Conference Agenda

Biofuels as a renewable energy source play a crucial role in the transition to a low-carbon energy system. Given the recently identified limitations of biofuels produced from food crops, largely related to the competition for land use and between food and energy, greater emphasis is placed on second-generation biofuels produced from non-food, lignocellulosic sources. Emerging research has pointed to the utilization of marginal lands for bioenergy crop production. Marginal lands, typically characterized by poor soil quality and other unfavorable growing conditions, present a promising opportunity for the cultivation of lignocellulosic crops, contributing to energy security and environmental sustainability without compromising food production. Studies have typically used mathematical programming and simulations to optimize the biofuels supply chain (SC) according to a range of objectives. Nonetheless, limited research has addressed the interactions between SCND and the upstream landscape-scale modeling associated with producing biomass. Broadening the SCND system boundary to include landscape design enables better control of feedstock supply and more precise estimates of GHG emissions.

Recent advancements in the fine-scale modeling of field-specific crop productivities and SOC sequestration potentials pose an opportunity for the use of integrated models. However, the highly geographically heterogeneous properties of fields introduce additional layers of spatial complexity. The inclusion of such details in optimization models leads to a significant increase in the number of variables and constraints. Models eventually become intractable, too big to be solved in an efficient amount of time. Consequently, the use of advanced techniques in data processing are needed for these large-scale models to be solved effectively and accurately.

We propose data processing methods to address spatial complexities in integrated biofuels SCND optimization models. Specifically, we employ two different methods to deal with the large number of fields for bioenergy feedstock production - composite curves and network reduction. The composite-curves-based approach serves to transform field-specific decision variables into lower-resolution ones without homogenizing properties of fields. This is achieved by establishing an order of selection for fields within that lower resolution and approximating the resulting composite curve, reducing the number of field variables. Network reduction is utilized to create clusters of fields that are nearby enough so that a single transportation arc to dedicated facilities can be assumed for them. This method aims to reduce the number of transportation-related variables in the model while ensure accurate representation of small fields. We also prepare an iterative linearization process for the estimation of composite curves and a two-step process in which the true field-to-facility transportation cost is recovered after a solution is obtained for network reduction. To demonstrate the feasibility and effectiveness of these methods in reducing the size of the model while maintaining its accuracy, a case study was conducted for the SCND optimization of second-generation biofuels in 8 states of the US Midwest. Finally, results reveal that while an optimization-based clustering method for network reduction leads to a more accuracy representation of the system, the use of composite curves is able to reduce the model’s run time by up to 83%.

Olefins are widely used as crucial precursors and essential building blocks in the manufacturing of chemical products, including plastic, detergent, adhesive, rubber, and food packaging. Ethylene is the most important olefin with global annual production exceeding 200 million metric tons. Currently, ethylene is almost entirely produced via steam cracking of gaseous and liquid hydrocarbon feedstocks such as ethane, propane, and naphtha. Steam cracking is one of the most energy and carbon-intensive processes in the chemical industry. As the U.S. energy landscape continues to transition toward clean, renewable electricity, one promising solution to decarbonize the steam cracking process is to implement electric cracking technology. Nevertheless, due to (1) the sheer size of most ethylene plants in the U.S., (2) the need to run these plants around the clock, and (3) the intermittent nature of variable renewable electricity (VRE) from solar and wind whose proportion in the U.S. electricity generation will continue to increase, it would be economically unrealistic and practically impossible to install massive energy storage systems (e.g., batteries) or perform complete plant reconfiguration to accommodate such a large power demand from electrified crackers.

Accounting for these complications and practical limitations, our vision for using electricity to provide process heat for steam cracking comprises diverse energy sources. We envision that the electrification of steam cracking will take place gradually due to the large capital investment associated with the decommissioning of existing crackers and the installation of new cracker units. Thus, both electrified and conventional crackers are present in the superstructure. Battery storage, electrolyzer, and hydrogen storage are used in conjunction with VRE generated onsite to support round-the-clock ethylene plant operation. Electrified crackers can be powered by electricity from the main grid, electricity generated in-house from dispatchable generators and fuel cell units, as well as from batteries. On the other hand, conventional crackers can be powered by fresh natural gas feedstock as well as the methane fraction byproduct (containing methane and hydrogen) from both conventional and electrified crackers. Essentially, the future ethylene plant becomes a microgrid, a local electric grid that acts as a single controllable entity with respect to the main grid. A microgrid can operate in either grid-connected mode or islanded mode, offering benefits such as improved resilience, economic operation, and flexibility.

In this work, we formulate a multi-objective optimization problem to minimize the total costs and carbon footprint associated with operating the ethane cracking microgrid. We build a differential-algebraic equation (DAE) numerical model that determines the energy requirement of conventional and electrified cracking. The energy demand obtained from this mechanistic model is then used to formulate a deterministic, steady-state operation of the ethylene plant. We also consider the uncertainties associated with VRE generation and market price predictions. This results in a scenario-based mixed-integer linear programming (MILP) model which is solved to optimality. By considering a hypothetical ethylene plant located on Texas Gulf Coast, we draw several insights regarding how decarbonized ethylene plants should be operated subject to the trade-offs between economic benefits and environmental impacts.