Comparison of Tramp Element Contents of Steel Bars from Japan and China

Abstract

In this study, a random sample of 107 steel bars from Japan and 26 steel bars from China were studied. Each specimen’s elemental composition of tramp elements, such as Cu, Cr, Ni, and Sn, was analyzed. By using the compositions of specific tramp elements, electric arc furnace steel bars were distinguished. The distributions of the tramp element composition were obtained for the Japanese steel bars and represent a larger number of samples than previous studies. Those for the Chinese samples are the first published data. The compositions of the four elements in the Japanese bars are statistically significantly higher than those in the Chinese bars. Owing to the large gap between the Cu contents of the Japanese and Chinese samples, and the statistically significantly differences between the ratios of Cu to both Ni and Cr in the Japanese and Chinese samples, it is highly likely that copper-based materials are separated from steel scrap at a higher rate in China. The relationship between the Cu and Cr compositions suggested that the mixing of copper-based materials and special steels, which contribute to Cr contamination, differs between Japan and China. The distributions of the Cr composition for each country had larger standard deviations than the distributions of other elements. The ratio of the Cr and Ni compositions in the Chinese samples is less than 2.25, resulting in 18% Cr and 8% Ni in stainless steel. It was found that Ni from materials other than stainless steel was likely to have been included.

1. Introduction

Tramp elements in steel materials are elements that cannot be removed easily by current smelting processes and can cause problems as impurities during working processes. In particular, Cu and Sn generate surface hot shortness and adversely affect the formability as their content levels increase.¹⁾ Cu contamination is believed to result from the unintentional introduction of copper materials into steel scrap when they are not separated during recovery of steel from end-of-life products.²⁾ Daigo et al.²⁾ estimated how Cu was mixed into steel scrap as of 2000 and approximated the change in Cu content in steel if the mixing rates are constant, suggesting that incrassation is possible. Although Ni and Cr are also present in carbon steel as impurities,^3,4) their presence is attributed to the mixing of unseparated special steels into carbon steel scrap when scrap is recovered from end-of-life products.⁵⁾ Oda et al.⁵⁾ focused on Cr and mentioned the possibility of its incrassation. Thus, impurities found in recycled carbon steels consist of two types of elements: those that were mixed into the steel cycle unintentionally as other materials than steel, and those that were intentionally added as alloy elements but became impurities because they were not separated by alloy type during recycling. Similar problems occur for materials other than steel. The former type occurs when steel screws are mixed into aluminum scraps, whereas the latter appears when several alloy types of aluminum are recovered in the same bucket.^6,7,8,9) Even if a sustainable society develops and recycling is promoted, there are concerns that the quality requirements for the impurity content of steel may no longer be met as long as a recycling system that allows the introduction of impurities is sustained.²⁾ Therefore, the current challenge is to elucidate the mechanism by which impurities are introduced and implement measures to reduce the possibility.

End-of-life products are disassembled and separated by type of material. The liberated materials are then recovered as material scrap. This disassembly and separation process can be roughly classified into two processing methods: manual dismantling and sorting, and shredding and mechanical sorting. The former incurs higher labor costs than the latter and is limited in its processing capacity.¹⁰⁾ These processing methods are believed to be selected on the basis of business profitability factors, such as labor costs, revenue of recovered material, and the necessary processing volumes. It is known that the processing method changes from manual dismantling to shredding in a country as it develops economically.¹¹⁾ Manual dismantling prevents the introduction of impurities more effectively because it is possible to separate each material type. Under manual sorting, materials are identified using human visual perception and product design information; such as, color, shape, and usage. On the other hand, under mechanical sorting, if foreign materials are caught in materials deformed by a shredder, it is difficult to separate these mixed materials from one another. Moreover, because mechanical sorting by magnetic separation and eddy current separation differentiate materials using their magnetic properties and electrical conductivity, it has a low separation efficiency for materials with similar characteristics or fails to separate these materials, allowing the introduction of other materials or the mixture of different alloy types. With regard to how tramp elements could be introduced, an individual case could be observed by applying time and effort. However, it is difficult to observe the every site scattered across Japan on which impurities could be introduced.

