C S Wiesner (1), S J Garwood (2), R Sandström (3), D M Street (4) and K J Coulson (5)
(1) TWI Ltd, Cambridge, UK
(2) Rolls Royce plc, Derby, UK
(3) Royal Institute of Technology, Stockholm, Sweden
(4) formerly ICI Engineering (UK), now retired
(5) formerly Air Products (UK), now retired
Following a brief historical introduction to brittle fracture prevention for pressure vessel applications in various countries, the derivation of the toughness requirements of the European pressure vessel code are described. The background to the specification with respect to yield strength range and the treatment of sub-size Charpy is presented. It is concluded that whilst the European requirements do represent a compromise of various national codes, their correct application will lead to fracture safe pressure vessels.
The consequences of a catastrophic failure of pressure vessels are very serious, both in terms of financial implications and the risk to human life. An essential consideration in the design and operation of vessels, particularly those made from ferritic steel and operating at low temperatures, is the avoidance of the possibility of brittle failure. The specification of material requirements in national standards throughout Europe was designed to minimise this risk.
Over the last 10 years or so, European Standards have been developed and are replacing national codes, prEN 13445 and prEN 13480 (currently in draft form) being the European specifications for pressure vessels and metallic industrial piping, respectively.
This paper describes the background to the fracture toughness requirements of the draft European unfired pressure vessel standard prEN 13445 which were also adopted for the standard on metallic industrial piping. Most of the development work of the requirements took place in the years 1991 to 1998, but final discussion continued into 2001.
2. Historical background
The first fabrication code for pressure vessels was developed by the American Society of Mechanical Engineers in response to the large number of boiler explosions in the early 1900s. Whilst poor design, fabrication and operation played a major role in these incidences, failure by brittle fracture was also of concern.
The combination of circumstances which lead to brittle fracture was studied extensively after the Second World War because of the high incidence of welded ship failures. The significance of the presence of cracks in welds or geometric notches was recognised. It was also realised that the resistance of the material to crack propagation by a brittle mechanism could not be ascertained from conventional tensile tests. Reducing areas of stress concentration by design, improving fabrication practices, and specifying material impact properties were the methods adopted by classification societies in their rules to help prevent catastrophic failures. International codes and rules therefore began specifying a minimum energy derived from impact testing of material (conventionally performed using Charpy specimens) in order to provide sufficient fracture toughness which combined with other measures would minimise the risk of brittle failure.
From studies of casualty material, a Charpy energy of 15ft pounds (20 Joules) at the failure temperature seemed to minimise the risk of failure. To improve margins of safety, 20ft pounds (27 Joules) became a common specification for low strength steels. Hodgson and Boyd  however recognised that it was insufficient to guarantee adequate performance using Charpy energy alone and suggested that a percentage of fibrous fracture should also be specified. They recommended 35ft pounds (47 Joules) and 30% fibrous fracture as a minimum requirement. Despite this, in many regulations, Charpy energy requirements alone are specified. However, the allowable chemistry of the steels is closely controlled to limit the variability of the ductile-brittle transition curve.
In many international codes the adoption of a specified Charpy energy at a given temperature based on experience and linked to the minimum operating temperature has become accepted practice to minimise the risk of brittle failure (e.g. AD Merkblätter  , ASME III  , etc.).
2.2. British, Dutch and German approaches based on engineering practice and experience and correlations with structural tests
Although the theory and recommendations made for ship steel selection and fabrication practice were carried over into the pressure vessel industry, there followed a spate of brittle failures of pressure vessels under hydrotest in relatively thick sections in the late 1950s and early 1960s, e.g  . Following these experiences, the significance of thickness and residual stresses on the risk of brittle fracture began to be recognised.
In search of a more quantitative approach to material selection to avoid brittle failures in pressurised components, a series of wide plate tests was commissioned by UK industry at the British Welding Research Association (the forerunner to TWI). Design curves for the avoidance of brittle fracture were developed, based on pioneer work by Woodley, Burdekin and Wells  in the early 1960s. They correlated Charpy transition temperatures, measured in the parent plate, with the temperature at which survival was obtained in welded notched wide plate specimens at an applied strain ε = 4 ε Y (approximately 0.5% plastic strain which was considered an adequate performance requirement). Correlations were obtained for a range of thicknesses in the as-welded and post-welded heat treated conditions. This approach assumed that wide plate tests are representative of structural behaviour in that the test specimen is large (typically 1m x 1m), contains a weld with near yield magnitude residual stresses (in the as-welded condition) and a crack-like flaw. The transferability of the Charpy test results was based on the assumption that the fracture toughness of the notched region in the wide plate specimens directly correlates with Charpy toughness of the parent plate.
