Cloth doing a job

A blinded filter cloth is a length of its own cake

Every filtration calculation carries a medium resistance, and the values tabulated for cloth run from about 10⁹ to 10¹¹ per metre. A clean woven filter cloth is more than three decades below that — under a hundredth of a single layer of the particles it is meant to hold. The tabulated number is the cloth with its holes full of cake, and once that is written down the particle cancels: a blinded cloth resists exactly as much as a fixed thickness of whatever has packed it, 0.86 mm for the filter cloth and a third of a millimetre for a voile.

Worth reading first: A woven filter beats its own rating · A hole is a channel, not an opening · A filter is rated by the hole it does not show.

A woven filter beats its own rating ended with the cloth handing its job over. Its hair layer catches a few per cent of particles ten times smaller than its holes, a few per cent is enough to start a cake, and once the cake exists the cake is the filter. That settles what stops the particles. It leaves open what the cloth contributes to the other half of the duty, which is how fast the liquid gets through.

Filtration engineering has a standard answer, and it is a number. Constant-pressure filtration through a cake follows a straight line in the right coordinates, and the line has an intercept that is credited to the filter medium — the medium resistance, tabulated for cloths at roughly 10⁹ to 10¹¹ per metre. The cloth’s own share of the duty is supposed to be that intercept.

The clean cloth cannot be that number. Worked out from the cloth’s own holes, the filter cloth among the named cloths resists 6.2×10⁵ per metre, more than three decades below the bottom of the band. What the tabulated resistance describes is the cloth after the slurry has got into it, and when that is written out the particle size drops out of the answer and leaves a length.

A filter cloth's resistance written as a thickness of its own cake, clean and blinded. The resistance of the filter cloth expressed as the thickness of cake that would resist as much, against the diameter of the particles in the cake, with one layer of particles drawn for scale. Clean, the cloth is worth 6.1e-4 micrometres of a one-micrometre cake and 0.06 micrometres of a ten-micrometre one, 0.0061 of a single layer. With its holes packed with the cake from the face down to the waist it is worth 0.43 mm, and packed through 0.86 mm, the same for every particle. Spun threads that the cake cannot enter would lower the packed figure, and are drawn separately. The chart cannot show whether a given duty packs the holes at all, which depends on how the first particles arrive.
Fig. 1 The filter cloth’s resistance written as the thickness of cake that would resist as much, against particle diameter. Clean, it is a vanishing fraction of one layer of particles at every size. With its holes packed, it is a fixed thickness, the same for every particle.

What a clean cloth resists

The filter cloth here is a 40 tex plain weave at 20 ends and 19 picks a centimetre, about half a millimetre thick and second only to the duck in weight. A hole is a channel, not an opening worked out the shape of its holes as they run through that thickness: wide at the face, pinched at a waist where the warp and the weft both transit, and wide again at the back. At its waist the filter cloth’s hole is 0.26 by 0.29 mm, open across 29 per cent of the cell it sits in.

A liquid going through that channel slowly loses pressure by viscous drag alone. The fourth power is a close cloth’s rule showed that an open cloth at a hundred pascals spends most of its pressure on the entrance and exit rather than on the drag, but filtrate comes through a cloth at a millimetre a second or less, and at that speed the drag is all there is. In that limit the resistance belongs to the holes and not to the fluid, so air in a permeameter and water in a filter see the same number: 6.2×10⁵ per metre for the filter cloth.

A cake is a bed of packed particles, and its resistance comes from the same law the cloth’s threads already obey. A cloth stops having holes before it stops passing air used Kozeny and Carman’s relation for the fibre inside a yarn; written for spheres of diameter d packed with a void share ε, it gives a permeability

k=d2 ε3180 (1−ε)2,k = \frac{d^2\,\varepsilon^3}{180\,(1-\varepsilon)^2},

about 10−13 m210^{-13}\,\mathrm{m}^2 for ten-micrometre silica at forty per cent voids. A single layer of those particles, ten micrometres thick, then resists d/k, about 10⁸ per metre.

So the clean cloth is worth 0.006 of one layer of its cake. At a hundred micrometres it is worth six hundredths of a layer, and at one micrometre six ten-thousandths. In the first figure that is the lowest line, rising with the square of the particle and nowhere near the dashed line that marks one layer. A cake overtakes a clean filter cloth before it has finished putting down its first particle.

