MODELLING OF CRUDE OIL DISTILLATION UNITS FOR RETROFITTING PURPOSES

Transcription

1 JICEC05 Jordan International Chemical Engineering Conference V -4 September 005, Amman, Jordan MODELLIG OF CUDE OIL DISTILLATIO UITS FO ETOFITTIG PUPOSES M. Gadalla, *, Z. Olujic, M. Jobson, +,. Smith Laboratory for Process Equipment, Delft University of Technology, 68 CA Delft, etherlands; m.gadalla@wbmt.tudelft.nl Center for Process Integration, the University of Manchester, PO Bo 88, Manchester M60 QD, UK * m.gadalla@wbmt.tudelft.nl + megan.jobson@manchester.ac.uk ABSTACT Crude distillation units are by far the largest capital and energy consug installations in the refining industry. Therefore, investors favor reusing eisting equipments more effectively rather than buying new units, providing that their retrofit objectives are satisfied with imum epenses. It is necessary to have robust and fast models that can reliably describe eisting crude distillation columns and heat recovery echangers in order to perform any retrofitting studies. In general, such systems can be modelled in a rigorous manner using commercial softwares, however shortcut models are quicker and with no significant convergence problems and can consider various optimisation design parameters. This work presents a new shortcut-based retrofit tool, which can simulate the performance of a real crude distillation tower and its associated preheat train. As such, it provides a basis for the optimisation of the operating conditions of the eisting units for energyrelated, economics and environmental benefits. It can also be used to assess and screen proposed equipment modifications for energy savings or capacity increase. In addition, within an optimisation framework, the proposed method can be used to perform a complete retrofit study on an industrial crude oil unit for various purposes. Application of the new method to some industrial plants indicated a significant potential for production capacity increase and reductions in both energy and emissions. ITODUCTIO Crude oil distillation is a major process and an important separation step in the refining industry. Their energy and capital intensive requirements play a vital role in future development projects. An eisting crude oil distillation unit is a comple structure with highly interlinked columns. The crude oil distillation column has strong connections with the associated heat recovery system, i.e. the distillation column is integrated with the heat recovery system. An atmospheric tower is a typical configuration for crude oil distillation processes. This configuration is etensively used in most refinery sites. It is a main tower with a set of side-strippers and pump-arounds; mostly three side strippers and three pump arounds. The column is connected to a series of echanger units for preheating the crude oil feed by recovering the heat from the hot column process streams and products end streams. The main distillation column uses a live steam at the bottom as a separating agent, while side strippers require reboilers for vapour generation. The first side stripper in some cases uses a live steam as well for striping. This is the case especially when there is a concern on the thermal decomposition temperature of the crude oil feed. Crude oil products are collected at the top of the column in the condenser and also from the side strippers, in addition to the bottom residue. Other column configurations also eist, such as progressive distillation sequence. This unit is a recent energy-efficient development, and was first installed at Mider refinery, Leuna, Germany. It consists of a main tower with a side-stripper, and uses two successive prefractionators (hode, 997). Few new crude oil distillation units are built; on the other hand, projects revamping eisting equipment are common. ather than installing new equipment, refinery managers prefer to eploit eisting units for more profit. eusing objectives are typically for energy saving and increasing production capacity, reducing greenhouse emissions, and processing new feedstock or changing product yields. These objectives are best achieved by using the eisting crude oil equipments and with or modifications.

