Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Save to My Library
Look up keyword
Like this
4Activity
0 of .
Results for:
No results containing your search query
P. 1
Consider New Analysis for Flares

Consider New Analysis for Flares

Ratings: (0)|Views: 79 |Likes:
Published by studyendless

More info:

Categories:Topics, Art & Design
Published by: studyendless on Oct 19, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

05/14/2013

pdf

text

original

 
PLANT SAFETY AND ENVIRONMENT
SPECIAL
REPORT
HYDROCARBON PROCESSING
NOVEMBER 2010 
I 
41
 
Consider new analysis for flares
Applying dynamic models in designing safety systemscan reduce capital costs
 
Z. URBAN,
 
M. MATZOPOULOS
and
J. MARRIOTT
, Process Systems Enterprise Ltd.,London,UK, and
B. MARSHALL
, Softbits Consultants Ltd., Medstead, UK
A
pplication of dynamic modeling for relief system design can sub-stantially lower capital expenditure(CAPEX) while simultaneously improving plant safety. This article considers using dynamic analysis to two areas: vessel depres-surization (or “blowdown”) and flare net- work design. New modeling methods canaccurately quantify relief loads and metaltemperatures to enable informed safety andCAPEX decision support.
CASE HISTORY 1:VESSEL DEPRESSURIZATION
Detailed dynamic analysis of the rapiddepressurization (blowdown) of high-pres-sure vessels is a key element of the safety analysis of oil and gas facilities and otherhigh-pressure installations.
Event description.
Depressurization of a vessel usually results in cold gas venting into the flare system. The cold gas can sig-nificantly lower the temperatures within theprocess equipment metal walls and pipe- work, as well as the relief system pipework immediately downstream of the blowdownvalves (BDVs). Low temperatures can leadto embrittlement of the equipment andpipework metal walls, and the differencein temperature between adjacent metal sec-tions can result in high thermal stresses.This condition has implications for theintegrity of process vessels, pipework andsections of the relief system, as well as forCAPEX. Accurate analysis of likely relief scenarios is essential to determine:
Relief loads entering the flare net- work.
For new designs, accurate infor-mation is needed to achieve an optimaldesign that minimizes the piping diam-eters required to meet Mach number andback-pressure constraints. Minimizing thepiping sizes also provides benefits in termsof reduced support infrastructure, whichcan be particularly important in the caseof offshore platforms where additional weight is heavily penalized. For revampsor expansions to an existing process plant,accurate data can help determine whetherthe current flare system can handle the new loads acceptably. In either design scenario,CAPEX savings can be considerable.
Temperature throughout the pro-cess and pipework metal walls
to identify areas of potential embrittlement, and where(and when) unacceptable thermal stressesare likely to arise. Such information can beused to mitigate potential problems eitherby controlling the relief rates or by rerout-ing the relief flows.
Temperature of the relieving “gas”streams
(which may actually contain evap-orating entrained liquids). This providesessential information for choosing theappropriate material of construction for thecritical sections of pipework immediately downstream of the BDV.The effects of low temperature can usu-ally be addressed by using suitable materi-als of construction. Unfortunately, in somecases, such materials can be expensive, andit is highly desirable to minimize their use without compromising safety consider-ations. This requires accurate quantifica-tion of flowrates and temperatures of therelieving stream, as well as the minimumtemperatures reached in the metal walls.
Complex phenomena.
Depressuriza-tion of a vessel involves a complex set of cou-pled physical phenomena that must be char-acterized accurately to understand behaviorand provide suitable design values.Current depressurization modeling isoften performed with off-the-shelf processflowsheeting simulators that use an equilib-rium thermodynamic approach. The latterprovides some indication of the flow andtemperature, but by no means adequately describe the complex thermodynamic andkinetic phenomena occurring as a result of rapid decreases in pressure. The fact thatmultiple phases can form within the vessel,and that these may not be in equilibrium with each other or with the vessel wall, canhave a significant effect on both the relief flows and metal temperatures of the vesseland relief system pipework.The sudden decrease in pressure in a gas-filled vessel results in a rapid change inthe thermodynamic state of the gas withinthe vessel. This can result in nucleationof liquids within the gas bulk to form a “droplet phase” as shown in Fig. 1. Someof the nucleated liquid leaves as entraineddroplets in the high-velocity gas exit stream(Fig. 2). Downstream of the vessel, and asthe pressure further reduces, this exiting entrained liquid evaporates into the bulk gas stream, lowering the temperature of the cold exiting stream even further. Thiscreates a risk of brittle fracture of the flaresystem pipework.Some proportion of the liquid remain-ing in the vessel drops to the vessel floor.
Nucleating liquiddropletsGas to relief system
Formation of droplet phase inblowdown event.
FIG. 1
 