The number the textbooks carry

Ruth’s analysis of 1933 is the standard account of what happens next. With a medium of resistance R and an incompressible cake whose resistance grows in proportion to the filtrate already collected, the time t to collect a volume v a square metre at constant pressure satisfies

tv=μ α c2 ΔP v+μ RΔP,\frac{t}{v} = \frac{\mu\,\alpha\,c}{2\,\Delta P}\,v + \frac{\mu\,R}{\Delta P},

a straight line whose slope belongs to the cake and whose intercept belongs to the medium. Here α is the cake’s resistance per kilogram, c the solids carried by each cubic metre of filtrate and μ the liquid’s viscosity. A filtration test is fitted to that line, and the intercept, turned back into a resistance, is what gets tabulated for a cloth.

What a filter cloth resists when its holes are packed, against the band quoted for a filter medium. The resistance of the filter cloth per metre of thickness, on a logarithmic scale, against particle diameter. Clean it is 6.2×10⁵ per metre whatever is being filtered. With its holes packed with cake it falls as the square of the particle: 8.7×10¹¹ at one micrometre, 8.7×10⁹ at ten and 8.7×10⁷ at a hundred. Its spun threads are a second path of 1.4×10⁹, open only to particles too coarse to enter the threads' own pores, about 10 micrometres here. Shaded is the range filtration texts tabulate for a cloth medium, roughly 10⁹ to 10¹¹, which the packed cloth crosses and the clean one sits more than three decades under. The chart cannot show a particle population, only one size at a time.
Fig. 2 The filter cloth’s resistance against particle size on a logarithmic scale. Clean, it is flat and far below the shaded band filtration texts tabulate. With its holes packed it falls as the square of the particle and crosses the band. Its spun threads offer a second path, drawn only where the particles are too coarse to enter them.

The clean cloth sits at the bottom of this chart and the tabulated band sits across its middle. If the clean figure were what a filtration test measured, the intercept would be invisible. For a five-micrometre silica at one per cent solids and half a bar it would be about a hundred-thousandth of a second per litre a square metre, and the cake would match it after two and a half millilitres a square metre had come through. Real filtration lines have intercepts that are plainly visible, and a cloth medium that measured 6×10⁵ would not have a table of its own.

What does land in the band is the same cloth with its holes full. Packed with ten-micrometre silica, the filter cloth’s holes resist 8.7×10⁹ per metre, inside the band; packed with one-micrometre silica, 8.7×10¹¹, above it; packed with a hundred-micrometre sand, 8.7×10⁷, below it. The band is a band of particle sizes. It is not a property of cloth, and the cloths that fill it in a test are cloths whose channels have taken in some of the cake.

The filtration trade has a word for that state, blinding, and treats it as a fault that develops in service, a cloth gradually clogging until it has to be washed or replaced. The arithmetic says the start of it is not a fault. It is the ordinary condition of any woven cloth under a cake, reached in the first minutes, and it is the condition the medium resistance was measured in.

A packed hole is a length of cake

The packed cloth’s resistance needs one more step, and the step removes the particle.

At a depth z through the cloth the hole leaves a share φ(z) of the cell open. A cake packed into that hole at the cake’s own voidage has the cake’s permeability k, but only over the share φ(z), so a thin slice dz of packed hole resists as much as a slice of open cake dz/φ(z) thick. Add the slices up:

Rpacked=1k∫dzφ(z).R_{\text{packed}} = \frac{1}{k}\int \frac{dz}{\varphi(z)}.

The cake’s own resistance per metre of thickness is 1/k. So the packed cloth resists as much as a layer of the same cake whose thickness is

L=∫dzφ(z),L = \int \frac{dz}{\varphi(z)},

and that thickness has no particle in it. It is read off the shape of the hole and nothing else. For the filter cloth, packed all the way through its 0.48 mm, L is 0.86 mm. A silt, a fine silica and a coarse sand packed into the same cloth each find it worth 0.86 mm of themselves.

How much cake a filter cloth is worth as the cake packs further into its holes. The thickness of cake equivalent to the filter cloth's holes packed from the face down to a depth, against that depth, with the depth itself drawn dashed for comparison. Where the hole is as wide as its cell the two run together; through the waist, where only 29 per cent of the cell is open, the packed length climbs faster than the depth. Packed down to the waist, 0.24 mm, it is worth 0.43 mm of cake; packed through all 0.48 mm, 0.86 mm. Every number holds for any particle size. The chart cannot show how far a real cake gets in, which is set by how the first particles bridge.
Fig. 3 How many millimetres of cake the filter cloth’s holes are worth as the cake packs further in from the face, against the depth it has reached. Near the face, where the hole is as wide as its cell, the two keep pace; through the waist the packed length runs ahead of the depth.

