Time Travel with an Experienced Process Expert

In this next installment, another TrendMiner process expert looks back on how he could have benefited from using self-service analytics in his previous roles as a Production Engineer and Engineering Specialist

According to the Cambridge Dictionary, hindsight is defined as the ability to understand an event or situation only after it has happened. As process experts, surely you can recall production situations in which you wished you’d had hindsight, situations that you could have solved easier and faster if only you had certain knowledge and tools. Perhaps, a tool like self-service analytics.

TrendMiner’s team includes a group of experts with diverse backgrounds and experiences in industrial manufacturing who understand what process experts have to contend with in their day-to-day operations. In a two-part webinar series “From Hindsight to Insight”, our process experts take you back to their previous jobs to show you what they could have accomplished with a self-service analytics tool.

This post features Data Analytics Engineer Eduardo Hernandez. He will show you how this tool would have given him the insight into his plant’s manufacturing production data and the context capabilities to make data-driven decisions and actions to streamline his work and maximize operations.

Look Who’s Talkin’

Process expert Eduardo Hernandez has certainly walked in your shoes. He’s held roles as a Production Engineer and Engineering Specialist. With his background, of course he can easily look back and recount examples and use cases of how he could have benefited from a self-service analytics tool. But first, he assesses the limitations of using Excel for time-series data and then explains the objective and benefits of bringing together process experts and data analysts with self-service analytics.

Eduardo Hernandez

“I really like the opportunity to interact with customers from different industries (oil & gas, chemical, pharmaceuticals, food & beverage) and help them solve their process problems using my knowledge of the self-service analytics. With this tool, they can solve their own production issues. For me, ‘self-service’ really means intuitive and easy to use.”

Eduardo Hernandez

Tales from an Oil Refinery

After graduating from the University of Houston, Eduardo started working for an oil refinery plant in Puerto Rico. This refinery would process and blend collected lube oil to create high grade lubricants. Additionally, the plant would package the lubricants into bottles to sell both locally and internationally. Eduardo’s responsibilities included overseeing the blending and packaging operations and helping the refinery engineers to solve process problems. Recently Eduardo recalled specific use cases where he could have used TrendMiner to allow him to do his job easier and better.

Use Case #1 – Quick Insights into Total Throughput & Setting Monitors for Premature Catalyst Deactivation

After the oil was collected, it would go to the upstream operation to be hydro-processed and refined. Next, it was moved to the midstream operation where different oil blends were created, for example SAE30, SAE10W-30, SAE5W-20, and others. The downstream operation included the filling and packaging lines. The first use case is around the hydro-processing and refining stage. Whenever lube oil is collected, it must first be pre-processed to remove some of the solids and water. After this, the oil was fed into two catalytic reactors. The pre-processed lube oil and hydrogen would flow into these reactors at high pressure and temperature over a catalytic bed to remove sulfur, nitrogen, and other contaminants. The next step generally consisted of separating the hydro-processed oil into three different cuts in a distillation column. During this process, one of the common issues the plant faced was premature catalyst deactivation, an issue not easily explained and solved.

A common question concerned throughput and whether or not the plant was running at the same throughput at previous production times. TrendMiner could have been used to answer this question quickly and easily. Another question was around inappropriate startups. Engineers suspected that some operators would start up their process differently from others, which is a logical scenario. TrendMiner would have done a great job at giving insight into and at monitoring the process at this stage.

To solve the throughput issue, process experts can use TrendMiner to do a value based search to find all events when the oil was flowing in the plant. In Eduardo’s use case, four events are found: a six-months event, a five-months event, and two seven-months events. These campaigns are different from one another in terms of duration, but the amount of throughput of these events needs to be compared to gain valuable information. Using TrendMiner, Eduardo adds a calculation on top of each search results to calculate the total throughput of the plant. For this first period of time, almost 1 million gallons were processed. There was a drop in the second period processing close to 800,000 gallons, and over 1 million gallons were processed during the last two campaigns. Clearly in the second campaign, something caused the catalyst to deactivate prematurely. The next step is to investigate the reason for this difference.

In TrendMiner, process experts can click on any of these search results to look at the data in more detail to investigate what was going on. Additionally, the first three events can be overlaid for comparison. The flow going into the reactor for these events can be looked at to see how the start up for these reactors were initiated. The results show that the flow rate going into the reactor and the temperature slowly increased with time. However, for the campaign with the lower throughput, it showed that at first, the flow rate went up steadily, but then there is a big increase in the flow rate. This spike would have definitely cost some heat spots in the catalyst, which would have led to its premature deactivation.

