Summary of Findings
Flame at a furnace top is not by itself a fault. On an open furnace, carbon monoxide rising through the charge burns with induced air between the charge surface and the hood, and virtually all of the CO and much of the organic matter burns there (U.S. EPA, AP-42 §12.4). A furnace of this class that shows no combustion at the top is the unusual case, not the reverse.
What matters is a flare that departs from the furnace’s own baseline: localised at the electrode rather than distributed over the charge, sustained rather than momentary, or newly present on a furnace class where undercover combustion should be minimal. Three routes produce that, and they leave different evidence: gas finding a path through the column or the casing, devolatilisation running later and higher than intended, and a change in the combustion conditions at the top. Elevated tar in the off-gas is read alongside the flare because it discriminates between the second route and the other two.
Key Findings
- Establish the furnace class before calling a flare abnormal. Open furnaces burn CO with induced air at the top by design; as enclosure increases, that combustion is suppressed and CO concentration in the furnace gas rises (EPA, AP-42 §12.4). The same visual signal means different things on the three classes.
- Three routes, distinguishable by evidence. A gas path through column or casing is localised and often steady; delayed devolatilisation raises tar and shifts where gas evolves; a change in top-side combustion conditions affects the whole furnace top rather than one electrode.
- A momentary flare at fracture is a different event. Flare-up accompanies hard breakage at the instant of fracture and is diagnosed with the breakage, not as a gas problem.
- Casing integrity belongs in the assessment. A hole in the electrode casing combined with unbaked paste can allow paste to flow out into the furnace room (Larsen, Platinum 2012), so a gas path and a paste path are the same defect at different severities.
- Cracks can predate the symptom. Where high slipping rates, high electrode current and increased electrode movement occurred together, cracks were generated before the electrode was fully baked and cured (Wilkinson et al., Infacon IX) — so the operating record of preceding days is part of the evidence.
Contents
- When is a flare abnormal?
- Route one: a gas path through the column or the casing
- Route two: devolatilisation running late
- Route three: a change in combustion conditions at the top
- What evidence should be collected?
- How is the cause attributed?
- Frequently asked questions
- Related reading
1. When Is a Flare Abnormal?
The first question is which furnace class the observation was made on, because the baseline differs across the three.
| Furnace class | Normal top-side condition | What makes a flare abnormal |
|---|---|---|
| Open | CO rising through the charge burns with induced air between the charge surface and the hood; virtually all CO and much organic matter burns at the furnace top | Combustion localised at one electrode rather than distributed; sustained jetting from a fixed point; a change from the furnace’s own baseline |
| Covered / semi-enclosed | Reduced air infiltration, therefore reduced combustion and a smaller gas volume to collect | Visible flame at all where the cover is intended to suppress it; combustion at a cover penetration |
| Sealed | Little if any air infiltration or undercover combustion | Any sustained combustion at the electrode seal region |
Table 1. Furnace class definitions and gas behaviour after EPA, AP-42 §12.4. The same visual observation carries a different meaning on each class, which is why class is established before the observation is classified as a fault.
Three further distinctions prevent the most common misreadings.
Momentary against sustained. A flare at the instant of an electrode fracture is part of the fracture event — audible crack, visible flare-up and instantaneous current interruption travel together, and that combination belongs to the breakage diagnostic rather than to a gas investigation.
Localised against distributed. Combustion spread across the charge surface on an open furnace is the design condition. A jet issuing from one point on one electrode is not.
New against baseline. The useful comparison is against what that furnace normally does, recorded at fixed points and times. Absolute descriptions travel badly between plants; a change against the furnace’s own history does not.
2. Route One: A Gas Path Through the Column or the Casing
Gas evolved inside the column has to travel upward through material that is still permeable. Where a continuous path opens to the outside — through a crack in the baked section, along a poorly bonded interface, or through a defect in the casing — the gas takes it, and combustion occurs where that path discharges.
The signature is position. A route-one flare issues from a fixed point, tends to persist rather than fluctuate with charging, and moves down the casing as the column advances, since the defect travels with the material. Where the discharge point is at or near a fin or a weld, the casing is the first thing to examine.
Two connections make this route more consequential than a gas leak alone.
