BRENNEN

Industries / Food, farming and forestry / Food and beverage processing

Cleaning and changeover

Cleaning and changeover is the work of removing food residue, germs and allergens from plant equipment, and of switching a line from one product or pack to the next without carrying the last product into it.

7

Milestones achieved, from 1920s to 2024.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1920s

    Dairy agrees on hygienic design

    The first 3-A Sanitary Standards set rules for the hygienic design of dairy equipment, meaning shapes, materials and joints that can be cleaned. They are named for the three groups that wrote them together: health inspectors, equipment makers and processors. A 3-A Symbol for equipment that meets them came into commercial use in 1956.

    3-a.org

    Credit

    US dairy regulators, equipment makers and processors, now 3-A Sanitary Standards, Inc

  2. 1955

    Cleaning without taking apart

    Before clean-in-place (CIP), dairies took pipes and tanks apart and cleaned them by hand. CIP pumps cleaning and rinse liquids through closed equipment in a fixed sequence. Dale Seiberling's obituary says he developed the world's first automated CIP system in 1955, for Hopewell Dairy. A 2023 trade article dates the first automated system to 1953, in a US family-owned dairy.

    dignitymemorial.comfoodengineeringmag.com

    Credit

    Dale Seiberling; Hopewell Dairy

  3. 1989

    Europe's hygienic design group

    Since 1989 the European Hygienic Engineering and Design Group (EHEDG) has brought equipment makers, food companies, universities and public health bodies together. It writes hygienic design guidelines and tests and certifies equipment.

    ehedg.org

    Credit

    EHEDG

  4. 2008

    Listeria hid inside slicers

    Listeria in deli meats from a Maple Leaf Foods plant in Toronto killed 22 people in Canada. The plant had found Listeria in 2007 and 2008 and cleaned with procedures standard in the industry. A House of Commons committee reported that organic material had built up deep inside the meat slicers, found only when they were fully taken apart. The independent investigator, Sheila Weatherill, called for the equipment maker, processor and Canadian Food Inspection Agency to work together more closely, including on validating sanitation procedures.

    Lessona machine that cannot be cleaned without full teardown gets cleaned only on the surface.

    noscommunes.cacanada.ca

    Credit

    Sheila Weatherill (2009 report); House of Commons agriculture committee

  5. 2020

    Codex names changeover risks

    The Codex code of practice on food allergen management deals with cross-contact, an allergen getting into a food that should not contain it. It lists inadequate equipment cleaning at product changeover among the causes, and says cleaning must be validated for each allergen, process and product. In July 2026 Codex added guidelines that allow "may contain" labels only after a risk assessment.

    Lessonfor an allergen changeover, "clean" has to be shown with a test for that allergen.

    gov.plfood-safety.com

    Credit

    Codex Alimentarius Commission (FAO and WHO)

  6. 2024

    Residue left on equipment

    Listeria in Boar's Head liverwurst made in Jarratt, Virginia, sickened 61 people in 19 states, of whom 60 were hospitalized and 10 died. FSIS named inadequate sanitation as a notable cause, including meat and fat residue left on equipment from the previous day, condensation dripping onto exposed product, and cracked floors that held water.

    Lessonsampling the product can miss what a look at the equipment shows.

    cdc.govfoodsafetynews.com

    Credit

    CDC (case counts); USDA FSIS (January 2025 review)

  7. 2024

    Cleanability test for open machines

    EHEDG published Guideline 57 in March 2024, a method for testing how well machines that are open to the room, such as conveyors and slicers, can be wet-cleaned. Testers apply a soil containing an optical tracer, clean it with an industrial robot arm so the cleaning is the same every time, and look for what remains. Fraunhofer IVV helped develop it. EHEDG calls it a screening tool for poor design that does not predict performance in industrial cleaning.

    ehedg.orgehedg.org

    Credit

    EHEDG; Fraunhofer IVV

Watch for

  1. Robot rinses and foams equipment

    CleanBotix of Grand Rapids, Michigan, a 2025 spin-off from Clean Logix, is building Rosie, a mobile robot arm meant to rinse, foam and sanitize equipment in meat and poultry plants. PickNik Robotics, which supplied its motion software, reports "80% of human cleaning speed, with fewer re-cleans and better ATP swab results", and says paid pilots have begun with large poultry processors. No customer is named and no outside test is published.

