Aluminium Scrap Feedstock: Classification, Contamination and Pre-treatment | 废铝炉料：分类、污染与预处理

Feedstock & Charge · Upstream of the melting furnace

1. Why an upstream page belongs on a dross site

The reason is a dependency, not a wish to broaden the site. Peer-reviewed work on aluminium slag processing states that the output of a recovery technology is determined primarily by the aluminium content of the initial material, the presence of pre-treatment operations, the type of main process, and resource costs. Reported aluminium recovery across published technologies spans 35–97%, with one study achieving 85% — a range so wide that quoting a technology’s headline rate without the feed behind it says almost nothing about a plant’s result.

So this page answers the question that comes before equipment selection: what, exactly, was charged? It links forward to the pages that answer what happens afterwards — the comparison of five dross treatment routes, the pressing and cooling equipment, and the hazardous-waste and residue routes.

2. Classification: scrap is a family, not one material

A formulation from an industry conference puts it plainly: scrap differs by source, by attached material, and by physical form; each combination forces a different question. Used beverage cans carry paint, ink and plastic; machining chips carry water, cutting oil and emulsion; architectural scrap carries connectors and mixed fittings; offcuts and closures raise particle size and iron contamination. Form matters as much as chemistry: bulky pieces, fine chips and different bulk densities change how a charge fills a furnace and how heat reaches it.

Chinese commodity research describes the intake sorting on three axes — alloy grade, impurity content and physical form — with optical emission or X-ray fluorescence spectrometry for grade identification and eddy-current separation for non-metallics. Cast aluminium alloy designations themselves follow a published national standard (GB/T 8063-2017), so grade is a defined field, not a trade description.

Stream · 料型	Typical attachments · 典型附着物	Question it forces · 必须回答的问题
Used beverage cans 废旧易拉罐	Paint, printing ink, plastic laminate and liner 油漆、油墨、塑料复合层	How much organic coating survives de-coating, and where the off-gas goes
Machining chips 机加工铝屑	Water, cutting oil, emulsion; high surface area 水分、切削油、乳化液；比表面积大	Moisture before it meets a molten bath; oil before it becomes off-gas
Architectural scrap 建筑废铝	Steel connectors, gaskets, glass, mixed profiles 钢连接件、密封件、玻璃、混合型材	Ferrous content and alloy mixing across batches
Offcuts and closures 边角料与瓶盖	Iron fines; wide particle-size spread 细铁屑；粒度分布宽	Particle size, bulk density and the iron limit of the target alloy
Coated or painted extrusions 带涂层型材	Organic coating on an aluminium substrate 铝基体表面的有机涂层	Whether thermal de-coating is complete, and what residue remains

The table is a classification, not a ranking. No stream here is “better” in general; each is better or worse for a stated furnace, alloy and permit.

3. What one tonne must answer before it reaches the furnace

The same industry source reduces the intake question to four items that must be known for every batch: moisture, organic matter, iron contamination, and batch-to-batch variation. It calls for a “quality passport” per batch — moisture content, organic residue, iron and particle size, plus batch identity and time — so that a later result can be traced back to the feed that produced it.

• Moisture. Water carried into a high-temperature melt is a safety issue before it is a quality issue. Drying to below 1% is described in commodity research as the intake condition before charging. This is the same water-reactivity family as the residue chemistry documented on this site: aluminium-bearing fines that meet water can generate hydrogen. Drying is therefore a control on a hazard, not merely a step for yield.
• Organic matter. Oil, paint and plastic become off-gas and a reducing atmosphere in the furnace. De-coating by thermal treatment in the 400–600 °C range, or by solvent routes, is described before melting rather than during it. The exhaust from that step is a separate permitted emission stream.
• Iron contamination. Magnetic separation removes ferrous pieces; iron that remains enters the alloy. One research note describes adding manganese or chromium to form high-melting compounds and hold melt iron below 1.2%. That figure is a reported process target, not a universal specification — the limit belongs to the alloy being made.
• Batch variation. The spread between batches often matters more than the average. A plant that records only a monthly mean cannot explain a bad week, because the feed identity has been averaged away.

