Get to know CORTEX
OnPoint is a portfolio of digital solutions that deliver enhanced system performance through optimization technologies developed by proven experts in your domain.
Ready to learn more?
Contact us
Advanced analytics
TO ADVANCE YOUR OPERATIONS.
Powerful and user-friendly.
OnPoint’s CORTEX™ is our advanced analytics platform that leverages historical data along with your process engineers’ expertise to drive profits by increasing operational efficiencies such as increased production and decreased downtime. In comparison to simple regression or statistical approaches, CORTEX combines machine learning with high compute power to enable models to learn from your complex process data.
PULL DATA
from historian or CSV files.
INPUT DATA
into CORTEX to cleanse, prep and enrich data.
MACHINE LEARNING ALGORITHMS
allows you to:
BUILD MODELS
VIEW OPERATIONS IN REAL TIME
WHAT-IF-ANALYSIS
ACTIONABLE INSIGHTS:
Reduce downtime, optimize uptime, boost ROI
PULL DATA
from historian or CSV files.
Click to see more
Where CORTEX delivers results:
ROOT CAUSE ANALYSIS
NOx Emissions
Compressor Deterioration
Weld Failures
Transfer Breaks
Quality Defects
Production Loss and Break Prevention
Flaring Event
Boiler Agglomeration
INITIAL USE CASE
CORTEX was implemented at a Flint Hills Resources facilitythat converts propane into propylene, which has a wide variety of manufacturing applications including the making of plastics. The scope of the project utilized advanced analytics to model the predictability of permitted emissions from the facility’s waste heat boiler. OnPoint’s software, combined with the knowledge of the company’s subject matter experts, used more than two years of data and 60 variables to identify opportunities to reduce emissions. Models were constructed to help the team evaluate the influence of each variable during a multitude of operating scenarios including various ambient temperatures and humidity levels.
Solution
Solution
INITIAL USE CASE
FHR senior Engineers used CORTEX to monitor the process dynamics using the model outcomes to avoid upset conditions. They then acted proactively to alleviate the transient conditions. The result allowed the plant to adjust various operational settings. This resulted in increased production (within permitted limits) and decreased emissions. Combining these tools with the knowledge of the facilities subject matter experts provided valuable insights - additional information they likely didn’t have access to before.
DOWNLOAD LITERATURE
DOWNLOAD LITERATURE
INITIAL USE CASE
The facility’s compressor was reaching maximum speed when extra capacity should have been available. This specific compressor was critical to the plant process and provided raw feedstock at rates directly correlating to plant production. To pinpoint what was happening, OnPoint retrieved data from the customer’s historian including flow rates, interstage pressures and temperatures, turbine variables (speed, governor position, steam conditions), cooling water and ambient temperatures, as well as overall plant production. CORTEX’s enrich function created a target variable based on the speed of the compressor consecutively reaching its maximum speed. CORTEX was then utilized to build a probability network to narrow down to the most impactful variables and perform analysis.
Solution
Solution
INITIAL USE CASE
Leveraging CORTEX, the facility’s subject matter experts, and OnPoint’s solutions and delivery group, a blockage in the first case was identified. When the compressor cases and surrounding equipment were opened for maintenance and further investigation during the next outage, the finding was confirmed. The cooler between the first and second compression stages was leaking water, which damaged the compressor internals and created a partial blockage in the suction of stage two. Debris from the cooling water reduced the 18-inch pipe to the capacity of an 8-inch pipe.
INITIAL USE CASE
OnPoint partnered with Big River Steel to deliver a solution around the industry-wide problem of strip breaks, occurring
in the continuous galvanizing process. Process experts often have hypotheses into issues such as these but have difficulty pinpointing the root cause. OnPoint’s CORTEX software, paired with Big River Steel’s process knowledge, provided unprecedented visibility into the data behind this process. OnPoint focused on analyzing variables pertinent to strip breaks from the failed welds, which can shut down a production line for as long as 36 hours depending on the location of the break within the process. OnPoint analyzed more than 6 months of data consisting of 80 variables.
