The Physics of Ceramic Fiber Formation in Centrifugal Spinning

Centrifugal force, melt viscosity, surface tension and solidification — the four physical quantities that decide whether a 2,000°C melt becomes 3–5 µm fiber or a stream of shot

September 15, 2026 · Process Fundamentals · 13 min read
Written and reviewed by the Jinyuan Machinery process engineering team. We build and commission the spinning machines and complete lines this article describes. To keep the physics verifiable, every number carries a source label: rheology theory means it follows from standard melt and glass rheology that any materials engineer can reproduce; industry reference means it is an empirical range quoted across industry technical literature and varies by process and product; field practice means it comes from our own commissioning records and operating lines. Where the industry disagrees with itself, we say so. Nothing here replaces the process documentation issued with a specific line.

Every ceramic fiber blanket starts as the same physical problem: how do you turn a pool of alumina-silicate melt at 1,900–2,200°C into filaments three to five micrometres across — and do it continuously, at tonne-per-hour rates, without the melt simply beading up into droplets?

The answer on a spinning line is a controlled fight between three forces. Centrifugal force at the roller rim does the stretching. Surface tension constantly pulls the melt back into spheres. Melt viscosity decides who wins — it is simultaneously the resistance the fibers must overcome and the property that lets a stretched filament survive long enough to freeze. Get the balance right and you get long, fine, uniform fiber. Get it wrong and the same machine, the same rollers and the same melt composition produce shot.

This article walks through that physics the way we use it on the line: what the centrifugal field actually does, why viscosity — not roller speed — is the first lever, where the alumina/silica ratio fits in, why shot forms, and why the same physics is why spun fiber outperforms blown fiber for needle-punched products.

Spinning machine and forming section of a ceramic fiber blanket production line in the assembly workshop
Spinning machine and forming section of a ceramic fiber blanket line — every parameter discussed here is set or stabilized within the few metres between the melt stream and the fiber falling onto the belt.

The Driving Force: A 15,000 g Field at the Roller Rim

Fiber forming on a spinning machine is an extensional process, and the extension is driven by centrifugal force. For a melt element on the roller surface:

F = m · ω² · r

where ω is the angular speed of the roller and r the distance of the melt element from the axis. Two engineering consequences follow directly from the formula (rheology theory):

  • Linear rim speed is what matters, not rpm. Since v = ω·r, the force scales as v²/r. A small roller running very fast and a large roller running slower are equivalent — which is why drive arrangements can be traded freely, and why quoting a roller speed without the roller diameter is meaningless.
  • The force peaks at the rim. The further the melt element sits from the axis, the stronger the field. Fiber is therefore thrown off and stretched at the roller edge, and the condition of the rim — grooves, buildup, wear — directly shapes the fiber.

Putting realistic numbers on it (industry reference): a spinning roller at roughly 9,500 rpm with a diameter near 0.30 m gives a rim speed of about 150 m/s and a centrifugal acceleration at the rim of approximately 1.5 × 10⁵ m/s² — on the order of 15,000 g. In a field of that magnitude, any melt whose viscosity is in the right range will be drawn into filaments. This is the physical confidence behind the whole method: the machine does not coax the melt into fiber, it forces it.

Why capacity scales with roller geometry. Because force grows with rim speed, a larger roller diameter raises the throwing and stretching capability directly, and single-machine output scales with both roller length and diameter. This is also the physical reason the industry moved from early three-roller staging to the modern two-roller arrangement: the same total stretch ratio can be achieved in one transfer with a correctly sized high-speed roller plus compressed-air assist, with fewer fiber-breaking transfers between rollers — see two-roller vs three-roller spinning machines.

The Opponent and the Tool: Melt Viscosity

Viscosity is the property that makes or breaks the process, and it has a steep, exponential relationship with temperature. Alumina-silicate melts follow the Arrhenius-type relation (rheology theory):

η = A · exp(Eη / RT)

with a viscous-flow activation energy Eη in the range of roughly 200–400 kJ/mol for alumino-silicate melts (industry reference, literature span across compositions). The exponential form has one practical consequence every operator should internalize: a small temperature error becomes a real viscosity error. At melt temperatures around 2,000°C, a ±10°C deviation shifts viscosity by roughly ±5–10%; ±20°C shifts it by roughly ±10–20% (derived from the Arrhenius relation with the Eη span above — rheology theory).

