The Material, Manufacturing and Computer Simulation of the Quenching Process of a Japanese Blade 


Tatsuo Inoue




Traditional methods of preparing a kind of steel called tamahagane used for the Japanese sword by tatara system and procedure of making the sword are briefly introduced with the discussions from the viewpoint of metallurgy and thermo-mechanical processing. Such traditional methods are also revealed to be consistent with the modern science and technology. The quenching process applied to the final stage of the procedure is focused to explain how the pattern of blade, the deformation and residual stresses are induced by the computer simulation based on the theory of metallo-thermo-mechanics relevant to the coupled fields among temperature, micro structural change and stress/strain.


1. Introduction


The Japanese sword originally used as fighting weapon is now one of typical traditional crafts with artistic characteristics, and so many monographs have been published in English [1-6] as well as in Japanese. The sword is also interesting from the viewpoint of modern science and technology [7-26] since the way of manufacturing the sword is really consistent with the science as is same as other surviving traditional products. Tawara, a professor of Japanese Sword Research Laboratory, the University of Tokyo, accomplished a monumental work in the framework of metallurgy[10]. Tawara measured the distribution of carbon density, precipitation and hardness in the cross section of the swords in relation with the pattern of blade and sori representing the mode of deformation during quenching. Successive works were made by Bain [7], Suzuki[11-12], Tsuwa[13], and others.


Very few works on the sword were made, however, from mechanical engineering aspect. Ishikawa[14-16] discussed the mechanism of cutting objects from theory of cutting and the shape of the sword from dynamics, and stress/deformation analysis after quenching by the finite element method was carried out by Fujiwara and Hanabusa[17-18] and the present authors[19-26].


As is well known, the Japanese sword is normally made of a traditional Japanese steel made of iron sand, called tamahagane[27-31], and manufactured by a special way, especially by folding the steel.In the first and second parts of the paper, the process of preparing the steel and the way of making the sword are briefly introduced.


One of the most attractive and important stage of manufacturing the sword applied to the end of the process before grinding and polishing is quenching, which induces the characteristic deformation pattern of bending called sori, and the formation of blade. The following parts treat some results of computer simulation of interesting bent shape of the sword and pattern of the blade simulated by a developed code 'HEARTS' [32-35].


The code was accomplished based on the theory of metallo-thermo-mechanics[36-39] relevant to describing the fundamental equations considering the coupling effect among microstructural change, temperature and stress/strain, which have been applied to the simulation of heat treatment processes considering phase transformation including quenching of the sword[19-26]. After discussing an paradoxical characteristics on the heat transfer coefficient between heated steel covered by a kind of thermally insulated clay, called yakibatsuchi, and water as the coolant, some results of simulation of a sword in the quenching process are presented.




Almost all Japanese swords with some exceptions are made of tamahagane steel, or noble steel, specially prepared by the tatara system by use of iron sand, but not by normal ore as seen in the old painting.


Fig.1 Old painting of tatara system.

Steels distributed in Japan before Meiji renovation in 1868 were produced by this method, while modern system of iron and steel making had been developed in Europe. The amount of yearly production in the time was approximately 10,000 tons being equivalent to that of Great Britain40

The steel was used not only for swords, but also for guns, cutting tools, nails for construction of old temples and shrines, and other products necessary for ordinary life.Around the period, the tatara system was replaced by the modern western system except for providing the steel to sword smiths.  

The Iron and Steel Institute of Japan constructed an experimental system of Tatara in 27-28in Sugaya, Shimane Prefecture, and accumulated interesting data of steel making technology. 

Due to the lack of the steel for the sword, the Japanese Sword Museum, Nippon Bijutsu Token Hozon Kyokai, started to organize the tatara system in Torigami, Shimane Prefecture, under the cooperation with Hitachi Metals, Ltd. in 1977, and provides the steel of 3-4 tons every year.


