Multi-scale approaches for gas-to-liquids process intensification: CFD modeling, process synthesis, and global optimization

Highlights

• A microchannel steam methane reforming process is modeled using model identification and parameter estimation.

• The reactor is incorporated into GTL as a process intensification alternative for utilizing stranded gas.

• An extensive process superstructure is globally optimized to evaluate process intensification at small scales.

• The microchannel process can reduce the break-even oil prices by $10/bbl.

Abstract

This paper presents a multi-scale framework for the intensification of small scale gas-to-liquids (GTL) processes. As the process intensification tool, a radial microchannel reactor is used to facilitate the catalytic steam reforming of methane. Due to the endothermicity of this reaction, a microchannel reactor serves as a promising alternative because of its enhanced heat transfer characteristics. However, the underlying mathematical model for the microchannel reforming process is complex. Since our aim is to elucidate the optimal process topology from a plethora of alternatives through a global optimization framework, we built a surrogate mathematical model to bridge this gap. Through a rigorous model identification, parameter estimation, and cross-validation analysis we have developed an accurate mathematical model that can predict the microchannel reactor output within 0.43%. We then implemented this mathematical model into a process superstructure that considers several novel and competing process alternatives for the production of liquid fuels from natural gas. Across different case studies ranging from 500 to 5000 barrels per day of total production, we have observed that the microchannel process can improve break-even oil prices (BEOP) by as much as $10/bbl. Since the small scale GTL process aims to utilize stranded natural gas with almost zero value, a parametric analysis is performed to evaluate the BEOP at different feedstock prices. We have observed that the microchannel reforming alternative is the superior process at all the scales investigated when a natural gas price of $1/TSCF is considered. The topological findings suggest that process intensification through microchannel steam reforming is a viable approach to monetize stranded natural gas.

Introduction

Recent advances in shale gas industry prompted an increase in natural gas production as well as a significant decrease in the spot prices. A recent perspective article (Floudas et al., 2016) has shown that natural gas based energy processes have garnered interest in recent years. However, it is evident that these processes exhibit multi-scale challenges that need to be addressed at different levels of characteristic time and length scales (Floudas et al., 2016). Among various process alternatives, the primary focus has been on the development of gas-to-liquid (GTL) technologies due to the significant potential to offset crude oil demand and imports (Floudas et al., 2012). The integration of these processes with biomass based feedstocks can also provide further synergistic benefits with improved economic and environmental performance (Onel et al., 2015). However, these type of systems require the development and investigation of optimal supply chains to account for feedstock availability and demand locations (Elia and Floudas, 2014).

Recent trends and future projections by the U.S. Energy Information Administration (see Fig. 1) also project natural gas to be abundantly available while crude oil net imports will stagnate around 5 quadrillion Btu/yr (Energy Information Administration, 2015a). However, a considerable amount of natural gas is stranded due to the low volume of production and a lack of necessary infrastructure and a demand market. Currently, stranded natural gas is either flared, vented, or re-injected. In the United States, flared and vented natural gas has been steadily increasing in recent years (Energy Information Administration, 2015b) (see Fig. 2). As such, stranded gas reserves do not provide any economic benefits while flaring and venting cause adverse environmental consequences. Viable alternatives to utilize stranded gas include GTL, liquefied natural gas (LNG), and compressed natural gas (CNG) (Wood et al., 2012, Khalilpour and Karimi, 2012, Dong et al., 2008). In this work, we focus on the production of liquid transportation fuels through a small-scale intensified GTL process.

