In blast cleaning and surface preparation, most plants cannot say precisely where the money goes.

Purchasing holds a price per tonne. But the real cost is decided by three things: how much you use, what you blend, and how long it holds. Those three answers have traditionally lived in one experienced foreman's head. When he leaves, the knowledge leaves with him. When the media supplier changes a batch, the profile drifts. And nobody finds out until the roughness reading fails and the coating is rejected.

We turned that knowledge into a tool that can be called on demand: Malizhipei Abrasive Grading — an AI grading assistant that works backwards from the surface result you need, instead of selling you a catalogue number and hoping.

What it does, in one sentence

Give it five inputs: workpiece material, plate thickness, target roughness Rz, blast machine parameters, and the abrasive grade currently in use. It returns four things.

1. A four-stage size-band plan. The mix is broken into coarse cutting, fine cutting, primary leveling, and final refinement — each with a specific share, not a vague instruction to "blend something in."

2. Three-layer validation. Industry prohibitions plus the machine's permissible grit size are checked one by one. If the plate thickness cannot take an abrasive that coarse, the system stops you before a batch of workpieces is ruined.

3. Consumption estimate. Initial charge, replenishment interval, and the ratio of new to recycled media. This is what actually governs your purchasing rhythm and the cash tied up in inventory.

4. Risk notes. Which step is likely to fail under this specific condition, and where to look first when it does.

The basis is not rules of thumb. It is enterprise standard Q/HX 003-2026, Graded Blast Media for Roughness-Controlled Surface Preparation, backed by 25 years of field data and 31 patents.

One distinction that matters: grading is not blending

This is the most common confusion in media selection, and the most expensive one.

Blending means mixing several grades together in proportions learned by experience. It is a noun. It ends at the factory gate, and from there it drifts on its own.

Grading means designing a complete particle-size distribution curve backwards from the target roughness, and then holding that curve through the blasting cycle. It is a system. The factory gate is where it begins.

The difference shows up in two places:

• What is controlled. Blending controls a ratio — 3:1, 4:2, whatever. Grading controls a distribution: the full window across the size spectrum, and the job each size band is there to do.
• How it is held. Blending is replenished by feel. Grading is held by air classification, screening, and topping up in proportion to each size band.

The outcome is measurable. Run a single grade in closed circuit and Rz typically drifts ±18–25 μm. Run a graded mix and it holds within ±10 μm.

Ask your current supplier for that number. If they cannot answer, they are selling you a batch of media, not a controlled ratio.

One practical rule worth keeping on the shop floor: work back from dry film thickness — Rz ≤ total DFT ÷ 3. Exceed it and the peaks stand proud of the coating; salt spray results will fail.

Three conditions, and what comes out

Ship inner and outer plating. Block shops need different profiles on each face: outer plating needs a coarser anchor pattern for adhesion, inner plating must not be over-blasted. Two separate distributions, two separate plans — steadier than running one grade across the whole hull.

Structural steel and bridge box girders. Plate thickness ranges from 8 mm to 40 mm on the same machine. Grading pulls both ends into the window without changing equipment.

Wind towers and pipe internal/external coating. Long-life coating systems are strict on cleanliness and anchor profile. The plans include SSPC cross-reference (SP1–SP18 plus water-jet WJ series), so overseas clients can specify by SP number directly.

Now in public trial on Xiaoping

The tool is live on Xiaoping, a skill evaluation platform, and is currently in open trial — closing 29 October 2026.

Xiaoping works on a simple principle: real users run real conditions, and the results are published as reviews, openly readable, with public scoring. We put the tool there so people can verify it themselves instead of taking our word for it.

Three things we invite you to do:

1. Look at the listing. https://xiaping.coze.com/skill/68b66492-7934-444f-b9be-e8071e1c3dd8

2. Run your own condition. No registration, no fee. Enter the actual parameters from your line and see how far the output differs from what you do today.

3. Write down what you find. If it works, say so. If it misses, say that too — which condition is unsupported, which template is missing. During the trial we change the tool based on these notes, and blunt feedback is more useful to us than polite feedback.

If you are a process engineer or a buyer, bring numbers: your current media consumption per tonne, how often you top up, and the measured spread of your Rz. Those three figures settle the question faster than any brochure.

About us

Wuhan Hongxin Lixin Metal Products Co., Ltd. — 25 years in metallic abrasive manufacturing and surface preparation engineering. 31 patents, member of SSPC (China), and lead author of the China Steel Construction Society standard Technical Specification for Intelligent Grading of Metallic Abrasives for Blast Cleaning.

Product range: steel shot S110–S780, steel grit G12–G120, cut wire shot, and grinding balls. 38 partner production bases for regional delivery.

Technical support and free samples: www.malgw.com