Iron sand with 2-5\% content of iron mined from Chugoku Mountains, which includes the best quality of iron sand in Japan, concentrated to the degree of 60\% by magnets system, while the mineral dressing method by gravity classification in flowing river had been adopted, which is no more popular to prevent water pollution problem. Such enriched iron sand (masa satetsu) contains 8\% of pure iron Fe and iron oxide Fe2O3 with very small amount of impurity such as 0.026% phosphorus P and 0.002% sulfur S being injurious for carbon steels. Chemical compositions are shown in Table 1. Here, alumina Al2O3 is so rare to be beneficial for low temperature refinement to be stated later.

Table 1 An example of chemical composition of iron sand in virgin and enriched states. 


The enriched iron sand is supplied alternatively to the furnace with charcoal by hands. Figure 2 illustrates the cross sectional view of the furnace under operation with some drainage mechanism constructed to three meters under the ground. Only a difference of the system from the classical one in Fig.1 is that electric motors are used for intermittent air blowing instead of manpowered bellows.

Fig.2 Cross sectional view of tatara furnace


Continuous burning is operated for 70h under the direction of a murage, or chief foreman. The temperature in the furnace is around 1200-1500 deg C lower than the melting point of the steel, which follows that the reduction process of the partly molten state is occurred between iron oxide Fe2O3 and silica SiO2 contained in the clay of furnace. During the process, the initial thickness of 200-400mm of the furnace is reduced to 50-100mm. After taking out the slag from the bottom of the furnace followed by destroying the furnace, a block of blister steel called kera in sponge state with dimension of 2.7m in length, 1m in width and 200-300mm in thickness and with 2-2.5ton containing steel of 1.5-1.8ton is obtained (see Fig.3), while necessary amount of iron sand and charcoal are respectively 8 and 13tons. (It is amazing that the block costs hundred thousand dollars, two hundred times much expensive of normal steel!)

Fig.3 Kera, a block of blister steel.

Steel produced on the both side of the block, where the enough deoxidization is completed by air supplied from kirokan (special wooden pipes) is called tamahagane, or noble steel, which is spelled as mother of metal in Japanese character. Other parts of the block with different chemical composition in Table 2 are also used for the sword making.

Table 2. Chemical composition of tamahagane, forged and core steels.


The chemical compositions of the best part of steel are 1.0-1.4% C, 0.02-0.03% P, 0.006% S, and 0.003-0.004% Ti, being very rare of sulfur and phosphorus even compared with industrial carbon steel (see Table 2).


The steel is cooled by cold environment since the operation is carried out in mid-winter followed by shattering, and distributed to about 300 professional and registered sword smiths in Japan.





The pieces of the steel with different carbon contents are heated in the carburizing or decarburizing environment, termed jigane-oroshi. This process is made in the furnace burnt by charcoal and ash of rice straws with the blowing air sent by fuigo (blowers). Decarbonization occurs in the part closed to the blower, while CO2 gas accelerates the sintering on the upper parts.


The successive process of making a sword is illustrated in Fig.4. The smith makes a flat plate with a handle termed as tekoita, on which the small pieces of broken flat pieces are piled up covered by a special Japanese paper dampened by water containing clay and rice straws to prevent oxidation on the surface of steel by insulating air. It is known that SiO2 in the clay contributes to increase the impurities including in the slag.

Fig.4 Process of manufacturing the Japanese sword.


Forging process is followed to obtain a block, where about ten to fifteen rounds folding called orikaeshi are repeated to get laminated materials with approximately 1,000 (=2**10) to 30,000 (=2**15) layers. The characteristic pattern of the laminated layers depending on the way of smiths is visible on the surface of the sword, some of which are depicted in Fig.5.

Fig.5 Laminated layers by orikaeshi forging.

Such bonding of each layers during orikaesi process is enhanced by the mechanism of so called mechanical alloying, for which so clean surface of the layers are necessary. This is achieved by dispersing impurities such as oxides and so on with sparks by hammering. The weight of the block decreases during the process to approximately 700-100g in the final shape of the sword.


