corrosion effects of hydrogen sulphide and carbon dioxide in oil production

Upload: rodrigo-figueiredo-chapouto

Post on 14-Apr-2018

225 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    1/22

    Section II/Gl, Paper 3

    CORROSION EFFECTS OF HYDROGEN SULFIDEAND CARBON DIOXIDE IN OIL PRODUCTION

    B Y

    WALTER F. ROGERS AN D J. A . ROU~E, R ( * )

    SYNOPSIS. Laboratory studie s have been ma de of the pred omi nan t factors in co rro sio n of steelby oil field brines c ontaining hydrogen sulfide or carbon dio xid e. Theoret ical electrochemical relation sat the anode and cathode have been combined to develop relations expressing the corrosion rates to beexpected. The se have been modified in accordance wit h cathode potential value s determined by exper-iment . Th e resulting equations show that corrosion rates und er sulfide conditions are greater th an thoseunder carbon dioxide conditions. Th is correlates wi th field experience. Th e experimental datushow that sulfide cathodes have a lower potentia l tha n those of carbonic acid enviro nmen ts. A s a nadditional imp ort ant factor, sulfide cathodes are more di flc ult to polarize th an cathodes in carbondioxide environments. The net result of a n anode of high potential and a cathode of low poten,tialwith difliculty polarizable characteristics is a high corrosion rate and low pitting factor. A theoryof corrosion und er sulfide co nditi ons has been developed whi ch considers corrosion to occur in twostages. Th e first, or initial stage, i s low and occurs unti l a sulfide cathode of low pol ari zat ion charac-teristics develops; this results in the second, or high corrosion stage. Field d ata are presentedsupporting this theory.

    RI%UM& Les uriteurs ont Ptudik nu lnboratoire les facteurs qui predom.inent duns la corrosion d~lacier pe r des enun: de gisenient c mt ena nt de (hy droge ne sulfurC o u de lanh ydrid e carbonique. D esrela,tion,s thCoriquPs dordre electrochimique, Ctahlies d l mode et U, la cathode, ont &e combin.Ces pourarriver U dautres qui exprimen.t les tnziz d e corrosion a privoir. Ces dernieres relations ont alors CtCmodifikes U la lumidre de valeurs d e potentiels cathodiques determ inis expkrimentalem ent. Les kquationsqui en rksultent montrent que les taux de corrosion duns le cas de H,S sont plus PlevCs que duns lecas de CO,. Ceci recoupe 1exp:lriencr ncqtiise sur le terrain. Les donnPes expdrimentales montrent queles cdhodes environnkes dH,S ont un potentiel p l u s bus que celles qu.i se diveloppent e n milieu dacidecarhon.ique. Fncteur i mporta nt q ui joue duns le m im e sens: les premieres sont p l u s difliciles 6, pola-riser que Irs secondes. L e rbsultat net dune an.ode a potent iel Cleve et d u ne cathode 2c bus potentiel,dailleurs diflc ilemen t polarisable, est est un e corrosion rapide , m ai s un faible facteur de piqilres.Une thkorie de In corrosion sous lnction de lhydrogdne sulfurk a PC divelo ppie, qu i la fait se passere n d e u r stades successifs: le pr em ie r .stcx.de est len t; il se pou rsui t jusq uh ce qu e se dkveloppe un.ecathode - H,F - U, fnihles cnractCristiques de polarisation; ce stade aboutit nu second o u stade decorrosion &levee. Le s a utheu rs prksentrnt des rPsultats p ris (sur le terra.in) pour itayer cette th,korie.

    RIASSUPJTO. Attraverso studi d i lahoratorio g l i A A . hanno indagato s ui principali fa ttori dellacorrosione dellacciaio per opera delle acque d i condensazione contenenti idrogeno solforato o anidridecarbonica ne i ca mp i petroliferi. L e cond izioni relative ai processi anod ici e catodici Sono state espressein formule matematiche e queste Sono state cornbinate in mod o da ric avarn e lespressione dei valo ri dellacorrosione prevedibili. Questi Sono stati adattati ai valori d i potenziale del catodo determinati speri-mentalmente. Le equazioni risultanti dimostrano che in ambiente d i H,S la corrosione 6 maggiore chen o n in ambiente d i CO, , i l che concorda con i dat i dellesperienza pratic a. I dati sperimentali dimostrano( * ) Gulf O il Corporation, Houston Production Division Chemical Lab., Houston, Texas, USA.

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    2/22

    480 P R O C E E D I N G S F O U R T H WORLD P E T R O L R U M C O N G R E S S - S E C T I O N I I / G

    che i catodi d i solfuro hanno un potenziale pia basso d i quelli immersi in ambiente carico d i acido car-bonico. Altro fattore importante t? il fatto che i catodi d i solfuro s i polarizzano meno facilmente che no n icatodi circondati da CO,. D a un anodo a potenziale alto e un catodo a potenziale basso e ditficilmentepolarizzabile, risulta una corrosione intensa e un (( pitl ing N poco rilevante. G li A A . espongono quindiuna teoria sulla corrosione in ambiente carico d i solfuri, secondo la quale tale corrosione avverrebbein d u e fasi: la p r ima fa se , o fase in iz ia le , dura s ino a q u a n do n o n s i f o m i un anodo ricoperto disolfuro il quale ha scarsa facilitd di polarizzazione: ha allora inizio la seconda fase, nella quale si haforte corrosione. A conferrna d i questa teoria, vengono anche riporta ti a lcuni dati ricava ti drrllesperienzaacquisita nei campi petroliferi .

    IntroductionCorrosion in hydrogen sulfide producing oilwells has been a problem of the U . S. oil in-dustry since the first sulfide production wasobtained in the Indiana-Illinois fields in 1889.At the present time hydrogen sulfide bearingoil and gas are produced in the United States

    in important quantities in Texas, Kansas, NewMexico, Arkansas, and Wyoming. Other coun-tries having a significant amount of such pro-

    duction or reserves are Canada, Mexico, andFrance.Damage to oil and gas well equipment undersulfide conditions may become manifest in sev-eral forms. The most prevalent is th at ofgeneral corrosion. General corrosion is definedas macroscopic attack and is manifest by thedevelopment of large deep pits, heavy ironsulfide scale formation, and loss of a large portionof the total metal volume. Fig. 1 illustratesthis type of attack. General corrosion occurson th e flow lines and tanks of th e producingsystem as well as the well tubing, casing, suckerrods and the like. This form of a ttack is responsible for the major corrosion losses whichoccur in siiltide production.A second form of sulfide at tack is corrosionfatigue. This form is restricted t o pumpingwells and occurs only on those portions of thewell equipment which are subject to cyclicstress, such as sucker rods.

    A thi rd form of at tack known as sulfides tress-corrosion cracking has recently been encountered in attempting to produce hydrogensulfide bearing high pressure gas-condensatewells. Sulfide stress-corrosion cracking has developed rapidly in some cases, resulting insharply fractured pipe within a few days. Theminimum hydrogen sulfide pressure required fothis type failure is not known, but failureshave developed in wells having a shut-in welhead hydrogen sulfide pressure of 297 psi. Inthe cases where this trouble developed, appre-ciable quantit ies of carbon dioxide were presenalong with the hydrogen sulfide. N o such failures have developed in wells producing carbondioxide to the exclusion of hydrogen sulfideA f ou rth ty pe of failure in th e presence ohydrogen sulfide is hydrogen embrittlement

    Fig. I. General Sulfide Corrosion of Oil W C ~ I Tubing24 Months Service

    This type failure has been principally encountered with sucker rods, drilling lines, wire lines

