Professional Documents
Culture Documents
Hydrogen Induced Cracking HIC Testing of
Hydrogen Induced Cracking HIC Testing of
Corrosion Science
journal homepage: www.elsevier.com/locate/corsci
Hydrogen induced cracking (HIC) testing of low alloy steel in sour environment:
Impact of time of exposure on the extent of damage
Jean Kittel a,*, Véronique Smanio b,1, Marion Fregonese c, Laurence Garnier a, Xavier Lefebvre a
a
IFP-Lyon, Rond-Point de l’échangeur de Solaize, BP3 69360 Solaize, France
b
Total, Avenue Larribau, 64018 Pau Cedex, France
c
Université de Lyon, INSA-Lyon, MATEIS CNRS UMR 5510, 21 ave J. Capelle, F-69621 Villeurbanne, France
a r t i c l e i n f o a b s t r a c t
Article history: Hydrogen induced cracking (HIC) of line pipe steel was investigated through immersion testing and
Received 14 September 2009 hydrogen permeation measurements. At constant pH and hydrogen sulphide partial pressure (pH2S),
Accepted 28 November 2009 the extent of HIC was found to depend on exposure time until a stable level was reached. The time to
Available online 5 December 2009
reach this stable value is affected by pH and pH2S. Results of permeation experiments confirmed that
HIC is linked with the increase of hydrogen concentration in the steel. It is also shown that low severity
Keywords: requires longer exposures to reach equilibrium. This must be taken into account for HIC testing in mildly
A. Low alloy steel
sour environment.
B. Hydrogen permeation
C. Hydrogen embrittlement
Ó 2009 Elsevier Ltd. All rights reserved.
C. Sulphide cracking
0010-938X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2009.11.044
J. Kittel et al. / Corrosion Science 52 (2010) 1386–1392 1387
n2
5
e gio
pH
r
C
SS
4
SSC region 3
-3 -2 -1 0
10 10 10 10
pH2S (bar)
2. Materials and methods Fig. 2. Microstructure (L-direction) of the tested steel in the centre line (a) and
close to the surface (b).
dium acetate. The bubbling gas was either pure H2S, or a mixture
The steel used for this study was an API X65 thermo-mechani-
of CO2 with H2S (1% or 10% H2S).
cally controlled process (TMCP) steel, coming from a sweet service
After a thorough deoxygenating step, the solution was saturated
pipe 1.07 m diameter and 25 mm thick. Its chemical composition
by bubbling gas at ambient pressure. The pH was then adjusted to
(Table 1) was analysed by optical emission spectrometry, except
desired level through addition of deaerated 1 N hydrochloric acid
for carbon and sulphur, analysed by a chemical method. The micro-
(HCl) or deaerated 1 N sodium hydroxide (NaOH). The solution
structure consisted mainly of ferrite with a portion of pearlite, with a
was then poured into the test cell. All the experiments were made
banded structure particularly pronounced in the centre line (Fig. 2).
at ambient temperature.
Specimens for HIC experiments were sampled following the
NACE TM0284-2003 standard test method [39]. The sample
dimensions were 100 mm long, 20 mm wide and 24 mm thick 2.3. HIC experiments
(0.5–1 mm were removed from the inner and outer surfaces). All
faces were machined to obtain a finish equivalent to that achieved HIC exposure tests were made in standard glass vessels. All tests
using 320 grit silicon carbide (SiC) paper. were performed following the general NACE TM0284-2003 stan-
Specimens for permeation experiments were 70 mm 70 mm dard [39], including three replicates for each condition and a vol-
plates, with a thickness of 10 or 0.5 mm. The surface finishing ume/surface ratio greater than 3 mL cm2. pH measurements of
was the same as that of HIC samples. the test solution were performed with a period of 48 h or less,
and adjustments were made if the value was not within 0.1/
2.2. Test solutions and exposure conditions +0.3 pH units from the desired level.
After each test, all specimens were inspected using ultrasonic
All solutions were prepared according to EFC16 document [40], testing (UT) method. The analysis was performed with a ULTRAPAC
solution A, and thus contained 5% sodium chloride and 0.4% so- system (Euro Physical Acoustics SA) and a 15 MHz transducer
Table 1
Chemical composition (wt%) of tested steel.
