FAQ: Modeling Heavy Oils Using Aspen HYSYS

Transcription

1 FAQ: Modeling Heavy Oils Using Aspen HYSYS 2010 Aspen Technology, Inc. AspenTech, aspenone, and the Aspen leaf logo are trademarks or registered trademarks of Aspen Technology, Inc. All rights reserved. 2089_20_0410

2 The following questions and answers were documented following an AspenTech Webinar held January 26, If you have scarce TBP data (up to 40%), how do you get the complete TBP curves? Aspen HYSYS will fill the rest of the curves using extrapolation methods. If the data does not look good (e.g., an undershoot or overshoot), use a guide point for interpolation for finishing off the TBP curve. 2. Why is viscosity going up with temperature in one of the plots? In this slide of the presentation, the plot on the right presents the viscosity of the cut (not the full crude) as a function of the boiling point (not the stream temperature) of each cut. So as the cuts (hypo components) get heavier, the viscosity goes up. 3. What EOS do you recommend for heavy oils? We recommend using Aspen HYSYS Peng Robinson in general. Though other methods may be needed in specific situations. 4. What is a cut and draw? The cut is the hypo component that comes from the lab data; the draw is a product from the column. 5. There is this myth that PRO/II is better than HYSYS when it comes to modeling heavy oils. What is your opinion about this? It is about getting the whole simulation right. In making changes, nothing beats HYSYS to accommodate using your own knowledge, giving you the ability to extend the simulation. Understanding best practices of using the tool is important because given the variety of crude oils and their heavy ends, and heavy crude oils in general that are available in the market now it is not realistic to expect that any simulator out of the box will be able to model every heavy oil. If users wish to extend the accuracy of predicting heavy end properties, they will need the ability to apply their lab analysis to the simulation. In this respect Aspen HYSYS offers the greatest flexibility to accommodate user input data to characterize cut properties. On a technical front, heavy end modeling constitutes the following: Hypo-component estimation. We have the improved Twu s method on heavy end estimation of molecular weights, critical properties, and acentricity. If the user uses the Twu estimation method for these properties, the hypo component estimation is quite comparable between both products. From the next version (V7.2) this estimation method will become the default. Viscosity handling of the assays has been improved both in the Aspen HYSYS Oil Manager and in Aspen HYSYS Petroleum Refining (previously known as Aspen RefSYS). These methods can accurately follow the viscosity data that the user provides. 1

3 Furthermore, with Aspen HYSYS V7.2 thermal conductivity prediction for hydrocarbons has been significantly improved through implementation of API 12A3.2. Beyond the ability to model heavy ends, AspenTech has been at the forefront of modeling bitumens and the current range of extremely heavy oils (e.g., from tar sands). This industry-leading capability is offered in tools like Aspen HYSYS Petroleum Refining. 6. How does one decide cut points from an input assay? In the case of Aspen HYSYS, the user has the choice to cut the assay into as many cut points as they wish depending on the problem they want to solve. In general 4-6 cuts are sufficient, but it depends on the process being simulated. More granularity is needed in the area of interest. In the case of Aspen HYSYS Petroleum Refining (RefSYS), we provide a predefined component list that works well over the entire refining range, thus making it easier for the user. 7. If we have lab data on emulsion viscosity, can the user input it into the assay and override what HYSYS uses as a default? That is, inputting coefficients for the waterin-oil and oil-in-water equations for the viscosity and also the inversion point? Emulsion viscosity is not part of standard crude data. Crude data needs to be as clean as possible and entered separately. Emulsion viscosity is for mixtures of oil and water and cannot be entered with crude oil data. It should also consider the impact of surfactants that may be added to aid in hydrotransportation (of heavy oils) and for enhanced oil recovery in upstream. 8. I have found PR under-predicts GOR and PRSV is better. Have you checked the performance of these two models? PR is the same in both. However, PR has a single interaction parameter, while PRSV has two interaction parameters, giving you more ability to deal with non-ideal systems. This makes PRSV more equipped for hydrocarbon-water systems and that is why it comes closer to modeling GOR. 9. I have found PR-Twu slightly better than PR for heavy oils. However, these methods fall short in accurately predicting flash characterization of oil. Interaction parameters of different models provide greater ability to deal with more non-ideal solutions. Go to flash calculations in properties, check stability analysis, and change variables if needed. You may also want to change fluid package. We suggest you use Twu s method for estimation of hypo component properties. This method has been significantly improved since Aspen HYSYS and is made the default in the upcoming V7.2 release. This would eventually improve flash characterization. 2