A gas path and a paste path are the same defect at different severities. A hole in the electrode casing combined with unbaked electrode paste can cause unbaked paste to flow out into the furnace room (Larsen, Platinum 2012). A casing defect that currently vents gas can vent paste if the material reaching it is not yet baked, which is why the position of the baking zone is part of this assessment rather than a separate question.
The crack may predate the flare by days. Where high slipping rates, high electrode current and increased electrode movement occurred together, cracks were generated before the electrode was fully baked and cured (Wilkinson et al., Infacon IX). Cracking set up under those conditions becomes visible only when the affected section reaches a height where gas can escape, so the operating record of the preceding period is evidence, not background.
3. Route Two: Devolatilisation Running Late
Volatile matter leaves across the baking transition at approximately 450–500 °C, and its departure is what converts paste into conducting carbon (Larsen, Platinum 2012). Where devolatilisation runs at a different position or time than the operating cadence assumes, more gas is evolving in the upper region, and more of it is reaching the top of the column rather than being consumed lower down.
This route is the one that shows in the off-gas rather than only at the electrode. Elevated tar content is consistent with volatile matter evolving later than intended, and it is the discriminator between this route and the other two: tar alone is not conclusive: gas paths, combustion conditions and sampling conditions can also change what reaches the off-gas measurement.
The contributors are the same ones that move the baking zone: volatile matter above the range for the grade family, a slipping rate that advances material faster than it bakes, and a current density below what the thermal balance requires. A thermal factor reinforces the effect. Green paste thermal conductivity falls with temperature up to about 400 °C, reaching roughly 0.5 W·m⁻¹·K⁻¹, while pre-baked material in the baking-zone temperature range conducts an order of magnitude better (White et al., 2020) — so material that is running behind on baking is also poorer at conducting the heat it needs to catch up.
4. Route Three: A Change in Combustion Conditions at the Top
The third route involves no change in the electrode at all. What burns at the furnace top, and how completely, depends on how much air reaches the gas — and that is a property of the cover, the seals and the draught rather than of the paste.
As the degree of enclosure increases, less gas is produced for capture by the hood system and the CO concentration in the furnace gas increases (EPA, AP-42 §12.4). A degraded seal, a displaced cover panel, a change in hood draught or a shift in charge distribution changes the local air supply and therefore where and how the gas burns. On a covered furnace, combustion appearing where the cover is intended to suppress it is a cover-and-seal finding, not a paste finding.
This route is identified by its distribution and by what did not change: the flare is associated with a physical opening or a draught change rather than with one electrode; the off-gas tar loading is unchanged; and no paste, cadence or load change coincides with its onset.
5. What Evidence Should Be Collected?
Collected while the condition is present, since most of it cannot be reconstructed afterwards. The observation is made from the plant’s normal safe positions; nothing in this article requires approaching a flare, and gas issuing under pressure at furnace temperature is a hazard governed by the plant’s own safety system.
- Furnace class and the current cover, seal and draught condition, including any recent change to any of them.
- Position of the discharge point relative to the casing, the shoes and the burden surface, and whether it moves down with the column.
- Distribution: one electrode or all three; one point or spread across the charge surface.
- Persistence: steady, or fluctuating with charging or slipping.
- Off-gas tar indication, against the furnace’s own baseline for the same load and product.
- The operating record for the preceding period — slipping rate, electrode current, electrode movement, and any load change.
- The batch in service: Certificate of Analysis (COA), volatile matter against the range for the grade family, and whether a batch change coincided with onset.
- Photographs from the plant’s normal observation position, with the time and load recorded.
Items 2, 3 and 5 are the ones that separate the three routes; items 6 and 7 are what make the finding actionable rather than descriptive.