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    3/22

    WALTER F. ROGERS AND J. A. ROWE, JR-CORROSION EFFECTS O F HYDROGEN SULFIDE 481

    etc. In these cases the rods or wires breakeasily after being in contact with the sulfideproduction for periods of time ranging from afew hours to months. Usually, aging th e metalin the atomosphere or at a low heat will renderit free of the brittleness. This typ e of failureis possibly closely kin to sulfide stress-corrosioncracking.A filth typ e of attac k, bu t one which seldomresult s in failure, is the development of hydrogenblistering. This becomes manifest through theformation of raised portions of t he met al inblister form. The at ta ck has been noted prin-cipally on the sheets of tan ks handl ing souroils and on test metal specimens exposed tohydrogen sulfide in the laboratory.Copson (1 ) has presented a review of theliterature on both sulfide and carbon dioxide

    oil well corrosion. The major ity of technicalwork on oil well corrosion has been on the prob-lem of sweet condensate wells, i. e., metalhandling carbon dioxide bearing gas under highpressure. At th e present time this problem islargely under control and an increasing amountof atte ntio n is being given to corrosian in sul-fide environments.Previous short time laboratory studies ofsulfide general corrosion have been reported byRogers and Shellshear ( 2 ) , who concluded thatsulfide bearing brines at pH 6.4 to 6.8 werenon-corrosive when free of oxygen. The testsincluded studies at fluid temperatures an d veloc-ities encountered in wells, and it was foundthat while these factors influence the corrosionrate th ey a re not of sufficient importance inthemselves to explain the high rates frequentlyencountered in the field. Neutral dissolved sol-ids at concentrations found in oil well brinesdid not affect the corrosion rat e greatly . Ginter( 3 ) reported that iron sulfide is cathodic tosteel and can result in increased corrosion rates.Ewing ( 4 ) has presented data showing thatiron sulfide is anodic to steel at pH valuesabove about 6.5. Ewing further concludes thatthe driving voltage for the corrosion reactiondepends only on the hydrogen sulfide activityof the solution and the partial pressure of mo-lecular hydrogen on the cathode . Rogers ( 5 )has investigated the protect ive effect of oilsproduced with sulfide bearing crudes and foundthem t o be non-wetting in character. Theproduction of high percen tages of oil, however,

    ma y result in protection through the mechanicalformation of oil films on the exposed metalsurfaces. Wescott and Vollmer (6, 7) havereported on the rat e of embrittlement of drillinglines and corrosion fatigue of sucker rods undersulfide conditions.It is now generally believed that the pHof the aqueous fluid produced by oil and gaswells is an important factor in the corrosionrate. This was first established for high pressurecondensate wells, when it was found that theproduced carbon dioxide rendered the well waterqui te acid. These wate rs also in some casescontain lower aliphatic acids, such as acetic,which further reduces the fluid pH (8). I t hasalso been shown that oil well brines foundneutral at the well head may be acidic in thewell when the acid gases released during produc-tion are recombined under reservoir conditions.The fluid pH in the well may be 2.5 units lowerthan at the well head (9). The development ofacid pH values in oil well brines and condensatewaters is believed to be one of the main causesof both sulfide and non-sulfide corrosion.In spite of t he work which has been doneon the causes of t he var ious forms of sulfidecorrosion there is still insufficient da ta t o explainproperly all th e phenomena encountered. Inview of thi s lack of complete in formation, labo-ratory tests have been made to determine betterthe mechanism of the corrosion reactions in-volved in general sulfide corrosion. I t is thepurpose of t his paper t o report upon theseinvestigations which, although primarily directedat the relative importance of factors involved ingeneral corrosion under sulfide and carbon diox-ide systems, still may shed some light on theassociated problems of stress-corrosion cracking,corrosion fatigue, embrittlement, and hydrogenblistering.Electrochemical Relations in General Corrosion

    The electrochemical theory requires the cor-rJding system to possess an anode, o r corrodingmember, and a cathode, o r non-corroding member.Both must be immersed in a conductive solutionand possess a metallic contact between them.An examination of the relations which occur atthe anode and cathode, and in the conductivemedium between, permits an understanding ofthe electrochemical behavior of the system.

    Proceedings 4th W . P . C . - Section 11 31.

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    4/22

    48 2 PROCEEDINGS FOURTI-I \I'ORT,II I'F:TROLEIJM CONGRESS-SECTION T1,'L:The total corrosion rate which occurs maybe related to the current which flows throughthe equation :

    [ I1326.8 I MC=-- zC = Corrosion rate, g per yearI = Current flow, amperesM = Molecular weight of corroding metalZ = Valence

    This relation shows the corrosion rate to bedirectly related to t he cu rrent flow in the system.The current flow in a corroding system is gov-wned by the relation :

    where

    i21E a - oR-whereIa = Anode current, aniperes

    E, = Anode potential, voltsEc = Cathode potential, voltsR = Resistance of the circuit, ohmsThe circuit resistance, R, is composed ofseveral elements. The more impo rta nt of theseare the resistance of the aqueous media, theresistance of the anode an d cathode metallic cir-cuits, and the resistant films which developon either or both the anode and cathode as theresul t of current flow. Polarization changeswhich affect E, and E, also can become of greatimportance in controlling the corrosion rate.

    Reac t ions al t h e A n o d eIn the case at hand the problem is concernedwith corrosion of steel in hydrogen sulfide andcarbon dioxide aqueous brine environments.The anode potential at any time must be thatof an iron anode in a solution of iron ions. Thesingle eloctrode potential a t any time for arerrous electrode is given by the 1 ~ 1 1 nownrelation :

    &e =where :

    FJFeRTzFaFP+ +

    131. 441 + ~ RT In i jFtl++Z F

    Wi th substituti on of th e values and transferrito common logarithms, the relation becomeE F ~ - .541 + 0 .0295 log aFe+ + [3-

    The value of the activity of the ferrous iois the governing force in determining the anopotential. Regardless or whether the corrosioenvironment is a sulfide, carbon dioxide, acido r basic solution, the potential of the anode wbe determined b y t he acti vity of th e ferrous ioat the anode surface. This activity will in turhowever, be governed by the so lubility of ferroions in the solution.Where the solution contains hydrogen sufide, or sulfide ions from other compounds, tlow solubility of i ron ha s a n appreciable effeon the anode potential. The solubility of irsulfide is calculated from the activity solubilproduct : a Fe ++ x aS= = 1 x lO-'9where :

    a F e + + or aS= is the activity of the rspective ions in gram-mol./liter.An estimate of the iron anode potential cbee made for sulfide brine concentrations fouin oil wells. The normal concentrati on of sufides ranges from 2 00 t o 750 p.p.m. as H2S. Nothe ac tiv ity of the sulfide ion is much less ththe stoichiometric concentrations, but fo r sim

    plicity of calculation the iron anode pot entiis calculated for sulfide activities of 1, 1 0 , an10 0 p.p.m. as - . From these values of S = anrelations 3-A and 4 , the anode potential becomeTABT,E

    as= nFe+ + aFc + + E:Fe(ppm) ( m m ) (an-mol/liter) (Volts)1 1.78 X l O - ' O 0 . 3 2 x 10 -14 - 84110 1.78 x 10-" 0.32 X 10-l6 - 870100 1.78 X 10-'* 0 . 3 2 x 10-ln - go0

    = Potential of the single ferrous= Gas constant, S.31 joules per= Absolute temperature, OC= Valence = 2-= Faraday, 96 ,500 Coulombs= Activity of ferrous ions; gram

    electrode, voltsdegree C per mol

    mol.,'liter a t P5oC

    It will be noted that the dissolved iron cocentration in the presence of sulfides is extremly low, being much less than 1 p.p.m., eventhe low sulfide activi ty of 1 p.p.m. This permth e iron anode to m aintain a high potentiwhich can lead to a high corrosion rate.Inasmuch as Relation 4 shows the mutuinterdependence of t he sulfide and fer rous i

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    5/22

    WALTER F. ROGEHS A N D J . A. ROWE, JR-CORROSION EFFECTS O F HTDROGEN SULFIDE 483

    activities, the sulfide activity can be substitutedin 3-A with the following result :[3-B]

    Relation 3-B requires a knowledge o l th eactivity of the sulfide ions in solution. Suchactivity data are not readily available in theliterature and are not easily obtained throughdirect determination.As corrosion occurs under the sulfide envi-ronment, iron sulfide is precipitated near theanode face. The development of this corrosionproduct may result in marked changes in thecorrosion rate through alteration of either E F ~rR o f Relation 2. Should sufficient mechanicalblocking of the anode face occur, the rate ofdiffusion of sulfide ions t o the anode surfacemight be reduced t o the point where the con-centration of dissolved ferrous ions would in-crease greatly. The development of only 1p.p.m.iron activity at the anode surface would alterits potential t o - .597 volts, a significant de-crease with respect to the calculated value of-0.841 volts in the presence of 1 p.p.m. sulfideion activi ty. A second result of the develop-ment of a heavy iron sulfide scale could be anincrease in resistivity of the conductive aqueouspath. This increase in R would then decreasethe corrosion rate accordingly.Where the corrosive environment containscarbon dioxide the corrosion iron has a relative-ly high solubility in comparison to iron sul-fide. Two corrosion products, i.e., ferrous car-bonate and ferrous bicarbonate, are present asfollows :

