Documente Academic
Documente Profesional
Documente Cultură
Fuel
journal homepage: www.elsevier.com/locate/fuel
a r t i c l e i n f o a b s t r a c t
Article history: Different approaches to calculate the hydrogen consumption during hydrotreating of petroleum fractions
Received 14 July 2010 are reviewed and discussed. Experimental information about the hydrotreating of various distillates
Received in revised form 18 November 2010 (naphtha, light and heavy gasoils, atmospheric and vacuum residues) obtained in a bench-scale unit at
Accepted 19 November 2010
typical operating conditions and commercial catalysts was used to determine the hydrogen consumption
Available online 5 February 2011
from global hydrogen balance and from the hydrogen balance in gas streams. The comparison of exper-
imental hydrogen consumption calculated with the different approaches indicates that there is not a
Keywords:
method that can predict hydrogen consumption with good accuracy. For quick calculation some correla-
Hydrogen consumption
Hydrotreating
tions can be used, but the error can be as high as 15%. The common approach used in reneries, which
Oil fraction does not take into account the dissolved hydrogen in the hydrotreated liquid, can give deviations up
to 6% with respect to the experimental value. To account for this dissolved hydrogen into the balance
all the correlations reported in the literature are suitable.
2011 Elsevier Ltd. All rights reserved.
0016-2361/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2010.11.047
3594 L.C. Castaeda et al. / Fuel 90 (2011) 35933601
Nomenclature
Gas0 total amount of gas feed (kg) N total nitrogen content (ppmw in Eq. (9), wt.% in Eq. (19))
Liq0 total amount of liquid feed (kg) O total oxygen content (wt.%)
Gas1 total amount of gas product (kg) Br bromine number (gBr/100 g)
Liq1 total amount of liquid product (kg) MA monoaromatics content (wt.%)
(H2)0 total amount of hydrogen entering the reactor (kg) BT benzothiophene sulfur content (wt.%)
(H2)1 total amount of hydrogen leaving the reactor (kg) DBT dibenzothiopene sulfur content (wt.%)
H0Gas total amount of hydrogen fed to the reactor (kg) PNA poliaromatics content (wt.%)
H0Liq total amount of hydrogen contained in the liquid TriA triaromatics content (wt.%)
feedstock (kg) DiA diaromatics content (wt.%)
H1Gas total amount of hydrogen in the gas reactor product (kg) Mw molecular weight
H1Liq total amount of hydrogen contained in the hydrotreated q density at 20 C (g/mL)
liquid product (kg) qL liquid density (g/mL)
DHgas difference between the amount of hydrogen entering Hdiss hydrogen solubility (scf/bbl)
and leaving the reactor determined by mass balance in Hpurity hydrogen purity (wt.%)
the gas streams (kg) PH hydrogen partial pressure (MPa)
H1diss amount of hydrogen dissolved in the liquid product (kg) GO gas/oil ratio (mL/mL)
Hcons total real hydrogen consumption (scf/bbl) T temperature (C)
H1HC-Liq amount of the hydrogen contained in liquid (kg) H Henry coefcient
HHC-Liq amount of the hydrogen contained in light c H2 concentration of hydrogen in solubility equation
hydrocarbons in liquid (kg) (mol H2/m3 oil)
H1H2 S amount of the hydrogen contained as H2S in gas (kg) gN molar gas volume at standard conditions (mol/L)
Hchem hydrogen consumption for consuming chemical k solubility coefcient ((NL H2)/(kg oil) (MPa))
reactions (Nm3/m3 oil) n catalyst remaining relative activity
HHDS hydrogen consumption in hydrodesulfurization reac- e catalyst bed void fraction
tions (scf/bbl in Eq. (20), Nm3/m3 oil in Eqs. (8) and VR catalyst bed volume (m3)
(15)) QL oil volumetric ow rate (m3/s)
HHDN hydrogen consumption in hydrodenitrogenation reac- CH remaining hydrogen demand of oil (mol H2/m3 oil)
tions (scf/bbl in Eq. (21), Nm3/m3 oil in Eqs. (8) and CT-cons total hydrogen consumption (mol H2/m3 oil)
(15)) HT-cons the experimentally measured hydrogen consumption
HHDO hydrogen consumption in hydrodeoxygenation reactions (mol H2/m3 oil)
