Biodiesel centrifuge for glycerin removal and purification in Indonesia

Source:

Author:

Hui Bo

Release time:

2026.11.18


A biodiesel plant in Indonesia lives or dies on what happens in the few hours after the transesterification reactor. The reaction itself is forgiving. The separation that follows is not: glycerin, methanol, soap, catalyst and wash water all have to leave the methyl ester stream, and the plant either meets its product limits or it recycles off-spec batches through the settlers again. A biodiesel centrifuge is the machine that turns that slow, weather-dependent gravity step into a controlled one, and it is now standard thinking in new lines and in debottlenecking projects across Sumatra and Kalimantan.

This guide is written for process and plant engineers who already know what FAME is and want the separation side explained properly: where centrifugal separation fits after transesterification, why glycerin removal by gravity alone stalls, how water wash and soap carry-over behave, how palm oil, refining by-products and used cooking oil change the machine's job, and how to choose between a three-phase decanter and a disc machine. Palm oil clarification is covered separately in the three-phase decanter article for palm oil mills, so it is not repeated here.

biodiesel centrifuge

Where a biodiesel centrifuge sits after transesterification

Most continuous plants run one or two reactor stages, each followed by a separation. Oil, methanol and an alkaline catalyst (usually sodium or potassium methylate or hydroxide) go in; what comes out is a two-liquid mixture. The light phase is methyl ester, still carrying dissolved methanol, traces of free glycerin, some soap, and residual catalyst. The heavy phase is crude glycerin, which holds most of the excess methanol, nearly all of the catalyst, and a good share of the soap. Because the reaction is driven with excess methanol, the heavy phase is far from pure glycerin; it behaves more like a viscous, alkaline, soapy methanol solution.

A biodiesel centrifuge can be placed at three points in that chain, and it helps to name them before discussing machines:

  • Glycerin split after each reactor stage. The mixture is fed to the machine, glycerin phase leaves through the heavy-liquid outlet and ester phase through the light-liquid outlet. Any solids (catalyst salts, polymerised gums, fines from low-grade feedstock) leave separately or with the heavy phase.
  • Wash-water removal. After neutralisation and water washing, the ester is mixed with water and must be split again. This is a liquid-liquid task with very little solid, and the emulsion risk is highest here.
  • Polishing and dehydration. A final pass before vacuum drying or storage removes entrained water droplets and haze, so the dryer is not asked to remove free water.

The table shows how the main components distribute after the reactor.

ComponentApproximate density (g/cm³)Where it mostly ends up after reactionWhat it does to separation
Methyl ester (FAME)0.87–0.89Light phaseThe product; carries dissolved methanol and traces of glycerin
Crude glycerin1.2–1.26 when pure; lower with methanol and soapHeavy phaseDensity difference with ester is large, so the split is physically easy once drops are big enough
Methanolabout 0.79Mostly heavy phase; some stays in the esterLowers heavy-phase density and viscosity; acts as a co-solvent that holds emulsions together
Soapabout 1.0–1.1 as a hydrated phaseMostly heavy phase; some at the interface or in the esterThe main emulsifier; stabilises small glycerin drops in the ester
Wash waterabout 1.0Separate phase after washingDensity difference with ester is modest, so drop size and temperature matter more

Typical Industry Reference Values for Biodiesel Liquid-Liquid Separation

Reaction and separation temperatureroughly 50–60 °C, below the boiling point of methanol (about 65 °C)
Ester / glycerin density differenceabout 0.3–0.4 g/cm³
Ester / wash water density differenceabout 0.1–0.13 g/cm³
Disc machine separation factorseveral thousand g, up to roughly 12,000 g
Water limit in finished ester (common European specification)500 mg/kg maximum
Heavy phase share of reactor outletvaries with methanol ratio; commonly on the order of one tenth to one fifth by mass

Why gravity settling of glycerin reaches a ceiling in Indonesian plants

Glycerin is roughly 40 percent denser than the ester, and on paper that makes a settling tank look sufficient. Stokes' relation explains why it often is not. The settling velocity of a drop scales with the density difference and with the square of the drop diameter, and it is inversely proportional to the viscosity of the continuous phase. The density difference here is generous. The drop size is the problem. Straight after a stirred reactor the glycerin is dispersed as droplets, and a large fraction of them sit in the range of a few micrometres to a few tens of micrometres. Halving the drop diameter cuts the settling rate to a quarter, so the finest fraction needs many hours, and the finest of all never settles in a realistic tank time.