These results, together with engineering practice outlined by Cotton  , were drawn together to derive operating temperature vs. impact test temperature vs. material thickness relationships, known as design lines. The need for a Charpy requirement plus a fibrous growth specification was not taken up by the pressure vessel rules in favour of limiting the choice of steels allowed by the standard and requiring 27 Joules at the relevant temperature (depending on thickness and stress category) for material with an ultimate strength <450N/mm2, but 40 Joules for higher strength levels. For plate materials these values relate to the longitudinal orientation.
Following the establishment of the design rules concept in BS 1515:Part 1 Appendix C in 1965, various revisions occurred leading to the publication of BS 5500:1979 Appendix D 'Recommended practice for carbon and carbon manganese steel vessels required to operate at low temperature'.
Dawes and Denys  re-assessed the requirements and confirmed the general safety for the as-welded condition. However, they noted that plates with thicknesses less than 6mm had reduced levels of safety. Re-examining the data presented by Dawes and Denys and taking into account additional results from thin section wide plate specimens led Garwood and Denham  to propose modifications to the BS 5500 Appendix D design curves. These included more conservative, but safe, design lines for post-weld heat treated conditions and the combination merging into a single line of the design rules for thicknesses of and below 10mm. These proposals formed the basis of the revisions to Appendix D issued in August 1988 and these are still in place in the current British Document PD5500  .
An approach similar to the BS 5500 concept was adopted for the Dutch pressure vessel code  . Charpy test temperature (requirement = 27J) in the Dutch rules are determined from minimum design temperature, pressure and thickness, similar to the British specifications. The German AD Merkblätter  generally use a 27J Charpy requirement combined with extensive workmanship and inspection requirements based on experience. Exact impact toughness requirements depend on steel type, and stress categories specified in the code. Charpy tests are performed at the lowest permissible operating temperature (a function of steel grade, thickness and stress). The toughness requirements of the applicable materials apply to both parent material and weldments. The reader is referred to Steffen  who described the development and organisation of the German boiler and pressure vessel codes.
In summary, the avoidance of brittle fracture in German, Dutch and British standard code requirements is achieved by ensuring adequate toughness of all materials and welds (e.g. by specifying a minimum Charpy energy of 27 or 40J); designing to prevent high stresses; and stress relief or (post weld) heat treatment of thick sections; adopting fabrication procedures and inspection methods to minimise occurrence of defects and maximise the chance of finding them; and carrying out a proof test.
2.3 Fitness-for-purpose approach and French and Swedish codes
Fitness-for-purposes approaches using fracture mechanics procedures tend to be much more specific than the methods based on correlation and/or experience. These methods quantitatively assess design conditions to calculate what flaw dimensions are tolerable (i.e. what flaw size can be missed without detriment to the safe operation of the vessel), leading to fracture toughness or Charpy requirements.
Existing Swedish pressure vessel design procedures  are directly based on Charpy-fracture toughness correlations and fracture mechanics assessments ofassumed defect sizes. They were derived using fracture mechanics using a ¼ thickness deep reference flaw (as per ASME) and the Charpy versus fracture toughness correlations by Barsom and Rolfe  . The approach is described in detail by Sandström  .
The French Unfired Pressure Vessel Code (CODAP)  is also based on fracture mechanics principles together with correlations  between Charpy and fracture toughness properties. It is applicable to unalloyed and non-austenitic alloyed steels in welded pressure vessel components. Sanz  and Thomas and Grandemange  describe the development and organisation of the French boiler and pressure vessel standards, including CODAP.
3. European pressure vessel toughness requirements
3.1. General Development
The technical committees CEN TC 54 and TC 267 were charged with the task of developing the European pressure vessel and industrial piping codes, respectively. Various joint working groups of the two committees, made up of representatives from EU countries were set up to write various aspects of the code. The low temperature performance of materials used, came under the remit of joint working group B (materials) and C (design calculations). These joint working groups set up a sub-group to look specifically at low temperature aspects.
