Design Tips 14

Flare

• Radiation - Wind Speed: Use 30 kmph as in API 521 sample. Some wrongly go for higher wind speeds, even gale speed specified for flare stack structure or flare pilots to stay lift. High speed winds cool and reduce surface temperature - high air heat transfer coefft. Ref: ‘Flare Radiation Flux Eqm Temperature Chart’ in Sizing. At 6.3 kW/sq.m, steel surface temperature is 175°C at 30 kmph and 270°C at 0 kmph. Flux with 30 kmph has to double to match still air temperature. Higher wind speeds may flatten flame but cool surfaces. Gale winds AND flare full blast AND an operator in open - zero chance

• Stack Height: Decided by tip velocity for radiation and dispersion. Minimum statutory heights in different countries, regardless of exit velocity. High tip velocity = High tip ΔP = Smaller tip = Less money on tip. High tip ΔP = high back pressure that reduces flow volume/ stack/ header diameters. Do not give all ΔP to tip or tail pipes. Equal ΔP/ meter run. In 2-3 iterations find best split - tip Vs header size. (1) Start with 30% ΔP for tip. In some plants, headers are longer. Factor accordingly. (2) ΔP/m is based on straight length and not equivalent length. For existing plant, eye-balled run lengths and correct count of bends + fittings will do. Bends + fittings take bulk of ΔP

• Sonic Flare: API 521 method and flame lengths are for sub-sonic pipe flares. There is no published or open literature correlation to estimate the length of well aerated sonic flame. Since flame envelope is an indication of air: HC mixture in LEL-HEL range, some estimate flame length based on dispersion calcs. Vendor literature shows good variation between measured and estimated lengths. Field feedback suggests that actual radiation is higher than predicted (“Tie the engineer who predicted to a pole at the foot of the flare bridge”) due to liquid carry over. Use API/ Flaresim to check supplier recommendations. API predicts longer flame; flame centre closer to observer. In shorter sonic flames, flame centre is away from observer. API flux = F*Heat output/Surface of sphere of diameter D. Do not buy supplier’s suggestion that in a multi-tip sonic flare, flames far away is shielded by the flames near. Radiation is from a hot gas cloud

• Blowdown: ESD isolation to limit HC inventory to 50 m³. BDVs based on PVgas > 100 bar. m³ or 4 m³ of LPG= inventory

• Blowdown: Vessels with HP gas or Oil don’t cooldown. Vessels with light ends and condensate may reach MDMT

• Blowdown: API - pressure brought down to 50%DP with thicker walls or to 100 psig in 15 minutes. Compressors with seal oil may require a faster rate. Safety may want it faster. Analyzing prevailing pressure + allowable stress at prevailing metal temperature, one can extend blowdown time

• Blowdown Rate Reduction: If peak blowdown > inflow, don’t waste money providing a bigger flare. Consider extended/ zoned/ staggered blowdown or via dynamic simulation or HIPPS to reduce load. See ‘Relief Systems - Sizing’ in Training

• Blowdown: BDV ROs selected on next standard size. Vessel may blow down faster with higher initial rates than 15 minutes assumed. All vessels are not at BDV rated pressure when blowdown is initiated. Considering ± errors in correlations/ calculations, it does not matter if a vessel is blown in 13 or 17 minutes. Due to header packing, tip gets less than total load. Exact 15 minutes precision run is not called for except to harass a junior engineer performing the calcs! Don’t sweat the small stuff. It takes attention away from key issues. Relief calcs were done in pre-Hysys and Flarenet days too. Plants were designed on Grote’s simple blowdown equation. Availability of tools instead of saving calculation time shouldn’t exponentially increase it. During reviews, where there is an obsessive attention to minor issues, usually there is always a big hole. Unlike downstream industries with known feed/ products/ properties, in oil & gas there are too many unknowns - in flow; composition etc. The wellfluid that flows in, is different from the ones on which multiple simulations for many future years were done. Seen cases where 15-year analyses were done but the field was dry after installation. One platform got shrunk to half capacity halfway thru the project and finally got mothballed. Spend time on transient and start-up issues that are not covered in steady state simulation

• BDV: 50% back pressure assumed for choked flow that gives the smallest orifice size and lowest flow. Blowdown flow rapidly declines and backpressure falls as square of the declining flow

• BDV RO: Single RO will do for this short duration service. No need for multiple ROs

• Load Summary: Tabulate loads for all scenarios. (1) Single Relief: Blocked outlet, Control valve failure, Gas Blowby, HX Tube Rupture, Reflux Failure, Excess Heat to Reboiler etc (2) Multiple Sources: Power, Cooling Water or Instrument Air Failure and for blowdown valves. See ‘Relief and Blowdown Summary’ in Sizing. API: It is not the largest mass flow but volume flow (low mol weight) that could govern sizing. Transient or declining load like blow down should not decide flare size. Wasted money. You don't have to route LP BDVs to LP header; can route to HP header too if capacity is available and vice versa for HP BDVs. Similarly, a single LP RV contingency like Gas Blowby can be routed to HP Flare

• Liquid Full Vessel Under Fire: Can burst under thermal expansion. C3 and light ends can lead to BLEVE. It will take long to vapor depressurize due to large vapor generation. Drain (or liquid blowdown) inventory in a fire zone with Remote Operated Valve (ROV) to unit blowdown drum or Flare KOD. Add an electrical heater to avoid LT issues and switch it on during liquid removal

• Purge Gas: Air ingress via (1) Open pipe LP tip (2) Vacuum pulled when a hot discharge cools or condenses or (3) Lighter than air purge gas fills stack; buoyancy effect. Purging maintains a positive pressure in flare header + stack

• Purge Gas - Tip: Wind eddies downwash air thru tip and create an explosive mixture in stack. API 521 purge rate maintains min tip velocity to reduce air ingress via tip. Valid for open pipe without seals. Tip seals reduce purge rate. Fluidic: 0.01-0.02 m/s; mol seal: 0.002-0.02 m/s; no seal 0.1-1 m/s

• Purge Gas: After a hot release, as gas cools, it pulls in vacuum. May condense if steam or heavies. See https://cdn.digitalrefining.com/data/articles/file/757959842.pdf. Hot releases are common in Refineries and Petchem plants. Sweep gas based on header volume, density of hot and cooled gas. As header metal absorbs heat, sweep gas demand is usually less. P-T compensated purge can reduce sweep gas demand. Water Seal Drum avoids air ingress and flame flash back to upstream as vacuum in header pulls up water in WSD maintaining a barrier to air via tips

• Purge Gas: If purge gas is lighter than air (mol wt < 29), then as it fills the stack, it creates buoyancy (-ive pressure) at stack base due to density difference. Can ‘theoretically’ suck in air. Based on stack height, calculate the buoyancy effect = H*(ρ-air minus ρ-purge). Maintain a purge rate so that its friction loss thru stack height exceeds buoyancy to have a +ive pressure at stack base