6. How Is the Cause Attributed?
| Observed combination | Most likely route | Next step |
|---|---|---|
| Fixed discharge point, one electrode, steady, moves down with the column; tar unchanged | Gas path through column or casing | Inspect casing condition at that height; review the preceding operating record for the crack-forming combination |
| Increased tar against baseline, flare distributed rather than localised, coincides with batch change or cadence change | Devolatilisation running late | Compare COA volatile matter against the grade family range; review slipping cadence and current density |
| Combustion associated with an opening, seal or draught change; tar unchanged; no paste or cadence change | Top-side combustion conditions | Cover, seal and draught inspection; not a paste question |
| Momentary flare with audible crack and instantaneous current interruption | Fracture event | Follow the breakage diagnostic path; this is not a gas investigation |
| Gas path present and the baking zone sitting low | Combined — treat as urgent | Casing defect with unbaked material adjacent can vent paste, not only gas |
Table 2. Attribution follows the combination of position, distribution and tar indication; a single observation identifies none of the three routes on its own.
The last row is the one that changes the priority. Where a discharge point coexists with a baking zone sitting low, the material adjacent to the defect may not be baked, and the same opening that vents gas can vent paste into the furnace room (Larsen, 2012). Where both conditions are present, the assessment stops being a diagnostic exercise and becomes an operating decision for the plant’s own standard.
7. Frequently Asked Questions
Q1. Our open furnace shows flame across the charge surface. Is that a problem? Not by itself. On an open furnace, CO rising through the charge burns with induced air between the charge surface and the hood, and virtually all CO and much of the organic matter burns there (EPA, AP-42 §12.4). The signals worth investigating are combustion localised at one electrode, a jet from a fixed point, or a change against the furnace’s own baseline — not the presence of combustion.
Q2. We see a steady flare at one point on one electrode. What is checked first? Whether the discharge point moves down as the column advances. A point that travels with the material indicates a defect in the column or casing rather than a condition of the furnace top. Record its height relative to the shoes and the burden, then examine casing condition at that height at the next opportunity, and review the preceding period for high slipping combined with high current.
Q3. Off-gas tar has risen but there is no visible flare. What does that indicate? Most consistently, devolatilisation running later than intended, which is the same mechanism that produces a route-two flare without necessarily producing a visible one. Compare the batch volatile matter against the range for the grade family, then review slipping cadence and current density. A tar rise with no paste or cadence change also warrants checking whether the off-gas measurement itself changed.
Q4. Can a flare be attributed to the paste supplier? Only through route two, and only with the evidence in Section 5. A fixed-point discharge is a casing or column defect; a combustion change at the top is a cover and draught matter. Where a batch change coincides with a tar rise and a distributed flare, the batch comparison is the next step rather than the conclusion — the operating cadence calibrated on the previous batch is a competing explanation and is checked in parallel.
Q5. Should the electrode be taken out of service? That decision belongs to the plant’s operating standard and safety system, not to any general guidance. The factor that raises urgency is the combination in the last row of Table 2: a discharge point coexisting with a baking zone sitting low, where the same opening can vent paste rather than gas. Establishing whether that combination is present is the useful contribution a diagnostic review can make to the decision.
8. Related Reading
- Paste Column Bridging and Uneven Sintering (companion article) — the discontinuous column that a gas path can follow
- Söderberg Electrode Hard Breakage vs. Soft Breakage — the momentary flare that belongs to a fracture event
- How the Baking Zone Forms and Migrates in a Self-Baking Electrode — the zone position that determines whether a defect vents gas or paste
- What Volatile Matter Actually Does Inside a Söderberg Electrode — the gas evolution behind route two
- Open-Top and Closed-Top Furnaces: Why Electrode Paste Specifications Diverge — why the baseline differs across furnace classes
- Irregular Current Transfer Around Söderberg Electrode Contact Shoes — the parallel condition at the same height
Author and Review
Written by Sterling Yang, Senior Vice President at Jiyang Carbon.
This article was reviewed by Mr. Yang, Technical Director at Jiyang Carbon, drawing on 40+ years of hands-on experience in carbon processing.
Furnace class definitions and gas behaviour are taken from the published sources cited in this article. No figures are given for flare frequency, flame dimensions or off-gas tar thresholds, because no verifiable general values exist and such indicators depend heavily on furnace class and configuration; the article gives comparison against the furnace’s own baseline instead. Observation is assumed to be from the plant’s normal safe positions, and no intervention method is prescribed.
Technical Consultation
For assistance in attributing a flare or a tar rise across the three routes, or in reviewing a batch against an onset date, please contact the technical team.