C Mn Si P S Cr Ni Mo Cu Nb V Fe
0.09 1.56 0.28 0.014 0.001 0.05 0.03 0.01 0.02 0.04 0.05 s.q.f. 100%
1388 J. Kittel et al. / Corrosion Science 52 (2010) 1386–1392
(6 mm diameter). For each specimen, the total area of the defects 6x10
-2
was calculated from the top view of the UT scans, and the crack
area ratio (CAR) was determined as the ratio between the area of
the defects and the total area of the specimen in the short trans-
verse plan. The CAR values reported in this paper are the average
-2
of three replicates. It should be noted that since cracking occurred 4x10
mainly in the centre line of the tested steel, CAR values are closely
J (A.m )
-2
linked with crack length ratio (CLR) values that would be deter-
mined from cross sections according to NACE TM0284-2003 proce-
dure [39]. -2
2x10
0.3 From all the experimental results, the evolution of CAR with
1 bar time was analysed. Fig. 6 presents CAR evolution at different pH
100 mbar and at constant H2S partial pressure.
As already illustrated in a previous paper [11], pH and H2S partial
pressure have a significant impact on the time necessary for HIC to
0.2 initiate. At constant H2S partial pressure, increasing the pH leads
to a delayed HIC initiation. This effect is particularly marked under
J (A.m )
-2
1 bar H2S: at pH 4.5, cracks are observed after only a few hours
immersion, whereas more than 200 h are necessary at pH 5.5. The
same tendency is observed for the results under 100 mbar H2S.
0.1
It also seems that H2S partial pressure has a more pronounced
influence than pH on the maximum extent of HIC: indeed, under
100 mbar H2S, the same CAR values are observed for the longer
immersion times for the tests at pH 5.5–3.5. However, this conclu-
0.0 sion requires further work on a range of sweet service materials.
0 5 10 15 Another interesting result was found for the experiments car-
Time (hours) ried out at pH 5.5 and under 0.1 and 1 bar H2S. HIC experiments re-
vealed that exposure under 1 bar H2S at pH 5.5 was less severe than
Fig. 4. Hydrogen permeation transient measured for 0.5 mm steel membranes 0.1 bar H2S at pH 5.5 (Fig. 6). Under 1 bar H2S, more than 200 h expo-
exposed to test solution at pH 5.5 and 1 bar or pH 5.5 and 100 mbar H2S.
sures were necessary before cracks were observed, while under
100 mbar H2S, cracks were detected after only 96 h exposure. This
Table 3 behaviour is quite unexpected: indeed, it is usually agreed that pH
Range of test durations (h) of HIC experiments at different pH and pH2S. and H2S partial pressure have a monotone impact on severity, i.e. in-
crease of severity when the pH decreases and H2S partial pressure
pH2S 10 mbar 100 mbar 1 bar
increases. This peculiar behaviour is in agreement with the perme-
pH 5.5 96–336 24–96–336 24–96–120 ation data of Fig. 4, showing a higher steady-state permeation flux
192–336
pH 4.5 96–336 24–48–96–336 5–8–14.5–24
at pH 5.5 and 100 mbar H2S than at pH 5.5 and 1 bar H2S.
44–96–120–336
pH 3.5 96–336–720–1080 24–48–96–336 24–96
4. Discussion: hydrogen diffusion vs. HIC
A A'
Fig. 5. US-scans after different times of exposure under 1 bar H2S at pH 4.5.
1390 J. Kittel et al. / Corrosion Science 52 (2010) 1386–1392
CAR (%)
40 40
20 20
0 0
0 100 200 300 0 100 200 300 0 500 1000
Test duration (hours)
Fig. 6. Evolution of HIC extent with time of exposure. Impact of pH at constant pH2S.
branes, which allowed to determine the diffusion coefficients of at pH 3.5 and 4.5, while it takes between 24 and 96 h at pH 5.5.
hydrogen in the steel (Table 2). For the comparison with HIC re- These values are in very close agreement with the timescales ob-
sults, these diffusion parameters were used for the determination tained from permeation data and diffusion modelling.
of hydrogen concentration profiles in HIC samples at different This result suggests that permeation measurements might rep-
immersion times. Those calculations were performed using finite resent an interesting method for the optimisation of HIC testing
element model software, with the following hypothesis and procedures. From permeation data, it is possible to determine dif-
boundary conditions: fusion coefficients, which can then be used for calculations of
hydrogen profiles in HIC samples, and thus give an indicative value
(a) The geometry of the modelled HIC samples was considered of the minimum time of exposure necessary to reach a critical
parallelepipedic (20 mm wide and 24 mm thick) with an hydrogen concentration level. Additionally, when permeation data
infinite length. shows hydrogen sub-surface concentration below the critical level,
(b) Hydrogen was supposed to enter from all four lateral faces of it might become useless to perform HIC tests.
the samples.
