Aluminium Dross Recovery: Five Routes and What Separates Them | 世界主流铝渣技术路线对比

Hot Dross Recovery · Published 2026-10-30

1. Why this comparison is harder than it looks

A dross plant manager comparing two vendor quotations is usually comparing numbers that were never measured the same way. Three structural reasons:

• Dross is not a specification material. Metal content in freshly skimmed dross spans a very wide range — one industrial gas supplier states 18% to 80% metallic aluminium depending on skimming practice and alloy [citation:24]. A route tested on 60%-metal dross and a route tested on 20%-metal dross are not comparable.
• Recovery has no default denominator. It can mean metal recovered as a share of the metal contained in the dross, or as a share of total charge mass. These differ by a factor of two or more on the same furnace.
• The residue is part of the product. A salt route and a salt-free route may recover similar metal and still differ decisively, because one produces a leachable hazardous salt cake and the other a saleable alumina-rich non-metallic product (NMP).

Everything below is therefore quoted with its stated basis, or flagged where the source does not state one.

2. The five routes, by mechanism

Route	Separation mechanism	Heat source	Salt	Typical feed state
Pressing	Static or dynamic pressure extrudes molten metal through a constriction; oxide shell fractures and metal drains	None — uses the dross’s own heat; cooling water in the ram	None	Hot, as skimmed
MRM / stirring	Mechanical stirring ruptures the oxide film; exothermic flux releases heat and maintains temperature; metal collects at the vessel bottom	Exothermic flux reaction	Flux added	Hot
IGDC	None during cooling. Inert gas stops oxidation so unreacted metal survives to later recovery	None — controlled cooling	None	Any temperature
Rotary salt furnace (RSF)	Molten salt destroys the alumina film, promotes droplet coalescence, and prevents re-oxidation	Gas or oil burner	Substantial	Cold or hot
Salt-free thermal (ALUREC / plasma / DROSCAR / DROSRITE / ECOCENT)	Heating above the aluminium melting point under a controlled atmosphere; rotation or tumbling ruptures the oxide film	Oxy-fuel, plasma torch, DC graphite arc, or the charge’s own exothermic reaction	None	Hot or cold

3. The denominator problem, demonstrated on one process

Figure	Source and character	Denominator	What it actually says
98%	Vendor statement and SEC filing — “recover 98% of the aluminium that was contained within the dross” [citation:36][citation:42][citation:46]	Metal contained in the dross	Of the metal fraction present, almost all is recovered. Says nothing about how much metal was there.
≈90%	Peer-reviewed review (Materials, 2024), tabulating salt-free processes [citation:11]	Not stated in the table	A summary figure; basis must be read from the cited primary work.
74%	Measured trial — TMS 2004 paper, white dross: a first tapping yielded aluminium equal to 74% of the initial charge; a second tapping 14 minutes later recovered nothing further [citation:40]	Total charge mass	An actual weighbridge result, not a model or a claim.
18–40%	Measured trial — same TMS 2004 campaign, seven consecutive runs on black dross: metal recovery varied between 18% and 40% [citation:40][citation:14]	Total charge mass	The same furnace, on lower-metal feed, returns roughly half of what it returns on white dross.

The reconciliation. Take a dross at the composition the vendor itself cites — roughly 60% metal, 40% residue [citation:42]. In a 100 kg charge containing 60 kg of metal:

• Recovering 98% of the contained metal means recovering 58.8 kg.
• Expressed against total charge mass, that same 58.8 kg is 58.8% — not 98%.
• The measured white-dross figure of 74% of charge implies more metal left the furnace than a 60%-metal feed contained, which tells us the charge was not pure dross or contained already-molten metal. That discrepancy is the point: without a stated charge composition, no recovery figure can be transferred between plants.
• Black dross at 18–40% of charge reflects a genuinely lower metal content in the feed, not a worse furnace.

