Put Manufacturing Optimization into practice

Anna-Katrina Shedletsky

Manufacturing Optimization is collecting manufacturing data from throughout the supply chain, analyzing the data to identify improvement opportunities, and then closing the loop with actions in the real world that improve core metrics like yield, gross margin, and throughput. Manufacturing Optimization has been the gold standard for improving a product or process in manufacturing for decades — and spans from one-time optimization exercises to incremental continuous improvements made bit-by-bit on factory floors. It’s possible to start with Manufacturing Optimization without significant infrastructure changes and limited investment.

Get the data: understand the inputs and outputs of your process

One of the first parts of Manufacturing Optimization is to aggregate useful data from across the supply chain. Useful is the operative word. Useful means the data has context and is high enough resolution to be meaningful. An example of a high-context, high-resolution data point is:

Unit DCG34GHX12, Speaker FR at 3kHz was 20.1 dB

Instrumental is a Manufacturing AI and Data Platform that enables brands to collect high-context data across their supply chain and pull it into one accessible place.

Do analysis: what changes to inputs will result in the outputs you want

The second part of Manufacturing Optimization is analyzing the data to identify opportunities to optimize and improve. It is possible to do this the old-fashioned way — with spreadsheets, statistics, and other analyses. If you’re only doing one optimization, doing it manually like this is fine. However, if you are running an ongoing process and expect to be doing continuous improvement, you can leverage modern software to help.

Instrumental’s Manufacturing AI and Data Platform leverages both statistics and machine learning algorithms to look at the problem from both directions:

  1. Identify inputs that will cause poor output
  2. Surface correlations between bad outputs and inputs you might be able to change

For the first item, the platform proactively identifies visual and test issues throughout the line and notifies engineers, who can immediately act to close the loop. These could be one-off defects found in visual data or a drift in functional test performance that triggers a Statistical Process Control (SPC) Alert.

For the second item, the platform can look at bad-performing units and then automatically identifies correlations to other unit parameters, test results, measurements, and even visual anomalies. This effectively helps engineers identify the potential root cause of the bad output to change the inputs needed.

Optimize: close the loop to get results

Once the key actions needed have been identified, you can close the loop by putting them into practice while monitoring to ensure you get the expected outcome.

For one customer who was concerned about product returns, Instrumental was able to help them to leverage data from their manufacturing process and return centers to identify opportunities for optimization. This resulted in the team finding an actionable root cause for some returns in minutes instead of days or months – this is the panacea of what optimization is about. Let’s explore how they used Instrumental to put manufacturing optimization into practice to achieve these results. 

A real-world example: preventing product returns

The customer engineering team wanted to understand why units were being returned and what they could change upstream in the factory to prevent those returns. With features in the Instrumental platform, the customer could pull data from their upstream suppliers in the form of images and functional test data. They also collected data from their final assembly, testing, pack-out facility, and returns center. Once all this information was aggregated, they could leverage Instrumental’s algorithms for analysis.

The customer had test data from their returns center, indicating poor performance for a specific test. They could look back at the performance of that test when it had gone through the manufacturing process and see it was marginal. They used Instrumental’s powerful Correlations feature to see if there were any other related properties between the failed units and those that were not returned. Using Instrumental’s unique Visual Correlations algorithms, the customer could see that the returned units actually had some damage to a specific component on the main circuit board. When they consulted their layout diagram, they realized that component was integral to providing the feature that had poor performance.

Beyond that, they could easily search through hundreds of thousands of production units they had already shipped and made to identify how many had the damage. This revealed another correlation: they all occurred around the same production week. The team was able to quickly understand the amount of exposure they had for the defect but also could build protections into their process to prevent the issue from coming back and causing additional returns.

Manufacturing Optimization today

Whether you engage with Instrumental to accelerate your Manufacturing Optimization journey or leverage existing tooling you have at your company — you don’t need to wait to be able to optimize your manufacturing and continuous improvement process. It’s easy to get sucked into the day-to-day fire-fighting. Our advice is to take a step back to get a broader view of how you develop and manufacture products — what in the process is causing the fire drills, delays, and poor field performance? Many times, it’s a combination of a lack of oversight (causing reactive posturing when problems happen) and an intended or unintended lack of access (making it harder to dive in and quickly get a resolution). What infrastructure or tooling could you build or buy to amplify or accelerate your existing team to deliver higher performance?

Related Topics

Manufacturing Intelligence

Don’t miss a post, subscribe today.