Odor Controls for Rendering Plants: Emission Inventory, Equipment, Sizing, Field Data

EPA's AP-42 Section 9.5.3 lists what that vapor carries: organic sulfides and disulfides, C4 to C7 aldehydes, trimethylamine and C4 amines, quinoline, dimethyl pyrazine and other pyrazines, C3 to C6 organic acids. Thresholds for several of them reach 1 part per billion, and quinoline is the only constituent on the list classified as a hazardous air pollutant. A design for rendering plant odor control works three to four orders of magnitude below any occupational exposure limit, which is why a stack that passes its VOC test still draws complaints at the fence line.

Ventilation air from receiving, crushing and meal handling is the mirror image: tens of thousands of cfm at a fraction of the concentration. Duct the two into one header and the train gets sized for combined flow rather than combined load.

Adsorbers and dry scrubbers suit dilute service, on ventilation air or as a polishing step behind a column, if relative humidity stays under roughly 70 percent. Impregnated grades add chemisorption capacity for hydrogen sulfide, and that capacity is spent in proportion to sulfur throughput, so media replacement is an operating cost.

Biofiltration puts a microbial population on packing that metabolizes the compounds to carbon dioxide, water and mineral salts. No fuel bill, no reagent bill, and it works where thermal routes are uneconomic: very high volume at very low concentration.

Meal grinding, screening and conveying release PM carrying sorbed odorants. Baghouse, cartridge and silo filters take that duty and protect the gas-phase equipment behind them.

Budgeting for odor emission control for rendering plants comes down to three things: capture airflow, contact time, consumables.

Capture airflow comes first, because everything downstream scales with it. ACGIH's Industrial Ventilation manual puts capture velocity at 100 to 200 fpm for contaminants released at low velocity into moderately still air, which covers hooding over drainers and presses. Buildings are commonly held at 6 to 12 air changes per hour, slightly negative to outdoors, so leakage runs inward.

Contact time follows from bed geometry, and the two figures have to agree. A carbon bed 2 ft deep at 60 fpm gives 2 seconds of empty-bed contact time, the usual polishing target. A biofilter sits in another range: 60,000 cfm at 4 ft/min surface loading needs 15,000 sq ft of bed, and 4 ft of depth over that area yields 60 seconds of empty-bed residence time. Move depth to 3–5 ft and loading to 2–5 ft/min and residence time swings between roughly 36 and 150 seconds, so quote depth and loading together, never residence time on its own.

Consumables need measured inlet data, and this is where published factors get misused. The AP-42 blood dryer numbers, 0.76 lb filterable PM-10, 0.08 lb hydrogen sulfide and 0.60 lb ammonia per ton of dried meal, are controlled figures taken downstream of a cyclone, a venturi and three hypochlorite packed beds, and they carry an E quality rating. They indicate what escapes an existing train, not what arrives at a new one. Media tonnage and reagent demand have to come from source testing on the actual stream.

On a meat and bone meal line we placed a 10,000 cfm TORNADO RP vertical scrubber downstream of a juice vapor condenser, so oily organics were knocked out before reaching the packing. Maximum pressure loss across the column is 6 in. w.c., and the unit is rated 40 to 180 °F. Laboratory measurement after commissioning, with plain water as the scrubbing liquid, returned these figures.