The problem with trusting the screw
A single-screw extruder is a pump with a deliberately leaky design. The output depends on drag flow minus pressure flow, and pressure flow depends on the die resistance, which changes as the screen pack loads, as the die heats and cools, and as the resin lot changes viscosity. The result is that die pressure drifts by 5–15% over a shift even when nothing is touched.
On a profile line that drift shows up as a slow dimensional wander. On a sheet line it shows up as gauge. On a coating line it shows up as coat weight. In every case the operator compensates by adjusting something else, which introduces a second variable and makes the process harder to control than it was.
The core insight: pressure stability and temperature stability are not the same thing, and most lines are instrumented for temperature while being limited by pressure. A line holding ±1 °C on every zone can still produce ±5% gauge variation if the die pressure is moving. The temperature loops are not the problem — they are just the only thing being watched.
Where pressure variation actually comes from
| Source | Time scale | Magnitude | Fixable by |
|---|---|---|---|
| Screw pulsation (flight frequency) | Seconds | 1 – 3% | Melt pump, or a different screw |
| Screen pack loading | Hours | 5 – 15% | Screen changer, or pump |
| Resin viscosity lot variation | Days | 3 – 10% | Pump (pressure-controlled), or incoming inspection |
| Die temperature drift | Hours | 2 – 5% | Die temperature control |
| Barrel zone drift | Hours | 2 – 4% | Barrel temperature control |
| Feed inconsistencies (flood vs starve) | Minutes | 3 – 8% | Feed control, or pump |
| Die build-up | Days | 5 – 20% | Die cleaning; nothing else works |
Table 1 — Pressure variation sources. The pump addresses the first three rows, which are usually the dominant ones. It cannot fix die build-up, and no equipment can.
The screen pack is the one that catches people. A fresh screen might run at 120 bar; after eight hours of production it is at 145 bar because it is partly blocked. The die sees 25 bar more, the output drops, and the wall thins. The operator notices the wall and turns the screw up, which raises pressure further. By the end of the shift the process is in a completely different state from where it started, and nobody can say when it changed.
The gear pump: what it does and does not do
A melt pump is a positive-displacement gear pump sitting between the screen changer and the die. It takes melt at whatever pressure the extruder delivers and pumps it out at a fixed volumetric rate, because the gears displace a fixed volume per revolution.
| Property | Without pump | With pump |
|---|---|---|
| Output stability | ±3 – 8% | ±0.3 – 1% |
| Die pressure stability | ±5 – 15% | ±1 – 2% |
| Response to screen loading | Immediate output loss | Absorbed; suction pressure rises, output unchanged |
| Response to viscosity change | Output changes | Output unchanged; pressure changes |
| Start-up time to stable | 20 – 40 min | 5 – 10 min |
| Scrap at changeover | High | Low |
Table 2 — What a melt pump changes. The output stability number is the one that justifies the capital, but the start-up time is the one operators notice.
The pump is controlled on suction pressure, not on discharge pressure. The extruder screw speed is trimmed to hold a constant pressure at the pump inlet — typically 30–80 bar — while the pump runs at a fixed speed set by the line. This is why the arrangement is sometimes called a "pressure-controlled, rate-controlled" system: the screw is pressure-controlled, the pump is rate-controlled.
What the pump does not do: it does not mix, it does not homogenise temperature, and it does not remove gels. Those are the extruder's and the static mixer's job. A pump installed to fix a gel problem will not fix it, and the disappointment is common enough to be worth stating.
The screen changer: continuous vs discontinuous
A screen changer holds the screen pack that filters out contamination and unmelted particles. There are two families and the choice depends on how much you value not stopping.
| Type | Operation | Pressure disturbance | Cost |
|---|---|---|---|
| Manual slide plate | Line stops, plate slid by hand | Full stop and restart | Lowest |
| Hydraulic slide plate (discontinuous) | Brief stop while the plate indexes | Short interruption, some scrap | Low |
| Continuous double-bolt | Second bolt already in line, screens changed alternately | Minimal; small pressure bump | Medium |
| Continuous rotary disc | Disc rotates slowly, screens changed one at a time | Very small | High |
| Backflush / self-cleaning | Reverse flow cleans the screen in place | Small | Highest |
Table 3 — Screen changer types. A continuous changer is what makes a pump worthwhile — a discontinuous changer reintroduces exactly the pressure step the pump was bought to remove.
The pairing matters. A melt pump with a manual screen changer is a half-solution: the pump holds output beautifully for eight hours and then the line stops to change screens, and the restart takes twenty minutes to stabilise. A pump with a continuous changer runs for weeks at steady state.