4 10. Any inputs on adaptation of PR for the K values of heavy cuts especially in the vacuum range, and any inputs on MW correlation? Make sure it is relying on the data that has been input and not just one piece of information. Make sure the correlation being used is using input data (e.g., specific gravity). Use property slate in HYSYS, looking for any anomalies at higher range. Go to the interaction parameter page; estimated parameters are better at lower boiling points. If there are any anomalies, try to fix them, but it is better setting at 0 than to use a bad one. For molecular weight, it is recommended to use Twu correlation. This method is more reliable for heavy cuts. Additionally API 2b2.1 method has been implemented for molecular weight estimation in HYSYS V Please speak to input of viscosity curves. It is my understanding that this is a mid vol% curve, but HYSYS does not return the correct properties from that form of input. Yes, viscosity curves correspond to mid vol% curve. In certain situations viscosity curves were not treated as mid vol% curve, and this defect has been fixed for Aspen HYSYS V As asphaltenes are polar molecules, does this have any impact on modeling heavy oils? Yes, asphaltenes are very polar. As long as you cover the heavy ends with an appropriate number of cuts, they will not have much impact on distillation, which is at the lighter end of the crude. If your process has to deal with asphaltenes, then have the appropriate granularity in the heavy ends and get them out of the simulation early. Also, make sure to create a wider cut that is not purely asphaltenes. Note that asphaltenes tend to precipitate in exchangers and plug them up. Calculation of precipitation of solid asphaltenes needs its own model and Aspen HYSYS does not have such a model. But HYSYS can track them in the flowsheet with the appropriate granularity of the heavy ends and thus mimic behavior in the liquid phase. 13. Does BK10 suit heavy crude oils? BK10 is typically good for conventional oils and in vacuum towers. But it is not good for heavy crude oils. BK10 is an empirical model and good for the range for which they are made. Using BK10 for heavy oils is risky and not recommended. 14. In the section where fluid properties of the individual components can be entered, what's the exact mixing formula, especially for viscosity? Is the indexed viscosity model available in this section? Aspen HYSYS Petroleum Refining allows users to enter viscosity at 38, 50, 60, and 100 C for different components. In order to get the bulk value, two options are provided (Indexed blending or HYSYS blending). Since Aspen HYSYS Petroleum Refining allows different assays with the same components on PFD, viscosity value for each component will be different. In that case component level blending is offered with the Refutas Blending rule. This is similar to Aspen HYSYS, but some blending rules are hard-coded. So Aspen HYSYS Petroleum Refining has more flexibility and complexity in handling this. 3