Three practical things follow from that in the workshop:

  1. The last glycerin is the problem. A settler removes the bulk heavy phase easily and then slows dramatically. Operators see a hazy ester above a clean interface and decide the tank is "done" when the clock says so, not when the physics says so. Free glycerin and soap then travel downstream into the wash.
  2. Tank time has to be built in, and it varies. A plant that settles overnight gets a different product on a cool, rainy night than on a warm, dry one, because temperature moves both viscosity and drop coalescence. At a humid Indonesian site, an open-air settler's shell temperature swings through the day even when the reactor outlet is held at a set value.
  3. Settling tanks hold a lot of inventory. That is a safety point as much as a process point, because every cubic metre standing in a tank is a cubic metre of methanol-bearing liquid. A centrifuge replaces hours of residence with seconds to minutes inside a rotating bowl.
Workshop rule of thumb: If the ester leaving a settler is still visibly hazy after the planned residence time, the haze is almost always a mix of micro-droplets of glycerin and soap micelles, not "dirt". Longer settling helps slowly and unreliably. A centrifugal field of several thousand g shortens the same separation by orders of magnitude, but only if the feed is not already a stable emulsion, which is why upstream conditions are discussed further below.

Glycerin separation by centrifuge: what actually controls the split

When the mixture enters a rotating bowl, the two liquids stratify into concentric layers: glycerin against the bowl wall, ester nearer the axis. What the operator controls is where the boundary between them sits and how long the liquids stay inside. Four variables dominate the result.

Temperature

Transesterification usually runs warm but below the boiling point of methanol, and the separation tends to run in the same range. Higher temperature lowers ester viscosity and speeds drop movement, but it raises methanol vapour pressure. In Indonesia, plants that heat the mixture too hard to "help" a lazy split usually find that methanol losses and vent odour rise while glycerin carry-over barely changes.

Methanol content of the feed

Methanol dissolves in both phases and acts as a bridging solvent. A feed with a large share of unrecovered methanol has a lighter, thinner heavy phase, a smaller density difference, and a stronger tendency to hold emulsion.

Interface position

The interface has to sit between the two outlets, and it moves whenever methanol content, temperature or feedstock changes. The adjustment mechanics are described in the palm oil clarification article; for biodiesel the point is simply that a set-up proven on fresh palm oil should be re-checked on the first batch of a different oil.

Residence time and feed rate

Pushing the feed rate up shortens residence and brings finer glycerin droplets into the ester outlet. The signs are a rising free-glycerin result in the lab or a haze that reappears after the wash. Feed-rate steps should follow product analysis, not the reverse.

For a plain glycerin split the interface is usually easy to hold. The work moves to the feed: stable temperature, a stable methanol ratio, and a short coalescence time in a pipe or small tank before the centrifuge.

Water wash and dry wash: where wash-water separation fits

After the glycerin split, the ester still holds residual methanol, catalyst, soap, and traces of glycerin. Most plants neutralise with a mild acid and then wash with water, typically warm, and often in two or three countercurrent stages. Each stage is a mixing step followed by a separation, and that separation is where many biodiesel plants run into the worst trouble, because soap in the ester acts as a surfactant and the mixing gives it the surface it needs.

The wash-water split is harder than the glycerin split in three specific ways. The density difference between ester and water is small compared to ester and glycerin. The water phase is the minor phase in volume, so only a thin heavy layer is present and the interface is sensitive. And the mixture is an emulsion by construction, because it has just been mixed on purpose. A centrifuge in this position does not remove emulsion-forming conditions; it gives the drops enough force to separate before they have time to merge into an unmanageable rag layer.

Separation stepPhases presentTypical difficultyWhat helps
Glycerin split after reactorEster / glycerin phase (with methanol and soap) / occasional solidsLarge density difference, but fine drops and soap at the interfaceStable temperature, controlled methanol level, short coalescence time before the machine
First water wash separationEster / soapy wash waterSoap-stabilised emulsion; wash water becomes cloudy and alkalineAcid neutralisation before washing, gentle mixing, warm wash water
Final polishingEster with trace waterTrace water that dryers should not have to handleFine separation with small solids load, clean feed

Soap, emulsions and carry-over: reading the symptoms

Soap is formed whenever the alkaline catalyst meets free fatty acid or water. Once present, it migrates to every interface and stabilises drops. Most separation trouble in a biodiesel plant is soap in disguise, so it pays to learn the symptoms by sight and by lab result.