Natural gas is a favorable feedstock for liquid fuels synthesis owing to the high hydrogen to carbon ratio of methane (Baliban et al., 2013b). Natural gas can achieve high yields toward liquid fuels synthesis as compared to other alternative feedstocks such as coal or biomass (Baliban et al., 2013b). In addition, hybrid feedstock refineries can be synthesized to exploit the synergistic benefits of integrating different combinations of coal, biomass, and natural gas (Baliban et al., 2010, Baliban et al., 2011, Baliban et al., 2012a, Baliban et al., 2012b, Baliban et al., 2012c, Baliban et al., 2013a, Niziolek et al., 2014, Niziolek et al., 2015, Onel et al., 2014). In the context of this work, we focus on a single feedstock natural gas to liquids system to evaluate the capabilities of a small-scale process that can utilize stranded gas in remote locations. Recent literature has focused on GTL processes (Fox et al., 1988, Gradassi and Green, 1995, Iandoli and Kjelstrup, 2007, Sudiro and Bertucco, 2007, Sudiro and Bertucco, 2009, Gao et al., 2008, Hao et al., 2008, Lee et al., 2009, Kim et al., 2009, Bao et al., 2010, Dillerop et al., 2010, Ha et al., 2010, Heimel and Lowe, 2009, Liu et al., 2011, Zhou et al., 2009, Peng et al., 1999, Bin et al., 2008, Hall, 2005, Suzuki et al., 1996, Horstman et al., 2005, Erturk, 2011, Vliet et al., 2009) via fixed topology process designs and a subsequent process simulation and economic analysis. Within the GTL process, each process alternative exponentially increases the number of possible flowsheet topologies. Therefore, it will be impractical to use fixed process designs to identify the optimal topology that minimizes the cost of production. Previous work within our group employed a rigorous process synthesis and global optimization framework to simultaneously compare novel and competing process alternatives for gas-to-liquids (Baliban et al., 2013b), gas-to-olefins (Onel et al., 2016), and gas-to-aromatics (Niziolek et al., 2016) refineries. In this work, we only focus on gas-to-liquids processes since the olefins and aromatics demand are concentrated in more industrial locations that will make the utilization of stranded gas more challenging.

The GTL process has been shown capable of competing with petroleum based processes, but the performance at smaller scales (less than 10 thousand barrels per day) was worse due to the economies of scale (Baliban et al., 2013b). Recently, process intensification has emerged as an enabling approach for these type of processes, since it has a potential to provide significant economic improvements to small-scale processes through the reduction of equipment size and energy consumption (Stankiewicz et al., 2000). Much of the efforts on process intensification involving chemical processes has focused on multifunctional reactors and microchannel reactors (Dautzenberg and Mukherjee, 2001, Vervloet et al., 2013, Lerou et al., 2010). Specifically, process intensification through microchannel reactors can provide order-of-magnitude improvement in heat duty per unit process volume; enabling smaller footprint or higher conversion process for the highly endothermic (e.g., steam methane reforming) and/or exothermic (e.g., Fischer–Tropsch) catalytic processes (Mbodji et al., 2012, Sousa-Aguiar et al., 2011, Tonkovich et al., 2004, Tonkovich et al., 2007, Stefanidis and Vlachos, 2008, LeViness et al., 2011, Butcher et al., 2014, Hessel et al., 2012, Sen and Avci, 2013, Stefanidis et al., 2009, Karakaya et al., 2012, Gumuslu and Avci, 2012, Almeida et al., 2011, Guillou et al., 2008, Todić et al., 2015). Therefore, there is a tremendous opportunity to utilize microchannel processes within a process superstructure to optimize and evaluate intensified GTL processes.

Inherent complexity in microchannel reactors necessitates using computational fluid dynamics (CFD) tools to accurately predict their performance and represent the direct competition between local heat and mass transfer rates with chemical kinetics (Zhai et al., 2010, Irani et al., 2011, Arzamendi et al., 2009b, Arzamendi et al., 2010, Murphy et al., 2013, Butcher and Wilhite, 2016). These simulations usually take several hours to converge because of the large number of variables and nonlinear equations. Although these models are rigorous and accurate, it would be impractical to implement them within a process synthesis and global optimization framework. Therefore, the gap between the intrinsic characteristic time and length scales of individual microchannel design and GTL process synthesis create the need for an effective bridging technique. One such possibility to eliminate this gap is the utilization of grey-box optimization. Grey-box optimization attempts to sample an underlying simulation/model to generate surrogate models so as to globally optimize the underlying complex model. A general purpose constrained grey-box global optimization tool can input a complex simulation to generate accurate input-output models and subsequently optimize the resulting mathematical model (Boukouvala and Floudas, 2016, Boukouvala et al., 2015).

In this work, we utilize the techniques related to model identification and grey-box optimization to generate a surrogate model for a SMR microchannel reactor. The resulting grey-box model is implemented within a process synthesis and global optimization framework to compare, for the first time, the intensified GTL process against the traditional counterparts and to assess the potential to avoid significant adverse environmental impact of stranded gas flaring while realizing an economic benefit. The paper will continue with (i) the multi-scale nature of the problem, (ii) the CFD modeling and simulation details of the microchannel reactor, (iii) grey-box modeling and model identification of the microchannel process, (iv) description of the GTL process superstructure and global optimization framework, and (v) computational studies before ending with our major conclusions.