Reprinted from HYDROCARBON ENGINEERING MARCH 2004

Process simulation has been widely used in refineries formore than 30 years. Steady state and dynamic simula-tions are established tools for the design of new process

units and optimisation (off and online) of existing facilities.While mathematical models of separation systems and otherunit operations are available in most commercial process sim-ulators, refinery reactors are very specialised models and, inmost cases, are simulated as standalone units. As a conse-quence, the interactions betweendifferent processing units are notfully explored and the true operat-ing window is never captured.

Many refiners have recog-nised the potential benefits ofextending rigorous modelling torefinery wide simulations thatmodel all of the process units asan integrated system. Thisapproach would provide a power-ful tool to help optimise refineryoperations, and would be animportant element in sustaining acompetitive advantage in light ofnew regulations, growing compe-tition and decreasing margins.However, such refinery wide mod-els were expensive and time con-suming to create.

In response to this need,AspenTech and UOP have devel-oped a new simulation tool calledAspen RefSYS, which enablesrefiners to create rigorous refin-ery wide models using a stan-dard, easy to use solution. Thisproduct combines AspenTech’srefining industry process simula-tion flowsheet technologies andthe refinery reactor modelsdeveloped by UOP andAspenTech.

This article illustrates a typicalexample of how rigorous refinerywide modelling can be used tohelp companies make moreinformed operational and

investment decisions. This article analyses different strategiesfor benzene control in gasoline: an important issue facingmany refiners because of its significant health and environ-mental implications.

Refinery wide simulationRefineries are very complex processing plants and each oneis unique with respect to its capacity, its processing units and

Refinery wide simulationDenis Westphalen and Hiren Shethna,Aspen

Technology, Inc., Canada, explore how a rigorous

refinery wide simulation tool can be used to analyse

different strategies for benzene control in gasoline.

Figure 1. Simulation model.

Figure 2. Naphtha complex.

53-56 1/7/04 10:59 Page 53

its feedstocks. Understanding the true economic potential of arefinery is a complex challenge, due to the intricate interde-pendency of its processing units. Linear programming (LP)models have been successfully used in the decision makingprocess for many years; however, refiners should be aware ofthe limitations of this approach. The LP model is an excellenteconomic evaluation tool, butbecause the model itself is builtwith parameters that are notalways kept up-to-date to trulyreflect the refinery capabilities,or is kept up-to-date with simpli-fying assumptions, it does notdeliver the optimum economicanswer and cannot adequately resolve ques-tions related to operating constraints and newparadigms. Due to these limitations, the incor-poration of rigorous models into existing workprocesses can help to ensure that daily opera-tional decision making is based on more com-plete and accurate information.

A refinery wide simulation tool can thereforeprovide a range of important benefits, includingthe ability to:

� Understand the impact ofchanges in unit operations.

� Find the best set of operating conditions.

� Quickly test the LP modelaccuracy.

� Update the LP vectors in atimely manner.

� Respond quickly to operational changes/upsets and new regulations.

� Identify process bottlenecks.� Evaluate different process configurations.� Maximise the efficiency of existing equipment.

Components of a refinery wide simulation toolTo provide companies with an effective and easy to use application, a refinery wide simulation tool based on rigorousmodels needs to incorporate a number of key components;these include:

� Thermodynamic engine: phase equilibria and the calculation of thermodynamic properties are necessary toany rigorous simulation.

� Assay management system: the access to assay librariesand to assay characterisation tools are necessary to represent the crude oils.

� Calculation of petroleum properties based on assay information: petroleum properties such as octane number,pour point, cloud point, flash point must be calculated atany point of the flowsheet. Special property blending cal-culations are essential.

� Library of standard unit operations: the user must be ableto build a simulation model using pre-defined unit operations as distillation columns, heat exchangers, etc.

� Refinery reactors: first principle models for refinery reactors must be available for the user. These modelsmust be ‘tuneable’ using plant data. Kinetic models mustbe developed from real pilot/plant data.

� Flowsheeting environment: unit operations and reactorscan be linked together in a flowsheeting environment. Theflowsheeting environment holds the model topology andinvokes all model calculations.

Analysis of benzene control strategiesTo demonstrate how refinery wide modelling can help compa-nies optimise their operational performance, a typical scenarioin which a refinery needs to identify the optimum benzene con-

trol strategy for a particular setof operating conditions hasbeen examined. The details ofthe refinery and its operationalstatus are outlined below.

Figure 1 is a simplifiedprocess flow diagram for therefinery model. Crude oil is fed

to the crude distillation unit that was modelledas a rigorous column with four side-strippersand four pump-arounds. The atmosphericresidue is sent to the vacuum distillation unit(VDU), where VGO is obtained and used asthe feedstock to the FCC unit. The FCC unitwas rigorously modelled including the reactor(riser, stripping and regeneration), the FCCmain fractionator and the gas plant. The HFalkylation unit was also part of the simulation. Naphtha is drawn from the top of the CDU and

sent to the naphtha stabiliserwhere light components areremoved. Stabilised naphtha ishydrotreated (not rigorouslymodelled in this work) and sentto the naphtha complex (Figure2). The naphtha complex com-prises the reforming and theisomerisation units.

