Sitech

Marc Pijpers I Sitech Services

“Our experience with TrendMiner is that it is indeed a proven efficiency tool that really adds value.”

Sitech logo whiteSITECH SUCCESS STORY
CP Kelco

Norman Ridgley I Senior Manager Operational Excellence

“TrendMiner was a key support tool in identifying process deviations.”

CP Kelko white logoCP KELCO SUCCESS STORY

Manu De Block I Project Engineer

“TrendMiner is a powerful search tool – the Google for industrial data.”

DEME SUCCESS STORYDEME white logo
Ashland

Jan Meireleire I Engineering Manager

“We wouldn’t be where we are today in over-achieving our goals without TrendMiner.”

ASHLAND SUCCESS STORYAshland logo

The rapid adoption of TrendMiner is due to the value it delivers to our customers. Our software is used for a wide variety of ways to improve overall profitability for batch, grade, and continuous production processes.

See how others use TrendMiner and find out how it can benefit your organization.

TrendMiner General Use Cases

Unplanned downtime has a huge impact on the overall profitability of the plant. Process expertise is not always sufficient or readily available to find the root cause.  With TrendMiner, you can use best process performance patterns to build “golden fingerprints”. When using multiple related tags up- and downstream in the production process, you can set early warnings in case of process deviations.

By using the lessons learned in the past, control room personnel or field engineers can get valuable information to take appropriate action and avoid unplanned process downtime.

Malfunctioning of the steam measurement in a column led to an increase of the column bottom temperature, and often resulted in a column trip. Through use of TrendMiner’s similarity search, the root cause was found to be peaks on the steam net. By setting fingerprints on both the steam measurement and the bottom temperature, early warnings were set to avoid trips of the column. This helped to avoid unnecessary process downtime and the related losses.

Advanced industrial analytics can help to assess which batch or grade shifts are the best and fastest, and monitor the shifts. If shifts are slow, TrendMiner can also be used to search for the factors influencing this bad behavior. The combination of root cause analysis and identifying the best performing shifts help to improve overall plant performance, resulting in higher on-time deliveries and overall higher production level with high product diversity.

For each production run of a specific alkane derivative grade, the conveyor system got jammed with small rubber particles. Because the cause was unclear, the only solution was to switch to another grade, which resulted in lost production time, additional scheduling overhead and labor to change additives.

With use of TrendMiner, the good and bad production runs were analyzed over large tag sets and  the root cause turned out to be a pressure drop in an upstream reactor before the onset of the pressure rise in the conveyor system.

The next step was to start monitoring the reactor pressure with TrendMiner in order to better control the reaction pressure. The overall result was a 15%+ efficiency increase during production of this specific grade, which directly impacted the plant’s profitability.

With TrendMiner you can analyze all historical production data and assess when the process was least efficient in using raw materials or delivered a poor-quality product. The factors influencing the anomalies are not always clear, but they must be identified in order to reduce or even avoid waste. TrendMiner’s advanced analytics capabilities can identify such factors without the need for data modelling, helping you achieve results faster.

Adding base to the mixing tank led to temperature increases which correlated with the success rate of the mixing. The inconsistent process resulted in an overly wide product quality variation and too much waste.

With use of TrendMiner, the direct correlation was proven between product quality and temperature in the mixing vessel during dosing. Once a process monitor for controlling optimum mixing conditions was created, it reduced product waste and also indicated when the mixing unit required maintenance.

Besides reducing waste in a production process, TrendMiner can help to identify bottlenecks in the production process. The user can look for influence factors based on their own assumptions, or they can ask TrendMiner to suggest the most likely influence factors for a specific process behavior. When the root cause is found, field settings can be optimized or enhancements to the production line can be implemented to increase overall production yield.

Within the production process of resins, foaming in the flush tank occasionally happens, blocking the subsequent pipes and depositing resins on the instruments. Expensive industrial cleaning is needed to remove the spilled resin.

