Water is food contact
From the harvest bin to the carton, produce meets water constantly — dump tanks, flumes, washers, hydrocoolers, drenches, final rinses — and the rules treat all of it as food contact. GLOBALG.A.P. (version 6, the scheme most South African export producers certify against) requires water used in harvest and post-harvest activities to meet the microbial standards for drinking water, backed by a documented water risk assessment reviewed annually and at least one laboratory analysis per season taken near the point of use — municipal supply included. South African law lands in the same place from another direction: a pack-house is food premises under R638 of 2018, whose definition of water is compliance with SANS 241. The reference criterion underneath both is simple and strict: E. coli not detectable in 100 mL. The PPECB's statutory export audits walk the same hygiene ground. In practice, then, a pack-house needs two water systems working at once: a supply train delivering drinking-standard water (filtration plus UV or chlorination on any borehole or surface source), and a process-water regime keeping the recirculating tanks safe all shift.
The dump tank is a shared bath
Recirculating water touches every fruit that passes through it, so its job is not to clean the produce as much as to stop one contaminated bin infecting the day's entire pack. The established regime is chlorine with pH discipline: around 50–75 mg/L of free chlorine held at pH 6.5–7.5 covers most post-harvest pathogens within minutes of contact. The pH is not a detail — it is the control: chlorine's killing form (hypochlorous acid) dominates below pH 7.5 and vanishes as pH climbs, while pushing below 6.5 off-gasses chlorine into the packing hall and corrodes equipment. That is why serious tanks run two dosing loops — one for chlorine, one for pH — each from its own stock tank, pump and injection point, with mixing distance between the two injectors and ventilation over the tank: acid and hypochlorite meeting undiluted release chlorine gas, and the acid should be a food-contact grade (citric, phosphoric or sodium bisulphate), not pool acid. Control the chlorine on a measured free-chlorine residual (amperometric probe or DPD kit), with ORP as the floor alarm: below about 650 mV the sanitiser is not working, and at the target residual and pH it reads well above 750 mV, so ORP proves the residual is active but cannot set it. Citrus pack-houses following the CRI regime run higher, 75–100 mg/L at ORP 800 mV or above. These are the published regimes for particular crops and particular schemes, not settings to copy across: the registered product's label and your own scheme's approved protocol set the concentration, the contact time and the fruit it may be used on. Manual test-kit checks are logged for the auditor either way.
Soil, leaves and fruit juice consume chlorine continuously, so demand rises through the shift; the scheme rules expect a defined change-and-replenish schedule for recirculated water, judged on sanitiser hold, turbidity and sight. Screening out debris and filtering the recirculation loop stretches every dosed litre. Where chlorine fights the organic load badly, peracetic acid is the accepted alternative: it holds its activity better than chlorine in the presence of organic matter and does not form chlorinated by-products, at the price of sharper handling. It is not indifferent to dirt — organic load still consumes it, and it still has to be dosed to a measured residual rather than to a recipe. Published pack-house concentrations sit in the region of 40–80 mg/L, but the registered product's own label and your scheme's approved protocol set the concentration, the contact time and the crop it may be used on. And produce has limits too: some lines mark or pit when chlorine runs far too high, which is one more argument for controlled dosing over bucket chemistry.
The Listeria lesson
The hard modern lesson of post-harvest water is that immersion can push water into fruit — through stems and stem scars, especially when warm fruit meets cold water — which is how a dump tank or hydrocooler without a maintained residual becomes an injection system for Listeria. The response written into current practice: never run immersion or drench water without its sanitiser residual verified, cool the water knowingly (temperature differentials drive uptake), and back it with the environmental swabbing programme the schemes now expect in packing areas. The final rinse, after the shared water, must itself be of drinking standard — the last water the fruit sees is the cleanest.
Rule of thumb
Two systems, both provable: drinking-standard supply water (tested each season at the point of use, treated at the source if it cannot pass) and process water held at 50–75 mg/L free chlorine (75–100 for citrus), pH 6.5–7.5, on separate, never-shared dosing loops controlled on measured chlorine, with logged checks. Refresh recirculated water on a schedule, filter the loop, keep the residual alive in anything fruit is dunked in, and finish on potable rinse water. The audit is passed at the dosing pump, not the paperwork.