    3EFe =- .002- .0295 log aS=

    Fe + 1 1 2 C O ~ -+ Fe ( H C 0 3 ) *+ 1!2Fe( I-ICO3)z- Fe + f + 2HCO3-HC03-- l I + 4- CO3=Fe+ + + C 0 3 = -+ Fe CO3

    Ferrous bicarbonate is relatively soluble, ferrouscarbonate much less so. Even ferrous carbonate,however, has sufficient solubility that its sol-ubility product is much higher than ferrous sul-fide. A solubility product which has been cal-culated from literature solubilities at 25oC is :Ks FeC03 = 3.4 x 10-7 is1

    Again a s in the case of a sulfide environ-

    ity ma y, therefore, be subs titu ted in Relation3 to determine its effect upon the anode poten-tial. Relation 3-C results :EFe =- .863- .0295 log aCO3= [3-c

    The solubility of ferrous iron in condensatewaters made acid by carbon dioxide is suchthat iron concentrations from 50 to 400 p.p.m.are measured analytically. Assuming thi s iront o have unit activity at these concentrationsthe ferrous anode potential by Relation 3-A becomes :

    TABLE 1Iron Iron EFe(p.p.m.) (gram-moliliter) (volts)

    50 9 x lo-' - .586400 7.2 x 10-* - .554

    These anode potentials are much less thanthose obtained in sulfide environments. Thedifference between a sulfide solution of 10 p.p.m.sulfide activity and a carbon dioxide solutionwith 50 p.p.m. iron acti vit y is 0.870- .588 o.284 volts. Such differences in potentials areone of the reasons why carbon dioxide environ-ments are less corrosive than those containingsulfide.Reactions at the CathodeWhen corrosion occurs to a steel surface asthe result of potential differences, in accord-ance with Equation 2, the first electron flowis probably t o a cathode which acts primarilyas a ferrous electrode of lower potential than theanode. Once current flow starts, however, hy-drogen is developed at the cathode in accord-ance with the relation :

    2H20 + 2e -+ H2 + 20H-This development of a cathode coated withhydrogen results in a cathode potential similart o that of a hydrogen electrode

    EH, = 0.059 log aH +- .0295 log PH, [6where PR, = pressure of hydrogen gas inatmospheres.ment, Relation 5 gives th e interdependence of Reiardless of the corrosive environment, theferrous and carbonate ions on each oth er in potent ial response of this cathode is that of acarbonic acid solutions. The carbonate activ- hydrogen electrode, i.e., it is influenced pri-

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    6/22

    484 P R O C E E D I N G S F O U R T H WORLD PETR O LEU M CO N GRESS-SECTIO N I I /G

    marily by the pH of th e solution. Althoughthe metal cathode acts in a similar manner tothe hydrogen electrode, there are differenceswhich affect its value. The hydrogen elec-trode is the standard against which electrode

    'potentials are measured and is assigned the ar-bitrary potential of zero. The hydrogen elec-trode consists of a platinum wire o r strip coatedwith platinum black to reduce polarization phe-nomenon. Pur e hydrogen gas is bubbled aroundthe electrode in the test solution to maintainits contact with hydrogen gas at one atmos-pheric pressure, the standa rd condition. Varia-

    cathode and hence decrease the corrosion rateRelation 6, therefore, requires as a first changthe addition of th e over-voltage effect. Thibecomes :EH,=0.059 log a H +- .02% lo g PH,- [6-A

    where E = Over-voltage of hydrogen undethe current density of the corrosion reactionReactions of the Coupled Cell

    From the relations which have been developed it is possible t o couple the anode and cathodand establish the resulting corrosion rate.

    L _ _ _ _

    I ' &2'-;-""'Fig. 2. Flow Diagram Corrosion Testing Apparatus

    G-4sELECTRICALTEST. BRINE

    ----______- - - - - -

    tions of the potential of the electrode with pres-sure of t he hydrogen gas as well as hydrogenion concentration are shown by Relation 6.When steel acts as a hydrogen electrode duringthe corrosion process the potential developeddiffers from that, of th e theoretical hydrogenelectrode. This results from the fact t ha t thepressure of the hydrogen gas developed may heless than atmospheric, i.e., gas evolution mayoccur too slowly, if at all: to simulate theoreticalconditions. In addi tion, discharge of the pro-ton to atomic hydrogen, or the combination ofatomic hydrogen to the molecular form ma:.require more electrical energy than for thestand ard hydrogen electrode. This added workrequired for discharge of gaseous hydrogen.whatever its cause, is known as the over-volt-age potential of the metal. Over-voltage po-tentials act t o increase the potential of t he

    Where the corrosive environment containhydrogen sulfide o r other sulfide ions, Reltions 2; 3-B and 6-A are involved. Reltion 2 is :

    After substituting 3-B and 6-A, Equation 2 fH2S environments becomes :

    ( a s = ) aH+)2IH,S= - - .002- 0295 logR'I PH,where 1H.S = Current flow in amperes

    E1 = Cathode over-voltage, volts.Where the corrosive environment contaisufficient carbon dioxide fo r th e solubility pro

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    7/22

    WALTER F. ROGERS AND J. A. ROWE, JR-CORROSION EFFECTS OF HYDRO GEN SUL FID E 485

    17cncJ02 14JJz0U0

    -a- I

    WU3a" 5cnI(Y

    2

    PHFig. 3. Cathode Potentials in H,S or CO , Brines

    p HH 4.5.0---'~..25Y

    r ~ ~~~~

    r * p H 9.0500 1000 1500 2000 2500 3(

    S U L F I O E S AS H2S. p p mEffect of H,S and pH on Corrosion Rateig. 4.

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    8/22

    486 PROCEEDINGS FOURTH WORLD P E T R O L E U M CONGRESS-SECTION II/G

    uct of ferrous carbon ate to hold, Relations 2,3- C and 6-A are used. These become:

    ~ 7 - wwhere ICO,= Current flow, amperesWhere the corrosion product, i.e., ferrous

    iron, has no solubility product, th e relations he -come:

    Ez = Cathode over-voltage, volts.

    (aFe++j PH,;-0.441 + 0.0295 log['7-C!

    where E3 = Cathode over-voltage, volts.Theoretically, in a ny given case of corrosionunder sulfide o r carbon dioxide conditions, equa-tions 7-A and 7-B would permit calculation o fthe corrosion cur rent. The da ta required wouldbe a knowledge of the corrosion cell circuit rt'-sistance, the activities of the sulfide or carhon-ate ions, the hydrogen pressure, the p11, and theover-voltage characteristics of the metal . Someof these facto rs are capable of direct dete rmi-nation, some are available in the litc.rature, andsome are dificult to obtain.

    I .7v)c-I0>3 1.4=i5a-0 1 . 1a0I-za 0 . aa

    ua 0 .5

    30

    00Ial0.2

    Relations 7-A and 7-B show that at equap H values and activities of carbonate or sufide ions, the sulfide corrosion rate is greateThis is borne out by field experience. Thequations further state thal the corrosion ratis a direct function of the activity of th e precipitating ion and of the squarc of the hydrogeion concentration.These equations are satisfactory so long athey describe the corrosion environment anelectrode properties. To a great extent thecan do this f o r oil well conditions, as the effecof velocity, temperature, and diffusion can bexpressed through adequate representation othe factors they affect in 7-A and 7-B. I t ialso of importance to recognize thal the initiacorrosion current flow can be markedly reduceas the result of polarization of the cathode oanode, the forinn t ion of protectivp films, etc

    Laboratory Test EquipmentThe laboratory test equipment used in thi

    stu dy is shown diagrammat,ically in Figure 2The appara tus consists of a circulating brinsystem which contains a calibrated burette ( Cof 250 cm3 capacity, a Model T-6 Sigmamoto

