Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

Novel Control of Hydrogen Cracking in Thick Steel Welds


Novel Control of Weld Metal Hydrogen Cracking in the Welding of Thick Steels

Joanna Nicholas and Richard Pargeter

TWI Ltd, Granta Park, Great Abington, Cambridge, CB21 6AL

Paper presented at the International Steel and Hydrogen Conference
28 September 2011


The incidence of upsets in hydrogen control during welding is often only identified at inspection as a fabrication hydrogen crack where the only remaining option is to carry out a repair. The work involved in excavating the crack and carrying out the repair can be very expensive and time-consuming, particularly for thick sections.

The application of a method to monitor hydrogen evolution to discriminate between “normal” and “high” hydrogen levels during welding is presented in the use of a modified Hydrosteel™ non-invasive hydrogen patch probe. This probe showed a difference in hydrogen evolution rate between welds made with differing initial hydrogen levels.

A technique for identifying upsets in hydrogen control during welding and applying immediate remedial action to avoid hydrogen cracking is presented. This technique has been shown to successfully identify upsets in hydrogen control and the ability to eliminate fabrication hydrogen cracking by extending the interpass time at the time of upset detection, has been demonstrated.


There are occasions when it would be beneficial to know whether there has been a problem with hydrogen control when welding a thick section steel before completion of the weld. It may be that a source of hydrogen is introduced during welding (inadequate cleaning or a region of damp flux, for example), and that if early detection and immediate action can be taken to avoid hydrogen cracking in the component, the requirement for repair and the related costs could be avoided. Such action may be increasing the interpass time to allow additional hydrogen diffusion prior to welding the next pass, or maintaining a postheat on completion of welding to allow hydrogen diffusion before the weld cools.

A probe that allows rapid measurement of hydrogen flux from steel surfaces offers the possibility of detecting such upsets. The Hydrosteel™ is a non-invasive hydrogen patch probe (Dean and Fray, 2000). It is particularly appropriate in that no other hydrogen monitoring systems can be used without an enclosed surface area, and with little or no surface preparation or can operate at elevated temperature. The probe is commonly used in corrosion monitoring situations, and the collection plate for this purpose is 150mm diameter. This is too large for application to welds but the higher fluxes from a hot weld mean that a smaller collection area would nevertheless be effective. Support has been received from the manufacturer to develop a smaller collection plate (~12mm diameter).

In a possible application, the hydrogen flux would be measured on welds at a temperature where the risk of cracking is still low. Provided this measurement can be interpreted, it would be possible to rapidly estimate the risk of cracking and allow the operator to decide ‘on the spot’ whether the weld should be maintained at some safe (high enough) temperature for a further length of time, or allowed to cool down. In essence, a weld procedure qualification can identify a set of parameters that are acceptable, and then this benchmark can be compared and correlated to the values measured during the welding operation.


A programme of work was therefore carried out in the first place to establish the feasibility of measuring hydrogen flux and temperature in-situ during welding for single layer welds, and then to demonstrate the use of the Hydrosteel™ for real welds, both in terms of actual hydrogen content and in terms of avoidance of cracking.

Practical Considerations

Hydrosteel™ probe modifications

The Hydrosteel™ is a non-invasive hydrogen patch probe that requires little or no surface preparation. It is commonly used for corrosion monitoring often at relatively low temperatures, but can be used at elevated temperatures. The probe works on the principle of analysis of gas volumes collected over a fixed area. The fixed area is 'swept' by using a vacuum to draw the gases into the probe. For the larger (150mm diameter) corrosion probe, the gases are guided to the centre of the probe with a spiral ridge in the plate. For higher temperature use, a smaller collection area can be used, and the manufacturer of the Hydrosteel™ probe has developed a “weld sized” probe (Figure 1), in conjunction with this project. Rather than the spiral design, fluted channels in the outer surface have been used to assist in effective gas collection. The rate of air flow into the probe for analysis is 0.5ml/s. The probe is coupled to the workpiece using magnets.

Pargeter 2 Figure 1
Figure 1.The Hydrosteel™ weld probe, showing the fluted channels for air suction to allow gaseous hydrogen to be collected from the surface of the weld

In order to record the workpiece temperature, a thermocouple was added to the upper surface of the probe. Thus, the temperature and hydrogen effusion could be recorded together.