(Nm3/m3 oil) k intrinsic reaction rate constant ((m3 oil)2/[(m3cat )
HHDM hydrogen consumption in hydrodemetalization reactions (mol H2) (s)])
(scf/bbl) a reaction order
HHGO hydrogen consumption in olens saturations reactions
(Nm3/m3 oil) Subscripts
HHDA hydrogen consumption in aromatics saturation reac- f feed
tions (scf/bbl in Eq. (23), Nm3/m3 oil in Eqs. (12), (17), p products
and (18))
HHDC hydrogen consumption in hydrocraking reactions (scf/bbl) Superscripts
sg specic gravity at 60 F 0 inlet
S total sulfur content (wt.%) 1 outlet
Y liquid yield (volumetric fraction)
(HDS) with a feed comprising 62 vol.% straight-run heavy gasoil, In the near future, practically all fractions of sour crudes will be
10 vol.% coker light gasoil, and 28 vol.% FCC light cycle oil, to subjected to catalytic processes that involve hydrogen. The total
produce ultra low sulfur diesel (ULSD). Stratiev et al. [3] found quantity of fractions directed to hydrotreating and hydrocracking
that a correct hydrogen balance in a renery may be obtained will amount to 90% of the total crude run [5]. It is therefore
by using chemistry of processes running at hydrogen consuming mandatory to determine the hydrogen consumption with high
units and those producing hydrogen such as catalytic reforming, accuracy.
steam reforming, and hydrotreating of petroleum fractions. They
developed equations to provide information about the consumed
Table 1
hydrogen for HDS, hydrodearomatization, hydrogenation and
Typical hydrogen consumption data for various renery processes.
hydrodenitrogenation reactions in commercial units, which allow
for optimization of renery hydrogen consumption. Process Wt.% on feed Wt.% on crude
Ramachandran and Menon [4] discussed a brief review on var- Hydrotreating
ious uses of hydrogen in the industry. They pointed out that in Straight run naphtha 0.05 0.01
Cracked naphtha 0.71.0 0.050.1
modern reneries hydrogen requirement is commonly about
1 wt.% of crude processed. Typical hydrogen consumption data Hydrodesulfurization
Low sulfur gasoil to <0.05%S 0.15 0.04
for various renery processes are presented in Table 1. Hydro-
High sulfur gasoil to >0.05%S 0.35 0.05
cracking of vacuum distillates requires a large amount of hydrogen,
Cycle oils hydrogenation 3 0.3
not only for desulfurization, but also for the increase in the hydro-
Hydrocracking vacuum gasoil 23 0.50.8
gen/carbon ratio of the products relative to the feed [5].
L.C. Castaeda et al. / Fuel 90 (2011) 35933601 3595
The aim of this work is to describe and analyze the different ap- composed by pure hydrogen, hydrogen contained in H2S, and
proaches that exist to calculate the hydrogen consumption in hydrogen contained in C1C6 hydrocarbons as shown in Fig. 1. H0Liq
hydrotreating of petroleum fractions. Experimental data obtained
and H1Liq are the total amounts of hydrogen contained in the liquid
in bench-scale are used to test the different methods.
feedstock and in the hydrotreated liquid product respectively. H1Liq
2. Hydrogen consumption includes the hydrogen content in the liquid H1HC-Liq , and the
amount of hydrogen gas dissolved in the liquid H1diss :
Hydrogen consumption during hydroprocessing is dependent
upon the feedstock properties, impurities removal, conversion le-
H1Liq H1HC-Liq H1diss 4
vel and properties of the catalyst. The heavier feed requires sub-
stantially more addition of hydrogen to attain a xed level of The total real hydrogen consumption (Hcons) is thus:
upgrading [6]. Developing detailed mass balances in order to know
not only that all streams in the process are well accounted but also Hcons DHgas H1diss 5
the distribution of hydrogen in those streams is of vital importance
The values of H1Liq
and H0Liq
are preferably obtained experimentally,
to perform further studies regarding the commercial application of
e.g., by ultimate (elemental) analysis, however, they can be esti-
a new process and catalyst. The usual manner for calculating
mated with empirical correlations.
