February 2022

Special Focus: Digital Technologies

Optimize refining operations using plant digital twin based on molecular modeling

A new era of the energy revolution has arrived with the rapid growth of innovative technologies that utilize alternative energy resources, such as electricity, solar, wind and hydrogen, and an increasing demand for sustainable use of natural resources.

A new era of the energy revolution has arrived with the rapid growth of innovative technologies that utilize alternative energy resources, such as electricity, solar, wind and hydrogen, and an increasing demand for sustainable use of natural resources.

Refiners are facing unprecedented challenges in maintaining their profit margins. For example, the demand for fuels, especially gasoline, is decreasing significantly due to the widespread adoption of new vehicles that run on various fuels. In the next decade, the refining industry is expected to shift from a fuel-oriented industry to a raw material-oriented industry; therefore, crude-to-chemicals is one of the major technology paths for refining operations. Moreover, strict governmental and societal regulations on carbon dioxide (CO₂), methane (CH₄), sulfur oxides (SO_x) and nitrogen oxides (NO_x) emissions are pushing refiners to reduce carbon emissions and upgrade to alternative feedstocks (e.g., biomass, pyrolysis oils) together with petroleum fractions. FIG. 1 shows a typical refining flowsheet for crude-to-chemicals.

FIG. 1. Representative refinery models for crude-to-chemicals.

This article will describe how molecule-based process simulation can help address the challenges of achieving crude-to-chemicals process benefits, developing refining processes to process alternative feedstocks, such as biofuels, and more.

Process description

As shown in FIG. 1, crude oil and biofuels both feed into a refinery. Crude oil first passes through a crude distillation unit (CDU), where it gets separated into various boiling fractions: straight-run naphtha, distillate, gasoil and atmospheric residue (AR). Biofuels are upgraded via a hydro-deoxygenation (HDO) unit and converted to paraffinic distillate materials. The effluent of the HDO is mixed with the petroleum distillate from the CDU and upgraded via a hydrotreater unit to produce high-quality diesel. Alternatively, the product of the HDO can be combined with the distillate and/or gasoil from the CDU and cracked through reactors (e.g., a hydrocracker or HCR) to produce HCR naphtha. Moreover, naphthas from different plants (e.g., straight-run naphtha and HCR naphtha) can be further upgraded through reactors (e.g., reformer, steam cracker) to produce more desirable products.

Challenges with the crude-to-chemicals process

Unlike the traditional refinery, the products of the crude-to-chemicals refinery are not solely fuels (e.g., gasoline, diesel, jet fuel)—the naphtha fractions are the feedstocks to petrochemical plants (e.g., ethane cracker and aromatics production). The typical scope of traditional refinery optimization is a single unit (e.g., CDU, reformer, hydrocracker) or several units. For the crude-to-chemical scope, the operator must move beyond the unit level optimization to consider optimization across both the refinery and the chemical plant.

However, the traditional lumped model approach cannot solve the optimization of the process simulation flowsheet shown in FIG. 1. The lumped models used in refinery process simulations (TABLE 1) usually define their species by physical properties, such as boiling point, specific gravity or solubility; however the species used to model a petrochemical plant are molecular components.

For example, a detailed paraffin-isoparaffin-olefin-naphthenes-aromatics (PIONA) carbon number breakdown is required to model ethylene cracking and/or aromatic chemicals. The lumped model cannot provide functions to precisely propagate the molecules from the refining units to the chemical units. Furthermore, the new feedstocks like the biofuel feeds shown in FIG. 1 are beyond the definition of the lumped model and cannot be described using the existing lumps. It is also necessary to describe refining upgrading processes at a highly granular level to get a better understanding of the chemistries, and provide optimal operations that comply with regulations on carbon and other emissions. This motivated the development of a novel approach to address this issue: molecular level modeling.

Molecular level modeling

Because molecules are the fundamental elements in any refining and chemical process—and can reveal the nature of chemical conversions—this novel molecular modeling solution approach provides refiners an optimal solution to address the aforementioned challenges. It begins with the company’s molecular characterization (MC) technology,^1,2,3 which has established a molecular library consisting of 22 distinct molecular compound classes and more than 60,000 molecules with a comprehensive rigorous thermodynamic property package driven by proprietary physical properties models^a. By leveraging the information from that library and the analytical data of crude oils, MC can characterize the molecular compositions of crude assays so that they optimally match molecular and bulk measurements.

