1. Get More From One Avantis Foxboro SimSci-Esscor Triconex Molecular Level Description of Crude Oils and Complex Hydrocarbon Streams Lanny Carr – Invensys Process System REFINING CHINA 2006 April 24 -26 Beijing, China

2. Invensys Confidential  … we could remove the uncertainty about using old assay data as the basis for how we process today’s feedstocks?  …we could reduce inherent assay errors by using measurements that we can take at the plant site?  …we could make those corrections on a daily basis?  …we could produce deeper information about true compositional nature of feedstocks and other process streams?  …we could use that detailed knowledge as more accurate input to planning and optimization tools for daily operational improvements?  …what if? What if…

3. Invensys Confidential  Laboratory measurements of distillation yield behavior.  A convenient way to lump crude oil into a manageable number of pseudocomponents.  Laboratory measurements of properties of these pseudocomponents.  Source of data for API correlations.  Source of data for Linear and Simulation Programs. Assays - What they are…

4. Invensys Confidential  High quality assays are costly and time-consuming to produce.  Library assays go out-of-date as fields mature.  Input data used to select assays are often just bulk properties that are insensitive to distributed properties.  Blending during gathering and transportation can create compositional characteristics difficult to mimic with available assays.  Even perfect assays can fail, for fundamental reasons, to fully account for compositional differences that govern the refining outcome. Assays - the Problems...

5. Invensys Confidential  Crude oil quality is difficult to measure.  Analytical measurements contain error.  Need to rely on data of unknown quality.  ‘Typical’ assays are commonly used in place of ‘measured’ assays.  Bulk property correlations contain deviations and are insensitive to important compositional variations.  LP models and simulators don’t always agree with the refinery operation. Issues when Evaluating Quality

6. Invensys Confidential The Z - Based Structural Index Correlation All compounds in crude oil are classified according to the general empirical formula: C n H 2n+z N u S v O w … –z-vectors are the classifying numbers –‘n’ is the primary correlating variable [Z-BaSIC tm ] Method

7. Invensys Confidential  All crudes are made up of a few dozen classes of compounds.  Each class has a mass distribution with respect to molecular weight.  Each class and molecular weight set are comprised of multiple isomers. In short, crude oil may be described by mass distributions with respect to 3 independent variables, ‘z’, ‘n’ and ‘i’. Crude Oil Composition - Simplified

8. Invensys Confidential Refining operations are processes to: –alter the classes (hydrogenation, reforming, HDS, HDN, etc.) –alter the molecular weight (FCC, HC, alkylation, coking, etc.) –alter the isomer structure (isomerization, cat. dewaxing, etc.) In Short, refining alters the ‘z’, ‘n’, and ‘i’ character of crude oil. Refinery Operations - Simplified

9. Invensys Confidential Correlation of Boiling Point with ‘n’ and ‘z’

10. Invensys Confidential Correlation of density with ‘n’ and ‘z’

11. Invensys Confidential Density vs Carbon Number

12. Invensys Confidential C16-C17 excerpt

13. Invensys Confidential Jet A Property Variations

14. Invensys Confidential Jet A Heating Value at Constant Density

15. Invensys Confidential Jet A Aromatics vs. Paraffins at Constant Density.

16. Invensys Confidential Speciation with Molecular Modeling

17. Invensys Confidential ROMeo Non Linear Planner Old System Cut-Point Base (Pseudo Component) 51 Pure- components 39 High Fidelity Pseudo- Components LPG LN HN KERO LGO HGO RC LPG LN HN KERO LGO HGO RC NL Planner System (Assay ）

18. Invensys Confidential Grouping of Hi Fidelity Pseudocomponents

19. Invensys Confidential Heavy Naphtha Cut at Constant Density

20. Invensys Confidential Component Estimation C4 C12 Assay Blend Crudes & generate pure components & high fidelity pseudo- components Crude Assay Library ROMeo Online Model Z-Basic Update individual crude ASSAYS from online measurements & lab data Data Reconciliation Performance Monitoring Crude Assay Data Compensation Method

21. Invensys Confidential NMR provides additional information for the Z-BaSIC convergence Total Hydrogen A, O, S Hydrogen Sulfur Data Inputs & Outputs

22. Invensys Confidential Conventional Assays Pseudocomponents ________________ LP Sim Opt Crude Oil or Monitors Reference file Pure components + Hi Fidelity Pseudocomponents ___________________ Props, Comp, Sim, Opt Interface with Conventional Systems

23. Invensys Confidential Weight % Species + Properties Pseudo Components Plus Distinct Species Weight % HC minus Species + Properties Each Pseudo-Component NL Planner /ROMeo Reconciled Assay Speciated Stream Preparation and Analysis

24. Invensys Confidential Simulated Unit Yield 2 & Properties + Components Yield 3 & Properties + Components Yield 5 & Properties + Components Yield 4 & Properties + Components Yield 1 & Properties + Components …What can we do with Speciation?

25. Invensys Confidential RoleActivityTechnology Process Unit Targets with Refinery Wide Representation Operator Set Operating Limits APC Targets Set Process Unit Detailed Targets Refinery Scheduler Schedule Refinery Approved Targets Process Unit Detailed Targets Set points Schedule Refinery Planner Refinery NLP or LP Update qualities Develop Operating Plan Daily Targets with Refinery Wide Representation Daily Crude Mix Daily Liftings Prices Constraints Operations Manager Lost Opportunity Price & Constraint Visibility

26. Invensys Confidential  Overcomes limitations of existing crude oil assay technology  Enables crude oil analysis on a fundamental molecular level  Potential benefits in excess of $0.30/bbl  Moves technology to a more scientific basis for component speciation Molecular Modeling - Advantages