4. Free metal content decides the denominator

This is the section that connects feedstock to the rest of the site. A peer-reviewed study on compaction and thermal pre-treatment prints its measurement definitions explicitly, and they are not interchangeable:

Measure · 指标	Definition · 定义	Denominator · 分母
Dross rate 铝渣产率	mass of dross / mass of scrap charged × 100	Scrap charged
Metal yield 金属收得率	mass of cast ingots / (mass of scrap charged + mass of heel) × 100	Charge plus heel
Dross metal content 渣中金属含量	metal recovered from dross / mass of dross charged to salt	Dross charged to salt
Total metal yield 总金属收得率	(charge − dross) + (dross metal content × dross), divided by charge	Original charge

The same study’s salt-remelting conditions are experimental, not a plant recipe: a 68.6 wt% NaCl / 29.4 wt% KCl / 2 wt% CaF₂ mixture, a salt-to-dross ratio of 2:1, and 800 °C in a resistance furnace. Reporting them as “the industry standard” would be a category error — they are the conditions under which those particular numbers were produced.

Two cautions follow, both documented in a technical review that cites ICSOBA 2023, Light Metal Age and a US DOE melting-efficiency reference. First, dross mass is not the same as metal lost: oxide films trap metal droplets, so weighing the skim tells you the weight of a mixture, not the metal inside it. Second, metal recovered from the metal contained in the feed is not the same as total mass yield from dross. An industry review quotes white or primary dross at 15–70% metallic aluminium and black or secondary dross around 12–18%, with dross generation commonly 15–30 kg per tonne melted and total melt loss cited within 1–10% of charge. These are ranges across different operations and test methods, not values for any single plant; this review is a vendor technical summary and the underlying primary texts were not opened here, so the figures are carried as ranges only.

5. Alloy family and geometry: a common claim that does not hold

A note often repeated in coverage of this topic says magnesium-rich 5xxx scrap produces dross with higher metal content. The primary literature does not support that wording, and the difference matters. An experimental study at the University of Michigan remelted clean sheet and machining-chip scrap of controlled thickness, roughness and composition, and found that magnesium-containing alloys (2xxx, 5xxx, 6xxx and 7xxx) accelerate dross generation and lower metal recovery; energy-dispersive X-ray spectroscopy found the magnesium content in the dross higher than in the parent alloy, consistent with magnesium’s higher reactivity. What is enriched in the dross is magnesium and its oxide — not a larger pool of recoverable metallic aluminium.

The same experiment isolates geometry as an independent variable. Recovery fell for thinner scrap, and was lower for machining chips than for sheet of the same thickness and composition, attributed to greater surface roughness increasing the area available for oxidation; smoothing the rough chip surface with a purpose-designed cutting tool increased recovery. A vendor technical review likewise cites casthouse guidance to keep preparation of high-magnesium alloys below 750 °C. So the accurate statement is: Mg-bearing scrap oxidises more readily, generates more dross and returns less metal; fine and rough feed does the same through surface area. The claim that its dross is richer in recoverable metal is recorded here as not supported.

This is also why a single flux recipe cannot be assumed across alloy families. A peer-reviewed study of multistage salt-slag recycling notes that a 5xxx alloy (EN AW-5754, approximately 2.60–3.60 wt% Mg) accumulates silicon and iron with each recycling loop, so later cycles require increasing additions of pure components. Recycling is not automatically a closed circle; feed classification is what decides how many times a given stream can go around.

6. Pre-treatment: what each step actually changes

Pre-treatment is best read as a set of variable changes, not as a list of machines. Each operation alters one measurable property of the charge and introduces one cost or trade-off.