With use of TrendMiner, it was discovered that foaming periods can be detected (post facto) by a decrease in the vapor temperature measurement, which was caused by resin deposited on the sensor. Further root cause analysis showed that foaming problems mainly occurred when a large quantity of flushing medium was pumped back into the tank, specifically after producing a very viscous product.

By changing the procedures and production schedules, foaming now rarely occurs. This resulted in reduced cleaning costs, increased asset and instrument reliability and increased overall yield.

Once a golden fingerprint is created based on best process performance profiles in the past, it can then be used to monitor the live process. With TrendMiner it is possible to receive automatic notifications if the actual process behavior deviates from the golden fingerprint for a certain amount of time.

For example, if the pressure is deviating from the fingerprint for more than 10 minutes, a notification can be sent to the control room to adjust the process and get it back within the boundaries of the golden fingerprint. This ensures products can be delivered at the quality that was ordered by the customer.

To further enhance quality control, lab test results can be imported into TrendMiner to validate the best production runs and ensure the golden fingerprint is based on lab facts.

Regular issues with the downstream drying step of a crystallization unit led to a loss in production time and product quality.

The subject matter expert developed an optimal multivariate crystallization profile with TrendMiner. The ideal drying step resulted in a collection of other optimal crystallization profiles, which were overlaid to create the golden multivariate crystallization fingerprint. This fingerprint included a pressure dependency for the drying step to get a steady production process and a reduction in blockages.

With the more reliable process and the created alerts in case of deviations, the overall production time was decreased and product quality could be better controlled.

TrendMiner can monitor the live process and even predict process evolution based on similar process evolution found in all available historical data. TrendMiner’s patent-pending pattern recognition technology can be used to send soft alarms, well before a hard alarm would go off. This helps to avoid hazardous situations on-site and prevents unnecessary production losses.

A large number of reactors were consuming cooling capacity from a utility network of cooling water. Sufficient cooling capacity is critical for many of these reactors, as thermal runaway could occur when the availability capacity is insufficient. This is an important safety risk.

To avoid this undesirable situation, TrendMiner was set up to monitor the cooling capacity in real time. Combining the process data with the process engineer’s knowledge about the process, warnings are triggered only on actual problem situations, avoiding false positive alarms that could be triggered by measurement noise or spikes in the data.

When the early warning has been triggered, there is ample time for the process engineer and operators to re-balance the reactors and deprioritize other equipment, so that the critical ones can consume the maximal cooling capacity. This is especially relevant in summer periods where cooling is a regular constraint.

Asset performance greatly depends on the process in which the assets are used. TrendMiner helps to contextualize asset performance with process data in order to predict when maintenance will be needed. Based on this information, the required maintenance can be aligned with the production schedule. This helps achieve production within the best operating zones of the equipment as well as reducing the maintenance, repair and overhaul costs.

drum

Hydrogen fluoride is used as catalyst in the alkylation unit at a petroleum refinery. It is regenerated in a stripper that is drained to a drum multiple times a day based on the level in the column. After a few days, four thermocouples at different heights in the drum indicate when the drum is full and must be drained.

With TrendMiner it was shown that the time until the drum is full (indirectly) depends on the acid flow to the stripper column. Using TrendMiner’s predictive mode it can now be predicted when the drum will require maintenance. This provides plenty of advance time to schedule the drainage work and lowers related maintenance costs.

Electricity consumption can easily be monitored, but energy is also consumed in (hot) Water, Air, Gas and Steam (also known in combination as WAGES). With TrendMiner, all processes related to WAGES can be monitored and even optimized. This reduces energy costs and contributes to achieving goals for reducing the carbon footprint of your organization.

As part of the energy savings goals and ISO50001 directives Covestro initiated three energy savings projects for their polyether plant in Antwerp. TrendMiner was used for online detecting, logging and explaining unexpected energy consumption and to compare the results with the reference year 2013.

Using specific formulas and calculated tags, steam consumption is monitored. Steam consumption can also be monitored using measurements from the steam valve opening. Control scheme is improved to prevent reactor temperature fluctuations caused by simultaneous heating and cooling.