2 To perform a retrofit project, retrofit models are an important element to fi the eisting distillation design Published grassroots design models and methods cannot be used directly for such a study since they are not capable of handling eisting equipment constraints such as number of stages, column diameters, locations of units, etc. On the other hand, no shortcut models are published for retrofit design, while rigorous models are widely applied within commercial simulation packages. Established rigorous models are time-consug and have convergence problems. And they fail to consider all design parameters simultaneously in the optimisation of the whole system, particularly in heatintegrated distillation columns. In contrast, shortcut models are quicker to solve and more robust for optimisation, especially when all design variables are to be considered simultaneously. They can also be combined with detailed heat-integration models for improving the energy efficiency of crude distillation systems. In this paper, modeling of crude oil distillation units for retrofit purposes is addressed. Shortcut models are developed for accounting for the details of an eisting crude oil distillation design, i.e. eisting configuration, real number of stages, and locations of side strippers, reboilers, condensers and pump arounds. These models are based on Underwood equation and grassroots shortcut design models by Suphanit (999). MODELLIG THEOETICAL BACKGOUD Shortcut models for grassroots design of distillation columns are well established and have been widely applied. The best known shortcut method for distillation design is the Underwood method (Underwood, 948). This method is based on two limiting assumptions, constant molar overflow within each column section, and constant relative volatilities throughout the column. It is easy to use, with only recoveries of two key components and thermal condition of feed needing to be specified. This method is applicable to simple conventional distillation columns, i.e. single-feed two-product columns with a single condenser and a single reboiler. The imum condenser and reboiler duties and also the imum vapour flow rates in column sections are obtained from the calculations. The method gives good results for distillation systems with relatively ideal mitures. However, for multicomponent mitures and for systems with nonideal vapour-liquid equilibrium behavior, the molar overflow is not constant and the relative volatilities change through the column (Seader and Henley, 998; Suphanit, 999). Where the underlying assumptions are not valid, the accuracy of the results will be compromised. Even in these cases, however, the estimation of imum vapour flow rates is good in regions of constant composition (pinch zones) (King, 980: p. 48; Kister, 99: p. 33). Some improvements (King, 980; andakumar et al., 98; ev, 990; Suphanit, 999) to the Underwood method have been suggested to etend its applicability. Based on Underwood equation, many researchers have proposed shortcut models for grassroots design of non-conventional distillation columns. The e- Underwood-Gilliland (FUG) model is the most popular shortcut model for design. Petlyuk et al. (965) and Stupin and Lockhart (97) used this model to analyse the energy consumption of fully thermally coupled distillation arrangements, compared with conventional columns. Cerda and Westerberg (98) developed shortcut models for the design of various comple distillation configurations. Glinos and Malone (985) developed a shortcut procedure for design of a distillation column with a side-stripper. Carlberg and Westerberg (989a) and Triantafyllou and Smith (99) applied the Underwood equation to a threecolumn model for the design and analysis of fully thermally coupled distillation columns. Suphanit (999) developed shortcut models for grassroots design of distillation columns using reboilers and stripping steam. These models are applicable for grassroots design of crude oil distillation columns. The model by Suphanit (999) is an improvement to the standard FUG method. The improvement concentrates on the limiting assumptions of the FUG method, i.e. constant molar overflow within each column section and constant relative volatilities throughout the column. The relative volatility of each component is taken to be the geometric mean of that at different locations in the column, i.e. top section, bottom section and feed stage. Enthalpy balances around column sections are carried out to accommodate the changes in vapour flow rates at both imum and actual reflu conditions. The modified shortcut design procedure at the imum reflu conditions is as follows:. Use the Underwood equation to estimate the imum vapour flow rates at the top and bottom pinch zones.. An enthalpy balance around the top section is performed to calculate the imum condenser duty and the imum vapour flow rate at the top of the column. Then, the corresponding reboiler duty is calculated by an overall enthalpy balance.