The same pattern appears in the other routes

Route	Figure as published	Stated or apparent basis	Contrasting figure from another source
Pressing	≈60% (review) [citation:11]; total aluminium recovery 62.5% (Altek “The Press”) [citation:19]	Charge basis; consistent between two independent sources	The most stable figure in the table — pressing has no thermal route to re-oxidise metal
MRM / stirring	≈90%, burning loss 4%; modified MRM under argon: 91%, burning loss 4% [citation:11][citation:37]	Not stated in the review table	A Japanese operator reports MRM accounts for no less than 60% of metal recovery across the whole UBC process [citation:43] — a different denominator entirely
IGDC	>80% of the unreacted aluminium metal recovered in trials (Alcan patent) [citation:12]	Unreacted metal present, explicitly	Practitioner account: in-house recovery 0–12%, secondary recovery 30–60% [citation:16] — charge-mass basis
Rotary salt furnace	≈80% (review) [citation:11]	Not stated	Salt slag tapped at 8.5% entrained metal (fixed-axis) or 12.5% (tilting) in the European mass-balance model [citation:21]
ALUREC	≈70% (review) [citation:11]	Not stated	—
Plasma (Alcan)	≈90% (review) [citation:11]	Not stated	Reported as higher than rotary kiln in a trade-journal comparison [citation:47]
DROSCAR (DC arc)	≈75% (review) [citation:11]	Not stated	70–80% across parameter variations; refractory crushing bodies raise recovery by ~10 percentage points [citation:14]
ECOCENT (centrifuge)	≈85% (review) [citation:11]	Not stated	—

Where a source does not state a basis, this site does not invent one. Rows marked “not stated” should be read as untransferable, not as comparable.

4. Salt: the second misused metric

Salt factor is defined as the mass of added salt divided by the mass of oxides in the charge — not per tonne of metal. This definition is the reason the published spread looks absurd until it is read correctly [citation:25][citation:21].

Furnace	Salt factor	Resulting salt slag	Source
Fixed-axis rotary furnace	0.9 – 2.0 (1.8 common)	≈50% salt content	Review [citation:17]; European mass-balance model uses 1.3 [citation:21]
Tilting rotary furnace	0.4 – 0.8	≈27% salt content	Review [citation:17]; model uses 0.4 [citation:21]
Dross treated in a rotary furnace	Salt added up to 50% of dross weight	—	Review [citation:17]

Downstream consequences of the salt

• Generation rate. Salt slag arising ranges from 200 to 500 kg per tonne of secondary aluminium depending on scrap mix [citation:13].
• Hazard classification. Salt slag is classified as toxic and hazardous waste — highly flammable (H3-A: evolves flammable gases in contact with water or damp air) and irritant (H4) under the European Catalogue for Hazardous Wastes. Landfill is forbidden in most European countries [citation:13].
• The residue carries metal away. Salt slag as tapped still contains entrained metal — 8.5% for fixed-axis, 12.5% for tilting in the European model [citation:21]. Metal lost here is metal the furnace never recovered, whatever the recovery figure says.
• Salt does not disappear. Evaporated salt of about 4.5 kg per tonne of charge leaves with the off-gas [citation:21].

A common salt flux composition cited for German rotary kilns is approximately 70% NaCl, 28% KCl, 2% CaF₂ [citation:47].

5. Energy: the third misused metric

The most useful published range for heating cold dross in a rotary furnace is 375 to 2,500 kWh per tonne of dross of external energy input [citation:45]. That 6.7-fold spread is not imprecision — it is the difference between processes, feed states and furnace designs. Compressing it into a “typical value” destroys the information.

Route	Energy character	Note
Rotary furnace, cold dross	375 – 2,500 kWh / t dross external input	The reference band [citation:45]
DROSRITE	No external energy input. Energy is extracted from the residue, stored in the refractory wall, released to the next charge	Vendor states savings can exceed 2,500 kWh/t versus conventional [citation:45]. Operates on the exothermic heat of its own residue and a controlled oxygen injection.
ECOCENT	Uses the inherent heat of hot dross; no additional energy input	Reported at less than 50% of conventional energy consumption [citation:38][citation:22]
Pressing	No furnace. Ram cooling water removes heat	Cited operating detail: dross falls from above 800 °C to below 450 °C, which is what stops oxidation [citation:19][citation:23]
IGDC	No heating; inert gas only	Argon flow cited in the patent at 0.5–1 scfm steady state [citation:12]

One independent estimate for white dross recycling gives roughly 72 kWh electricity plus 454 kWh thermal per tonne of white dross, yielding about 0.45 tonnes of metallic aluminium [citation:41]. Because the output is stated, this figure can be converted — which is more than can be said for most published numbers.

