Superstructure Optimization with Embedded Neural Networks for Sustainable Aviation Fuel Production

TL;DR

The paper develops a multi-objective superstructure optimization framework for sustainable aviation fuel production that embeds neural network surrogates within a MIQCP to handle mixtures and variable operating conditions. By representing key subprocesses (biomass gasification, RWGS, FT) with ANNs and integrating them into a target FT kerosene pathway, the approach captures nonlinearities while remaining tractable. Results show unconstrained optimization favors fossil ATR, whereas emission constraints shift designs toward biomass gasification with DAC-CS, achieving net-zero or net-negative emissions at higher costs; biomass-based configurations offer favorable abatement economics (roughly $390 per tonne CO2 avoided) compared to DAC-only routes. Crucially, allowing adaptive, ANN-driven operating conditions yields up to 20% cost savings over fixed designs, underscoring the value of simultaneous optimization of system topology and unit-level operating parameters. The study highlights the importance of biomass and carbon management strategies for cost-effective SAF and provides a framework for evaluating policy-relevant abatement costs and pathway trade-offs.

Abstract

This study presents a multi-objective optimization framework for sustainable aviation fuel (SAF) production, integrating artificial neural networks (ANNs) within a mixed-integer quadratically constrained programming (MIQCP) formulation. By embedding data-driven surrogate models into the mathematical optimization structure, the proposed methodology addresses key limitations of conventional superstructure-based approaches, enabling simultaneous optimization of discrete process choices and continuous operating parameters. The framework captures variable input and output stream compositions, facilitating the joint optimization of target product composition and system design. Application to Fischer-Tropsch (FT) kerosene production demonstrates that cost-minimizing configurations under unconstrained CO2 emissions are dominated by the fossil-based autothermal reforming (ATR) route. Imposing carbon emission constraints necessitates the integration of biomass gasification and direct air capture coupled with carbon sequestration (DAC-CS), resulting in substantially reduced net emissions but higher production costs. At the zero-emission limit, hybrid configurations combining ATR and biomass gasification achieve the lowest costs (~2.38 \$/kg-kerosene), followed closely by biomass gasification-only (~2.43 \$/kg), both of which outperform the ATR-only pathway with DAC-CS (~2.65 \$/kg). In contrast, DAC-only systems relying exclusively on atmospheric CO2 and water electrolysis are prohibitively expensive (~10.8 \$/kg). The results highlight the critical role of the embedded ANNs: optimal process conditions, such as FT reactor pressure and gasification temperature, adapt to changing circumstances, consistently outperforming fixed setups and achieving up to 20% cost savings.

Figures (12)

• Figure 1: Generic model of a chemical conversion process within the superstructure network. Each process is characterized by inlet and outlet ports through which mass flows can enter and exit the process. The general numbers of inlet or outlet ports are denoted by $\mathrm{iN}$ and $\mathrm{oN}$, respectively. A process can contain several unit operations (e.g., compression, heating/cooling, reaction, separation). The process boundaries are indicated with solid lines, while the boundaries for balancing the inlet and outlet ports are shown with dashed lines.
• Figure 2: Simplified representation of the superstructure network for the production of kerosene via Fischer-Tropsch (FT) synthesis. Processes that are represented by an artificial neural network (ANN) are indicated. Components and processes not considered in this study are greyed out. (DAC: direct air capture; ASU: air separation unit; RWGS: reverse water-gas shift; CS: carbon sequestration; FT: Fischer-Tropsch; MtO: Methanol-to-Olefins).
• Figure 3: Flowsheet of the gasification process, simulated in Aspen Plus®AspenPlu.24.08.2023, taking into account three types of biomass. The gasification process is divided into three principal stages: thermal decomposition and gasification of the biomass feedstock, pretreatment of the gasifying agents, and subsequent gas purification (removal of solid components and other impurities). (MIS: miscanthus; WS: wheat straw; PC: pine chips).
• Figure 4: Flowsheet of the reverse water-gas shift (RWGS) process, simulated in Aspen Plus®AspenPlu.24.08.2023. The RWGS process consists of three parts: pretreatment of the feed mixture, the RWGS equilibrium reaction, and cooling of the reaction mixture and associated water separation.
• Figure 5: Flowsheet of the Fischer-Tropsch (FT) process, simulated in Aspen Plus®AspenPlu.24.08.2023. The FT process consists of the following steps: pretreatment of the synthesis gas mixture, the exothermic FT reaction, product separation (water removal, separation of light gas components from long-chain hydrocarbons), and recycling of the reactant components CO and H2.