Sizing the pump and the screens
Pumps are sized by displacement per revolution and by maximum pressure. The displacement should be chosen so that the pump runs at 20–60% of its maximum speed at the design output — running at 90% wears the gears and leaves no headroom for a formulation change.
| Parameter | Typical value | Consequence if wrong |
|---|---|---|
| Displacement | Sized for 40–50% speed at design output | Too small: no headroom. Too large: poor low-speed control |
| Suction pressure | 30 – 80 bar | Too low: cavitation and starvation. Too high: extruder working hard for nothing |
| Discharge pressure | Set by the die, typically 80 – 250 bar | Determines pump size and bearing life |
| Temperature control | ±1 °C on the pump body | Viscosity change across the pump changes the leak rate |
| Screen mesh | 20/40/60 or 40/80/120 depending on resin | Too fine: pressure builds fast. Too coarse: contamination passes |
| Screen area | Sized for 8–24 h between changes | Too small: constant changes. Too large: pressure drop is fine but the housing is big |
Table 4 — Pump and screen sizing. The screen mesh choice is a compromise between filtration quality and how often the line stops.
Screen mesh is usually specified as a stack: a coarse breaker screen at the front to take the load, then progressively finer screens, then a coarse support screen at the back. A typical PVC stack is 20/40/60; a typical PE film stack is 40/80/120/20. The fine screen does the filtering; the coarse ones protect it from being torn by pressure.
When it is worth the capital
A melt pump and a continuous screen changer together cost a meaningful fraction of a small line. Whether it pays back depends on what the variation is costing.
| Situation | Payback case |
|---|---|
| Tight dimensional tolerance (under ±3%) | Strong. Often the only way to meet it repeatably |
| High material cost (engineering resins, filled compounds) | Strong. Gauge saving of 2–3% pays back quickly |
| Long runs (weeks on one product) | Strong. Start-up scrap and drift both fall |
| Frequent colour or product changes | Moderate. Faster stabilisation helps, but purge dominates |
| Loose tolerance commodity profile | Weak. The tolerance does not need it |
| Very short runs (hours) | Weak. Start-up is not the constraint |
Table 5 — When the capital is justified. The strongest case is tight tolerance plus expensive material, which is exactly where most technical profiles sit.
The payback arithmetic is usually done on material saving alone and that understates it. The bigger number is often the reduction in off-spec product and the elimination of the operator's habit of adjusting the line every hour. A process that does not need adjusting is a process that can be documented, and a documented process is what allows a second shift to produce the same part as the first.
Retrofitting to an existing line
Retrofits are common and usually successful, but there are three constraints that decide whether it is straightforward.
Space. A pump and a changer add roughly 1.2–2.0 m to the line between the extruder and the die. On a tightly packed line this is the whole problem, and it may require relocating the die and re-running the downstream.
Extruder pressure capability. The pump adds 30–80 bar of suction pressure that the extruder did not previously have to generate. An older extruder with a marginal thrust bearing may not be happy. Check the allowable head pressure before committing.
Die and adapter heating. The extra hardware adds residence time and thermal mass. The adapter between pump and die needs its own temperature control, and the start-up sequence changes because there is now a volume of melt that has to be brought up to temperature before the pump can turn.
Does a melt pump improve mixing?
No. It is a positive-displacement pump, not a mixer. It improves pressure and output stability. If you have a dispersion or gel problem, the fix is screw design, a static mixer, or better incoming material control — not a pump.
Can I run a melt pump without a screen changer?
You can, but you lose much of the benefit. The pump holds output constant while the screen loads, which means the pressure simply rises until the screen blinds or the extruder reaches its limit. Without a changer you have moved the problem rather than solved it.
What suction pressure should the pump run at?
Typically 30–80 bar. Too low and the pump starves, which produces output fluctuation and gear wear. Too high and the extruder is working against a resistance it does not need to. The set point is chosen so that the extruder runs at a comfortable specific output.
Why does my die pressure still vary with a pump installed?
Because the pump holds volumetric output, not pressure. If the resin viscosity changes, the pressure needed to push that volume through the die changes. That is normal and it is not a fault — the output is still constant, which is what matters for dimensions.
How often do screens need changing?
On a continuous changer, one bolt changes roughly every 15–60 minutes depending on contamination level, and the whole pack is replaced once a shift or once a day. The signal is suction pressure reaching its upper limit, not a clock.
Does a pump help with colour changeover?
Indirectly. It shortens the time to reach stable output after a change, so less material is run before the product is on spec. The actual colour transition is still governed by purge and by the dead volume in the die and adapter.
What happens if the pump stops?
The die pressure collapses and the product goes off-spec immediately. Most lines have an interlock that stops the haul-off and diverts the product. The extruder must also be stopped or put into a recirculation mode, or the suction pressure will spike and trip the overpressure protection.
Is a pump worth it on a small line?
If the tolerance is tight or the material is expensive, yes — the economics do not depend much on line size. If the product is a commodity profile at ±10%, no. The decision is about what the variation costs, not about the size of the machine.
Trying to hold a tolerance your line cannot repeat?
Tell us the product, the tolerance band, the resin and the current variation you are seeing. We will tell you whether the limit is pressure stability, temperature stability or tooling — and which of those is worth fixing first.
EN 
ES 
FR 
DE 
PT 
AR 
RU 
JA 
KO 
IT 
NL 
TR 
PL 
ID 