Predicting scatter in the ductile to brittle transitional fracture in steels

Lead Research Organisation: University of Bristol
Department Name: Mechanical Engineering


Understanding the complex interaction of several competing temperature dependent physical processes resulting in the ductile to brittle transition, and of the inherent, temperature-dependent uncertainty in experimental data, is the major drive, and the major novelty, of this proposal. Inclusion of this complex interaction into a single fast and flexible engineering fracture prediction model is another novelty.The applicant has proposed a multi-scale model predicting fracture from microstructure and flow stress, i.e. not relying on experimental fracture data. In addition to predicting the full transition curve, the model was shown to predict temperature dependent scatter, fracture propagation path, crack arrest and fracture surface. The levels of scatter predicted with the model showed qualitative agreement with limited experimental Charpy data for a thermomechanically controlled rolled (TMCR) steel. This was achieved by taking the grain size distribution into account, and by introducing a temperature dependent grain mis-orientation threshold.In contrast to many complex microstructure deformation models, like e.g. crystal plasticity finite element model (CPFEM) the prototype multi-scale fracture model gives predictions of immediate relevance to industry. The proposed approach is a much more advanced that the current scatter predicting models, which are typically based on neural network or fuzzy logic, and therefore are completely divorced from the physics of fracture. Very few models predict the uncertainty of the transitional data.Despite considerable success, however, the prototype model is not yet suitable for quantitative fracture prediction. At present the crack propagation speed in the model is fixed to one CA cell per time increment. This restriction leads to slower fracture propagation, and ultimately to unrealistically high total energies at the lower shelf. However, the major obstacle remains lack of understanding of the temperature and strain rate sensitivity of the major micro-mechanical parameters in the transition area.This proposal would advance the engineering science because correct prediction of scatter is possible only if all the complexity of several interacting fracture processes at various temperatures and strain rates is understood and modelled. The fact that there is still no adequate physical explanation of high transitional scatter means that there are gaps in our understanding of fundamental physics of fracture. This project would fill these gaps on the way to developing a scatter prediction model.


10 25 50

publication icon
Mahalingam S (2012) The Role of Grain Size on Deformation of 316H Austenitic Stainless Steel in Key Engineering Materials

publication icon
Shterenlikht A (2015) Fortran 2008 coarrays in ACM SIGPLAN Fortran Forum

publication icon
Shterenlikht A (2012) Levenberg-Marquardt vs Powell's dogleg method for Gurson-Tvergaard-Needleman plasticity model in Computer Methods in Applied Mechanics and Engineering

Description 1. Experementally measured grain size distribution cannot be described by a single standard probability distribution function, even in simple cases, like normalised mild steel. Popular PDFs, such as Weibull's and gamma underpredict the tails. What is needed are combinations of several PDF, e.g. Weibull for the bulk of the data + Pareto for the tails

2. Twinning in BCC crystals, accompanying cleavage in TMCR steel during the lower shelf Charpy test has been quantified. The number of full and partial twins was measured along the crack propagation path and normal to it. However, the contribution of twinning to the total absorbed lower shelf energy is still not known.

3. A cellular automata grain evolution code has been developed. The model correctly predicts the onset of abnormal grain growth.
Exploitation Route steelmakers will be interested in these results, because it directly influences their understanding of the probabililty of a catastrophic cleavage failures. Based on the data collected, two journal publications are planned.
Sectors Aerospace, Defence and Marine,Construction,Energy,Manufacturing, including Industrial Biotechology

Description The project led to a number of open source research software codes:, >1000 downloads to date + interface to powerful FE library ParaFEM:, >10 downloads/week.
First Year Of Impact 2010
Sector Electronics,Energy,Environment
Impact Types Economic

