The Glass Cullet Bound: Furnace Energy Keeps Falling to 100 Per Cent Cullet, but Contamination Sets the Usable Limit at One Third

Adding recycled glass (cullet) to a container-glass furnace saves energy, and a common expectation is that some intermediate cullet fraction must be optimal, beyond which further cullet stops helping. This paper tests that expectation with an explicit energy and contamination model, computed in the accompanying `compute.py`. The result is a negative result for the energy bound and a positive result for the quality bound. On energy alone there is no interior optimum: net energy per tonne of good glass falls monotonically from 5.306 GJ/t with no cullet to 4.395 GJ/t at 100 per cent cullet, a saving of 17.2 per cent, even after sorting energy and a rising reject rate are charged. The binding limit is contamination. A village-scale colour-sort and wash line that removes 88 per cent of ceramic, stone and porcelain contamination still delivers cullet containing 3.0 kg of that contamination per tonne, three times a 1.0 kg/t low-reject specification. The cullet fraction at which the feed reaches that specification is therefore 0.33. The same line reaches 92 per cent single-colour purity, below the 98 per cent required for clear (flint) glass, so it cannot supply flint production at any cullet fraction. The line is nevertheless economically sound for green and amber production at 5 tonnes per day: 1,250 tonnes per year, an annual margin of $52,500 on the stated prices, and a payback of 2.82 years. SIMULATED, NOT MEASURED.