    CA RB ONA T E S AS H2CO3, ppmFig. 5. Effect of H,CO, and pH on Corrosion Rate

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    9/22

    WALTER F. ROGERS AND J . A. ROWE, JR-CORROSION EFFECTS O F H Y IIH O G EN SULFII IE i 3 7

    pump (K), a therinoiiieter ( G ) , glass and calo-mel electrodes for pN measurement (E) , a1000 cm3/min. rotameter (F) for fluid volumemeasurements, a cell equipped with platinum andsteel electrodes, and a calomel cell for study ofcathode potentials ( D ) , and a corrosion cell con-taining a steel-platinum galvanic cell (H) formeasurement of the corrosivity of the circulatedfluid. .A Beckman Model H2 A C pH Meter isused t o indicate pH values. The burett e iswater-jacketed (B) so that the circulating fluidmay be kept at an17 desired temperature be-tween 750F and 1600F, utilizing heated water(A) and pump ( J ) . *Iacuum pump (L) isincluded as auxiliary equipment . This pump isused to exhaust through an acid gas removaltrain (Pi) air, acid gases, i.e., carbon dioxide orhydrogen sulfide, as required during the test runs.Known quanti tie s of carbon dioxide or hy-drogen sulfide can be introduced into the systemby displacing them from cylinders through thepurification train (Pz) nto burettes ( N ) and thendisplacing into the circulating brine as desired,using mercury displacement from levelling bulbs.The corrosivity of a circulating fluid is de-termined through the use of galvanic cell ( H ) .This cell consists of a cylinder of platinumfoil 1/2 inch in diameter by 2 3 / 8 inches irilength. Centered inside the platinum tube isa No. 10 common steel nail insulated from theplatinum by a band of rubber tubing top and

    hottorn. The nail and platinum are connectedelectrically through a 6. 7 ohm resistor externalto the cell. The corrosion current passes throughthis resistor and is determined by measuringthe potential drop across the resistor, using a0-18 mv millivoltmeter (M) . This potentialdrop was used as th e index of corrosion inFigs. 3, 4 , 8, an d 9. The cathode anode ratioof this cell is 4 : 1. This low ratio was laterfound to restrict corrosion current under car-bonic acid conditions. The potential of thecathode or anode may he measured with respectt o the calomel cell positioned at (E) .The use of a galvanic cell to measure corro-sivity possesses many advantages over the com-monly used method of measuring weight lossesof coupons. Steel was used as the anode forobvious reasons. Plat inum was chosen as thecathode because of its low potential, relativeinertness, and ability to act as a hydrogen elec-trode, t he normal response of the cathode evenin steel-steel corrosion phenomenon. Couplingtwo such materials tends to insure that the ironis truly anodic to the cathode, permitting ameasurable current to flow even under relativelynon-corrosive conditions. Steel-steel anode-cathodes could not be made to perform satisfac-torily in this service. In tests made usingsteel-steel couples the larger size steel specimeninvariably became t he anode regardless of thecomposition of th e two metals.

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    10/22

    488 P R O C E E D I N G S FOURTH WORLD PETROLEUM CONGRESS-SECTION I J I GI

    v)c

    a'0anczWUUa00100

    6LOG (oH+)2

    7 8PHFig. 7. Effect of (aH+) ' on Corrosion Rate

    Another important portion of the test equip-ment is cell (D), in which the potential responseof a steel cathode was measured under sulfideand carbonic acid environments at different cur-rent densities. The cell contains a circular1.623 square inch horizontally mounted cathodeof 1020 steel. Positioned 1.5 inches above i t is ,a platinum sheet anode of equal diameter. This tanode has a centrally positioned 3/8-inch hole :anode and cathode are connected to a 1.5 voltbattery, variable resistor, and fixed resistor.The current flow to the cathode is controlled bythe variable resistor and measured as a potentialdrop across the fixed resistor by means of a milli-voltmeter with full scale o f 0-18 mv. The potentialof the cathode was measured by means of a 0-4volt potentiometer (Q ) . This cell was placedin the circulating system only when necessary.

    Experimental TechniqueA total of three separate types o f studieswas made utilizing the equipment of Figure 2 .

    The first consisted of a study of corrosion rates

    0

    5g5"

    Fig. 8. Cathode Polarizatioll in 12,s Brines

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    11/22

    WALTER F. ROGERS AND J . A. ROWE, JR-CORROSION EFFECTS O F H Y D R O G E N SULFIDE 489

    I!I, .o .7 5 .80 .85 .90 .95 I .ooC A TH0 E -C A LOM EL E L ECTRODE, VOLTSFig. 9. Cathode Polarization in CO , Brine

    obtained from five corrosive and non-corrosiveoil field brines, with and without hydrogen sul-fide or carbon dioxide. These consisted of anon-corrosive sweet brine from the East Texasfield, two corrosive sulfide brines from WestTexas, one non-corrosive sour brine from theWinkler Field of West Texas, and a non-corro-sive sour brine from the Darst Creek, Texasfield.

    Approximately 350 cubic centimeters of wellbrine were introduced into the circulating ap-paratus and brought to 1000F, the standardtest temperature, and to the standard pumpingrate of 500 cm3 per minute. Vacuum was ap-plied to the top of the recombination cell andgaseous components in the system removed.Evacuation at th e vap or pressure of water wascontinued until the p H of the circulating brinereached the max imu m value. This removal ofhydrogen sulfide and carbon dioxide resulted inmajor changes in the fluid pH . Some values ob-tained through evacuation are shown in Table IIT.

    TABLE 11 pH AfterWell Head Evacuation TgpeWell PH of Acid Gas BrineT. G. Hendricks 7 6.60 7.80 SourThomas Dix 3 6.85 8.29 SourTubb 11-T 6.60 7.50 s o u rL. D. Grim 3 6.70 7.73 SweetCrier-McElroy 9 6 . 7 5 8.42 SOW

    These upward changes in pH result from thepresence of alkaline components in the wellbrine. These ar e principally sodium and cal-cium bisulfides and bicarbonates. Sulfide brinescontain, besides bicarbonates and possibly freecarbon dioxide, alkaline bisulfides, free hydro-gen sulfide, mercaptans, and possibly alkalinesulfides and polysulfides. Thu s, in Equation7-A the equivalent activ ity of t he to tal sulfidecompounds is required rather than only that ofthe hydrogen sulfide per se .

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    12/22

    4930 PROCEEDINGS FOURTH WORLD PETROLEUM CONGRESS-SECTION II/GThe corrosion rate of each brine was deter-mined, both containing H2S at varying partialpressure up t o approximately 1400 mm ofHg H2S pressure without C 0 2 , and at a similarcarbon dioxide pressure without H2S. In makingeither a CO2 or FIBS run the circulating brine

    was evacuated until the pH reached a maxi-mum. The system was then brought directlyto the desired gas pressure using the gas t o betested. After equilibrium was reached the sys-tem was evacuated. This permitted the com-

    Fig. 10. General Carbon Dioxide Corrosion of Oi lWell Tubing. 17 Months Servicc.

    plete removal of residual sulfides or carbonateswhen tests with carbon dioxide and/or hydrogensulfide were madp. Having thus removed, forinstance, all residual carbon dioxide, an incre-mental amount of hydrogen sulfide gas was ad-mitted. and the system permitted to come toequilibrium. The pH, corrosion rate of thegalvanic couple, and separate potentials of theanode and cathode to the saturated calomel cellwere noted. In this manner a series of pHvalues, corrosion rates, and anode and cathodepotentials were obtained. The corrosion rateof the system at the typical well sulfide or car-bonate content and pH value could be relatedto the known corrosion rates of the brinestested.

    In these tests the anode potentials wereround t o remain relztiv

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    13/22

    WALTER F. ROGERS AND J . A. ROWE, JR-CORROSION EFFECTS OF HYDROGEN SULFID E 491

    D A Y SFig. 11. Coupon Corrosion in H,S Environments

    The procedure for dete rmination of the pol-arization characteristics of the cathode was es-sentially the same in each of th e three casestested. A synthetic brine containing 50,000p.p.m. sodium chloride was charged t o the system,brought to 1000F, evacuated, and circulated at500 cm3/min. The circulating fluid was adjust-ed to contain approximately 500 p.p.m. sulfideo r carbonate content at the pH values of 4-5,6.5-7.5, and 9.5-10.0. This required the addi-tion of some sodium hydroxide to obta in thehigher pH values. After the desired pH andcarbonate o r sulfide content was obtained thesystem was allowed to circulate until the poten-tial of the cathode reached equilibrium. In thecase of the sulfide condition a black iron sulfidefilm was present on th e steel surface. In addi-tion to the test using a sulfide surface in asulfide fluid and a non-sulfide surface in a car-bon dioxide fluid, a test was also made of asulfide surface in a fluid containing carbon diox-ide but no sulfide.