Description An Experimental and Modelling Approach to Engineering the Stability of Mixed Micro- and Nano-Grain Size Polycrystals to Improve Durability
Amount £442,297 (GBP)
Funding ID EP/H006729/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 01/2010 
End 10/2013
Title Multi-scale modelling framework for supercomputers 
Description We have combined MPI, OpenMP and Fortran coarrays to make a generic multi-scale modelling framework, where computational kernels can be easily swapped, e.g. Ising magnetisation, microstructure solidification, grain boundary migration, fracture, etc. This framwork can be used to benchmark modern HPC architecture and software stack. 
Type Of Material Computer model/algorithm 
Year Produced 2013 
Provided To Others? Yes  
Impact A large number of dowloads, new collaborations, research input from PRACE (POP), Software sustainability institute, ARCHER eCSE, Invitation to contribute to a panel event at SC17. 
Description multi-scale modelling collaboration with University fo Manchester 
Organisation University of Manchester
Department School of Physics and Astronomy Manchester
Country United Kingdom 
Sector Academic/University 
PI Contribution Manchester have developed a highly scalable MPI FE solver. We coupled this with Bristol cellular automata fracture model to deliver a flexible multi-scale modelling framework for HPC systems.
Collaborator Contribution Manchester's code is open source. However, it would not have been possible for us to use it efficiently without their expertise, advice and hands-on collaboration
Impact For full details see 2016: A Shterenlikht, L Margetts, L Cebamanos, JD Arregui-Mena, Multi-scale CAFE framework for simulating fracture in heterogeneous materials implemented in Fortran coarrays and MPI, PGAS Application Workshop (PAW), held in conjunction with SC16, MON 14-NOV-2016, Salt Lake City, UT, USA, PDF paper, PDF talk. 2016: A progress report on using TAU and CrayPAT for profiling CGPACK and hybrid CGPACK+ParaFEM (coarrays + MPI) programs. 2016: A progress report on using TAU for profiling CGPACK and hybrid CGPACK+ParaFEM (coarrays + MPI) programs with the Intel 16 compiler and Intel MPI library. 2016: L Cebamanos, A Shterenlikht, D Arregui, L Margetts, Scaling hybrid coarray/MPI miniapps on Archer, Cray User Group (CUG2016), London, MAY 2016, paper and talk. 2015: Coarrays from laptops to supercomputers, a talk by Harvey Richardson, with several CGPACK results, from OCT-2015 joint meeting of BCS specialist Fortran group and IoP Computational Physics Group. 2015: Lee Margetts' talk on ParaFEM given at University of Bristol in 2015, with several CGPACK+ParaFEM examples, 2015: A. Shterenlikht, L. Margetts, S. A. McDonald, N. K. Bourne, Towards mechanism-based simulation of impact damage using Exascale computing, in Shock Compression of Condensed Matter 2015, R. Chau, T.C. Germann, and T.D. Sewell, Editors. 2017, AIP: Melville, New York. p. 080009. DOI: 1 page PDF flyer, PDF talk. 2015: PDF paper, A. Shterenlikht, L. Margetts, Three-dimensional cellular automata modelling of cleavage propagation across crystal boundaries in polycrystalline microstructures, Proc. Roy. Soc. A 471:20150039, DOI: 10.1098/rspa.2015.0039. 2015: Invited talk at the 1st ParaFEM workshop, 13-APR-2015, Oxford, UK, for printer or for screen. 2015: PDF paper, A. Shterenlikht, L. Margetts, L. Cebamanos, D. Henty (2015) Fortran 2008 coarrays, ACM SIGPLAN Fortran Forum 34:10-30, DOI: 10.1145/2754942.2754944. 2015: L. Margetts, A. Shterenlikht, Proc. of NAFEMS seminar: CAE based fatigue, UPM, Madrid, 2015, PDF talk.
Start Year 2013
Title abumpack 
Description A library with material constitutive model routines in modern Fortran and Fortran 77, including multi-scale CAFE modelling. 
Type Of Technology Software 
Year Produced 2012 
Open Source License? Yes  
Impact New collaboration with the University of Trondheim, Prof. Zhilian Zhang. Several papers, e.g. %A A. Shterenlikht %A N. A. Alexander %T Levenberg-Marquardt vs Powell's dogleg method for Gurson-Tvergaard-Needleman plasticity model %J Computer Methods in Applied Mechanics and Engineering %V 237-240 %P 1-9 %D 2012 
Company Name CMPLX Engineering Ltd 
Description An engineering consultancy focussing on Fortran, HPC and solid mechanics. We offer services locally, nationally and globally, including 1-2-1 support, long term software development and optimisation projects and training. 
Year Established 2016 
Impact Assistance to NTNU, Trondheim, Norway with ABUMPACK software - new capability enabled and PhD student trained
Description 2016 Hartree Centre summer school on engineering simulation 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact About 30 people have attended this week-long school, a mix of PhD students, RAs and non-academic professionals. The mode was a mix of lectures and practicals, with full participation of all course leaders and attendees. This lead to discussions and questions every day. There was a social event, which helped to bring the course tutors and attendees closer together. The number of downloads of our software increased after the course.
Year(s) Of Engagement Activity 2016