For this manufacturer of high-tech imaging systems, monitoring accuracy and product quality are critical. To help ensure its processes are top notch, the company aggregates and analyzes large volumes of streaming data from its manufacturing and testing processes, and inspects the materials it receives from several suppliers of highly calibrated technologies. Any quality problem could mean a part fails sooner than expected, or triggers a false alarm at a customer site that causes unnecessary panic. The worst case scenario would be technologies failing to detect real threats.

"One of our biggest challenges is our supplier processes," says the manufacturing quality engineer. "We monitor incoming parts because even a slight variation can have huge consequences. We have very strict criteria, and if we see deviations in specifications, other unexpected issues can emerge."

The quality team was spending 15 minutes to an hour trying to interpret the data in each test report. "During manufacturing, we could have four or five systems streaming data for 10 hours a day—mostly lists of numbers. In Excel, it initially looks like nonsense. We once had a problem that we didn't discover for a long time, and it cost us a lot of money."

By setting up automated manufacturing analytic workflows with the TIBCO® Data Science platform, the company can now complete complicated processes in just a few minutes. "With the TIBCO Data Science system, we catch problems very early on in the process. The number of improvements we've made is immense, and the overall cost of doing business has decreased significantly.

Increased Quality and New Knowledge

"The TIBCO Data Science system helped us improve product quality by decreasing the variability of everything we produce. It's allowed us to discover things that we weren't able to detect previously."

One business unit's TIBCO Data Science installation references the company's main database, which receives data from various input sources. Automated alerts and reports are triggered and sent out to various stakeholders as needed to handle issues before they become problems. "With TIBCO Data Science automated analyses, I get the results I need in a matter of minutes, so I can spend much more time on other areas of my job."

Cost Savings and Production Insights

TIBCO Data Science automated analytics has also reduced one business unit's cost per unit by 50 percent. This is due to the team now being able to better understand key drivers of the production process and make adjustments. "The biggest impact the TIBCO Data Science system has had is in preventing production downtime. In the past, we’ve had issues with downtime that could take months to fix, and that's a huge loss in revenue."

Now, if something looks odd in a report, the quality team is better equipped to ask and answer questions before shutting down a manufacturing process. Is the issue critical? Will it autocorrect? The next round of data provides fast answers. If something is truly out of specification, the system can then be shut down to avoid rework and allow further investigation and analysis to find out what happened and what to do about it.

"I'd say the TIBCO Data Science system has reduced product revenue losses by about 30 percent," says the engineer.

Better and More Efficient R&D

The research department also uses the TIBCO Data Science system, particularly for its Design of Experiments (DOE) feature and predictive analytics. Before, some R&D experiments could take a week or more to complete. Now, the same experiments can be completed in just a few hours. Predictive analytics is used to test a particular scenario, and then predict the result of continuing further along that course. This technique helps the team identify when to change course, or when to keep going because the design appears promising.

"The DOE feature has been huge in helping us develop new products and further our capabilities. It's one of my favorite features," says the engineer.

"The TIBCO Data Science platform has already paid for itself several times over. In fact, I was just asked if we could expand its use to some of the other processes and divisions."