Chocolate business

Chocolate Production Line: Machines You Need, in Order

On this page
  1. First decide what you will make (it changes the line)
  2. The line in order
  3. Bean processing (if bean-to-bar): roasting, winnowing, grinding
  4. Mixing and refining
  5. Conching
  6. Storage tanks and tempering
  7. Forming: molding, enrobing, depositing drops
  8. Cooling tunnels
  9. Packaging
  10. Standalone machines vs a turnkey line
  11. How to compare vendors
  12. Sizing the line
  13. Sources and open questions

A full bean-to-bar chocolate line runs in this order: bean cleaning, roasting, cracking and winnowing, grinding the nibs into liquor, mixing, refining, conching, holding, tempering, forming, cooling, demolding where there are molds, and packaging with a foreign-body check. Start from couverture instead of beans and the line begins at melting and tempering.

If you’re still deciding whether to build a plant at all, and at what size, start with how to start a chocolate factory, which covers business models, registration and costs.

First decide what you will make (it changes the line)

The product sets the shape of the line. Bars, filled shells, enrobed centers and drops share the chocolate-making half, but after tempering each needs different forming and cooling equipment, and your starting material sets how much of the first half you need at all.

Two questions matter most.

What do you start from? A bean-to-bar plant roasts, winnows and grinds its own beans. A plant that buys couverture leaves all of that, and the raw-bean food safety work that comes with it, to the supplier. In between is a plant that starts from cocoa mass and cocoa butter. A public research institute’s model project in India, published by CSIR-CFTRI, is one: it needs a mixer, a three-roll refiner and a conche ahead of tempering.

What shape leaves the line? A 2023 review of chocolate processing by Ditchfield and colleagues describes the last stage as a fork: tempered chocolate goes into molds, as bars or shells, or onto products as a coating, then gets demolded where needed and packed. Drops for bakeries are a third branch.

If compound chocolate is part of the range, the tempering question changes too. Coatings made with fats other than cocoa butter mostly don’t need it, as real vs compound chocolate explains.

Write these answers down before you talk to vendors. Every quote you get back will depend on them.

The line in order

The order below comes from published process descriptions. Which steps a given plant needs is an engineering decision for that plant, so treat the table as a starting point.

Step Bean-to-bar bars From couverture or cocoa mass
Bean cleaning, roasting, cracking, winnowing Yes No
Grinding nibs into liquor Yes No
Mixing, refining, conching Yes Only if starting from mass
Tempering Yes Yes (compound mostly not)
Forming Bar molds Shell molds, cold stamping or an enrober
Cooling Tunnel One or more tunnels
Demolding Yes With molds; not after an enrober
Packaging and foreign-body check Yes Yes

The process order comes from the Ditchfield review. Shell molding and cold stamping are described in a University of Birmingham doctoral thesis by Le Révérend (2009), and enrobing in a 1974 US patent. Drops have their own line, covered in how chocolate drops and chips are made.

The bean steps don’t have one fixed order. The same review says beans are roasted, broken and winnowed, but that “these operations can be performed in different order and under different conditions according to chocolate type and manufacturer preference.” Some plants roast whole beans; others shell first and roast the nibs.

Bean processing (if bean-to-bar): roasting, winnowing, grinding

Bean processing turns fermented, dried beans into cocoa liquor. It needs a cleaner, a roaster, a cracker and winnower, and a mill, and it’s where the raw-bean food safety work sits. The last step, grinding, takes nibs from roughly 0.5 cm to below 40 µm in several stages, using knife mills, disc mills, ball mills or refiners, among other equipment, according to the Ditchfield review.

Incoming beans carry stones, metal and debris. The FDA’s 2024 draft guidance on preventive controls lists magnets, sifters and screens alongside metal detectors as devices that keep physical hazards out of food.

Roasting

Roasting is done in rotary drum roasters or continuous roasters, heating the beans from ambient to between 110 and 160 °C depending on the bean type, per the Ditchfield review. A 2022 roasting review by Peña-Correa and colleagues describes the common industrial setup as a large rotating drum through which hot air of 110 to 150 °C is blown for 20 minutes to 2 hours. The same review calls drum roasters batch in one place and continuous in another; both exist.

Scale pushes toward continuous machines: large companies handle tons of cocoa a day, the review notes, so continuous processes suit them better than batch ones. Faster methods such as fluidized bed roasting have been known for more than 40 years and still aren’t common.

