In recent years several research projects have been carried out at MPA Stuttgart to investigate the leak-before-break (LBB) behaviour of pressure-bearing components which are relevant to plant safety. In these investigations the test pipes have for the most part been made of ferritic material. International research programmes such as, for example, the Degraded Piping Programme (Wilkowski et al., 1986, Wilkowski et al., 1989. Degraded Piping Program, Phase II. Report NUREG/CR-4082, vol. 4, Sept. 1986, and vol. 8, March 1989, Battelle, Columbus, Ohio, USA) or the IPIRG-Program (Schmidt et al., 1991. The International Piping Integrity Research Group (IPIRG), Program—An Overview. SMiRT 11 Proceedings, Paper G23/1, Tokyo, Japan, August 1991) have also dealt with pipes made of austenitic materials. However, they were fabricated of not stabilized quality. To take into account the material of comparable components of German nuclear power plants, the experiments reported in the following are focussed on pipes made of Ti- and Nb-stabilized austenitic material. The results presented below relate to pipes containing circumferential defects subjected to internal pressure and external bending loading. As regards the ferritic components an overview of the experimentally determined results is presented. The predictive capability of engineering calculational methods are presented by way of example. The current programme of investigations is presented together with the testing techniques and the initial results.

Large plates of fine grained constructional steels of differing strength and toughness properties were loaded to fracture in tensile tests at room temperature. The unwelded plates had a test cross section of (21-28) x 500 mm and were weakened at their centre try transverse through-wall slits of various lengths. Accompanying the tensile tests. calculations based on fracture mechanics concepts were performed.

High temperature seamless pipes of dimensions O.D. x T x L = 457 x 15 x 5500 mm(3) made from 15 NiCuMoNb 5 material were loaded by internal pressure and a superimposed external bending moment. 27 tests with different geometries and positions of circumferential flaws in the base metal or in circumferential welds were performed at room temperature.Depending on the geometric conditions, the load bearing and failure behaviour of the investigated pipes could clearly be represented in a common failure diagram. Significant differences in maximum loads due to the position of the flaws in the structure, or whether inside or outside, could not be detected. Maximum loads were compared with calculations using applied engineering failure approaches, based on concepts of local flow stress and plastic limit load. Elastic-plastic fracture mechanics was applied to some of the component tests and the initiation and instability criteria verified.

Pipes made of 20 MnMoNi 5 5 steel and an MnMoNiV special melt, with an external diameter of 800 mm, wall thickness of 47 mm and length of up to 5500 mm, were provided with circumferential defects of defined length and depth. The pipes were loaded by internal pressure and a superimposed alternating bending moment. During the tests, deformation and crack growth were determined in the wall thickness and circumferential direction, and these were compared with calculated values. Pipes with an outer diameter of 226 mm and a wall thickness of 20 mm were used to investigate the leak-before-break behaviour in the dynamic sphere. These pipes also were made of 20 MnMoNi 5 5 steel and an MnMoNiV special melt, and were loaded with internal pressure and an alternating bending moment. The excitation took place at the resonance frequency of the pipes. The pipes also contained circumferential defects of defined length and depth.

A post-calculation by means of four engineering approaches, based on toughness, yield stress, plastic instability and ligament stress criteria, was made of the failure pressure on 134 pipes and vessels. The calculation was assessed by comparing the calculated with the experimentally obtained results. A statistical-based evaluation was made, since the results from the calculation and the experiment are affected by natural scattering of characteristic values, such as material properties and geometrical dimensions, among others. It was possible to find for each equation an individual weighting factor, which helped to improve considerably the approximation of the calculation to the experimentally determined failure pressure.

Over the last 35 years, researchers worldwide have conducted hundreds - if not thousands - of pipe fracture experiments. In the early years, researchers focused their attention on studying the failure pressure and crack propagation behavior of axially cracked pipe loaded by internal pressure. The earliest work was sponsored by the oil and gas industry and, as such, involved relatively thin-walled, low toughness carbon steel pipes. This work was eventually followed up by efforts in the USA and Germany on nuclear piping with axial cracks. In recent years, attention has turned to understanding the behavior of circumferentially cracked nuclear piping subjected to both pressure and bending loads. The loading histories for these experiments range from the relatively simple case of quasi-static, monotonic displacement control to the more complex cases of dynamic cyclic loading, and pipe system experiments. In this paper, two of the leaders in this research, i.e. Battelle in the USA and MPA Stuttgart in Germany, have collaborated to develop a database of pipe fracture experiments. The database includes data from other organizations as well as the data from Battelle and MPA. In addition, as part of this paper, an example of how the database was used to assess the failure pressure of axially cracked pipe is given.

Pipes made of steel 20 MnMoNi 5 5 having an external diameter of 800 mm, wall thickness of 47 mm and length of 5 m were provided with circumferential flaws in defined length and depth. They were loaded simultaneously by internal pressure, alternating or pulsating external bending moment. Deformation and the crack growth were determined in the wall thickness and circumferential direction and compared with calculated values.

