NEWS

Nov 20 , 2008
First Intellect 3.0
"Autonomous Analytics"
Implementation in
December

more news...


IntelliDynamics
A Division of BioComp Systems
1.952.746.5761
info@intellidynamics.net

Solutions Implemented
 
 

Virtual Sensor Examples

The following examples are real. They are taken from screenshots of actual customer applications. The customers' names are withheld to maintain their confidentiality.


 

Impending Device Failure Seen Before Actual Failure

 

 

On-line fault estimators enable you to see real-time estimates of the probability of particular problems occuring. Trip an alarm or gracefully shut down if the probability exceeds a threshold in order to protect your key assets.

 

Benefits:

  • See process problems in real-time
  • Take corrective / recovery actions before the problems become catastrophic.

 


 


 

 
 

Oil Platform Production Estimates

 

 

The above graphic shows the total liquid production from a 5 well platform in the South China Sea. The blue trace is the virtual sensor, the red is the actual production as measured by production meters. While most platforms have liquid production meters, this same approach is applied to individual wells, which typically do NOT have production meters. This enables the customers' asset owners to see individual well production in real-time without going on-test, as well as be alert to total platform metering issues.

 

Benefits:

  • See individual well or total platform production rates in real-time
  • Reduce well testing to an audit level.

 


 


 

 
 

Distillation Product Properties

 

 

The above graphic shows Reid Vapor Pressure of petroleum distillation bottoms products. While the customer had an RVP meter on line, it is expensive to maintain and was too far downstream for control. The virtual sensor is predictive, estimating RVP 15-30 minutes in advance based on tower conditions. This estimate, after 2 years running, continues to be 99.7% accurate, more accurate than the vendor's certification of the instrument itself. How can this be? It is because many instrument readings were used to create the model and thus the model represents a mean reading for given conditions, reducing the spurious noise produced by the RVP instrument.

 

Benefits:

  • See product performance in real-time
  • Enable responsive cascade product control by removing dead-time.

 


 


 

 
 

Powerplant NOx Emissions

 

 

The above graphic shows the NOx emissions from a dual fuel oil / natural gas power plant boiler. While most plants have stack emissions sensors, the actual sensor is "backed up" by the virtual sensor and their difference can be compared to illuminate problems with the physical sensor or the boiler itself.

 

Benefits:

  • See environmental conformance in real-time
  • Provide an alternative technology redundancy to your physical sensors.
  • Detect abnormal conditions or faults via real-time comparative analytics.

 


 


 

 
 

Paper Properties

 

The graphic shows a virtual sensor's estimate (red) of product performance vs. actual lab test results (blue). The customer is able to see on-line estimates of their paper's quality characteristics in real-time, enabling faster corrective action and reduced product variance and non-conformance.

 

The virtual sensor is likely more accurate than the test results, providing more detail up to the second.

 

 

Benefits:

  • Reduce testing to model-audit levels, freeing resources and reducing costs.
  • See product performance problems sooner, react and correct immediately.
  • Virtual Sensor reduces test procedure variance, with potentially higher accuracy.