3 3. The imum vapour flow rate at the bottom of the column is calculated by enthalpy balance around the reboiler. At the actual reflu condition, the vapour flow rate in the top section is calculated by material balance around the condenser, assug a reasonable value for the reflu ratio. Then, the vapour flow rate in the bottom section is calculated by an enthalpy balance around the reboiler. The shortcut model equations for distillation columns with reboilers are listed below: The grassroots design model uses Underwood equation at the feed stage as given by Eq. (): αi f, i Σ q α i () The imum vapour flow rate (V ) and distribution of components with volatilities between those of the key components at the top pinch location is given by: α idi Σ α i V () The imum vapour flow rate at the bottom pinch (V' ) location is then calculated as follows: (3) V V ( q) F The imum reflu ( ) ratio is then calculated: V top / D) (4) (, The e equation (e, 93) is used to detere the imum number of stages ( ) at total reflu (Cited in Seader and Henley, 998; Chapter 9): /( ) ln ( ) / (5) ln[ α /α ] Gilliland correlation relates the number of actual stages to the imum number of stages, and imum and actual reflu ratios, as in Eq. (6) (Molokanov et al., 97) ξ ξ ep ξ ξ 0.5 (6) where, + ξ (7) After calculating the total number of theoretical stages () inside the column, the location of the feed stage can be identified using the empirical equation of bride (944) (Cited in Seader and Henley, 998; Chapter 9): S B D f f b d 0.06 (8) Following the calculations of the total number of stages and the feed location, an energy balance is carried out to calculate the condenser and reboiler duties (Seader and Henley, 998). For distillation columns using stripping steam, because the separation characteristics, such as temperature profile and vaporisation mechanism, are different from those of reboiled columns, the grassroots design model is as follows:. Underwood equation and enthalpy balance are applied to estimate the imum vapour flow rate in each section, and the imum reflu ratio and the condenser duty.. At actual reflu, the vapour flow rate in the top section is calculated by material balance around the condenser, assug a reasonable value for the reflu ratio. 3. An enthalpy balance around the top section calculates the condenser duty; while an overall enthalpy balance finds the enthalpy of the bottom product. 4. The dew point temperature is calculated for the top product for a partial condenser; the bubble point temperature is calculated for a total condenser. The temperature of the bottom product is calculated by bubble point calculation with known enthalpy. Then, the vapour flow rate at the bottom of the column is calculated by an overall enthalpy balance. The number of stages is calculated separately for each section of the column. In the rectifying section, the e equation is applied to detere the imum number of stages at total reflu condition, as given by Eq. (9). Then, the Gilliland correlation deteres the number of stages in this section. 3

4 d / ln / ln α /α d FS FS (9) [ ] In the stripping section, the number of stages is counted from the bottom stage until the stage vapour flow rate reaches or eceeds the vapour flow rate below the feed stage. The vapour flow rate below the feed stage is assumed to be the imum vapour flow rate at the bottom pinch of the column. EW SHOTCUT MODELS FO ETOFIT OF CUDE DISTILLATIO COLUMS Modelling of eisting distillation columns, especially crude oil distillation units is more difficult and shows more challenges than in grassroots design. In grassroots, the design models work in a logic direction that is the calculation of the required number of stages for a given set of conditions and separation requirements. However for a retrofit case, the eisting column design is fied and the models need to simulate the eisting design and calculate the separation and heat duties. Since crude oil distillation columns use the combination of reboilers and stripping steam for separation, and have a structure of a main column with side columns, the retrofit models are developed individually, based on the work of Suphanit (999), for columns with reboiler and others with stripping steam. A retrofit design algorithm is proposed in order to combine all column models together so that an eisting comple crude distillation tower can be simulated. etrofit Shortcut Models for Columns with eboilers Fig. shows an eisting simple distillation column with a reboiler; the number of stages in the top and bottom sections is and S respectively. Feed S Top product Bottom product Fig.. Distillation column with reboiler Material balances are carried out around the column for the light key () and heavy key () components: ( ) B f F b (0) b f ( ) F B ( ) D () f F d () d f ( ) F D (3) For an eisting distillation column with a reboiler, the e equation is rewritten in a new form in order to give the recovery of the key components as a function of the imum number of stages, as follows: where (4) α (5) α The relative volatilities of the key components, which are substituted in Eq. (5), are calculated as the geometric mean of the values at the top and bottom of the column and the feed stage. The compositions of these streams are given as a function of the component recoveries. From the eisting total number of stages inside the distillation column and for the given operating conditions, the imum number of stages, that is required to calculate the term is calculated from a new form of the Gilliland correlation, as follows: ( Gill ) ψ Gill ψ (6) where ξ ξ ep ξ ξ ψ Gill (7) 5 4