Initially, the design lines were based on the Swedish approach  . The low temperature sub-group compared the Swedish design lines  with the French code  , and also with the British  and Dutch  rules based on correlation between Charpy and structural (wide plate) test results. Checks were carried out comparing the British/Dutch rules (based on longitudinal Charpy specimen orientation) with the proposed design lines based on transverse properties (but allowing for the temperature shift between transverse and longitudinal Charpy specimens orientation), which confirmed that the European proposals were satisfactory.
The results showed that for PWHT conditions, the design lines were virtually identical in all cases whilst the European proposals were somewhat less onerous than the British and Dutch codes for the as-welded conditions. However, the European specification requires testing of all weld locations (including weld metal and HAZ) whilst parent steel testing was considered sufficient (up to a certain welding heat input) for the British and Dutch rules.
Allowing for the factors that all Charpy properties are to be determined using the more conservative transverse orientations in the European code, and after careful comparison of the thickness ranges, yield strength requirements and reduction factors, taking into account applied stress levels, it was possible to derive design lines for steels which are reasonably consistent with all four national codes.
The design lines of this so-called 'Method 2' approach applicable to ferritic steels (C, C-Mn, fine grained steels) and 1.5-5% Ni alloys are shown in Fig.1 and 2 based on the 1998 Draft European Unfired Pressure Vessel Standard prEN 13445-Part 2: Materials, Annex B: 'Requirements for Prevention of Brittle Fracture'. The Design Reference Temperature, TR, is defined as the Minimum Metal Temperature, TM (i.e. the lowest temperature occurring during operation, start up and shut downs, process upsets and pressure and leak tests), adjusted by a temperature term which is dependent on the applied stress (see below for details). The lines in prEN 13480-Part 2: Materials, Annex B are identical with the exception that a line for 6mm thickness is also included. This has implications for the required Charpy impact energy levels for sub-size specimens which are discussed in Section 3.3.
Fig.3 Comparison of code of practice (Route 2, now referred to as Method 1) and design line (Route 1, now referred to as Method 2) toughness requirements: difference between design reference temperature TR and material impact test temperature TKV for different thicknesses
However, after achieving overall consensus with respect to the design lines, the code of practice approach (now called 'Method 1') was re-introduced and can now be used as an alternative to quantify toughness requirements for pressure vessels. The reason for this was a direction to the sub-group from the Technical Committees 54 and 267 to recognise a simple procedure aligned to the European Pressure Equipment Directive. 'Method 1' therefore applies a simple Charpy requirement of 27J or 40J at the design reference temperature broadly following the German code. This method is applicable to the steels mentioned above and also includes requirements for 9% Ni alloy steels, austenitic-ferritic (duplex) stainless steels and austenitic stainless steels. The relevant requirements are shown in Table 1. Obviously, for those steels covered in both Methods 1 and 2, different toughness requirements will result when applying either method, similar to those shown in Fig.3.
For both Methods 1 and 2, there are stress-dependent temperature adjustments for the PWHT conditions which can be invoked if the applied membrane stresses are below 75% of the specified minimum yield strength (SMYS) (10°C adjustment), below 50% SMYS (25°C adjustment) or below an absolute value of 50N/mm2 (50°C adjustment). For the as-welded condition, an adjustment can only be invoked (because yield magnitude residual stress always apply) if the absolute value of membrane stress is below 50N/mm2 (40°C adjustment), this value is based on operational experience rather than technical justifications and can be reconciled by invoking crack arrest arguments, i.e. at such low applied stresses, crack arrest away from welds is likely if initiation should occur.
Table 1 prEN 13445-Part 2 and prEN 13480-Part 2 Code of Practice ('Method 1') Toughness Requirements. Reference Thickness >5mm
|Material||Required Charpy impact energy, J||Impact test temperature, TKV, °C||Valid for reference thickness||Conditions|
|Ferritic steel, 1.5 - 5% Ni alloyed steels
||Design reference temperature
||For as welded ≤30mm
||Yield strength ≤310N/mm2
||Design reference temperature (1)
||For post weld heat treated ≤60mm
||310N/mm2 < yield strength ≤460N/mm2
|9% Ni alloyed steels
|Weld materials (3) for austenitic steels and castings made from austenitic steels
|Austenitic-ferritic ('duplex') stainless steels
||Minimum metal temperature, TM
||Limited to minimum metal temperatures, TM ≥ -30°C
|(1) Impact values in the weld, HAZ and parent material of 27J are required.