A bar of shingane (core steel) with low carbon content is wrapped by kawagane or hagane (skin steel) with high carbon for which the tamahagane steel is normally used (see the cross sectional views in Fig.4). This process is called tsukurikomi. After rough grinding by the smith himself, the sword is transferred to the final process of yakiire (quenching), which is the main topics of numerical simulation in the following sections.


Before quenching, a kind of clay, yakiba-tsuchi, mixed by charcoal powder and so on is pasted on the surface of the blade to control the heat transfer intensity to be discussed in Sec.6 as presented in Fig.6.

Fig.6 Tsuchioki, pasting a kind of mixed clay on the blade.


Most interesting situation is that the pasted clay is thick on the ridge while thin on the blade part as illustrated in Fig.7 . Finally the quenching operation of the sword heated up to 800-850 degC into water is carried out. (The temperature of heated sword and cooling water depends on the school of smiths and the material property as well as the dimension of the sword.)

Fig.7 Pasted pattern of the thickness of yakiba-tsuchi


During the quenching process, a white hard part with martensite structure is induced near the blade, while other shining part remains pearlite and ferrite structures. The border of the parts is called hamon as seen in Fig.8.

Fig.8 Hamon, shape of border between quenched and unquenched parts.


Here, wavy or zigzag pattern of the hamon is realized by cutting the clay by a spatulas. A computer simulation how the hamon appears and how the stresses are induced will be treated in the following sections.




In such cases of quenching of the Japanese sword, and other machine parts in general, incorporated with phase transformation, fields of metallic structure, temperature and stress/deformation are coupled each other as schematically illustrated in the diagram of Fig.9 [36-39].

Fig.9 Coupling effect among metallic structures, temperature and stress/deformation.


Each field is to be described by the coupled fundamental equations of kinetics of phase transformation, heat conduction equation and constitutive equation combined with kinematic relation and equation of motion, which are summarized in separate page (see separate page of






Brief introduction of the developed CAE system 'HEARTS' is presented in this section to be used for the simulation of the quenching process of the Japanese sword.


+++++ 5.1. Finite Element Scheme and Method of Numerical Calculation ++++


Finite element scheme is applied to the fundamental equations developed above, and a new version 2.0 of 'HEARTS' [35] approximately with 35,000 steps consisted of 250 subroutines in several levels is coded by FORTRAN 77. For three dimensional problem as well as two dimensional and axisymmetrical problems (plane stress and strain problems including that of generalized plane strain for stress analysis), which were available in the version 1.0. The 2-D and 3-D isoparametric elements with variable-number-nodes are selected from an element library.


A skyline scheme and modified or full Newton-Raphson method are employed to solve these nonlinear equations in each time step. In order to treat unsteady heat conduction equation depending on time, a numerical time integration scheme 'step-by-step time integration method' is introduced, while an incremental method is used for deformation and stress analysis.


++++++++ 5.2. Architecture of 'HEARTS' ++++++++


The heat treatment simulation code ''HEARTS'' is utilized in the CAE circumstance as illustrated in Fig.10 \ref{Architecture}, being combined with the solver, and pre/post processor such as PATRAN, I-DEAS, or other popularly used processors, and the interface. The data necessary for the simulation is generated by the pre-processor, is output in the form of intermediate file. The data in the file is transferred into the data file for control and initial-boundary conditions as well as the file for the element and node data, while the material data file is constructed separately.

Fig. 10 System architecture of CAE system 'HEARTS'


The solver of ''HEARTS'' is divided mainly into four parts corresponding to the equations, and they can be connected by the user's requirement what kind of solutions, coupled or uncoupled, to be solved. The output of the numerical results calculated by the solver are transferred into the files for post-treatment, list image and final results. The data for post-treatment is again stored in the intermediate file through the interface to convert into the final data for post-processor, and several kinds of illustration are available by the user's requirement.




Before quenching the sword into water, the yakiba-tsuchi clay is pasted on the surface as shown in Fig.6 to control the cooling condition of the surface of the steel. Since the temperature distribution is to be calculated in the body of the sword, it is necessary to identify the relative heat transfer coefficient on the metal surface as the function of the thickness of the clay.