Although it is possible to assume factors which may affect this process as above, there have been few studies that analyze the impurity element content of steel products which is the result of introducing these elements. If tramp elements are mixed into steel because of the separation efficiency, the tramp element contents of steel produced in China are expected to be lower than those of materials produced in Japan because China uses primarily manual dismantling, and Japan practices mainly shredding and mechanical sorting.^12,13) This study aims to comprehend the process by which tramp elements could be introduced by comparing the contents of these elements in steel produced in Japan and China.

2. Research Methods

2.1. Previous Analyses of Tramp Element Contents of Steel Materials

We summarize former studies on the tramp element content of steel in Table 1 and provide concise reviews below. Nafziger et al.⁴⁾ mentioned the change in the average contents of Cu, Cr, Ni, and Sn in common steel for which steel scrap was the raw material in 1954, 1962, and 1977 using the results of analyses conducted by the American Iron and Steel Institute in the U.S. With the exception of Cr, the contents of the other three elements declined from those in 1954 and 1977. We postulated that, with increased interest in controlling the tramp element content in steelmaking processes, this decline is due to the use of scrap with low impurities, which diluted the tramp elements, and the implementation of more efficient scrap processing; such as separation and sorting. Although a major investigation was conducted in 1981 targeting cast iron to examine the impurity content, it is not included in Table 1 because the cast iron differs from steel analyzed in this study. In Japan, the Non-Integrated Steel Producers’ Association has been investigating the tramp element content of EAF (electric arc furnace) reinforcing bars every few years since 1987.³⁾ The results show no significant changes in the Mn, S, Si, Sn, and P contents in a long-term period from 1987 to 2007. On the other hand, the Cr contents have exhibited an increasing trend, rising from approximately 0.14% in 1987 to approximately 0.19% in 2007. Moreover, a comparison of impurity contents by region in Japan revealed that the contents vary because different types of scrap are available in these regions. Toi et al. analyzed the Cu, Sn, and Cr contents of steel scrap obtained from 1996 to 1997 and examined not only the average contents of these elements but also their distributions.¹⁴⁾ In addition to the studies described above, a few other published studies have analyzed the impurity contents of steel products and scrap, although these do not mention the sampling methods or ensure representativeness.¹⁵⁾

Table 1. Tramp element contents of steel reported in previous studies.

		Cr	Cu	Ni	Sn
1954	U.S.	0.52*1	0.17	0.107	0.035	Cold scrap shop only*3	[4]
1962	U.S.	0.71	0.151	0.085	0.017	Cold scrap shop only*3	[4]
1977	U.S.	0.115	0.162	0.086	0.011	Cold scrap shop only*3	[4]
1987	Japan	0.14	0.28	0.08	0.03	EAF steel bar*4	[3]
1989	Japan	0.159	0.327	–	0.024	EAF steel bar*4	[3]
1991	Japan	0.169	0.325	–	0.024	EAF steel bar*4	[3]
1996*2	Japan	0.158	0.297	–	0.031	EAF steel scrap (n=50)	[9]
1997	Japan	0.179	0.289	–	0.018	EAF steel bar*4	[3]
2001	Japan	0.174	0.288	–	0.019	EAF steel bar*4	[3]
2004	Japan	0.192	0.318	–	0.021	EAF steel bar*4	[3]
2007	Japan	0.185	0.281	–	0.020	EAF steel bar*4	[3]

*¹  The figure with one decimal place is illegible because of the age of the study. The number could be 0.32.

*²  Samples include new scrap and old scrap that might have been produced earlier.

*³  Survey procedures are unknown.

*⁴  Average of average contents of 32 EAF steels during the year obtained by questionnaire.

The introduction of impurities is believed to vary depending on how end-of-life products are disassembled and how steel scrap is processed. However, there have been few available studies that have observed and compared the impurity contents of steel in regions where different disassembly methods and processing are implemented.