With TrendMiner, energy consumption is effectively decreased and performance against the reference year is easily monitored.

Process expertise is scarce and lessons learned in the past may be hard to access when needed. TrendMiner helps to capture process knowledge and store it in relation to process behavior. When a similar process behavior occurs, it is logged in the system. Optionally the appropriate stakeholders can also be notified, including the previously captured knowledge. In this way knowledge is preserved, can be used to educate new resources and can be used to increase process expertise in the team.

A burner oven was suffering from many trips which caused production loss and increased gas consumption. With TrendMiner, multiple previously unknown root causes were found for the trips.

With this increased understanding of the process, monitors were created to alert the key stakeholders. The monitors allow them to take appropriate action when a specific cause of a trip occurs and thus avoid a trip actually happening.

The events are now also automatically annotated with the explanation of the root cause. In this way TrendMiner helps the organization to actively learn based on combining actual process behavior with subject matter expertise.

Variations in process behavior impact equipment performance over time. Besides fingerprints, scatter plot based “best operating zones” can be created with TrendMiner. By controlling the process, the asset can perform within its best operating zone, resulting in extended longevity, increased performance reliability and lower lifetime costs.

Within a water distribution network, only 5% of all pumping stations were responsible for over 50% of the total maintenance costs. With TrendMiner, expected malfunctioning of pumps could be detected by rising hydraulic head and increased energy consumption.

Using TrendMiner’s scatter plots and statistical fingerprints, the pumps are monitored to perform most efficiently and reliably within their best operating zones. Alarms have been added and users are now notified when equipment is expected to fail, allowing them the time in advance to take action. This has improved performance reliability and lowered overall costs.

Production losses or periods of high energy consumption can be detected with TrendMiner and listed automatically. After annotating the periods of interest with their causes, daily/weekly/monthly reports on Overall Equipment Efficiency (OEE) or energy consumption can be created. The loss accounting reports can be discussed with the shift team or other review meetings to take appropriate action for overall plant improvements.

For the MDI production, three plants were running at different production rates and had different amounts of downtime, resulting in less than optimal plant performance. In order to locate where and how the most issues were arising, a general impact analysis of downtime per MDI plant was performed using TrendMiner.

With TrendMiner’s value based search and reusable filters, each individual downtime (with its own downtime length) is captured. A report for further analysis is generated per plant with start-end date/time and duration, in order to calculate downtime costs. In addition, TrendMiner’s calculations are used to create average production rates and the percentage of production per MDI plant.

Using the alert functionality and automatic annotations in TrendMiner, downtime situations are automatically labeled and automatic loss accounting reports are generated. Using these reports, fast and efficient impact analysis based on time-series data contributes to developing plant improvement projects to reduce downtime and increase overall production.

In addition to setting up monitors for predictive maintenance, TrendMiner allows users to predict the evolution of the production process based on best matching behavior patterns seen in the past. The high-performance pattern recognition capabilities allow the subject matter experts to predict performance themselves, without requiring data modeling by data scientists.

Using the predicted process evolution, appropriate action can be taken by control room personnel in case of deviations from the golden fingerprints. In this way, overall profitability can be increased by controlling on-point production and reducing costs.

The circulation flow through a plate cooler was seen to drop at the end of the dosage, causing an increase of the reactor temperature. To dose faster and reduce the cycle time, faster cooling would be needed, so this problem caused a bottleneck in production.

Using TrendMiner’s search, overlay and comparison functionalities revealed that the circulation flow always dropped around a pressure of 2.8 bars in the reactor. The cause of the problem was identified to be an internal security of the circulation pump around this pressure. The related safety security setting could be increased, freeing up extra cooling capacity and making faster dosage possible.

Following this discovery, a record production was achieved due to cycle time reduction analyses in TrendMiner. With only a few minutes cycle time reduction per batch, hundreds of thousands more kilograms of product is produced per year, directly impacting the overall profitability of the plant. On top of this, products become cleaner and more consistent in quality, which means that more batches of a product grade can be produced before cleaning is needed.