(2) If higher impact values are required in the material specification, these requirements shall be fulfilled for the weld material at -196°C, if the design reference temperature < -196°C.
(3) For design reference temperatures ≤ -196°C. Requirements for base material shall be as given in the appropriate material standard.
A third approach ('Method 3') permits the application of fracture mechanics to determine the suitability of a pressure vessel. Any of the three methods can be applied by agreement of all parties.
3.2. Yield strength ranges
The Method 2 approach initially only differentiated between base materials of less than or greater than 310N/mm2 yield strength (approximately equivalent to the 450N/mm2 ultimate tensile strength limit in the existing Dutch and British rules). However, this meant that the as-welded requirements were over-conservative compared to some existing national standards. To overcome this, an additional figure was introduced to cover the yield strength range between 310 and 360N/mm2, see Fig.1. For a given design reference temperature, the design lines for the 310 to 360N/mm2 yield strength range are shifted by 20°C to colder material impact temperatures.
This modification was reconciled with fracture mechanics calculation by applying a PD6493:1991 Level 2 approach using an ASME reference flaw (¼ wall thickness deep by length six times depth) at a nominal applied stress of 2/3 the yield strength and yield magnitude residual stresses.
Under these circumstances, for an assumed thickness of 28mm (which is the approximate demarcation at which the design reference temperature and material impact test temperature are the same for as-welded vessels), the following stress intensity factors KI apply at the deepest point of the assumed flaw.
Thus, to prevent fracture of the assumed flaw in 28mm thickness, a material toughness of >70MPa √m is required for 310N/mm2 yield strength. If the yield is increased to 360N/mm2 an approximately 14% higher fracture toughness is required, for an increase in yield strength from 310 to 460N/mm2, the required increase is about 47%.
The original proposals for yield strengths ≤310N/mm2 for as-welded applications relate to 27J Charpy impact energy. For yield strengths greater than 310N/mm2 up to 460N/mm2, 40J impact energy was at first specified. Whilst Charpy impact energy versus fracture toughness (KIc) correlations are very variable, three common correlations give the following results:
|Correlation||Charpy impact energy, J||Fracture toughness,|
|Barsom and Rolfe 
|Marandet and Sanz  (Note a shift in temperature is also required)
|Sailors and Corten 
There are many others, but the overall trends indicate that for these strengths of pressure vessel steel, 27J gives an average of 85MPa √m and 40J give approximately 107MPa √m. This is broadly similar with the applied stress intensities given above and confirms that the original as-welded requirements give about the same level of tolerance to flaws for 310N/mm2 yield strength with 27J as for 460N/mm2 yield strength with 40J.
Inevitably, this means that a 360N/mm2 yield strength material tested to 40J is more flaw-tolerant and the step change in toughness requirements between 310 and 360N/mm2 may be over-restrictive.
The variation in applied KI for the 310N/mm2 yield strength level at the different thicknesses given in the draft code was therefore calculated and compared with the decrease in design reference temperature (ΔTR) due to thickness (read off the relevant design lines). Results are shown below:
|Thickness, mm||Increase in KI, %||ΔTR, °C|
These results indicate an approximately constant relationship between the increase in KI and the design reference temperature with increasing thickness. It can be seen that an increase in KI of approximately 18% (average value in above table) is equivalent to a temperature shift of approximately 12°C (average value in above table), and the modified design lines for the 310 to 360N/mm2 yield strength range (which are shifted by 20°C) are therefore reconcilable with fracture mechanics considerations.
3.3. Sub-size Charpy specimens
There are two occasions where sub-size Charpy specimens are employed:
(i) if the material thickness of interest is below 10mm, and
(ii) if the extraction of 10mm thick Charpy specimens is impractical.
(Sub-size Charpy specimens referred to in the following are defined as those of identical width but smaller thickness.) When relating sub-size Charpy results to full-size values, there are two issues of concern. One is the value of impact energy, and the generally accepted method adopts a simple net section area-scaling rule to calculate impact energy values for thinner specimens.
