Bleaching Earth Filtration in Edible Oil Refining: Leaf Filter Design and Operation

The short answer

The bleaching earth filter takes spent clay out of hot bleached oil, and an AOCS Lipid Library article lists the horizontal leaf filter and the vertical leaf filter among the processing filters for this step. Clear oil is only half its job. The other half is how much oil goes to the dump inside the spent earth, and the figures published for it run from about 20 to 50 percent, stated on different bases.

Part of that spread is real, because clay, dose, feedstock and purge differ between plants. Part is how the number is stated: the method, the point in the cycle, and what the percentage is taken of. Of the six variables that article names for oil retention, three are wholly or partly set at the filter. The oil left in the cake is also what spent earth heats itself on after discharge, so purge and fire risk belong in the same design review. We build both leaf filter configurations, so read this with that in mind.

Where the filter sits in the bleaching step

Bleaching mixes an adsorbent clay into neutralised or degummed oil, holds it hot under vacuum, and filters the clay out. The AOCS article, by Brooks, Berbesi and Hodgson, gives typical bleaching temperatures of 90 to 125 °C (AOCS Lipid Library). A Vandemoortele patent, citing Bailey's, gives a dose of for example 0.2 to 2.0 percent of the oil (US 5,256,613), and a 2024 review cites 5 to 10 kg of earth per tonne of crude palm oil in Malaysia (Sigauke et al.), which sits inside that range.

Two details from the AOCS article put the filter inside the process rather than after it:

• Contact time includes the filter. The article counts contact time from the slurry tank through the filter, typically 15 to 45 minutes and most often 20 to 30, and notes that excessive contact time can darken the deodorised oil. In our reading, filtration time is part of contact time, so a slow filter is a process variable, not only a throughput one.

• Water shows up at the filter. The article notes that excess moisture slakes the clay and can plug the filter. The cause is upstream; in our reading the symptom shows at the filter as a rising pressure drop.

The AOCS article describes two stages of filter after the bleacher: processing filters (horizontal and vertical leaf filters) and then polishing filters (bag, cartridge, paper). The Perlite Institute's edible oils guide describes the same order: a safety filter after the cake filter, often bags, then cartridges where oil quality demands it (Perlite Institute). The same guide notes that bleached oil filtration on a continuous line uses two filters, one filtering while the other is cleaned or on standby.

How much oil is left in spent bleaching earth?

These are the figures we fetched for this article, each with its source:

• Up to 35 percent by weight as solvent-extractable oil, and up to 50 percent as total organic matter, "of the earth used" (AOCS article).

• 20 to 40 percent by weight (Sigauke et al., review, 2024).

• About 25 to 30 percent (w/w) after bleaching (a 2022 refining-equipment patent, US 12,421,463).

• 30 to 40 percent after the cake is blown with steam (US 5,256,613).

• About 25 percent after steam blowing, about 20 percent after hot water circulation (Ong, JAOCS, 1983, abstract).

• 30 to 50 percent by weight, "thus ... about 50-70% clay" (US 11,517,030).

Read side by side, they look like disagreement. Three things change the number without any change in the cake:

Method. The AOCS article reports oil retention two ways: weight loss on Soxhlet extraction with hexane, and total organic matter by ashing. US 5,256,613 notes that the organics content found by burning off is "usually considerably higher" than the figure from Soxhlet extraction, because solvent extraction "does not lead to complete removal of organics". That is the likely reason the AOCS article's upper figure by ashing (50) sits above its extractable figure (35).

Point in the cycle. A cake sampled before the purge and one sampled after steam or hot water has gone through it are different cakes. Ong's 25 and 20 percent are post-treatment figures; the 2022 patent's 25 to 30 percent describes earth after bleaching with no purge stated.

Basis. Some figures are a share of the cake, some a share of the earth. US 11,517,030 states that its 30 to 50 percent oil leaves 50 to 70 percent clay, so it is a share of the cake. The AOCS figures are "of the earth used". Our arithmetic, taking the cake as oil and earth only (no filter aid or water): a cake that is 30 percent oil by weight carries 0.30 / 0.70 = 0.43 kg of oil per kg of earth. Quoted per kg of earth, the same cake reads 43 percent.

Most of the sources above state neither method nor basis, so these three cannot be applied to each figure after the fact. Before comparing two clays, two filters or two plants on "oil retention", ask for the method, the sampling point and the basis.

For scale, our illustration, not a plant figure: at a 1 percent dose and 0.43 kg of oil per kg of earth, the oil leaving in the spent earth is 4.3 kg per tonne of oil bleached.

Three of the six retention variables sit at the filter

The AOCS article lists what sets total oil retention: clay dosage, clay characteristics (particle size distribution and mineral type), permeability of the filter bed, incoming feedstock quality, cleanliness of the filter screen, and the conditions used to purge the filter before the spent cake is discarded. The refinery chemist owns the first, second and fourth. The other three sit at the filter, one of them shared with the clay.

