The Power of Hindsight: Tales from Process Experts

TrendMiner process experts look back on how they could have benefited from using self-service analytics in their previous roles as process, technology, control, and production engineers

At one time or another, almost everyone in the process industry has dealt with an issue and later looked back and thought, “If only I’d known!” Ah, hindsight – it’s a powerful thing.

Luckily we can use that power to make better decisions going forward. When you look at the knowledge and tools that are available now, you can understand how these would have helped you in the past, especially with your work. You can start to understand how these tools would have made your life so much easier.

Our team here at TrendMiner includes a wide variety of process experts and engineers who have walked in your shoes and are always coming across different ways they could have benefited from having a self-service analytics tool in their previous roles. (So much so that we recently presented a two-part webinar series entitled “From Hindsight to Insight” to share those examples with people who may be wondering: “What could I have done differently? How could I have made better – or faster – decisions using my data?”)

Team members often recount ways a self-service analytics tool would have helped them to solve problems faster and work more efficiently – ultimately making their jobs easier. They reflect upon their own use cases to talk about the insight into the manufacturing production data they could have had using this tool. In particular, they look at how they analyzed data in the past and the how they could have analyzed that data with self-service analytics.

A Couple of Time Travelers

Fredérick Motte
Fredérick MotteVP of Customer Success
Nick Petrosyan
Nick PetrosyanCustomer Success Manager

TrendMiner (and process) experts Fredérick Motte, VP of Customer Success, and Nick Petrosyan, Customer Success Manager, have been where you are now, and they recount tales from their days working and trouble-shooting in plants. Fredérick spent six years as an APC engineer and is an expert in public data analytics and process control input. He traveled the world as an analysis consultant for manufacturing processes. Nick spent six years as a production/technology engineer working in polymers and in a continuous gas plant and has a strong passion for manufacturing, working with operators, and optimizing processes. He especially enjoyed working on capital projects like debottlenecking, control system upgrades, and especially troubleshooting and working with operators to tweak the plant for optimization. Both experts are now part of the TrendMiner team specializing in showing how a self-service analytics tool can help process experts with their day-to-day operations. It’s only natural they would come across specific use cases they know they could have benefited from having a tool like this.

Saving Time & Contextualizing that Data

VP of Customer Success Fredérick Motte – being from Belgium – spent a lot of time in Europe, but his career also took him the United States, China, Russia, and even as far as Africa and Australia. So he’s seen a lot of different companies, a lot of different ways of working with data, and a lot of different levels of data maturity. Everything from companies that were still laying the foundation and setting up their data acquisition systems, to companies that were 90% automation driven, and everything in between.

Excel was and is still the most commonly used tool for diagnostics and troubleshooting in industrial manufacturing processes. However, experts struggle using Excel – it’s time-consuming, error prone, and unscalable for larger data sets. When it crashes some files can become corrupt. Moreover, mistakes are common with the slicing and dicing of the addition of formulas and data. These mistakes become compounded when working in a team because work visibility can be unclear, and mistakes made by one expert may not be easily picked up by the rest of the team. Also, Excel is not necessarily designed for time-series analytics so has limited analytics capabilities.

Self-service analytics democratizes the process data allowing experts at all levels to access and gain actionable information. It brings together the expertise and experience on the shop floor with the expertise and experience of the engineers, allowing personnel to tackle the data in all production systems for increased process insights.

TrendMiner’s self-service analytics software can be used to annotate and capture process context as part of the data. This information can be saved on the system and is then available for all personnel to access. He points out that one of his main challenges was making sure that operators were in the loop at all times.

“All those analytics features …Now what would have been the key benefits for me, I think, is always having the context of what’s going on. Literally, you’re never alone in the control room, but also figuratively, there’s a lot of things going on…Don’t hide it in Excel sheets, don’t hide it emails, make it part of the data itself, make it part of that one unified ‘Data Silo’ that everybody has access to,” recalled Fredérick, as he wistfully looked into the horizon. “TrendMiner would’ve helped me do in minutes and hours what cost me days. And I would’ve been able to save those days to actually do in depth analysis, not just do the mundane work of exporting data prepping and making sure that my Excel didn’t crash.”

For example, when changing tuning parameters, it is important to keep track of the  modifications. TrendMiner’s capability keeps all the experts in the loop at all times allowing for constant visibility of the whole production process. Frederick identified three key benefits of using a self-service analytics tool:

  • It allows for the contextualization of the operation making it part of the data, not just part of Excel sheets and emails, so important process information is accessible to all.
  • It saves time, doing in minutes or hours what would have taken days, freeing up time for experts to do the actual analysis.
  • It gives a head start on reporting and leaves a trail for successors.

Looking back, Frederick knows TrendMiner could have made his job easier and more efficient and can show how this tool allows experts to label, filter out, and visualize data. At the same time, the process team can more effectively collaborate and build KPI dashboards for monitoring production. Now as the head of TrendMiner’s Customer Success department, Fredérick makes sure our customers are just that – successful.

Speeding Up Root Cause Analysis & Improving Day-to-Day Workflows

‘Twas the winter of 2018

Customer Success Manager Nick Petrosyan was probably a little bit closer to the plant in his previous roles than Frederick was. Nick spent six years as a production/technology engineer at a continuous gas plant and prior to that he worked in polymers. But one thing is for sure – he has a strong passion for manufacturing.