    After establishment of the desired metal sur-Face and circulating fluid condition, current wasallowed to flow to the metal cathode. Whenthe cathode potential was found to have reachedequilibrium the potential was recorded and the

    current increased. This was repeated in stepsuntil the desired maximum potential vasreached.

    Cathode Potentials in Hydrogen Sulfideand Carbon Dioxide EnvironmentsThe data resulting from the first series oftests were of particular importance in demon-strat ing th e action of the cathode in hydrogensulfide and carbon dioxide environments. Sev-eral complete runs using both of these acidgases were made with each of the five oil well

    brines. Since each complete run consisted o fa number of acid gas pressures, together with ameasure of th e developed corrosion rate andcathode potential, a large amoun t of d at a wereobtained. These are shown in Fig. 3. Eachpoint represents the measured potential of theplatinum cathode to a satu rated calomel electrode.The potential points are plotted against thep H of the solution in which they were measured.A series ol parallel lines have been drawnon the figure and these represent the potentialof the hydrogen electrode in equilibrium withvarying pressures of hydropen gas, as calculatedfrom Relation 6.

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    14/22

    492 PROCEEDINGS POURTH WORLD PETROLEUM CONGRESS-SECTION I I / GIt will be noted th at the carbon dioxide Utilizing the value of cathode potentiafound here, equation 7-A expressing corrosion

    currents under sulfide conditions becomes :cathode potentials fall relatively well along a linedescribing the potential of the hydrogen elec-trode, plus an over-vol tage potential of 0.062 ( as=) aH+)2olts. This average over-voltage potential cor- I H , ~ 1- 1.002 - 0295 log 1 -5

    relates reasonably well with the over-voltage [7-E

    ode in iron-iron corrosion under carbon diox- [ 7 -

    value of 0.09 volts of smooth pla tinum givenin the literature (10). If this cathode poten-

    and~ H , S= [- 1.1495 - 0295 log (a s = ) (aH+)2tial can be transferred directly t o a steel cath- R

    ide conditions, Relation 7-I3 becomes :ICO,=

    Relative Corrosion R>atesUnder Sulfideand Carbon Dioxide Conditions[- .601 - 0295 log (aCOj= ) (aH+)2]i7-w

    The hydrogen sulfide cathode potentials showconsiderable scatter. The points, however, fallinto fairly well marked groups. It will be notedthat the cathode values are all much lower thanthose found for the carbon dioxide conditions.The potential values at the varying pHs fallalong hydrogen electrode values at partial pres-sures of hydrogen varying from 1 x 10-2 t o1 x 10-5 atmospheres. The major ity of poten-tial values might be considered as represented bythe hydrogen electrode potential in equilibriumwith 1 x 10-5 atmospheres of hydrogen. Thislatter condition is taken as the most probablecathode potential value, although the greatervalues are believed t o be significant as repre-senting intermediate stages in the developmentof the minimum sulfide cathode potential . Thesereduced cathode potentials represent an impor-tant property of sulfide corrosion, i.e., they de-monstrate the fact that part of the increasedcorrosion rates under sulfide conditions are duet o the great potential difference resulting fromthe low cathode potential.

    In the experimental work with sulfides theplatinum cathode developed a thin film of ironsulfide. It is believed this iron sulfide coveredthe platinum cathode t o develop a continuousfilm and exhibit the properties measured. Itis believed the low potentials so developed onplatinum carry over t o iron cathodes under sul-fide conditions, as the result of the developmentof iron sulfide surIaces, either through directreaction of the m2tal with sulfide o r the mechan-ical deposition of iron sulfide on the metalsurf ace.

    The second series of tes ts furnished da ta othe relative corrosion rates of steel under carbodioxide and sulfide conditions. The tests wermade at pH values from approximately 4 to and acid gas contents up t o 3000 p.p.m., a slightlygreater range than normally found in oil welbrines.The data for sulfide corrosion rates are showin Fig. 4. They have been plotted t o correlate with Relation 7-F, which has been deriveto express the maximum corrosion current undesulfide conditions.

    1IH,S = - - 1.1495- .0295 log (a s = ) aH+)RThe relation is :

    [ 7 -Relation 7-F sta tes th at the sulfide corrosion rate is controlled by changes in the ac tivity of the sulfide and hvdrogen ions of the corrosion cell environment. Figure 4 shows thathese factors are important in controlling thcorrosion rate.Relation 7-F states that the corrosion current at constant pH is a straight line functioof the log as=. Relation 8 shows that at constant pH the sulfide ion concentration is a direcfunction of the bisulfide concentra tion. Therefore, if the total dissolved sulfides from Figureare plotted against corrosion current at constan

    p H straight line relations should result. Thiassumes, of course, that only bisulfides are present , which is not entirely the case. The datof Figure 4 are not sufficiently extensive foplots t o be made at each test pH, but data havbeen used at the pH values of 6.25 and 9.0 anare shown in Fig. 6. It will be noted thastraight lines are obtained showing correlationbetween Relation 7-E and the sulfide activity

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    15/22

    WA L T E H. F. ROGERS AND J. A . ROWE, J R - C O R R O S I O N EFFECTS O F HYDROGEN SULFIDE 493Relation 7-E also states that at constanta s = the corrosion current is a straight line func-tion of the log (aH+)2. The cS= was calculatedfrom the relation :

    where :S= = the stoichiometric concentration of sul-HS- = the stoichiometric concentration of bi-pK2 = second dissociation constant of H2S =By selecting the pH and knowing ~ K ~ , H , s ,heratio __ can be calculated for any HS- con-centration. From this ratio the concentrationof S = . can be calculated. By plotting the cor-rosion rate at this S = value against pH or (aH+)z,a test of the correlation of these factors with7-E is obtained. Following this procedure andselecting the S = value from the maximum HS-concentration tested at pH 6.0 as the constantS = value, calculations were made at pH 6.0,7.0, 8.0 and 9. The data are given in Fig. 7.Again, the assumption is made that only bisul-fides are present, The curve shows that thecorrosion increases at a rate greater than at astraigh t line relation required by 7-F, partic-ularly a t the lower pH values. This couldresult from the difficulty of ascertaining equalactivi ty values of S = required for selection of thecorrosion currents taken from the curves.The experimental data obtained for corro-sion rates in carbonic acid environments areshown in Fig. 5. ReIation 7-D has been pre-viously derived t o express the corrosion currentdeveloped under carbonic acid conditions. Therelation is :IH,CO,=

    fide ionssulfide ionsi x 10-15

    S =HS-

    1 [-0.601-0.0295 log (aCOa-) (aH+)2]E ~ - D I

    This equation is quite similar t o 7-F withthe exception that the anode potential is lowerand the cathode potential higher than foundwith sulfide environments. This should resultin lower corrosion currents, which is borne outby the curves of Fig. 5. The curves also showthe dependence of the corrosion rate on thecarbonate ion and hydrogen ion concentrationsas called for by the relation.

    Equation 7-D states that the corrosion cur-rent at constant pH is a direct function of thelog aCO3=. This relation is similar t o that ofsulfide corrosion. The corrosion currents takenCrom Fig. 5 were plotted against the carbonatecontent for pH values of 7.0 and 8.0 in Fig 6.The data show that straight lines are obtainedas called for.Relation 7-D also states that at constantaCO3= the corrosion current is a straight linefunction of the log (a H+ )z as under sulfide con-ditions. To test this relation a concentrationof C 0 3 = ions was selected, and calculation ofthe HC03- concentration made at pH values of6, 7, and 8 in the same manner employed forsulfide environments. The corresponding valuesof corrosion current a t the aCO3= selected weretaken from Fig. 5 and plotted in Fig. 7 againstthe log (aH+)z. The data are similar t o thoseobtained under the sulfide condition, in thatthe corrosion rate increases at a more rapidrate with increase in (a H+ )z than called for bythe equation.It is particularly noteworthy that 7-F and7-D call fo r hydrogen sulfide corrosion rates tobe greater than for carbonic acid conditions.The corrosion curves of Figs. 4 and 5 show thatthis difference is borne out experimentally.