    Achieved whena named processor or an independent party publishes results from routine sanitation shifts, such as ATP pass rates and re-clean counts against a human crew.

    picknik.aipicknik.ai

    Claimed by

    CleanBotix and PickNik Robotics, November 2025

  2. Sensors end CIP washes sooner

    Ecolab launched CIP IQ, which reads sensors on clean-in-place circuits and uses "real-time soil load data" to judge when cleaning is complete. Ecolab says dairy trials cut alkaline wash time by 10 to 20%, from its own trials. A 2016 UK project led by Martec with the University of Nottingham noted that CIP runs "blind" and is set for the worst case.

    Achieved whena customer or an independent party publishes measured savings in wash time and water over months, with cleanliness confirmed by swabs or rinse tests.

    foodanddrinktechnology.comnottingham.ac.uk

    Claimed by

    Ecolab, September 2025

Open

  1. 01

    Residue map after washing

    A camera with ultraviolet and violet lights, carried on a cart or a cobot, that photographs every surface of a washed line in the dark and returns a map of spots that still glow, since many food residues glow under these lights. Test it on a mock line soiled with diluted meat juice, milk or a fluorescent tracer.

    The brief
    Credit · unclaimed
  2. 02

    Wash-down coverage tracker

    Fixed cameras over a wash-down area and a vision model that records, panel by panel, which surfaces the foam and rinse spray reached during manual cleaning, and shows the crew a list of missed areas before they leave. Test it on a rig of numbered panels, using coloured foam or a harmless fluorescent tracer in the water to show the true coverage.

    The brief
    Credit · unclaimed

Open 01 · the brief

Residue map after washing

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A camera with ultraviolet and violet lights, carried on a cart or a cobot, that photographs every surface of a washed line in the dark and returns a map of spots that still glow, since many food residues glow under these lights. Test it on a mock line soiled with diluted meat juice, milk or a fluorescent tracer.

Why it is still open

The 2024 International Dairy Foods Association paper says visual inspection remains the industry standard for the first check after cleaning, with ATP swabs as spot checks. The Danish Meat Research Institute calls this mandatory inspection "people-reliant". A handheld imager, SafetySpect's CSI-D, shows a person where to swab, but someone has to aim it. No shipped system that scans a whole line turned up in searches for this file.

idfa.orgdti.dkfrontiersin.org

Built with what exists

A cobot such as a UR5e, or a push cart, a machine-vision camera with LED lights at 275 nm and 405 nm (the wavelengths CSI-D uses) and a filter that blocks the lamp light, an image-segmentation model trained on labelled photos, and a commercial ATP meter and protein swabs to check its calls.

  1. First step

    Soil stainless-steel coupons with three dilutions of meat juice, photograph them under the lights, and record the weakest dilution the camera sees beside ATP readings.

  2. Done looks like

    On a mock line with 50 marked sites, 25 left with a known low level of soil and 25 properly cleaned in an order the operator does not know, the map's dirty or clean call matches the ATP swab result at 45 or more sites on each of three days. The scan takes less time than a technician's visual walk-down.

Open 02 · the brief

Wash-down coverage tracker

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

Fixed cameras over a wash-down area and a vision model that records, panel by panel, which surfaces the foam and rinse spray reached during manual cleaning, and shows the crew a list of missed areas before they leave. Test it on a rig of numbered panels, using coloured foam or a harmless fluorescent tracer in the water to show the true coverage.

Why it is still open

PSSI, a large US contract cleaning firm, logs time, water temperature, chemical strength and water pressure on tablets across about 500 plants, and the 2019 report on its system does not say it records which surfaces were reached. A May 2025 trade article lists "missed zones" among common human errors in manual cleaning. No product that maps coverage during manual wash-down turned up in searches for this file.

food-safety.comprovisioneronline.com

Built with what exists

Two or three cameras rated for wash-down, an open video segmentation model such as Segment Anything 2 run on a local GPU workstation, and an ultraviolet torch and tracer to produce the true coverage for scoring.

  1. First step

    Film five people foaming and rinsing a table-sized rig with coloured foam, mark by hand which panels were covered, and check whether a simple colour threshold already gets most panels right.

  2. Done looks like

    On a rig of 40 panels, across 10 runs by at least five different people, the tracker's reached or missed call agrees with an ultraviolet photo taken after each run on at least 36 of 40 panels per run, and it flags every panel that got no foam.