Control corrosion in washwater systems

ause equipment failures

Intermittent or continuous water iujection?Water

cools to the salt-deposition temperature. Unforthe resulting aqueous solution can be very corunits are operating at rates above may be a good choice. If continuous washwater is not used, the NH4C1 andlor NH4HS may build up in the reactor pressure drop and lower heat transfer. These dry salts are not considered as very corrosive. When an intermittent
'.

hydroprocessing reactors by converting feed. Hydrochloric acid can enter the unit

Where should the water be injected? Wakr should be injected into the reactor effluent upstream of where the effluent is cooled down to the deposition temperature.

, ,

If chlorides are not a problem, the normal injection

Fig. 2. Typical reactor effluent.configuration with hot separator.

Fig.. 3. Common washwater recyc

point is just before the effluent air cooler (Points A and C in Figs. 1,2). Typically, the effluent is cooled to 275OF to 400F'. This temperature may be even higher if a hot separator is used. The injection point should always be downstream of the hot separator. Some engineers like to design an intermittent injection ahead of the last exchanger before the air cooler (Points B and D in Figs. 1,2). However, this method can impact the water rate required ifthe system is designed for a higher injection temperature. If parallel exchangers are used before the air cooler (Point B in Fig. I), exchangers czin be cleaned intermittently without increasing the design washwater rate. Process conditions. After the washwater injection, the process stream is in three phases: vapor, liquid hydrocarbon and an aqueous phase. Normally, the vapor phase will be 70% to 90% of the total stream volume. With a high-pressure hot separator configuration, only a small amount of hydrocarbon liquid is present. If there is no hot separator, then the stream (typically 60%-85%) will be mostly hydrocarbon liquid by weight. Hydrogen sulfide by itself is not very soluble in water; NH3is much more soluble. When large quantities of N & l and H2S are present in the vapor phase and water is introduced, t h e NH3 dissolved in the water causes a n electrolytic reaction and an approximately equivalent number of moles of H2S a s moles of NH3 will be absorbed into the aqueous phase. Be careful when using simulation software to calculate sour water composition, because not all simulation programs have correlations that correctly model this relationship. ' \ have excess aqueous water, the hydrocarbon liquid I, and vapor must be saturated with water. Normally, only a small quantity of water will be dissolved in the hydrocarbon liquid. 9% have free water, calculate how much water is required to saturate the vapor phase. This calculation can be done by simulation. Alternatively, the Appendix shows how to estimate saturating the vapor phase by water injection.
Mditional design guidelines. Construction materials

since it has the highest likelihood of maldi water, plugging or local high-velocity proble Washwater source. The two sources use water are steam condensate and stripped sour w refiners prefer steam condensate. Stripped so often recommended because i t contains no diss gen. However, most refiners elect to use s t sate due to its availability and reliability. Ca taken to ensure that the washwater is oxyg than 15 ppm to 50 ppm). Most licensors spec washwater is also free of cyanides.

engineers defined the NH4HS concentration by measuring the total concentration of NHs the water and summing them. Others have rep lines based solely on the NH3 content. Many studies indicate that carbon steel c related to the NH4HS concentration. Most lic industry experts have threshold guidelines for,

cient washwater is injected and some wat the aqueous phase, then the resulting aqueo

for the effluent system after the water injection point is a n economic decision. The cost of alloy systems have a trade-off in reduced inspection costs and reduced risk. Materials are often selected based on the perceived severity of expected NH4HS corrosion and some of the industry guidelines are dependent on the material Downstream of the washwater injection, modern designs use either carbon steel, Alloy 825 (UNSN08825), or Alloy 2205 (UNS S31803) materials for exchangers.
-

Superficial fluid velocity. I t is well acc the superficial, fluid velocity affects the rate 0 solution corrosion. The normal velocity for materials is 20 ft/s maximum. For alloy mat is normally used, but this is considered a les guideline. The actual threshold corrosion vel alloys has not been determined. In air cooler is often used as the minimum velocity to ens tribution. Normally, minimum velocity is no since 20 ft/s is a very low velocity for these rn streams. Superficial velocity for a multiphase be easily calculated by taking the total stream flow and dividing by the cross-sectional area. : Piping design. The goal of the washwater,
-