2.2. Obtaining Steel Materials and Conducting Elemental Analyses of Impurity Contents

In this study, we gathered Japanese and Chinese EAF steel and conducted elemental analyses to obtain the Cu, Cr, Ni, and Sn contents. EAF steel bars were used for this study because they have greater allowable concentrations of impurities^2,15) and can consist of steel scrap as raw materials, such as heavy melt scrap, that could contain relatively large amounts of impurities. Japanese samples were collected in the Chubu region of Japan in 2011.¹⁶⁾ One hundred seven steel bars were randomly sampled from a scrap yard that recovered from various locations. An Olympus portable X-ray fluorescence analyzer was employed to conduct elemental analyses after coatings and oxides were removed from the scrap surfaces. Quantitative results for the elemental contents were obtained by applying the fundamental parameter method¹⁷⁾ with the device. Chinese samples were collected from 2012 to 2013 in Shanghai, Jiangsu Province, and Zhejiang Province. Twenty-six steel bars were obtained from scrap yards and construction sites. Emission spectrochemical analysis was used to conduct the elemental analyses. Although processed scrap and obsolete scrap were mixed in the specimens obtained in both Japan and China, they were not completely distinguishable. As a result, it was not feasible to specify the times at which the obtained steel bars were produced. In the U.S., a decrease in the impurity content was observed when awareness of the impurity element content rose in 60’s and 70’s. In Japan, no significant changes in elemental content, except for that of Cr, were found over time after 80’s. Therefore, we treated the specimens without identifying the production date.

2.3. Comparative Analyses of Impurity Contents of Japanese and Chinese Steel Bars

Because it is not clear what steelmaking processes were employed and what types of raw materials were used for the obtained steel bars, they were classified into natural-resource-based BOF (basic oxygen furnace) steels and steel-scrap-based EAF (electric arc furnace) steels according to the elemental analyses. Toi et al.¹⁴⁾ focused on the Cu contents of steel and classified them as BOF steels with Cu contents of 0.05% or lower and EAF steels with Cu contents of 0.11% or higher. Additionally, another study¹⁵⁾ mentioned setting the Cu content of BOF steel at 0.03% or lower. On the basis of previous studies and analysis results, the obtained steel bars were classified as BOF steel bars and EAF steel bars. For the Japanese specimens, the impurity contents in the steel bars categorized as EAF steels were compared with previous studies. This confirmed that the obtained specimens were representative.

Not all the impurity elements in EAF steel bars originate from contaminants; some of them were alloyed into the steel scrap used as a raw material. Therefore, the presence of impurity elements cannot necessarily be attributed to the introduction of other materials containing them or special steels. However, we compared the mechanism by which these elements could be introduced to Japanese and Chinese steel bar specimens in terms of the content distribution of each element and the relationship with other elemental contents.

3. Results and Discussion

Histograms of the Cu, Cr, Ni, and Sn contents with a 0.01% interval on the x-axis based on elemental analyses of the steel bar specimens obtained in Japan and China were shown in Figs. 1(a)–1(d). To allow comparison of the results for the Japanese and Chinese specimens, the frequency is calculated with respect to all of the specimens from each country. Although two Japanese steel bar specimens contained 1.12% and 1.16% Cr, whereas a Cr content of 0.53% was found in one Chinese steel bar specimen, these specimens are not displayed in Fig. 1(b) for better visibility. The results for the Cu, Cr, Ni, and Sn contents of the Japanese specimens contain errors because they were defined by the fundamental parameter method.¹⁷⁾ The errors for the Cu, Cr, Ni, and Sn contents were 0.006%, 0.004%, 0.010%, and 0.006%, respectively, in the average. The errors for the contents of these elements were 0.010%, 0.008%, 0.016%, and 0.009%, respectively, in the maximum cases.

Fig. 1.

Histograms of (a) Cu, (b) Cr, (c) Ni, and (d) Sn contents of steel bar specimens from Japan and China.

Because information on the raw materials and steelmaking furnaces for each specimen was not available, we selected primarily scrap-based EAF steel bars from the sampled steel bars using the analyzed impurity contents. It seems very unlikely that impurities were intentionally added to the steel bars, and the specimens were classified according to the contents of Cu and Ni, which cannot be removed once they enter from steel. We set a Cu content of 0.05% and a Ni content of 0.02% as the thresholds of which we considered that the specimen is likely a EAF steel bar if it exceeds either. As the scatter plot of the Ni and Cu contents in Fig. 2 shows, some Japanese and Chinese specimens had Cu contents of greater than 0.05% and Ni contents of 0.02% or less. On the other hand, none of these samples had Cu contents of 0.05% or less and Ni contents greater than 0.02%. Both the Cu and Ni contents in eight Japanese specimens and ten Chinese specimens fell below the thresholds. We determined that the raw materials of these specimens were composed primarily of natural resources, such as pig iron and direct reduced iron. Many more of the specimens from China were classified as BOF steel bars than the Japanese specimens; this is attributed to the fact that the percentage of steel bars produced by BOFs is higher for Chinese steel bars. In addition, the ratio of steel bars produced by EAFs to those produced by BOFs is about 9 to 1 in Japan.¹⁸⁾ This figure is close to the ratio of the classified specimens, indicating that the thresholds set above are appropriate.

