It’s 2 AM. Do You Know What Your Parts are Doing?

By Alex Benham

Picture this:

You’re lying in bed, fast asleep, when your phone rings. First you’re mad at yourself that you forgot to put the darn thing on mute, but then you pick it up and look at the caller ID.

It’s your boss.

Or, even worse, it’s your boss’ boss.

You almost drop the phone in surprise, but immediately answer.

“Listen,” a voice says in your ear. “Something’s happened.”

You sit up, throwing the covers onto the floor. This can’t be good.

Does this sound a little too familiar? Maybe it wasn’t at 2 in the morning. Maybe it was in the middle of dinner, or while you were grocery shopping on a Saturday afternoon, or sitting at your kid’s little league game. Whatever the situation and whatever the message, 9 times out of 10, getting an after-hours call from a higher-up is not a good thing.

Most likely, a part has failed, and, aside from figuring out how to fix it, the big wigs are trying to find two things:

1. What caused the failure
2. Whose fault it is

To put it another way, I had a friend who’s dad always said there was just one fundamental equation in aerospace that has to remain constant:

The number of takeoffs have to equal the number of landings.

If you don’t have confidence in the quality of your parts, your landings are not going to equal your takeoffs. (And you’re going to get a lot more after-hours calls than you’d like.)

That’s a problem that we’re trying to solve here at Sigma Additive Solutions. Our solution? A suite of software that will give you the in-process monitoring and data you need to send each part you manufacture out of the door with the confidence (and proof!) that no 2AM call will be necessary.

Our Machine Health module takes a look at variance within production fleet machines. Using the machines’ built-in sensors and simple thresholding, standard WACO Run Rules, and advanced statistical analysis, it can monitor the machine’s performance to identify sudden changes, steady trends, and non-conformance of key parameters of the machine environment.

Even better, the module standardizes the data across machines so you can understand variance within a build, across machines, and throughout production runs.

Key features:

• Convert varying machine log files and live-streaming API data to a standards-based format
• In-app data analytics, monitoring, and dashboarding
• Build to build and machine to machine data comparison
• Automated reporting tools for ease of qualification and production

The Process Health module uses a machine’s existing cameras to monitor powder bed builds and identify quality failures in real time. Paired with machine learning and image recognition, this module can identify the overwhelming majority of build failure causes–from streaking to chatter to protrusions–in real-time, without the need for constant human oversight.

With the ability to customize the alerting criteria, Process Health can provide a comprehensive analysis of powder bed disturbances, and help you quickly identify and prevent build failures.

Key features:

• Camera and Thermal Camera modules for image based defect detection
• Sigma training sets offer standards-based definitions for a defect detection library
• Allows for use with 3rd party machine learning and artificial intelligence approaches, as well as Sigma’s sensor fusion approach
• Correlates to layerwise machine health data with pixel correlation between cameras and melt pool data

Using on-axis photodiodes, Part Health detects micro-scale defects in melt pools in real time. With the analysis capabilities for multi-laser systems, targeted defect analysis, and layer to layer connection filtering, this module offers deeper insight into the root causes of defects. A perfect complement to CT scanning, it is cost-effective for in-process inspection and early detection.

Using the data that the photodiodes capture, Part Health also constructs 2D and 3D visualizations of individual parts or the entire build to and creates automated reports with a simple pass/fail judgment. ​

Key features:

• Sigma’s proven melt pool monitoring and analytics applied to any OEM data, Sigma retrofit, or integrated hardware (like Novanta Firefly 3D)
• Provides standards-based comparison tools across varying fleet monitoring
• Links to Machine Health layer data and Process Health camera-based pixels for data fusion
• Machine learning and artificial intelligence approaches included for defect detection at a part level (lack of fusion, porosity, etc.)