88

KYDROCARBON PROCESSING / JUNE 1997

&rve showing simulated NH4HSconcentration with tem-

air cooler i s to get an even distribufluid and the washwater through each air e. To accomplish this goal, these design guidelines distribution piping-commonly called

Fig. 5. Nomograph relating NH4CLdepositiontemperatures with varying CI-concentrations.

water i

of the two guidelines is controllingfor a given application, since it determines which process variables are most critical and what design alternatives are permitted. If the NH4HS concentration is controlling, the only way k ~ a r a t flow orifice for each water iniection ~ o i n t . to decrease the amount of required water is to reduce the e NH3 content in the reactor effluent. This is not normally done, since it would require significant modification to &high salt concentration could build up. ~ h i s j s the unit or reactor conditions. Operating engineers must &y important in units using an intermittent water be aware of the feed nitrogen levels. Changing from design &ce the salt concentration during washing will to higher-nitrogen feeds should not be done without conBy be very high. sidering potential corrosion implications. b of thumb. The washwater rate is ~ r o ~ o r t i o n a l If the second specification is controlling, refiners can optimize the process condition at the injection point and the washwater rate. For example, depending on the calculated NH4HS deposition temperature, it may be possible to decrease temperature a t the injection point, thereby &ed solely by a "rule of thumb" basis. decreasing the washwater requirement. Another option may be to install a washwater recycle pump, taking some K," factor. In the late 60s and early 70s. R. separator water and pumping it back to the injection point. This requires installing a high-pressure, low-head pump (Fig. 3). The concentration of the separator water will remain unchanged, but significantlymore water is available at the injection point to saturate the vapor and ensure gon charts).-The "Piehl KPnfactor is: an aqueous phase is present. This scheme decreases the ! I amount of make-up water, and reduces the quantity of = (mol% NH3 in reactor effluent) X (mol % produced sour water. Both reductions are benefits. : H2S in reactor effluent) (1) Process conditions at water injection point. There are two theories about the NJ&HS solution conditions at 1's proposed guidelines, Kp less than ervice, above 0.15 is severe service and the injection point and in the downstream heat transfer equipment: severe service. At the injection poin<(which is typically 300Fto s. When applying industry design r injection amount is determined 400Fequilibrium temperature), essentially all vapor phase NH3 will be absorbed into the aqueous phase, creating a high NH4HS solution or ntration in the separator Only a small amount of NH3 will be dissolved,creating a modest NH4HS concentration solution. required to ensure that the eous phase" guideline is met. The two theories differ only in the NH4HS solution con01' the washwater rate; the centration at the injection point and temperatures around be exceeded. It is important to understand which 150F 200F. Below this temperature, both theories to
HYDROCARBON PROCESSING 1 JUNE 1997

89

agree that essentially all of the NH3 is absorbed into the aqueous phase. T h e o r y 1-All of t h e ammonia i s in the w a t e r phase. Essentially, if all of the NH3 is absorbed into the water as N a H S at the injection point, then the NH4HS concentration can be easily calculated by ratioing the amount of NH3 to the amount of aqueous water at the injection point or by determining the amount of NH4HSin the separator water and dividing injected water remains in the.aqueous phase, and almost all of the NH, is absorbedas NI&HS. The NH4HS con-

cant. There is

drum conditions, if the H2Sconcentration i is larger than the NH3 concentration, the

injection point. If Theory 1is true, the calculated NH4HS concentra-

separator. Plant data and licensor have co the calculated NH4HSconcentratio

sour water produced. Theory %most of the ammonia stays in the vapor phase. Two major commercial simulator programs support Theory 2. A simulation analysis indicates that the NH4HS concentration at the injection point is typically less than in the downstream separator drum water (Fig. 3). More or less washwater has no impact on the calculated NH4HS concentration. More NH3 does affect the NQHS

ing a hydroprocessing effluent system. This is change. In the past, some licensors and ope the K, factor to determine materials and was tern requirements. could convince an engineer to install an in the reactor loop. An amine absorber can

