TDHX Minerals · Dalian Taiding Hengxin International Trade Co., Ltd.info@tdhxtrade.com  |  Call / WhatsApp +86 18142387636
Technical insight

Mineral Powder Factory Audit: What to Check Inside the Plant

Mineral Powder Factory Audit: What to Check Inside the Plant

Most factory audits produce a folder of photocopies and a recommendation that answers nothing. This checklist treats the audit as a measurement: decide whether you are approving a supplier or releasing a batch, follow the material in process order, and test the five points that decide whether the certificate is real.

Most factory audits of a mineral powder plant end the same way: a full day of walking, a folder of photocopies, and a closing sentence that recommends approval without answering the only question that mattered — will the next container behave like the sample? The gap is rarely effort. It is that the auditor arrived with a general supplier questionnaire, and the plant answered it truthfully and uselessly.

This is the checklist we hand to buyers who visit our plant on the Haicheng talc belt, and it is the same list we would use at someone else’s plant. It treats the audit as a measurement with a stated purpose, follows the material in process order rather than the route the host prefers, and ends with findings written so that they can actually be closed. The steps before and after the visit — legal entity, export rights, batch documents, commercial signals — belong to our supplier verification guide. Everything below happens inside the gate.

First decide which audit you are running

“Factory audit” covers two different examinations, and running them together is the most common reason a report ends up long but undecided.

  • Supplier approval audit. You are deciding whether this plant can make your grade repeatedly. You audit the process, the equipment and the quality system. The output is a judgement on capability that should survive a change of order.
  • Shipment or order audit. The goods are already made. You are deciding whether one batch matches the agreed specification. You audit records and samples, and the output is a release decision on one lot.

The checklists overlap; the questions do not. An approval audit asks what would make this line drift. An order audit asks what proves this lot did not drift. With one day on site, run the approval audit on the line that will make your grade, and settle the shipment question from the batch record of the lot you are buying. Do not try to do both across a whole plant — you will do neither properly.

Request the document pack before you book the flight

Half of a factory audit can be finished at a desk, and doing it first makes the site visit much sharper. Ask for five items and read them before you travel:

  • The specification sheet for the exact grade, with the test method named beside each line and not only the limit.
  • Certificates of analysis for the last three batches of that grade, so you can see whether the numbers move together or wander independently.
  • The batch numbering rule on one page: how a batch number is built, and what each part of it can be traced back to.
  • The equipment list, with mill types and the fineness range each one is set up for.
  • Calibration and verification records for the balance, the drying oven, and any particle-size instrument used on your grade.

Two things deserve attention at this stage. If the specification names a mesh size but no method, the number is not yet a specification: mesh is a statement about a sieve, and a sieve has requirements of its own. The general test methods behind extender grades are standardized too — ISO 3262-1 defines what an extender is and sets out the methods that the later parts of the series rely on — so a specification sheet without method names is missing half of itself. And if three consecutive certificates are identical to the last decimal, ask whether those are per-batch measurements or a reused template — a real run of three batches in a mineral plant almost always moves somewhere.

Walk the line in process order, not the order you are offered

Plants route visitors past the clean parts, usually starting at the mill floor and finishing in the showroom. Ask instead to start at the ore pile and end at the loading bay, following the material. Our own published flow is a workable skeleton because it matches most Chinese non-metallic plants: quarry, raw material sorting, primary breaking, jaw crushing, feed to the grinding line, packaging, truck loading. Each stop has one specific failure mode worth asking about.

Where the grade is made: the classifier, not the mill

Buyers look at the mill. The mill sets how fine the material can become; the classifier — the separator sitting on top of the circuit — decides what actually leaves as product. This is the most important stop on the tour and the easiest one to disguise with a swept floor.

Ask three questions at the separator: what the cut point is currently set to, who is permitted to change it, and where that change is written down. Then ask what happens when the line switches grade. The answer you want involves a written changeover sequence — stop the feed, run the circuit empty, purge before the new grade is collected. The answer that should worry you is “the operator adjusts it and we check the first bags”. A line that changes grades by feel will ship a container whose first tonnes and last tonnes are not the same product, and a certificate drawn from the top of the silo will never show it.

Iron is an invisible process step

Grinding wears metal. Liners, grinding media and the internals of the circuit all contribute fine iron to the powder, and in a white mineral that iron shows up as brightness loss and as a chemistry line the buyer never asked for. Magnetic separation is therefore a real process step rather than decoration, and it is almost never on the visitor route.

Ask to see the magnetic elements themselves, not the housing: the magnets or the electromagnetic matrix, the cleaning schedule, and the log that proves the schedule was kept. Then ask what happens to the collected iron, and whether the magnets are checked for strength on any interval. A plant that cannot produce a cleaning log is telling you the separator exists but the procedure does not. Our own magnesia line removes iron, silicon and foreign matter by magnetic separation before the material is sold, for exactly this reason: it is cheaper to remove it at the separator than to argue about it after arrival.

