This paper summarizes the development and use of a non-intrusive, patch probe to monitor hydrogen flux in wet H₂S piping, vessels and pipelines. It utilizes an electrochemical cell with a Ni layer on the external surface of the equipment. Laboratory tests were conducted in several wet H2S environments, including the standard NACE TM 0177 and TM 0284. These tests were used to evaluate the response of the probe and to correlate hydrogen flux measurements with susceptibility to hydrogen induced cracking in conventional steels. This paper also describes the successful in-plant use of the probe in a PETROBRAS refinery for a two years period.

Failures of piping, vessels and pipelines exposed to wet H2S service can result from hydrogen charging that results from sulfide corrosion. Therfore, monitoring and assessment criteria for hydrogen induced damage are important methodologies to minimize the risk of failures. In order to achieve this goal, the critical defect size has to be established. This is generally achieved with fracture mechanics and finite element analysis, once the toughness of the hydrogen charged material is known. Then, inspection has to be performed and, if no critical size defect is found, the equipment is considered safe for continued operation.

Another tool in control of wet H₂S damage is hydrogen flux monitoring (HFM). It can be used to assist plant operators to control system conditions to minimize hydrogen flux produced by aqueous sulfide corrosion and thereby also minimize the development and growth of wet H₂S damage into critical size defects. To achieve this goal, the hydrogen concentration through the thickness of the equipment must be kept below a threshold value for initiation and/or propagation of wet H₂S cracking. This means that the threshold values of hydrogen concentration for crack nucleation and for crack propagation have to be first determined.

The electrochemical permeation measurement technique is a natural choice for the HFM procedure since its readings shall have the capability for capturing the high fraction of hydrogen needed for measurements of high accuracy and precision. In most cases, a thin Pd plating is used on the exit surface as the method to assure that all the hydrogen which permeates the membrane is measured^1,2 Nevertheless, Pd plating is a complex process to be used on equipment of an industrial plant. Moreover, this surface plating often degrades with time³. Alternatively, Ni plating has been considered a good alternative⁴. Like Pd, it allows low background currents, on the order of 0.03 microA/cm2, it provides an oxide that it is stable with time, and it allows high efficiency of hydrogen monitoring. The use of a thick Ni electrodeposited layer (30 micro-m has the advantage of requiring less stringent surface preparation (usually grinding to a 400# grit surface is good enough) than a Pd layer and it also requires little or no maintenance for long periods of time. The HFM procedure incorporating the Ni layer is also preferable to the Pd foil technique used in some existing probes, since the Ni-layer provides for a much higher hydrogen monitoring efficiency on the exist surface than does the Pd foil.

Electrochemical permeation cells of the Devanathan type, made of PTFE and with two different sizes, were used in this program. The smaller cells were used to test the thinner specimens, while the larger ones were used to test the thicker samples (>=9 mm). Their design was such that they could be used to make either laboratory or plant/field measurements. The attachment of these cells to larger scale laboratory specimens or to the external surface of real equipment, was made through the use of an external steel cable or a magnetic clamping device. The volumes of these cells were 0.5 and 2 liters. The permeation procedure was similar to that given in ASTM G148 ⁵. In the entrance side, hydrogen was formed on the steel surface through either cathodic charging or by exposure to an H₂S-containing solution. The following solutions were used:

• 1M acetic acid solution with and without a cathodic current of 0.22 mA/cm2 imposed to the specimen on the entrance side; and solutions
• NACE TM 0177 5 (Method A / Solution A)
• NACE TM 0284 (Solution A) 6
• NACE TM 0177 with 60 and 200 ppm of a filmic amine inhibitor
• NACE TM 0284 with 30, 50 and 200 ppm of bicarbonate.

In all cases, the charging solutions were continuously bubbled with H2S and the monitoring solution was a 3M NaOH solution deaerated with nitrogen. A Saturated Ag/AgCl reference electrode (Ag/AgCl) and a platinum counter electrode used in the monitring solution. All tests were conducted at 28 C.