PLANT SAFETY AND ENVIRONMENT
SPECIAL
REPORT
42 
I
 
NOVEMBER 2010
 
HYDROCARBON PROCESSING
Initially, this evaporates instantly due to the warm temperature of the metal it encoun-ters. The effect is similar to a drop of waterfalling on a hotplate. However, once themetal has cooled sufficiently (typicallafter a few tens of seconds), liquid beginspooling (Fig. 2) and forming a continuousliquid phase.The pool boils vigorously, cooling andreducing in size, and, in turn, reducing thetemperature of the metal beneath it. Thisevent can lead to significant temperaturedifferences between the metal immediately below the pool and its surroundings—pre-senting a very real threat of brittle fractureand rupture of the vessel base.The effect of the phenomena can beseen graphically in Fig. 3, which shows theresults of depressuring a vessel filled withlight-hydrocarbon gas at 120 bar. In theinitial phase of depressurization, the gastemperature (black line) drops rapidly. Thetemperature of the metal wall in contact with the gas (green line) begins to drop, butmuch more slowly because of the resistanceto heat transfer between between wall andgas and heat conduction within the wall. After about 80 seconds, a droplet phasebegins to form throughout the gas. Initially,droplets in contact with the metal heat uprapidly and vaporize (red spike). When coolliquid droplets (at a temperature close tothat of the bulk gas) begin to pool on thevessel floor, the liquid temperature increasesfurther above the bulk gas temperature asthe liquid is heated by the metal wall andchanges in composition. After a while, thegas bulk temperature begins to rise becauseof heat influx from the metal wall.The items of most concern are the rap-idly decreasing temperature of the metalin contact with the liquid pool (blue line)and the difference between the tempera-ture of this metal and the adjacent metalcontacting with the gas (green line). Themetal temperature can be seen to dropto nearly –30°C, approaching the brittlefracture temperature for carbon steel. Thetemperature difference between vessel floorand sides rapidly increases to over 20° andis nearly 40° by the end of the blowdown, which may give rise to unacceptable stress.Because of the rapid change of condi-tions, the three phases coexisting in thevessel (gas, droplet and a pool of liquid,)and the vessel walls are not in equilibrium with each other throughout most of theblowdown event. In comparison, the dot-ted line shows the equivalent temperaturecurve obtained using an equilibrium modelfor the same blowdown, which predicts a much less severe drop in temperature. Thismodel fails to identify the most significantsafety-related aspect—the cooling effect of the liquid on the vessel bottom.Fig. 4 shows the resulting vessel walltemperatures and associated thermal stressesfor the vessel vertical walls as color tempera-ture plots. This information would not beavailable without rigorous modeling of thenonequilibrium mass and energy transferbetween phases. This example describes just one scenario. Other scenarios may develop depending on the initial inventory and state of the material in the vessel. Forexample, there may be “bubblet” nucle-ation in super-critical fluid, rather than thedroplet nucleation described here.
CASE HISTORY 2: FLARE NETWORK
Conventional flare header design tech-niques use peak relief flows in steady-statesimulation to assess system capacities anddetermine back-pressures downstream of blowdown valves (BDVs) and pressuresafety valves (PSVs), Mach number in theheaders, and radiation at the flare tip.This steady-state assumption is highly conservative. While conservative approachesmay be desirable in safety system design,they can nevertheless lead to gross overde-sign throughout the system. Key areas of over design include:
Oversized flare header.
Sizing theheader for the sum of the maximum flowstakes no account of effects such as:System packing, where the gas pres-surizes the available volume in the flarenetwork 
Condensed liquid dropletspooling and evaporatingEntrained liquiddroplets leavingwith gas
Droplets exiting in the gas streamand forming a continuous liquidphase on the vessel bottom.
FIG. 2
2300 100 200 300 400 500 600
Time, sec
   T   e   m   p   e   r   a   t   u   r   e ,   K
700 800 900 1,000 1,100 1,200240250260270280290300
Bulk gasEquilibriumMetal wall incontact with gasMetal wall incontact with liquidLiquidFormation of liquidphase and poolingOnset of nucleationThree-phase(gas, droplet,liquid)Two-phase(gas, droplet)Single-phase(gas)
Temperature profiles over the duration of the blowdown event.
FIG. 3
 