Series of experiments based on Japan Industrial Standard, JIS-K2242, were made to measure the cooling curve of a cylinder made of silver coated by the clay with different thickness. The reason of the usage of a silver is that the material is not undergone any phase transformation during the heating and cooling process. A thermocouple is mounted on the surface as shown in Fig.11. The cylinder is heated up to 800 degC by a reflection type electric furnace, and cooled in the water.

Fig. 11 A silver rod mounting a thermocouple


Obtained cooling curves are demonstrated in Fig.12 as the parameter of thickness of the pasted clay[43]. It is so interesting that the curves for thick clay (t=0.7-0.8 and 0.75-0.9mm) show typical mode with moderate cooling rate due to film boiling followed by severe cooling stage by nuclear boiling, the shape of which are similar to the case without the clay. When the thickness is small (t=0.1-0.15 and 0.2-0.3mm), on the other hand, no film boiling stage is observed, which means that the cylinder is cooled severely from the beginning. This is also confirmed by the observation of bubble nucleation by video camera.

Fig. 12 Cooling curves on the surface of a silver rod depending on the thickness of pasted clay.

Inverse calculation is carried out by perturbation method to identify the heat transfer coefficient on the surface of the cylinder. Results are represented in Fig.13. It is paradoxes to be noted from this figure that the coefficient in the case with thin clay gives higher value than without clay during 800-400 degC being most important temperature range for quenching. This data will be employed as the boundary condition when solving the coupled heat conduction equation.

Fig. 13 Temperature dependent heat transfer coefficient.




7.1. A Sword Treated and the Condition of Simulation


The shape and dimension of the sword treated here is illustrated in Fig.14, which is a model of a classical and famous sword termed Bizen-Osafune. 

Fig. 14. Shape and dimension of a sword treated.


Three dimensional finite element mesh division of the sword is represented in Fig.15, where the division is made for a half part in the width direction due to symmetry. Figures 15(a) and (b) respectively denote the whole region and the enlarged part near kissaki(tip).

(a)Global view.

(b)Near the tip.

Fig. 15. Finit element mesh.


Total number of the elements is 828, and that of the nodes is 1230. This model is supposed to consist of two regions, (see Fig.16(a)), core steel with 0.2\% carbon content and skin steel with 0.65%C to which different material data are applied. To differentiate the relative heat transfer coefficient depending on the thickness of the yakibatsuchi clay, the surface of the sword is divided into two parts shown in Fig.16(b) with different value indicated in Fig.13.

Fig. 16 Division of the sword for two materials with different carbon content (a) and for two kinds of surface area with different heat transfer coefficient (b).



The sword is uniformly heated up to 850 degC, at which temperature the whole region is changed into austenitic structure, and the sword is quenched into the water of 40 degC.


7.2. Effect of Pasted Clay on the Formation of Quenched and Unquenched Border


To know the effect of the thickness of clay on the induced hamon (border between quenched and unquenched regions), simulation of quenching under several different conditions were carried out. Red parts of Fig.17 show the volume fractions of martensite after quenching for different way of pasted clay. When the sword is quenched by pasting thick clay of 0.8mm, martensite hardly appear except for the part near the blade (see Fig.17(a)), which follows that very thin hamon occurs. However, almost whole region become martensite as seen in Fig.(b) when thin clay with 0.1mm thickness is pasted on the whole surface. If the clay is pasted thin on the blade side, and thick on other part, on the other hand, ideal distribution of martensite is obtained by the simulation with hardened blade by martensite and with ductile main body by pearlite as is so realistic as the normal sword. Hereafter, the simulation below is made with the pasted yakibatsuchi clay of the final pattern.

Fig. 17 Martensite fraction corresponding to hamon, depending on the way of thickness of 

pasted clay.



7.3. Variation of Temperature, Metallic Structures, and Associated Deformation

Figure 18 shows the temperature distribution of the sword with successive time from the beginning of the quenching, and the mode of deformation is also depicted in the figure. The part of blade near the edge with thin thickness shrinks due to thermal contraction by severe cooing, which leads to the bending to the downward termed as gyaku-sori or reverse bending at t=1s as is shown in Fig.(b).