Fig. 2.

Cu content versus Ni content of steel bar specimens from Japan and China.

On the basis of the distribution of the tramp element contents of the specimens classified as EAF steel bars, Table 2 shows the average, standard deviation, and median of the impurity contents for Japan and China. For comparison, Table 2 also shows the average impurity contents of reinforcing steel bars investigated by the Non-Integrated Steel Producers’ Association in 2007³⁾ and the standard deviation for the contents of each element studied by Toi et al. in 1997.¹⁴⁾ Each of the elemental contents of the steel bars obtained in Japan was almost equivalent to the average values from the investigation conducted in 2007. Regarding regional characteristics within Japan, steel bars produced in the Chubu region, where the samples for this study were collected, are considered to have lower impurity contents than those produced in other regions.⁵⁾ Although the impurity contents observed in this study may have been lower than the national average reported in the former study due to the regional difference, they are believed to be representative of those of steel bars in Japan. In this study, it was possible to obtain the distributions of the impurity contents for more EAF steel bar samples than in any previous study and obtain information that is representative of steel bars in Japan. The standard deviations were compared with the figures reported previously.¹⁴⁾ The standard deviations for Cu and Sn were found to be equivalent to the former study. The standard deviation of Cr was larger than the former study. One of the reasons for this is that specimens with no Cr content were removed from the statistical population in the previous study. Even considering this fact, the standard deviation in the report is 0.097, which is still smaller than the standard deviation observed in this study. This is attributed to increased fluctuations in the Cr content over time with the 15-year gap in the timing of assessments. No reference materials are available for the Chinese specimens. Therefore, we hope that the values mentioned in this study will serve as a benchmark for reference and comparison in future studies.

Table 2. Average, standard deviation, median, and maximum of observed distributions of impurity contents of steel bars by country.

		Cu	Cr	Ni	Sn
Japan	Average	0.278	0.175	0.072	0.017
	Standard deviation	0.081	0.120	0.039	0.023
	Median	0.279	0.159	0.068	0.000
	Maximum	0.565	1.122	0.379	0.102
China	Average	0.098	0.093	0.048	0.012
	Standard deviation	0.034	0.126	0.020	0.004
	Median	0.09	0.04	0.05	0.01
	Maximum	0.16	0.53	0.10	0.02
Average during 2007 in Japan [3]		0.281	0.185	0.08*	0.020
Standard deviation in Japan [9]		0.081	0.075	–	0.02

*  Measured in 1987

A comparison of the impurity contents of the Japanese and Chinese specimens revealed that the steel bars obtained in China had lower contents of all four elements than those obtained in Japan. The differences were considered significant at a significance level of 1% for the average contents of Cu and Ni, and at a significance level of 5% for the average Cr and Sn contents. The maximum values of the impurity contents of the Chinese specimens were lower than those of the Japanese specimens. We thought one of reasons is that in China, there are less chance for other materials than carbon steel to be mixed into steel scrap. The lower usage ratios of obsolete scrap in China are also believed to explain that trend. Although it is not possible to specify the type of raw material used in steelmaking when the obtained specimens were produced, steel scrap accounts for 98% in Japan and approximately 50% in China of the average raw material for EAF steel.¹⁹⁾ In addition, steel scrap used as a raw material also includes processing scrap, i.e. factory-generated new scrap that contains no other materials and in-house scrap, i.e. steelmaking-factory-generated new scrap. Accordingly, in-house scrap and processing scrap of BOF steel rarely contain impurities, so they serve as a dilution source, as in pig iron and direct reduced iron. Thus, data on the ratio of obsolete scrap in steel scrap are required in order to precisely quantify the relationship between the raw material and the elemental content in steel products. Further studies are required for quantitatively discussing the impurities that are mixed into obsolete scrap. Our comparison of the average contents of the four impurities in the Japanese and Chinese steel bars showed that the Cu contents differed significantly between the two countries compared with the other elements. Although this could be due to the low Cu content as an alloying element in steel, it appears likely that in China, copper materials are more frequently liberated without allowing them to be mixed into steel scrap. As mentioned in the introduction, the reason for this is believed to be that manual dismantling is performed in China, as opposed to the mechanical separation primarily used in Japan. Copper-precipitation-hardened steel was also considered as one of possible ways of copper mixing. Although any statistics on it could not be found, it was found by that the annual production of the steel is negligible small.^20,21,22) In addition, the average Sn content is higher in the Japanese steel bars, but the difference between the two countries was smaller than that for the other impurities. The median for Sn is larger for the Chinese steel bars; because the median for Sn in the Japanese steel bars is 0.00, this is attributed to the greater availability of Sn-free steel in Japan that prevents the entry of Sn into the system of cyclic use of steel.

