Case study: Recycling systems & material recovery — a city or utility pilot and the results so far

Municipal recycling systems across the developed world face a paradox: public support for recycling remains at historic highs, yet actual material recovery rates have plateaued or declined in many jurisdictions since China's National Sword policy disrupted global waste trade in 2018. Cities that once exported the complexity of sorting and processing now confront the full cost of building domestic recovery infrastructure. Among the most instructive responses is the City of Houston's transformation of its curbside recycling program through the One Bin for All pilot, a multi-year initiative that replaced conventional single-stream collection with advanced sorting technology, AI-driven contamination detection, and new downstream partnerships. The Houston case, along with complementary pilots in Flanders (Belgium) and Kamikatsu (Japan), offers founders and circular economy practitioners a grounded view of what municipal material recovery modernization actually requires, what it costs, and what results it delivers.

Why It Matters

The global recycling industry processes approximately 600 million metric tons of secondary materials annually, representing a market valued at over $250 billion according to the Bureau of International Recycling. Yet the effective recovery rate for municipal solid waste in the United States remains stubbornly low at roughly 32%, a figure that has barely moved since 2015 despite billions in infrastructure investment. The European Union performs better at approximately 48%, but even that figure masks enormous variation, from Germany's 67% to Romania's 11%.

These numbers matter because material recovery sits at the nexus of three converging pressures. First, extended producer responsibility (EPR) legislation is expanding rapidly, with 12 US states now considering or implementing packaging EPR laws modeled on programs already operating in France, Germany, and Canada. Second, corporate sustainability commitments from companies including Unilever, Nestle, and PepsiCo have created demand for recycled content that existing systems cannot reliably supply. Third, the Inflation Reduction Act and EU Green Deal have unlocked public financing for circular infrastructure at scales not seen since the original recycling boom of the 1990s.

For founders building in the circular economy space, the gap between recycling aspiration and recycling reality represents a substantial market opportunity. The US EPA estimates that capturing the 50 million tons of recyclable material currently landfilled each year would generate $7.2 billion in commodity value and support over 150,000 jobs. But capturing that value requires fundamentally rethinking how cities collect, sort, and market recovered materials.

Key Concepts

Single-stream recycling allows residents to place all recyclable materials into one container without pre-sorting. While this approach increased participation rates by 25-35% when widely adopted in the 2000s, it also introduced contamination rates of 15-25%, degrading material quality and reducing the value of recovered commodities. The trade-off between convenience and quality remains the central tension in municipal recycling system design.

Optical sorting uses near-infrared (NIR) spectroscopy, visible light cameras, and increasingly AI-powered image recognition to identify and separate materials at high speed on conveyor belts. Modern optical sorters process 8-12 tons per hour with purity rates exceeding 95% for target materials, compared to 85-90% for manual sorting. The technology has matured significantly since 2020, with companies like TOMRA, Pellenc ST, and Machinex offering systems that can distinguish between polymer types (PET, HDPE, PP, PS) and even food-grade versus non-food-grade containers.

Contamination rate measures the percentage of non-recyclable material present in the recycling stream. US single-stream programs average 17-25% contamination, while European dual-stream and deposit return systems achieve 3-8%. Every percentage point reduction in contamination increases the market value of recovered bales by approximately $15-25 per ton, making contamination management one of the highest-leverage interventions in recycling economics.

Material recovery facility (MRF) is the centralized processing plant where collected recyclables are sorted, cleaned, baled, and prepared for sale to end markets. Modern MRFs operate on capital investments of $25-80 million depending on throughput capacity (typically 20-60 tons per hour) and incorporate multiple stages of mechanical, optical, and robotic sorting. The economics of MRF operation depend heavily on commodity prices, contamination levels, and the availability of downstream buyers for sorted materials.

What Happened: The Houston One Bin for All Pilot

Houston, the fourth-largest city in the United States with 2.3 million residents, launched the One Bin for All pilot in 2019 through a partnership between the city, the American Chemistry Council, and technology provider TotalRecycle (now part of Closed Loop Partners). The pilot addressed a stark reality: Houston's curbside recycling program, launched in 2010, had stalled at a 22% diversion rate with contamination levels consistently exceeding 20%.

The pilot operated in phases across three Houston neighborhoods encompassing approximately 15,000 households. Rather than asking residents to sort materials, the program collected all household waste in a single bin and used advanced sorting technology, including optical sorters, robotic pickers, and density-based separators, to extract recyclable materials from the mixed waste stream. This "dirty MRF" approach represented a fundamental departure from conventional recycling, which relies on residents to pre-sort materials.

Phase 1 (2019-2020) focused on technology validation. The pilot processed 2,400 tons of mixed waste and achieved a 40% material recovery rate, nearly double the city's existing curbside program. Recovered materials included plastics (PET, HDPE, PP), metals (aluminum, steel), paper and cardboard, and glass. Critically, the technology also captured material streams that conventional curbside programs miss entirely, including flexible packaging, textiles, and small electronics.

Phase 2 (2021-2023) expanded operations and focused on improving material quality. The addition of AI-powered contamination detection systems from AMP Robotics reduced the contamination rate in sorted bales from 12% to under 4%, bringing recovered material quality in line with virgin material specifications for many applications. Processing costs during Phase 2 averaged $85 per ton, compared to $65 per ton for Houston's existing single-stream MRF, but the higher recovery rate and improved material quality generated $45 per ton more in commodity revenue, partially offsetting the cost difference.

Phase 3 (2024-2025) addressed the economics of scaling. Analysis by the city's Solid Waste Management Department projected that a full-scale implementation serving all 400,000 Houston households would require $120-160 million in capital investment for two new advanced MRFs but would increase citywide diversion from 22% to an estimated 45-50% while generating $18-24 million annually in commodity sales.

Parallel Cases: Flanders and Kamikatsu

The Houston pilot did not operate in isolation. Two international programs offer complementary evidence about different approaches to the same challenge.

Flanders, Belgium has achieved a municipal recycling rate of 71%, the highest of any major region globally, through a system built on extended producer responsibility, mandatory source separation into 12 material categories, and pay-as-you-throw pricing. Residents purchase official waste bags at prices that embed the full cost of collection and processing, creating a direct financial incentive to minimize residual waste. The system costs approximately $115 per household annually but generates commodity revenues of $65-80 per household, resulting in net costs well below those of most US programs. Critically, Flanders invested three decades in public education and enforcement infrastructure that cannot be replicated quickly in other jurisdictions.

Kamikatsu, Japan, a town of 1,500 residents, implemented a zero-waste program in 2003 that requires separation into 45 categories. By 2024, the town achieved an 81% recycling rate, the highest documented rate for any municipality globally. While the extreme sorting requirements would be impractical for large cities, Kamikatsu demonstrates that high recovery rates are technically achievable when collection systems align with downstream processing capabilities. The town's approach informed the design of Japan's broader Act on Promotion of Resource Circulation for Plastics, enacted in 2022.

Key Metrics and Results

Metric	Houston (Pre-Pilot)	Houston (Phase 2)	Flanders	Kamikatsu
Material Recovery Rate	22%	40%	71%	81%
Contamination Rate	20-25%	3.5%	4-6%	<2%
Processing Cost ($/ton)	$65	$85	$78	$95
Commodity Revenue ($/ton)	$25	$70	$62	$45
Net Cost per Household/Year	$95	$72	$115	$130
Material Categories Captured	5-6	12-15	12	45

What Worked

Technology-driven quality improvement proved to be the single highest-impact intervention. The integration of AI-powered sorting, particularly robotic systems from AMP Robotics and optical sorters from TOMRA, reduced contamination rates by 80% compared to manual sorting. This quality improvement directly translated to higher commodity prices, with sorted PET bales commanding $380-420 per ton versus $220-260 for conventionally sorted material.

Resident convenience drove participation. Houston's one-bin approach eliminated the confusion and effort that suppress participation in conventional programs. During Phase 2, contamination complaints from downstream buyers dropped 65%, even as the volume of material processed increased by 35%.

End-market partnerships established before scaling ensured recovered materials had buyers. The pilot pre-negotiated offtake agreements with Indorama Ventures (PET recycling), Novelis (aluminum), and Georgia-Pacific (mixed paper), providing revenue certainty that conventional spot-market selling cannot offer.

What Did Not Work

Cost parity with landfilling remained elusive. At $85 per ton processing cost, the advanced MRF approach still exceeded Houston's landfill tipping fee of $42 per ton. Until landfill costs internalize environmental externalities or EPR fees shift processing costs to producers, the economics will continue to favor disposal over recovery in many US jurisdictions.

Glass recovery proved problematic across all three case studies. Glass containers break during collection and contaminate other material streams, particularly paper. Houston's pilot achieved only 55% glass recovery compared to 90%+ for plastics and metals. Flanders addresses this through dedicated glass collection points, but this requires additional infrastructure investment.

Scaling workforce requirements presented challenges. Advanced MRFs require technicians capable of maintaining robotic sorting systems, AI model calibration, and optical sensor arrays. Houston struggled to recruit qualified maintenance staff, leading to 15-20% downtime during Phase 2 that reduced effective throughput.