Experts can use TrendMiner’s self-service analytics to monitor flow rate and temperature and to avoid this situation in future runs. First, a fingerprint for a good startup needs to be created. This fingerprint is the ideal area of operation for the reactor start up. Second, a monitor needs to be set. If at any point in the future, there is a deviation from this fingerprint, an email is sent to the operators alerting them that there is an issue with the flow rate and also suggesting corrective measures.

Use Case #2 – Improved Analysis of the Blending Process for More Efficient Production Scheduling

For this refinery, more than 40 different types of products were blended in different tanks having different sizes and geometries. The operation was complex. The challenge was to have a better way to review previous batches and also to estimate blending times.

Process experts would meet every week to plan the operations and the different batch productions for the coming week. They needed an accurate estimation of blending time for each product, a critical aspect for the continuous operation of the business. TrendMiner could have helped.

For each batch, TrendMiner would have let experts see the data and look at what was going on in the process. The start and end times as well as the total time it took to create the batch is shown. Context information about each batch could have been added and saved – information such as the which vessel and additives were used, the lab approval, the total production volume, and the product that had the highest demand. All of this information gives an accurate overview of the time required to make a new batch of each product and would have allowed experts to more efficiently plan the coming week’s production schedule.

Use Case #2 – Improved Analysis of the Blending Process for More Efficient Production Scheduling

For this refinery, more than 40 different types of products were blended in different tanks having different sizes and geometries. The operation was complex. The challenge was to have a better way to review previous batches and also to estimate blending times.

Process experts would meet every week to plan the operations and the different batch productions for the coming week. They needed an accurate estimation of blending time for each product, a critical aspect for the continuous operation of the business. TrendMiner could have helped.

For each batch, TrendMiner would have let experts see the data and look at what was going on in the process. The start and end times as well as the total time it took to create the batch is shown. Context information about each batch could have been added and saved – information such as the which vessel and additives were used, the lab approval, the total production volume, and the product that had the highest demand. All of this information gives an accurate overview of the time required to make a new batch of each product and would have allowed experts to more efficiently plan the coming week’s production schedule.

Use Case #3 – Contextualizing & Tracking Downtime as Part of the Preventive Maintenance Program

The blending operation was complex; however, the filling and packaging line operation was even more complex. More than 100 different products filled bottles/containers ranging from small four ounce bottles, to conventional one gallon bottles, to 55 gallon drums and larger. This operation was carried out in four different filling lines and had many challenges. One was to remember the production of the previous week since its parameters changed weekly, for example, from one production to another, from one bottle type to another, and from one oil blend to another.

Having a quick and easy way to review past productions was critical for planning purposes. This is where TrendMiner could have been extremely useful. Experts could have looked at each week’s data to review production over the week in order to plan the next week’s production. Additionally, experts could have input and saved important context information about production such as which line was used to fill and package each product.

Another production challenge was tracking downtime as part of the preventive maintenance program. Traditionally, downtime was tracked using either an Excel file or even a pen and paper. Part of this tracking was figuring out the notes that different operators made about the reason a line had stopped. Experts needed this information to find solutions and prevent additional downtimes from happening.

TrendMiner could have helped with process aspect as well. Using TrendMiner, personnel could have recorded relevant context information. With a simple click on any context item, experts could have retrieved information on what caused the filling line to stop for that event. Some examples included problems in the capper, labeling machine, and filling station. More significant were the downtime events due to problems with a poor nozzle-bottle seal caused by a broken spring and a worn O-ring.

Having this last piece of information would have been particularly helpful as experts would have known to place an order for spare parts like springs and O-rings as a proactive approach to ensuring sufficient stock parts for the filling lines. TrendMiner also allows personnel to upload documents, e.g. purchase orders or invoices, thus preventing this information from getting lost in emails or in  computer files.

Self-Service Analytics – the Insight Process Experts Need

Eduardo showed how TrendMiner allows process experts to provide more efficient and effective contribution to day-to-day challenges – afterall it was designed by engineers, for engineers.

No matter what your manufacturing process area, you don’t have to have that “A-ha, hindsight” moment, rolling your eyes in frustration. You can use self-service analytics to truly get a handle on your operational processes allowing you to streamline your work. Take it from an experienced process expert, “Self-service analytics is the insight you need.”

Want to see these examples for yourself and get the full slide deck?

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

Additional Resources