Crude Evaluation Best Practices

Crude Evaluation Best Practices IDTC 2017 May 16 & 17 Dubrovnik, CROATIA Malek Masri www.haverly.com

Crude Evaluation Optimized Process IDTC 2017 May 16 & 17 Assay accuracy and update Importance of crude data in various evaluation processes Crude Evaluation in Marketing and Planning December 7, 2016

website: www.haverly.com Haverly Systems : Oil and Gas Softwares provider since 1962 Africa Asia Europe Americas Offices : - New Jersey - California - Texas - United Kingdom - Singapore - Lebanon Al Friedrich Haverly Systems Inc. email: Al@Haverly.com cell: +19736271424 Malek Masri Haverly Systems Europe email: malek@haverly.com cell: +96171428 579

Haverly Systems Inc. Oil industry softwares provider since 1962 Refinery Modeling Systems Crude Assay Management Systems Scheduling Optimization Systems Crude Oil Management Evaluation Tool H/COMET Crude Assay Management System H/CAMS Refinery LP Planning GRTMPS (G5) Refinery Scheduling Applications H/Sched Crude Assay Libraries (Chevron, Haverly..)

Crude data in various processes Key calculations based on Crude data : - Catalyst performances - Process Units capacities - Reactor models/process Simulators calibration - Final Products blending (Jet..) -

Crude Evaluation Optimized Process IDTC 2017 May 16 & 17 Crude assay accuracy and update December 7, 2016

Assay quality

Assay quality Check list : sampling, laboratory, calibration Have predefined analysis scheme - measured vs calculated point! Check correct complete and Smooth assays Softwares added value : - mass balance - properties monotonicity - ASTM, IP.. correlations - blending indexes - graphical check - warnings - references Crude Assay Management System H/CAMS

Ex.1: Nitrogen Crude Assay Management System H/CAMS

Ex.4: Yield Curve Crude Assay Management System H/CAMS

Ex.2: Viscosities Crude Assay Management System H/CAMS Similar : F > C > P rule!

Ex.3: Gas Chromatography Crude Assay Management System H/CAMS - GC data available up to 160-170C - Used in Reformer simulator calibration (N+2A)

Assay update Update assays using cargo data (Flash assay tool) Cusiana crude has got 8 API lighter in 14 years due to mixing of Cupiagua crude (45 API) and gas injection API Sulfur 1995 36.25.25 1998 39.39.17 1999 41.0.17 2004 42.5.12 2009 44.1.11

Flash Assay update

Assay update Crude sample full analysis vs. using flash assay update Full assay lab analysis Needs sampling Analysis requires several weeks Costs 6000 to 15000 $ / assay Requires full rework of lab data in crude manager softwares Not necessary if crude doesn t change a lot through time Flash Assay using crude manager softwares as H/CAMS No sampling needed. Cargo data is enough Use base assay (as previous year one) distributions Commonly used to update assays with new : Yield curve (SimDis) Whole crude measurements GC data

Assay update SimDist vs. TBP (Flash assay) TBP Distillation ASTM D2892 uses 15 plate column, 5:1 reflux ratio method only applies to crude oil distillation starts at atmospheric pressure and then switches to vacuum conditions at approx 350 C and continues to 650 deg F remaining charge transferred to vacuum potstill method ASTM D5236 distillation continues at 0.5 mm Hg to obtain an AEBP limit of 1000-1050 deg F Reproducibility of methods : ASTM D2892 1.2 wt% ASTM D2892 (vac) 1.4wt% ASTM D5236 1.5wt% SimDist (HTSD) HTSD is an ASTM proposed method which is basically an extension of ASTM D2887 key difference between HTSD & D2887 is the ability to analyze residue containing samples produces the boiling range distribution of hydrocarbons by gas chromatography up to final temp of approx. 1380ºF (750ºC) Analysis calibrated by correlating the C5 to C120 n-paraffins elution time to their AEBP Results obtained very quickly (few hours) compared to 2-3 days for physical distillation (D2892/5236)

Assay update SimDist vs. TBP (Flash assay) Some observations There is general assumption that SimDist is the equivalent to Wt% TBP. This is not entirely correct. ASTM states that it is % eluted matches WT% TBP quite well in the middle, underestimates at the front and over estimates yields at the back end. ideal situation is to also do a whole crude DHA (GC) so that the front of the yield curve is well defined H/CAMS is very well suited to receive this data since the front of TBP curves are built with a full GC if available

