Welding Consumables — Supplier Qualification Guide

When the Approval Passes and the Production Lots Don’t #

A structural fabricator in Southeast Asia qualified a Chinese E7018 supplier in Q1. The initial sample approval went cleanly: tensile strength 560 MPa, elongation 26%, diffusible hydrogen under 4 mL/100g deposited metal, all within AWS A5.1 limits. By Q3, their NDT team was flagging a 3× increase in hydrogen-induced cracking on fillet welds. No process parameters had changed. The root cause, confirmed after three weeks of investigation, was a moisture re-absorption issue in the flux coating — not from the buyer’s storage, but from the supplier’s. Electrode flux moisture content had drifted from 0.15% at initial sample to above 0.40% at production volume, well past the 0.30% threshold that triggers diffusible hydrogen elevation in low-hydrogen grades.

The fabricator had no incoming moisture verification step. Their COA acceptance was visual — they checked that the form existed and the numbers were present. They never cross-referenced consecutive batch COAs, never performed spot-check re-baking, and never requested the supplier’s pre-shipment electrode conditioning records.

That’s the pattern we see most often in our QSP-14 audit program: failure at the process discipline level, not at the chemistry level. The initial material may be exactly what it claims to be. What changes at volume is process control — flux drying protocols, baking temperature consistency, storage humidity between production and packaging. None of that shows up on a standard COA unless you know exactly which fields to interrogate.

The COA Fields That Actually Predict Production Quality #

A welding consumable COA from a Chinese supplier typically contains six to ten fields. Procurement teams tend to focus on the obvious ones: tensile strength, yield strength, elongation. These matter, but they are also the easiest to represent favorably, since a single compliant melt can generate reference data that gets applied to many subsequent batches.

The fields with genuine predictive value are:

Diffusible hydrogen content — for low-hydrogen grades (E7018, E7016, E308L-16), this should be tested per AWS A5.1 Annex A or ISO 3690, reported in mL/100g deposited metal, and flagged if it exceeds 4 mL/100g for H4-designated electrodes or 8 mL/100g for H8. A COA that lists tensile strength to three significant figures but reports hydrogen as simply “meets H8” without a numeric value is a red flag.

Flux moisture content at time of packaging — this is rarely included on standard COAs unless you specifically request it. We ask for it as part of our incoming qualification form (our internal document references this as INS-22C, Supplementary Flux Characterization). Acceptable range for low-hydrogen electrodes is ≤0.20% moisture by weight. Anything above 0.25% at time of packaging means the buyer is absorbing storage risk that the supplier should have eliminated.

Chemical analysis lot traceability code — every heat of core wire has a melt certificate. The COA should contain a traceable melt code that can be cross-referenced to a core wire mill certificate. If the supplier cannot provide mill certificates on request, the chemical analysis data on the COA is unverifiable. In our audit program, roughly one-third of new suppliers could not produce traceable mill certificates within 48 hours of request — which tells you something about their raw material traceability posture.

Coating concentricity and eccentricity values — per AWS A5.1, maximum eccentricity for E7018 is 3% for 4.0mm diameter. Coating eccentricity causes arc wander, uneven bead profile, and increased spatter. It rarely gets tested by buyers at incoming, but deviations are a clear signal of manufacturing consistency problems.

Batch re-baking temperature and duration — if the supplier re-bakes electrodes before packaging (as they should for low-hydrogen grades), the COA or a supplementary document should confirm the baking cycle: typically 300–350°C for 1–2 hours for E7018. Absence of this record doesn’t mean re-baking didn’t happen. It means the supplier doesn’t document it — which is a process discipline signal, not a materials signal.

The mechanical properties table — the one most procurement teams read first — is actually the lowest-risk field on the COA for established chemistries. AWS A5.1 tensile requirements for E7018 are 480 MPa minimum. Any competent compounder hits that with margin. The risk is in the process fields, not the property fields.

Decision Framework: How Qualification Rigor Scales with Application Risk #

If the application is structural — pressure vessels, lifting equipment, load-bearing weldments subject to code inspection under AWS D1.1 or EN ISO 15614-1 — the qualification protocol needs to include a formal weld procedure qualification (WPS/PQR) using the candidate consumable. This means destructive testing: bend, tensile, Charpy impact at the specified test temperature. For low-temperature service rated to −40°C or below, Charpy impact energy must be confirmed at temperature, not inferred from room-temperature data. Chinese suppliers frequently provide impact values tested at 0°C and present them as applicable to −40°C service. They are not.