5 15. If I have an appraisal well being drilled and samples taken, what inputs should I ask to have tested from the lab and in what formats to model the oil to the highest quality in HYSYS? Make sure with the lab that your data is mud-free and surfactant-free. Also confirm the lab knows you are not talking about black oil calculations. If the oil analysis is given in standard format by the lab, it should be easy to use in HYSYS. If there is a wide range from different samples, they need to be characterized separately. They may also need to run an uncertainty analysis on the data. 16. What is the upper limit of pseudo-components you would recommend for a cut? That is, how many pseudo-components are too many? Look at the process you are simulating. If higher granularity in a certain part of the crude is of interest to you, use more pseudo components. Rule of thumb: 6-8 is typical. In heavier range, you could possibly go down to 2-4 components. In Aspen HYSYS it can be possible that if you have a large number of blends in the flowsheet, the Oil Manager can end up generating large number of hypo components. This situation is well-handled in Aspen HYSYS Petroleum Refining, where the number of hypo components is controlled by the user. In Aspen HYSYS Petroleum Refining, users can input the laboratory assay data directly on stream view and generate as many assay streams as required without modifying the predecided component list. Finally, users can blend these stream assays in the feed blender to generate a blend in required proportions. 17. Does Aspen HYSYS have solids handling ability? Aspen HYSYS can handle solids such as non-interactive agents like sand, mud, and surfactants. You can use hydrocyclones to separate these. It should not be used with heavy metal content in crudes. Metals content should be represented differently. 18. Please clarify how you can modify the calculated viscosities for emulsions to get something closer to what one is expecting. For emulsions there are three main elements: oil, water, and the surfactant agent that needs to be entered as hypothetical components. You need to define viscosity of oil and the water/surfactant mix and may need to change experimental data to get the mixture right. 19. Can Aspen HYSYS be helpful for modeling "Crude Tar Distillation." Crude tar is a product of coal carbonization in coke ovens. Generally speaking, yes. Set up of distillation can be done in HYSYS. 20. Do you have experience in using HYSYS with Mexican crudes? Yes, Mexican crudes are moderately heavy and HYSYS can handle them. It can also handle heavier North American crudes like California crudes. Gulf of Mexico crudes are actually lighter. All Aspen HYSYS users on the US Gulf Coast have experience with common Mexican crudes. 4

6 21. We used PR-Lee Kesler enthalpy and had problems with convergence, and then we used Chao Seader (CS) and had no problems for the Peregrino (Brazil) crude. Any comment on CS usage? Chao Seader is good, but PR-Lee Kesler is better. There were some defects with Lee Kesler enthalpy calculations, which have been addressed in HYSYS V What recommendations do you have on SimDist conversion methods in HYSYS to TBP for high boiling point cuts? Should SimDist be treated as TBP directly or converted to TBP? SimDist input can be treated as TBP input in Mass Basis. HYSYS also allows you to use API correlations for TBP conversion. In HYSYS and HYSYS Petroleum Refining (formerly Aspen RefSYS) V7.2, you can enter data directly as SIMDIST. 23. How do I use different assays and blend them together? Aspen HYSYS allows input of multiple assays at the same time. On blending page, you can blend any assays. If you install large number of blends in the PFD, the number of hypo components can grow rapidly. When the user has a large number of assays to handle on PFD, the preferred way is to use Aspen HYSYS Petroleum Refining. This allows the user to handle large number of assays on PFD while maintaining a much lesser number of hypo components. Additionally, you can enter all the assays directly on the stream in the PFD using the new macro-cut feature on the stream. 24. What is your recommendation for estimating sulfur in products from Mexican crudes? Define a user property in your oil called sulfur (or any other property like vanadium or nickel content). Then provide a distribution for that property in Aspen HYSYS. You need estimates from lab or correlations that estimate sulfur as a function of NBP or density for each component. Then you can enter this as a curve for each assay. As crude gets blended and distilled, you can track these properties throughout the flowsheet. Note: Aspen HYSYS Petroleum Refining handles these properties in a different fashion. Contact AspenTech for more information. 25. How would you characterize a heavy oil for which you have an ASTM D7169 distillation data? You would have to convert to TBP or one of the other methods. You would do this in Excel. 26. Please comment on the use of the GS method? Have you compared PR vs. GS? For a wide range of simulation, PR is a better method. Other methods may be very good for certain specific conditions. 5