The Temperature Control Target — Calibrated, Not Quoted

Industry articles often state that furnace temperature must be held within "±5°C". We would not ask a production line to achieve that, and the physics says it is unnecessary (field practice, calibrated against furnace-supplier data). An industrial furnace melting at 2,000°C has a large bath, large thermal inertia, continuous charging disturbances, and an internal temperature field that is simply not uniform — lateral gradients of tens of degrees exist inside the bath as a matter of course. Whole-furnace fluctuation in real plants runs around ±10–±30°C, and a well-tuned system (PID tuning plus power balancing) converges to roughly ±8–±10°C.

The correct target is narrower and smarter: stabilize the melt temperature at the flow outlet, because that is the melt that actually gets spun. With stable charging and melt level, closed-loop power control and heated, insulated flow outlets, holding outlet temperature within about ±10–±20°C is achievable and fully sufficient for the viscosity window (field practice). When outlet drift widens beyond roughly ±30°C, the symptom is unmistakable on the belt: visible diameter variation in the fiber and rising shot.

The Forming Window

Melt conditionViscosityPhysical result
Too thin< ~1 Pa·sDroplets cannot be extended — surface tension wins, output turns to shot and coarse beads
Forming window~1–15 Pa·sFilaments stretch and survive; commonly quoted optimum ~2–5 Pa·s
Too thick> ~15 Pa·sMelt cannot be drawn — coarse, short fiber, frequent breakage

Honesty about this table (industry reference): the literature does not agree on a single window. Technical sources quote 1–10 Pa·s, 5–15 Pa·s and 10–15 Pa·s depending on whether the product is blanket or board and on the machine type. All of these are empirical values, not standard specifications. The window that matters is the one calibrated on your own line with your own charge materials during melt trials — and once it is calibrated, the fastest diagnostic in the plant is knowing where in the window the melt currently sits.

The Chemistry That Sets the Window: Al₂O₃ / SiO₂

The viscosity window is not just a temperature problem — the melt composition decides where the window sits and how wide it is.

  • Raising Al₂O₃ increases the network connectivity of the melt, which raises viscosity and, after forming, the continuous-use temperature of the fiber (rheology theory; the alumina ladder in GB/T 3003 grades reflects exactly this).
  • Raising SiO₂ builds a more continuous glass network — better drawability, smoother forming — at the cost of temperature resistance (rheology theory, silicate glass chemistry).

The workable balance for melt-spun fiber is an Al₂O₃/SiO₂ ratio of roughly 45/55 to 50/50 (industry reference), which is the compromise point between drawability and temperature resistance, consistent with the 42–55% Al₂O₃ span across standard fiber grades. Production chemistry control then becomes a guardband exercise: Fe₂O₃ and alkali oxides creeping upward suppress viscosity and quietly shift the forming window — one of the reasons a line can drift out of spec without any equipment change.

High alumina is a one-way street with a limit. Higher Al₂O₃ buys temperature resistance but raises both viscosity and surface tension. Standard melt-spun high-alumina fiber tops out around 52–60% Al₂O₃ — near the ceiling of what centrifugal spinning can draw (industry reference). Beyond that the industry changes method, not chemistry: polycrystalline fiber made by colloidal routes. And where a 1,400°C-class product is needed from a melt-spun line, the practical answer is zirconia- or chromia-bearing grades — adding ZrO₂ or Cr₂O₃ to lift temperature performance without pushing Al₂O₃ past the spinnable limit.

Surface Tension: The Reason Shot Exists

Surface tension is the quiet adversary of the whole process. A liquid minimizes its surface area, and for a small volume of melt the minimum-area shape is a sphere. Every microsecond a filament exists, surface tension is working to contract it back into droplets. Fiber forming succeeds only while the stretching stress outpaces that contraction.

The magnitude comparison (rheology theory, order-of-estimate): the centrifugal tensile stress available scales as ρ·v² — with v the rim speed — while the surface-tension back-pressure on a filament of radius r scales as 2γ/r. At 150 m/s rim speed the centrifugal term dominates comfortably for a melt in the viscosity window, and the filament keeps thinning. When it does not, the physics announces itself immediately:

ConditionPhysical behaviourResult on the belt
Balance normalCentrifugal stress beats contraction; filament keeps thinning until it freezesUniform fiber, low shot
Al₂O₃ above ~60%Surface tension elevated; available tensile stress insufficientShot only — the melt refuses to form fiber (well-documented industry phenomenon)
Gas bubbles in meltInterface disturbances as bubbles burst through filamentsFiber with bubble inclusions, breakage, higher shot

This is the physical answer to a question every buyer eventually asks — why can't we just specify 65% alumina for more temperature headroom? Because past the spinnable ceiling the line does not produce coarser fiber; it produces no fiber at all.

