Design Tips 13

Flare

• Flares: Finalize first number and type of flares - HP, LP and AP. Acid Gas flared or vented. Spared or not. Demountable flare. Drone inspection to avoid stairs and ladders. Load reduction by HIPPS; staggered/ extended blowdown; steam drives & backup power. Dynamic Simulation helps reduce loads as all PSVs don’t pop at the same time + flare header packing reduces load to flare tip

• Adjacent Flares: (1) Will the existing flare operate when installing the second. If not, closer location can avoid flame lick from each other (2) Can both be taken out at the same time for maintenance to avoid radiation from operating flare. Otherwise keep them apart. (3) Operating flare stack metal is kept cool by flared gas flowing inside. If only one operates, expansion issues for the other (4) If the 2 divide total flow, then Water Seal Drum (WSD) or PCV to avoid void U tube phenomena on low flows. Read API 521

• Staged Flares: Water seal drums for LP and PCV/SDV backed up by buckling pin for HP. Read API 521

• Flare location: Based on plant layout. Some prefer upwind to avoid plant releases are not blown into open flame. API 14E downwind. API 51R downwind. Some prefer crosswind or in sides

• Flare Gas Recovery Unit (FGRU): Standard packages available. Talk to suppliers.  Read API 521. FGRU recovers continuous low pressure relief and PSV/PCV leaks. See typical leak rates discussed a few pages further down. FGRU is not designed for short duration, high flow and high backpressure releases. On high backpressure, FGRU is bypassed by blowing the seal in a Water Seal Drum or opening an on/off valve. To cater to the on/off valve not opening a RD or buckling pin is provided. RD requires an inlet/outlet LO valves + flare system shutdown to replace RD. Buckling pin is easy to replace and low MTTR (mean time to repair)

• Headers: Service conditions decide MOC. MOC decides number of headers. (1) HT/LT - High or low temperature (2) Corrosive - Acid/ Sour gas, separate heat traced (3) CO2 Removal membrane rejects

• LT/HT Interface: LT minimum 10D or 3 m upstream of LT/HT spec break. With LT sources, some specify min 50D or 10 m to cater to cold creep via metal conductivity. Cold upstream piping can cool inflow gas and reduce downstream temperature further, theoretically. Based on experience, each industry has its norm. For a definite answer, do transient analysis

• KOD: PAHH against twisted or damaged flare tip to avoid ‘blocked outlet’. ESD-1. Causes: Flame lick from adjoining flare. Internal coking/ burning damaging tip. Google Images <Damaged Flare Tip>

• KOD Sizing: Go by API 521. Droplet size: 300-600μ per client. Higher in steam-assisted/ HP flares

• KOD Liquids: Practices differ. Oil & Gas: Route to Inlet Sep. Valuable condensate that drops out on JT chilling or hot flared gas cooling due to heat loss to flare header metal. FPSO: Route to Slop Tanks. Refineries: Not to crude tanks, as light condensate flash adding to vapor load. Heavy liquids go to tanks. Ref: API 521 on hold-up time

• Liquid Carry Over to Flare: API 521 5.7.9.4 - likely droplet sizes that cause burning rain. Estimate liquid fall-out distance to decide sterile area, apart from radiation. Some take it as 60m

• DP: LP KOD/CDD operate at 0 to 34 kPa. DP min 345 kPa (50 psi), based on peak explosion pressure is 7-8 times operating pressure; legacy number from coal mines. Read API 521. Higher DP for HP Flare based on OP + margin, say 10 bar

• DT: Max relieving or black body temp in upstream Oil & Gas. Fire case temp ignored. Release temp is high in downstream Refinery or Petchem. Allowable DP drops with DT for 150# piping. At 40°C = 20 bar. 150°C = 16 bar. As the change is not drastic, a few take DT 150°C. Suits system that get purged with steam for maintenance. DT impacts piping stress + expansion loops on large headers

• FCCU Flare KOD Sludge: O2 reacts with unsaturates and forms polymer. O2 via PSVs or PCVs. Merox unit vent gas has O2. Add RDs upstream of passing PSVs. Consider (a) Routing vent stream with O2 to a furnace firebox via a few floor nozzles or a special burner with an additional nozzle (b) Diesel circulation to make the sludge flow and route to oil recovery (c) Electrical heater in KOD or pump suction to keep sludge in solution and (d) Route sludge to delayed coking or incinerator. Rather than get rid of or live with the sludge, isolate air or O2 reaching KOD to avoid the sludge formation. A small weir with a manhole upstream will help to scoop out. Projecting pump suction nozzles above vessel bottom may avoid muck to KOD Pumps

• Header LT: HP PSV/BDV tail pipe temp may be low - not all the way to flare tip due to heat given to header metal + ambient. BDV release goes down quick. PSV release doesn’t continue as operators respond. Transient heat transfer analysis or Hysys Imperial College Blowdown Model. SS Tail pipe with LTCS/CS headers

• Header Liquid Pockets: Review and avoid. Hydrate blockage can reverse - rupture RDs sending gas/ condensate/ oil to utility system. See Safety Alert

• Subsea U tube header to Flare: Provided to far-off flare to avoid expensive tripod supported bridges to the flare. They get filled with condensing liquid HC + seawater entry due to corrosion. Cause obstruction in relief path + high backpressure on PSVs. Results in burning liquid overflow via flare tip whenever there is a relief + floating burning oil drifts towards platform. Avoid. Results in high purge rate + periodical HP gas sweep to clear the U tube

• Radiation Calcs: Many factors get factored or combined conservatively in the equations used (1) Wind blows towards observer at the highest speed at the same time as peak rate. Wind is free to blow in any direction and there may be no wind in a given instant or during peak flow (2) Peak or noon solar radiation at all times. Line-of-sight coincidence of sun and flame only during early morning or late evening with a weak solar component (3) Ignoring 15-20% attenuation due to humid air (4) 100% heat release. Up to 50% may go as unburnt and soot. Unburnt gets accounted in F, Fraction Transmitted (5) Sonic flares aerate better leading to higher combustion. Interestingly lower F (6) F = 0.1-0.15 for Sonic; 0.2 to 0.3 for LP. More C1, less is F. One may assume F = 0.15 for a wellhead sonic tip but in reality, it is high due to oil carry over - soot and unaccounted additional heat release. Read "environment.gov.ab.ca/info/library/6694.pdf" on observed variations - up and downwind; far and near. “Furnace Operations", Dr RD Reed - 2 points (i) As H/C declines from 0.25 to 0.13 (Paraffins to Acetylenes), unburnt increases from 5 to 50%. Unburnt reduces heat output. We always take 100% combustion (ii) LP flares at peak loads and < H/C 0.25 smoke more, reducing heat release. Soot absorbs heat, reducing radiation. Dr Reed suggests F = 0.11 for H/C 0.33 (methane); 0.12 for H/C 0.25; 0.07 for H/C 0.17 (7) Papers at flare supplier websites: F measured vary (7) Loads: Flare header packing reduce net output to flare tip (8) During a blowdown or cumulative PSV loads, all sources do not start at the highest pressure nor release at the same instant (9) 20% variations in load estimate by different process engineers. Not simple as Current = Voltage/ Resistance. Empirical co-relations with ± 15% accuracy. Some process engineers enjoy doing it to fourth order accuracy not realizing these are estimates. Note: Use LHV (Lower Heating Value) to calculate Q. Not HHV (Higher Heating Value). HHV applies when water in flue gas condenses and releases its latent heat