Fig.18 Successive deformation associated with temperature distribution.


When martensitic transformation starts to occur in that part, however, normal bending called sori to the upper direction is observed due to the volumetric dilatation by martensite formation (see Fig.(c)). Gyaku-sori again appears at t=3-4s, because of the pearlitic transformation in the part of ridge. In the successive stage of cooling, hot ridge side shrinks gradually because of thermal contraction, and finally, the normal bending can be obtained.


Thus simulated deformation gives the good agreement with the actual bending mode of sori. Such mode of successive deformation due to martensitic and pearlitic transformation is shown in Fig.19.

Fig. 19 Successive development of structures



7.4. Stress Distribution and Residual Stresses


Stress distribution in the longitudinal direction in the course of quenching is represented in Fig.20

Fig. 20 Longitudinal stress distribution and residual stresses.


The simulated residual stresses after complete cooling are compared with measured data by X-ray diffraction technique on the lines along Hasaki (edge), Shinogi (side ridge) and Mune (ridge)(see Fig.14) as shown in Fig.21.

Fig. 21 Comparison of calculated residual stresses with experimental data.


It is also noted that the maximum stress near the top of the sword during quenching reaches the fracture stress, which sometimes leads to cracking or breakage of the sword during the operation.



Procedure of preparing the traditional Japanese steel, tamahagane, followed by the method of making the Japanese sword is summarized in the first part of the paper from the scientific point of view. Theory of metallo-thermo-mechanics relevant to the simulation of quenching processes and the brief introduction of the finite element computer code 'HEARTS' are also stated.


As an example of the application of the simulation of quenching processes, a Japanese sword is focussed, and the change in temperature, metallic structure and stress/deformation are calculated. The results reveal to represent such real situations. The discussion from the viewpoints of metallurgy and mechanics are carried out in each section of preparing Japanese steel and manufacturing the sword, especially on the effect of pasted clay.


In conclusion, the technology surviving for over thousand years is really consistent with the modern science and technology.





The author wish to express his hearty acknowledgement to Prof. K. Ishikawa, Kanazawa Institute of Technology, Mr. J. Nozaki, Metal Museum, Mr. T. Suzuki, Nippon Bijutsu Token Hozon Kyokai, for their providing instructive information on the science of Japanese sword. Cooperation to develop the CAE system ''HEARTS'' and identify the heat transfer coefficient are made respectively by Mr. K. Arimoto, CRC Research Institute (now moved to SFTC Co.) and Mr. H. Kanamori and co-workers, Idemitsu Kosan Co., respectively. The numerical calculations by use of the system are carried out by Mr. T. Uehara and Mr. H. Ikuta, graduate students of Kyoto University.



*********** References ***********


1. L. Kapp, H. Kapp and Y. Yoshimura, The Craft of the Japanese Sword, Kodansha International, Tokyo (1987).

2. S. Kanzan (Translated by J. Earle), The Japanese Sword --- A Comprehensive Guide, Kodansha International, Tokyo (1983).

3. J. M. Yumoto, The Samurai Sword --- A Handbook, Charles E. Tuttle Co., Tokyo (1991).

4. N. Ogasawara (Translated by D. Kenny), Japanese Swords, Hoikusha, Tokyo (1991).

5. Museum of Fine Arts, Boston (ed.), Japanese Master Swordsmith ---The Gassan Tradition, Museum of Fine Arts, Boston, (1989).