hydrogen consumption is by means of experimental data either
Based on all these equations, which are derived from mass bal-
by a hydrogen balance in gas streams or with hydrogen content
ances, the following approaches can be established for calculating
in the liquid feed and products. Fig. 1 shows all the streams in-
hydrogen consumption during hydrotreating operations.
volved in a general mass balance in an HDT reactor, where Gas0
and Liq0 are the total amounts of the gas stream entering the reac-
tor, and liquid feedstock to be hydrotreated respectively. Gas1 and 2.1. Mass balance of hydrogen in gas stream
Liq1 are the total amount of the gas product, which contains the
unreacted hydrogen, hydrogen sulde as by-product and light The most-known method for calculating hydrogen consumption
hydrocarbons ranging from C1 to C4, and the total amount of in hydrotreating units is by measuring the amount of hydrogen at
hydrotreated liquid product respectively. the inlet (H2)0 and at the outlet (H2)1 of the reactor system. The dif-
From Fig. 1, the following mass balances can be derived: ference between these two amounts, according to Eq. (2), is just a
delta of hydrogen and not the chemical hydrogen consumption.
0 1
Global balance : Gas0 Liq Gas1 Liq 1 This approach, although being the most commonly used in com-
mercial hydrotreating plants, has been reported to be not accurate
Hydrogen balance in gas streams : H2 0 H2 1 DHgas 2 [2,5]. The application of this method requires data of ow rates of
the gas streams entering and leaving the reactor, which are fre-
Global hydrogen balance : H0Gas H0Liq H1Gas H1Liq 3 quently reported in volumetric basis as well as the gas stream
composition measured by online gas chromatography (GC).
where (H2)0 and (H2)1 are the total amounts of hydrogen entering The real H2 consumption can be then calculated by adding the
and leaving the reactor respectively. If the hydrogen purity is as- dissolved H2 in the liquid product to the delta of hydrogen (DHgas)
sumed to be close to 100%, Gas0 = (H2)0, DHgas is the difference be- determined by the gas stream balance, according to Eq. (5).
tween the amount of hydrogen entering and leaving the reactor
determined by mass balance in the gas streams, which is commonly 2.2. Global hydrogen balance
(and erroneously) reported as hydrogen consumption. H0Gas and H1Gas
are the total equivalent amounts of hydrogen contained in the gas The calculation of hydrogen consumption by measuring the
streams entering and leaving the reactor respectively which are hydrogen content in the liquid feed (H0Liq ) and in the hydrotreated
Table 2
Empirical correlations to estimate the hydrogen content in petroleum fractions.
%H: hydrogen weight percent, CH: carbon-to-hydrogen weight ratio, q: liquid density at 20 C, n: refractive index at 20 C, Mw: molecular weight, Tb: average value of boiling
points calculated with 10, 50, and 90 vol.% of distilled product in Kelvin, sg: specic gravity 60 F/60 F, xA: fraction of aromatics in the mixture, AP: aniline point in K, m:
viscosity at 98.9 C in cSt, P: pressure in MPa, and GO: H2/HC ratio, HPurity: hydrogen purity.
HHDADiTriA
2 3
10Y Yp 10Y
2 TriAf 1000 TriAp Mwpp TriAf TriAp 100 DiAf 100
DiAp Mwpp qf
6 Mw f Mwf 7
4 522:4
10Y
3 MAf MAp Mwpp 100
18
Nf Y p Np qf These authors established a differential mass balance equation
HHDN 5 22:4 19
140 100 140 100 for hydrogen using an isothermal plug-ow reactor model. The to-
tal hydrogen consumption represents the concentration of all the
These authors assumed that the amount of H2 consumed by HDO bonds likely to react with hydrogen at the reaction conditions be-
reactions in HDT of diesel is negligible. It was reported that feed fore any treatment occurs. The observed reaction rate of hydrogen
quality has considerable inuence on the hydrogen chemical con- consumption comprises the sum of the reaction rates of hydrogen
sumption and conrmed earlier reported studies. The difference with the bonds likely to react in oil. In the absence of intraparticle
of experimental and calculated hydrogen consumption with these diffusion effects the following equation was derived:
equations was found to be 29.2%.