Step · 工序	Variable changed · 改变的变量	Cost or trade-off · 代价与权衡
Sorting by grade 按牌号分拣	Alloy chemistry of the charge	Spectrometry per batch; mixed streams cannot be fully recovered by downstream refining
Size reduction 破碎	Particle size, bulk density, heat transfer	Finer material exposes more surface to oxidation; reported intake sizes of 10–50 mm are design targets, not optima
Magnetic separation 磁选除铁	Ferrous content	Multiple stages needed; fine iron embedded in chips is harder to remove
De-coating / de-oiling 脱漆脱脂	Organic load and off-gas	Thermal routes around 300–450 °C in one study and 400–600 °C in a research note; energy and a permitted exhaust stream
Drying 干燥	Moisture; hydrogen and splash hazard	Energy; overdrying fine chips raises dust
Compaction / briquetting 压块	Bulk density and exposed surface area	A real trade-off: compaction lowers oxidation area but also hinders de-coating — see below

That last row is the most useful finding on this page, because it runs against the usual “more preparation is always better” story. In the compaction study, loose chips lost 1.51% of their mass on thermal treatment, uniaxially compacted briquettes 1.43%, and torsion-compacted briquettes 1.32%; the compacted forms showed a lower degree of de-coating, explained by reduced surface area and internal porosity limiting gas transport and contact between oxygen and organics. The same study uses colour as a process indicator: a light brown/yellow residue indicates complete de-coating with oxide remaining, while darker colours reveal carbonaceous residue. Compaction and de-coating pull in opposite directions. Whether a plant should briquette depends on which loss dominates for its feed — oxidation of a large surface, or incomplete removal of a coating. There is no universal sequence.

7. The commercial reading

For a buyer or a plant manager, the point of the preceding six sections is not metallurgical elegance. It is that an unmeasured feed makes every downstream number unmanageable. A recovery rate, a salt consumption, a dross invoice and a product’s alloy certificate all trace back to what was charged; if the charge is described only by weight, none of them can be compared between two suppliers or two months.

One technical review expresses the economics as a chain rather than a headline rate: net recovered value per tonne of dross equals dross mass × metallic aluminium content × process recovery × product-acceptance factor − treatment cost. Feed condition touches at least three of those terms. A supplier quotation that quotes only “recovery” has left out the parts that determine whether the metal can actually be sold into the intended alloy.

Two consequences for a commercial reader:

• Pre-treatment is a measurement investment. Sorting, drying and assaying do not just clean the material; they produce the batch record that makes a later dispute resolvable. A plant cannot manage a number it has never recorded — the same point made on the casthouse equipment page about in-house metal losses.
• Feed is the recycler’s input, so it sets the business model. How a third-party recycler buys, grades and finances scrap is the subject of the commercial models page. Recovery technology begins where the purchase specification ends.

There is also a boundary to this optimism. The 5xxx example above shows impurity accumulation over successive loops, and the 35–97% spread in reported recovery shows how strongly the starting material governs the result. “Recycled” is a description of origin, not a guarantee of grade. What keeps a stream at grade is classification at intake, done every batch.

8. Boundaries and where this connects

This page is not a supplier recommendation, a purchase specification, or a treatment-route comparison. It does not quote a recovery rate without its basis, and it does not extend the site’s scope into melting practice. The pages that carry those subjects:

Question · 问题	Page · 页面
Which dross treatment routes exist, and how do their claimed rates differ?	Global Mainstream Aluminium Dross Technologies · Hot Dross Recovery
What equipment measures and cools hot dross at the furnace door?	Dross Presses, Coolers, Pans and Moulds · Pressing & Cooling
What happens to melt quality after the charge is molten?	Melt Degassing and CFF Filtration · Melt Refining & Filtration
How is the residue classified and treated?	Hazardous Waste Valorization · Classification Codes
How do regulations differ by region, and how is scrap distinguished from dross?	Regional Policy Differences
Where do the trade codes and the recycling business meet?	Markets & Resources · Recycler Business Models
What do the terms and the source rules mean?	Glossary · Editorial Standards