5 ξ (8) + The number of theoretical stages in Eq. (6) is related to the actual total number of stages in the distillation column Actual by the overall tray efficiency, as given by: η OTray Actual (9) The tray efficiency is provided from the design data available for the distillation column. However, when no data are available the efficiency can be calculated from correlations found in many tetbooks such as Seader and Henley (998), Peters and Timmerhaus (980), and Kister (99). In the calculation of the imum reflu, the change in the vapour flow rate in the column is accounted for by applying the procedure of Suphanit (999). So, Underwood equation estimates the imum vapour flow rates at the top and bottom pinch zones (King, 980: p. 48; Kister, 99: p. 33). Then, by enthalpy balances around the column sections, the imum vapour flow rates at the top and the bottom of the column are obtained. The bride correlation is rearranged for an eisting distillation column, with given number of eisting stages in each column section. The new form gives the ratio of the product compositions of key components, as a function of the number of eisting stages in each section. b d B D f f S (0) Similarly, the numbers of theoretical stages in the rectifying and stripping sections are related to the eisting stages in each section, as follows: S η () Tray STray Actual η () SActual Dividing Eq. () by Eq. (3), the ratio of the key component mole fractions can be obtained: b d f f D B (3) By substituting Eq. (3) into Eq. (0), the recovery of the key components can be related to the number of eisting stages in each section: (4) where /.47 B f (5) D f S Eq. (4) fies the split of the two key components for eisting distillation columns with given number of stages in each section. The term, can be calculated for the eisting distillation column, where the number of eisting stages in each section is given and the top and bottom product molar flow rates and the mole fractions of the key components are known. Eq. (4) is rearranged in order to be solved with Eq. (4): ( ) (6) ( ) (7) ( ) ( ) Then, Eq. (4) is rearranged as follows: (8) (9) Eqs. (8) and (9) are solved simultaneously to calculate the recovery of the heavy key component as follows: ( ) ( ) (30) ( ) + ( ) 0 (3) 5

6 (3) ( ) + ( ) + 0 Eq. (3) is quadratic, in one unknown,, as a function of various constants. The equation is rearranged to the form: ax + bx + c 0, where X -. The constants a, b and c can be etracted from equation 3. The two roots of the quadratic b + b 4ac equation are X, a b b 4ac X. a Since the recovery of the heavy key component, must be positive and less than unity, the variable X must be positive. Therefore, only the positive root of the solution is acceptable. Thus, the solution to Eq. (3) is: 4 ( + ) ( ) + ( ) / + ( + ) ( ) (33) This equation calculates the recovery of the heavy key component, given the values of the terms and. Then, the recovery of the light key component can be estimated from Eqs. (7) and (33), as follows: 4 ( + ) + ( ) ( ) / ( ( + + (34) After calculating the key component recoveries, the product key compositions (mole fractions) can be calculated from Eqs. () and (3), and hence the key component flow rates. On the other hand, the recoveries of the non-key components are calculated as follows:. For the components lighter than the light key, they are assumed to appear completely in the distillate and at zero mole fractions in the bottom product. Similarly, the components heavier than the heavy key appear completely in the bottom product and at zero mole fractions in the distillate (King, 980; p. 35).. The Underwood equation is used to detere the distribution of non-key components with ) ) volatilities between those of the key components in the top and bottom sections at imum reflu. 3. After calculating the imum number of stages, the e equation is used in another form to calculate the distribution of intermediate-boiling nonkey components at total reflu as follows (King, 980; p. 46): ( D) ( D) i α i α (35) Where (D) i is defined as the distribution ratio of component i and is equal to the ratio of the recovery of the component i between the top and bottom products. The distribution ratio of the heavy key component can be obtained, providing the recovery to the top and bottom products obtained previously. 4. Then, the distribution of the non-key components for the given operating condition can be obtained by linear interpolation between their distributions at imum and total reflu as suggested by Treybal (979) and King (980; p. 434) (Suphanit, 999). The recovery of non-key components can then be calculated from the distributions. 