7 27. Do you need a special Aspen HYSYS package to use the Peng-Robinson Sour EOS? Yes, Peng-Robinson Sour or SRK Sour can be used when working with acidic systems containing H2S. 28. What is the minimum laboratory data required for refining simulation, as well as any additional recommended analyses? The minimum recommended lab data is distillation curve (TBP/D86/D2887/D1160) and key bulk properties such as density, viscosity, and Watson K factor. 29. Are activity models used instead of equation of state for heavy oils? No. 30. What is the minimum number of data points required for a good curve? For better results in curve fitting algorithms, it is recommended to have at least three points. 31. Do you prefer interpolation or approximation via least squares when you construct the TBP from scarce data? It is recommended to use Aspen HYSYS interpolation method (Probability Extrapolation for TBP, least square extrapolation for other properties) in absence of data. 32. We have found for flash calculations that PR works OK, but for low-pressure systems we tend to use the Zudkevitch-Joffee EOS. This improves our heavy oil flash calculations. As we said above, the key to flash calculations would be hypo component estimation for which we recommend Twu method for heavy oils. This will be changed as a default in V What options/approach do you recommend to modeling vacuum residues where a TBP curve is not available? If you use Aspen HYSYS Petroleum Refining (formerly Aspen RefSYS), the TBP curve will be extrapolated using a normal distribution at the heavy end. 34. When blending light oil diluents with a heavy oil, how well do the blending rules in HYSYS represent the properties of the resultant blend. In the Oil Manager, HYSYS uses a mixing rule that is well established (sum(molfrac, power(dynvisc, 1/3)))^3. But if the user has more information about blend viscosity, there is an option in HYSYS to input bulk values for blend on the blend page. In this case, HYSYS will NOT use its own mixing rule, but it will shift the blend viscosity to a user-supplied value. 6

8 35. How does HYSYS handle asphaltene behavior? For instance, many times we are forced to use a very low aromatic diluent to reduce the viscosity of a heavy (10 API) high asphaltenic oil. Aspen HYSYS can make it appear as a liquid phase with good interaction parameters. Generally you can model the assay using the new macro-cut feature for ashaltene on a HYSYS stream and then add a diluent like pentane on a different stream and mix them together. Several users are using this approach to evaluate different bitumen upgrader processes with HYSYS. 36. Please comment on the reliability of transthermal dynamic simulation/surge analysis of heavy crude oil pipeline? You can use Aspen HYSYS and Aspen HYSYS Petroleum Refining to perform this study. Additionally in V7.2, you can also track the petroleum properties of crude dynamically along a pipeline. 37. Do you recommend kij tuning using the limited experimental data for heavy oils? No. 38. Is there a utility that automatically generates the amount of each cut in the PFD so I can generate one stream in the PFD that replicates the assay? Oil Manager can install each assay in the blend as well as blend in PFD. Also, in Aspen HYSYS Petroleum Refining, you can use the macrocut feature that will enable you to create an assay directly on the stream. 39. What is the best way to predict metals distribution (Ni, V) on products from heavy crude? Metals distribution is usually an input to an assay in HYSYS and HYSYS Petroleum Refining. 40. If all I have is light ends composition and bulk properties, can I characterize heavy oil? It is required to have some estimation of the distillation curve in order to use the light ends feature in HYSYS. Alternatively, the user can input only bulk properties first and have some estimation of TBP distillation curve. Then the user can use this distillation assumption along with light ends and bulk properties input values. 7

9 About AspenTech AspenTech is a leading supplier of software that optimizes process manufacturing including oil and gas, petroleum, chemicals, pharmaceuticals and other industries that manufacture and produce products from a chemical process. With integrated solutions, process manufacturers can implement best practices for optimizing their engineering, manufacturing and supply chain operations. As a result, AspenTech customers are better able to increase capacity, improve margins, reduce costs and become more energy efficient. To see how the world s leading process manufacturers rely on AspenTech to achieve their operational excellence goals, visit Worldwide Headquarters Aspen Technology, Inc. 200 Wheeler Road Burlington, MA phone: fax: info@aspentech.com EMEA Headquarters AspenTech Ltd. C1, Reading Int l Business Park Basingstoke Road Reading UK RG2 6DT phone: +44-(0) fax: +44-(0) ATE_info@aspentech.com 2010 Aspen Technology, Inc. AspenTech, aspenone, the Aspen leaf logo, and the 7 Best Practices of Engineering Excellence are trademarks of Aspen Technology, Inc. All rights reserved APAC Headquarters AspenTech - Shanghai 3rd Floor, North Wing Zhe Da Wang Xin Building 2966 Jin Ke Road Zhangjiang High-Tech Zone Pudong, Shanghai , China phone: fax: apac_marketing@aspentech.com