    Polarization of Steel Cathodes in HydrogenSulfide and Carbon Dioxide Environmentsand Effect on Pitting Factors

    The third series of tests was made t o deter-mine the ease of polarization of th e cathodeduring corrosion of steel in sulfide or carbonicacid brine environments. Such polarizationphenomena are of considerable importance indetermining long time corrosion rates, as con-trasted with Equations 7-D and 7-F whichhold fo r initial corrosion rates. The greater theease with which ei ther corrosion current o r exter -nally applied cu rrent, as in cathodic protection,increases the cathode potential the less the cor-rosion which will occur.The results of the data obtained in thesetests are shown in Figs. 8 and 9. The first da tawere obtained under sulfide conditions at threebrine pH values, 4-5, 6.5-7.5, and 9.5-10. Theseencompass the pH range of all but most acidbrines. In considering the results it is of im-portance that the steel cathode surface was

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    16/22

    494 PROCEEDIN(iS FOUHTH \TORLT) PETROI .Cl7Pf CONGRESS-SECTION II jGcoated with a film of corrosion ferrous sulfide.The d ata of Fig. 8 indicate that cathodewith iron sulfide coatings have low potentials,are cathodic t o steel, and do not polarize readily.The star ting potentials of th e steel cathodesvaried from .650 t o .702 volts. Fig. 9 showsthe s tart ing po ten tid s of steel cathodes in car-bonic acid environments. These potentialsrange from .680 to .739 volts, an average of.705 volts as compared to .671 volts f o r thesulfide condition. While the difference is notgreat, the steel cathode of iron sulfide surfacepossesses the lower potential. The sulfide po-tentials do not represent the true potential offerrous sulfide cathode because the metal sur-face contains anodic as well as cathodic areas.In such cases the measured potential lies be-tween the two single electrode potentials, de-pending largely upon the electrical resistancesof th e anodes and cathodes involved.

    Figs. 8 and 9 show th at much higher currentdensities are required t o alter the potentials ofthe iron sulfide cathode than are required forth e carbon dioxide environment cathode. Twocriteria were used t o select the potentials atwhich the current densities for the differentenvironments are compared. These are baseclupon methods used in electrical protection todetermine the current required for protection.One method requires t,hat current be applieduntil the cathode potential reaches the valueof - .790 volts to a calomel electrode (- 0.850volts to a saturated copper sulfate electrode).Another method is to select the current valuedetermined from an intersection of the tangentsto the two portions of a plot of potential (E)versus the logarithm of th e current ( log I)applied. A tangent intersect results from thefact that a break in the E-log I curve occurs.The data are given in Table IV .

    Similar data are given in Table IV for thecase where an iron sulfide film was present ona steel cathode in a carbonic acid environment.Such cathodes appear at times in pumpingwells producing fluid containing sulfate reduc-ing bacteria. The resulting bacterial actionresults in t he development of iron sulfide cor-rosion products on sucker rods and tubingunder otherwise non-sulfide conditions.

    Data are also given in Table 1V on th e effecton polarization characteristics resulting from

    mechanically removing the iron sulfide film, olarge portions of it, from the cathode.The values of current required for the twcriteria differ from each other bu t give the samgeneral results. The amount of current requireto polarize a sulfide surface varied from 3.2 t11.9 times as much as required to polarize

    TABLE IV. - POLARIZATIONHARACTERISTOF STEELCATHODES IN HYDROGENIJLFI

    AND CARBONICA C I D ENVIRONMENTSA. Current Deriszty Required lo DevelopCathode Potential of 0.790 volts

    pH 4.5 pH 6.5 pH 9.5Environment maisq. f t . maisq. f t . ma/sq. f t .Carbonic Acid - Brine 74 30 11Hydrogen Sulfide - Brine 24 0 140 130Carbonic Acid - Brinewith Iron Sulfide Sur-face 180 15 110Hydrogen Sulfide - Brinebefore Scraping Surface - 137Hydrogen Sulfide - Brine 9 4fter Scraping Surface -

    --

    B. Curretjt Density Required f o r Br c a k inE-Log I CurvePH 4.5 PH 6. 5 pH 9.5Environment ma/sq. f t . ma/sq. ft. m a i m f t .

    Carbonic Acid - Brine 24 10 9Hydrogen Sulfide - Brine 12 0 80 (est.) 52Carbonic Acid - Brinewith Iron Siilfide Sur-

    face 80 48 64Hydrogen Sulfide - Brinebefore Scraping Surface - 8 2Hydrogen Sulfide - Brineafter Scraping Surface ---50

    steel surface in carbon dioxide environmentsIt was also found that a sulfide cathode incarbonic acid environment exhibited this samrequirement of a large current density t o deveop polarization. The da ta furt her show th athe current density required for polarizatiocan be reduced if the sulfide film is removemechanically.I t is of interest to speculate upon the effecof polarization of steel cathodes upon pittincharacteristics of steel under carbonic acid anhydrogen sulfide environments. A rapid, thougnot uncommon, corrosion ra te of steel pipe ioil wells is one year for penetrat ion of pipe o0.250 inch wall thickness. A current outpu

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    17/22

    of 3.5 ma is required for one square inch of 0.250inch wall pipe t o corrode in one year.Table V shows the amount of current re-quired for polarization of steel cathodes inhydrogen sulfide or carbonic acid environmentsof different pH values. If this polarizationcurrent in malsq. in. is divided into the cur-rent output of 3.5 ma, the number of squareinches of cathode required t o permit 1 sq. in.of anode to corrode through in one years time

    pitting factor is high and is approximately15 to 20.These cathode-anode values of 3 or 4 to 1for sulfide attack and 15 or 20 t o 1 for carbondioxide attack correspond to the experimentaldata reported in Table V.Polarization of the cathode hy corrosioncurrent and the maximum pitting factor devel-oped are functions of extreme importance incorrosion attack. For instance, the corrosion

    TABLE . - M I N I M U M ATHODE A N O D ERATIOS OR jT.4RYI?JG CORROSION RATESA . Protpction Current Based on Current Req u i red to D w e l o p 0.790 vol t s ( * )

    Aqueous Condition5 % Sodium Chlorideand 488 to 548 p.p.m.-HCO, Io n5 % Sodium Chloridean d 536 to 599 p.p.m.-HS Ion

    Current Requiredfor Protection Potential Sq . in . R y dto ReceivegH malsa. ft. majsq. in. 3.5 ma4.5 74 0.51 6. 96.5 30 0.21 16.69. 5 11 0 . 0 7 7 45.44.5 240 1.67 2.16. 5 150 1.04 3.49.5 130 0.90 3.11

    B. Pro tec t io n Current Based on Breulc in E-Log I Cirroe

    Aqueous Condition5 % Sodium Chloridean d 488 to 548 p.p.m.-HCO, Io n5 % Sodium Chloridean d 53 6 to 599 p.p.m.-HS Io n

    Current Requiredfor Protection PotentialpH ma/sq. f t . maisq. in.4.5 24 0.176. 5 10 0.0699.5 9 0.0624.5 120 0.836.5 80 0.569. 5 48 0.33

    (* ) Referred to Saturated Calomel Electrode.