Although the hydrogen probe does not require extensive surface preparation, a surface which is flatter than the undulating surface of a manual weld is needed to ensure good coupling of the probe to the weld cap.

The aim of surface preparation was to remove a minimal quantity of weld metal whilst achieving a flat surface and only a fairly small flat region (16mm diameter) was needed. This was achieved by drilling a flat bottomed hole using a milling cutter.

The drill used was electro-magnetically clamped to the workpiece, with manual control on the drill depth. Trials were carried out using cold and hot welds. With the cold welds, the flat region was achieved easily, but with the hot welds, the electromagnetic clamping mechanism was not always sufficient to ensure that the drill did not skate over the weld surface. An additional stabilising clamp was used and good surface preparation was achieved. The cutter was selected to give a 16mm flat bottomed hole, and it is noted that some skill was required to achieve a flat area which did not remove excessive amounts of weld metal. The surface preparation took a few minutes, and affected the maximum temperature and hydrogen profile recorded (i.e. shorter preparation time resulted in higher temperature and therefore higher initial hydrogen effusion rate).


As it is known that hydrogen diffusion rates vary with temperature (Coe and Moreton, 1967), any measurement of hydrogen effusion or flux must also be accompanied by simultaneous temperature measurement if the results are to be related to hydrogen content. A series of experiments was therefore carried out, firstly, to demonstrate that the probe can be modified to make successful measurements of hydrogen effusion from a weld, and secondly to show whether a fixed level of hydrogen effusion (taking temperature effects into account) can be related to a diffusible hydrogen level in the weld.

A three bead single layer was deposited for each of three manual metal arc (MMA) consumables; a vacuum packed low hydrogen electrode, a basic electrode and a cellulosic electrode. These were selected to give a wide range of hydrogen levels. All the welds were made at approximately 0.4kJ/mm heat input on 8mm thick C-Mn steel plate, without preheat. Shortly after the welds were deposited, a flat surface was prepared. The probe (with thermocouple attached) was then held onto the flat region and the temperature and hydrogen effusion measured.
Initially, the feasibility of using a modified probe was examined to ensure that temperature and hydrogen effusion data could be collected. A further series of experiments was then carried out, using the same conditions, but this time allowing the response of the hydrogen monitor to reduce to 25nl/cm2/s, approximately the effusion rate observed for low hydrogen consumables after about four minutes, irrespective of the recorded temperature. The welds were chilled rapidly in liquid nitrogen to freeze in the remaining hydrogen. These were then stored in solid CO2 prior to analysis to determine the diffusible hydrogen remaining in the weld. This was carried out using gas chromatography employing a collection temperature of 150°C for six hours.

Precise computation of hydrogen concentration (even assuming totally uniform initial distribution) from effusion rate and temperature under dynamic cooling conditions is a very challenging concept. It was however, considered likely that a sufficiently reliable assurance of a safe hydrogen level could be achieved by ensuring that the effusion rate fell to that measured at elevated temperature for a known safe hydrogen level.

To explore this, manual metal arc weld deposits using consumables of different hydrogen potential were made for hydrogen analysis, using the equipment used for standard hydrogen determination in weld metal. Sets of three deposits were made at heat inputs of approximately 1.5kJ/mm to achieve a wide weld bead and were tested in accordance with ISO 3690, except that the welds were allowed to cool in the jig (shown in Figure 2) to 80°C prior to quenching. The welds were analysed for hydrogen concentration using gas chromatography.

Pargeter 2 Figure 2
Figure 2 The jig used for hydrogen determination to ISO 3690

Following an initial set of rutile welds, three further rutile weld deposits were made and the caps milled to allow the probe to be positioned on the sample. The hydrogen effusion rates were recorded at 80°C prior to quenching to allow analysis in accordance with ISO 3690.

Three more very high hydrogen (cellulosic) weld deposits were then made The welds were allowed to cool to 80°C, but this temperature was maintained until the hydrogen effusion rate reached the average level for the rutile deposits at 80°C. The weld was then quenched for analysis in accordance with ISO 3690.