VR
2.4. Hydrogen consumption by reaction average contributions C H 1a C T-cons 1a a 1n1 ek 25
QL
The calculation of hydrogen consumption when experimental where CH is the remaining hydrogen demand or, the concentration
information is not sufcient or for quick estimations can be done of the bonds which are bound to react with hydrogen at the reaction
according to acquired experience. Edgar [16] reported the follow- conditions, expressed in mol H2 per m3 of the oil, CT-cons the total
ing average contributions of each HDT reaction to the hydrogen hydrogen consumption, n the catalyst remaining relative activity,
consumption (scf/bbl): e the catalyst bed void fraction, VR the catalyst bed volume, QL the
HDS 95100 per each 1 wt% removed 20 oil volumetric ow rate, k the intrinsic rate constant that includes
the hydrogen partial pressure term (k = kPH2 ) and a the reaction or-
HDN 300350 per each 1 wt% removed 21 der which depends on the concentration and reactivity of the react-
ing molecules.
HDC 25 per each 1 vol:% removed 22 The reaction order and the total hydrogen consumption were
determined by regression analysis of the experimental data. The
total hydrogen reaction rate was found to be of second order,
HDA 27 per each 1 wt% removed 23
a = 2, while the total hydrogen consumption was determined to be
257 Nm3 H2/m3 oil; within the temperature range of 350430 C
HDM 26 per each 1 ppm removed 24
for two catalyst types. The predicted value of hydrogen consumption
In addition to Edgar equations, Speight [17] presented a correction from diagrams in the literature based on the API gravity of the
in hydrogen consumption by metal content (Table 3). residue, is 207 Nm3 H2 =m3 oil [19].
Beuther and Schmid [20] measured the hydrogen consumption
2.5. Hydrogen consumption by kinetic modeling during a vacuum residue HDS and found a value of 277 Nm3
H2 =m3 oil for 99.2% hydrodesulfurization. For a second-order
In general, the calculation of hydrogen consumption by means reaction of hydrogen consumption, Eq. (25) takes the form:
of kinetics is done by the summation of the rates of all H2-consum- 1 Hcons VR
ing reactions, of course, the stoichiometry of each reaction, i.e. stoi- 1 ek 26
n HT-cons Hcons HT-cons QL
chiometric coefcients must be taken into account. Thus, the more
reactions the kinetic model includes in which hydrogen is present The authors reported a graphical correlation of the experimental
the better and more precision in the estimation of hydrogen data for the calculation of the intrinsic reaction rate constant k
consumption. in the temperature range of 350430 C, and another correlation
Various simplied approaches based on kinetics have been re- of Arrhenius plot of the intrinsic rate constants for hydrogen
ported in the literature. Papayannakos and Georgiou [18] devel- consumption.
oped a kinetic model for hydrogen consumption during catalytic
hydrodesulfurization of a residue in a trickle-bed reactor. The
atmospheric residue of Greek petroleum deposits in the Aegean 3. Solubility of hydrogen
Sea was used as feedstock. They found that the kinetic model cor-
related well the experimentally measured hydrogen consumption. To properly determine the hydrogen consumption in the hydro-
gen mass balance, the solubility of the hydrogen in the liquid
streams on Fig. 1 is required. The mass-balance equations are
Table 3 based on the assumption that the gasliquid equilibrium can be
Correction in hydrogen as function of metals content in petroleum fractions.
described by Henrys law as follows:
Ni + V (ppm) Correction (%) Ni + V (ppm) Correction (%) gN
H 27
0100 2 700 12 kqL
200 1 800 16
300 2.5 900 21 where H is the Henry coefcient, gN the molar gas volume at stan-
400 4 1000 28 dard conditions, k the solubility coefcient and qL the density of the
500 6.5 1100 38
liquid under process conditions. The concentration of hydrogen
600 9 1200 50
cH2 can be then estimated by the following equation:
3598 L.C. Castaeda et al. / Fuel 90 (2011) 35933601
Table 6
Properties of feed and product of hydrotreating experiments.