If the application is MRO maintenance welding on non-critical mild steel — structural repair, bracket fabrication, general fitment — the qualification burden is lower, but the COA verification discipline should remain the same. The distinction is that you accept wider mechanical property variation (within AWS A5.1 Class limits) but you do not relax the moisture and hydrogen controls, because the failure mode (HAZ cracking in restrained joints) is the same regardless of whether the joint is critical or maintenance.

If you are consolidating multiple welding consumable grades under one supplier — which global MRO teams increasingly do for logistics simplicity — the qualification scope changes because you are now evaluating manufacturing breadth, not just single-grade compliance. A supplier who makes excellent E7018 may have a weak stainless program. I’d prioritize a multi-grade sampling protocol here: qualify each major grade class independently, not on the basis of E7018 performance alone.

One boundary condition worth stating explicitly: this framework assumes SMAW (stick) and GMAW (MIG) consumables. For SAW flux-wire combinations, the qualification approach changes materially because the flux is a separate supply chain variable with its own basicity index, moisture absorption characteristics, and bulk storage requirements. The same supplier quality posture doesn’t transfer without additional flux-specific checks.

A specific, non-obvious recommendation: before volume commitment, request six consecutive production batch COAs — not six individual sample COAs. The difference matters because a supplier can cherry-pick samples. Consecutive lot COAs over a 90-day window show production process stability in a way no single sample approval can.

Practical Guidance for Buyers #

When qualifying a Chinese welding consumable supplier, don’t start with the mechanical properties datasheet. Start with the COA format itself — specifically whether diffusible hydrogen is reported as a measured numeric value or as a classification claim (“meets H8”). That single field tells you more about the supplier’s testing discipline than anything else on the form.

The risk scenario to watch: a supplier who passes initial qualification on small-run production, then transitions to higher-volume output using a different core wire source or flux batch. Our QSP-14 program has logged this pattern in 4 out of 11 suppliers evaluated over the past two qualification cycles. The trigger is almost always raw material sourcing pressure at volume — the compounder switches flux supplier or wire rod mill without notifying the electrode manufacturer, and the change propagates through without requalification. Incoming hardness and moisture spot-testing catches this; visual COA review does not.

Before volume commitment, insist on a 3-lot incoming inspection using your own lab or a third-party per ISO/IEC 17025-accredited facility. Test parameters: diffusible hydrogen per ISO 3690, coating eccentricity per AWS A5.1, and at minimum one Charpy impact set at your service temperature. Run this on three non-consecutive production lots, not three samples from the same pallet. That’s the gate before any volume PO.

For buyers managing MRO welding consumable programs across multiple facilities, the additional complexity is lot synchronization — ensuring the qualification data you gathered in one plant applies to what arrives at another. It doesn’t always. Lot traceability documentation is the only control that works across distributed sites, and it needs to be a contractual requirement, not a request.

Is AWS certification from a Chinese supplier reliable?

AWS certification is a self-declaration for most A5-series classifications — there is no mandatory third-party audit required for the classification to be valid. Some Chinese suppliers have independent lab verification of their classifications; many do not. Requesting the test report behind the classification, not just the classification stamp, is the only way to verify it.

What’s the right AQL level for incoming welding electrode inspection?

For low-hydrogen grades in structural applications, we use AQL 1.0 per ISO 2859-1 as our standard incoming plan — stricter than the AQL 2.5 most MRO buyers default to. The rationale is that a single hydrogen-cracked weld in a restrained joint costs more to repair than the entire incoming inspection program.

How many consecutive COAs should I request before qualifying a supplier?

Six, covering at least 90 days of production. Single-point COAs are nearly useless for evaluating lot consistency. The number six isn’t arbitrary — it’s what our qualification protocol requires before a supplier moves from provisional to approved status on our internal AVL.

Do Chinese suppliers test to GB/T or AWS standards — and does it matter?

It matters more than most specification sheets suggest. GB/T 5117 allows slightly different impact testing conditions than AWS A5.1, and a supplier reporting “compliant to GB/T 5117” is not necessarily meeting the impact toughness requirements on your AWS-referenced drawing. Always specify the standard explicitly in the PO, and ask the supplier to confirm which test standard their COA data reflects — they are not always the same document.

What about suppliers who offer both AWS and EN ISO dual certification?

Dual certification is possible and legitimate, but the test conditions for EN ISO 2560 and AWS A5.1 differ in ways that affect Charpy impact test temperature and minimum energy requirements. A COA showing dual classification should cite test results under both sets of conditions — if it only shows one set of test data with two classification stamps applied, that’s a documentation shortcut that warrants a direct question to the supplier’s QA team.

Published by sinoraw.com Technical Team | Eng. Robert Chen, Metalworking and Fabrication Consumables Engineer | Request a sourcing consultation