6. What the residue becomes

This is where routes actually diverge, and it matters more than recovery in most jurisdictions.

Route family	Residue	Hazardous?	Known outlets
Salt routes	Salt cake / salt slag — salts, alumina, entrained metal	Yes — toxic and hazardous, leachable, H3-A flammable in contact with water	Salt recovery by aqueous leaching; the insoluble black mud (≈60–70% Al2O3) goes to refractory or cement uses [citation:13][citation:25]
Salt-free thermal	NMP — non-metallic product, mostly alumina with variable AlN and MgO	Residual Al and AlN remain reactive	NOVAL (Alcan plasma product) has an energy content on the order of 4,500 kJ/kg, suited to high-temperature applications [citation:44]; SEROX (ALSA) is the water-leached counterpart, lower energy content [citation:44]; calcium aluminate for steel ladle metallurgy [citation:44]
Pressing	Pressed dross cake, salt-free	Lower reactivity than salt cake	Further processing or sale; no salt cake treatment step required
IGDC	Cooled dross, low fines fraction	—	Conventional secondary recovery; low fines makes recovery “technically simple and economic” [citation:12]

Why AlN dominates residue management. Aluminium nitride hydrolyses on contact with water, releasing ammonia — the reaction set recorded for dross dust fractions includes AlN + 3H₂O → Al(OH)₃ + NH₃ [citation:25]. Gas from leaching such fractions is reported at 85–95% H₂, 4–10% CH₄, up to 1% NH₃ and is described as highly explosive [citation:25]. This is a property of the material, not of any one route — but salt-free routes concentrate it, because they do not dilute the residue with salt.

7. Selecting a route: what actually decides it

Order	Question	Why it comes first
1	What is the metal content of your dross, measured?	It spans 18–80% across the industry [citation:24]. Every performance claim is conditional on it. Measuring it is cheaper than any equipment decision.
2	Hot or cold at the point of processing?	Pressing and MRM need hot dross; IGDC accepts any temperature; rotary routes accept both but pay 375–2,500 kWh/t to heat cold feed [citation:45]. Moving from cold to hot processing is often a larger gain than changing equipment.
3	Can you lawfully handle the residue?	Salt cake landfill is forbidden in most of Europe [citation:13]. If you cannot place salt cake with a licensed processor, a salt route is not available to you regardless of its recovery.
4	Is there a salt-cake processor within economic distance?	Processing capacity and toll fees are regional. This is frequently the binding constraint.
5	Alloy and magnesium content?	NMP composition varies with alloy — alumina, spinel (MgAl2O4), and minor oxides [citation:44]. High-Mg feeds produce a different residue with different outlets.
6	Only then: capital and operating cost.	Coming last is deliberate. The two salt-free electric routes — plasma and DROSCAR — were reported as technically salt-free but burdened by high CAPEX and OPEX [citation:46].

Two claims this site does not repeat

• “Simply remove the salt from an RSF.” A processor-side account states that while metal can be recovered from an RSF with zero salt flux, recovery rates become extremely low, and the residue remains hazardous — containing significant unrecovered metallic aluminium and high nitrates from prolonged high-temperature exposure to an uncontrolled furnace atmosphere. Removing the flux does not convert an RSF into a salt-free route [citation:46].
• “A higher headline recovery means a better plant.” Given the denominator evidence in §3, a 98% figure and a 40% figure can describe the same furnace on different feed. The comparison cannot be made from the headline.

8. Scale, for context

Quantity	Value	Source
Global dross generation	≈3 million tonnes / year, containing up to 80% aluminium	PyroGenesis SEC filing, citing an AlCircle sector report [citation:42]
Typical dross composition	≈60% metal / 40% residue	Same filing [citation:42]
Black dross composition	Metal 10–20%; salt-flux mixture 40–55%; balance alumina	Review [citation:13]
Salt slag per tonne of secondary aluminium	200 – 500 kg	Review [citation:13]
Tolling fee for dross treatment	Reported range US$250 – 450 per tonne treated	Vendor estimate from customer discussions [citation:42]

9. What this site could not verify

• No independent, plant-scale, side-by-side trial of all five routes on a single characterised dross was found. Every cross-route figure in §3 comes from a different campaign, on a different feed, at a different scale. The table is therefore a map of what has been reported, not a league table.
• Recovery bases are unstated in the most-cited review tables. We mark them unstated rather than inferring a basis that would make the numbers comparable.
• Vendor figures are labelled as vendor figures. The 98% recovery and the 2,500 kWh/t saving are both vendor-sourced and are presented next to the measured TMS trial results rather than instead of them.
• Current installed counts per route were not found from a primary source and are not stated.