---

---

中文精炼版 · 废铝炉料：分类、污染与预处理

本站记录从熔炼炉到填埋场之间发生的一切，而废铝炉料在更前一格。本页写它，是因为一座炉子能回收的金属不会超过装进它的金属：不说明分母的回收率不是数量，分母由炉料的成分、洁净度与几何形态决定。本页把上游边界往前挪一格，不写熔炼操作，也不写熔炼后的残渣处理。可检验的主张很窄：废铝不是单一物料，而是一个物料家族；两批同样重量的料，会因附着物、水分、铁杂与颗粒形状不同，生成不同量的铝渣、不同的熔损与不同的合金成分。预处理就是把不确定的来料，变成行为可测量、可重复的入炉料——它首先是经济职能，其次才是环保职能。

1. 上游页为什么属于一个铝渣站。同行评议研究指出，回收技术的产出主要由初始原料的含铝量、有无预处理、主工艺类型与资源成本决定；已发表技术的铝回收率跨度达 35–97%，其中一项研究为 85%。区间宽到这种程度，只报技术名称下的”头条回收率”而不报炉料，几乎不说明任何一座工厂的结果。所以本页回答选设备之前的问题：究竟装进去了什么？另须守住两个词：废铝是买来、分拣、重熔的原料；铝渣是熔炼生成的废物，二者不共用贸易编码或废物编码，详见区域政策篇。

2. 分类：废铝是一个家族。行业会议的表述是，废铝按来源、附着物、物理形态三个维度不同：易拉罐带油漆油墨与塑料复合层，机加工铝屑带水、切削油与乳化液，建筑废铝带钢连接件与玻璃，边角料带细铁屑且粒度分布宽；形态与成分同等重要，大块、细屑与不同堆密度改变装炉与传热。入厂分拣按合金牌号、杂质含量、物理形态三轴进行，用直读光谱或 X 射线荧光识别牌号、涡流分选去除非金属；铸造铝合金牌号依 GB/T 8063-2017。

核心对照表：下表按料型归位；文字说不清的地方，用表格说清。

料型	典型附着物	必须回答的问题
废旧易拉罐	油漆、油墨、塑料复合层	脱涂层后残留多少有机物，尾气去哪里
机加工铝屑	水分、切削油、乳化液；比表面积大	水在遇到熔池之前如何处理，油在变成尾气之前如何处理
建筑废铝	钢连接件、密封件、玻璃、混合型材	铁含量与批次间的合金混杂
边角料与瓶盖	细铁屑；粒度分布宽	粒度、堆密度与目标合金的铁限值
带涂层型材	铝基体表面的有机涂层	热脱涂层是否彻底，残渣是什么

这是分类，不是排序。没有哪种料”更好”，只有在给定炉型、牌号与许可条件下更合适或更不合适。

3. 一吨料进炉前必须回答的四件事。水分、有机物、铁杂、批次差异；并应逐批建立”品质护照”：含水率、有机物残留、铁杂与粒度，加批次与时间，使日后的结果能追回产生它的那批料。水分首先是安全问题——水进高温熔池会引起喷溅，且含铝细料遇水可产氢，与本站残渣篇是同一类反应；研报给出的入炉条件是干燥至水分低于 1%。有机物在炉内变成尾气与还原气氛，脱漆按 400–600 ℃ 热处理或溶剂路线在熔炼前完成，其尾气是另一条需要许可的排放流。铁杂先磁选；一份研报记载加锰或铬形成高熔点化合物，把熔液铁控制在 1.2% 以下——这是该工艺的目标值，不是通用规格，限值属于所生产的牌号。批次差异往往比月均值更要紧：只记月均值的工厂解释不了某一个坏周，因为料已被平均掉了。

需要明说：流传很广的除铁率 ≥98%、脱油率 ≥95%、回收率由 90–92% 提升到 95–98%、能耗 80–100 降到 50–70 kWh/t、吨铝降本 150–250 元等数字，出自行业汇编与厂商材料，本站未能追到原始测量，故正文不把它们当作可用数值；凡本页给出阈值，都同时写明来源与”报告目标值”的身份。