The integral says where the cost comes from. Near the face and near the back the hole is as wide as its cell, because the threads beside it have crossed to the other side, and a packed slice there costs exactly its own thickness. Through the middle, where the open share falls to 29 per cent, each tenth of a millimetre packed costs more than three tenths of a millimetre of cake. The waist is where a blinded cloth’s resistance lives, which is the same place a filter is rated by the hole it does not show found its retention to live, for a different reason.

How far the cake gets in decides how much of that is collected. A cake whose first particles bridge the waist and pack the funnel above it fills the hole from the face down to the waist, 0.24 mm, and is worth 0.43 mm. A cake that bleeds fines through for its first minutes, and lodges some of them below the waist, can pack the whole channel and is worth 0.86 mm. Between those two is the range a duty can actually be in, and both ends are lengths of the order of the cloth’s own thickness.

Six cloths, ranked twice

The length depends on two things: how thick a cloth is, and how narrow its waist is as a share of the cell. A clean cloth’s resistance depends on something else again, the widest channel and the fourth-power law that charges heavily for every narrowing. So the two rankings need not agree, and on six of the named cloths they do not.

Six cloths clean and blinded, which do not rank the same way. Six named cloths, each with its clean resistance beside the thickness of cake its holes are worth when packed through. voile 1.6×10⁵ per metre clean, 0.35 mm packed; muslin 3.6×10⁵ per metre clean, 0.52 mm packed; poplin 6.6×10⁵ per metre clean, 0.51 mm packed; sheeting 1.3×10⁶ per metre clean, 0.75 mm packed; duck 4.4×10⁵ per metre clean, 1.02 mm packed; filter 6.2×10⁵ per metre clean, 0.86 mm packed. Clean, from least: voile, muslin, duck, filter, poplin, sheeting. Packed, from least: voile, poplin, muslin, sheeting, filter, duck. The clean figure is set by the widest channel and the packed one by the cloth's thickness over its narrowest open share, so a thick cloth with a narrow waist does worst. The chart cannot show release or bleed, which are the other two things a cloth is chosen for.
Fig. 4 Six named cloths, each with its clean resistance beside the length of cake its holes are worth when packed through. The voile is least in both columns. The others reorder: the duck and the filter cloth move to the bottom once their holes are full.

The voile is the easiest cloth in both columns, at 1.6×10⁵ per metre clean and 0.35 mm packed: it is thin and its waist is half open. The sheeting is the hardest clean cloth on the table, 1.3×10⁶, because its holes are small, but packed it is worth 0.75 mm and sits fourth. The duck is the third easiest clean cloth and the very worst blinded, 1.02 mm, because it is thick. The filter cloth, which is on the table because it is the kind of cloth a filter press uses, moves from fourth clean to fifth packed.

A cloth chosen by its clean permeability has been chosen by the wrong statistic. The clean figure is what a permeameter reads and what a datasheet quotes. Under a cake the cloth’s holes are packed within minutes and the clean figure stops describing anything, while the figure that governs the run — thickness over open share, weighted towards the waist — is not on the sheet.

What a blinded cloth costs in litres

Ruth’s line makes that cost concrete. Take a five-micrometre silica at one per cent solids by mass, pressed at half a bar, through the filter cloth.

Ruth's line for a 5 µm silica through a filter cloth, clean and blinded. Constant-pressure filtration of a one per cent 5 µm silica slurry at half a bar, drawn as time over filtrate against filtrate. Every medium gives a straight line of the same slope, rising 0.25 seconds per litre for every hundred litres a square metre collected, set by the cake, and the medium is the intercept and nothing else: clean 0.000012 seconds per litre, packed to the waist 0.35 seconds per litre, packed through 0.69 seconds per litre. The clean cloth's intercept is indistinguishable from zero; it has been matched by the cake after 2.4 millilitres a square metre. Packed through, the cloth holds its own until 136 litres a square metre, when the cake is 0.86 mm thick. The chart cannot show a cake that compresses, whose line curves upward.
Fig. 5 Ruth’s line for a five-micrometre silica through the filter cloth with three media: clean, packed to the waist and packed through. The slope belongs to the cake and is the same for all three; the media differ only in their intercepts, and the clean cloth’s is too small to see.

The three lines are parallel, because the cake is the same cake. The clean cloth’s line starts at the origin for any purpose a filtration engineer has. The blinded cloths’ lines start well up the axis: 0.35 seconds per litre a square metre packed to the waist and 0.69 packed through, against a slope that adds a quarter of a second for every hundred litres collected. Packed through, the cloth carries as much of the pressure drop as the cake until 136 litres a square metre have come through, by which time the cake on it is 0.86 mm thick, its own L again, since cloth and cake are counted in one currency.