10. What this page is not

• Not a supplier ranking. No route is recommended. See Mission & Independence.
• Not design data. Parameters quoted from sources are reported values, not specification or purchase data. See Terms of Use §6.
• Not an endorsement of any named process. Named processes appear because they are the ones with published, traceable data — including data that does not flatter them.

Related: Hot Dross Recovery · Pressing & Cooling · Hazardous Waste Valorization · Classification Codes · Editorial Standards

---

中文摘要 · 世界主流铝渣技术路线对比

铝渣不是一种物料，也不存在唯一最优路线。工业实践中五类工艺占据主导——压渣、机械搅拌（MRM/炒灰）、惰性气体冷却（IGDC）、加盐回转炉，以及无盐热处理路线（ALUREC、等离子、DROSCAR、DROSRITE、ECOCENT）。区分它们的不是那个头条回收率数字，而是残渣变成了什么，以及那个回收率用的是什么分母。

1. 为什么这个对比比看上去难。三个结构性原因：铝渣不是规格物料——刚扒出的渣金属含量跨度极大，一家工业气体供应商给出 18%–80% 的区间，取决于扒渣操作与合金；在 60% 金属含量的渣上做的试验，与在 20% 上做的不可比。回收率没有默认分母——它可以指”回收金属占渣中所含金属的比例”，也可以指”占总炉料质量的比例”，同一座炉子上两者相差一倍以上。残渣也是产品的一部分——加盐路线与无盐路线可能回收相近的金属，但一个产出可浸出、属危废的盐饼，另一个产出可销售的富氧化铝非金属产物（NMP）。因此下文每个数字都连同其声明的基准一起引用；来源未声明基准的，本站标注”未声明”而非替它补一个。

2. 五条路线的机理。压渣：静压或动压把熔融金属从收缩口挤出，氧化壳破裂、金属排出，无热源、利用渣自身热量，压头通冷却水，不加盐，需热态进料。MRM/炒灰：机械搅拌破坏氧化膜，放热熔剂放热并维持温度，金属沉于容器底部，加熔剂，热态进料。IGDC：冷却过程中不做分离，惰性气体切断氧气使未反应金属存活至后续回收，无热源、为受控冷却，不加盐，任何温度均可。加盐回转炉（RSF）：熔盐破坏氧化铝膜、促进液滴聚并、阻止再氧化，燃气或燃油烧嘴，大量加盐，冷态热态均可。无盐热处理：在受控气氛下加热至铝熔点以上，靠旋转或翻滚破坏氧化膜；热源为富氧燃烧、等离子炬、直流石墨电弧，或炉料自身放热反应；不加盐，冷态热态均可。

3. 分母问题——用一个工艺来证明。DROSRITE 在文献中出现过 98%、约 90%、74%、18–40% 四个数字。四个都是真实的，彼此并不矛盾。98%：厂商表述及 SEC 备案文件——”回收渣中所含铝的 98%”，分母为渣中所含金属量，它说明金属组分几乎全被回收，但没有说明渣里本来有多少金属。约 90%：同行评议综述（Materials, 2024）在无盐工艺表中的汇总值，未声明基准。74%：实测——TMS 2004 论文，白渣首次出铝量相当于初始炉料的 74%，14 分钟后第二次出铝再无回收，分母为炉料总质量，是地磅实测结果而非模型或宣称。18–40%：同一次 TMS 2004 试验，七次连续处理黑渣，金属回收率在 18%–40% 之间，分母同为炉料总质量。如何调和对账：取厂商自身引用的典型成分——约 60% 金属、40% 残渣——100 kg 炉料含 60 kg 金属时，回收所含金属的 98% 即回收 58.8 kg；若改以炉料总质量表达，这 58.8 kg 是 58.8% 而不是 98%。而白渣 74% 的实测值意味着出炉金属量超过了 60% 金属含量炉料所含的金属——这说明该炉料并非纯渣，或含有已熔金属。这个偏差本身就是要点：没有声明炉料成分，任何回收率都无法在工厂之间移植。黑渣 18–40% 反映的是炉料本身金属含量更低，不是炉子更差。可移植的事实不是”哪条路线回收最多”，而是：同一座炉子，一种料报 74%，另一种料报 18–40%。炉料性质对选型的影响，大于设备选择的影响。

