(325a) Resource-Circular Manufacturing Systems Optimisation

The projected 50% increase in global population in the 21st century combined with non-OECD economic growth is expected to increase demand stress and lead to land-water resource scarcity and rising waste generation. Take agro-food sector as an example. Current agriculture contributes significant greenhouse gas emissions and occupies 37% of global land and 70% fresh water use. Moreover, rising waste generation from agro-food sectors brings additional stress, which is equivalent to waste of 8.5% of annual water withdrawn and 28% of agricultural lands globally. Global municipal solid waste (MSW) growth is projected to exceed 11 million tons per day (59%-68% organic fraction (OFMSW)) by 2100 under ‘business as usual’ [1] and a significant amount of (330km³ per year) municipal wastewater (WW) is generated [2]. Increasing waste trends are particularly intense in less developed countries and contribute to resource stress, greenhouse gases and environmental degradation. Resource-circular manufacturing underpinned by technology innovation and a sustainable energy-resource-waste nexus, if fully realised, will signal the manufacturing transformation including agro-food sectors.

In conventional manufacturing systems, waste streams have been regarded as by-products (carrying zero or low-value) rather than marketable commodities with well-defined grades. In fact, waste streams such as OFMSW and WW streams are not only carbon-rich resources as energy carrier but also may contain high nutrient values (e.g. nitrogen, phosphorus). The waste resources present promising opportunities for recovery of value-added products via thermochemical and biochemical technologies under circular economy. To exploit waste resource value requires a transformative waste value chain and resource-circular manufacturing, which calls for robust projection of waste quantity and quality (composition) and waste-recovery process integration into manufacturing systems.

Efficient planning (e.g. sizing and logistics) and operation of waste recovery technologies requires continuous and consistent waste feedstock supply and integrative solutions considering the existing manufacturing facilities. However, it is complex to quantity waste feedstock from diverse sources due to highly varying composition and low traceability, which are dependent on spatially-resolved factors (e.g. local industry and behaviour), temporal and environmental variables. Despite the research advances in MSW volume forecast by using material flow models, regression analysis, machine learning and artificial intelligence techniques, robust projection of waste including OFMSW and WW composition remains open.

Moreover, fragmented sub-systems hinder the transformation towards resource-circular manufacturing. Empirical advances across energy production, manufacturing, waste treatment, environmental and natural resource management represent fragmented sub-systems, where conflicting decision criteria are involved (e.g. economic feasibility vs. sustainability). Research remain open on multi-scale systems modelling to enable optimal design of resource-circular manufacturing by capturing the interdependencies across sub-systems.

This study presents a data-driven optimisation modelling approach which integrates machine learning techniques, mathematical optimisation and chemical process design to inform the transformation of resource-circular manufacturing systems. We use two studies to demonstrate the applicability of the modelling approach. Specifically, we adopted gradient boosting model to project waste composition based on their non-linear relationships with spatially-temporally resolved variables. Beyond industrial waste, we also tested the modelling approach on OFMSW composition projection considering 327 UK local authorities [4]. The predictive performance and spatial granularity of model projections offer a promising approach to inform decision-making on waste commoditization and utilisation. Further, a closed-loop single cell protein manufacturing system underpinned by fermentation technology is presented. Our original laboratory data (including WW recovery and separation technologies) are integrated into a process design tool. Our research derived the scaled-up technology performances to effectively co-recover energy and N/P/protein from fermentation waste. To link the process design with multi-objective optimisation, a data driven approach was followed. Surrogate modelling techniques were applied, which bridged the technology process design with multiple sustainability criteria across manufacturing sub-systems. The process optimisation model was further expanded to include spatial and temporal dimensions, which highlights the spatially-explicit solutions on technology choice, sizing, location, and logistics of waste recovery and its integration with existing protein manufacturing facilities. The optimal solutions account for waste recovery through anaerobic digestion and other chemical and thermochemical technologies to meet the economic feasibility and environmental sustainability simultaneously. Overall, the data-driven optimisation models developed in our research represent a promising approach to tackle the open challenges hindering resource-circular manufacturing.

REFERENCES

1. Hoornweg, D., P. Bhada-Tata, and C. Kennedy, Environment: Waste production must peak this century. Nature, 2013. 502(7473): p. 615-7.
2. Hoornweg, D. and P. Bhada-Tata, What a Waste - A Global Review of Solid Waste Management. 2012, World Bank.
3. WRAP, Food Waste Chemical Analysis. 2010.
4. Adeogba, E., et al., Waste-to-Resource Transformation: Gradient Boosting Modelling for Organic Fraction Municipal Solid Waste Projection. 2019.