The second issue is the shift in transition temperature, ΔT, for thinner specimens. Two relations have been developed in the last ten years or so to address this issue, based on the measured transition temperature shift of normalised Charpy energies between 25J/cm 2 and 50J/cm 2 (corresponding to 20J to 40J in full-size specimens).
The relations are given below:
||= -0.7 (10-t)2
||(Ref.  )
||= 51.4 1n (2 (t/10)0.25-1)
||(Ref.  )
where t is specimen thickness.
Both give similar results in thickness range between 3 and 10mm.
For Charpy requirements to be equivalent, the measured Charpy energy (normalised by the specimen net section area in J/cm2 units) has therefore to be specified at a colder temperature for sub-size specimens. The table below gives two examples using ΔT1.
Table 2 Examples of equivalent sub-size Charpy requirement
| ||10x10mm specimens||7.5x10mm specimens||5x10mm specimens|
|Equivalent Charpy properties
|Equivalent Charpy properties
The design lines in prEN 13445 have been derived for, and are applicable to, full-size (10x10mm) Charpy specimens. This means that the material impact test temperature at which a given Charpy level is to be achieved applies to full-size Charpy specimens only. For vessel thickness below 10mm (case i) above), design lines below the current lines are therefore theoretically applicable. However, as it is impossible to extract full-size Charpy specimens from material of less than 10mm thickness, and results from sub-size Charpy specimens are shifted to colder temperatures as per above relations, the design line would have to be corrected upwards by ΔT. Further details are given by Towers  .
To avoid this cumbersome approach and because the two effects of lower design lines and upwards correction for transition temperature shift cancel each other out approximately, the low temperature sub-group has therefore agreed that prEN 13445 specifies the use of the 10mm design lines for all material thicknesses below 10mm, with impact energy requirements proportional to the specimen area, as in Table 2 above, but without the temperature adjustment, as the temperature adjustment are implicit in the use of the 10mm line for thicknesses below 10mm.
As mentioned in Section 3.1, the design lines in prEN 13480 do include a separate line for 6mm thickness. To compensate for this potential non- conservatism, the Charpy requirements for thinner piping are greater than suggested by the scaling of specimen areas as shown in the table below:
|Full-size requirement, J||Sub-size requirements, J|
|10 x 7.5mm||10 x 5mm||10 x 2.5mm|
This higher than proportional Charpy requirement for thinner specimens therefore compensate for the shift to warmer Charpy transition temperatures for thinner piping components. It is considered unfortunate that the pressure vessel and piping standards did not adopt identical requirements due to time pressures during the final drafting stages, but it may be possible to correct this in future revisions and both approaches are considered to result in safe components.
If sub-size Charpy specimens are extracted from thicker components, (case (ii) above), then the design line of the actual component thickness should be applied and the sub-size Charpy test temperature shall be reduced by the following amounts to allow for the shift in transition temperature for sub-size Charpy specimens. The values specified were rounded from the above ΔT1 and ΔT2, the results are given below:
|Reference size, mm||Sub-size, mm||Reduction in impact|
test temperature, °C
|10 x 10
10 x 10
10 x 10
10 x 7.5
10 x 7.5
10 x 5
|10 x 7.5
10 x 5
10 x 2.5
10 x 5
10 x 2.5
10 x 2.5
In this case, the impact energy requirements are proportional to the thickness reductions, as indicated for instance in Table 2 above. Note that prEN 13445, after initially including references to 2.5mm thickness makes no comparison for thicknesses below 5mm because such thicknesses are not foreseen in relevant European product codes.
When vessels are fabricated by welding, the toughness requirements outlined above (using either the code of practise Method 1 or the Method 2 design lines) are to be achieved in both weld metal and heat affected zone regions. In addition, welding production tests are required at all material impact test temperatures for thicknesses greater than 12mm and for material impact test temperatures below -30°C for thicknesses greater than 5mm.
4. Summary and conclusions
The European Codes for unfired pressure vessels (prEN 13445) and industrial metallic piping (prEN 13480) are approaching publication. The toughness requirements for these codes have been developed during 1992-2001 as a combination of relevant existing national codes. Three independent methods have been identified. These cover simple but conservative code of practice type Charpy stipulations; fracture mechanics-based design lines taking into account the effect of thickness and stress relief condition (including an allowance for cases where low applied stresses apply); and, if so agreed by all parties, a specific fracture mechanics treatment using fitness-for-purpose approaches as for instance specified in BS 7910  . Whilst the requirements do inevitably represent a compromise of various pre-existing national codes, their correct application will result in fracture-safe pressure vessels.