    Sq. in. Rcqdto Receive3.5 ma20.65 0 . 756.4

    4.26. 39. 0

    Ratio CIA Reqdfo r Penetrationof 0.250 inchWall in 1.0 Year6. 916.6

    65.42.13.93.9

    Ratio CIA Reqdfor Penetrationof 0.250 inchWall in 1.0 Tear20.65 0 . 756.44. 2

    6. 39. 0

    results. These figures are given in Table Vand show th at while the pH of the solutionaffects the results, the presence or absence ofa sulfide film has the greatest effect. The datashow that at a fixed corrosion rate the sulfidesurface, by virtue of requiring only a smallcathode for maintenance of corrosion rate, willdevelop the greatest number of pits. The car-bonic acid condition in such case leads t o avery high pitting factor. Cases of these twotypes of corrosion are shown in Figs. 1 and 10.Fig. 1 is an example of sulfide corrosionwhich developed in 24 months. The pittingfactor ranges from 3 t o 5. Figure 10 is anexample of corrosion under carbon dioxide con-ditions which developed in 17 months. The

    currents postulated for H2S and CO2 condi-tions are :IR,S= - - 1.1495- og (a s= ) aH+)2] [7-F]and :ICO,= - - .6010- og (aCOs=) (aH+)2]

    1R1R [7-D]

    A t equal pH values and activities of thesulfide and carbonate ions, the relative corro-sion currents are about twice those for sulfideenvironments as fo r carbon dioxide environments.As the corrosion current flows, however, thecathodelanode ratio controls the speed of attack.For instance, consider sulfide and carbon dioxideenvironments a t p H 4.6, and let there be suffi-

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    18/22

    49 6 PROCEEDINGS F O U R T H WORLD PETROLEUM CONGRESS-SECTION I I / Gcient anodic spots on a section of tubing forthe cathode/anode ratio t o be 5 ; using the.E-log I break data, the required minimumcathode/anode ratio fo r maximum corrosionrate is 4.2 for a hydrogen sulfide environment,an d 20.6 for a carbon dioxide environment.The sulfide anode, because of a sufficientlysized cathode, will corrode through in a yearstime. The carbon dioxide anode, however, be-

    5cause of a cathode only __ o r one-fourththe required size, will be reduced t o a currentflow requiring four years for pipe penetration.Thus, because of the difficulty of polariza-tion of sulfide cathodes, numerous pits candevelop rapid ly. In carbon dioxide environ-ments, because of the large cathode area re-quired for high corrosion rates, the presence ofseveral anodic spots closely spaced results ina low cathodelanode ratio and a reduced pit-ting rate.The carbon dioxide corrosion rates obtainedfrom the recombination cell demonstrate thisfact. The cathode/anode ratio of the galvaniccell was about 4 : 1. This is a sufficient ratiofor sulfide corrosion currents t o he high, whichthey were. This ratio, however, is insufficientt o support high corrosion rates in carbon diox-ide environments, and the corrosion rates foundin the experimental cell confirmed this point.

    20.6

    Theory of Sulfide CorrosionThe data developed in these studies haveresulted in a theory of corrosion in sulfide en-vironments. Field experience has demonstra tedthe fact that corrosion under oxygen-free oilwell sulfide conditions is often rapid and serious.Several factors, some of which have been devel-oped here, show why rapid corrosion rates t osteel should be expected.The principal factors are :

    1. The potent ial of the anode is highdue t o low iron solubility.2. The poten tial of the sulfide cathodeis low, approximating for a platinum cathode,tha t of the hydrogen electrode of partia l pres-sure of hydrogen of 1 x 10e5 atmospheres. Thepotential of the sulfide cathode is variable,depending upon the status of the iron sulfidedeposit .

    3. The iron sulfide cathode surface maintains a low potential during corrosion becausof its ability t o receive large quantities of current without appreciable polarization phenomenon.When a steel specimen is placed in a sulfidenvironment the anode has a high potentiadue t o easy contact of the sulfide ions witthe metal anode. The cathode, however, haa relatively high potential at the start becausit is more nearly a non-sulfide than a sulfidcathode at this time and its initial polarizatiocharacteristics are similar t o those in a carbodioxide environment. The net initial corrosiorate is relatively low. A s corrosion proceedsiron sulfide is deposited around and away fromthe anode as a result of continued development of corrosion product. The iron sulfidwhich accumulates over the steel surface begint o act as a cathode because of its low potentialThe gradual development of the iron sulfidcathodic surface permits more and more anodiareas t o become active because of the potentiadifference between steel and iron sulfide anthe small iron sulfide cathodic surface requiret o maintain an iron anode in active conditionAs heavy concentrations of iron sulfide develothere are probably more cathodic surfaces olow potential available than there are anodiareas capable of polarizing them. In othewords, the corrosion system becomes one oanodic control. Thus, corrosion rates undesulfide system start low and increase, wheand if a low potent ial , polarization-resistaniron sulfide cathode develops. While in somcases sharp pitting with a high pitting factodevelops, the tendency is fo r a rapid corrosiorate to develop with the metal surface havina low cathode/anode ratio.

    Correlation of Sulfide Corrosion Theorywith Field ExperienceThere are many instances of corrosion rate

    developing in the field which follow the patterof the theory presented. The theory calls foan original corrosion rate which is relatively lountil the development of an iron sulfide cathodof proper potential and resistance t o polarization, at which time the corrosion rate increaseOne such pattern was obtained during thtesting of a series of 1.5 in. x 4.5 in. x 1

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    19/22

    WALTER F. ROGERS AND J . A . ROWE, JR-CORROSION EFFECTS O F HYDROGEN SULFIDE 497

    gage steel specimens installedin the waste waterdisposal line of a lease in the Winkler Pool,Winltler County, Texas. The produced waterhas the following analysis :pI-1 6.6

    Magnesium 237 Sodium 56G p.p.m.Calcium 768 Chloride 1065 Sulfate 2330 Bicarbonate 345 )Hydrogen Sulfide 300

    The specimens were removed from the testspool at varying time intervals in groups of four.The maximum weight loss from each set of fourspecimens is plotted against time in Fig. 11.I t was found for the first 62 days of test that alow uniform corrosion rate was obtained. Be-tween the 62 and 123 day test the corrosion rateincreased slightly. The specimens removed after180 days had developed serious corrosion losswith the formation of massive soft black ironsulfide.Another such pattern was obtained whiletesting the effectiveness of ammonia treatmentof the vapor space of Tank A, Crane, Texas, acone roof pipe line tank handling sour oil. Tenpounds of anhydrous ammonia were injectedinto the vapor space daily for the first 365 days,followed by 18 pounds per day for 115 days.The effectiveness was determined from the weightloss of 4 in. x 6 in. steel panels placedin the vapor space and withdrawn at intervalsup to 480 days. The da ta are shown in Figure 11.The curve shows that for the first 120 days theweight losses of the test coupons were negligible.Subsequent specimens showed greatly increasedcorrosion rates.A third such patt ern was obtained duringsimilar testing of the effectiveness of anhyd-rous ammonia treatment of the vapor spaceof Tank B, another large tank at Crane, Texas,handling sour oil. The ta nk was treated with15 pounds of anhydrous ammonia per daythroughout the test. The specimens during thefirst 240 days developed negligible weight lossvalues; subsequent specimens which were run upto 480 dayss liowed a greatly increased rate ofcorrosion.The tests made in these two tanks are con-sidered separate and independent ones. Exam-

    The data are shown in Fig. 11.

    ination of the treatmen t method, handling ofthe tanks and their contents during the periodof high weight losses were not different fromoperations prior to the break in the corrosioncurves. The change in corrosion ra te is ascribedt o a change in the corrosion cell cathodes as aresult of iron sulfide formation.

    Other data are available which support thecorrosion theory bu t contain othe r factors whichundoubtedly af fect th e corrosion rate. Forinstance, it is known t ha t the frequent use ofscrapers in pipe lines handling sour oils resultsin l o w internal corrosion rates. While part of thismay result from the continual destruction ofiron sulfide on the pipe surface, another factorof great importance is th at water pockets in thepipe are also destroyed and carried away.Summary and Conclusions

    Studies have been made in th e laboratory ofcorrosion rates of steel by oil well brines con-taining hydrogen sulfide or carbon dioxide underoxygen-free conditions.From the electrochemical theory and theexperimentally determined data, expressionshave been developed for the corrosion rates ofsteel under either of these environments. Theyare: Corrosion Current under Sulfide Conditions:

    JH,S = - - 4.1495 - 0295 log ( a s = ) aH+)2]Corrosion Current under Carbon Dioxide

    1R

    Conditions:1ICO, = -[-..Solo- .0295 log ( a C O j = ) (aH+)2]R

    The relations developed cover the initial cor-rosion rates where the cathodes are under zerocurrent flow. The steady state corrosion cur-rents are reduced from the initial rate by polar-ization of the cathode from corrosion currentflow.Sulfide surfaces were found t o resist polar-ization, and as a result a small sulfide cathodecan support a relatively large anodic area at ahigh corrosion rate.Carbon dioxide environments develop cath-odes which polarize relatively easily. Thisresults in low corrosion rates where the cathode/

    Proceedings 4th W . P . C . - Section 11 32.