Symptom on the plantLikely causeFirst thing to checkUsual remedy
Ester outlet hazy, clears slowly on standingFine glycerin or water droplets held by soapSoap content of the ester before the wash; temperature in the feed lineRaise acid neutralisation accuracy, trim feed rate, hold temperature steady
Ragged third layer between ester and heavy phaseSoap-rich emulsion collecting at the interfaceFree fatty acid of the oil; catalyst dose; water in the methanolCorrect upstream chemistry first; then re-set the interface
Thin, pale glycerin with ester lossesInterface too close to the heavy outletInterface setting; methanol content of feedMove the interface toward the axis side or change back pressure; re-check after the methanol level stabilises
Wash water stays cloudy after the machineEster droplets held in the waterMixing intensity in the wash; water temperatureReduce shear in the wash mixer; slow the feed to the centrifuge
Two limits to keep in mind: a centrifuge removes separate phases, not dissolved material. Dissolved methanol and glycerin in the ester come out by washing and by vacuum drying, not by spinning faster. And no machine fixes a feed that was converted badly; high free fatty acid in the oil, wet methanol, or a catalyst overdose will give a soapy product regardless of which separator follows.

Feedstock differences for Indonesian producers: palm oil, refining by-products and used cooking oil

The same centrifuge gets a different job depending on what went into the reactor. The three feedstock families seen in Indonesian plants shift the load in recognisable ways.

Refined palm oil and fractions

Refined, bleached and deodorised palm oil gives the cleanest conversion: low free fatty acid, modest soap, a relatively clean glycerin phase. The main concerns are saturated fractions that can give a hazy ester in cool conditions, and interface stability when the plant switches oil grades.

Refining by-products and high free fatty acid streams

Indonesian refiners that take palm fatty acid distillate, acid oil, or soapstock-derived fatty material as feed face a different chemistry. High free fatty acid means either an acid-catalysed esterification stage first, or a heavy soap load in a single alkaline step. In both cases the separation after each stage matters more, because water is generated by esterification and the intermediate mix carries a strongly changed water and methanol balance. Solids are also more likely: gums, polymerised material and fines may appear in the feed. That raises the question of whether the machine must handle solids in the heavy phase. A decanter can discharge them continuously; a disc machine with intermittent discharge needs a plan for how often to release solids.

Used cooking oil

Used cooking oil is the least predictable feed. Free fatty acid, water, polymerised material, food fines, and degraded triglycerides vary from one collection batch to the next. Plants processing it pre-treat heavily by settling, filtering, heating and often dehydration, and the centrifuge task after the reactor is typically harder than with refined oil: more soap, more solids, and a heavy phase that is darker and thicker. The practical effect is that interface and flow settings should be treated as a range of recipes, not a single set-and-forget value.

FeedstockFree fatty acid and water levelEffect on soap and emulsionEffect on centrifuge task
Refined palm oil / oleinLowLow soap, clean heavy phaseEasiest job; keep temperature steady to avoid wax haze
Palm fatty acid distillate, acid oilHighEsterification water plus soap if alkaline route is usedSeparation after each stage matters; solids handling and discharge plan needed
Used cooking oilVariable, often highHigh soap, variable emulsionsWidest swing in feed behaviour; settings need a documented range

Sampling and lab checks that show the split is working

A centrifuge on a biodiesel line is only as good as the feedback around it. Plants that run steadily for weeks tend to have a short, fixed sampling routine, and the useful samples are taken at the machine outlets rather than in the product tank, where several hours of production have already blended together.

  • Ester outlet after the glycerin split. Free glycerin and soap content tell whether the split is clean before the wash is loaded with it.
  • Heavy-phase outlet. Colour, thickness and methanol content show whether ester is being lost with the glycerin. A heavy phase that has turned thin and pale is the usual warning that the interface has moved.
  • Wash-water outlet. Alkalinity of the discharged water shows whether neutralisation is keeping up. Water that leaves cloudy points to ester droplets being carried out.
  • Ester after polishing. Water by Karl Fischer titration and a haze check after cooling to ambient temperature, which in an Indonesian workshop is a gentler test than in a cold-climate plant but still catches wet ester.

Logging these results against feed rate, temperature and feedstock batch builds the range of settings that actually works, and most off-spec results then trace back to upstream events such as a wet methanol delivery.

Biodiesel plants in Indonesia: site conditions and operating realities

Indonesia's national policy of mandatory blending, with the blend proportion raised step by step over the years, has made biodiesel a steady base-load product for palm refiners rather than an occasional side line. In Indonesia, many plants therefore run closer to continuous operation than their early designs assumed, and separation equipment that was acceptable at part load is tested at full load. That is where a centrifuge conversation usually starts in practice: the settlers were sized for a lower output, and the plant is now asked to run for long periods without stopping.