Specimens (permeation membranes) were made of conventional (non-HIC resistant, high sulfur) C-Mn steel made to either ASTM A516Gr60 or ASTM A516Gr70. Chemical composition and mechanical properties of steels are presented in Tables 1 and 2, respectively. In the case of specimens exposed to the NACE solutions, the exposed material was the ASTM A516Gr60; whereas, for the acetic acid environments the ASTM A516Gr70 was used. The entrance side was polished to 100 grit and exit side to 600 grit. The exit side was either Pd plated (0.4 micro-m) according to the procedure proposed by Poeperling⁽¹⁾ or Ni plated using the Watt solution and procedure described by Kushida⁽⁸⁾. For the Ni plating, time was set to allow the Ni layer thicknesses of either 0.1 micro-m, 0.5 micro-m, 5 micro-m or 30 micro-m.

After Ni or Pd plating, specimens were cleaned and the cell was assembled. The potential of the plated surface was kept constant at +200 mV relative to the saturated Ag/AgCl electrode until a background current lower than 0.05 microA/cm² was achieved. After the low background current was achieved, deaerated charging solution was added to the entrance side, followed by 1-hour additional N₂ purge. When it was the case, either H₂S was purged or a cathodic charging current was applied depending on the type of experiment. At the same time, the computer program identified the beginning of the permeation test. A potentiostat coupled to a multiplexer with the necessary software was used to make all the measurements in the laboratory and also in the field. Before and after each permeation test, specimens of 9 mm, 15 mm and 36 mm were Ultrasonic (UT) inspections were conducted to identify and to quantify HIC nucleation and propagation. Details of the UT procedure were given elsewhere⁽⁹⁾.

RESULTS

(Click on the images for bigger images)


FIGURE 1 - Hydrogen permeation curves normalized by the specimen thickness

FIGURE 2 - Hydrogen permeation curves normalized by the specimen thickness for tests in TM0177 solution.

Figures 1 and 2 show the influence of the Ni and Pd layer thickness on the permeation curve of ASTM A516Gr60 specimens with 1.6 and 3 mm thickness, respectively, exposed to the NACE TM 0177 solution.

FIGURE 3 – Hydrogen permeation curves normalized by the specimen thickness for various Ni plating thicknesses.

Figure 3 shows the effect of Ni layer thickness for ASTM A517Gr70 specimens with 1 mm steel thickness, exposed to the 1M acetic acid solution with and without a cathodic charging current of 0.22 mA/cm².

FIGURE 4 - Hydrogen permeation curves for different environments.

Figure 4 reveals the influence of the environment on the permeation curve of the ASTM A516Gr60 specimens with 3 mm thickness and 30 m m Ni layer.

FIGURE 5 – Hydrogen permeation curves normalized for peak flux and time for 1 mm thick specimens.

FIGURE 6 - Hydrogen permeation curves normalized for peak flux and time for 3 mm thick specimens.

Figure 5 shows the impact of the Ni layer thickness on the normalized permeation curves (J/J_max versus the dimensionless time parameter  = D_eff t/l²) for the 1 mm thickness specimens exposed to the 1M acetic acid solution with and without a cathodic charging current. Whereas, Figure 6 shows the same effect for the 3 mm thickness specimens exposed to the NACE TM0177 solution.

FIGURE 7 - Normalized hydrogen permeation curves for TM0284 solutions.

Figure 7 indicates the influence of the exposure of 3 mm thickness specimens with 30 m m Ni layer to the NACE TM 0284 with bicarbonate additions of 30, 50, 80 and 200 ppm, on the normalized permeation curves. The normalized curves were plotted from the beginning of the test until the peak current value was achieved.

FIGURE 8 – Effective Diffusivity as a function of Ni layer thickness.

Figure 8 shows the effective diffusion coefficient D_eff, calculated according to the T_lag formula as a function of the Ni layer thickness. This had shown for the specimens that were exposed to the 1M acetic acid solution, with and without a cathodic charging current.

FIGURE 9 – Normalized hydrogen permeation current versus severity of HIC measured at various charging level.


FIGURE 10 –Hydrogen concentration versus severity of HIC measured at various charging levels.

Figure 9 shows the relationship between the cracked area that grew during the permeation test, detected by UT after the permeation test and the peak current. Figure 10 reveals the relationship of the same cracked area but now versus the hydrogen concentration on the entrance side. This was calculated assuming steady state, and adopting the diffusion coefficient of 5 x 10^-6 cm²/sec.

FIGURE 11 – Hydrogen permeation measurements for plant FCC Unit. (Condition 1)

FIGURE 12 – Hydrogen permeation measurements for plant FCC Unit. (Condition 2)

FIGURE 13 – Hydrogen permeation measurements for plant FCC Unit. (Condition 3)

Figure 11, 12 and 13 show permeation curves for of a cell coupled to a high pressure vessel in an FCC light-end recovery unit of a Petrobras’ refinery where the HFM technique was is being being used.

Figures 1 and 2 show that when thin Ni and Pd layers are used on the exit side of the permeation cell, the peak current is about 9 times higher than the one achieved by bare steel. In the case of high hydrogen charging environments and for specimens of 1.6 mm and 3 mm thickness, ; Figures 1 and 2 show that thin Pd and Ni plating leads to the same range of peak current - 8 to 12 A/cm. There is no noticeable significant influence of the higher Ni layer thickness (30 m), either on the measured peak current, or in the time required to reach the peak current. The difference in peak current is much more affected by surface factors on the entrance side, such as FeS layer formed during the test, and by the formation of irreversible traps (i.e. blisters) in the steel during the test. It is expected that in the case of equipment with thicknesses above greater than 3 mm, the influence of the Ni layer thickness will be even less pronounced.

Figure 3 shows that, in the case of a less severe hydrogen charging environment, where the peak permeation current is on the order of 1 m A/cm. It can also be seen that there is no influence of the thick Ni layer in the peak current achieved, although in this case there is a pronounced influence in the time required for the peak current to show up.

The influence on the time required for the peak current to be achieved reached is shown in the normalized permeation curves (J/J_max versus  = Dt/l²) presented in Figures 5 and 6. In Figure 5, for mild hydrogen charging environments, the higher Ni layer thickness causes a shift to the right of the entire permeation curve, with a delay in the detection time. It is also shown that the Ni layer thicknesses in the range from 0.1 to 0.5 m m had similar behavior and the most responsive current readings. Ni layer thickness of 5 m m resulted in a delay in the normalized permeation curve, and Ni layer thickness of 30 m m caused an even further delay in the permeation process. Figure 6 shows that, in the case of higher hydrogen charging environments, the influence of the Ni layer thickness is restricted to the beginning of the permeation curve. At currents higher than 50% of J/J_max all curves merge, irrespective of the thickness of the Ni layer.

Results showed clearly that, for engineering purposes, it would be possible to use the thick Ni layer procedure to monitor the permeation current of equipment. There would be a delay in the permeation current reading, which would be more or less pronounced as the hydrogen charging was less or more severe. Nevertheless, the peak permeation current would be correctly read. The benefit would be seen is a practical one manifested in the reduced maintenance needs of the HFM cell, since the thick Ni layer would result in less deterioration with time and longer term performance.

Once the Ni layer thickness was chosen, sSeveral hydrogen charging environments had to be evaluated in order to set threshold values of hydrogen charging severity that would assure that above which actual defects would not grownot propagate by HIC. Figure 4 shows the permeation curves for a number of environments. It can be seen that the addition of a filming amine inhibitor to the NACE TM0177 solution or of bicarbonate to the NACE TM0284 solution generally allowed reduced hydrogen charging resulting in permeation currents in the range of 0 to 12 m A/cm.

An interesting result was that, for the NACE TM0284 solution with and without bicarbonate additions, the normalized permeation curves (Figure 6) had the same behavior. This was not expected since the peak current changed from 1.8 m A/cm, in the case of 200 ppm bicarbonate addition, to 5 m A/cm, when no addition was done.

Figure 8 presents the diffusion coefficient calculated through the D_Tlag formula given in ASTM G148, as a function of the Ni layer thickness, for the ASTM A516Gr70 steel exposed to the 1M acetic acid solution, with and without a cathodic charging current of 0.22 mA/cm². As expected, the diffusion coefficient decreaseds with the increase in the Ni layer thickness. For smaller thinner Ni layers thickness (0.1 m m), the diffusion coefficient of 8.0 x 10^-6 cm²/sec was not much lower than the one measured by Yoshizawa⁽¹³⁾ for the same Ni thickness; whereas, in the case of 1 m m Ni layer the diffusion coefficient dropped to 2.0 x 10^-6 cm²/sec and correlated even better with his results⁽¹³⁾.

To define hydrogen permeation currents below which no cracks would either nucleate or propagate, specimens of conventional ASTM A516Gr60 steel were UT inspected before and after exposure to various wet H₂S environments of measured hydrogen charging severitiesy. Details of the UT procedure are given elsewhere⁽⁹⁾. Figures 9 and 10 present the percent of cracked area versus

• the peak permeation current, and
• against the hydrogen concentration

calculated using the peak current and a diffusion coefficient of 5 x 10^-6 cm²/secg.

As it can be seen, no cracks were observed when the permeation current normalized by the steel thickness was lower than 3.5 m A/cm or corresponding to a subsurface hydrogen concentration of less than 20 m mol/cm³ on the entrance side of the specimen. The hydrogen concentration on the entrance side is was calculated per ASTM G148 as C₀= I_peakl /FD_eff (where C₀ is the hydrogen concentration on the entrance side, I_peak is the permeation peak current and F is the Faraday constant).

Although no cracks have been detected for the abovementionedose hydrogen concentrations during the relatively short term testslevels, it was considered safer to specify the a more conservative level of 1 m A/cm was taken as the a possible threshold for in- service equipments that operates under sour conditions for much longer time. This was done because the normalized permeation curves presented in Figure 5 do have a steeper angle than the theoretical curve, which denotes trapping from possible crack nucleation⁽¹⁴⁾. No clear correlation was obtained in this work between the exposure time to the sour environment and the cracked area ratio, although further evaluation should be carried out.

This work was aimed at the The goal of this program was the development of the a hydrogen permeation procedure probe and methodology that would allow the integrity assessment of high- pressure vessels vessels in the FCC light-end unit of a Petrobras’ refinery. during its remaining lifetime. In order to allow the use of the technique, UT inspection was performed to select a place in the vessel initially free of defects where the hydrogen probe could be mounted. Other regions of the vessel were UT inspected in order to correlate the peak permeation current with the crack growth. During the operation cycle of the equipment, UT inspection could be initiated once the hydrogen threshold level was achieved.

In a future phase of this program, it is planned to couple expand more sensitive (real-time) HFM techniques with the use of fracture mechanics and a finite element model. It is hoped that this effort will provide guidance on the inspection plan for the equipment. The model will consider previous defects and then should inform expected crack sizes. It will also attempt to give unit operators a monitoring device that can be used to monitor hydrogen flux along with other process conditons. This will enable operators to minimize the damage caused by process upsets and optimize process conditions. Examples of work in progress involve use of these techniques to minimize water wash requirements and associated expenditures.

The sensitivity of the technique, and its ability to predict critical process conditions, were considered adequate, as shown by the following example: A permeation current of 1.0 A/cm was considered as the one below which there is no crack propagation. With the background current is oin the order of 0.05 A/cm²; the sensitivity of the technique should was required to be at least 5 times higher than the threshold permeation current. Therefore, for a vessel thickness of 40 mm, the background current would be 0.05 A/cm² times 4 cm which leads to a value of 0.2 A/cm, and the threshold value of 1.0 A/cm is 5 times higher than the background current, therefore easily measurable by the technique.

In fact, Figures 11, 12 and 13 show the permeation curves of an actual refinery vessel that was exposed to a wet H₂S environment. The vessel was has been monitored for the last two years and the cell has clearly shown process changes that caused different hydrogen charging severities. Figure 11 corresponds to a period where the unit had a feed stock, composed of a parafimic parafinic crude with extremely low N and S content, of 4000 m³/d, and the total wash water was of 150m³/d. Figure 12 corresponds to a period where the feed stock was raised 15%, to 4600 m³/d, due to the addition of 600 m³/day of a naphthenic crude with high N and S content., and t The total wash water was also increased to 320 m³/d. Although there was a significant increase in the wash water, compared to the increase in the feed stock, the latter case had caused a much higher N and S, due to the naphthenic crude. Figure 13 shows the effect of keeping the feed stock at 4600 m³/d and decreasing the total wash water to 180 m³/d. The reduction in the amount of the wash water, caused an obvious increase in the amount of CN^- and HS^- in the vessel.

This hydrogen permeation probe, as previously mentioned, was not intended to be used to monitor control the process conditions (i.e. work in-progress) to monitor but the integrity potential for damage and resultant integrity of the plant equipment. Nevertheless, the probe was clearly able to show periods of process miscontrolupset leading to high hydrogen flux of sufficient levels to potentially cause damage., when When these miscontrol upsets extended for long periods of time the resulting in high hydrogen flux and possible would likely lead to wet H₂S cracking. However, it should be emphasized that, in order to implement control of the process, another permeation probe, with a much faster response should be used. This probe is currently under development and undergoing in-plant tests with the intent to be more responsive probe would and allow process control by the balance of water injection amount or by the use of inhibitor injections when needed. With the use of both cells, both process control and integrity assessments would be possible.

Based on the studies described herein, the following conclusions were made:

References  Top

1. Rolf K.Poeperling, Gerd Sussek, Effect of Metallurgical and Testing Variables on Hydrogen Induced Corrosion Behaviour of Structural and Pipelines Steels. Paper No. 51, Corrosion/95, NACE International, Houston, Texas, March 1995.
2. P. Manolatos, M.Jerome, J.Galland. “Necessity of a Palladium Coating to Ensure Hydrogen Oxidation during Electrochemical Permeation Measurements on Iron”, Electrochimica Acta, V.40, n° 7, pp 867-871, 1995.
3. P.Manolatos, M.Jerome, “A Thin Palladium Coating on Iron for Hydrogen Permeation Studies”, Electrochimica Acta, vol.41, n°3, 359-365, 1996
4. S.Yoshizawa, K.Tsuruta, K.Yamakawa, G.Boshoku. Corrosion Engineering, vol. 24, pp 511, 1975.
5. G148, “Standard Practice for Evaluation of Hydrogen Uptake, Permeation, and Transport in Metals by an Electrochemical Technique”, ASTM, West Conshohoken, PA, 1997.
6. TM 0177, “Laboratory Testing of Metals for Resistence to Specific Forms of Environmental Scracking in H2S Environments, NACE International, Houston, Texas, 1996.
7. TM 0284, “Evaluation of Pipeline and Pressure Vessel Steels for Resistence to Hydrogen Induced Cracking”, NACE International, Houston, Texas, 1996.
8. T.Kushida, T.Kudo, I.Takeuch, “Effect of Environmental and Metallurgical Factors on Full Ring Test”, Corrosion, n°67, 1994
9. Carnival R.O., Damasceno S.D., Joia C.J.B., Bezerra P.S.A., “Ultrasound and Permeation Techniques Applied to Evaluate Structural Integrity of Equipment Used in Refinery Units”, ASNT Fall Conference, Nashville, Tenesse, October, 19-23, 1998