其他路线同样存在这个模式。压渣：综述约 60%，Altek “The Press” 法铝总回收率 62.5%，两者基准一致，是表中最稳定的数字——压渣没有热工路径导致金属再氧化。MRM：综述约 90%、烧损 4%；改良 MRM 在氩气保护下达 91%、烧损 4%；而日本运营方称 MRM 贡献了整个 UBC 流程不少于 60% 的金属回收量——完全是另一个分母。IGDC：Alcan 专利称试验中回收了未反应金属铝的 80% 以上，明确以”未反应金属”为分母；而从业者记录为厂内回收 0–12%、二次回收 30–60%，属炉料质量基准。加盐回转炉：综述约 80%；欧洲物质流模型给出出炉盐渣夹带金属 固定轴 8.5%、倾动式 12.5%。ALUREC 约 70%；等离子（Alcan）约 90%；DROSCAR（直流电弧）约 75%，参数变动范围内 70–80%，且加入耐火破碎体可使回收率提高约 10 个百分点；ECOCENT（离心）约 85%。来源未声明基准的，本站不替它编一个——标为”未声明”的行应读作”不可移植”，而不是”可比较”。

4. 盐耗：第二个被误用的指标。盐因子（salt factor）的定义是”加入的盐量 ÷ 炉料中氧化物的量”，不是”每吨金属用多少盐”。这正是公布值看起来离谱、读对之后才合理的原因。固定轴回转炉：盐因子 0.9–2.0（常见 1.8），产出盐渣含盐约 50%；欧洲物质流模型取 1.3。倾动式回转炉：盐因子 0.4–0.8，盐渣含盐约 27%；模型取 0.4。回转炉处理铝渣时加盐可达渣重的 50%。“回转炉”这一个词覆盖了盐耗相差二至四倍的两种机器——不指明固定轴还是倾动式的对比，不是对比。下游后果：产生率为每吨再生铝 200–500 kg 盐渣，随废铝料配比而变；危废归类上，盐渣被列为有毒有害废物，遇水或潮湿空气释放易燃气体（H3-A）并具刺激性（H4），欧盟多数国家禁止填埋；残渣带走金属——出炉盐渣仍夹带金属，模型取固定轴 8.5%、倾动式 12.5%，这部分金属是炉子从未回收的，与回收率数字无关；盐不会消失——约 4.5 kg/吨炉料 的盐随烟气蒸发。德国回转窑常用盐熔剂成分约为 70% NaCl、28% KCl、2% CaF₂。

5. 能耗：第三个被误用的指标。回转炉加热冷渣的外部能源输入，最有用的公布区间是 375–2,500 kWh/吨渣。这个 6.7 倍的跨度不是精度不足，而是工艺、进料状态与炉型设计的差异——把它压缩成一个”典型值”等于销毁信息。DROSRITE：无外部能源输入，能量取自残渣、储存在耐火墙中、释放给下一炉料，厂商称较常规工艺节能可超过 2,500 kWh/t，靠自身残渣放热与可控注氧运行。ECOCENT：利用热渣自身显热，无额外能源输入，报道为常规技术能耗的不到 50%。压渣：无炉子，压头冷却水带走热量；被引用的操作细节是渣温自 800 ℃ 以上降至 450 ℃ 以下，这正是终止氧化的机制。IGDC：不加热，仅通惰性气体；专利给出稳态氩气流量 0.5–1 scfm。另有一项独立估算：回收白渣约需 72 kWh 电 + 454 kWh 热能/吨白渣，产出约 0.45 吨金属铝——因为给出了产出量，这个数字可以换算，而多数公布数字做不到。