In a survey, about N of the data points indicated maximum corrosion a t 320For above, but about 9 indicate 6 maximum corrosion at 225OF or less. In addition, there are reported failuresjust upstream of the separator drum, which would have a much lower NH4HS concentration than upstream of the air cooler if Theory 1is true. If Theory 2 is accepted, is the "percent aqueous solution rule" invalid? Empirical evidence shows if all of the water is vaporized, the point where water first condenses will be very corrosive. However, this corrosion may not be caused by NH4HS, but by NH4Cl and other salts in solution. Hydrochloric acid is much more soluble in water than H2S,so the very first condensed liquid could be very high in NH4Cl concentration. Because only a small amount of chlorides are present in hydroprocessing effluent streams, very little water is needed to drive the NH4HS concentration down to nil, making the NH4HS corrosion controlling. Probably, significantly less than 20% aqueous water would meet this requirement; however, the 20%calculated aqueous water guideline is probably a good rule. Future unit expansion or different heat exchanger profiles can very quickly consume this 20% safety margin. In addition, once enough water has been injected to satu-

are many reasons for using an amine ab hydroprocessing unit reactor loop. Besides im tor performance, corrosion in the feed prehe be decreased. It can lower the temperature NH4HSbegins to deposit. However, has no noticeable impact on NXHS c tor effluent. Rule of thumb. The required washwater ra determining the NH4HS concentration in the water and the water quantity remaining aque injection point. If the NH4HS concentration is then calculationsindicate that 1gpm/1,000 bpd 4 wt% NH4HS if the net reactor NH3 yield is based on fresh feed. If the NH3 yield is greater; NH4HS concentration will be proport NH3 yield can vary widely depe gen level and the denitrification wt% yield might be a typical value, values si higher and lower are also encountered. If t h was 0.4 wt% and the water injection rate was bpd, the calculated NH4HSconcentr water is 32 wt%, which is obviously excessive. Ifthe of water remaining aqueous" specification is co

then the water rate depends on process conditions, not feed

water rate is presented in gpm/1,000 b

Process conditions Case 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Hot sep. present ? N N Cold sep. pressure psig 900 900 900 900 900 900 900 900 900 900 900 1,800 1,800 1,800 1,800 500 500 900 900 1,800 1,800 500 500 500 900 900 900 900 900 900 Process temp., O F 400 300 500 400 300 500 400 300 500 400 300 400 300 400 300 400 300 400 300 400 400 400 400 500 500 400 300 500 400 300 washwater temp., "F 200 200 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 200 200 100 100 100 100 100 100 100 100 100 100 100 Recycle rate, 2,000-1,000 scflbbl Sep. vapor recycle rate, scffbbl 2,000 2,000 2,000 2,000 2,000 . 1,000 1,000 1,000 500 500 500 2,000 2,000 1,000 1,000 1,000 1,000 1,000 1,000 1.000 500 1,000 500 500 1,000 1,000 1,000 500 500 500 Recycle rate, 1.000600 scflbbl gpmf 1,000 bpd 1.7 0.7 3.0 1.6 0.6 2.2 1.1 0.4 1.7 0.8 0.3. 1.2 0.4 0.8 0.3 1.5 0.6 0.4 0.2 0.3 0.1 0.5 0.3 0.4 0.6 0.4 0.2 0.3 0.2 0.1

N
N

N N N
N N

N N
N

N N N
N N Y, 500F Y 500F , Y, 500F Y, 500F Y 500F , Y,500F Y 500F , Y 5OO0F , Y 500F , Y 500F , Y 500F , Y 500F , Y,5OO0F

Ratios of water rates Water inj Proc temp., 5oO-400 temp. 2W10O0F "F 1.06 1.03 1.88 Proc temp.,