“I really love manufacturing. I love working with operators, I love optimizing processes. And I really enjoyed working on capital projects like debottlenecking, or even control system upgrades. My office was pretty much in the control room. So a lot of my time was spent on day-to-day troubleshooting, working with operators to make little tweaks to the plant and optimizing things. When I learned about TrendMiner I really saw that it was a tool that I would have loved to have for that type of day-to-day workflow.”

Nick feels the root cause analysis diagnostic features would have been able to add the most value in his previous roles. The last two months of his previous job were spent dealing with plant issues due to an unusually harsh winter in the Gulf coast. His plant was halfway through startup when a critical valve at the top of a distillation column locked, which would have prevented the start-up. After several attempts at trouble-shooting, the team was successful at getting the valve to move and pop. He didn’t, however, have the time to look back at the data to see if there was an early indicator that the valve was starting to stick and malfunction. He didn’t have the time to download the data into Excel and study it. And if he did find a pattern in the data, how could he have set up a monitor to notify personnel in the future about a possible valve malfunction? With a self-service analytics tool, he could have taken immediate actions to analyze the data and set a monitor.

 Use Case #1 – Sticky Valve

Nick’s first example of “Hindsight to Insight” is how he would have used a self-service analytics tool when dealing with a sticky valve. Quintessentially, when a valve starts to stick, there is a delay between the valve output changing and the actual process responding. TrendMiner lets experts do different types of searches to identify and set up monitors for that type of behavior such as an area search or similarity search.

First, periods of normal and bad operation need to be identified. Normal operation should be labeled, so any period outside this zone can be found. Next, a search should be performed to find these periods. A monitor can be set, and if such a period occurs, the monitor can send emails to experts alerting personnel about the situation and also suggesting possible corrective actions. So when using a self-service analytics tool, experts can find an early indicator and take it all the way to a real time monitor that can alert personnel. Experts do not have to perform an incident investigation. Very simply, very easily, and very quickly, they can have these types of results.

Use Case #2 – Compressor Startup

The second use case Nick described is a two-day period that covers all of the states of a compressor startup. Using TrendMiner, Nick can see the shutdown phase with a lot of the flows and pressures reading zero or near zero and with the recycle valves open. During startup, these readings transition, first with some transient behavior until steady state is reached. Nick looks at the full data set of six of these cycles: six startups, six shutdowns, and six periods of steady state operation. He uses TrendMiner to key in on those startup periods to compare them to one another. Using the software’s searching capabilities, he selects a startup mode tag and tells it to find periods where it’s in the startup stage. All six startups are shown.

He simply clicks on one to have a visual look through the first startup and then can go to the second startup, and so on. Using the Google-like search results of the software, Nick shows how an expert can zip and fly through the data, just by clicking. An expert can also layer the search results on top of each other and can switch to the stack trend view which will help clean up all of the overlapping data. Each tag goes into its own swimming lane where experts can compare the startups to see what’s different.

Nick wants to know which startup out the six had the lowest vibrations throughout which is an early indicator that startup is running smoothly. Using TrendMiner, Nick can see if there’s any differences in the vibrations between these startups. By adding calculations on top of the searches to calculate the average vibration, the results show that three startups had high vibrations and three had low vibrations. Nick layers the good startups on top of each other to see what a typical low vibration good startup looks like. He saves this result as a fingerprint and labels it as a “Good Startup”. Now, Nick can set a monitor to alert personnel for startups outside this fingerprint.

With TrendMiner’s self-service analytics tool, Nick was able to perform

  • Discovery analytics and impact analyses.
  • Area and similarity searches.
  • Pattern recognition and advanced monitoring capabilities.

After interpreting the results, he could set monitors for process early indicators and suggest corrective actions to be taken.

Process Experts – Say No to Hindsight & Yes to Insight

Frederick and Nick make the case that TrendMiner’s Self-Service Analytics software is analogous to a catalyst in chemistry. With this tool, process experts can speed up their discovery, diagnostic, and impact analyses to gain insight into production issues. They can make data-driven decisions and actions based on this information.

This self-service analytics tool lowers the “human energy” required to tackle problems, so experts can incorporate much more “solution finding” time into their day-to-day roles. Both Frederick and Nick explain how they would have used TrendMiner to improve their work and thus plant operation.

Frederick shows how a self-service analytics tool can be used to contextualize process data keeping all personnel in the production loop. Using the software, Nick looks at finding an early indicator for a valve sticking which could prevent future plant shutdowns. He also studied the best rotating equipment startups and compared them to ones which had higher vibrations and machine stress.

Engineers can apply a self-service analytics tooling to a myriad array of manufacturing problems. This tool greatly increases the problem solving bandwidth of the typical plant expert and streamlines improvements to reliability, efficiency, overall asset effectiveness, and environmental and safety performance.

Now, you don’t just have to think about what you could have done if only you had a tool like this…

Stay tuned for next week’s blog where we bring you tales from two more TrendMiner experts about how they would have used a self-service analytics tool.

Want to see these examples for yourself?

Check out our free webinar on demand “From Hindsight to Insight” and see firsthand how Fredérick and Nick could have benefited from using TrendMiner.

Additional Resources