Ready to Eat Food Manufacturer OEM/ODM One-Stop Customisation for Canned Grain Red Bean, Mung Bean and Oat Congee

A ready-to-eat congee in a can is a starch product before it is a flavour product. Three bases behave differently under the same process: red bean holds its shape and thickens slowly, mung bean breaks down fast and releases liquid, and oat brings a soluble fibre that changes with every minute of heat. One line can run all three. It cannot run them on one profile.

This page follows what actually decides whether a can of congee holds together. That means the cook endpoint, the fill and the retort, the checks that catch a drifting lot, and what has to change when a buyer moves from one base to another. The most expensive mistake is not the wrong sugar level. It is a cook profile that was set on one bean lot and never re-confirmed on the next.

A buyer lists a congee because it fills a slot: breakfast, a lighter meal, a lunchbox component, a shelf-stable item for a channel that has no cold storage. The first order follows the listing. The second order follows the can.

That second order is where most new grain products stall, and it stalls for reasons that have nothing to do with taste in a sample cup. The problems show up weeks later, in a warehouse, in a can that was fine when it was approved.

Congee is a suspension. Starch that retrogrades pulls water out of the gel, and the water collects at the top of the can or around the lid seam. A buyer opening a can from the back of the pallet sees a clear layer and concludes the product is unstable, even when the microbiology is clean and the label is correct.

Beans are an agricultural input. Moisture, size and hardness move between lots, and a cook set by clock time produces a softer or firmer product as the input moves. The buyer's tasters notice before the buyer's specification does, because the specification usually names ingredients rather than texture.

A congee positioned as a meal replacement has to carry a nutrition story on the front and a nutrition panel on the back. When the story is a weight claim, the product enters a regulatory conversation that a food factory cannot win, and the label gets refused at the port or pulled from the shelf.

A can of congee is a fixed portion of cooked starch and legumes in a sealed container, stable at ambient temperature. The commissioning decision is not only which bean goes in. It is the texture at opening, the texture at the end of the declared life, and the number on the front that the consumer can check with a spoon.

Those three are process outputs, not recipe choices. A brief that names only the ingredients leaves the most important half of the specification unwritten.

The plant behind this line was founded in 2018 and runs more than 30 intelligent high-speed production lines, with 350 to 400 employees. Between 80% and 90% of output is exported, and the company has served more than 2,000 brand customers as of April 2024.

R&D spending has stayed above 10% of annual revenue, and the recipe library holds more than 5,000 formulas, many of them built for retail and foodservice clients across grain, dessert and soup categories.

The grain and dessert lines share a plant, a quality system and a batch numbering scheme. A congee project sits next to dessert and soup projects, which matters in practice: the same retort vessels, the same seam inspectors, the same release procedure.

The company holds utility model patents granted in 2020 covering a packaging bowl, a sterilisation device and a production device, and design patents granted in 2021 covering a rotating self-heating bowl shell and a visible bowl-shaped container. Those patents are not about congee, but they come from the same engineering group that sets the retort profiles.

More than 30 lines is a statement about scheduling capacity rather than about any single product. It means a run can be planned around an SKU change instead of being forced into whatever slot is left, which is what makes a three-base range practical on one site.

A recipe library above 5,000 formulas is useful for a buyer who wants a starting point rather than a blank page. It is also a caution: a library is only as good as its records, and the value in a congee project is the record of what each formula did after retort, not the formula itself.

Three figures are deliberately absent from this page: floor area, daily output and annual revenue. The numbers published elsewhere for each of them disagree with one another, and a supplier that quotes one figure in an article and another in an audit is not worth the time it takes to check.

What a buyer can check instead is a site registration, the scope on a certificate, and the batch file from a comparable run. Each of those has a date and a scope attached to it, which is more useful than a capacity number in a brochure.

Congee fails in three predictable ways, and each has a different control. The table below maps the failure a buyer sees to the cause on the line and the action that prevents it. It is the section to bring to a supplier conversation, because it turns a vague complaint into a question with an answer.

The table is deliberately short. A longer list would look thorough and would be harder to use. Each row names one measurable item, and each measurable item has a file behind it.

Four of the five rows are about the can being different from the sample, not about the can being unsafe. That grouping matters when a buyer is choosing a supplier, because a factory that only talks about microbiology has answered the smaller half of the question.

The row about the label is the exception, and it is commercial rather than technical. A refused label costs a print run and a delay, and it is the one failure on the list that a buyer can prevent entirely before a single can is filled.

It moves the conversation from ingredients to endpoints.

A buyer who asks for "red bean congee, mildly sweet" has described a direction. The factory then chooses the texture, the solid content and the cook profile, and the buyer discovers the result at the sample stage. That is workable if the buyer is willing to accept what comes back, and expensive if they are not.