Bed permeability. The cake on the leaf is the filter medium once it has formed. The article ties its permeability to the clay's particle size, to the filter aid used as precoat, and to the mesh of the filter medium. There is a trade-off in the clay itself: finer grades are more active, but a mineral industry reference on bleaching clays warns that too much fine powder slows the filter rate and can push oil retention high enough that losses in the cake become excessive (911Metallurgist). The AOCS article states the same trade-off between permeability and activity. In our reading, a clay change can raise the pressure drop and the oil loss without any change at the filter, so a filter trial should use the clay the plant will actually buy. For how precoat and body feed build the bed, see precoat vs body feed.

Screen cleanliness. The article lists the cleanliness of the filter screen as a retention variable in its own right, so leaf cleaning between cycles belongs in the operating procedure, not only in maintenance.

Purge conditions. What the cake is blown with, and for how long, is the next section.

Steam blowing the cake: what it recovers and what it does not

When the filtration cycle ends, the cake still holds a large share of oil. US 5,256,613 says the cake is commonly blown with steam to cut oil loss, and US 11,517,030 repeats the same sentence. Two published figures after steam blowing, basis not stated, are about 25 percent (Ong) and 30 to 40 percent (US 5,256,613, which describes a filter press cake). Ong reports hot water circulation reaching about 20 percent. A purge reduces the loss; it does not remove it.

Going much lower needs a separate extraction step:

• Ong found hexane extraction of spent earth economically feasible only for expensive and relatively stable oils, and extraction with an aqueous solution of soda and salt feasible for its simplicity and low cost.

• A Kewalram Oils patent describes solvent extraction of spent earth in agitated tanks and porous metal filters, outside the bleaching filter, reaching about 3 to 5 percent oil in the deoiled earth (US 6,780,321).

• The 2022 patent describes the attempt to do the solvent wash inside a vertical leaf or candle filter and lists what goes wrong: the cake loosens as oil is replaced by solvent, cake thickness stops being uniform, voids form and the wash short-circuits through them, and a filter washed with solvent under pressure risks leakage. It also calls steam desolventising of that washed cake inside the filter questionable, because steam does not distribute evenly through it.

That last point is narrower than it first reads. The patent is describing a cake already loosened by solvent, not ordinary steam blowing of an oily cake. It does not say what follows here, so the general point is ours: a purge medium, like the solvent in the patent, takes the easiest path, so in our reading a blow works best through a cake that is even in thickness from leaf to leaf. That makes cake build-up worth designing for: even feed distribution across the leaves, enough space between leaves for the cake the dose produces, and a purge path that reaches every leaf.

Spent earth heats itself, and the filter affects how much fuel it carries

Spent bleaching earth can self-heat and ignite. US 5,256,613 calls it a fire hazard "because the oil may ignite spontaneously", and says landfilled earth must be covered to keep air out. The 2022 patent says the oil makes spent earth "prone to auto-ignition", so it is treated as flammable waste and its storage and transport raise safety concerns. The 2024 review says the same.

The clearest study we found is Taylor and Jenkins (JAOCS, 1990), who made bleaching clay and oil cakes and tested them for self-heating (abstract). They found a two-step process: a low-temperature heating stage, then a high-temperature stage with charring and, in some cases, flaming combustion. The variables that significantly affected it were clay type, filter cake age, oil retention, moisture and the presence of antioxidant. Oil type was not significant in their tests. US 5,256,613 says the hazard is greatest "when the oil is highly unsaturated". We have only the study's abstract and the patent's sentence, so we report both and do not resolve them.

Three of their variables are touched by the filter. Oil retention and moisture both change with the purge, since steam and hot water add water to the cake, and cake age starts when the cake forms. The abstract does not give the direction of each effect, so in our reading purge and fire risk belong in the same design review rather than being settled by one rule.

On the discharge itself, our selection rule is short: a wet cake discharge suits a horizontal leaf filter, and a vertical leaf filter is designed for dry cake and slurry cake discharge. Which one a refinery wants depends on where the spent earth goes next. For the orientation comparison, see horizontal vs vertical leaf filter.

What to specify for a bleaching earth filter

Send the filter supplier the duty, not a size. Our suggested list:

1. Oil and throughput. Oil type, batch or continuous, tonnes per day.

2. Earth. Clay type and grade, dose, and whether activated carbon or filter aid is added.

3. Temperature and vacuum at the bleacher outlet.

4. Filter arrangement. One filter or two, alternating, on a continuous line.

5. Purge. Steam, hot water or none, and the oil retention you are aiming for, stated with its method and basis.

6. Discharge. Dry, wet or slurry cake, and where it goes.

7. Downstream. The polishing filter after the leaf filter, so the two are sized together.

The type of filter is then decided on the process data. For an application new to us, we offer a trial filtration test before an order.

Our filters, spares and plant capability

Our leaf filters have been used for bleaching earth filtration in edible oil refineries in India and abroad. We build both horizontal and vertical leaf filters for edible oil duty, and on our filters, filter media are selected to suit the application. ~1,000 units installed worldwide across 460+ installations (all applications). Most of them are leaf filters.

If your refinery runs another maker's filter, we supply replacement filter bags, nozzles, leaves, elements, baskets and seals for other makes of filter.

We also design, build and deliver complete edible oil refining plants on a turnkey basis, in the range 5 TPD to 50+ TPD.

Send us your duty

Send the seven points above and we will say whether a horizontal or vertical leaf filter fits your bleaching step, and say so if neither does. Contact our engineers.