6. J. Homma, Japanese Sword --- The National Museum Series, Kogeisha, Tokyo (1949).

7. B. C. Bain,Trans. JISI,26 (1978) 265.

8. H. Tanimura,J. Metals,32 (1980) 63.

9. T. Kato, Bulletin of the Metals Museum, {\bf 26} (1996) 60.

10. K. Tawara, Scientific Research on Japanese Swords (in Japanese), Hitachi-hyoronsha, Tokyo (1953).

11. T. Suzuki, Science on Tatara and Japanese Sword (in Japanese), Yuzankaku, Tokyo (1993).

12. T. Suzuki, Traditional Technology of Making Japanese Swords (in Japanese), Rikougaku-sha, Tokyo (1994).

13. H. Tsuwa, Proc. 3rd Int. Conf. Production Eng. (1979), 365.

14. K. Ishikawa, The Science of Japanese sword (in Japanese), Kanazawa Institute of Technology, Kanazawa (1989).

15. K. Ishikawa, The Science and culture of Japanese sword (in Japanese), Bulletin of Culture Lecture Series, Atsuta Shrine, No.10 (1997).

16. K. Ishikawa, Nippon-Toh (Japanese Sword) --- High-Technology Threading Through History, Lecture Note, Rochester Institute of Technology, Rochester (1993).

17. H. Fujiwara and T. Hanabusa,Research Report, Faculty of Eng., Tokushima Univ.

(in Japanese), {\bf 38} (1993) 1.

18. H. Fujiwara and T. Hanabusa, Proc. 3rd Int. Conf. Residual Stresses,2 (1991) 1537.

19. T. Inoue, Proc. Int. Seminar on Microstructures and Mechanical Properties of New Engineering Materials (1993) 515.

20. T. Inoue, T. Ohmori, T. Hanabusa and H. Fujiwara, Proc. 4th Int. Conf. Residual Stresses (1994) 970.

21. T. Uehara and T. Inoue, J. Soc. Mater. Sci., Japan (in Japanese),44 (1995) 309.

22. T. Inoue and T. Uehara, Proc. Int. Symp. Phase Transformations during the Thermal/Mechanical Processing of Steel (1995) 521.

23. T. Inoue, T. Uehara, H. Ikuta and I. Miyata., Proc. Int. Conf. Materials and Mechanics-97 (1997) 137.

24. T. Inoue, Trans. Japan Soc. Mech. Eng. (in Japanese ),97 (1994) 132.

25. T. Inoue, Boundary (in Japanese),11 (1995) 36.

26. T. Inoue, Materia (in Japanese),135 (1996) 81.

27. Committee on Restoration Project of Tatara Steelmaking System, Restoration of Tatara Steelmaking Process and the Blister Steel (in Japanese), ISIJ-Special Report No. 9 (1970).

28. Y. Matsushita, Proc. Int. Conf. Science and Technology of Iron and Steel, Trans. ISIJ,11 (1971) 212.

29. J. Kozuka, Trans. ISIJ,8 (1968) 36.

30. K. Kubota, Coll. Int. Inst. CNRS, No.538 (1970) 577.

31. K. Horikawa, Bull. Metals Museum,12 (1982) 29.

32. T. Inoue, D. Y. Ju and K. Arimoto, Proc. 1st Int. Conf. Quenching and Control of Distortion (1992) 206.

33. T. Inoue, K. Arimoto and D. Y. Ju, Proc. 3rd Int. Conf. Residual Stresses (1991) 226.

34. T. Inoue, K. Arimoto and D. Y. Ju, Proc. 8th Int. Cong. Heat Treatment of Materials (1992) 569.

35. T. Inoue and K. Arimoto, J. Mater. Eng. and Perfor., ASM,6 (1997) 51.

36. T. Inoue and B. Raniecki, J. Mech. Phys. Solids,26 (1978) 187.

37. T. Inoue, S. Nagaki, T. Kishino and M. Monkawa, Ing.-Archiv.,50(1981) 315.

38. T. Inoue and Z. G. Wang, Mat. Sci. Tech.,1(1985) 845.

39. T. Inoue, Berg-und Huttenmannische Monatshefte,132(1987) 63.

40. T. Kuroiwa, Distance between Tatara and Kouro (in Japanese), Agne, Tokyo (1993).

41. W. A. Johnson and R. F. Mehl, Trans. AIME,135(1939) 416.

42. C. L. Magee, Nucleation of Martensite, ASM, New York (1968).

43. H. Kanamori, E. Nakamura, S. Koyama and T. Inoue,J. Japan Soc. Heat Treatment,36(1997