Roasting whole beans or nibs is a real choice for the line. Many manufacturers still roast whole beans because the shells come off more easily afterward, even though heat moves more efficiently into smaller pieces, Peña-Correa and colleagues write. The food safety side of that choice, including why Salmonella data for whole beans can’t simply be applied to nibs, is in the bean-to-bar process.

Don’t expect a vendor to settle the science: a 2022 review of roasting physics by Rojas and colleagues says the effect of heating rate on cocoa is still unknown. Ask for control and repeatability instead.

The industry’s cocoa quality guide (CBQR), in its sample-roasting protocol, says conditions have to be mapped by temperature, time and loading for each bean variety and each type of roaster, and that raw and roasted beans must be kept apart to prevent cross-contamination. So ask how the roaster measures air and bean temperature, what load it repeats at, and how raw and roasted product are kept separate.

Cracking and winnowing

The beans are broken and the shell is blown away from the nibs, a step called winnowing. The shell matters beyond yield. The quality guide calls it often the most contaminated part of the bean, for microbes, molds and some residues. Nib yield depends on whether the winnower sizes the broken pieces; the figures are in the bean-to-bar process.

Airflow sets the trade-off. A 2016 study by Firmanto and colleagues of a small winnower at an Indonesian public research institute measured shell left in the nibs at different suction speeds: 3.70% at 0.54 m/s, 1.06% at 0.72 m/s and 0.53% at 0.90 m/s. At low airflow, shell pieces about the size of the nibs stay in. The machine worked well only with well-dried, fermented beans.

The study doesn’t report how many nibs were lost into the shell stream, and stronger air can carry good nibs out with the shell. So ask a vendor for both figures, shell in the nibs and nibs in the shell, measured on beans like yours. One number tells you half the story.

Mixing and refining

Mixing combines liquor, sugar, milk powder and other ingredients into a paste; refining then grinds that paste until the particles are fine enough.

The two machines have to fit each other: a roll refiner needs a paste of the right consistency, and the mixer has to deliver it. Chocolate refining explained compares melangers, roller refiners, ball mills and the refiner-conche, including the conditions behind vendor throughput claims.

One point belongs in the line plan. During refining, the friction can push the chocolate to 60 °C, and the machine is cooled to keep the viscosity workable, according to the Ditchfield review. That cooling is a utility: ask each vendor how much cooling water the refiner needs, at what temperature.

Conching

Conching is a long stirring step that changes the chocolate’s flavor and flow. The Ditchfield review describes it as stirring at temperatures above 50 °C for a few hours to remove undesirable volatile compounds and develop flavor, viscosity and texture.

It also breaks up the fat-trapping clumps a roll refiner leaves behind, which is why a roll refiner always needs a separate conche while a melanger or a refiner-conche does part of this work in the same machine. The mechanism is in chocolate refining explained.

Conches are heated, so energy is part of the comparison. The processing review gives one example of heat recovery: heat from the roasting medium reused in conching. Batch or continuous, capacity, cycle time: those answers will come from vendors, so ask for them in the same units (see sizing below).

Storage tanks and tempering

Between conching and forming, chocolate is held in heated storage tanks and then tempered, so the cocoa butter sets in a stable crystal form. For the tanks, US hygiene law sets the baseline: holding, conveying and manufacturing systems must be designed so they can be kept clean, and the instruments that measure and record temperatures must be accurate and precise (21 CFR 117.40).

Tempering

Tempering means taking the chocolate through a temperature curve so it sets with a gloss and a clean snap. Why that works is in what is chocolate tempering.

A 2025 review of tempering by Stobbs and colleagues describes the classic cycle. The chocolate is heated past 50 °C, often to 60 °C. It’s then cooled under controlled stirring, typically to around 27 to 29 °C for dark chocolate, and gently reheated to around 31 to 32 °C, just below the melting point of the stable Form V crystals. Industrial plants do this in continuous tempering machines, as a 1998 comparison by Bolliger and colleagues notes.

For a buyer, the first finding is that there’s no single setting. The Stobbs review says tempering temperatures and times depend on the origin and composition of the cocoa butter, so the same conditions can give bloom with a different cocoa butter source. A machine you’ll run with several recipes needs temperature zones you can set per recipe.