Solution
Solution
INITIAL USE CASE
Using Big River’s data, OnPoint pinpointed an equipment issue and then determined the optimal set points that produced the strongest weld. Since implementing those recommendations, the galvanizing line has gone from multiple strip breaks per month to zero, keeping employees safer and improving bottom line. Equipping Big River’s operations teams with the power of the CORTEX platform has continued to generate improvements across several areas of the mill. The facility has seen both an increase in production efficiency and a reduction in unplanned downtime, capturing millions in additional value.
DOWNLOAD LITERATURE
INITIAL USE CASE
Since 2013, a plant experienced difficulties with a machine that would frequently break when pushing additional
throughput. The number of transfer breaks varied each day and were seemingly random in nature. A transfer break is categorized as the failure of the undried sheet to “transfer” from the felt and onto the backing roll. This resulted in the sheet breaking and downtime being incurred to rethread it onto the machine. If this machine ran reliably, it would mitigate a loss of $1M per month.
Solution
Solution
INITIAL USE CASE
OnPoint's delivery team initially utilized over 4,500 variables provided for analysis. Through evaluation, these variable lists were reduced to 600. A predictive model was built with our client’s process knowledge experts in order to determine the probability of these transfer breaks. With the amount of reduced variable counts and use of correlations, the delivery team and the client were able to find a weekly anomaly causing a significant amount of these breaks. Through this effort, the client’s process knowledge experts were able to identify a relationship using four variables, ultimately saving them $12M annually in wasted production. The transfer break severity and frequency were drastically reduced, which in turn increased process stability and predictability.
DOWNLOAD LITERATURE
INITIAL USE CASE
Product defects are a costly but common occurrence in the steel industry. Steel process experts understand key variables impacting defects, but it is often difficult to point to a clear cause. Through the combination of OnPoint’s CORTEX™ platform and engineering expertise, and Big River Steel’s process knowledge, OnPoint sought to create a solution to reduce defects before they occur. OnPoint compiled more than 4,000 variables from multiple data sources. Leveraging the CORTEX platform through work with Big River Steel, the team reduced the data set to less than 50 key variables. While delivering the solution, our engineers discovered key insights enabling OnPoint to provide operators real-time guidance for reducing process defects.
Solution
Solution
INITIAL USE CASE
Incorporating the knowledge of Big River Steel’s subject matter experts, OnPoint developed dynamic models using key process variables. This gave the facility the ability to predict their quality defects. The models were developed to reflect physical relationships and ensure optimal performance. Based on new knowledge from the model, changes have been made to the controls programming in order to reduce ongoing defects. This real-time model alerts operators of changed conditions that increase the likelihood of defects, highlighting the changes in the variables driving the increase. End results indicate the ability to minimize defects and present opportunities to Big River Steel’s facility, equating to significant savings.
DOWNLOAD LITERATURE
INITIAL USE CASE
Over the course of four years, a facility faced difficulties with their overall production rate. Their paper production was reduced to around 40-50 tons per day. The production decrease was believed to be due to varying grades of paper. Previous work helped the site identify that the machine was performing “normally.” Market conditions caused a change in grades that were less effective at producing. The goal was to help the plant identify ways to improve the quality while running these newer grades. This immense loss of production caused $225K in annual profit losses.
Solution
Solution
INITIAL USE CASE
OnPoint's delivery team started off with 700 data points hourly for a time span of five years. Our analysis stayed highlevel until finding a theory and diving right into a solution based off of that theory. The delivery team found that market conditions were forcing the plant to run the machine to create products it wasn’t sufficient at making. Ultimately, this resulted in breaks and pulls in the paper. Currently, CORTEX is providing a model that delivers a better understanding of their processes and provides the conditions needed so they can consistently prevent these breaks and pulls.
DOWNLOAD LITERATURE
INITIAL USE CASE
Solution
Solution
INITIAL USE CASE
OnPoint's delivery team analyzed key variables to ultimately reduce their downtime and increase efficiencies. Through OnPoint’s CORTEX, a live dashboard monitor was deployed to alert engineers and operators of a future event in real time. This monitor allowed the facility to predict the event up to 40 minutes in advance and pointed the operators toward causes and potential remedies. Ultimately, this helped process engineers and operators identify the source of the event and take preventative actions for future events.
