter delivery times.
  • The left axis shows the total moves per 24 hours (MP24) using a stacked chart representing three move types: stocker-to-stocker, stocker-to-tool, and tool-to-tool. The stack total represents total moves within a 24-hour period. The stacked chart is interesting data to visualize because this is where we can glean clues of move type trends over time. For example, some days a rush to storage is evident, while on other days there is a greater emphasis on tool moves. These trends are typically aligned with factory-specific milestones that can be highlighted as required.
It’s important enough to explain these chart elements—MP24, average DT per day, and trends over 30 days—because then can we realize the value of it. A fab manager, for example, might wonder why on the 19-Sep mark that the average DT spiked compared to other days, and here you can answer your Ops team with specific reasons, such as a new product introduction (NPI), a run-on at Wafer Starts with new test wafers, or a new bay coming online. Often the gap between the 95th percentile and the average delivery time can also indicate that on that specific day, a bottleneck likely occurred, or some odd disruption triggered this event, which is seen in the day-over-day trend.
If we can agree this chart has value and is common in most Ops Reviews, I will next show how we create this chart in CLASS MCS 5® and then visualize it using APF Reporter.

Step 1 – Analysis of a Move

We need to first define what a move is and have a clear understanding of all the elements that contribute to a move from the AMHS perspective. Figure 2 shows a breakdown of the anatomy of a move in a 300mm semiconductor factory—from first requested to transfer complete.
Figure 2 Move Anatomy
Figure 2. Anatomy of a move and event triggers

In Figure 2 the Delivery Time is the amalgamation of several events occurring, most notably the sum of assign-wait, retrieve, and delivery times. The other key point here is the definition of a move. A move begins with the MES move Requested and ends with the overhead transport (OHT) event of Transfer Complete. It’s important here to point out that a stocker-to-tool move will consist of one macro-Transfer Complete move (from the MES perspective), which simultaneously means that we will have two micro device complete moves (one from the stocker, one from the OHT). Distinguishing between MES moves and device moves is key to how we store this information in MCS statistics tables.

Step 2 – Data Organization

Here then is the secret sauce to all of this—the data organization. The CLASS MCS 5 data storage structure, illustrated in the anatomy-of-a-move in Figure 2, is all inclusive into one single table called SCARMOVE_SUMMARY. This table represents the infusion and feedback from our customer user groups, where we achieved alignment on a single storage organization strategy. The table looks something like the one shown in Figure 3.

Figure 3 Scarmove
Figure 3. Sample SCARMOVE SUMMARY table
This table identifies all the relevant elements of a move: the MES request time, vehicle assign time, arrival time, delivery time, command ID, and so on. Factory data analysts who need to build this type of chart will quickly realize the value. A single table to represent all data in the anatomy of a move is a unique accomplishment of SmartFactory Material Control (CLASS MCS 5).

Step 3 – Data Visualization

And now the sleek part—taking the data visualization to its completion. With all the elements of the move properly defined, categorized, and stored in a single database table, we can now leverage the APF Reporter tool using the MCS Delivery Time Report template, available with CLASS MCS 5.14. The APF Reporter template has been designed with our users in mind, considering the Ops Review meetings they’ll need to attend. Figure 4 shows a sample chart created using APF.
Figure 4 Delivery Time Report
Figure 4. MCS Delivery Time Report created using the APF Reporter tool
Our current customers have been asking for this reporting capability for several years, and now the APF template used to create this chart will be available with MCS 5.14 in March 2023.

Ready to contact us to learn more about this or other charts to improve productivity?

About the Author

Picture of John Robinson
John Robinson
Connect with us on LinkedIn to begin your journey towards creating a world class factory today.