5. The mole fractions of the non-key components in the top and bottom products can then be calculated from similar relationships to those in Eqs. () and (3), for the calculated recoveries from the previous steps. Hence, the non-key components flow rates are then calculated. The temperatures of the products are calculated by carrying out bubble and dew point calculations. Then, an enthalpy balance around the various column sections calculates the condenser and reboiler duties. The retrofit shortcut model for reboiled distillation columns is represented through Eqs. (5) to (9), (), (), (5), and (33) to (35). The shortcut model calculates the product recoveries and compositions, condenser and reboiler duties, and product temperatures, for a given column and operating conditions. etrofit Shortcut Models for Columns with Steam The retrofit modelling for columns with steam is carried out separately for each column section (Fig. ). For the rectifying section with a given number of stages, the shortcut model equations are as follows: A material balance is carried out around the column for the light key component, resulting in: f F d D (36) 6

7 f F d (37) D The e equation is rewritten for an eisting distillation column to give the composition of the key components in the distillate as a function of the imum number of stages, as follows: d d where ϕ (38) Steam FS α ϕ Steam (39) FS α Fig.. Distillation column with stripping steam To account for the change in the volatility, the volatility of the key components is calculated as the geometric mean of the values at the top and bottom of the column and the feed stage. The imum number of stages in the rectifying section can be calculated from the Gilliland correlation, given the eisting number of stages and operating conditions: Feed ( Gill ) ψ Gill ψ (40) where S Steam Bottom product Water Top product ξ ξ ψ Gill ep 0. 5 (4) + 7.ξ ξ ξ (4) + The number of theoretical stages in the rectifying section is obtained from the actual stages in that section by a similar relationship to that in Eq. (). As mentioned previously, the vapour flow rate is not constant throughout the column. Thus, to account for this aspect, the procedure of Suphanit (999) is applied. The vapour flow rates at the top and bottom pinch zones are calculated by the Underwood equation as suggested by King (980). Then, the vapour flow rate at the top of the column is calculated by an enthalpy balance around the top section. Thus, the imum reflu ratio required in Eq. (39) can be calculated. By solving Eqs. (37) and (38), the recovery of the light key component can be calculated as a function of the eisting stages in the rectifying section and the bottom composition: d Dϕ Steam (43) F f The retrofit shortcut model for the rectifying sections is represented through Eqs. (39) to (43). The model calculates the recovery of the light key component for a given number of eisting stages in the rectifying section. For the stripping section, the retrofit model is based on consecutive flash calculations, as shown in Fig. 3. The model deteres the bottom product composition for a given number of stages in the stripping section and steam flow rate. The calculation starts by assug a bottom product composition, in terms of a heavy key component recovery. Then, in an iterative procedure, consecutive flash calculations are carried out from the bottom stage towards the feed stage. In each step, the number of stages is counted from the bottom stage to the feed stage. Then, the bottom product composition, which is related to the heavy key component recovery, is updated through a linear relation with the previous bottom composition. The new recovery value of the heavy key component is equal to the old value of the heavy key component recovery us a recovery difference. This recovery difference is directly proportional to the ratio of the difference of the eisting number of stages and the current number of stages to the difference of the current number of stages and the number of stages from the previous iteration. At the first iteration, the recovery difference is assumed to be a small value. The iterations terate when the calculated number of stages and eisting number of stages are identical. Hence, the correct bottom product composition is obtained. 7