These two units play an important role in the benzene contentsand final quality of gasoline. The reformer is a CCR platform-ing unit operating at 500 kPa (72.5 psia) and all four reactorsat 510 ˚C (950 ˚F). The use of a refinery wide simulation toolallows the study of the interdependency of those two units, theinteractions with the different crude blending options, the influ-ence of the operation of the CDU and the connection with theother units that produce gasoline.

Two different crude oils are blended before being fed to theCDU. Table 1 shows some properties of crude oils ‘A’ and ‘B’.

Base caseOn the base case, crude oils ‘A’ and ‘B’ are blended using a1:1 ratio on volume basis. Table 2 shows the properties of thefeed stream to the reformer unit and Table 3 shows the gaso-line pool composition.

Influence of pre-fractionationOne of the strategies employed to control benzene in gasolineis the prefractionation of naphtha so that benzene and ben-zene precursors are removed from the reformer feed. Oncethose components are removed, they are sent to theIsomerisation unit, where benzene is saturated. Choosing theright benzene reduction strategy requires an understanding ofthe impact that each strategy has on such factors as overallgasoline production, refinery balance and economics. The rig-orous simulation model of the refinery makes it possible toanalyse different alternatives and process conditions and iden-tify the optimum approach.

Benzene concentrations in the reformate depend on theamount of benzene and benzene precursors in the feed to thereformer. Benzene formation occurs by conversion of cyclo-hexane and methylcyclopentane to benzene and through thedealkylation of heavy aromatics to benzene.

Reprinted from HYDROCARBON ENGINEERING MARCH 2004

Table 1. Properties of crude oils ‘A’ and ‘B’Property Crude ‘A’ Crude ‘B’API 41.28 29.78Paraffins by volume (%) 55.92 36.99Naphthenes by volume (%) 21.42 34.37Aromatics by volume (%) 22.66 28.64

Table 2. Reformer feed streamProperty ValueAromatics by volume (%) 14.42Benzene (volume %) 0.01RON 58.3D86 IBP 93.53D86 10% 102.92D86 30% 106.18D86 50% 114.12D86 70% 121.73D86 90% 145.82D86 FBP 175.1

Table 3. Gasoline pool compositionSource Volume % RON Aromatics Benzene

(vol %) (vol %)Reformate 41.7 101.3 76.7 4.92Isomerate 13.4 80.2 0 0Alkylate 12.4 91.8 0 0FCC naphtha 32.5 97.5 17.9 0.96Total 100 96.5 19.7 2.36

53-56 1/7/04 10:59 Page 54

Using the simulation model, the influence of the initial boil-ing point of the reformer feed on some key properties wasanalysed. Since all operations (distillation columns, reactors,heat exchangers and others) are part of the same simulationmodel, this study can be performed in a straightforward way.Figures 3, 4, 5, and 6 depict the influence of the D86 IBP of thereformer feed on the reformate research octane number, ben-zene concentration, hydrogen yield, and reformate, isomerateand gasoline yields, respectively.

It can be concluded that the benzene concentration onreformate, and therefore on gasoline, can be significantlyreduced if the initial boiling point of the reformer feed isincreased. The decrease in the hydrogen and reformate yieldsmust be analysed in the context of the overall refinery, takinginto account the hydrogen demand in other processing units.The increase in the reformate octane number suggests thatthe severity of the reformer could be reduced.

Further opportunitiesWhilst this analysis has identified the potential for significantperformance improvements by making minor alterations to theoperating strategy, the refinery wide simulation model alsomakes it possible to evaluate the impact of a broader range ofpotential changes to explore whether they could offer greaterbenefits.

This could include using model to examine the detailedimpact of different feedstocks on the downstream units, including the reactor units. For instance, the blending ratios ofcrudes ‘A’ and ‘B’ can easily be changed and those changesare automatically propagated through the flowsheet. The opti-mum reforming temperature and initial boiling point of thereformer feed stream can thus be determined for a differentblending ratio. In this way, it is possible to view the model as adesktop virtual refinery that allows refiners to gain a much bet-ter understanding of their complex operations. This greaterunderstanding can help to identify improvement opportunities,and can ultimately lead to more efficient and profitable refineryoperations.

References1. ‘Aspen RefSYS 1.0’, User’s Manual, Aspen Technology, Inc..2. KEESOM, W. H., KUCHAR, P. C., Penex and Platforming Synery for

Efficient Naphtha Processing and Benzene Control, UOP, LLC.3. WIER, M. J., UTLEY J., ELSTEIN, J., SCHWAKE, D., ‘Strategies for

Maximizing Profits from Reforming Units’, NPRA Annual Meeting, SanFrancisco, California (1998).

4. SUDKAMP, R., ‘Linking Crude Units with UOP Reactor Models in a Multi-Plant Refinery HYSYS Model’, Presented at the AspenTech Users’ GroupMeeting, Paris (2003).

5. FORREST, J., REYES, E., ‘Refinery-Wide Rigorous Simulation SolutionDrives New Value for Refiners’, NPRA Plant Automation and DecisionSupport, San Antonio, Texas (2003).

Reprinted from HYDROCARBON ENGINEERING MARCH 2004

Figure 3. Reformate RON. Figure 4. Benzene concentration on reformate andgasoline.

Figure 5. Hydrogen yield. Figure 6. Reformate, isomerate and gasolineyields.

53-56 1/7/04 10:59 Page 55