In order to quantify the difference in the possibility for each tramp element to be mixed into steel scrap during recycling processes in Japan and China, we set the content of one tramp element as a reference and analyzed the difference between the Japanese and Chinese specimens in terms of the volume of another tramp element relative to that of the reference element. The ratios for a set of paired tramp element contents were calculated for each obtained specimen. The distribution of the ratios in the Japanese and Chinese steel bars was compared in Table 3. Because many of the specimens had no Sn content, the Sn content was used as a numerator only when calculating ratios involving the Sn content. The ratios of the Cu and Cr contents showed no statistical difference between the distributions in the Japanese and Chinese steel bar specimens. On the other hand, the average Cu content was considerably higher (1.59 times) than the average Cr content for the Japanese specimens, whereas it was almost equivalent to the Cr content (only 1.05 times) for the Chinese specimens. This suggests that the Cu and Cr contents were distributed almost independently of each other. The reason is thought to be that the Cr originated from ores, and the routes of entry for special steel—a source that introduced the Cr originating from scrap—differ from those for Cu. The ratios of Cu to Ni and of Cu to Sn were both significantly higher in the Japanese steel bar specimens (at a significance level of 1%). On the basis of these results, the introduction of Cu is assumed to be significantly higher for the Japanese steel bars if the introduction of Ni and Sn are assumed to be at an equivalent level for the Japanese and Chinese steel bars. As the examination of the Cu contents revealed, copper materials are more frequently liberated in China instead of being mixed into steel scrap. Significant differences were found (at a significance level of 10%) for the ratio of Cr to Ni. These differences were significant at a higher significance level because they are both alloy elements found in stainless steel, and their content ratios in stainless steel do not differ greatly by country. Whereas the Cr-to-Ni ratio is 2.25 in the most major type of stainless steel, with a Cr content of 18% and a Ni content of 8%, the average ratio for the Chinese steel bars was 1.72, indicating a higher Ni content. This suggests the possibility of other sources of Ni in China in addition to stainless steel.

Table 3. Comparison of content ratios for each pair of elements in Japanese and Chinese samples.

	Average in Japan	Average in China	Result of t-test	Significance level for t-test
Cu/Cr	2.12	2.08	No significant difference	25%
Cu/Ni	4.10	2.35	Significant difference	1%
Sn/Cu	0.057	0.130	Significant difference	1%
Cr/Ni	2.71	1.72	Significant difference	10%
Sn/Cr	0.133	0.244	Significant difference	5%
Sn/Ni	0.252	0.289	No significant difference	25%

4. Conclusion

In this study, 107 Japanese steel bar specimens and 26 Chinese steel bar specimens were obtained, and elemental analyses of the Cu, Cr, Ni, and Sn contents were conducted. According to standards for the impurity contents obtained by comparison with the results of previous studies, 99 of the 107 Japanese specimens and 16 of the 26 Chinese specimens were classified as EAF steel bars. This study obtained the distribution of the impurity contents for the largest sample of EAF steel bars to date as information representative of Japanese steel bars. Because no reference materials are available for the Chinese specimens, we hope that the values shown in this study will serve as a benchmark for reference and comparison in future studies.

For all four of the evaluated elements, the EAF steel bars obtained in China had significantly lower tramp element contents than those obtained in Japan. This study showed a considerable disparity in the Cu contents and a statistical difference in the ratio of Cu to the other elements between the Japanese and Chinese specimens. This is attributed to the fact that copper materials are more frequently liberated in China without being mixed into steel scrap. The relationship between the Cu and Cr contents suggested the possibility that special steels and copper materials are introduced by different processes. In addition, this study revealed that the standard deviation of the Cr contents was higher than that for the other elements. The ratio of Cr to Ni in the Chinese specimens indicated that the other sources of Ni than stainless steel are mixed into steel scrap.