4. 游离金属含量决定分母。这是本页与全站相接的一节。一项同行评议研究把测量定义印得清楚，且彼此不可互换。

指标	定义	分母
铝渣产率	铝渣质量 / 入炉废铝质量 × 100	入炉料
金属收得率	铸锭质量 /（入炉料 + 炉底余料）× 100	装料加炉底余料
渣中金属含量	从渣中回收的金属 / 加盐处理的渣质量	加盐处理的渣
总金属收得率	（装料 − 渣）+（渣中金属含量 × 渣），再除以装料	原始装料

该研究的加盐重熔条件是实验条件而非工厂配方：68.6% NaCl / 29.4% KCl / 2% CaF₂、盐渣比 2:1、电阻炉 800 ℃；写成”行业标准”就是范畴错误。另有两条须守住：渣重不等于损失的金属——氧化膜夹带铝滴，称的是混合物质量；从进料所含金属中回收的份额，不等于从渣中得到的总质量产率。一份厂商技术综述（转引 ICSOBA 2023、《Light Metal Age》与美国能源部熔炼效率资料）给出白渣金属铝 15–70%、黑渣约 12–18%、铝渣生成 15–30 kg/t、熔损 1–10% 的跨厂区间；本站未打开其所转引原文，故只作区间参照，不作定值。回接全站：五条路线篇核过”回收 98%”之说，其分母是渣中金属组分，不是渣总质量，也不是炉子装料；本页补的是这条链的前半段——装料决定生成多少渣、渣里含多少金属。没有分母的百分比不是数量，该约定维护在归类编码与计量约定页。

5. 合金族与几何形态：一个不成立的说法。常见说法是”5xxx 高镁废铝产生的渣金属含量更高“。一手文献不支持这个措辞，而且差别要紧。密歇根大学的实验重熔了厚度、粗糙度与成分受控的板材与铝屑，发现含镁合金（2xxx、5xxx、6xxx、7xxx）加速铝渣生成并降低金属回收率；能谱显示渣中镁含量高于母合金，与镁更活泼一致。渣里富集的是镁及其氧化物，不是更多可回收的金属铝。同一实验把几何形态独立出来：薄料回收率更低；同厚同成分的铝屑低于板材，归因于表面粗糙度增大氧化面积；用专用刀具把铝屑背面压平后，回收率上升。厂商综述转引的车间指导是高镁合金制备温度低于 750 ℃。所以准确表述是：含镁废料更易氧化、渣量更大、回收金属更少；细料与粗糙表面通过比表面积产生同样效果。“其渣更富可回收金属”在本页记为不成立。另有一层：同行评议研究指出 5xxx（EN AW-5754，镁约 2.60–3.60%）每循环一次硅、铁杂质上升，后续循环需要越来越多纯料——回收不自动等于闭环，能循环几圈由入厂分类决定。

6. 预处理：每一步究竟改变了什么。把预处理读成一组”变量改变”，而不是一串设备名：分拣改变装料成分，代价是逐批光谱检测；破碎改变粒度、堆密度与传热，代价是细料氧化面积变大（研报的 10–50 mm 是设计目标，不是最优值）；磁选改变铁含量，细屑中嵌着的铁更难去；脱漆脱脂改变有机负荷与尾气，热处理温度一项研究为 300–450 ℃、一份研报为 400–600 ℃，代价是能耗与一条许可排放流；干燥改变水分与产氢／喷溅风险；压块改变堆密度与暴露面积。最后一行最有用，因为它反着”准备越充分越好”的常识：压实研究中，松散铝屑热处理失重 1.51%、单向压块 1.43%、扭压块 1.32%，压块形式的脱涂层程度更低，原因是表面积与内部孔隙率下降，限制了气体传输以及氧气与有机物的接触；该研究还用颜色作判据——浅棕／黄表示脱涂层完全、留下氧化物，深色表示仍有含碳残留。压实与脱涂层方向相反：该不该压块，取决于这批料的主导损失是”大面积氧化”还是”涂层没除净”，没有通用顺序。

