What:
Produce oil at specified Reid vapor pressure (RVP).

How:
We modeled the separators and the stabilization column then predict and optimize RVP in real-time.

Results:
RVP predictions are 97-99% accurate for years running.  Product variance is cut in half.

Take Action

	Call 1-952-746-5761
	Request Information
	Tech Requirements
	Get a Quotation

Not Finding What You are Looking For?
   Tell Us!
(that's how we improve)

"Fresh" oil from wells is often under pressure (naturally or pumped) and usually contains substantial dissolved gases.  This oil needs to be "stabilized" prior to storage, shipment or refining.  This means depressurizing it, flashing off the gases, then running it through a column to "dial-in" gas concentrations to specification.  With 2-phase operations, the oil flow to the stabilization units can vary substantially.  Feed-forward, cascade control is very useful to maintain product consistency on target.

The Methods
Data was extracted from a real-time data historian and Process Modeler was used to model tower conditions vs. RVP sensor readings.  These models were put on-line in real-time to estimate RVP 15 to 30 minutes in advance.

Once the RVP models were validated by operating them for a period, iImprove was used to create optimization schemes to target RVP to setpoint by manipulating tower temperature setpoint in a Yokagawa DCS (Distributed Control System).

Results
This virtual sensor trend chart below shows the estimated (blue) and actual (red) RVP (normalized to non-disclose actual values) downstream from a stabilization unit.  This and other models are in use 24 hours a day, 7 days a week estimating product performance between RVP sensor readings and while the RVP sensor is removed.  Data shown is "out-of-sample" (actual vs. estimated production on data not used to build the virtual sensor) and reflects true performance (99.59% accurate).

The chart below illustrates the reduction in variance of the controlled RVP values.  On the left is the "Before" state, when controlled by the operators.  The portion on the right is the "After" state, when it is controlled by Process Intellect.

Refining Operations 
In a similar way, refining operations, such as decanting, distilling, drying and cracking, all involve creating products, intermediates or final, with particular characteristics.  Depending on the dynamics, keeping these characteristics under control can be problematic, particularly when each unit operation has its own unique behavior.  Typical unit operation optimization schemes rely on a series of theoretical equations to express the behavior of systems.  The key word is theoretical.  Our technology models the unique process behavior and then uses those cause-and-effect models to understand and determine how to operate the process, or a combination of processes at near optimal conditions.

More Oil and Gas Applications
Lift Gas Injection Optimization
Trunk line Transients Prediction
Well Treatment Optimization