A buyer who asks for a drained solid content, a bean-to-liquid ratio, a target firmness and a declared life has described a product that can be validated. The second brief takes longer to write and much less time to make. It also survives a change of factory, which the first one does not.

Every supplier in this category lists certificates. Almost none of them can tell a buyer who issued a certificate, which site it covers, and which products fall inside its scope. Those three questions separate a document from a claim.

Ask for the certificate itself, not the logo. The document names the certification body, the site address and a scope statement, and the scope is where most surprises sit. A certificate covering a neighbouring plant, or covering a category that does not include your product, is not evidence about your can.

The second question is the audit cycle. A management system certificate is a statement about how a site operates at a point in time, reviewed on a fixed rhythm. It is not a statement about the contents of a particular batch. Product-level evidence is the batch file and the release sheet.

Three documents travel with a shipment in the normal case: the commercial invoice and packing list, the health or export certificate required by the destination market, and the batch release documentation the buyer asks for.

Which of those is mandatory depends on the market, and the list changes. A buyer entering a market for the first time should ask for the current requirement before the artwork is approved, because a label change after printing costs more than a label change before it.

The line does not put a certification logo on a label unless the certificate names the product and the site. That rule has cost projects time, and it has kept products out of ports.

A congee project has three changeable layers, and they cost very different amounts. Knowing the order is the difference between a two-week adjustment and a two-month project.

A formula change inside the same base is the cheapest. Sugar, salt, the bean-to-liquid ratio and the addition of a second grain are all adjustable, provided the change does not move the fill weight or the texture target far enough to invalidate the cook profile.

Changing the ratio shifts the viscosity, and viscosity shifts the heat penetration. A modest change is confirmed on a trial run. A large one, such as doubling the solids, is a new profile and a new validation.

A new fill weight is a new geometry and a new heat penetration study. A label change is cheap in production terms and expensive in time, because the artwork has to be re-approved and the ingredient order changes whenever the recipe does.

ODM, OEM and private label orders run on the same line, with the same parameters and the same batch file format. That matters because the most common worry about contract manufacturing is that the branded work is done properly and the private-label work is subcontracted.

It is not subcontracted here. A private-label run uses the same retort vessel, the same seam check and the same release criteria as an ODM run, and the batch file looks identical apart from the label code. A buyer who wants to verify that can ask for a redacted batch file from either type of run and compare them.

A Singapore client needed a ready-to-eat product for people who have difficulty swallowing, and the market offered very little in that category that was also shelf stable. The brief was not a flavour brief. It was a texture brief with a nutritional floor.

The product had to be spoonable without lumps, soft enough to swallow without a second stage of chewing, and stable at ambient temperature so it could sit in a care setting without a cold chain. It also had to look like food rather than a supplement, because the people eating it were not patients in a clinical sense.

Two things had to be solved together: the particle size and the starch system. Milling the grain finer solved the lumps and created a new problem, because fine particles settle faster and the can separates during the hold. Raising the starch level solved the separation and made the product gluey.

The answer was a base cooked to a softer endpoint and a starch system that held the particles in suspension without thickening further after the retort. The result was a range of ready-to-eat rice and grain bowls that entered a special-diet setting rather than a mainstream shelf. The same cooking and suspension logic carries over to the red bean, mung bean and oat bases on this line today.

The three bases share a line and a release procedure, and they do not share a cook profile. Red bean, mung bean and oat absorb water at different rates, gelatinise at different temperatures, and respond differently to the same retort cycle.

Starch behaviour under heat is not a detail behind the product. It is the product. Wang and colleagues reported in the International Journal of Biological Macromolecules in 2017 on a heat-moisture treatment applied to red adzuki bean starch. The treatment raised the pasting temperature and suppressed granule swelling during heating. It changed both the thickness and the digestion behaviour of the finished material.

The practical reading is that the same bean, cooked with a different moisture history, does not produce the same can. A cook profile is therefore tied to a lot specification, not to a product name.

The oat base adds a second variable. Its soluble fibre changes under heat as well as its starch. Kivelä and colleagues reported in Carbohydrate Polymers in 2012 that oat beta-glucan in aqueous solution undergoes oxidation and chain cleavage during heat treatment, with measurable degradation at 120 °C. The oat base therefore loses molecular weight through the retort, and its viscosity has to be measured rather than assumed. Two oat lots that analyse the same on paper can produce cans of different thickness.

Three checkpoints carry most of the risk on this line.

• Incoming beans. Moisture, 1,000-grain weight and defect rate are measured per lot against a purchase specification.
• Cook endpoint. Firmness is measured by compression on a fixed sample, and the cook stops on the measurement rather than on the clock.
• Post-retort hold. The can is held before release, and the release decision includes free liquid and texture against a reference sample.