7. 商业读法。对采购与工厂管理者，前六节的意义不是冶金上的漂亮，而是未被计量的来料会让下游每个数字都不可管理：回收率、盐耗、铝渣处置发票、产品牌号证书，全部追到装进去的东西；若装料只用重量描述，两家供应商之间、两个月之间就无从比较。一份技术综述把经济写成链条而非头条率——每吨渣的净回收价值 = 渣量 × 金属铝含量 × 工艺回收率 × 产品可接受系数 − 处理成本；炉料状况至少影响其中三项。只报”回收率”的报价，恰好漏掉了决定金属能否按目标牌号售出的部分。两条推论：预处理是计量投资——分拣、干燥、化验不只清洁物料，还生成了日后可据以解决争议的批次记录；炉料是回收商的投入端，它决定商业模式，见第三方回收商商业模式篇。也要给乐观划界：5xxx 的杂质累积 与 35–97% 的回收率跨度 都说明“再生”描述的是来源，不是牌号保证；让一批料始终留在牌号内的，是每批入厂都做的分类。

8. 边界与衔接。本页不是供应商推荐，不是采购规格，也不是处理路线对比；不引用没有基数的回收率，不把本站范围扩进熔炼操作。路线对比见五条路线篇与热渣现场回收页；炉口装备见压榨机与工装篇、压榨与冷却页；熔后质量见熔体精炼与过滤页；残渣归类见危废无害化与资源化页、归类编码页；区域法规与”废铝／铝渣”之辨见区域政策篇；贸易编码与商业模式见市场与上游资源页、回收商商业模式篇；术语与来源规则见术语页与编辑标准。

资料来源与出处。引用是”指称”而非翻译：叙述是元语言，被引文件语言是对象语言，中文法定题名逐字保留并标注语言，英文表述一律标为本站释义，不冒充官方英文题名。同行评议文献、官方标准与专业研究给出全称；厂商材料具名不链接；未能回溯至原始测量的数字据实写明缺口，不以估计填补。《Recycling》（MDPI）——产出由初始原料决定、85% 与 35–97% 区间、5xxx 循环杂质累积；《Journal of Sustainable Metallurgy》（Springer，2023）——四个测量定义与分母、盐配比与实验条件、压实降低脱涂层程度、残渣颜色判据；ASME MSEC 2022（密歇根大学）——含镁合金加速渣化并降低回收率、渣中镁富集、薄料与粗糙表面回收率更低；2026 SMM 再生金属产业高峰论坛演讲 《重新定义熔炉入口——从不确定的废料，到可信任的入炉料》（对象语言，未翻译；本站释义：”重新定义熔炉入口”，非官方英文题名）——废铝按三维度分族、入炉四问、逐批品质护照；新湖期货研究所 《新湖有色专题（铝合金）：铸造铝合金期货系列报告》，2025-05-21（对象语言，未翻译；本站释义：”铸造铝合金期货系列报告”，非官方英文题名），经新浪财经转载——分拣三轴、破碎粒度、脱漆温度、干燥、除铁目标、GB/T 8063-2017；Continental Furnaces 技术综述（转引 ICSOBA 2023、《Light Metal Age》与美国能源部资料）——渣重不等于金属损失、两个分母、白渣／黑渣区间、净回收价值链，仅作区间参照。未能核实：除铁率 ≥98%、脱油率 ≥95%、回收率 90–92% → 95–98%、能耗 80–100 → 50–70 kWh/t、降本 150–250 元/吨、回收期 1.5–2.5 年，压块使烧损 20% → 5%、双室炉 8% → 3%，以及”含镁渣金属更多”之说（已被 ASME 实验否定）与逐批品质护照的实际执行比例——均不以数字填补缺口。