(c) The initial hydrogen concentration was supposed to be zero
4.2. Evaluation of severity from permeation experiments
inside the specimen.
(d) Diffusion was supposed to follow Fick’s law of diffusion,
The second interesting comparison between HIC and perme-
with a constant sub-surface concentration C0 and a constant
ation tests lies in the experiments performed at pH 5.5 and under
diffusion coefficient D, both determined from permeation
100 mbar and 1 bar H2S. HIC experiments revealed that exposure
measurements.
under 1 bar H2S at pH 5.5 was far less severe than 0.1 bar H2S at
pH 5.5 (Fig. 6). Under 1 bar H2S, more than 200 h exposures were
Fig. 7 illustrates the calculated evolution of hydrogen concentra-
necessary before cracks were observed, while under 100 mbar
tion with time in the section of a HIC sample at pH 5.5 and with
H2S, cracks were detected after only 96 h exposure. The lower
100 mbar H2S. Similar calculations were performed for the diffu-
severity of pH 5.5 and 1 bar H2S towards pH 5.5 and 0.1 bar H2S
sion parameters determined from the experiments at pH 4.5 and
is confirmed by permeation measurements (Fig. 4). Indeed, perme-
3.5 (Table 2).
ation current measured at pH 5.5 under 1 bar H2S rapidly de-
Then, since all ultrasonic evaluations of HIC samples revealed
creased well below the values under 100 mbar H2S. Similar
that cracking was located near the mid-thickness, the analysis of
trends have already been observed by Kimura et al. [48]. These
hydrogen profile was restricted to the centre line, i.e. at a depth
authors observed that above a certain H2S partial pressure, corro-
of 12 mm. Evolutions with time of concentration profiles in the
sion rate of steel was decreasing, thus leading to a decrease in per-
centre line are presented in Fig. 8. Obviously, hydrogen concentra-
meation rate. This decrease might probably be explained by the
tion increases more rapidly at pH 3.5 than at pH 5.5, and the final
build-up of a more protective iron sulphide layer at 1 bar H2S,
equilibrium level is also higher.
and a reduced corrosion rate and hydrogen reduction reaction.
Finally, for the analysis of this modelling towards HIC, we con-
Consequently, hydrogen charging in the metal takes more time un-
sidered a critical hydrogen concentration, above which cracking
der 1 bar H2S, and the time to reach the critical hydrogen concen-
might occur. From a previous study using the same steel [11], this
tration is strongly delayed.
threshold hydrogen concentration was found to lie between 0.9
Obviously, H2S plays an important role on the surface state of
and 1 ppm by weight. This threshold value was reported in
the steel, hence influencing strongly hydrogen charging rate and
Fig. 8. For each time increment, we also calculated the average
hydrogen sub-surface concentration C0. This result also suggests
hydrogen concentration in the centre line. The corresponding re-
that permeation measurements could be used in order to define
sults are plotted in Fig. 9.
pH–pH2S severity regions as already suggested by Duval et al.
For the discussion of time to initiate HIC, comparison between
[49].
Figs. 9 and 6 (results at 100 mbar) is extremely interesting. Mod-
elling of hydrogen diffusion in the centre line of HIC specimens
(Fig. 9) shows that the average concentration reaches the critical 5. Conclusions
cracking level (0.95 ppm by weight) after 3 h at pH 3.5, after 7 h
at pH 4.5, and after more than 40 h at pH 5.5. On the other hand, These results confirm that HIC assessment of sweet service line
the evolution of HIC extent with time (Fig. 6) reveals that cracks pipe steels in mild sour environments requires precautions. For the
are observed after only a few hours immersion under 0.1 bar H2S lower H2S partial pressure and high pH combination, more than
J. Kittel et al. / Corrosion Science 52 (2010) 1386–1392 1391
Fig. 7. Calculated hydrogen concentration profile in the 20 mm 24 mm section of a HIC sample exposed at pH 5.5 and 100 mbar H2S (D = 7.5 1010 m2 s1; C0 = 0.97 ppm
by weight).