Acknowledgments

Part of this study was made possible by assistance from JSPS Grant-in-Aid for Scientific Research (22686084 and 15H02860).

References

• 1)   M.  Hatano,  K.  Kunishige and  Y.  Komizo: Tetsu-to-Hagané, 88 (2002), 142.
• 2)   I.  Daigo,  D.  Fujimaki,  Y.  Matsuno and  Y.  Adachi: Tetsu-to-Hagané, 91 (2005), 171.
• 3)  Non-Integrated Steel Producers’ Association: Electric Furnace Reinforcing Steel Bar Quality Inspection Report, Non-Integrated Steel Producers’ Association, Tokyo, (2008), 139.
• 4)   R. H.  Nafziger,  A. D.  Hartman and  R. F.  Farrell: Trends in Iron Casting Compositions as Related to Ferrous Scrap Quality and Other Variables, 1981-86, Bureau of Mines, United States Department of the Interior, Washington DC, (1990), 70.
• 5)   T.  Oda,  I.  Daigo,  Y.  Matsuno and  Y.  Adachi: Tetsu-to-Hagané, 95 (2009), 720.
• 6)   H.  Hatayama,  H.  Yamada,  I.  Daigo,  Y.  Matsuno and  Y.  Adachi: Mater. Trans., 48 (2007), 2518.
• 7)   H.  Ohno,  K.  Matsubae,  K.  Nakajima,  S.  Nakamura and  T.  Nagasaka: J. Ind. Ecol., 18 (2014), 242.
• 8)   I.  Daigo,  Y.  Matsuno and  Y.  Adachi: Resour. Conserv. Recycl., 54 (2010), 851.
• 9)   Y.  Igarashi,  I.  Daigo,  Y.  Matsuno and  Y.  Adachi: ISIJ Int., 47 (2007), 758.
• 10)   I.  Daigo,  M.  Sasaki,  K.  Fujisaki,  Y.  Matsuno and  Y.  Adachi: J. Jpn. Soc. Mater. Cycles Waste Manage., 21 (2010), 178.
• 11)   Y.  Takahashi,  I.  Daigo,  Y.  Matsuno and  Y.  Adachi: CAMP-ISIJ, 25 (2012), 91, CD-ROM.
• 12)   F.  Wang,  R.  Kuehr,  D.  Ahlquist and  J.  Li: E-waste in China: A country report. United Nations University Institute for Sustainability and Peace, Bonn, Germany, (2013), 59.
• 13)   J.  Li,  K.  Yu and  P.  Gao: J. Mater. Cycles Waste Manag., 16 (2014), 31.
• 14)   A.  Toi,  J.  Sato and  K.  Kanero: Tetsu-to-Hagané, 83 (1997), 850.
• 15)  The Iron and Steel Institute of Japan: Effects of Cu and Other Tramp Elements on Steel Properties, ISIJ, Tokyo, (1997), 13.
• 16)   N.  Fujitsuka,  I.  Daigo,  Y.  Matsuno and  Y.  Adachi: CAMP-ISIJ, 25 (2012), 88, CD-ROM.
• 17)   K.  Sugiyama,  J. W.  Essel,  T.  Kitamura and  Y.  Waseda: Bull. Inst. Adv. Mater. Process., Tohoku Univ., 48 (1993), 140.
• 18)  Committee on Iron and Steel Statistics, Japan Iron and Steel Federation (JISF): Handbook for Iron and Steel Statistics 2010, JISF, Tokyo, (2010), 306.
• 19)  The Japan Ferrous Raw Materials Association: Quarterly Ferrous Raw Material Statistics. Vol. 48, Spring, Tokyo, (2011), 63.
• 20)   S.  Tanaka and  T.  Nakamura: Kobe Steel Eng. Rep., 63 (2013), No. 1, 2.
• 21)  KOBELCO: Gijutugaido, http://www.gijutsugaido.jp/catalog/word/html/0163.html, (accessed 2015-4-9).
• 22)   Y.  Okamura,  T.  Kasuya,  R.  Yamaba,  M.  Tanaka and  H.  Tamehiro: Steel Const. Eng., 1 (1994), 53.