PLANT SAFETY AND ENVIRONMENT
SPECIAL
REPORT
44 
I
 
NOVEMBER 2010
 
HYDROCARBON PROCESSING
• Potential for sequencing of flareevents. For example, depressurization ini-tiated deliberately by an operator may becomplete well before a fire causes PSVs tolift. Steady-state peak flow analysis, on theother hand, assumes that all events occursimultaneously.Reducing the peak flows used as thedesign basis by judicious analysis can signif-icantly reduce pipe sizes and materials andfabrication costs, which can be substantialfor large-diameter headers. Reducing thesize also creates knock-on savings related tothe support structure and flare stack size.
Oversized flare stack.
The flare stack sizing depends on radiation emitted by theflame, which is a function of the volumetricgas flowrate through the flare tip. Using unrealistically high flowrates determinedfrom peak flows results in an over-long stack, creating weight problems in offshorefacilities or adding stack support costs (orunnecessary additional header length) inonshore facilities. Similarly, a lack of accu-rate temperature information leads to a widespan between the minimum and maximumdesign temperatures used for gas arriving atthe stack, resulting in unrealistic allowancesfor thermal expansion and contraction.
Over-use of expensive alloys.
 Although flare system pipework may be incontact with gas at extremely low tempera-tures, this typically occurs for a relatively short duration. The use of steady-stateflows does not consider the duration of such exposures to low temperature, whichmay result in very conservative and expen-sive application of alloys.It can be argued that a good flare net- work design is one that minimizes capitalexpenditure while meeting all safety con-straints. Overdesign should be avoided wherever possible.By making simple dynamic analysesusing data that is mostly already availablein some form, it is often possible to refinenetwork designs to arrive at systems witha significantly lower capital cost whiledemonstrably meeting safety require-ments. Similarly, it is often possible to findadditional capacity during retrofits, thusremoving the need for additional capitalexpenditures.Typical examples of where dynamicanalysis can bring significant new informa-tion that has an impact on capital cost are:
Peak flowrates.
The actual relief flow through any PSV is at the maximum only for a short period. Using steady-state meth-ods based on peak flows is equivalent tomaking the assumption that all relief flowsstart at the same time and go on forever. Inreality, it is often possible to take credit forstaged or staggered relief. Shifting depres-surization of certain units by a few tens of seconds can make a significant differenceto the peak flows through the system—aneffect that cannot be represented at all by steady-state simulation.
Packing.
Steady-state approaches makethe implicit assumption that the flare sys-tem has no volume—what goes in comesout, instantly. For larger systems, theimpact of relief flows is partially “absorbed”by pressurization of the flowing lines andthe dead volumes in non-flowing parts of the system. This “packing” effect can reduceboth the calculated peak back-pressures orMach numbers and the peak flows seen atthe flare tip, allowing reduction in headerand tailpipe diameters and flare stack lengths, respectively. Dynamic simulationallows this important buffering effect to betaken into account in the design.
Duration.
Equally important, dynamicsimulation can be used to determine theduration of peak flare loads. Engineering  judgment can then be used to assess therisks of any infringements. For example,a 5-second violation of back-pressure orradiation constraints may well be accept-able, especially given the capital costs of oversizing the flare system to avoid such a contingency.
Temperature.
Relief system pipework that is likely to encounter low temperaturesneeds to be constructed from expensivealloys such as Inconel to avoid the possibil-ity of embrittlement and consequent frac-ture. The true extent of pipework that truly needs to be constructed of such materials isimpossible to gauge with steady-state simu-lation, as low-temperature flows are con-sidered to continue forever, ensuring thatcalculated metal temperatures reach theirminimum. In reality, such flows may only last for a few minutes; the thermal inertia of the pipework metal and heat gain from theenvironment prevent the pipework fromreaching the gas temperatures during thistime (a similar effect can be seen in the bulk gas and metal temperature plots in Fig. 3).It is frequently possible to reduce the usageof alloy significantly based on the moreaccurate information from the dynamicanalysis. One oil company reported saving $1.5 million on a single vessel this way.
Flare-stack temperatures.
Dynamicscan also help provide a true picture of thetemperature of gas arriving at the flare tip.Proper calculation of the effect of low-tem-
4a. Wall temperatures at the end of the blowdown;4b. Wall thermal stresses from the effects of pooling liquid.
FIG. 4
Example of a flare network showing active sources.
FIG. 5

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->