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    20/22

    49 8 P R O C E E D I N G S FOURTH WORLD PETROLEUM CONGRESS-SECTION II /Ganode ratio is small. If th e cathode/anode ratiois sufficiently large, then carbon dioxide corro-sion rates become very high. Where rapid cor-rosion rates are obtained in both environmentsthe sulfide pit ting factor is small an d the carbondioxide pitting factor high. The fluid p H is ofmuch less importance in sulfide corrosion thanin carbon dioxide corrosion.

    A theory has been developed covering thecauses of high corrosion rates in sulfide environ-ments. This theory considers the prime fac-tors in sulfide corrosion to be a high anode po-tential and a relatively high cathode potentiala t t he start of corrosion. The cathode duringthis stag e primarily acts as a hydrogen electrodeand the corrosion rate is relatively low. Ascorrosion continues the iron sulfide corrosionproduct becomes the cathode because of it svery low potential. Since a n iron sulfide cath -ode is dificult to polarize, as well as havinga low potential, a high corrosion rate is develop-ed a nd maintained. Sulfide corrosion rates,therefore, go through two stages. The first, orlow rate, occurs initially during the develop-ment of th e sulfide cathode, arid the second,or high rate, afte r the low potential, polarization-resistant cathode is formed.

    Data are presented showing the results o fcoupon testing under sulfide environments whichsubstantia te this theory of sulfide att ack . I tis believed th at the use of mechanical or chem-ical means to remove cathodic iron sulfidecorrosion products will reduce t he ra te of cor-rosion.

    AcknowledgementThe authors wish t o tha nk the Management o l

    Gulf Oil Corporation, Houston Production Divi-sion, for its kind permission to publish this report.

    Bibliography(1 ) H. R. Copson: Literature Survey on Corro-sion in Neutral Unaerated O il Well Brines,Corrosion, 7, 123 (1951).( 2 ) W. F. Rogers, W. A. Shellshear: Corrosion of Steel by O il Well Waste Waters, Ind. Eng.

    Chern., 29, 160 (1937).(3 ) E. G. Woodruff, R. L. Ginter, L. G. E. Bignell:O il Field Steel Tank Corrosion, Private Pub-lication, Tulsa, Oklahoma, 1926.

    ( 4 ) S. Ewing: Electrochemical Studies of t h e Hdrogen Sulfide Corrosion Mechanism, SouCentral Regional M e c t . NACE, Tulsa, Okhoma, Oct., 1953 (Unpublished).(5) W. F. Rogers: Influence of O i l in SubsurfaCorrosion 01 Oil-Well Equipment. Oil G

    JournaZ, 48 , 32, 73 (1949).(6) U . B. Wescott, C. H. Rowers: Economical Slection ot Sucker Rods, Trans . A. I .M .M.E1 1 4 , 177 (1935).(7 ) L. W. Vollmer, B. B. Wescott: Effect 01 Hdrogen Sulfide on Wire Rope, Petr. Me

    Engr., 5 7 - 6 1 , Oct. 6-8 (1930).(8 ) Condensate Well Gorrosion , 143, NatuGasoline Association of America, Tulsa, Okhoma, 1935.

    W. F. Rogers, J . A. Rowe, P. J. Kahsh: Laoratory Apparatus for Studying Oil WSubsurface Corrosion Rates and Some Rsults, Corrosion, 9 , 25 (1953).Glasstone: Textbook of Physical Chemistry, D. Van Noslrand Co. Inc. 1940.

    This paper was presented on June 14th, 19by M. -1. H A N N A (G i t l / O il Corp . - HoustoT e r a s , U s A ) o n behalf of the authors.Discussion

    ,I.I . J,. C ~ L J G T T O I SS.N . P. A . Pari s , - FJe fais simplement deux remarques pratiqusur des questions particulikres relatives aconclusions de ce rapport tl ans lensemhle et r h e m e n t interessant. Les auteurs concluequun nettoyage mecanique o u chimique ayapour but denlever les produits de corrosion p132s devrait recluire le taux de corrosion. Mremarques sont les suivantes:On sait, que lorsquon conduit en laboioire des experiences de corrosion soils tensioplus on augmente la frequence des nettoyagplus on augmente la rapiditi! des ruptures.Deuxikmement: on sait 6galement qquand un puits produit du qaz nature1 connant de lhydrogkne sulfure & haute pressiplus la vitesse d u gaz est rapide et plus la corsion du tubing augmente; alors que le nettoyades produits de corrosion doit aussi augmenavec la vitesse du gaz.J e demande au x auteurs si ces deux marques ne sont pas en contradiction avec conclusion citee plus haut.

    1,. Z A M B O N I Nalco Italiana S. p . A . -The method which has been developed lllessrs. Rogers and Rowe for determin

    (9)

    (10)

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    21/22

    WALTER F. R O G E R S AN D J. A . ROWE, JR-CORROSION EFFECTS O F HYDROGEN S U L F I D E 4the corrosion rate in the well by the brines isvery interesting since it can really give immediateinformation on corrosion as a function of manyseparable variables.A different method for evaluating the cor-rosion by well brines which can easily be ap-plied to field operation, has been proposed andtested by my company; it consists in deter-mining the iron concentration in th e oil phase ofthe !iquid produced. In fac t, corrosion decayproducts, mainly iron oxide and sulfide, arefound to be easily dispersible in the oil, at con-centrations which seem to be quite representa-tive of t he corrosion process in general.The analytical method is suitable for fielddetermination, since the final readings are madeonly with colorimetric and volumetric proced-ures. Wit,h reference to t he recommendationof the authors. that mechanical and chemicalmeans be used to remove cathodic iron sulfidecorrosion products in order to increase cathodepotential and consequently to reduce corrosionrates, may I recall that a line of corrosion inhib-itors for producing wells have been recentlydeveloped.These inhibitors, in liquid or solid form(oil-and-water-dispersible) are fed into the pro-ducing well annulus or directly into the string.It has been ascertained that their action istwofold; in fact they have a detergency actinnfor sulfide deposits and a wetting action be-tween metallic equipment and oil phase: whilet,he first action reduces the corrosion electro-motive forces, the second one increases theohmic resistance of t he corrosion cur rent paths.M. A . HANNA epl ies . I feel very sorry notto have been able to answer the questions on th epresentat ion of th e paper I have given, but itis completply out of my field and I beg yourpardon.W r i t t e n r e p l y *

    J. A. ROWE r. Tn regard to the first re-mark, Mr. Cauchois is correct in his statementregarding cleaning if the corrosive agent isSent in by the authors after the Congress.

    ____

    carbon dioxide alone. However, thi s does napply to hydrogen sulfide-steel systems sinthe corrosion reaction driving force is the cathodiron sulfide which is formed as a result of tpreliminary solution of iron. In practice, manhydrogen sulfide systsms are neutral o r evalkaline. Therefore, the initial corrosion ratdue to hydrogen ion concen1,ration alone. mhe very low. Only after a tight, layer of irsulfide has formed on the steel surface does tcorrosion and pitting attack become greataccelerated. Hence the reduction in corrosira te by mechanical or chemical removal of tsulfide scale.As to the second remark, an increased flivelocity mould do more t o accelerate corrosiby virtue of increased depolarization than would to reduce corrosion by v irtue of irsulfide removal. The iron sulfide scale is geerally quite firmh- adhered to the metal suface and is removed only by a very efficiemechanical scraper or bv chemical means suas acidizing.When high partial pressures of hydrogen sfide are involved the pitting attack is frequentaccompanied or superseded by blistering apickling due t o the high hydrogen ion concetration at, the metal surface.Tn regard t o Mr. Zambonis first remat,he iron determination method developed his company can be a very useful tool in evuating corrosion rates of steel systems carryioil and water. I t is especially of interest sulfide systems where the corrosion producare insoluble in water and apparently dispermore uniformly in oil.

    In regard to reduction of sulfide corrosirates by chemical means, the authors feel ththose chemicals which promote oil wetting anor which adsorb on the metal surface give goprotection by virtue of p reventing water froproviding a low resistance path between tcathodic and anodic areas. However, it hnot been their experience that organic corosion inhibitors wjll remove an appreciabportion of that tight iron sulfide film whiaccelerates corrosion.

  • 7/29/2019 Corrosion Effects of Hydrogen Sulphide and Carbon Dioxide in Oil Production

    22/22