UC Davis

In 2016, California passed Senate Bill (SB) 1383 which requires a 75% reduction in organic waste disposed of in landfills by 2025 as part of a larger mandate to reduce greenhouse gas (GHG) emissions in the state. These targets will be achieved primarily through diversion to compost and anaerobic digestion (AD), however significant infrastructural investments are needed to capture and treat this waste stream. Food waste (FW) is the largest portion of the municipal organics waste stream and is an ideal target for diversion to AD, which can be built as self-contained, scalable units deployed throughout urban areas. Using spatial models, an optimal network of ADs can be developed that not only reduces GHG emissions associated with FW disposal, but also reduces those associated with the collection and transportation stages of the waste system. Within this context, this research presents a novel method of estimating commercial FW generation in Los Angeles County, California that can be used to model a network of containerized ADs for FW.Following a review of three classes of spatial models and their practical use in waste management modelling, a simulated “FW Geography” (FWG) dataset is developed that consists of 273,023 points representing FW generators from 16 industry groups that in total generate 1,046,713 tons FW/year. This dataset was developed using non-spatial waste generation data from California Department of Resources and Recovery (CalRecycle) in the form of Tons Per Employee Per Year (TPEPY) values as well as spatial, Census-tract level employment and business data from ESRI Business Analyst (BA) and parcel-scale land use data from the Southern California Association of Governments (SCAG). Significant preprocessing of the datasets was needed to match the production-oriented industry groups of the ESRI BA and SCAG data to the waste-oriented industry groups of the CalRecycle TPEPY values. The FWG is less spatially aggregated than the municipal level waste generation estimates released by CalRecycle and can be used as an input to spatial models to develop a network of ADs. Using the FWG to develop a network of ADs requires careful consideration of the strengths and weaknesses of the spatial models, balancing model runtime with solution quality in real world instances. The time it takes to run a model is dependent on the number of points in the input dataset; with over 200,000 points in the FWG, these models will take an unreasonable time to solve. This problem is addressed in the discussion, which outlines methods of data aggregation that not only reduces the size of the FWG, but also increases its accuracy. These methods make use of a Tons Per Business Per Year (TPBPY) value developed for this study which captures the competing modelling goals of maximizing FW treatment while minimizing collection points and the Vehicle Miles Travelled (VMT) between them. The fine spatial scale of the SCAG zoning dataset can also be leveraged to reduce the size of the FWG; by aggregating FW generator points in close proximity to one another, modelling shared collection bins among FW producers, the spatial accuracy of the FWG can be increased. The resulting FWG can be used to develop a network of containerized ADs for FW that reduces overall GHG emissions of the waste management system and creates a circular economy of food.