The Four Stages of Fiber Formation

Between the flow outlet and the fiber curtain landing on the belt, forming proceeds through four physically distinct stages. Each has its own controlling variable and its own characteristic failure mode:

StageWhat happensControlling parameterFailure mode
1. Film spreadMelt spreads into a thin film on the distribution rollerDistribution roller speed, roller surface temperature and wettingUneven film → diameter variation along the fiber
2. Viscous drawingFilm transfers and stretches between rollersSpeed matching between rollers (stretch ratio)One excessive stretch step → breakage, fiber length drops
3. Filament throwingFilaments leave the rim; centrifugal force vs surface tensionRim linear speedSurface tension wins → shot
4. FreezingFlying fiber cools to glass; diameter locked inCooling air volume and placementToo fast → brittle breakage; too slow → fibers stick together

Two of these stages deserve emphasis from commissioning experience (field practice). First, the stretch ratio between stages matters more than the absolute speed of any one roller — the speed matching is where fiber length is won or lost. Second, cooling is a "fast but not harsh" operation: the air curtain must freeze the fiber before fibers touch each other, but over-cooling chills the glass below its working point while it is still under tension and produces brittle, dust-prone fiber. The equipment-level details are covered in our guide to spinning machine technology.

Spinning vs Blowing: Same Melt, Different Physics

The melt can also be fiberized without rollers, by hitting the stream with a high-pressure air jet. The two methods differ not in degree but in the nature of the applied force — and every practical difference between spun and blown fiber follows from that:

Physical dimensionCentrifugal spinningAir blowing
Stretch force sourceCentrifugal force from a roller surface, directed along the filament axisShear and drag from a 0.6–0.8 MPa air jet
Force–time characterContinuous, gradual gradient, controllableInstantaneous, turbulent, random
Force uniformityGood — filaments drawn evenlyPoor — force varies drop to drop
Surface tension compensationCan be managed through speed and viscosityDifficult to compensate finely
Typical fiber3–5 µm diameter, 150–250 mm long, low shot2–3 µm diameter, short, higher shot
CoolingBallistic flight plus assist airForced air cooling dominates
Single-machine capacityHigh (10,000-tonne-class lines)Low to medium

The one-line physics summary we use internally: blowing smashes the stream with air; spinning draws it out with force. Blowing is fast but crude; spinning is steady and uniform. That is why needle-punched blanket and folded modules — products whose strength depends on fiber length — are made on spinning lines. A fuller equipment-level treatment is in our spinning machine technology guide.

Practical Rules From the Line

The physics above condenses into a short list of operating rules that we hand to every customer during commissioning (field practice):

  1. Viscosity before speed. Stabilize furnace outlet temperature and get viscosity into the window first; only then tune roller speed. Most "the fiber is too coarse" complaints trace back to melt temperature, not to the machine.
  2. Move temperature and speed together. Raising melt temperature lowers viscosity — and the correct compensation is to raise roller speed to keep the strain rate matched. The two setpoints are a pair, not independent knobs.
  3. Read the shot. Shot content is the cheapest fiber-quality sensor on the line. Sudden increase → measure outlet temperature drift first. Persistent increase at stable temperature → check melt chemistry (Fe₂O₃, alkalis).
  4. Cool "fast but not harsh." Cooling air set to just enough to freeze; excess air trades a sticking problem for a brittleness problem.

For the diagnostic side of quality — how diameter and shot are actually measured and specified — see our guides to fiber diameter control and shot content in spun fiber.