(6).
Fig. 2 shows the hydrogen consumption calculated from exper- Experimental Riazi-Daubert
Dhulesia Jenkins-Walsh
imental data for the different feedstocks. With the global hydrogen Goossens ASTM D-3343
balance it is conrmed that hydrogen consumption depends on the Bureau of standards Goodger
nature of the feed. This already known behavior is because heavy
fractions contain higher concentration of impurities such as sulfur,
nitrogen, asphaltenes, aromatics and metals, and their removal re-
quires higher amounts of hydrogen. The different values of hydro-
gen consumption are also due to the type of reaction occurring
during the HDT of each feedstock. Hydrotreating of light and mid-
dle fractions mainly comprises HDS reaction. Although other reac- HDSN HDSD HDSG HDAR HDVR
tions, (hydrodenitrogenation, hydrodearomatization and olens Fig. 3. Hydrogen consumption by global hydrogen balance with correlations for
saturation) occur in less extent, while for heavy fractions hydro- hydrogen content in liquid streams.
demetalization and hydrocracking also take place.
The global hydrogen balance requires experimental ultimate
analyses to determine hydrogen consumption that is why it is Some correlations exhibited better performance for heavy feeds
rarely reported by reneries, and they prefer the use of a quick ap- while others did for light feeds, which is due to the range of appli-
proach as that given by Eq. (2). Alternatively, the correlations re- cation in which they were developed. In general, the use of corre-
ported in Table 2 can be applied to calculate hydrogen content in lations to determine the hydrogen content of liquid streams to
liquid streams. further calculate the global hydrogen balance tends to underesti-
Fig. 3 shows a comparison of the experimental hydrogen con- mate the hydrogen consumption as compared with the experimen-
sumption and that determined from the global hydrogen balance tal value. For HDS of light distillates the correlations yield quite
using correlations to calculate the hydrogen content in liquid similar hydrogen consumption with the lowest error respect to
streams. the experimental one. The highest errors are found for HDS of hea-
vy feeds.
The correlations to calculate hydrogen content from only one li-
Experimental hydrogen consumption,
Hydrogen consumption,
1200 Experimental Edgar-Speight
solved hydrogen from correlations and calculations using Hysys
simulator. 1000
ft3 std/bbl
The dissolved hydrogen from KorstenHoffman and Mapiour 800
Sundaramurthy correlations and that determined by Hysys led to 600
similar results. Hydrogen consumption determined from Eq. (5) 400
showed deviations lower than 0.23% respect to the global hydrogen
200
balance on the HDT of light and heavy gasoils, while the observed
0
error in the HDT of heavy fractions was between 1.14% and 2.73%. HDSN HDSD HDSG HDAR HDVR
In general, hydrogen consumption calculated from the hydrogen
balance in gas streams is close to that obtained by the global Fig. 6. Hydrogen consumption by reactions average contribution.
hydrogen balance.
The dissolved hydrogen accounts for 0.16% of total hydrogen
consumption from hydrogen balance in gas streams as expressed underpredicted the hydrogen consumption and they have the low-
by Eq. (2). This certainly not a high contribution, is the most likely est error for HDS of light gasoil, which is in agreement with their
reason why reners use the difference between the amount of range of application.
hydrogen entering and leaving the reactor in the gas streams,
DHgas, to report the hydrogen consumption. 4.5. Hydrogen consumption by reaction average contributions
Experimental Korsten-Hoffman
rate data, GC analysis, liquid analysis and dissolved hydrogen cal-
Mapiour-Sundaramurty Hysys
culations is required to determine the real H2 consumption.
The best correlations to be used for hydrogen content calcula-
tions in liquid streams were the following: JenkinsWalsh correla-
tion for naphtha, Dhulesia correlation for light and heavy gasoils,
and Goosens correlation for AR and VR.
The dissolved hydrogen obtained from KorstenHoffman,
MapiourSundaramurthy and Hysys calculations was similar; hence,
the use of any of these approaches is suitable.
For quick calculation the correlations of Lee and Stratiev
Tzingov can be used to estimate hydrogen consumption by
HDSN HDSD HDSG HDAR HDVR
chemical reactions contributions for HDT middle fractions, while
Fig. 4. Hydrogen consumption by hydrogen balance in gas stream with correlations the approach developed by Edgar and Speight is better for heavy
for dissolved hydrogen. fractions.
L.C. Castaeda et al. / Fuel 90 (2011) 35933601 3601
References [11] Dhulesia H. New correlations predict FCC feed characterizing parameters. Oil
Gas J 1986;13:514.
[12] Jenkins GI, Walsh RE. Quick measure of jet fuel properties. Hydrocarbon
[1] Mapiour M, Sundaramurthy V, Dalai A, Adjaye J. Effects of hydrogen partial
Process 1968;47(5):1614.
pressure on hydrotreating of heavy gas oil derived from oil-sands bitumen:
[13] Winn EW. Physical properties by nomogram. Pet Rener 1957;36(21):
experimental and kinetics. Energy & Fuels 2009; [Published online].
1579.
[2] Lee CK, Magalhaes C. da Silva Luiz E, Osowski Carlos A. Study compares
[14] Baird CT. Crude oil yields and product properties. Ch. De la Haute-Belotte 6,
methods that measure hydrogen use in diesel hydrotreaters. Oil Gas J
Cud Thomas Baird IV, 1222 Vezenaz, Geneva, Switzerland; June 1981.
2008;13:5863.
[15] Goodger E, Vere R. Aviation fuels technology. London: MacMillan; 1985.
[3] Stratiev D, Tzingov T, Shishkova I, Dermatova P. Hydrotreating units chemical
[16] Edgar MD. Hydrotreating Q&A. Paper presented at the NPRA Annual Meeting
hydrogen consumption analysis a tool for improving renery hydrogen
1993, San Antonio, Texas; 1993.
management. In: 44th International petroleum conference, September 21
[17] Speight JG. The chemistry and technology of petroleum. 3rd ed. berlin: Marcel
22, Bratislava, Slovak Republic; 2009.
& Deckker; 1999.
[4] Ramachandran R, Menon RK. An overview of industrial uses of hydrogen. Int J
[18] Papayannakos N, Georgiou G. Kinetics of hydrogen consumption during
Hydrogen Energy 1998;23(7):5938.
catalytic hydrodesulfurization of a residue in a trickle-bed reactor. J Chem
[5] Aitani AM. Processes to enhance renery hydrogen production. Int J Hydrogen
Eng Jpn 1988;21(3):2449.
Energy 1996;21(4):26771.
[19] Nelson WL. Oil Gas J 1977;28:126.
[6] Ancheyta J, Speight JG. Hydroprocessing of heavy oils and residua. Taylor &
[20] Beuther H, Schmid BK. In: 6th world petroleum congress, Section III; 1963. p.
Francis Group, CRC Press; 2007.
297.
[7] Riazi MR, Daubert TE. Prediction of molecular type analysis of petroleum
[21] Korsten H, Hoffmann U. Three-phase reactor model for hydrotreating in pilot
fractions and coal liquids. Ind Eng Chem Process Des Dev 1986;25(4):100915.
trickle-bed reactors. AIChE J 1996;42:1350.
[8] Riazi MR, Nasimi N, Roomi Y. Estimating sulfur content of petroleum products
[22] Riazi MR, Roomi YA. A method to predict solubility of hydrogen in
and crude oils. Ind Eng Chem Res 1999;38(11):450712.
hydrocarbons and their mixtures. Chem Eng Sci 2007;62:664958.
[9] Goossens AG. Prediction of the hydrogen content of petroleum fractions. Ind
[23] Riazi MR, Vera JH. Method to calculate the solubilities of light gases in
Eng Chem Res 1997;36(6):25004.
petroleum and coal liquid fractions on the basis of their P/N/A composition.
[10] ASTM Annual Book of Standards. ASTM International; West Conshohocken, PA;
Ind Eng Chem Res 2005;44:18692.
2002.