DOWNLOAD LITERATURE
Although already an industry leader in reducing flaring events, Flint Hills Resources engaged OnPoint to help them go even further. Flaring events were both frustrating and costly to operations, and the pre-event notification time was only five minutes. The facility had previous theories regarding the unit’s operations and potential causes for these flaring events but couldn’t pinpoint a root cause. These flaring events resulted in a loss of production and profit.
INITIAL USE CASE
A facility’s power boiler was experiencing stack plugging issues, requiring days of downtime and hazardous removal. This boiler is designed to burn 100% petroleum coke, but also permitted to burn biomass. The combustion of biomass occasionally forms calcium sulfate ash, which causes the ash to stick inside the furnace. When this ash builds up, the boiler and other parts of the plant have to shut down for a period of up to 12 days. Gaining the ability to run their process consistently would mitigate a $1M per month loss. The latest outage incurred a total cost of $18M.
Solution
Solution
INITIAL USE CASE
Starting with 15 months of data that included 250 data points at one minute intervals, OnPoint's delivery team began its evaluation. After initial findings looked promising, the team took five years of data and included five similar outages. This process enabled the delivery team to reduce the number of variables needed to predict the event to 20, and a predictive model was built alongside the client’s process knowledge experts. Previous methods provided a five-minute warning, which was not enough time to prevent the outage. the CORTEX platform identified and provided a model that gave days to weeks advanced notice and allowed the plant to alleviate incidents, saving them $14M in lost production. These findings assisted the facility in identifying issues before an event would occur in the boiler. Two agglomeration outages were avoided in the six months following the implemented solution.
DOWNLOAD LITERATURE
Transfer Breaks
Production Loss and Break Prevention
Flaring Event
Boiler Agglomeration
INITIAL USE CASE
Since 2013, a plant experienced difficulties with a machine that would frequently break when pushing additional
throughput. The number of transfer breaks varied each day and were seemingly random in nature. A transfer break is categorized as the failure of the undried sheet to “transfer” from the felt and onto the backing roll. This resulted in the sheet breaking and downtime being incurred to rethread it onto the machine. If this machine ran reliably, it would mitigate a loss of $1M per month.
Solution
Solution
INITIAL USE CASE
OnPoint's delivery team initially utilized over 4,500 variables provided for analysis. Through evaluation, these variable lists were reduced to 600. A predictive model was built with our client’s process knowledge experts in order to determine the probability of these transfer breaks. With the amount of reduced variable counts and use of correlations, the delivery team and the client were able to find a weekly anomaly causing a significant amount of these breaks. Through this effort, the client’s process knowledge experts were able to identify a relationship using four variables, ultimately saving them $12M annually in wasted production. The transfer break severity and frequency were drastically reduced, which in turn increased process stability and predictability.
DOWNLOAD LITERATURE
INITIAL USE CASE
Over the course of four years, a facility faced difficulties with their overall production rate. Their paper production
was reduced to around 40-50 tons per day. The production decrease was believed to be due to varying grades of paper. Previous work helped the site identify that the machine was performing “normally.” Market conditions caused a change in grades that were less effective at producing. The goal was to help the plant identify ways to improve the quality while running these newer grades. This immense loss of production caused $225K in annual profit losses.
Solution
Solution
INITIAL USE CASE
OnPoint's delivery team started off with 700 data points hourly for a time span of five years. Our analysis stayed highlevel until finding a theory and diving right into a solution based off of that theory. The delivery team found that market conditions were forcing the plant to run the machine to create products it wasn’t sufficient at making. Ultimately, this resulted in breaks and pulls in the paper. Currently, CORTEX is providing a model that delivers a better understanding of their processes and provides the conditions needed so they can consistently prevent these breaks and pulls.
DOWNLOAD LITERATURE
INITIAL USE CASE
Although already an industry leader in reducing flaring events, Flint Hills Resources engaged OnPoint to help them go even further. Flaring events were both frustrating and costly to operations, and the pre-event notification time was only five minutes. The facility had previous theories regarding the unit’s operations and potential causes for these flaring events but couldn’t pinpoint a root cause. These flaring events resulted in a loss of production and profit.
