1. Dominique Carrouge Houston February 2002 Phase Transformation Group H. K. D. H. Bhadeshia MCAS Technology Group Dr. P. Woollin

2. Microstructural Change in High Temperature Heat-Affected Zones of Welded

3. Why this project ? Super Martensitic SS are : Increasingly employed in the Oil & Gas industries Cost effective alternative to Duplex SS Engineered to be : High strength, Corrosion resistant, Weldable But : Work on understanding microstructure development in HAZ could improve service properties

4. Phases in HAZ Low carbon martensite (ok) Retained austenite (ok) Delta-ferrite (?) Can reduce toughness and corrosion resistance !

5. Aim Understand HAZ microstructure Locate delta-ferrite in HAZ of single-pass and multi-pass welds Investigate effect of some welding parameters Predict ferrite retention

6. HAZ & Phase diagram   Ae 5 Ae 4 Ae 3

7. Single-pass weld Plate : 5 mm thick Mechanised TIG process No filler material Heat-input : 0.7 kJ/mm Steel A (UNS S41426) : 0.01 C, 12 Cr, 6.5 Ni, 2.5Mo

8. Microstructure Electrolytic sulphuric acid Villela’s reagent

9. Coarse-Grained HAZ

10. Widmanstätten Austenite Ferrite (now martensite) Amount of ferrite negligible

11. Dual Phase HAZ ~ 20 % ferrite DPHAZ maximum length : 1 mm

12. X-ray Mapping of FerriteMo Cr Ni Partitioning during welding !

13. Less ferrite with high heat-input 0.7 kJ/mm0.3 kJ/mm Slow cooling rateFast cooling rate Steel A : 0.01 C, 12 Cr, 6.5 Ni, 2.5Mo

14. Situation in multi-pass weld Steel A : 0.01 C, 12 Cr, 6.5 Ni, 2.5Mo 12 mm thick pipe HI : 0.4 - 1.2 kJ/mm 25 Cr Duplex filler wire 7 passes TIG + MMA

15. Ferrite distribution Root Cap Fast cooling rate Slow cooling rate

16. Modelling ferrite content Physical model need reliable thermodynamic data MT DATA software has been used Prediction of equilibrium phase transformation temperatures and volume fractions A : 0.01 C, 12 Cr, 6.5 Ni, 2.5Mo B : 0.01 C, 11 Cr, 1.5 Ni, 0.5 Cu C : 0.01 C, 12 Cr, 3 Ni 1223 o C 1255 o C 1235 o C Ae 4

17. High temperature dilatometry Heating rate : 10 K/min Ae 4

18. Comparison Poor agreement Lack of data in the database Physical Modelling cannot be used

19. Neural network model Non linear regression tool that can be used to make extrapolation and predictions Not enough data to create a neural network model due to difficulty to measure ferrite content in Martensitic SS But a model has been developed for predicting ferrite retention in Austenitic stainless steels For a given composition the Ferrite Number can be estimated http://www.msm.cam.ac.uk/phasehttp://www.msm.cam.ac.uk/phase-trans/2001/ferrite.number.html/index.html

20. To avoid ferrite Minimum level of Nickel required to minimise ferrite :

21. To avoid ferrite Minimum level of Nickel required to minimise ferrite :

22. To avoid ferrite Minimum level of Nickel required to minimise ferrite :

23. Summary &Conclusion HT-HAZ microstructure has been studied Volume fraction of ferrite trapped in HAZ are small Mo containing alloy retain more ferrite Effects of this ferrite on properties are not well established But if ferrite content is to be minimised : Slower cooling rate would reduce ferrite content Alloy chemical composition could be adjusted Thank you for your attention