0.5 0.5
1 hour
1 hour
1 h.
0.0 0.0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
X position (mm)
Fig. 8. Modelling of hydrogen concentration profiles in the centre line (Y = 12 mm) of HIC specimens in test solution under 100 mbar H2S and at different pH, using diffusion
constants of Table 2 (time increment between each profile = 2 h).
4 weeks might be necessary before HIC initiation. Therefore, it tion of diffusion coefficients can be used for calculations of hydro-
might become necessary to adopt long term immersion tests for gen profiles in HIC samples, and give a correct evaluation of the
the qualification of sweet service steels in terms of HIC resistance. minimum time of exposure necessary to reach a critical hydrogen
For this purpose, permeation measurements might be an inter- concentration level. Additionally, when permeation data shows
esting and complementary experimental means of investigating hydrogen sub-surface concentration below the critical level, it
the impact of environmental parameters on HIC. The determina- might become useless to perform HIC tests.
1392 J. Kittel et al. / Corrosion Science 52 (2010) 1386–1392
[19] L.W. Tsay, Y.C. Chen, S.L.I. Chan, Sulfide stress corrosion cracking and fatigue
1.5 crack growth of welded TMCP API 5L X65 pipe-line steel, Int. J. Fatigue 23
(2001) 103–113.
Average [H] in centre line (ppm)
[20] NACE MR0175/ISO 15156-2, Petroleum and natural gas industries – materials
for use in H2S containing environments in oil and gas production – Part 2:
Cracking-resistant carbon and low alloy steel, and the use of cast iron, 2003.
[21] J.R. Scully, P.J. Moran, The influence of strain on hydrogen entry and transport
1.0 critical [H] (0.95 ppm)
in a high-strength steel in sodium-chloride solution, J. Electrochem. Soc. 135
(1988) 1337–1348.
[22] J.R. Scully, P.J. Moran, Influence of strain on the environmental hydrogen-
assisted cracking of a high-strength steel in sodium-chloride solution,
Corrosion 44 (1988) 176–185.
0.5 [23] C. Mendibide, T. Sourmail, Composition optimization of high-strength steels
for sulfide stress cracking resistance improvement, Corros. Sci. 51 (2009)
pH 5.5 2878–2884.
pH 4.5 [24] L.W. Tsay, M.Y. Chi, Y.F. Wu, J.K. Wu, D.Y. Lin, Hydrogen embrittlement
susceptibility and permeability of two ultra-high strength steels, Corros. Sci.
pH 3.5 48 (2006) 1926–1938.
0.0 [25] J. Kittel, F. Ropital, J. Pellier, Effect of membrane thickness on hydrogen
0 10 20 30 40 50 permeation in steels during wet H2S exposure, Corrosion 64 (2008) 788–799.
[26] A. Turnbull, M. Saenz de Santa Maria, N.D. Thomas, The effect of H2S
Time (hours)
concentration and pH on hydrogen permeation in AISI 410 stainless steel in
5% NaCl, Corros. Sci. 29 (1989) 89–104.
Fig. 9. Average hydrogen concentration in the centre line (Y = 12 mm) of HIC [27] A. Turnbull, M.W. Carroll, The effect of temperature and H2S concentration on
specimens in test solution under 100 mbar H2S and at different pH, using diffusion hydrogen diffusion and trapping in a 13% chromium martensitic stainless steel
constants of Table 2. in acidified NaCl, Corros. Sci. 30 (1990) 667–679.
[28] C. Azevedo, P.S.A. Bezerra, F. Esteves, C.J.B.M. Joia, O.R. Mattos, Hydrogen
permeation studied by electrochemical techniques, Electrochim. Acta 44
(1999) 4431–4442.
Acknowledgement
[29] L. Coudreuse, J. Charles, The use of a permeation technique to predict critical
concentration of H2 for cracking, Corros. Sci. 27 (1987) 1169–1181.
