From the outside, a separator is the least interesting object on a facility: a steel cylinder lying on its side, doing apparently nothing. But many of the quality problems a facility encounters, whether in its oil, gas, or water, start inside this vessel. Often the first major process vessel after the wells, it supplies all three downstream trains. Its mistakes travel far.

The whole trick is density and patience

The feed arrives as gas, oil, and water together, often with considerable momentum. The first task is a reliable bulk split that gives each downstream train a manageable feed. Density differences provide the driving force, but fluid properties, operating conditions, internals, and residence time determine the result. A separator’s water outlet is not automatically ready for discharge, nor is its oil outlet automatically ready for sale.

The internals serve that basic idea. An inlet device reduces the incoming jet’s momentum so it does not keep remixing the fluids. A settling zone slows the flow and allows the phases to stratify; some designs add coalescing internals. A mist extractor removes entrained droplets from the gas, with performance depending on its design and operating envelope. In a common horizontal arrangement, a weir allows oil to overflow into its own compartment while water leaves from below. Vortex breakers at liquid outlets help prevent gas being drawn into the liquid stream.

Stokes' law, priced in steel

Watch one droplet of water in oil. Gravity pulls it down; drag resists. For a small spherical droplet settling in the low-Reynolds-number regime, their balance is described by Stokes’ law. The consequential part: settling speed scales with the square of droplet diameter. Double the droplet, four times the speed. That is why coalescence, the merging of droplets, is the most valuable thing that can happen inside the vessel, and why heat is a genuine separation tool: viscosity falls fast with temperature, and a separator that worked in summer can quietly struggle in January.

Now connect that to money. A droplet needs time to travel through the continuous liquid phase. At a specified liquid flow, the required retention time helps determine the working liquid volume. The target depends on the fluid, emulsion behaviour, temperature, droplet size, and separation duty; a single range is not a universal design rule. Gas capacity, liquid settling, surge allowance, internals, and geometry then have to work together. Horizontal vessels are common in three-phase service, but the final dimensions come from the duty, not a fixed shape rule.

Three controllers, one frozen picture

A running separator is three feedback loops holding one picture still: a pressure controller on the gas outlet (the fast loop, where a compressor trip arrives here within seconds), a level controller on the oil compartment, and an interface controller holding the oil-water boundary. The interface loop is the honest one: when separation is clean its instrument sees a sharp boundary; when separation degrades, the boundary smears and the instrument starts to guess.

The failure modes will find you

Operators encounter a familiar set of problems. Foam can disrupt level measurement and travel toward the gas outlet. A rag layer (an emulsion band at the oil–water interface) can make interface control unreliable. Sand accumulates and reduces useful volume, requiring a suitable solids-management and inspection approach. Carryover sends liquid toward downstream gas equipment. Gas blowby allows high-pressure gas through a liquid outlet into downstream equipment and must be considered in the protection design. Control, shutdown, and relief provisions follow the assessed scenarios; no single instrument substitutes for that review.

Simple on the outside. Three loops and one stubborn equation on the inside.

I’m a practising oil & gas facilities engineer. If you design or operate these vessels and disagree with anything here, the YouTube comments are exactly where that belongs.

Further reading

• SLB: separator
• SLB: residence time