That last coincidence is no coincidence. The volume at which cake and blinded cloth are equal is the volume that deposits a cake of thickness L, so it depends on the solids loading and on nothing about the particles’ size, the pressure or the liquid. A slurry carrying ten times the solids reaches it in a tenth of the volume. A clean cloth would have been matched after the first two and a half millilitres a square metre; the blinded one holds its own for more than fifty thousand times as long.

The threads are a second way through, sometimes

A cloth is more than its holes. At its waist every hole sits in a cell that is seventy per cent thread, and if the threads are spun from staple fibre they are porous: sixty per cent fibre and forty per cent space, the same structure that gives every cloth a floor to its permeability. For the filter cloth that path resists 1.35×10⁹ per metre, which is far above the clean channels and does not matter at all while the holes are open. It is the same split a fabric to fill and a fabric to load found in a reinforcement, a fast scale between the threads and a slow one inside them, with the roles about to reverse.

Once the holes are packed it can matter a great deal. In the second figure the packed holes fall below the threads’ path only for particles coarser than about twenty-five micrometres; below that the threads are the easier route, and at one micrometre they are six hundred times easier. A blinded spun cloth would pass its filtrate through its yarns, if the yarns stayed clear.

They stay clear only for a particle that cannot get into them. The space between fibres fifteen micrometres across, packed at sixty per cent, is about ten micrometres across, so a particle finer than that goes into the yarn with the liquid and packs it too. The second path is open exactly where it is least needed. For particles coarser than ten micrometres it lowers the packed resistance by up to a factor of seven, which is the dip in the second figure’s lowest curve; for anything finer the threads blind with the holes and the cloth is the length L and nothing less.

That is a reason for monofilament filter cloths which is not usually given for them. A monofilament’s thread is solid, so it offers no second path and nothing for fines to blind. A cloth woven from it is the length L from the start and stays it, while a spun cloth is sometimes better than L for coarse duties and, for fine ones, no better, and liable in addition to hold fines inside its yarns where washing reaches them least.

When the cloth stops mattering

Because the packed cloth and the cake on top of it are made of the same thing, the cloth’s share of the pressure drop is simply its length over the total:

share=LL+c,\text{share} = \frac{L}{L + c},

with c the thickness of cake above it.

How much of the pressure a blinded cloth takes as the cake on it grows. The share of the filtration pressure spent across a cloth whose holes are packed through, against the thickness of cake above it, for voile, poplin, filter. Since both are the same cake, the share is L/(L + c) with L the cloth's packed length — voile 0.35 mm, poplin 0.51 mm, filter 0.86 mm — and no particle size, solids loading or pressure enters. Each cloth takes half until the cake is as thick as its own L, and a quarter at three times it. The chart assumes the threads offer no second path, which holds for any particle fine enough to enter them; a clean cloth's curve would fall to a thousandth inside the first micrometre and is not drawn.
Fig. 6 The share of the filtration pressure spent across three blinded cloths as the cake on them grows. Each takes half until the cake is as thick as its own packed length — 0.35 mm for the voile, 0.51 mm for the poplin, 0.86 mm for the filter cloth — and a quarter at three times that.

No particle size, solids loading, pressure or viscosity appears in that share. Every cloth hands over half of the work when its cake reaches L and three quarters at 3L. A press that builds cakes of twenty or thirty millimetres spends nearly all of each cycle with the cloth at a few per cent of the drop, and its throughput is set by the cake. A continuous vacuum filter that discharges at one or two millimetres spends most of each revolution with the cloth carrying a third of the pressure or more, and there the choice of cloth sets the throughput directly. The same cloth is a detail on one machine and the bottleneck on another, and the deciding number is the cake thickness the machine discharges at, compared with the cloth’s own L.

What a cloth for thin cakes should be

For the machines where the cloth matters, the arithmetic reverses some of the usual reasoning.

Thinner beats more open. L is at least the packed depth, so a cloth half as thick starts with half the floor. Opening the sett lowers L too, but through the open share at the waist, and an opener sett also raises the largest hole that the rating is read from. Thickness costs retention nothing, which matters because a filter cloth has two jobs and the sett is already committed to the other one.

Keep the cake out of the waist. A cake that bridges at the face and never enters the channel is worth almost nothing, while one that packs the waist is worth most of the cloth’s L. That is the case for a raised or calendered face whose job is to catch the first particles before they reach the channel, and it is a different job from the one the hair layer was found to do: there the hairs had to catch enough to start a cake at all, here they have to catch it high enough to keep it out of the hole.

Do not read the datasheet’s permeability as the running resistance. It belongs to a state the cloth leaves in its first minutes. The number to ask for is the cloth’s thickness and the open share at its waist, from which L follows.