The method varies too. Seeding, adding chocolate or cocoa butter that already holds stable crystals, speeds up crystallization, the same review notes. In an ETH Zurich doctoral study by Grob (2020), chocolate seeded with a cocoa butter crystal suspension mostly at 0.5% of the cocoa butter crystallized faster; at 1.25% it was over-tempered, and its melt viscosity stayed high. And the 1998 comparison found that a high-shear crystallizer matched a conventional tempering machine’s crystallization quality with much less residence time and a lower outlet temperature.

You’ll see figures for how much of the fat should be in stable crystals. Sources give different ranges, measured in different ways, so don’t compare machines on that number. Ask how the output is checked instead: with a temper test or a tempermeter, how often, and what happens when the reading drifts. Batch and continuous machines are compared in our guide to tempering machines.

Forming: molding, enrobing, depositing drops

Forming gives the chocolate its final shape, and it’s where the configurations split. Molding fills molds, enrobing coats centers on a belt, and depositing shapes small drops straight onto a belt, a line covered in how chocolate drops and chips are made.

Molding bars and shells

For molding, tempered chocolate is deposited into molds, cooled until it sets, then demolded, per the Ditchfield review. Solid bars follow that sequence directly. Solid, shell and one-shot methods are compared in chocolate molding.

Hollow shells take more steps. The Birmingham thesis lays out the traditional sequence:

  1. Freshly tempered chocolate is deposited into the molds.
  2. It’s cooled from the mold side, so a layer sets against the mold.
  3. The mold is turned upside down and the chocolate that hasn’t set runs out, to be melted and tempered again.
  4. A filling can be added.
  5. Tempered chocolate closes the piece, or two halves are joined.
  6. The pieces set in a cooling tunnel.

The thesis is frank about the weaknesses. Most of the chocolate has to be reworked. Because the shell crystallizes somewhat at random, the space left for filling varies, which leads to under- or overfilling, and an even shell thickness is hard to hold.

Cold stamping is the alternative it describes. A chilled plunger, typically at −10 °C, presses into the freshly deposited chocolate for about 3 seconds, so the mold sets the outside shape and the stamp the inside. Process time drops and shell volume stays within tight tolerances. The chocolate can cool locally at up to 100 °C a minute, fast enough for unstable crystals to form, yet the thesis reports a plant making up to a million mini eggs a day with the right crystal structure. Ask how the stamp’s cooling and the room’s humidity are controlled.

Mold design affects cooling too. For typical 125 g plates, a 2021 study by Grob and colleagues found the chocolate came away from the mold wall soonest at an apparent heat flux of 550 W/m², not at the most intense cooling.

Enrobing

In enrobing, centers travel on a belt over a wave or under a curtain of chocolate held at the right temperature, which coats them completely, then pass straight into a cooling tunnel or onto a cooling belt. That description comes from a 1974 US patent, US 3,987,751.

The coating you get depends on the chocolate as much as the machine. In a 2023 study by Işık Ertural and colleagues of dark chocolate on wafers, the share of coating picked up fell from 72.3% without emulsifiers to 45% with a mix of lecithin and PGPR, because the chocolate flowed differently. So a vendor’s coating figure only means something for a stated chocolate, with a stated viscosity, on a stated product.

Cooling tunnels

A cooling tunnel takes the heat out of molded or coated chocolate under controlled conditions, so it sets with stable crystals before packing. Its length and zones can set how fast the whole line runs.

Grob’s ETH Zurich thesis groups tunnel designs into longitudinal, pater noster, nested and spiral types. Longitudinal tunnels are mostly used for coated products such as pralines. Larger and filled products that need longer cooling usually go into pater noster, nested or spiral coolers, or a combination.

Heat leaves the product by conduction, convection and radiation. Earlier work cited in the thesis says chocolate should first cool gently, in radiant or gently moving air, with contact cooling the most efficient after that, and the air should be dry to prevent condensation.

The thesis gives a sense of the load. Cooling 1 g of properly tempered dark chocolate from 30 °C to 12 °C means removing 75 J. By our arithmetic, that’s about 10 kW for 500 kg of chocolate an hour, for the chocolate alone. Molds, belts and air leaks come on top, so treat it as a minimum.

It also measured a real tunnel: about 500 kg of product an hour, air at 12 °C, three cooling zones plus a pre-zone, and a total dwell time of 20 minutes 15 seconds. In one longitudinal zone, heat transfer at the top of the mold was much lower and warm air was trapped beneath it, which affected the fat crystal network and the quality of the chocolate. Better airflow raised overall heat transfer by 10%; with highly conductive materials, cooling was up to 15% faster.