Case 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

400300 "F

Pressure, 900-1,800

psis

2.48 2.00 2.80 2.1 5 2.96 2.68 2.82 1.62 1.08 1.06

1.34 1.43 1.62 1.30 1.40 1.48

1.36 1.47 1.40 1.41

1.48 1.55

1.62 2.05 1.64 2.03 1-60 1.32

les the net amount of NH3, RCl produced. Reactor conn a hydrocracker is normally 00% for nitrogen and sulfur. rotreater, the conversion can blv depending on unit feed@&d the denitrification con$& 30% to 65%. The total con,,-,n.s of these chemicals in the Eimuent is equal to the net 8~;roduced the reaction and by P l e gas. &most all refinery hydropro-

nt on the NH3in the effluent water injection rate. ent vapor rate. The reactor @ vapor rate a t the injection L very important variable. I t $nes how much injection water F e d to ensure 20% to 25% of &r in the aqueous phase. ReacGent is a fairly dry stream; only B amount of water is carried kith the effluent stream, some rater is required to saturate the ihase. Water will also be disi'ihto the hydrocarbon liquid
I exist until the vapo; phase 1 ted, assuming complete equis reached. Therefore, if the fate is increased by 25%, then h e water is required to satu-

I
I

arrangements with major customers

Please write us for our full color brochure, comprehensively describing KEMCO's capabilities and recently completed heat transfer euuipment

$s. phase, if the vapor rate is $ed by 25%.

Z9ffluent vapor rate is fairly the separator's vapor rate, $.susually measured or calcu,From Table 1, if the "water Png aqueous" specification is &!ing, doubling the separator ... . $gate will cause a 30% to 50% 9 in the required water rate if 6.' ,..separator is present. Nearly Ethe required washwater rate K* if a hot separator is present.
6: 7 .

important parameter when ?$.ing how much water injecCircle 11 1

93

tion is needed for 20% water to remain in the aqueous phase. From Table 1, if the process temperature before injection increases by 100F, required the washwater rate to keep the same % water remaining in aqueous increases by a factor of 1.5 to 2.5 times. The unit heat integration and exchanger performance can affect the washwater system. If the design is controlled by the "percent water remaining aqueous" specification, then the operator must be aware of how changes in the process temperature will affect this basis. Changes in the unit throughput, feed temperature or heatexchanger performance can all affect the process temperature and greatly impact washwater operation. Many refiners have increased unit throughput without considering the effects on the washwater system. Separator configuration.Hydroprocessing units normally fall into two broad categories:Units that have a rector effluent hot separator and those that do not. In units using a hot separator, the reactor-effluent liquid is removed a t a particular point in the effluent cooling train (typicallybetween 450Fto 550F).This improves the

unit's overall energy efficiency and minimizes heat transfer equipment. In these units, normally no or avery small NH4HS guideline and that & amount of hydrocarbon liquid is ahead of the washwater injection point. Thus, the water injection can have a signifi- very high percentage of the cant quench impact, sometimes cooling the effluent by 150For more at that point Therefore, less water is required more critical, since less vat; to satisfy the "water remaining aqueous" guideline. The data in Table 1show the impact by using a hot separator. At the same water. As shown in Table 1,$ injection conditions, three to five times the temperature from 1 0 0 " ~ ~ ~ as much water is required to maintain same water amount remaining aqueous without a hot separator than with one. Without a hot separator, t h e entire reactor effluent must be quenched by the water injection. Since ~ e r a t u r e causes the washwid this mass flowrate is relatively high, water injection cannot cool the effluent very much (25Fto 75F).However, with a hot separator, most of the effluent stream has been removed. so the water can quench by a much This greater degree (100Fto 200F). equilibrium temperature contributes to the large difference in washwater then a change in fresh fee8 required.to create aqueous water a t the injection point. - the unit feedrate can also: impact on the process tempei the injection point. ,* If the aqueous water at4 specification is controlling,j charge rate can still a 4 required washwater rate ev:

not altered, a change in the f drate can still alter the equ reduce the total effluent st$ same temperature. Also, mo will be required before ex< water is present. ~alculati: cate that, for a constant tota amount of water required to. the vapor. If a hot separator this effect is minimal.
Unit specific studies. To h ators understand the impact variables on potential NH& sion, graphs can be prepared t these effects when key vaq washwater design are alter1 are several examples:

Circle 112
94

r;;lbosition temperature. Fig. 5 depicts a nomovm--

F:

gadfresh feed ratio. This chart is a re-statement

system and process variables. By understanding'the impact of the process conditions, this unit can be optimized in the design phase and ensure proper system operations as conditions change.
ACKNOWLEDGMENT Based on a paper originally presented at the 1997 Gulf Publishing Co.lHydrocarbonProcessing Process Optimization Conference and Exhibit, April 8-10,1997, Houston, Texas.
LlTERATURE CITED NH.,HS Deposition Chart, U.S. Bureau of Mines Bulletin 406,1937, p. 56. NH&l Deposition Chart. Journal of Chemistry and Physics, 1944. p. 318. Piehl, R L., "Survey of Corrosion in Hydmracker Emuent Air Coolers,"Maferials Performance, January 1975, p. 15. Sheri-er, C. and M. Dumue, m a t e and Resid Hydmproeessing: Copingwith Corrosion yith High Concentrationsof Ammonium Bisulfide in the Pmcess Water,"MaterialsPerformance, November 1980, p. 25. . and Damin, D. G . J. D,McCoy, %vention of Corrosion in Hydrodesulfurizer Air Coolers and Condensers,"M a t e d Performance, December 1978, p. 23. Pie& R L."Camsion by SuUide-Containing Condensate in Hydrmacker EffluentCoolem." API 33rd Midyear Meeting, May 1968. Alvarez, A M. and C. A Robertson, "Mataials and Design Considerationsin High Pressure HDS Effluent Caolem,"Materials Performance, May 1993, p. 16. . Shargay. C. A, A J. B a g d e a n , J. W. Coombs and W KJenkins, "Corrosion in Hydmpmeessing Units," Corrosion in the Oil Refining Industry, Eds. L Kaley, J. E. . Feather, N.Coble, R. Strong, Houston, NACE 1996).pp. 1011-10118.

b l &deposition chart1 modified to reflect unit

&rs such as configuration, operating pressure, ratio, etc., for a particular unit. These assump-

aversfon factor can be derived to account for chlo,the make-up gas and fresh feed. This chart was id based on process simulations and spreadsheet ~ o n for the various cases. A similar chart could s pared for NH4HS deposition. ater remaining aqueous. This nomograph prepared to explain the interaction between the &r remaining aqueous and washwater rate and

'

The author

'operating pressure, recycle rate, effluent distillas involved in design of new or revamped units on of existing u i sshould have a basic undernt design guidelines relating to the waterwash

James Turner is a Principal Process Engineer at Fluor Daniel, Inc., in Houston. He has over 15 years of industry experience, with most of it in process design and optimization of new and revamped Refinery Units. Mr. Turner has extensive experience in the design and troubleshooting of hydroprocessing units. He has a BS degree in chemical engineering from Texas A&M University and is a member of AlChE and is a registered professional engineer.

APPENDIX
deposition charts. To use the chart for NH4HS ion, calculate the NH4HS dissociation constant, product of the partial pressure times the pressure in the reactor effluent:

Estimating amount water to saturate the vapor phase. The flash calculation determines how much water

partial pressure of each component, i, can be cal)y taking the moles of the component in the vapor hided by the total moles in the vapor phase mulb the absolute pressure in the process. y
2'.

-hi vapor phase)l(nhtd vapor phase) x P

(3)

@ ~ the is i

he same procedure can be used to determine the

mole fraction (in the vapor) of the ith com-

$peposition temperature. g,pet reactor yield can be calculated from the trogen level &d the percent denitrification. If a or catahst vendor is involved, they wili normally . @,the reactor net yields.

is needed to saturate the vapor phase and is normally done by a process simulator. However, the required rate can be hand calculated. This calculation is only an estimate and may differ by 5% from the value calculated by simulations. Estimate the equilibrium injection temperature. This temperature is typically 270F 350F. ~ l l to It typically be 30F 100F than the process temperato less ture before injection if a hot separator is not used. I f a hot separator is present, the temperature may be 200F or more than the process temperature. Using steam tables, determine the steam saturation pressure at the above temperature. Estimate the molar flowrate of hydrogenhydrocarbon in the vapor phase a t the injection point. (This is normally very close to the vapor flowrate off of the cold highpressure separator). Use Eq. 4 to estimate the number of water moles required to saturate the vapor at given conditions:
Washwater molar flow = Fcl x vapor molar flow HC x

m3

e N& content in the reactor effluent is the same in the reactor. Essentially all of the
where: molar flow HC= molar flow of Hz and hydrocarbon in the vapor phase a t injection point. PSm S m =Absolute pressure of saturated
HYDROCARBON PROCESSING1 JUNE 1997