8 The shortcut model for stripped columns includes the model for the rectifying section and the procedure shown in Fig. 3. etrofit Shortcut Models for Comple Crude Oil Distillation Columns etrofit design of comple crude oil distillation columns is more difficult than simple columns, as is the case with grassroots design. etrofit modelling of such columns is performed on the decomposed sequence. The comple crude distillation column is decomposed into the thermodynamic equivalent sequence of simple columns with thermal connections (Carlberg and Westerberg, 989b). Update new bottom composition Assume a bottom composition Carry out consecutive flash calculations from the bottom stage towards the feed stage Yes umber of stages is counted until the stage vapour flow reaches the vapour flow below the feed stage ( S) ' S S >ε o Actual Bottom composition Fig. 3. etrofit algorithm for stripping sections Eisting stages ( S) An atmospheric crude oil, i.e. column with three side strippers and three pump arounds (PA) and using live steam at the bottom and for the first side stripper, is typically decomposed into an indirect simple sequence of four single columns, connected thermally, as shown in Fig. 4. Eisting number of stages in the comple column are distributed according to the analogy of the comple configuration and the simple sequence. Side strippers and pump arounds are relocated as shown in Fig. 4. Using this well-known decomposition technique, the retrofit modelling becomes more easy and systematic. Each simple column in this equivalent sequence can be designed individually, using the retrofit shortcut models presented previously for reboiled and stripped distillation columns. Then, a retrofit procedure for the sequence of columns is proposed, as follows:. etrofit shortcut model for stripped columns is applied to the first two single columns, For each column, the retrofit model calculates the product stream flow rates, temperatures and compositions, and the duties of the condensers for the given operating conditions and steam flow rates.. Shortcut models for columns with reboiler are applied to the third and fourth columns, calculating the product flow rates, compositions and duties. 3. Due to the presence of thermal couplings and pump arounds, an iterative procedure is carried out. The iterations start with the simulation results obtained in step. ew recoveries for key components are updated in each iteration step through a linear relation with the previous recoveries. The iterations terate when the calculated number of stages and the eisting stages are identical. 4. For the whole sequence, the iterative procedure results in the product compositions, flow rates and temperatures, the flow rates of the liquid and vapour streams of the thermal coupling connection and the duties of all reboilers and condensers and pump arounds. PA3 PA PA Feed Water L Steam H LD 3 Steam HD ES Single column with steam 8 7 Single column with reboiler Fig. 4. Crude atmospheric tower and its equivalent sequence etrofit design procedure presented above is applicable to any crude oil distillation configurations, such as progressive distillation units. CASE STUDY A ATMOSPHEIC CUDE OIL DISTILLATIO TOWE An atmospheric crude oil distillation unit processes 00,000 barrels/day (60 kmol/h) of crude oil at 5 o C and 3 bar into five products: light naphtha (L), heavy naphtha (H), light distillate (LD), heavy distillate (HD), and residue (ES). Superheated steam at 60 o C and 4.5 bar is used as a stripping agent. The distillation column configuration is shown in Fig. 4. 8

9 The true boiling point data are based on a tetbook eample (Watkins, 979). The crude assay is represented using 5 pseudo-components, using the oil characterisation technique embedded within a rigorous simulation package (HYSYS, 999). The physical and thermodynamic properties of each pseudo-component (e.g. molecular weight, vapour pressure, boiling temperature, critical properties, etc.) are calculated using the Peng obinson model. Table gives the eisting stages in each section of the columns, the operating conditions, the steam flow rates, and the pump around temperature differences and the cooling duties. ESULTS AD DISCUSSIO The eisting atmospheric unit is simulated using the retrofit shortcut model. The retrofit model calculates the product flow rates and temperatures, the flow rates of the liquid recycled through the pump arounds, the duties of the condenser and reboilers, and the key component (KC) flow rates. The results are summarised in Table, compared with rigorous simulation (HYSYS) results. Table. Specifications of atmospheric crude tower Column Col. Col. Col. Col, specifications 3 4 ectifying stages Stripping stages Pressure (bar) Steam (kmol/h) PA ΔT ( o C) PA duty (MW) Top flow (kmol/h) Bottom flow (kmol/h) eflu ratio 4.77 Table. esults of shortcut and rigorous models Parameter Shortcut model igorous model Column ES flow (kmol/h) 633.9* ES temperature ( o C) KC flow (kmol/h) * PA duty (MW).87*.87* PA flow (kmol/h) 87 87* PA ΔT ( o C) 30.0* 7. Steam flow (kmol/h) 00* 00* Column HD flow (kmol/h) 49.8* HD temperature ( o C) KC flow (kmol/h) * PA duty (MW) 8.03* 8.03* PA flow (kmol/h) * PA ΔT ( o C) 50.0* 47.0 Steam flow (kmol/h) 50* 50* Column 3 LD flow (kmol/h) 65.8* LD temperature ( o C) KC flow (kmol/h) * PA duty (MW).5*.5* PA flow (kmol/h) * PA ΔT ( o C) 0.0*.4 eboiler duty (MW) * Column 4 L flow (kmol/h) L temperature ( o C) KC flow (kmol/h) H flow (kmol/h) H temperature ( o C) KC flow (kmol/h) * Condenser duty (MW) eboiler duty (MW) * eflu ratio 4.77* 4.79 *: specified + : water-free basis The table shows that the results predicted by the retrofit shortcut model are in good agreement with those obtained from the rigorous simulations; no