Click here for the final report.

 Introduction

In 2020, 47.0 million tons of municipal solid waste (MSW) were generated in Florida by single-family dwellings (31% of the total generation), multi-family residences (13%), and commercial entities (56%) (FDEP 2021a). Approximately 42% of the total MSW stream was recycled (FDEP 2021b). Collection programs are established by waste management divisions (cities, municipalities, or counties) to provide waste collection and management services for residents. Residential curbside collection (RCC) programs usually provide garbage, recyclables, yard waste, and in some cases, food waste collection lines, and positively impact recycling by diverting recoverable materials from the waste stream (US EPA 2011). The design of RCC programs varies significantly in Florida. The major differences are the number of collection lines, the collection frequency, the type of recycling collection system, the type of containers, and the fuel used for collection vehicles.

Inefficient collection and scheduling procedures negatively affect RCC efficiency, GHG emissions, and cost. Recently, municipalities in Florida have trended to single-stream recyclables collection and provided multiple or larger recycling containers to encourage recycling and increase recycling efficiency. At the same time, many collection providers are switching to less frequent garbage collection because of diversion to other service lines (e.g., recyclables, yard waste) and the rising cost of collection. However, contamination of recyclables has become an emerging issue, reducing the efficiency and benefits of recycling. Recently, the US Environmental Protection Agency (US EPA) released the National Recycling Strategy to bolster the MSW recycling systems of the nation. In the strategic plan, the EPA especially emphasized that "Contamination in a single-stream system continues to be a deterrent to improving America's recycling rate” (US EPA, 2021). Recent communications with the Orange County Solid Waste Division (OCSWD) indicated rising contamination of recyclables potentially attributable to the reduced waste collection frequency. Therefore, as municipalities change their collection programs, studies are needed to assess the environmental and economic aspects of such changes, including reducing the collection frequency of service lines (once per week vs. twice per week), collection containers, collection system design (single & dual-stream collection, and food waste collection), and the fuel types of the collection vehicles (natural gas, diesel, electric, etc.) on the economics, efficiency, and environmental effects.

 Motivation and Objectives

Task 1: Waste Generation Characteristics of RCC programs (Months 1-2)

The generation rates of waste, recyclables, yard waste, and food waste vary by season and year, depending on the state of the economy, population growth or decline, and geographic location. Therefore, it will be essential to look at the generation rates for multiple communities over the same period. Communities in the state of Florida, representing different waste collection programs, were identified during the previous UCF study and will be further studied in the proposed project. The key parameters for the calculation of the waste generation rate are the demographic characteristics and the waste collection data. For demographic characteristics, the 2020 US Census data containing the residents per house, GIS data containing the addresses registered, and data from the county municipal waste divisions containing the collection routes will be fused into one map by ArcGIS software to calculate residents per house (Fig. 4). The average number of residents per house will be determined and used for later life-cycle models. Waste collection data representing different waste collection programs have been requested from their solid waste divisions during the previous UCF studies. These datasets will be updated with a focus on contamination of recyclables, and will be used in Task 2. The combination of waste collection data and residents per house will be used to determine the waste generation rate for residual waste, recyclables, yard waste, and food waste 

Task 2: Evaluate the recycling efficiency, contamination rate, net carbon footprint, and collection cost of the different RCC programs (Months 2-7)

It is expected that the waste recycling efficiency, defined as the amount of waste generated divided by the total waste generated, varies depending on the implemented waste collection program. The recycling efficiency of each program will be calculated as the fraction of the total residential waste that represents recyclable waste. Finally, the composition of the recycled material obtained from the MRFs will be used to compute the contamination rate (the proportion of contaminants relative to the total recyclables). The estimated quantities of collected recyclables from Task 1 and the composition of the recycled material will be used to calculate the carbon offset and collection cost of the waste collection programs A to J* in the context of the entire waste management program, using the life-cycle assessment (LCA) model-Solid Waste Optimization Life-Cycle Framework in python (SWOLFpy) (Levis et al. 2014).

SWOLFpy is a free, open-source solid waste management LCA optimization framework with built-in parametric and Monte Carlo sensitivity and uncertainty analysis capabilities. SWOLFpy simulates the emissions, energy, and material use associated with the processes throughout the life-cycle of MSW management from collection to final disposal or energy/material recovery, and it includes default models for common solid waste management processes.