Plants cluster where refined palm products, deepwater or river ports, and bulk storage are close together. On Sumatra, the east coast around Riau and North Sumatra is the obvious example; on Kalimantan, plants sit near river ports and the major plantation belts. Java hosts several units near the Jakarta and Surabaya industrial areas, where marine terminals and power utilities are close.

Heat, humidity and rain

Ambient temperatures in these regions seldom fall below the range where palm methyl esters stay clear and liquid, which spares some wax-haze trouble, but that does not make the separation easier by itself. High humidity matters in three ways. Methanol is hygroscopic, and storage and transfer lines that breathe humid air bring water into the reaction, which makes soap and shifts the interface. Heavy tropical downpours change the temperature of outdoor tanks and pipes within minutes, which shows up as a sudden change in viscosity and in the machine's separation behaviour. And washing water drawn from surface sources may carry hardness or suspended material that affects soap behaviour during the wash.

Feedstock logistics

Plants fed by nearby refineries receive a consistent product; plants taking by-product streams or collected used oil see variation from truckload to truckload. Specify the machine for the harder case and log which settings worked for which batch.

Three-phase decanter or disc machine: choosing the configuration for Indonesian biodiesel plants

Two machine families are used for biodiesel liquid-liquid separation. The right choice depends on the solids load, the step in the line, and how much emulsion the plant has to live with. The internal mechanics of the three-phase decanter are explained in the palm oil three-phase decanter guide, and the disc stack, interface rings and discharge types are covered in the disc-type oil-water separator article. Here the focus is purely on how each fits the biodiesel line.

Three-phase decanter

A horizontal three-phase decanter has a second liquid outlet in addition to the solids discharge, so it splits light liquid, heavy liquid and solids in one pass. Its strength is tolerance of gum, catalyst salt and fines, which makes it a natural fit for the glycerin split with variable feedstocks. Its limit is moderate centrifugal force, so very fine droplets in the last wash stages may stay in the light phase.

Disc stack machine

A disc stack machine reaches much higher separation forces and splits fine droplets better, which suits wash-water removal and final polishing when solids are low. Its limit is solids: intermittent discharge needs periodic release, and unfiltered feed blocks it quickly.

QuestionThree-phase decanterDisc separator
Best position in the lineGlycerin split; feed with solids or variable qualityWash-water removal; final polishing
Solids toleranceHigher; continuous solids conveyingLow to moderate; needs discharge plan
Fine droplet separationModerateBetter, because of higher separation force
Sensitivity to feed variationLowerHigher; clean, steady feed pays off
Common arrangementSingle machine for the split in smaller plantsSecond machine after the wash in plants that need low water in the ester

Pre-treatment ahead of either machine

Whichever machine is chosen, its performance is set largely upstream. A strainer catches rags and agglomerates that would block feed ports. A short residence tube lets micro-droplets merge. Temperature control keeps the feed in the range the interface was set for, and gentle feed pumps avoid creating fine droplets that the centrifuge must then remove again. Most plants prefer separate machines for the glycerin split and the wash, because the setting changes between the two tasks are large and a shared machine needs a thorough flush at every changeover.

Polishing the ester: water content and haze before storage

The last separation step defines what the quality laboratory sees. A common European specification for biodiesel limits water to a few hundred milligrams per kilogram, and total glycerol is limited to a fraction of a percent by mass; the plant's own product standard may differ, so the numbers should be taken from the applicable specification rather than from a general guide. What matters for the centrifuge is that free water droplets are removed mechanically before the ester goes to vacuum drying or storage, and that nothing in the line re-wets it.

Plants often discover water problems by a seasonal pattern. In the wet season, ester stored in partly vented tanks absorbs moisture and the lab result drifts upward. A polishing centrifuge removes free droplets but cannot stop a tank from breathing, so breathers or nitrogen blanketing on storage matter as much as the machine.

Methanol and safety around a biodiesel centrifuge

Methanol is flammable, its vapour is heavier than air, and the liquid is toxic. Any machine that handles methanol-bearing streams sits inside the hazardous-area planning of the plant, and the centrifuge adds its own considerations: rotating equipment, sealing, and a liquid flow that may be warm. These are general engineering points, and the plant's own hazard study and the applicable local regulations decide the specifics.

  • Area classification and equipment rating. Motors, sensors and control panels near methanol-bearing equipment need the protection class the area classification calls for.
  • Closed handling. Enclosed feed and discharge lines and a vent route to a safe location limit vapour release into the workshop, which in a hot, humid building otherwise lingers at floor level.
  • Inert gas. Some plants blanket the machine or its receiving tanks with nitrogen to keep oxygen out of the vapour space, especially during start-up and shutdown.
  • Shutdown and cleaning. Methanol-bearing liquid should be flushed out before opening a machine, and wash liquid should go to a controlled drain, not a trench.