Solution
Solution
INITIAL USE CASE
OnPoint's delivery team analyzed key variables to ultimately reduce their downtime and increase efficiencies. Through OnPoint’s CORTEX, a live dashboard monitor was deployed to alert engineers and operators of a future event in real time. This monitor allowed the facility to predict the event up to 40 minutes in advance and pointed the operators toward causes and potential remedies. Ultimately, this helped process engineers and operators identify the source of the event and take preventative actions for future events.
DOWNLOAD LITERATURE
INITIAL USE CASE
A facility’s power boiler was experiencing stack plugging issues, requiring days of downtime and hazardous removal. This boiler is designed to burn 100% petroleum coke, but also permitted to burn biomass. The combustion of biomass occasionally forms calcium sulfate ash, which causes the ash to stick inside the furnace. When this ash builds up, the boiler and other parts of the plant have to shut down for a period of up to 12 days. Gaining the ability to run their process consistently would mitigate a $1M per month loss. The latest outage incurred a total cost of $18M.
Solution
Solution
INITIAL USE CASE
Starting with 15 months of data that included 250 data points at one minute intervals, OnPoint's delivery team began its evaluation. After initial findings looked promising, the team took five years of data and included five similar outages. This process enabled the delivery team to reduce the number of variables needed to predict the event to 20, and a predictive model was built alongside the client’s process knowledge experts. Previous methods provided a five-minute warning, which was not enough time to prevent the outage. the CORTEX platform identified and provided a model that gave days to weeks advanced notice and allowed the plant to alleviate incidents, saving them $14M in lost production. These findings assisted the facility in identifying issues before an event would occur in the boiler. Two agglomeration outages were avoided in the six months following the implemented solution.
DOWNLOAD LITERATURE
EVENT PREVENTION
OPTIMIZATION
Cement Kiln NOx Savings
NOx Emissions
Weld Failures
Yield Performance
Condenser Recovery
INITIAL USE CASE
CORTEX was implemented at a Flint Hills Resources facilitythat converts propane into propylene, which has a wide variety of manufacturing applications including the making of plastics. The scope of the project utilized advanced analytics to model the predictability of permitted emissions from the facility’s waste heat boiler. OnPoint’s software, combined with the knowledge of the company’s subject matter experts, used more than two years of data and 60 variables to identify opportunities to reduce emissions. Models were constructed to help the team evaluate the influence of each variable during a multitude of operating scenarios including various ambient temperatures and humidity levels.
Solution
Solution
INITIAL USE CASE
FHR senior Engineers used CORTEX to monitor the process dynamics using the model outcomes to avoid upset conditions. They then acted proactively to alleviate the transient conditions. The result allowed the plant to adjust various operational settings. This resulted in increased production (within permitted limits) and decreased emissions. Combining these tools with the knowledge of the facilities subject matter experts provided valuable insights - additional information they likely didn’t have access to before.
DOWNLOAD LITERATURE
INITIAL USE CASE
OnPoint partnered with a leading company in the cement industry to apply CORTEX™ Advanced Analytics software. The objective was to decrease the amount of reducing agent required to chemically treat NOx at one of their sites, while also maintaining or improving their already low NOx emissions. The facility provided OnPoint with process data, product and feedstock chemistry data, and process documentation. OnPoint performed data preparation, enrichment steps, and model development using CORTEX.
Solution
Solution
INITIAL USE CASE
Leveraging CORTEX, OnPoint was able to analyze the facility’s historical data and provide operating recommendations to minimize combustion NOx created in the process, without sacrificing energy efficiency or product quality. The models developed were reviewed with the facility’s process experts to validate the model performance, improve the models and results, and develop recommendations for process operation to decrease reducing agent consumption. By following the recommendations, OnPoint estimates this site will net significant savings from a more than 20% decrease in reducing agent required.
