What Is the Best Method of Separation for Oil and Water: A Selection Guide

Answering the core question: What is the best method of separation for oil and water? There is no single best method; the right one is set by the droplet size of the oil in the water, and everything else follows from that. Free oil, droplets above 150 micrometres, separates by gravity alone in an API 421 separator, reaching 50-150 mg/L. Dispersed oil of 20-150 micrometres needs a plate pack interceptor or a hydrocyclone, reaching 20-60 mg/L. Emulsified oil below 20 micrometres cannot settle at all and requires dissolved air flotation with chemical coagulation, reaching 10-30 mg/L. Dissolved oil below 5 micrometres requires a coalescer or a membrane, reaching below 5-10 mg/L. Selection therefore starts by measuring the droplet size distribution, not by choosing equipment: the governing physics is Stokes law, where the rise velocity scales with the square of droplet diameter, so halving the droplet size quarters the settling rate.

1. The Four Oil States and the Method Each Requires

Oil in water exists in four physically distinct states, and each is separable only by a method matched to it:

• Free Oil: Gravity Separation: Free oil is present as droplets large enough to rise by gravity within a reasonable residence time, typically above 150 micrometres, and it is what an API 421 separator or a simple settling basin removes. Because the rise velocity is proportional to the square of the diameter, a 300 micrometre droplet of 850 kg/m3 oil rises at roughly 12 mm/s while a 75 micrometre droplet manages only 1 mm/s, which is why the cut size defines the vessel. A conventional API basin limits horizontal velocity to 0.9 m/min and keeps the Reynolds number below 2,000, delivering 50-150 mg/L from an inlet of several hundred to a few thousand. This is always the first stage in a treatment train, because it is the cheapest per tonne of oil removed, it tolerates slug loads and solids, and it protects everything downstream.
• Dispersed Oil: Plate Packs and Hydrocyclones: Dispersed oil of 20-150 micrometres settles too slowly for an open basin, so the design shortens the distance the droplet must travel rather than lengthening the time. A corrugated plate interceptor stacks plates 20-40 mm apart at 45-60 degrees, so a droplet rises only to the plate above, coalesces into a film and drains upward to the collection channel; this cuts the required footprint to 20-50% of an API basin and reaches 20-60 mg/L. A liquid-liquid hydrocyclone instead uses centrifugal acceleration, spinning the stream so the heavier water moves to the wall and the oil core is drawn off, in a package a fraction of the size of any gravity device. Hydrocyclones are the choice offshore and where space is at a premium, but they need 200-500 kPa of pressure drop, are intolerant of solids and gas, and lose efficiency at turndown below 40-50%.
• Emulsified Oil: Flotation and Chemical Treatment: Below about 20 micrometres, droplets are stabilised by surfactants, by natural surface-active compounds, or by fine solids at the interface, and gravity will not separate them in any practical vessel. The answer is to change the physics: a coagulant such as ferric chloride or aluminium sulphate neutralises the surface charge, a polyelectrolyte bridges the droplets into larger flocs, and dissolved air flotation then attaches 10-100 micrometre air bubbles that carry the floc to the surface for skimming. Dosing has to be determined by jar testing on the actual stream, because the optimum is narrow and overdosing restabilises the emulsion, making the effluent worse than no treatment. Correctly dosed DAF reaches 10-30 mg/L and is the standard secondary treatment in refineries and petrochemical plants, with induced gas flotation preferred where solids are high or where a saturator is undesirable.
• Dissolved and Polishing: Coalescers, Membranes and Adsorption: Where the discharge limit is very low, or where the water is to be reused, a polishing stage is added. A coalescing filter passes the stream through a fibrous or a specially wetted medium where fine droplets attach, merge and grow until they can be separated by gravity in a downstream chamber, reaching 5-20 mg/L. Ultrafiltration or ceramic membranes reject oil almost completely, giving below 5-10 mg/L and often below 1-5 mg/L, at the cost of a concentrate stream that must be handled and of regular cleaning. Organoclay or activated carbon adsorption removes dissolved hydrocarbons that no mechanical method can touch, and is used as a final guard bed. For shipboard bilge water, the limit is 15 ppm under IMO MEPC.107(49), met with a dedicated separator, a coalescing and filter stage, an oil content monitor and an automatic overboard valve that closes if the limit is exceeded.

2. How to Select: The Decision Sequence

Selection is a sequence of five questions, and answering them in order almost always yields the right train:

• Measure the Droplet Size Distribution First: This single measurement determines everything, yet it is the step most often skipped. A laser diffraction particle size analyser or a simple microscopic count on a fresh, un-sheared sample gives the distribution, and the d50 and d90 values tell you which technology will work. If most of the oil is above 150 micrometres, gravity is sufficient and any further investment is wasted. If the d50 is below 20 micrometres, no gravity device will work at any size and the money must go into chemical conditioning and flotation. Where the sample cannot be taken without shearing, which is common because sampling itself through a valve breaks up droplets, the correct approach is to take the sample at low velocity from a dedicated sample point and to analyse it immediately rather than after transport.
• Check Whether the Emulsion Is Being Created Upstream: Very often the problem is not the separator but the pipework. A centrifugal pump, a throttling control valve, a partially closed valve or a sharp restriction shears free oil into a stable emulsion of 5-20 micrometre droplets that no gravity separator can recover. Before specifying any new equipment, walk the line and identify every shear point between the source and the separator; replacing a centrifugal pump with a progressive cavity or a low-shear screw pump, or replacing a throttling valve with a variable speed drive, frequently improves performance more than any capital project and at a fraction of the cost. Similarly, check that the temperature has not fallen, since water viscosity rises sharply as it cools and the settling rate falls in proportion.
• Define the Discharge Limit and the Duty Profile: The required outlet concentration sets how many stages are needed: 50-150 mg/L needs gravity only, 10-30 mg/L needs flotation, below 10 mg/L needs a coalescer or a membrane, and 15 ppm needs a dedicated oily water separator with monitoring. The duty profile matters as much as the limit. A steady flow is easy; a storm-driven or batch-driven flow with peaks of three to ten times the average requires either a much larger vessel or an equalisation basin upstream, because separators sized on average flow fail during every surge. Slug loads, solids content and the presence of surfactants or cleaning agents all push the design towards a more robust, less optimised configuration.
• Evaluate Footprint, Utilities and Operating Cost: Capital cost is rarely decisive once footprint and utilities are counted. An API basin is cheap per unit of performance but needs a large plot, which may not exist at an existing plant or offshore, so a plate pack or a hydrocyclone wins despite higher capital per cubic metre per hour. Flotation needs chemicals, a saturator pump, a recycle stream and skimmings handling, so it carries a continuing operating cost and a labour requirement that gravity does not. Membranes give the best water quality but need pretreatment, cleaning chemicals, periodic element replacement and a route for the concentrate. The correct comparison is total cost of ownership over the plant life at the actual duty profile, including the cost of disposing of the recovered oil and sludge, which is often the largest single operating item.
• Design the Train, Not the Unit: Real plants use a sequence, and the stages are complementary rather than competitive. A typical refinery train is an API separator for free oil, an equalisation basin to damp surges, dissolved air flotation with chemical dosing for emulsified oil, and a sand filter or a coalescer for polishing, followed by biological treatment for dissolved organics. Each stage is sized for what the previous one leaves behind, and each protects the next from overload. Where the recovered oil has value, the train is designed to keep it clean and water-free rather than merely to remove it, since oily sludge is a disposal cost while recovered oil is a revenue item. And every train needs instrumentation: inlet and outlet oil content monitoring, differential pressure across each element and flow measurement, so that a performance decline is detected and diagnosed rather than discovered at the discharge point.

Oil Water Separation Methods Comparison Matrix

Frequently Asked Questions (FAQ)

Q: Can you separate oil and water by gravity alone?

A: Yes, but only for free oil. Gravity separation works when the oil droplets are large enough to rise to the surface within the residence time of the vessel, which for practical designs means above about 150 micrometres. The physics is Stokes law: the rise velocity is proportional to the square of the droplet diameter and to the density difference, and inversely proportional to the water viscosity. A 300 micrometre droplet of typical mineral oil rises at about 12 mm/s and is easily captured, but a 75 micrometre droplet rises at only 1 mm/s and a 20 micrometre droplet effectively never separates in a real vessel. This is why gravity is always the first stage and why it is always followed by something else when the discharge limit is below about 50 mg/L or when the oil has been emulsified by pumping or by surfactants.

Q: Why does my oil water separator work in summer but fail in winter?

A: Almost always because of viscosity. Water viscosity roughly doubles between 25°C and 5°C, and because the droplet rise velocity is inversely proportional to viscosity, the settling rate halves over the same range. A separator that was designed with only a modest margin, or that is already operating near its limit, will therefore carry oil over as soon as the water cools, even though nothing else has changed. The same mechanism explains why heating a heavy oil feed improves separation: the oil becomes less viscous and the density difference increases. The practical responses are to size the separator using the viscosity at the lowest expected operating temperature rather than at ambient, to insulate or heat the vessel and the inlet line where the climate demands it, and to check whether the wider problem is that the design margin was too small in the first place.

Q: What breaks an oil in water emulsion?

A: Four approaches, in order of increasing intensity. Chemical: a coagulant such as ferric chloride or aluminium sulphate neutralises the surface charge that keeps droplets apart, and a polyelectrolyte bridges them into larger flocs; the dose must be determined by jar testing because it is specific to the water chemistry. Thermal: heating reduces viscosity, increases the density difference and destabilises the interfacial film, which is why some plants heat the feed to 50-70°C. Mechanical: gentle, low-shear mixing after dosing promotes floc growth, while high-shear mixing destroys it, which is why flocculation basins use slow-speed paddles. And electrical: an electrostatic coalescer applies a high-voltage field that polarises and stretches the droplets, rupturing the film and merging them, and is used in crude oil dehydration treating rather than in wastewater service. The common element is that breaking the emulsion is a pre-treatment that then allows a gravity or flotation stage to work.

Q: What is the difference between removing oil and removing dissolved hydrocarbons?

A: Mechanical separation removes oil that exists as a separate phase, whether as free, dispersed or emulsified droplets, and its performance is described by a droplet cut size and an outlet concentration in milligrams per litre. Dissolved hydrocarbons are genuinely in solution at the molecular level, typically below about 5 micrometres and usually measured as total petroleum hydrocarbons or as a specific compound such as benzene; no gravity, flotation or filtration method will remove them. Dissolved hydrocarbons are addressed by stripping with air or steam, by adsorption on activated carbon or organoclay, or by biological treatment in an activated sludge or a membrane bioreactor. This distinction matters because a plant that meets a 10 mg/L oil and grease limit may still fail a dissolved hydrocarbon limit, and because the analytical methods differ: oil and grease by gravimetric or infrared extraction, dissolved hydrocarbons by gas chromatography.