Huibo LWS decanter and DBX-3 disc separator for biodiesel plants in Indonesia

Two Huibo machine families fit the separation steps described above.

LWS three-phase horizontal decanter. Biodiesel refining is one of the applications this series is made for. It adds a second liquid outlet to the standard decanter, so it separates liquid-liquid-solid mixtures as long as the phases differ enough in density, and a guide disc sets the liquid-liquid interface. On a plant that makes ester from palm feedstock, this is the machine for the glycerin split after the reactor, where a heavy phase with some solids has to leave continuously.

DBX-3 disc separator. A liquid-liquid-solid machine for oil, water and small amounts of solids, with automatic intermittent solids discharge. The disc range covers drum diameters of roughly 330 to 785 mm. The DBX-3 suits the wash-water removal and polishing steps, where the flow is mostly ester carrying a little water, and its stainless steel contact parts and PLC-controlled discharge keep operation consistent from one shift to the next.

To start, send the reactor outlet and washed ester data (flow, temperature, glycerin and water content), and the technical team will recommend between the LWS decanter and the DBX-3 disc separator.

Frequently asked questions

Why not rely on gravity settling for glycerin removal?
Gravity settling works for the bulk of the glycerin and then slows sharply, because settling speed falls with the square of the droplet diameter and the fine droplets formed in the reactor are very small. The result is a hazy ester above a clean interface, and the haze carries glycerin and soap into the wash. Settling time also varies with tank temperature, so a plant in Indonesia can see different ester quality between a hot afternoon and a cool, rainy night. Large settling tanks also hold a lot of methanol-bearing liquid. A centrifuge applies several thousand g, which shortens the separation from hours to minutes and makes the result less dependent on weather. Settlers are still useful as a first stage before a centrifuge in many plants.
What causes emulsions and soap carry-over in biodiesel?
Soap forms when the alkaline catalyst meets free fatty acid or water, and it acts as a surfactant that stabilises droplets of glycerin in ester or ester in water. High free fatty acid oils, wet methanol, excess catalyst, and strong mixing in the wash all raise the risk. Symptoms are a hazy ester, a rag layer at the interface, and cloudy wash water. A centrifuge helps by separating drops quickly, but it cannot remove soap that is dissolved or held in micelles. The reliable fixes are upstream: control free fatty acid, keep methanol dry, neutralise accurately before washing, and mix gently.
Where does the water wash separation fit in a centrifuge-based line?
It follows neutralisation, once the ester has been mixed with warm water to extract residual methanol, catalyst, soap and glycerin. The mixture is an engineered emulsion, so the centrifuge there faces a harder liquid-liquid split than the glycerin step, with a smaller density difference and the water as the minor phase. Many plants use two or three wash stages, each with its own separation, and a disc machine is often chosen for the later stages where solids are low and fine droplets must come out. Some plants wash and centrifuge only once if the upstream split was clean. Dry wash with adsorbents removes this step altogether but adds a filtration stage.
Does used cooking oil feedstock change the machine job?
Yes. Used cooking oil varies in free fatty acid, water, polymerised material and food fines, so the reaction produces more soap, more solids and a darker, thicker heavy phase than refined palm oil does. The centrifuge must therefore tolerate solids and a less stable interface, which usually favours a three-phase decanter at the glycerin split, with a disc machine reserved for polishing after the wash. Interface and feed-rate settings should be recorded as a range linked to the batch quality instead of one fixed value. In Indonesia, where used oil may come from several collectors, this variation from load to load is the main operating difficulty.
Can one centrifuge handle both the glycerin split and the water wash?
Technically a single machine can do both on different runs, but the two tasks are different enough that the changeover is significant. The glycerin split has a large density difference and possible solids; the wash split has a small density difference and a high emulsion risk. The interface position, back pressure and feed rate all change, and residues of one stream contaminate the next unless the machine is flushed thoroughly. Smaller plants sometimes share one machine with a written changeover procedure, while continuous plants usually dedicate one machine to each step so that settings stay stable for weeks at a time.
What safety points apply when methanol is present?
Methanol is flammable, toxic, and gives off vapour heavier than air, so equipment must follow the plant's hazardous-area classification, including motor and sensor protection ratings. Closed feed and discharge lines with a vent to a safe location prevent vapour building up at floor level, a real concern in warm, humid Indonesian workshops with limited air movement. Nitrogen blanketing, bonding and earthing against static, and a documented purge before opening any machine are standard practice. Seals should be checked routinely, since a small leak in an enclosed space is both a fire and an exposure risk.