These are figures from single plants and experiments, so don’t copy them as settings. Faster isn’t automatically better either: cooling too fast can produce unstable crystals and fine surface cracks, according to Lonchampt and Hartel (2004), one route to the bloom explained in why chocolate turns white.

When you compare tunnels, ask for the tunnel type and why it suits your product, the number of zones, the air temperature, speed and direction in each, the dwell time at your line speed, how humidity is controlled, and whether the belt or mold is cooled by contact.

Packaging

The packaging machine wraps the product in a material that protects it, and the material matters as much as the machine.

Start with what the film has to do. In a 2023 study by Canzoneri and colleagues of milk chocolate stored for up to 12 months, oxygen-impermeable packaging cut the formation of cholesterol oxidation products by up to 34% compared with unsealed standard packaging.

Fiber-enriched milk chocolate needs a medium or high water vapor barrier to keep its flow properties in storage, a 2022 study by Verde and colleagues found. Decide the film first, then ask whether the machine runs it and how seal integrity is checked. Parts and films that touch chocolate fall under food contact materials rules, covered in the hub guide.

The foreign-body check is more than a detector at the end of the line. Refiners, mills and crackers have metal-to-metal contact, and where magnets, screens, metal detectors or X-ray belong, with the FDA texts behind them, is set out in the RFQ checklist’s foreign-body section.

Standalone machines vs a turnkey line

Buying a turnkey line means one supplier designs, installs and starts up the whole line and answers for how the machines work together. Buying standalone machines means you, or an integrator you hire, take that job on. In the EU, that job comes with legal responsibility.

Under the EU machinery rules, machines linked under a common control system to make one product form an “assembly of machinery,” and whoever puts the assembly together is treated as its manufacturer, with the conformity assessment, declaration and technical file for the whole line. With a turnkey line that is normally the supplier; with standalone machines it can be you. The tests, the paperwork and what changes when you add a machine to an existing line are set out in the RFQ checklist, under who is the manufacturer of the line. The CE marking basics are in the hub guide.

Whichever route you take, write acceptance tests into the order: a factory acceptance test before shipment and a site acceptance test after installation. What to put in them is in the RFQ checklist.

How to compare vendors

Compare vendors on written answers to the same specification. Send everyone the same request for quotation and ask for a line-by-line reply, or the quotes will arrive with different scopes.

The 17 questions to put in it are in our chocolate equipment RFQ checklist, with the hygienic design rules the machines have to meet. Its first row is the capacity claim, stated the way the next section sets out.

Sizing the line

Size the line from the slowest step, not from the fastest machine’s brochure. As a Kansas State guide on buying food equipment puts it, “a process is only as fast as the slowest unit operation” (Rausch and Erickson, MF-2096). The same guide notes that equipment capacity can’t be matched exactly to the process.

The method for turning a sales target into kilograms per hour, with losses and equipment effectiveness, is in the capacity section of the hub guide. Here the problem is different: comparing what vendors claim.

A capacity claim can only be compared with another if it states four things:

  1. The product and recipe, including its viscosity. In the wafer study, coating pickup ranged from 72.3% to 45% with the recipe alone, and one refiner maker’s spec sheet notes its figure refers to dark chocolate and varies with the product (see chocolate refining explained).
  2. The format: mold, piece weight, fineness. The Kansas State guide notes that capacity can be stated per item or by weight, so molds per minute and kg per hour convert only through piece weight and mold layout.
  3. How time is counted: net running hours or hours including cleaning and changeovers, and for batch machines the full cycle time. Without a cycle time, no kg per hour can be worked out.
  4. Which machine is the bottleneck. A molding machine’s molds-per-minute figure means nothing if the cooling tunnel can’t take that rate. In the ETH Zurich example, a 500 kg/h tunnel held product for just over 20 minutes.

Put them in the first row of your RFQ and ask every vendor to state their figure the same way.

The Kansas guide adds that every product interacts with the equipment differently, so its viscosity and stickiness should be known before you buy. For a chocolate line, that argues for a trial with your own chocolate before you sign.

Sources and open questions

  • We found no independent source comparing bean cleaners, liquor mills, mixers, conches, holding tanks, drops depositors, wrapping formats or demolding mechanisms, or describing a bar line’s vibrating and mold-heating stations or an enrober’s extra stations (bottoming, blowing off excess, decorating).
  • Cost, lead time and commissioning time of single-supplier against multi-supplier lines: no independent study yet.