significant deviations are observed. The maimum temperature difference is 0 o C, and the deviations in all flow rates are within 7%. However, the other variables have very close results. As seen, the distillation column configuration is very comple with many side strippers and pump arounds; moreover it has a large number of pseudo components. So this good agreement is an important result for such a column compleity and large component numbers. In refinery distillation products, the product composition and the true boiling curves are of great importance since these features detere the product le fraction Comp mo Shortcut igorous Pseudo-comp. no. specifications and Fig. 5. L product composition hence the product qualities to meet the market requirements. Fig 5 shows the L product composition, obtained from the shortcut model, and compares it with rigorous simulation as an eample. The figure is plotted as the pseudo-component mole fraction in the product against the pseudo-component number. As shown, the retrofit models predict product compositions in a good agreement with rigorous simulation. Furthermore, in Fig. 6, the true boiling curve (TBP curves) of the same product is compared with rigorous simulation. It is clear that there is good agreement between the results. etrofit shortcut models presented above can provide a basis for the optimisation of process 9

10 conditions when combined with detailed models for the heat recovery process in a single framework. This framework may change all process conditions simultaneously in order to 40 improve the energy efficiency and increase the production capacity of an eisting refinery distillation unit. Boilin g tem perature o ( C) % Volume distilled igorous Shortcut Fig. 6. L product TBP curve Furthermore, by changing the objective function, unconventional objectives can be achieved such as reducing atmospheric emissions or increasing profit. In addition, the retrofit models can simulate an eisting crude unit for new conditions, such as changing pump arounds duties, replacing reboiler with steam or increasing capacity. They can also be modified in order to calculate column diameters required for revamping projects. SUMMAY AD COCLUSIOS ew shortcut models for retrofit design of crude oil distillation units have been presented. These models were based on previous improvements to the Underwood design method. Developed models include shortcut design of distillation columns with both reboilers and steam for stripping. etrofit design procedure is applicable to different crude oil distillation configurations, including atmospheric towers. The application of the models to an eisting crude unit provided a very good agreement with established rigorous simulations. EFEECES Carlberg,. A. and A. W. Westerberg. 989a. Temperature-heat diagram for comple columns. 3. Underwood s method for the Petlyuk configuration. Ind. Eng. Chem. es., 8, 386- Carlberg,. A. and A. W. Westerberg. 989b. Temperature-heat diagram for comple columns.. Underwood s method for side strippers and enrichers. Ind. Eng. Chem. es., 8, 379- Cerda, J. and A. W. Westerberg. 98. Shortcut methods for comple distillation columns.. Minimum reflu. Ind. Eng. Chem. Proc. Des. Dev., 0, 546- e, M Ind. Eng. Chem., 4, 48- Glinos, K. and M. F. Malone Minimum vapour flow in a distillation column with a side stream stripper. Ind. Eng. Chem. Proc. Des. Dev., 8, 087- HYSYS Process Simulation Version.., Hyprotech Ltd King, C. J Separation Processes. McGraw-Hill Inc., ew York, nd edition, Chapter 9 bride, C. G Petroleum efiner, 3(9), 87- Kister, H. Z. 99. Distillation Design. McGraw-Hill, ew York, Chapters 3, 6, 7 Molokanov, Y. K., T. P. Korablina,. I. Mazurina and G. A. ikiforov. 97. Int. Chem. Eng., (), 09- andakumar, K., and. P. Andres. 98. Minimum reflu conditions. Part I. Theory. Part II. umerical solution. AIChE J., 7, 450- Peters, M. S. and K. D. Timmerhaus Plant Design and Economics for Chemical Engineers. 3 rd edition, McGraw-Hill Book Company Petlyuk, F. B., V. M. Platonov and D. M. Slavinskii Thermodynamically optimal method for separating multi-component mitures. Int. Chem. Eng., 5(3), 555- ev, E The Constant heat transport model and design of distillation columns with one single distributing component. Ind. Eng. Chem. es., 9, 935- hode, A. K Environmentally advanced refinery nears start-up in Germany. Oil & Gas Journal, March 7 Seader, J. D., and E. J. Henley Separation Process Principles. John Wiley & Sons, Inc., Chapters 6, 9 Stupin, W. J. and F. J. Lockhart. 97. Thermally coupled distillation - a case history. Chemical Engineering Progress, 68(0), 7- Suphanit, B Design of comple distillation system. PhD Thesis, UMIST, Manchester, UK Treybal,. E Mass Transfer Operations. McGraw-Hill Inc., nd edition Triantafyllou, C. and. Smith. 99. The design and optimisation of fully thermally coupled distillation columns. Trans IChemE, 70(A), 8- Underwood, V Fractional distillation of multicomponent mitures. Chemical Engineering Progress, 44, 603- Watkins, Petroleum efinery Distillation. Gulf Publishing Company, nd Edition, Teas, USA 0