Cracking in multipass welds

To assess whether the methodology can be applied to real welds to prevent hydrogen cracking, a submerged arc welding procedure was developed to weld a multipass bead in groove weld in Q1N type low alloy steel in order to avoid cracking. A preheat of 120°C was applied, with an average heat input of 2.3kJ/mm. The hydrogen effusion was measured at three points (after the first, fifth and final passes) during deposition of the simulated procedure qualification weld (Weld A), at the minimum preheat/interpass temperature.

For a second weld, flux that had been exposed to moisture in order to increase its hydrogen potential was used for the first pass, which was also deposited without preheat, to simulate a serious deviation from the intended welding procedure (Weld B). The remainder of the procedure was as for Weld A. The hydrogen effusion was measured at the same key points in the welding.

For Weld C, the procedure for Weld B was repeated, (i.e. the first pass was deposited with high moisture flux and no preheat), and the hydrogen effusion was measured at the key times. In this case, however, the preheat was maintained (interpass time extended) until the hydrogen effusion rate had reduced to the same level as for the first weld at that stage. All welds were allowed to cool naturally on completion, and left for 72 hours before inspection. They were then examined for hydrogen cracking using ultrasonic testing and metallography.


Relationship between temperature and effusion rate

The initial trials confirmed that the method could be used to successfully record the hydrogen effusion and temperature over time. The Hydrosteel™ has two methods of operation - continuous measurement and measurement over one minute; in both cases, measurements are taken at five second intervals. Using the one minute monitoring option, the drop in hydrogen and temperature were recorded over ten minutes (Figure 3), for a basic electrode. The gaps in the data correspond to the resetting of the probe.

Pargeter 2 Figure 3
Figure 3 Hydrogen and temperature response for a basic electrode, deposited at approximately 0.4kJ/mm
This was repeated using the continuous monitoring option for the cellulosic electrodes and for vacuum packed basic electrodes (Figures 4 and 5). The undulation of the traces is principally a result of some instability of the probe, giving some variability in coupling.
Pargeter 2 Figure 4
Figure 4 Hydrogen and temperature response for a cellulosic electrode, deposited at approximately 0.4kJ/mm. The maximum level of flux that can be recorded by the probe is 500,nl/cm2/s
Pargeter 2 Figure 5
Figure 5 Hydrogen and temperature response for a vacuum packed basic electrode, deposited at approximately 0.4kJ/mm

As indicated in Figure 4, the hydrogen flux from the cellulosic weld was very high, in excess of 500nl/cm2/s until the temperature of the weld had reduced to below 130°C. For the basic electrodes, the hydrogen released had dropped to 25-30,nl/cm2/s by the time the weld had reached this temperature.

Hydrogen analysis showed that the diffusible hydrogen remaining when the Hydrosteel™ measurement reached 25nl/cm2/s was undetectable for the basic electrodes, and was at the limit of detection (0.1ml/100g) for the cellulosic electrode. A higher value would be expected for the same effusion rate recorded at a lower temperature.

Thus, it is possible (for single layer welds) to record the time, temperature, hydrogen flux profile, and to identify high and low initial concentrations of hydrogen in the weld from the measured hydrogen flux.

Assessment of hydrogen levels

The first set of samples was not subjected to measurement of effusion rate and did not have cap dressing. For the rutile electrodes, the values of hydrogen were between 10.1 and 23.0ml/100g, and for the cellulosic electrodes, the hydrogen level was in excess of the calibration limit of the analyser.

The average hydrogen effusion rate recorded at 80°C for the second set of rutile welds (with local cap dressing) was 33.1nl/cm²/s, from the range 19.8 to 44.9nl/cm²/s. The values recorded for the cellulosic electrodes on completion of the hold at 80°C were in the range 33 to 42nl/cm²/s. The plot of the change in hydrogen effusion rate with time for the cellulosic welds is shown in Figure 6.