How the numbers were reached

The hole’s profile is the sampled channel that a hole is a channel, not an opening set up in place of the hole between four threads seen from above: four hundred and one levels through the cloth’s thickness, each with the clear width left by the warp and weft beside the hole at their own heights in a plain weave. The clean resistance is the viscous limit of that channel’s flow at a thousandth of a pascal, where the inertial share of the drop is under a thousandth, with the threads’ own Kozeny–Carman bed in parallel. The packed length is the integral of the reciprocal open share, taken from the face down to a stated depth. The cake’s permeability is Kozeny–Carman for spheres at forty per cent voids, written so that it is the threads’ own law at equal surface, and checked to be. The slurry is a silica of density 2,650 kg/m³ in water at one per cent solids by mass, pressed at half a bar, and Ruth’s line is the standard incompressible form.

Each of the six cloths is at its own named construction in plain weave, the one weave whose holes are all one size, as only a plain weave has one size of hole found; in any other weave L would be a distribution, and its smallest member would carry the flow. Nothing in the result depends on the slurry, which enters only the Ruth figure; the length L depends on the cloth alone.

What the cake arithmetic cannot see

The packed hole is assumed to hold the cake at the cake’s own voidage. A cake that forms inside a narrowing channel may pack more tightly or more loosely than one on an open surface, and L scales with the ratio of the two permeabilities. If the hole packs denser, the blinded cloth is worth more than L; if looser, less.

How far the cake enters is not computed. The two depths drawn — to the waist and through — bracket it, and a real duty sits somewhere between them depending on how the first particles arrive and bridge. Bridging of a channel by particles a twentieth of its width is a question about arching, not about flow, and is not attempted here.

The channel’s profile ignores the crowns. The open share at a level is taken from the four threads around the hole at their transiting heights, which leaves the hole full width at the face and the back. Real crowns intrude there, so L near the faces is slightly understated.

The cake is incompressible and of one size. A compressible cake makes Ruth’s line curve, and a real particle population packs more tightly than one size does, lowering k for the cake and the packed hole alike. Neither changes L, which is a ratio of the two, but both move every resistance quoted here.

Bleed and release are left out. A cloth is chosen for how clear its first filtrate is and for how cleanly its cake lets go, and neither appears in a resistance. A cloth with a small L may bleed more or hold its cake harder; the arithmetic here says nothing about either.

Who found which part

Darcy’s law is from 1856, Kozeny’s relation from 1927 and Carman’s form of it from 1937. Ruth set out the constant-pressure filtration line in 1933 and 1935, with the medium as its intercept, and the medium resistance has been fitted from that intercept ever since; the observation that a cloth’s fitted medium resistance exceeds its clean resistance by orders of magnitude, and that the difference is particles lodged in the cloth, is standard in the filtration literature and is the reason precoats and cloth washing exist.

What is put together here is the length. Written as a thickness of the same cake, a packed hole’s resistance is a property of the hole’s profile alone, ∫dz/φ\int dz/\varphi, which the channel arithmetic already in hand delivers for any construction; it ranks cloths differently from their clean permeability; and it makes the cloth’s share of a filtration a function of the cake’s thickness and of nothing else. Whether a filtration text has written the medium resistance this way is not known here, and the separate behaviour of monofilament and spun cloths is offered as an account, not as a claim of priority.

Still open: whether a wet cloth’s conductance reads its length

The integral ∫dz/φ\int dz/\varphi is not peculiar to flow. It is also the resistance of the same channel filled with a conducting liquid, divided by the liquid’s resistivity: electrical current in a channel of varying section obeys the same slice-by-slice addition as Darcy flow through a packed one. So a clean cloth soaked in brine between two electrodes should read its own blinded length before it has seen a slurry, provided its threads are monofilament and carry no current. A spun cloth adds its yarns’ wet pores in parallel, which is the same second path the filtration arithmetic found, and separating the two would need a cloth measured with its yarns sealed.

That would turn L from a computed quantity into a measured one, cloth by cloth, from a conductivity cell and a micrometer. Whether the measured lengths rank a set of real filter cloths the way their running intercepts do is a comparison against filtration tests that has not been made.

What links here

Computed from the collection rather than written here: the essays that point at this one.

Reads more easily once this is understood

Essays that name this one as worth reading first.

Shares its objects with

Essays naming at least two of the same things, that neither author linked.

Named objects

A flat tag is an object no other essay names yet.

BlindingChannel waistFilter cakeHydraulic radiusMedium resistanceOpen areaPermeabilityRetention