Flare Network Sizing and API 521 Blowdown

Introduction

Sizing a single PSV is a contained problem with a clear answer. Sizing the flare network that carries every relief load to a safe disposal point is not — it is a system-level exercise governed by API Standard 521 (Pressure-relieving and Depressuring Systems). The decisions made here drive the diameter of every relief sub-header, the size and orientation of the knock-out (KO) drum, the height of the flare stack, and ultimately the radiation exclusion zone around the unit.

This post walks through the logic of flare network design — contingency selection, simultaneous relief, blowdown sizing per the 50%-in-15-minutes rule, and the backpressure check that ties it all together.

Step 1 — Define Every Credible Relief Contingency

API 521 Section 4 lists the canonical overpressure scenarios. For each protected piece of equipment, the engineer asks: what credible upset could push pressure above MAWP, and at what rate?

Common contingencies:

• External fire (pool fire engulfment) — drives latent heat input through wetted area
• Blocked outlet — closed valve while inlet flow continues
• Tube rupture in a heat exchanger — high-pressure side breaches into low-pressure side
• Control valve failure — full open or full closed in the failure mode
• Reflux failure in a column
• Loss of cooling or loss of utility (instrument air, power)
• Thermal expansion of a blocked-in liquid

Each contingency produces a relief rate, a relief temperature, and a fluid composition. The PSV is sized for the governing case for that vessel — but the flare network must handle the simultaneous combination of all PSVs that could lift together under one initiating event.

Step 2 — Simultaneous Relief Logic

This is where many designs go wrong. You do not simply sum every PSV in the unit at its rated capacity. You ask: what single root-cause event would lift these valves at the same time?

API 521 Section 5.3 frames it as a tree of common-cause events:

The governing global contingency — usually general power failure or cooling water failure for unit-level studies — sets the design flare load. The fire case is evaluated zone by zone and rarely governs the main header (it governs the local sub-header and the KO drum liquid handling).

A common mistake is double-counting: summing the fire-case relief from a vessel and its blocked-outlet relief in the same scenario. They do not occur together — pick the larger.

Step 3 — Sizing the Flare Header

With the design relief load defined, the flare header is sized for the worst-case mass flow at the worst-case temperature and composition. Two constraints govern:

Mach Number Limit

API 521 recommends keeping flare header velocity below Mach 0.7 at peak relief flow, with Mach 0.5 as a more conservative target for long headers. Above this, the noise, vibration, and erosion risks rise sharply, and the assumption of incompressible-style pressure-drop calculations breaks down.

Backpressure Limit

The total backpressure at any PSV outlet — built-up plus superimposed — must stay within the valve's tolerance:

• Conventional spring-loaded PSVs: built-up backpressure ≤ 10% of set pressure
• Balanced bellows PSVs: built-up backpressure typically ≤ 30–50% of set pressure
• Pilot-operated PSVs: tolerant of higher backpressure but check the vendor curve

Backpressure is calculated by walking the network from the flare tip upstream, summing line losses at the design simultaneous flow. If the calculation shows a 12% backpressure on a conventional PSV, the choices are: increase the header diameter, or change the valve type to balanced bellows.

Step 4 — Knock-Out Drum

The KO drum sits between the flare header and the stack. Its job: remove any liquid carried over with the relieving vapour, so droplets do not reach the flare tip and rain burning hydrocarbon downwind.

Sizing rules (API 521 Section 7.3):

• Drop-out velocity calculated for a 300–600 micron droplet (depending on tip type)
• Liquid hold-up volume sized for 20–30 minutes of the largest single liquid relief, or 10 minutes of the simultaneous liquid relief — whichever governs
• Horizontal drum is the standard configuration for offshore and most onshore plants

The drum should never run liquid-full. A high-level alarm + automatic transfer pump is standard, with manual fallback.

Step 5 — Emergency Depressurisation (Blowdown)

This is the part that most surprises people new to relief design. API 521 Section 4.6 introduces a separate requirement beyond PSV protection: for vessels in fire-exposed service, the operator must be able to reduce vessel pressure to 50% of design pressure within 15 minutes of the fire being detected.

Why this rule? Because vessel wall material loses strength rapidly under fire exposure. By the time the wall metal reaches roughly 593°C (1100°F) — typical at 15 minutes of pool-fire engulfment — carbon steel retains only about half its room-temperature yield strength. Holding the vessel at MAWP at that point risks rupture before the PSV can prevent it. Depressurising to ~50% MAWP gives the wall an additional safety margin.

Sizing the blowdown valve and orifice:

Required orifice diameter is iterated such that the integrated
mass flow over 15 minutes brings vessel pressure from initial
(typically PSV set or operating) to 50% of design.

The calculation is dynamic — pressure, temperature, density, and choked-flow status all change as the vessel depressurises. A spreadsheet or dynamic process simulation model is the practical tool. The blowdown line discharges into the same flare header, so its peak flow must be checked against the simultaneous-relief envelope.

Step 6 — Flare Stack Sizing

With the simultaneous load and tip exit Mach known, stack height and exclusion zone follow from radiation calculations (Brzustowski, Simple, or detailed CFD). The accepted radiation limits at grade per API 521 Table 5:

The 1500 BTU/hr·ft² contour typically defines the manned-area exclusion radius around the flare base.

Common Pitfalls

• Summing relief loads without a common-cause event. Always identify the initiating event first; the simultaneous list follows from that, not from arithmetic.
• Using PSV rated capacity instead of required capacity for header sizing. Rated is what API 526 standard orifices deliver; required is what the contingency calls for. The difference can be 30–50%.
• Ignoring the blowdown contribution to the flare header. Operators may open BDV and PSV simultaneously during a fire — the network must handle both.
• Designing for one phase only. Multi-phase relief (gas + liquid carryover) changes the velocity profile and KO drum sizing materially.
• Skipping the dynamic depressurisation check. A static "average flow over 15 minutes" calculation will underpredict the orifice required, because the flow tails off as pressure drops.

Conclusion

A flare network is the physical reconciliation of every protective device in the unit. PSV sizing is the input; the network design is where contingency logic, simultaneous relief, depressurisation, and disposal converge.

Done well, the network has consistent backpressure margin, a KO drum that handles the credible liquid load, and a stack that meets radiation limits with the right tip selection. Done poorly, it shows up as PSV chatter, liquid carryover at the tip, or a unit that cannot meet the API 521 depressurisation requirement during the very fire scenario it was designed for.

The discipline is system-level thinking: every PSV, every blowdown valve, and every disposal line as one network — not a stack of individual sizing exercises.