- Hodgson J and Boyd G M: Irons Inst. Of Naval Arch, 1958, Vol.100, 141.
- AD Merkblätter W10: 'Materials for pressure vessels' November 1987, and HP5/2 'Manufacture and testing of pressure vessels', July 1984, DIN, Germany.
- ASME III: 'Rules for construction of nuclear power plant components' and ASME VIII: 'Rules for construction of pressure vessels'. American Society of Mechanical Engineers, New York, 1989.
- Smith N and Hamilton I G: Paper No 591, Journal of the West of Scotland Iron and Steel Institute, Vol.76 (1968-1969) 117-187.
- Woodley C C, Burdekin F M and Wells A A: British Welding Journal, March 1964, 123-136.
- Cotton H C: ISI publication, April 1967, Vol.104, 31-38.
- Dawes M G and Denys R: 'BS 5500 Appendix D: An assessment based on wide plate brittle fracture test data'. International Journal Pressure Vessel and Piping, 15, 1984, 161-192.
- Garwood S J and Denham J B: 'The fracture toughness requirements of BS 5500'. Proc Conf ASME PVP 1988, Vol.PVP-7, Paper 88, ASME, June 1988.
- PD5500:2000: 'Specification for unfired fusion welded pressure vessels', British Standards Institution, London, 2000.
- Dutch Rules for Pressure Vessels: Sheet No.110, The Netherlands, 1989.
- Steffen H P: 'German boiler and pressure vessel codes and standards: materials, manufacture, testing, equipment, erection and operation', Proc. Conf. 'Pressure Vessel Codes and Standards - Developments inPressure Vessel technology-5', Ed. R W Nichols, Elsevier Applied Science, London, 1987.
- ECISS/TC22 N74: 'Fracture toughness concept of the Swedish pressure vessel code', February 1989.
- Barsom J M and Rolfe S T, ASTM STP 466, 281-302, American Society for Testing and Materials, 1970.
- Sandström R: 'Minimum usage temperature for ferritic steels', Scandinavian Journal of Metallurgy, Vol.6 (1987), 242-252.
- CODAP: 'French Code for the construction of unfired pressure vessels, Part M, Appendix MAZ, 1984.
- Marandet B and Sanz G: 'Evaluation of the toughness of thick medium strength steels by using linear elastic fracture mechanics and correlations between KIc and Charpy V-notch, Proc. Conf. 'Flaw growth and fracture'. STP 631, ASTM, Philadelphia, 1977, 72-95.
- Sanz G: 'Proposal of a quantitative method for the choice of steel qualities with regard to the risk of brittle fracture'. IRSID/AFNOR document, 1981.
- Thomas A and Grandemange J M: 'French Codes and Standards on boiler and pressure vessel technology', Proc. Conf. 'Pressure Vessel Codes and Standards - Developments in Pressure Vessel Technology-5', Ed. RW Nichols, Elsevier Applied Science, London, 1987.
- Sailors R H and Corten H T: 'Relationship between material fracture toughness using fracture mechanics and transition temperatures, in fracture toughness'. Proceedings 1971 National Symposium on FractureMechanics, STP 514, Part II, ASTM, Philadelphia, 1972, 164-191.
- Towers O L: 'Testing of sub-size Charpy specimens: Part 1 - the influence of thickness on the ductile/brittle transition'. Metal Construction, March 1986, Vol.18 (3), 171R-176R.
- Wallin K: 'Methodology for selecting Charpy toughness criteria for thin high strength steels: Part 1 - determining the fracture toughness'. Jernkontorets Forskning, Report from Working Group 4013/89, 28December 1994.
- Towers O L: 'Testing of sub-size Charpy specimens: Part 3 - The adequacy of current code requirements'. Metal Construction, May 1986, Vol.18(5), 319R-325R.
- BS 7910:1999 (incorporating Amendment No.1): 'Guide on methods for assessing the acceptability of flaws in metallic structures', British Standards Institution, London, 2000.
Originally published by Elsevier Science Ltd.