Pargeter 2 Figure 6
Figure 6 Change in hydrogen effusion rate with time at 80°C for cellulosic welds, with measured hydrogen content marked

The average hydrogen level in the rutile welds after cooling to 80°C with hydrogen effusion rate monitoring was 23.1ml/100g. This compares well with the range of 10.1 to 23.0ml/100g obtained for the rutile electrodes assessed in a manner similar to that described in the standard. The average hydrogen level in the cellulosic welds after cooling to 80°C and holding until the hydrogen flux was similar to the level in the rutile welds was 22.5ml/100g. This compares favourably with determination of hydrogen after cooling to 80°C in the rutile electrodes (23.1ml/100g).

Cracking in multipass welds

Observations during welding

The damp flux passes were notable due to the appearance of a blue flame around the wire, compared only a little smoke around the wire for the dry flux, and a porous weld surface, particularly at the start and stop locations. It is considered unlikely that such a severe upset would occur unnoticed during any form of production, and remedial preheating could be applied at that stage. However, this very severe case was used to encourage cracking.

Hydrogen measurement

The measurements on these welds were made at the minimum preheat temperature of 120°C. For the control sample (Weld A), when the weld had cooled to 120°C, the hydrogen effusion rate was 10nl/cm²/s for the first pass, 12.5nl/cm²/s for the fifth and 22.5nl/cm²/s for the final pass.

For Weld B, the hydrogen effusion rate at the same temperature and points of measurement were 41nl/cm²/s, 17.6nl/cm²/s and 24.3nl/cm²/s respectively. After the first pass of Weld C, the hydrogen flux was allowed to decay to 10nl/cm²/s, whilst the preheat/interpass temperature was maintained at 120°C. The hydrogen flux measured at other locations in the weld was slightly less than that measured for Weld A, at 11.4nl/cm²/s and 21.1nl/cm²/s respectively.

Ultrasonic testing

The TOFD inspection was applied using techniques to identify longitudinal and transverse flaws, as for the applied procedure, any fabrication hydrogen cracking would be expected to be transverse to the weld. Weld B, which had the first pass made with damp welding flux but no remedial measures applied, exhibited some isolated indications, and an indication similar to lack of fusion. All the other welds also showed isolated indications.


At all the locations where indications were identified in Welds A and C, only collapsed pores, or entrapped slag were found on metallographic sectioning. However, some weld metal cracking was identified associated with other indications in Weld B (Figure 7). The crack path was consistent with hydrogen cracking, and was transverse to the welding direction.

Pargeter 2 Figure 7
Figure 7 Weld metal hydrogen cracking identified in Weld B


The practical trials have demonstrated that it is possible to measure the hydrogen effusion from a cooling weld, along with the temperature profile, using the modified Hydrosteel™ probe. The difference between the hydrogen effusion from nominally high hydrogen and low hydrogen consumables was also significant (at 150°C, >500nl/cm²/s for high hydrogen (scale A) and 50nl/cm²/s for low hydrogen (scale D or E)). This gives confidence that the modified probe could discriminate between adequate and inadequate hydrogen control during welding.

The monitoring of hydrogen evolution to a common, low value shows that when the level of diffusible hydrogen is measured, the remaining diffusible hydrogen is at a low level. However, the significance of hydrogen contents in the completed welds (for example, 0.1ml/100g recorded in this case) on the likelihood of cracking needs to be explored. Kasuya, Hashiba, Ohkita and Fuji (2001) indicate that for a consumable with a measured hydrogen content of 5ml/100g, the hydrogen level in the completed multipass weld will drop to approximately 0.4ml/100g after seven days. The significance of this was not considered in terms of crack susceptibility, but only in terms of hydrogen distribution.

The results from the hydrogen level assessment have demonstrated that the time needed to hold a weld at a certain temperature to allow hydrogen effusion prior to cooling to achieve a desired hydrogen level can be identified by measuring the hydrogen effusion rate. Specifically, the results showed that the hydrogen content of cellulosic weld metal (with an expected starting hydrogen level of 60-100ml/100g) could be reduced to a level comparable to that of rutile electrodes by holding at 80°C until the hydrogen effusion rate was at a similar level as that recorded for the rutile consumables at 80°C.

Fabrication hydrogen cracking identified in a weld which had a significant deviation from the intended welding procedure (Weld 3) was avoided in the weld with remedial interpass time (Weld 5). The Hydrosteel™ successfully identified a deviation in the weld procedure in terms of hydrogen control, and enabled remedial action to be undertaken shortly after welding, along with the fixed end point for the extended interpass time.

