(234e) Reducing Dimensionality in Many-Objective Optimization for Planetary Boundary-Informed Sustainable Decision Making

As the global community increasingly strives to shift toward a more sustainable economy, the urgency of informed and sustainable decision-making has never been greater. This process often requires navigating complex trade-offs among a multitude of environmental, economic, and social goals. One useful framework for quantifying the sustainability of a system are planetary boundaries (PBs), which define global limits on human-induced disruptions to critical Earth system processes that, if exceeded, could dramatically affect the planet’s ability to support humanity [1]. Nine boundaries are currently defined, ranging from climate change to biosphere integrity and freshwater use [2]. Out of them, seven have already been quantified through thresholds with associated measurable control variables that allow us to assess and monitor the progress made towards sustainable development. Research has identified supply chain configurations that meet transport demand while minimizing PB transgressions and reducing the risk of exceeding Earth’s ecological capacity [3].

Most of the current studies about PBs use life cycle assessment to calculate if PBs are transgressed or not. A typical approach is to assign a certain proportion of PB limits to a specific process, and to identify solutions that minimize the sum of transgressions over all PB’s considered. As an alternative approach, one could consider each planetary boundary metric as its own objective to be minimized and solving the resulting many-objective optimization problem (MaOP). Such an approach would give rigorous information about the tradeoffs between all planetary boundaries in the decision making space; however, such solutions are limited by the current state of the art in solving and interpreting MaOP’s. In this work, we seek to overcome these limitations by utilizing our novel objective reduction community algorithm (ORCA) to systematically reduce the high-dimensional MaOP with PB objectives to a 2-objective problem whereby correlating PB objectives are grouped together., thus preserving critical tradeoff information [4].

In this work, we apply the ORCA algorithm to a representative sustainable hydrogen supply chain design problem. This sustainable hydrogen supply chain model considers nine objectives, including cost and eight planetary boundary objectives associated with six Earth system processes (e.g., climate change, biogeochemical flows, land use). Our results show that key tradeoff structures are maintained in the reduced problem, allowing for identification of promising compromise solutions at Pareto frontier knee points. Namely, we identify renewable-powered electrolysis as the major driving force governing objective tradeoffs, as PB objectives that are aided by including this process (i.e. climate change, ocean acidification) are placed in a separate group from those that are hindered (i.e. land system change, freshwater use). Overall, this approach offers a scalable and interpretable framework for integrating planetary boundaries into optimization-based decision making and systematically assessing their tradeoffs and highlights the importance of dimensionality reduction methods in advancing sustainable system design.

[1] Rockström, J., Steffen, W., Noone, K., Persson, Å., Chapin III, F. S., Lambin, E., ... & Foley, J. (2009). Planetary boundaries: exploring the safe operating space for humanity. Ecology and society, 14(2).

[2] Steffen, W., Richardson, K., Rockström, J., Cornell, S. E., Fetzer, I., Bennett, E. M., ... & Sörlin, S. (2015). Planetary boundaries: Guiding human development on a changing planet. Science, 347(6223), 1259855.

[3] Ehrenstein, M., Galán-Martín, Á., Tulus, V., & Guillén-Gosálbez, G. (2020). Optimising fuel supply chains within planetary boundaries: A case study of hydrogen for road transport in the UK. Applied Energy, 276, 115486.

[4] Russell, Justin M., and Andrew Allman. "Sustainable decision making for chemical process systems via dimensionality reduction of many objective problems." AIChE Journal 69.2 (2023): e17962.