DOWNLOAD LITERATURE
INITIAL USE CASE
OnPoint partnered with Big River Steel to deliver a solution around the industry-wide problem of strip breaks, occurring
in the continuous galvanizing process. Process experts often have hypotheses into issues such as these but have difficulty pinpointing the root cause. OnPoint’s CORTEX software, paired with Big River Steel’s process knowledge,
provided unprecedented visibility into the data behind this process. OnPoint focused on analyzing variables pertinent to strip breaks from the failed welds, which can shut down a production line for as long as 36 hours depending on the location of the break within the process. OnPoint analyzed more than 6 months of data consisting of 80 variables.
Solution
Solution
INITIAL USE CASE
Using Big River’s data, OnPoint pinpointed an equipment issue and then determined the optimal set points that produced the strongest weld. Since implementing those recommendations, the galvanizing line has gone from multiple strip breaks per month to zero, keeping employees safer and improving bottom line. Equipping Big River’s operations teams with the power of the CORTEX platform has continued to generate improvements across several areas of the mill. The facility has seen both an increase in production efficiency and a reduction in unplanned downtime, capturing millions in additional value.
DOWNLOAD LITERATURE
INITIAL USE CASE
A facility wanted to optimize the production yield for their polymer manufacturing and fiber spinning processes. OnPoint focused on delivering a real-time prediction of their yield based on the facility’s key process drivers. The goal was to provide an early indication of suboptimal performance and highlight the variables influencing poor yield.
Solution
Solution
INITIAL USE CASE
OnPoint's delivery team collected 18 months of historical data for over 1,000 variables from several different data
sources. OnPoint’s analytics engineers then built dynamic data models for 10 major production lines. During this process, the variables were narrowed down to 100 key variables to maximize the yield prediction. To enable real-time deployment, OnPoint developed a custom data extraction and transfer process. This process pulls raw data from several customer data sources, automates calculations and delivers a clean data feed to the real-time monitor. This real-time monitor continues to provide a prediction of the yield and alerts when low yield is forecasted. These alerts highlight key variable changes causing the low yield and inspire action from operators with the ability to improve the overall yield performance.
DOWNLOAD LITERATURE
INITIAL USE CASE
A Flint Hills Resources facility was experiencing operational difficulties with a partial condenser unit downstream of
their reactors. The product recovery was not only lower than expected but also unpredictable. Shut down times due to frequent condenser cleanings resulted in a loss of production and overall profit that totaled roughly $1M annually. The facility’s goal was to improve the percent knockout of the unit. This would extend the time between unplanned shutdowns and create a more stable knockout rate over time. Analytics engineers from OnPoint partnered with the Flint Hills Resources facility to optimize the various process parameters associated with the partial condenser and upstream reactors.
Solution
Solution
INITIAL USE CASE
OnPoint delivered a summary report identifying key variables and variable ranges to improve product recovery in the
unit. The optimum effluent cooler operating temperature, reactor feed rate and composition were identified and a
live dashboard monitor was deployed. Leveraging CORTEX, plant engineers and operators can monitor and optimize their process in real time. This implementation also provided new insights into the unit’s overall operation to increase production outside of product recovery.
DOWNLOAD LITERATURE
Root Cause Analysis
Find out the why, and address it
See case studies
Event Prevention
Predict and overcome failures
See case studies
Optimization
Fine-tune for peak performance
See case studies
Data Cleanse
Data Visualization
MaGE
Speedy Model Creation
Real-time Monitor
Data Cleanse
Load your data as-is and CORTEX will clean it. impute missing values and handle categorical variables.
Data Visualization
Visualize and remove outliers. Add rows and columns to your data and learn what variables are important to the process.
MaGE
CORTEX's proprietary algorithm eliminates the need for you to hunt for the best model. MaGE builds a variety of models in the platform plus an optimized, ensemble model. Then it provides scores for all of them.
Speedy Model Creation
CORTEX will create your model in minutes, no matter how much data you upload.
Real-time Monitor
See detailed, operational data in real time and be alerted to changes as they happen. Our alerts pinpoint the slightest change in variables so you know exactly what the issues are and when they occurred.
![](https://media-s3-us-east-1.ceros.com/koch-industries/images/2020/10/21/3458f6ca2b43ff101d79b6a159080681/plus-icon.svg)