The authors thank G. Parrain, who performed all HIC experi- [30] W.K. Kim, S.U. Koh, B.Y. Yang, K.Y. Kim, Effect of environmental and
metallurgical factors on hydrogen induced cracking of HSLA steels, Corros.
ments and participated in the permeation measurements.
Sci. 50 (2008) 3336–3342.
[31] G.T. Park, S.U. Koh, K.Y. Kim, H.G. Jung, Effect of steel microstructure on
hydrogen permeation behavior and the determination of critical hydrogen flux
References for hydrogen embrittlement, NACE Corrosion/2006, San Diego, CA, 12–16
March 2006, paper 06438.
[1] T.V. Bruno, C. Christensen, R.T. Hill, History and development of TM0284, NACE [32] G.T. Park, S.U. Koh, H.G. Jung, K.Y. Kim, Effect of microstructure on the
Corrosion/99, San Antonio, TX, 25–30 April 1999, paper 99422. hydrogen trapping efficiency and hydrogen induced cracking of linepipe steel,
[2] T. Hara, H. Asahi, H. Ogawa, Conditions of hydrogen-induced corrosion Corros. Sci. 50 (2008) 1865–1871.
occurrence of X65 grade line pipe steels in sour environments, Corrosion 60 [33] H. Asahi, M. Ueno, T. Yonezawa, Prediction of sulfide stress cracking in high-
(2004) 1113–1121. strength tubulars, Corrosion 50 (1994) 537–545.
[3] R.D. Kane, M.S. Cayard, NACE committee report 8X294: review of published [34] S.U. Koh, J.S. Kim, B.Y. Yang, K.Y. Kim, Effect of line pipe steel microstructure on
literature on wet H2S cracking, NACE Corrosion/99, San Antonio, TX, 25–30 susceptibility to sulfide stress cracking, Corrosion 60 (2004) 244–253.
April 1999, paper 99420. [35] S.U. Koh, B.Y. Yang, K.Y. Kim, Effect of alloying elements on the susceptibility to
[4] S.E. Mahmoud, C.W. Petersen, R.J. Franco, Overview of hydrogen induced sulfide stress cracking of line pipe steels, Corrosion 60 (2004) 262–274.
cracking (HIC) of pressure vessels in upstream operations, NACE Corrosion/91, [36] J. Marsh, Comparing hydrogen permeation rates, corrosion rates and sulphide
Cincinnati, OH, 11–15 March 1991, paper 91010. stress cracking resistance for C-110 and P-110 casing steel, NACE Corrosion/
[5] E.M. Moore, Hydrogen-induced damage in sour, wet crude pipelines, J. Petrol. 2007, Nashville, TN, 11–15 March 2007, paper 07109.
Technol. 36 (1984) 613–618. [37] C. Mendez, I. Martinez, L. Melian, J. Vera, Application of hydrogen permeation
[6] L. Smith, An overview of European federation of corrosion documents EFC16 and for monitoring sulfide stress cracking susceptibility, NACE Corrosion/2002,
EFC17, NACE Corrosion/99, San Antonio, TX, 25–30 April 1999, paper 99423. Denver, CO, 7–12 April 2002, paper 02342.
[7] S.N. Smith, M.W. Joosten, Corrosion of carbon steel by H2S in CO2 containing [38] J.R. Vera, R. Case, A. Castro, The relationship between hydrogen permeation
oilfield environments, NACE Corrosion/2006, San Diego, CA, 12–16 March and sulfide stress cracking susceptibility of OCTG materials at different
2006, paper 06115. temperatures and pH values, NACE Corrosion/97, New Orleans, LA, 9–14
[8] C. Bosch, J.P. Jansen, T. Herrmann, Fit-for-purpose HIC assessment of large March 1997, paper 97047.
diameter pipes for sour service application, NACE Corrosion/2006, San Diego, [39] NACE TM0284-2003, Evaluation of Pipeline and Pressure Vessel Steels for
CA, 12–16 March 2006, paper 06124. Resistance to Hydrogen-induced Cracking, NACE International, Houston, TX, 2003.
[9] T. Hara, H. Asahi, Y. Terada, T. Shigenobu, H. Ogawa, The condition of HIC [40] EFC Publication No. 16, Guidelines on Materials Requirements for Carbon and
occurrence of X65 linepipe in wet H2S environments, NACE Corrosion/99, San Low Alloy Steels for H2S-containing Environments in Oil and Gas Production,
Antonio, TX, 25–30 April 1999, paper 99429. The Institute of Materials, London, UK, 1995.
[10] T. Hermann, C. Bosch, J. Martin, HIC assessment of low alloy steel line pipe for [41] M.A.V. Devanathan, Z. Stachurski, The mechanism of hydrogen evolution on
sour service application – literature survey, 3R International 44 (2005) 409–417. iron in acid solutions by determination of permeation rates, J. Electrochem.
[11] J. Kittel, J.W. Martin, T. Cassagne, C. Bosch, Hydrogen induced cracking (HIC) – Soc. 111 (1964) 619–623.
laboratory testing assessment of low alloy steel linepipe, NACE Corrosion/ [42] P. Manolatos, M. Jerome, J. Galland, Necessity of a palladium coating to ensure
2008, New Orleans, LA, 16–20 March 2008, paper 08110. hydrogen oxidation during electrochemical permeation measurements on
[12] R.W. Revie, V.S. Sastri, G.R. Hoey, R.R. Ramsingh, D.K. Mak, M.T. Shehata, iron, Electrochim. Acta 40 (1995) 867–871.
Hydrogen-induced cracking of linepipe steels. 1. Threshold hydrogen [43] P. Manolatos, M. Jerome, C. Duret-Thual, J. Le Coze, The electrochemical
concentration and pH, Corrosion 49 (1993) 17–23. permeation of hydrogen in steels without palladium coating. Part I:
[13] R.W. Revie, V.S. Sastri, M. Elboujdaini, R.R. Ramsingh, Y. Lafrenière, Hydrogen- Interpretation difficulties, Corros. Sci. 37 (1995) 1773–1783.
induced cracking of line pipe steels used in sour service, Corrosion 49 (1993) [44] P. Manolatos, M. Jerome, A thin palladium coating on iron for hydrogen
531–535. permeation studies, Electrochim. Acta 41 (1996) 359–365.
[14] G.P. Tiwari, A. Bose, J.K. Chakravartty, S.L. Wadekar, M.K. Totlani, R.N. Arya, R.K. [45] J. Crank, The Mathematics of Diffusion, second ed., Oxford University Press,
Fotedar, A study of internal hydrogen embrittlement of steels, Mater. Sci. Eng. Inc., Oxford, UK, 1975.
A Struct. 286 (2000) 269–281. [46] ASTM G148-97, Standard Practice for Evaluation of Hydrogen Uptake,
[15] J. Leyer, P. Sutter, H. Marchebois, C. Bosch, A. Kulgemeyer, B.J. Orlans-Joliet, SSC Permeation, and Transport in Metals by an Electrochemical Technique, ASTM
resistance of a 125 KSI steel grade in slightly sour environments, NACE International, West Conshohocken, PA, 2003.
Corrosion/2005, Houston, TX, 3–7 April 2005, paper 05088. [47] J.L. Crolet, M.R. Bonis, Revisiting hydrogen in steel, Part I: Theoretical aspects
[16] C.M. Liao, J.L. Lee, Effect of molybdenum on sulfide stress cracking resistance of of charging, stress cracking and permeation, NACE Corrosion/2001, Houston,
low-alloy steels, Corrosion 50 (1994) 695–704. TX, 11–16 March 2001, paper 01067.
[17] J. Mougin, M.S. Cayard, R.D. Kane, B. Ghys, C. Pichard, Sulfide stress cracking [48] M. Kimura, N. Totsuka, T. Kurisu, T. Hane, Y. Nakai, Effect of environmental
and corrosion fatigue of steels dedicated to bottom hole assembly factors on hydrogen permeation in line pipe steel, Corrosion 44 (1988) 738–744.
components, NACE Corrosion/2005, Houston, TX, 3–7 April 2005, paper 05085. [49] S. Duval, R. Antano-Lopez, C. Scomparin, M. Jerome, F. Ropital, Hydrogen
[18] K.E. Szklarz, Sulfide stress cracking of a pipeline weld in sour gas service, NACE permeation through ARMCO iron membranes in sour media, NACE Corrosion/
Corrosion/99, San Antonio, TX, 25–30 April 1999, paper 99428. 2004, New Orleans, LA, March 28–April 1, 2004, paper 04740.