6. 残渣变成了什么——这才是路线真正分道扬镳之处，且在多数司法辖区比回收率更要紧。加盐路线产出盐饼/盐渣（盐类、氧化铝、夹带金属），属危废：有毒有害、可浸出、遇水释放易燃气体；出路为水浸回收盐类，不溶黑泥（约 60–70% Al₂O₃）用于耐火材料或水泥。无盐热处理路线产出 NMP（非金属产物，主要为氧化铝，含不定量 AlN 与 MgO），残余铝与氮化铝仍具反应性；NOVAL（Alcan 等离子产物）能量含量约 4,500 kJ/kg，适合高温应用；SEROX（ALSA）为其水浸对应物，能量含量较低；亦可制取铝酸钙用于钢铁炉外精炼。压渣产出无盐压渣饼，反应性低于盐饼，无需盐饼处理环节。IGDC 产出冷却渣、细粉比例低，专利称因此后续回收”在技术上简单且经济”。为何 AlN 主导残渣管理：氮化铝遇水水解释放氨——铝渣粉尘馏分记录的反应集包括 AlN + 3H₂O → Al(OH)₃ + NH₃；浸出此类馏分产生的气体报道为 85–95% H₂、4–10% CH₄、最高 1% NH₃，被描述为高爆炸性。这是物料属性而非某条路线的属性——但无盐路线会浓缩它，因为它们不用盐稀释残渣。

7. 如何选型：真正起决定作用的顺序。本站不给路线排名，也不推荐供应商。下列是决定适配性的提问顺序，而非裁决。第一，你的渣金属含量实测值是多少？全行业跨 18–80%，任何性能宣称都以它为条件；测出它比任何设备决策都便宜。第二，在处理点是热态还是冷态？压渣与 MRM 需要热渣；IGDC 任何温度都可；回转路线两者皆可，但加热冷料要付 375–2,500 kWh/t；从冷态处理改为热态处理，收益往往大于更换设备。第三，你是否能合法处置残渣？盐饼填埋在欧盟多数国家被禁；若无法把盐饼交给有资质的处理商，加盐路线对你而言根本不可用，与它的回收率无关。第四，经济距离内是否有盐饼处理商？处理能力与加工费是区域性的，这常常是真正的约束条件。第五，合金与镁含量如何？NMP 成分随合金而变——氧化铝、尖晶石（MgAl₂O₄）及少量其他氧化物；高镁炉料产出不同残渣、有不同出路。第六，然后才是投资与运行成本——有意放在最后：两条无盐电路线（等离子与 DROSCAR）被报道技术上无盐，但受制于高投资与高运行成本。两条本站不转述的说法：其一，”把 RSF 的盐去掉就行”——处理方记录指出，RSF 在零盐熔剂下虽能回收金属，但回收率极低，且残渣仍属危废（含大量未回收金属铝，以及因长时间高温暴露于不受控炉气而产生的高浓度硝酸盐）；去掉熔剂并不能把 RSF 变成无盐路线。其二，”头条回收率越高工厂越好”——按第 3 节的分母证据，98% 与 40% 可以是同一座炉子在不同炉料下的结果；仅凭头条数字无法比较。

8. 规模参照。全球铝渣产生量约 300 万吨/年，含铝量最高可达 80%；典型铝渣成分约 60% 金属 / 40% 残渣；黑渣成分为金属 10–20%、盐熔剂混合物 40–55%、余为氧化铝；每吨再生铝产生 200–500 kg 盐渣；铝渣处理加工费报道区间为 每吨 250–450 美元。

9. 本站未能核实的部分。没有找到在同一批已表征铝渣上对五条路线做的独立、工厂规模、并排试验；第 3 节的每个跨路线数字都来自不同试验、不同炉料、不同规模——该表是”已报道内容的地图”，不是排行榜。最常被引用的综述表中回收率基准未声明，本站标注”未声明”而非推断一个能让数字变得可比的基准。厂商数字已标明为厂商数字——98% 回收率与 2,500 kWh/t 节能均出自厂商，与实测的 TMS 试验结果并列呈现，而不是取而代之。各条路线的当前装机数量未找到一手来源，故不予陈述。

10. 本页不是什么。不是供应商排名——不推荐任何路线，见《使命与独立性》。不是设计数据——引自来源的参数是报道值，不是选型或采购数据，见《使用条款》第 6 节。不是对任何具名工艺的背书——具名工艺之所以出现，是因为它们有可追溯的公开数据，包括那些对它们不利的数据。