Pipes with an outer diameter of 406 mm, a wall thickness of 12.5 mm and a length of 1000 mm were tested under tensile loading. They were prepared with part-circumferential flaws of uniform length but different depth. A fine grained constructional steel having a low upper shelf ISO-notch impact energy ≤ 50 J at test-/room temperature was chosen as test material. With the aid of the test results, the predictions of established approximation methods for the calculation of maximum load and, as far as possible, the load at crack initiation were checked. The applied methods which are based on modelling of the pipe by means of flat tensile specimens proved to be unsuitable for estimation of the maximum load. The methods for maximum load evaluation which had been specially developed for pipe geometries showed mostly a good agreement with the experiments. The calculations based on elastic-plastic fracture mechanics gave results which underestimated maximum load, whilst the crack initiation loads were to some extent overestimated.

For the testing of large scale specimens, a 12 MN high-speed tensile testing machine was designed and built at MPA Stuttgart. The aim was to determine the influence of high loading rates on the stress and strain behaviour of unwelded and welded components of ferritic and austenitic materials. This new generation of testing machines is driven by a propellant charge, and generates a maximum tensile force of 12 MN with a piston velocity of 25 m/s after a stroke of 20 mm, or a maximum velocity of 60 m/s after a stroke of 400 mm.

During the recent phase of the project “vessel failure” tests have been carried out on pipe sections under fast external cyclic bending. In connection to former examinations with slowly alternating bending loading now the resonance effect with high accelerations of masses together with energy dissipation due to material plastification was taken into account.

Experimentally determined failure curves for pipes weakened by surface longitudinal or circumferential defects, were compared with results calculated with the aid of engineering approximation methods. Considering the scatter bands of the mechanical properties and the geometrical dimensions, then by use of the engineering approximation methods, one can make only rough estimates of the load bearing behaviour.

The paper presents development of an expert system prototype applied for the leak-before-break analysis at MPA Stuttgart. The expert system prototype developed in the research allows successful introduction and use of standard and/or heuristic engineering knowledge (procedures, code calculations, etc.), of the preceding operational experience and of the testing results and experimental evidence in the leak-before-break analysis of future cases. The system has been developed, applied and verified on the basis of the results of over 50 large scale MPA-tests performed in the recent years.

For the testing of large-scale specimens, a 12 MN-High Loading Rate Tensile Testing Machine was designed and built at MPA Stuttgart. The aim was to determine the influence of high loading rates on the stress and strain behaviour of unwelded and welded components of ferritic and austenitic materials. This new generation of testing machines is driven by a propellant charge, and generates a maximum tensile force of 12 MN with a piston velocity of 25 m/s after a stroke of 20 mm, or a maximum velocity of 60 m/s after a stroke of 400 mm. In a first test programme, welded and unwelded wide plate specimens made of material X 6 CrNi 18 11 were tested at room temperature with different strain rates from 10−3 /s to 63/s. In addition to a description of the 12 MN-High Loading Rate Tensile Testing Machine, the results of the high loading rate tensile tests performed will be presented and compared with quasistatically tested wide plate specimens.

To investigate the strength behaviour, the influence of toughness to crack initiation and instability, the strain behaviour, the crack opening as well as the leak-before-break behaviour of artificially cracked piping, pipe bend tests were carried out. The dimensions of the pipes were 800 mm O.D. and 47 mm wall thickness, similar to the dimensions of the main cooling piping system of a 1300 MWe German PWR plant. Two types of nuclear grade materials similar to A 533 Cl 2 were used with high upper shelf impact energy of 50 and 150 J. The pipes were weakened by circumferential oriented surface notches respectively by through-wall flaws having a length of 60 degrees and various depths.

For the determination of the load bearing capacity, the deformation, leak-before break- and fracture behaviour, bending tests on vessels out of ferritic and austenitic materials containing a circumferential defect were performed. The vessels were loaded over a temperature range extending from 20°C to 580°C by internal pressure and additionally by an externally applied quasi-static or dynamic bending moment. For the test a 4-point bending test rig providing a bending moment up to 3 MNm was used. Parallel to the tests the failure curves as well as the leak-before-break curves were calculated with the aid of engineering type calculational methods. No significant difference as far as load bearing capacity is concerned, can be established between quasi-static and dynamic loading. In most cases fracture occurred after considerable plastic deformation in the test cross section. The engineering type methods applied to the theoretical calculation of the leak-before-break curve on the corresponding defect curve proved to be conservative with respect to safety.

For the determination of the strength-, deformation- and fracture behaviour of the material 17 MnMoV 6 4 (WB 35) which is used for piping components, tensile tests were carried out at different loading rates (monotonic and impact-type) on smooth and notched pipe strip specimens over a temperature range extending from − 30°C to 250°C. For the conduct of the tests a hydraulic high speed tensile machine having a free motion device was used; the velocity of impact was preset at ca. 7 m/s. With impact-type (dynamically) loaded specimens in general higher strength and deformation values were obtained than with monotonic (statically) loaded ones. In all of the specimens having low deformation values which were investigated microfractographically, ductile portions were found adjacent to the notch on the fracture surface.