In principle, cracking could have been prevented by a postheat treatment on completion of welding, but for mid-weld postheating, the extended time usually associated with postheating is greatly reduced due to the much shorter diffusion path for hydrogen to a free surface than if it were trapped in the same location for the entirety of welding. Thus not only can the effusion rate monitoring identify the problem, but also it can reduce the time required for remedial action. A further reduction in time can also be achieved as the technique can identify the end point (minimum time) for the mid-weld postheating, at which the hydrogen level has reduced to (or below) a ‘safe’ level. Conventional postheating does not have a minimum time associated with it, and can be many hours, owing to the limited knowledge of the distribution of hydrogen within the weld.

The measurement of hydrogen effusion was matched to the minimum preheat temperature in the control weld, as the effusion rate will vary with temperature and time. From a practical perspective, this results in a relatively straightforward monitoring of the weld, as after welding and preparing the surface for the probe, the temperature does not need to be continuously recorded, just maintained at or above the measurement temperature. The Hydrosteel™ probe can be used to monitor continuously, or at intervals, until the hydrogen effusion rate achieved for the weld procedure qualification is reached. Initial work with the probe explored the monitoring of temperature and time simultaneously, the aim being to calculate hydrogen concentration in the steel using an algorithm relating effusion rate and diffusivity, which is a function of temperature. However, if the monitoring of a procedure qualification weld occurs at the specified minimum interpass (preheat) temperature, such calculations are not needed, and monitoring of a production weld does not need to be additionally complicated by monitoring and recording of temperature as well as hydrogen effusion rate in the same instrument.


Following a programme of work to determine a practical means by which hydrogen cracking can be prevented by remedial action during welding, thereby avoiding the costs of repair, the following conclusions can be made:

  1. The technique has been shown to successfully identify the time needed to achieve a required diffusible hydrogen level.

  2. The Hydrosteel™ was able to identify an upset in hydrogen control during welding, on a multipass weld.

  3. Application of an extended interpass time at the minimum preheat level, for a time determined from measurement of hydrogen effusion rate prevented hydrogen cracking in a multipass weld.


The results are encouraging, and suggest that the technique can be developed into a practical tool for use in fabrication. The methodology could be applied as follows:

During procedure qualification, use the Hydrosteel™ to monitor the hydrogen effusion rate at a number of locations throughout the thickness of the weld, including after the final pass is made, making note of the hydrogen effusion rate at the minimum preheat level. Record these on the weld procedure qualification.

During welding, use the Hydrosteel™ to monitor the hydrogen effusion rate at the same stages of welding. If the hydrogen effusion rate is greater than that recorded during weld procedure qualification, increase the interpass time (maintain preheat) until the level of hydrogen effusion rate is at or below the values recorded during weld procedure qualification. Resume welding at this point. If the hydrogen effusion rate is already at or below the value recorded during weld procedure qualification, welding can proceed until the next hold point, when this is repeated. After completion of welding, use the Hydrosteel™ and hold the weld at the minimum preheat level until the hydrogen effusion rate is at or below that recorded on the cap of the weld procedure (effectively a postheat treatment), after which cooling can progress (using the methodology recommended in the procedure). Carry out all other operations (NDT, PWHT, etc) as normal after this.


  1. BS EN 1011-2, British Standards Institution 2001: ‘Welding and allied processes. ‘Determination of hydrogen content in ferritic steel arc weld metal’ March

  2. Coe F R and Moreton J, 1967: 'Estimation of diffusivity coefficients for hydrogen in ferrous materials' British Welding Journal Vol. 14, no. 6, June pp. 313-20.

  3. Dean F W H and Fray D J, 2000: ‘Ultrasensitive technique for detection of hydrogen emanating from steel and other solid surfaces’, Materials Science and Technology, January, Volume 16 No 1. p.41-46. ISSN 0267-0836

  4. Kasuya T, Hashiba Y, Ohkita S, and Fuji M, 2001: 'Hydrogen distribution in multipass submerged arc weld metals' Science and Technology of Welding and Joining 2001 Vol. 6 No. 4 pp.261-266.

For more information please email: