Sorting the waste problem
The waste management industry, in my opinion, is a peculiar and hard industry to be in. It is fairly concentrated and heterogenous globally.
Through these years, I never imagined the journey of waste once I sorted it and put in the bins. Little did I know that this so-called unpopular industry is going through its own transformation by virtue of big policy changes and the push by #zerowaste and #climatechange campaigns.
So, the waste we produce ends up in the landfills, incineration or at material recovery facilities (MRFs for recycling) depending on the type of material, sorting accuracy and where it is produced. For instance, a major chunk of plastic produced by US and Europe used to be recycled in China until 2018 when it imposed a ban on importing waste.
Only 13.5% of the total waste generated gets recycled (source: World Bank). Unattractive economics, low recovery rates, poor collection infrastructure and low awareness on waste disposal are some contributing factors.
We live in the times of climate change and these numbers couldn’t be more at odds with the need of the hour. There is little blame on the waste management companies, as, recycling is inherently an unattractive business. It remains a strong pursuit of the bigger waste managers, gaining from the scale needed at MRFs. On the other hand, the smaller ones experience periodic shutdowns of MRFs (especially during slump in oil prices), making it unsustainable and inconsistent. To make such industries fly, strong policy regulations and frameworks are needed. But the application of emerging technologies seems promising; AI has been instrumental in transforming multiple industries and waste management isn’t an exception.
Different parts of the value chain have adapted to this change differently.
Waste Collection & Transfer:
For waste collection, the developed countries at least have the requisite infrastructure with segregation at the bin level, such a thing doesn’t even exist in a lot of countries. Despite that the level of awareness of recyclable vs. non-recyclable waste is unimpressive, leading to very high contamination in bins. Waste managers end up using a lot of resources (labor, time and money) to re-sort manually. Mix-up of wet and dry waste makes even the recyclable non-recyclable due to contamination. All these factors make the cost of collection highest contributing ~45% (highest among landfill, composting, incineration, etc.) to the total cost of processing waste of ton.
Key challenges:
- Incorrect sorting of waste at source;
- Unoptimised routes of trucks;
- Lack of monitoring of bin levels
Progress so far:
There are numerous start-ups of the likes of Compology (US), Resourcify (DE), BigBelly (US) that are optimising pick-up routes by monitoring and compacting fill-levels of bins through machine-based sensors, helping them reduce the cost per pick-up. There is also a focus on monitoring contamination at least at the community-bin level by triggering alarms to trucks. However, this is yet to see the mass adoption because of high cost of investment per bin and lack of solid proof-of-concept.
There is still more work left …
The councils, waste managers and consumer goods companies have a role to play in this at least for creating awareness, enforcing fines and having stringent targets on using recycled inputs. Well-defined targets on re-use of input materials, elaborate and clear labels of RECYCLING or NO RECYCLING are some measures in the right direction.
Solutions addressing incorrect sorting at source (households or offices) are yet to see the light of day. The harder nature of the problem here makes waste collection an interesting space for community councils, entrepreneurs and investors.
- Data collection using AI at a bin or truck level will help in training models and also minimising costs down the chain;
- CPG companies can gain insights from customer panel/SKU data from research firms to track down reprocessed material. In no way I imagine a blockchain platform De Beers setup to track diamonds, but this surely adds up to solid bottom-line and SDG goals for them (re-processed plastic uses 50% less electricity than virgin; source: British Plastics and Rubber Magazine)
Lesser-known-fact: Greasy pizza boxes, soiled food cardboard plates/ containers are not recyclable
Waste Processing (MRFs):
Once the waste reaches MRFs, it is separated manually (to plastic, paper, metals) along conveyer belts, baled and sent to be re-processed as inputs.
Key challenges:
- Human errors and safety on conveyer belts;
- High costs/resources for plant operations;
- Lack of data on waste composition
Progress so far:
This use-case has seen the highest traction so far with AI-based computer vision. With players such as Sadako, AMP Robotics, Zen Robotics and Greyparrot.ai, robots-based sorting is becoming mainstream among MRFs. The hardware is a robot equipped with AI-based computer vision. Setting up robots is capital intensive and also limiting in terms of reach; however start-ups are coming up with unique business models to overcome this. The robots save costs in the long-term, improve accuracy and the AI captures data and learns from every API dip during sorting. Even though there is little disclosure on the benefits yet, the adoption of it seems promising!
Looking to the future….
- The optimised operations promise increase in efficiency, waste managers may look at setting up production of glass, paper or plastic using reprocessed inputs;
- The CPG companies will look for long-term partnerships with waste managers for reprocessed material, this is already happening as we speak (source: British Plastics and Rubber Magazine);
- As we gain scale, technologies aimed at efficient waste processing and energy recovery will be needed
The wider adoption of recycling and its economics pose a classic chicken-and-egg problem, I am curious to see what will come first!