Assay update SimDist vs. TBP (Flash assay) if the 1995 assay is used in a planning LP Cusiana is being under valued by $0.57/bbl c.f. the more current 2009 quality Updating the 1995 assay with the Flash Assay tool to a 2009 vintage compares very closely the actual full TBP assay. Only $0.03/bbl difference API Sulfur 1995 36.25.25 1998 39.39.17 1999 41.0.17 2004 42.5.12 2009 44.1.11 LP analysis using G5 model Mbbls Obj Fn (M$) Value Base Case 1995 assay (TBP) 823 3884 Case 1 2009 assay (TBP) 823 4357 +$0.57/bbl Case 2 1995 assay - Flash update with 09 HTSD data 823 4332 +$0.54/bbl

Fouling factors and Trays Efficiencies Crude Towers simulators calibration using Efficiency factors modeling adds significant accuracy

Crude Evaluation Optimized Process IDTC 2017 May 16 & 17 Crude Assay in Planning December 7, 2016

Assay data in planning LP Matrix Generator Distillate Yields and Properties at Pre-determined Cut-point Temperatures Crude Assay and Distillation Simulation Software LP Model Recursion Procedures - Engineer needs to guess optimal cut points - Trial and error process (iterative) - Intermediate template needed - Input errors risk - Data to be updated for each change of crude - Data to be updated for each change of cut points Converged Solution

Assay data in planning By Modes Crude C D U Naphtha A Naphtha B Kero A Kero B Naphtha Pool Kero Pool AGO A AGO B AGO Pool AR B AR A Atm Resid Pool V D U

Assay data in planning Using Swing Cuts Crude A Crude B AR B C D U Naphtha A Naphtha B Swing A Swing B Kero A Kero B Swing A Swing B AGO A AGO B Swing A Swing B AR A Naph / Kero Swing Pool Kero / AGO Swing Pool AGO / AR Swing Pool Naphtha Pool Kero Pool AGO Pool Atm Resid Pool V D U

Assay data in planning Modal and Swing cuts technique limitations Modal technique Large number of operations used in combinations difficult to interpret Assumes linear averages over modes or crudes predict results for mixes and intermediate cut points Can over optimise crude processing using different cutting patterns for material that would be run together Can over optimise blending using material separately that would be running down together unless you pool the rundowns. Swing cuts technique Assumes uniform quality in whole of swing Need multiple swings where cut point range is wide. Then need MIP so swing cuts move in step. Blends properties linearly just like modal model Difficult to get good data for narrow cuts Difficult to converge if material reaches the same products by swinging up or down.

Assay data in planning LP Matrix Generator (GRTMPS) Distillate Yields and Properties at Assumed Cut-point Temperatures and with respect to changes in Cutpoint Temperatures Crude Assay and Distillation Simulation (H/CAMS) Cut Point Optimisation LP Model Recursion Procedures (GRTMPS) Converged Solution LP Solution Cut-point Temperatures Distillation Yields and Properties at Solution Cut-point Temperatures and with respect to changes In Cut-point temperatures

Assay data in planning Crude 1 Crude 2 Crude 3 Base Mode C D U Naphtha Kero Diesel Delta Temp Delta Yield / Delta Quality Cut Point Optimisation Naphtha Pool Kero Pool Diesel Pool Atres Pool Base Mode LVGO AR V D U Delta Temp HVGO VR Delta Yield / Delta Quality LVGO Pool HVGO Pool

Assay data in planning Cut Point Optimisation Use Adherent Recursion to link LP to Distillation Simulator Set up a single base mode per crude Automated tracking of cuts qualities All producing the same output pools and using the same cut point quality (so cut first, blend afterwards, for crude mix) When converged, results match those of simulation as if you had generated the mode at those cut points Switch of crude is made easy selecting in one click another crude file Change of cut points made easy as no need to reset a template

Crude Evaluation Optimized Process IDTC 2017 May 16 & 17 Crude Assay in Trading & Marketing December 7, 2016

Assay data in trading Which model to use? Refinery Excel model is still frequent Crude assay Refinery specific model Third party refinery Simplified model Full model Generic model

Assay data in trading Use accurate refinery model for Netback GPW calculations If refinery model is available always better to use full specific model Time consuming to tune many crudes data Often third party model is NOT available Use Generic model as in HCOMET it includes : Pre-defined trading regions (platts, argus..) Pre-defined products specifications (platts, argus..) Pre-defined refinery configurations (O&GJ, PIRA..) Automated links to crude, freight and products pricing systems (platts, argus, Bloomberg..) Recursed process units simulators with flexible modes (gasoline/diesel) Consider some specificities as Tan discount : reduction of 1 $/bbl for each TAN number above 0.60

Assay data in trading Crude Oil Management Evaluation Tool H/COMET

Conclusion Check list : Have Laboratory crude analysis done right Use softwares to check and complete assays Regularly update your assays (as using Flash Tools) Automate as much as possible crude input data to LP, scheduling and trading tools Evaluate crudes on accurate and flexible models (trading) Analyze refinery yields and operations against crude qualities to get to know its behavior and impact (high/low tan, paraffinic/naphthenic, asphaltenes precipitation..).