Five control points that decide whether the certificate is real

A certificate is only as good as the laboratory that signs it and the sample trail behind it. Five points are worth more than a wall of framed documents.

  1. Test sieves. Certificates, last verification date, and the physical condition of the cloth.
  2. Particle-size instrument. Which method is used below the sieve range, which reference material verifies it, and when it was last verified.
  3. Balance, oven and moisture determination. Calibration labels, and whether moisture is measured on the basis the specification assumes.
  4. Retained samples and the batch trail. Whether a lot you received can be reconstructed from records plus a physical sample.
  5. Production and changeover log. Who moved the cut point, when, and what was done to the material in between.

The sieve cabinet and the laser unit

If the grade is specified by mesh, the plant must own test sieves, and those sieves have a standard behind them. ISO 3310-1 sets the technical requirements and corresponding test methods for test sieves of metal wire cloth, and it applies to aperture sizes from 125 mm down to 20 µm. That lower figure matters more than it looks: below roughly 20 µm, woven wire stops being a dependable measuring instrument, so fineness has to be expressed by another method — laser diffraction, for instance, under its own standard. A plant that claims 2000 mesh while showing you a stack of sieves is telling you something about its method whether it intends to or not.

In the cabinet, look at three things: the certificates supplied with the sieves, the date of the last verification, and the condition of the cloth. Sagging, torn or dented mesh changes the result even when the paperwork on the wall is perfect. If there is a laser unit, ask about the dispersion method, the reference material used for verification, and the date of that verification. Two plants can both report a D50 and mean two different things by it.

The sample room and the batch trail

Ask to see retained samples for the last six months, then pick a shipment you actually received and ask the plant to walk backwards from it: loading record, packing record, silo or blend record, mill run, raw ore batch. Every link should carry a number that connects to the next one. The finding you are looking for is not “no records” — most plants have records — but a link that can only be bridged by someone’s memory.

Retained samples deserve a second look. A sample room full of bags without labels, or labels without date and batch, cannot settle a dispute. The useful question is not whether samples are kept, but whether the sample for container X can be on the table within a day, and who signs for it.

The packing scale is the last quality step

Packing is treated as logistics and audited as quality. A 25 kg bag that weighs 24.6 kg is a commercial problem; a bag whose inner liner is torn is a quality problem that surfaces as caking six weeks later in a warehouse. Watch one full cycle at the packing head: the scale calibration check, the liner insertion, the sealing, the batch code applied to the bag, and the pallet wrap. Then ask whether the batch code on the bag is the same code that appears on the certificate. If the two systems do not meet, the traceability chain breaks at the very end, which is where it is most expensive to repair.

Sampling: witness the draw and seal a duplicate

Most quality disputes are not about analysis. They are about the sample. If you did not see the sample taken, you are relying on a bag that could have come from anywhere in the silo, or from the previous run.

  • Agree the sampling plan in writing before the visit: how many increments, from which points, combined or kept separate, and what quantity per sample.
  • Witness the draw on the lot you are buying, not on a display pile. Note where in the silo or the stack the increments came from.
  • Split and seal in front of you: one for the plant, one for your laboratory, and a third retained if the value justifies it. Label with batch, date, point of draw and the initials of both parties.
  • Keep the seal intact until analysis. A sample that arrives open has no evidentiary value, however good the laboratory.

The same logic applies to pre-shipment inspection by a third party: the inspector’s value is in the witnessed draw and the sealed duplicate, not in the report format. We have written separately about when third-party inspection pays off and how to word the clause.

Write findings so they can be closed

An audit report is a work instruction, not an opinion. Every finding should carry four parts: what was observed, which requirement or specification line it touches, what evidence would show it has been fixed, and a date. Anything without a closure test becomes a paragraph that both sides agree to ignore.

  • Critical. The plant cannot currently make the grade to specification, or the traceability chain is broken. No order until the item is closed and re-verified.
  • Major. Capability exists but a control is missing or not followed: no changeover purge record, magnets not cleaned on schedule, sieves past their verification date. Order may proceed with a written corrective action and a date.
  • Minor. Documentation and housekeeping items that do not change the product today but will next year.

Resist the phrase “recommend improvement”. It cannot be closed, so it will not be. Write “fit a record card at the separator and log the cut point at every grade change; provide the first month of cards with the next shipment” instead — that is a finding someone can finish.

What changes by mineral

The process skeleton is shared; the failure modes are not. Five adjustments to the same checklist:

  • Talc. Whiteness and lamellar structure carry the commercial value, and both are sensitive to grinding and to iron. Ask how brightness loss across the mill is controlled, and keep mineralogy — tremolite, the asbestos group — on a separate evidence track, because a chemistry panel cannot answer it. Our talc powder guide covers the grade ladder from 325 to 3000 mesh.
  • Ground calcium carbonate. Purity and brightness are set largely by the ore, so the audit question shifts to consistency: ore source control, classifier stability and iron pickup. Moisture deserves close attention because the specification window is tight and the oven method matters.
  • Dolomite. Two oxides have to land inside their windows at the same time, so ask how ore is blended to hit both, and confirm that raw and calcined material run on separate lines. A plant that puts both through one circuit will eventually mix them. See the dolomite powder range for the grades we hold.
  • Magnesia. The kiln is the process. Calcination temperature defines the reactivity of caustic calcined magnesia, while dead burned and fused grades depend on much higher temperatures. Ask for kiln records, not kiln photographs. Then ask how the finished powder is protected from humidity, because a hygroscopic product stored in an open shed changes before it ships.
  • Aggregate products. Pebbles and white sand are audited on size distribution, colour sorting and clay or fines content, which makes the sieve and washing stages more important than the mill. The mesh size guide shows how those size bands are written.