Transferable Lessons

First, contamination management is more valuable than volume growth. Reducing contamination from 20% to 4% increased commodity revenue by $45 per ton, while increasing collection volume by 20% added only $5-8 per ton in revenue. Cities should prioritize quality over quantity.

Second, technology selection must align with local waste composition. Houston's mixed-waste approach works for cities with low existing recycling rates and limited public willingness to sort. Flanders' source-separation approach works where strong public institutions can sustain education and enforcement over decades. Neither approach is universally superior.

Third, EPR legislation changes the economics fundamentally. In jurisdictions with producer responsibility (most of Europe, parts of Canada), producers pay $80-150 per ton for packaging recovery, making advanced sorting economically viable. In jurisdictions without EPR (most US states as of 2025), municipalities bear these costs directly and must rely on commodity revenues and avoided landfill costs to justify investment.

Fourth, data infrastructure matters as much as physical infrastructure. Houston's pilot generated granular data on waste composition, sorting efficiency, and material quality that enabled continuous optimization. Cities that invest in real-time monitoring and analytics achieve 15-20% better outcomes than those relying on periodic manual audits.

Action Checklist

• Conduct a comprehensive waste characterization study to understand local material composition before selecting sorting technology
• Benchmark current contamination rates and set target reductions with specific timelines
• Evaluate AI-powered sorting technologies from providers including AMP Robotics, TOMRA, ZenRobotics, and Machinex
• Negotiate multi-year offtake agreements with end-market buyers before committing to infrastructure investment
• Assess EPR legislation status in your jurisdiction and factor producer fees into financial models
• Develop workforce training programs for advanced MRF maintenance roles
• Implement real-time composition monitoring to enable continuous process optimization
• Model full lifecycle costs including capital depreciation, maintenance, and commodity price volatility

FAQ

Q: What is a realistic timeline for a city to modernize its recycling infrastructure? A: Plan for 4-6 years from initial feasibility study to full-scale operation. This includes 12-18 months for waste characterization, technology evaluation, and procurement; 18-24 months for facility design and construction; and 6-12 months for commissioning and ramp-up. Houston's pilot took five years from concept to Phase 2 completion.

Q: How much capital investment is required for an advanced MRF? A: Modern advanced MRFs processing 30-50 tons per hour require $40-80 million in capital investment, depending on the level of automation and local construction costs. Smaller facilities (10-20 tons per hour) can be built for $15-30 million but have higher per-ton operating costs due to reduced economies of scale.

Q: Can AI sorting technology work with existing MRF infrastructure, or does it require a greenfield facility? A: Robotic sorting arms and optical sorters can be retrofitted into existing MRFs, typically for $2-8 million per unit depending on throughput requirements. However, facilities built before 2010 often lack the conveyor configurations, electrical capacity, and data network infrastructure needed for AI systems, making retrofit costs 30-50% higher than greenfield installation.

Q: What contamination rate should a city target for its recycling program? A: Target below 5% contamination for access to premium commodity markets. Rates of 5-10% are acceptable for most domestic end markets. Above 10%, material quality degrades rapidly and some buyers will reject loads entirely. Achieving sub-5% contamination requires either advanced sorting technology or rigorous source-separation programs with enforcement.

Q: How do commodity price fluctuations affect the economics of advanced recycling? A: Commodity prices for recycled materials can swing 40-60% year over year. Sorted PET ranged from $180 to $480 per ton between 2020 and 2025. Multi-year offtake agreements with floor prices, diversified material recovery (capturing 12+ streams rather than 5-6), and EPR fee revenues all help buffer against price volatility. Cities should model economics at 25th-percentile commodity prices to stress-test financial viability.

Sources

• City of Houston Solid Waste Management Department. (2025). One Bin for All: Five-Year Pilot Results and Scaling Analysis. Houston, TX.
• US Environmental Protection Agency. (2025). Advancing Sustainable Materials Management: 2023 Fact Sheet. Washington, DC: EPA.
• European Environment Agency. (2025). Municipal Waste Management in Europe: Country Profiles and Benchmarks. Copenhagen: EEA.
• Bureau of International Recycling. (2024). World Recycling Industry Annual Review. Brussels: BIR.
• AMP Robotics. (2025). State of Recycling AI: Performance Data from 200+ MRF Deployments. Louisville, CO.
• Closed Loop Partners. (2024). Advancing Circular Systems for Packaging: Investment and Infrastructure Analysis. New York.
• TOMRA Systems ASA. (2025). Optical Sorting Technology Benchmarks: Purity, Throughput, and Economics. Asker, Norway.