Everything downstream, from the seam check to the retort chart, protects the can. These three protect the product inside it.

A mung bean lot arrived with a moisture content at the top of the accepted window. It passed the incoming check, because the window had been set on moisture alone. On the line, the beans absorbed the soak water more slowly, the cook ran to its usual time, and the centre of the batch came out firmer than the reference.

The cans were filled, seamed and retorted normally. Forty-eight hours into the hold, the release sample showed free liquid at the top of the can and a texture that was grainy rather than smooth. The lot had passed every measurement taken before the retort and failed the one taken after it.

The lot was blocked in the warehouse and not released. Committed units were replaced from the next validated run, and nothing shipped against the failed texture.

Two things changed. The incoming bean specification gained a moisture ceiling tied to the soak ratio, so a wetter lot is soaked in less water rather than for the same time. The cook endpoint moved from a time setting to a compression measurement at the end of the cycle, and that measurement is now written on the batch sheet and signed. A new bean lot also triggers a bench cook before it can enter a production batch.

A pallet of 168 g cans is heavy, and a metal can dents. Most of the damage in this category happens between the pallet and the shelf, and most of it is preventable at the packing stage rather than at the carrier.

Production is planned by SKU rather than by incoming order. A three-base range is scheduled so that the changeover between bases is planned in advance, because a mung bean run after a red bean run requires a cleaning cycle and a fresh cook profile confirmation. An order that arrives late is fitted into the next planned slot for that base rather than inserted into the middle of another run.

That structure is why a delivery date is a function of the base, the pack and the current schedule, and why a date quoted without those three is a guess.

Three measures carry most of the protection, and each has a test behind it rather than a preference.

A carton is specified with a compression strength that is measured, not assumed. The test stacks cartons to the height the container will see, at the humidity the route will produce, and the result is written into the pack specification. A carton that passes on the factory floor and fails in a container usually failed the humidity condition rather than the load.

The stacking pattern keeps the load inside the pallet footprint, so the bottom cartons carry compression rather than shear. A new pack also goes through a drop and vibration check before it ships. A carton that fails a stack test fails in a container, usually at the bottom of a stack, and usually after the vessel has sailed.

The third measure is the simplest. A can that is filled and seamed correctly survives handling that a can with a marginal seam does not. That is why the seam check is a production control rather than a shipping control.

Three categories do not go onto this line, and saying so early saves a meeting.

• Medicines and products presented as medicine. A can of congee is a food. A label that implies treatment moves it into a regulatory category this plant does not make.
• Concepts that only work with a disease claim. If the product needs a health condition in its name to sell, the claim has to come off the label and the product has nothing left to say.
• Ingredients with no documented identity. A botanical or a grain blend with no species name, no supplier specification and no certificate of analysis cannot enter a batch, because the batch file has to name what went into the can.

Anything else, including formats this line does not currently run, is a development question rather than a refusal.

The fastest enquiry contains five items: the destination market, the pack size, the base or bases you want, the texture and sweetness direction, and the declared shelf life you need. With those five, the first reply can be a process recommendation rather than a capability list.

A reference sample is worth more than a page of description. A can of the product you are trying to match tells us the particle size, the free liquid and the sweetness in one object, and it lets the first trial start from a real target.

If you want to see the line, the plant is open for a visit, and a redacted batch file from a comparable congee run can be shown under the usual confidentiality terms. Send the five items through the contact page and ask for samples of the red bean, mung bean and oat bases.

The endpoint is set on a compression measurement taken on a fixed sample at the end of the cook, not on a clock. Each base has its own target, and the target is confirmed on a trial run for each incoming lot that falls outside the reference window. Red bean, mung bean and oat sit at different targets because they absorb water and gelatinise differently.

A new fill weight is a new geometry, so it needs a new heat penetration study before production. The study identifies the slowest heating point in the can and confirms the F0 at that point. The retort then runs at 121 °C with a hold set by that validated value rather than by a fixed time.

Each batch carries an incoming inspection sheet, a cook record with the compression result, a filler log, a seam record, a retort chart and a release sheet, all filed under the batch number. Retained samples are held beyond the declared shelf life so a complaint can be traced to the lot and the cook that produced it.

This line does not run aseptic filling, and it cannot take a pack that will not survive a retort cycle. A chilled product, a pouch format or a powder that the consumer reconstitutes are all possible as separate development projects with their own tooling and validation. The limit is the line, not the company.

Send the destination market, the pack size, the base or bases, the texture and sweetness direction, and the shelf life you need. Those five let us recommend a process and a trial plan instead of sending a brochure. A reference sample shortens the first round more than any other single item.