Remote audits: what video proves and what it cannot

A video call has become a normal substitute for the first visit, and it is genuinely useful: you can confirm that equipment exists, read nameplates, watch a sample being drawn, and look at documents held up to the camera. It saves a trip and it filters out plants that cannot produce basic paperwork.

What video cannot prove is everything that requires walking. Whether two silos are actually isolated from each other, whether the floor between them is clean, whether the log was filled in this morning or last month, whether the operator adjusts the cut point by feel once the camera is off — none of that survives a screen. If the order value is material, plan a physical visit before the second container rather than before the first. The first container can be protected by a third-party inspection; the relationship cannot.

Re-audit triggers, and how we work with buyer auditors

An approval audit has a shelf life. Re-audit when the ore source changes, when a new grade or a new line is introduced, after a significant complaint, or simply when the last visit is old enough that the people you met have moved on.

At our own plant we would rather remove the friction than defend it: the document pack above is available before a visit, the retained sample room opens on request, sampling can be witnessed at the packing head or at the silo, and a third-party inspector is welcome at any stage — we would only ask that the scope and the sampling plan be agreed in writing first, so that the report answers the question you actually have. If you are still choosing between suppliers, the grade selection guide is a better starting point than an audit. If you already know the grade and want the specification in writing, send the target and we will match it against the product lines we run.

Frequently Asked Questions

Can we approve a supplier on documents alone, without visiting?

Documents will carry you a long way: legal entity, a specification with named methods, three batches of COA, the batch numbering rule, the equipment list, calibration records. What they cannot show you is separation - whether two grade silos are physically isolated, whether the floor between them is clean, whether a changeover purge actually happened, and whether the log was filled in on the day it claims. Those are the findings that decide whether the second container matches the first. A practical split: approve the paperwork remotely, then visit before the second order rather than before the first, and use a third-party pre-shipment inspection to protect the first.

When should we book the visit, and how long does it take?

Book it for a day when the plant is actually running your grade - a visit during a changeover, a maintenance stop or a quiet week shows you a building, not a process. One full day is enough to walk one line properly from the ore pile to the loading bay, plus the laboratory and the sample room; two days if you want a second line or the kiln section as well. Tell them in advance which grade you are buying and which records you will ask to see, because a plant that knows the scope will have the batch file ready, and a plant that does not will spend the morning looking for it.

If the audit finds a major finding, can we still place the order?

It depends on which finding, and that is exactly why findings should be graded rather than listed. A major item - magnets not cleaned on schedule, no changeover purge record, sieves past their verification date - describes a control that is missing while the capability is present. Those can usually be ordered against, provided the corrective action is written, dated, and verified on the pre-shipment sample or inspection. A critical item - the plant cannot currently hold your specification, or the traceability chain is broken - is not a discussion about timing; the material you would receive is not the material you specified.

Related articles

Who is responsible for this page

  • PublisherThis article is published by Dalian Taiding Hengxin International Trade Co., Ltd., the export arm of Haicheng Taiding Huaxin Mineral Products Processing Co., Ltd., whose plant is in Haicheng, Liaoning, China.
  • Basis of the dataMesh, whiteness, chemistry and moisture figures come from retained production samples and batch inspection records of our own plant. Statements about standards are quoted verbatim in the sources section at the end of this page, each with a link to the issuing body's official page.
  • Updates and correctionsThis article was last updated on . If you find an error in a figure or a statement, write to info@tdhxtrade.com and we will verify and correct the page, updating the date at the top.

Standards and sources quoted

The quotations below are taken verbatim from published standards; each link goes to the issuing body's official page.

“This document gives the definition for the term extender and specifies test methods that are required for most of the subsequent parts of ISO 3262.”

— ISO — International Organization for Standardization, ISO 3262-1:2020, Extenders — Specifications and methods of test — Part 1: Introduction and general test methods, official source

“ISO 3310-1:2016 specifies the technical requirements and corresponding test methods for test sieves of metal wire cloth. It applies to test sieves having aperture sizes from 125 mm down to 20 µm, in accordance with ISO 565.”

— ISO — International Organization for Standardization, ISO 3310-1:2016, Test sieves — Technical requirements and testing — Part 1: Test sieves of metal wire cloth, official source

Need a custom specification?

Tell us your target mesh size, whiteness and application. We send samples and a quotation within 24 hours.

Request a Quote WhatsApp