Using the molecular assays from MC as the initial values, a general molecule-based (MB) reactor^4,5 modeling framework was developed to simulate and optimize molecular reactor models and propagate the molecular information throughout the flowsheet. The MB reactor can organize O (10,000) molecular components in refining streams as a set of homologous series and describe O (10,000) distinct reactions in a given refining or petrochemical reactor. Employing a linear-free energy relationship (LFER)^6,7,8 technique, refiners can simulate and calibrate a complex refining reactor model with only O (30) kinetic parameters and obtain the high-fidelity and high-granularity molecular data of the products. The benefits of the MB reactor are summarized in TABLE 1.

An MB model builder was developed for users to include their own detailed molecular models into the MB reactor without requiring hard coding. The MB model builder has been used to develop a MB hydrocracker/hydrotreater model in terms of more than 2,400 molecules and 5,700 reactions.¹⁰ Refiners can use this unit operation to simulate and calibrate a hydrocracker model and propagate high-fidelity molecular data through a flowsheet of a refinery.

SINOPEC Research Institute of Petroleum Processing (RIPP) leveraged its own molecular reformer models to the MB reactor via an MB builder and successfully simulated a four-bed industrial CCR plant, as shown in FIG. 2.

FIG. 2. Customized MB reformer model in MB reactor.

In addition to modeling petroleum feedstocks at the molecular level, the MB reactor can also model alternative hydrocarbon feedstocks such as biomass, lignin, cellulose, hemicellulose, plastics, coal, etc., allowing the MB reactor to support modeling sustainable feedstocks in a refinery flowsheet. A green diesel⁹ model to upgrade biofuels was developed via an MB reactor.

Takeaways

Molecular modeling can address the upcoming challenges for refiners, and it is possible to track the molecular compositions of any stream in a flowsheet across a wide range of refining and chemical models, as shown in FIG. 3. Users can utilize high-fidelity molecular data, not only for a sustainable crude-to-chemicals steady-state simulation or calibration, but also for the integration with online optimization or for planning and scheduling.¹⁰ Molecular modeling provides an optimal plant digital twin solution to optimize refining operations for decades to come. HP 

FIG. 3. Molecular sustainable HPR across a wide range of refining and chemical models.

NOTES

^a AspenTech PC SAFT models
^b Aspen HYSYS Petroleum Refining V12.0
^c Aspen PIMS

LITERATURE CITED

1. Chen C-C and H. Que, “Method of characterizing chemical composition of crude oil for petroleum processing,” U.S. patent: 20130185044A1.
2. Watanasiri, S., S. Wang and L. Yu, “Method to represent metal content in crude oils, reactor feedstocks and reactor products,” U.S. patent: 20160162664A1.
3. Watanasiri, S., S. Wang, L. Yu and C. Quan, “Molecular characterization method and system,” European patent: 1086.2060-000.
4. Hou, Z. and D. Campbell, “Molecule-based equation oriented reactor simulation infrastructure and its model reduction,” U.S. patent application No. 16/250445.
5. Hou, Z. and D. Campbell, “Hybrid attribute reaction model (ARM) in molecule-based EO Reactor (MB EORXR),” U.S. patent application No. 16/739291.
6. Mochida, I. and Y. Yoneda, “Linear free relationships in heterogeneous catalysis: I. Dealkylation of alkylbenzenes on cracking catalysts,” Journal of Catalysis, April 1967.
7. Mochida, I. and Y. Yoneda, “Linear free energy relationships in heterogeneous catalysis: II. Dealkylation and isomerization reactions on various solid acid catalysts,” Journal of Catalysis, July 1968.
8. Klein, M. T., et al., Molecular modeling in heavy hydrocarbon conversions, CRC Press, Boca Raton, Florida, 2006.
9. AspenTech Sustainability Library, online: https://esupport.aspentech.com/S_Article?id=000098501

10. Aspen HYSYS Petroleum Refining V12.0

The Authors

Related Articles

From the Archive

Comments

Comments

{{ error }}

Add Comment

{{ comment.comment.Name }} Editor • {{ comment.timeAgo }}

{{ comment.comment.Text }}