Four Misconceptions the Physics Corrects

Common beliefWhat the physics actually says
"Faster rollers always make finer fiber"Diameter is set by stretch ratio and viscosity window together. Out of the window, no speed produces good fiber
"Hotter is easier to spin"Too hot means viscosity below the window — the melt beads instead of drawing. The goal is a precise temperature, not a high one
"Just add alumina for higher temperature grades"Alumina raises surface tension along with viscosity; past ~60% the melt forms shot only. Temperature grades come from ZrO₂/Cr₂O₃ additions or a different process
"Finer fiber is always better fiber"Over-fine fiber is weak and dusts. Needle-punched blanket needs a controlled 3–5 µm distribution for strength and handling

Frequently Asked Questions

Why doesn't faster roller speed always produce finer ceramic fiber?
Because fiber diameter is set by the combination of stretch ratio and melt viscosity, not by roller speed alone. If the melt viscosity is outside the forming window, the extra centrifugal force cannot be converted into useful extension: a melt that is too thin forms droplets and shot regardless of rim speed, and a melt that is too thick cannot be drawn out at all. Speed only helps when viscosity is already inside the window, which is why furnace temperature should be stabilized before roller speed is touched.
What melt viscosity is needed to spin ceramic fiber?
Published industry practice places the workable forming window at roughly 1 to 15 Pa·s, with different sources citing 1–10, 5–15 or 10–15 Pa·s depending on the product form and machine type; a commonly quoted optimum sits near 2–5 Pa·s. These are empirical industry values, not standard specifications — the practical window for a given line is established by melt trials with that line's own charge materials. Below about 1 Pa·s the melt tends to form shot; above about 15 Pa·s it cannot be drawn and produces coarse, short fiber with breakage.
How precisely must the furnace temperature be controlled for spinning?
On an industrial furnace melting at around 2,000°C, holding the whole furnace bath to ±5°C is not realistic — large thermal inertia, continuous charging and an uneven temperature field put real-world fluctuation at roughly ±10 to ±30°C, with well-tuned systems converging to about ±8–±10°C. What actually matters is the melt temperature at the flow outlet, because that is what sets viscosity. Keeping outlet melt temperature within roughly ±10–±20°C is sufficient for stable fiber forming; when drift grows beyond about ±30°C, fiber diameter variation becomes visible and shot content rises.
Why does high alumina content cause shot instead of fiber?
Alumina raises both the viscosity and the surface tension of the melt. Above roughly 60% Al₂O₃ the surface tension effect dominates: when the melt is flung from the roller rim, the tensile stress available cannot overcome the melt's tendency to contract into droplets, so the discharge turns into shot rather than continuous filaments. That is why melt-spun high-alumina fiber tops out around 52–60% Al₂O₃, and higher temperature grades are achieved by adding zirconia or chromia instead of ever more alumina.
Why does centrifugal spinning produce longer fiber than air blowing?
The two methods apply force in fundamentally different ways. Spinning draws the melt continuously along the filament axis from a roller surface, with a gradual force gradient and a controllable direction, so filaments are stretched evenly and stay long — typically 150–250 mm at 3–5 µm diameter. Blowing tears the stream apart with a turbulent, intermittent air jet, so the force is random in direction and duration; fibers come out shorter, with a wider diameter spread and more shot. That is why needle-punched blanket, which depends on fiber length for tensile strength, is made by spinning.
What should I check first when shot content suddenly increases?
Check furnace temperature drift first. A sudden rise in shot almost always means the melt viscosity has moved out of the forming window, and temperature is the fastest-moving influence on viscosity. If outlet temperature is stable and shot remains high, look at melt chemistry — rising Fe₂O₃ or alkali content suppresses viscosity and shifts the window. Speed changes come last: adjusting roller speed before viscosity is back inside the window treats the symptom and usually makes the fiber distribution worse.

The Bottom Line

Ceramic fiber formation is not a craft secret — it is a measurable contest between centrifugal stress, surface tension and a viscosity window set by chemistry and temperature. A 15,000 g field at a 150 m/s roller rim provides the force; a melt held at roughly 2–5 Pa·s (inside the 1–15 Pa·s working band) provides the formability; and controlled cooling locks the result in as glass. When a line produces shot, breaks fiber or drifts out of diameter spec, the cause is almost always one of these four quantities moving — and the fastest route back is to check them in the order the physics dictates: outlet temperature, then chemistry, then speed.

If you are evaluating a spinning line, ask the supplier not just what the equipment does but where its calibrated viscosity window sits and how the melt outlet temperature is held. The answers separate a line that was engineered around the physics from one that was assembled around a price.

Related Reading

Continue with these guides and equipment pages:

Related Equipment

Need a Line Engineered Around the Physics?

Tell us your target fiber grade and capacity, and our process engineers will walk you through the viscosity window calibration, roller configuration and forming parameters we would apply.

Contact Us