95

steam a t process injection temperature Ps,,,, =Absolute pressure of process a t injection point Fc is defined by these values (interpolate for other values):

PsATSTM for 3 10F 78 psia is

Vapor flow at injection point-assume separator vapor rate: Flow = 1.05x 90 MMscfd x
(I mole1hr)

5% g~

Operating pressure, psig 500 1,000 1,500 2,000

9,108 scfd

Fc
1.1 1.2 1.3 1.4

= 10,375 rnolelhr

vapor rate

So the water rate to saturate vapor is: Water (molelhr) = 1.2 x

(1 - 7811,015)
+

This calculation yields estimated required water rate to saturate the vapor phase. To estimate how much water would be required to provide 20% excess (20%of the washwater remaining aqueous), simply multiply the calculated water rate by 1.2.
Estimate NH,HS concentration in separator water. Eqs. 5 and 6 can be used to estimate the wt% NH4HS in the separator water for all cases where the net reactor H2S yield is greater than the net NH3yield:

Assume 5% of the water will be dissolved in tl carbon liquid, so for 20% excess, multiply by 12 . Water rate = 1,034 x 1.25 = 1,295 moVhr = 4i By simulation, the amount of water required 20% aqueous is 47 gpm, the same as the calculai Rearranging Eq. 6 yields: 2

6
: ;

Calculate water rate to yield 4 wt% N ~ H S ~

wt%NH4HS in solution = 511 14 xWf x FN x CN ~ ~ 1 0 0 100 x 100 x WWR x 500

(5) In this example, the water rate is set by the & concentration mideline. -?d 3

where:

Wf= Mass flowrate of unit feed, lblhr

FN=wt% nitrogen in unit feed

CN=% Denitrification in reactor (net nitrogen

conversion)

WWR= washwater injection rate, gpm This calculation assumes that all of the NH3is absorbed as NH4HS in the water a t separator conditions. This should be a reasonable assumption, as plant data and simulation results indicate that typically 99'Z&t of the NH3 will be dissolved in the water.
Example 1 Calculations for a FCCU feed hydrotreater. : For this example, assume these conditions: + Assume minimum 20% aqueous water a t injection + Assume maximum 4 wt% NKHS in separator sour water A hot separator is not used 50,000 bpd feed (670,000lb/hr) 3% sulfur in feed + 0.1% nitrogen in feed 90% desulfurization + 50% denitrification + 1,800 scfibl recycle rate (90 MMscfd separator vapor rate) Before water injection point, process temperature = 350F,process pressure = 1,000 psig

Example 2: Calculations for a FCCU feed hydi For this example, assume these conditions: 1 Assume minimum 20% aqueous water at 3 Assume maximum 5 wt% NH4HS in sepm; water A hot s e ~ a r a t ois used r
+

0.05% nitrogen in feed 90% desulfurization -3 4 + 50% denitrification p + 1,800 scfibl recycle rate (90 MMsefd sepa14 + rate) 3 Before water injection point, .$ process temperature = 480F, process pressure = 3
+

Estimate water to produce 20% aqueous.::

about 310F.Therefore, the estimated water d same as in Example 1. By simulation, the a3 water required to produce 20% aqueous is 45
Calculate water rate to yield 4 wt% N ~ H S ;

Using Eq. 7 yields:

Estimate water requiredto pmduce 20% aqueous. Eq. 4 can be used to estimate the amount of water required to saturate the vapor phase. The process temperature before injection is 350F.For this calculation, assume that at the injection point, the temperature is quenched to 310F.
96
ElYDROCwON PROCESSING lG/JUNE 1997

WWR (gpm) = 7.3 x = 24 gpm

x 670,000 x 0.05 X

In this example, the water rate is set by the" water remaining aqueous" guideline. The make-up water make could be optimized if the heat integratio] were modified or if a water recycle pump were in<