RCC Programs A-J*:

A. 2-day waste, 1-day recycling (SS), 1-day yard waste (2, 1, 1-SS),

B. 2-day waste, 2 day recycling (SS), and 1-day yard waste (2, 2, 1-SS),

C. 1-day waste, 1-day recycling (SS), 1-day yard waste (1, 1, 1-SS),

D. 1-day waste, 2 day recycling (SS), and 1-day yard waste (1, 2, 1-SS),

E. 2-day waste, 1-day recycling (DS), 1-day yard waste (2, 1, 1-DS),

F. 2-day waste, 2 day recycling (DS), and 1-day yard waste (2, 2, 1- DS),

G. 1-day waste, 1-day recycling (DS), and 1-day yard waste (1, 1, 1- DS),

H. 1-day waste, 2 day recycling (DS), and 1-day yard waste (1, 2, 1- DS),

Additional waste collection programs for food waste (organics) will be assessed:

I. 1-day waste, 1-day recycling (SS), 1-day yard waste, and 1-day food waste (1, 1, 1, 1-SS),

J. 2-day waste, 1-day recycling (SS), 1-day yard waste, and 1-day food waste (2, 1, 1, 1-SS).

Task 3: Recycling composition (contamination) sensitivity analysis and Advanced Sensing Technology for Contamination Inspection (Months 4-12) 

In this task, a sensitivity analysis of contamination rate on the life-cycle cost and GHG emission will be evaluated to identify the value gained or lost at a range of contamination rates of recyclables (30%, 40%, 50%, and 60%). Life-cycle cost and GHG emission for recycling commodities versus life-cycle cost for disposing of them at a range of contamination rates (from 10% to 50%) will be performed using SWOLFpy to identify the “value/cost” (in dollars) lost when contaminated loads are disposed of versus when they are accepted and recycled. To aid the better sorting of the recyclables and contaminations, an image-based sensing technology for the inspection of contaminated recyclables will be developed and evaluated. A hyperspectral camera (spectral range: 1000-2500 nm) will be used to analyze the contamination of the recyclables (will be purchased by the University of Central Florida (UCF) team).  Hyperspectral imaging allows detecting and mapping the presence of contaminants, e.g., moisture (Zhu et al. 2020), food residuals (Chen et al. 2021), and mixtures of different polymers (da Silva et al. 2020). The PIs will conduct laboratory tests on hyperspectral imaging with various waste types and contamination. In subsequent years, a deep-learning algorithm will be developed for inspection purposes using a high-speed workstation at UCF and field applications will be evaluated at a nearby collection system and MRF. For the current proposal, the UCF team will setup the laboratory tests using the setup shown in Fig. 6. The physical set-up is an integrated system of hardware (hyperspectral camera, computer, conveyor belt-or a plate) and software. Using the setup, the UCF team will take numerous images for a variety of types of recyclables that are accepted by in Florida, including pasteboard, brown paper grocery bags, corrugated cardboard, newspaper, box board, carrier stock, white and colored paper, plastic containers #1-7, lids, glass bottles and jars, aluminum cans, steel cans and tin household containers, Telephone Books, etc. The types of contaminants will include but not limited to, food or liquids residuals, hazardous and/or unwanted materials (e.g., needles, batteries), and items incorrectly tossed into recycling (e.g., grocery store bags, straws, coffee lids). The images will be used to train a machine-learning-based model for object and contamination identification using neural network algorithms. A similar model was constructed by the co-PI on a previous project (Chen et al. 2022).

Eco-environmental assessment of the waste collection programs A to J, using various fuel types and containers (Months 8-12) 

In this task, the fuel consumption (using natural gas, diesel, electricity, etc.) and type/size/number of recycling containers for the waste collection programs A through J, will be estimated on a monthly basis to account for the seasonal and operational variations. The estimated fuel consumption will be reported as liter per ton of collected waste. Finally, the wheel-to-wheel emissions for vehicles with various fuel types will be estimated using the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) software (US Department of Energy 2012). The estimated emissions will be reported as kg CO2 eq per ton of waste, recyclables, yard waste, and food waste for each RCC program. Alternative fuels will be identified based on the literature review. We anticipate considering twelve alternative fuels or fuel blends for the waste collection vehicles (WCVs) in the US based on fuel type and source; (1) diesel, (2) CNG (North American), (3) CNG (Non-North American), (4) LNG (North American), (5) LNG (Non-North American), (6) hydraulic-hybrid, (7) CNG (LFG sourced), (8) LNG (LFG sourced), (9) BD20 (Algaculture), (10) BD20 (soybean), (11) BD100 (Algaculture), and (12) BD100 (soybean). The fuel performance data (a quantitative measure of the fuel performance with respect to each selection criteria) will be obtained from the literature. Two MCDA methods, Simple Additive Weighting (SAW) (Churchman and Ackoff 1954) and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) (Hwang and Yoon 1981), will be used to rank fuel alternatives for the waste collection industry using the multi-level environmental and multi-criteria approach (Read et al. 2013). The selection of these two methods was based on their ability to handle multi-attribute decision-making problems. Fig. 7 presents a previous study done by the PI of this nature, however given the dynamics of the energy sector, we will be updating our study.