No abstract

Synopsis :A model to estimate future steel stock and demand was developed under the assumption that the steel stock per capita saturates following a sigmoid curve for Japan, China, South Korea and Taiwan. Per capita GDP was used as the variable. Different sigmoid curves were assumed for each end-use, i.e. buildings, infrastructure and automobiles. As in a previous study on automobiles ownership forecast, country-specific saturation values were introduced for steel use in buildings and infrastructure. As an indicator of regional difference, net population density was introduced. The relation between the saturation value and net population density was examined using the Japanese prefectures as samples, and the relation was applied to the East Asian countries. Future population and GDP were substituted into the sigmoid curves and the steel stock and demand was estimated up to 2050. As a result, steel demand for buildings and infrastructure was estimated to reach its peak around 2020 in China, the amount approximately being 330 million tons and 200 million tons respectively. Steel demand for automobiles in China was estimated to continuously increase until 2050 exceeding 100 million tons.

The present flows of steel scraps in Japan, China, South Korea and Taiwan are described, and a dynamic model that analyzes future scrap flows was developed. To estimate obsolete scrap recovery, a population balance model (PBM) was used for Japan, South Korea and Taiwan. The PBM dynamically estimated the amount of discarded steel by taking into account steel input into a society by end use and the lifetime distributions of each end use. For China, obsolete scrap recovery was estimated using a leaching model, which used the steel stock and the recovery ratio of obsolete scraps. Three different methods were applied to forecast future steel input for each country. The first method applied the assumption that steel demand in the future remains at the present level. The second method applied a logistic curve to estimate future steel stock. The third method applied regression equations to future steel input by end use. GDP and population were used as variables. Finally, the steel input forecasts produced by each method were substituted into the obsolete scrap recovery estimation model. The logistic curve method estimated that in 2030 obsolete scrap recovery would be 29 million tons in Japan, 83 million tons in China, 20 million tons in South Korea, and 3.7 million tons in Taiwan.KEY WORDS: China; Japan; South Korea; Steel scrap; Taiwan. Development of Material Flow Model Estimation of Steel InputThe residence time of steel in society and the recovery ratio of obsolete scraps differ among the end uses of steel. 2)Therefore, steel input in Japan before 2004 was estimated by each end use.Steel consumption was estimated by adding steel imports to and subtracting steel exports from steel production. Then, steel input into society was estimated by subtracting process yield scrap shipments and indirect exports from steel consumption and adding indirect imports. Indirect imports and exports are the steel imported and exported, respectively, by finished steel product industries.Steel input by end use in Japan from 1945 to 2004 was estimated in the following manner. Steel consumption by end use from 1971 to 2000 was obtained by conducting an interview survey.6) Process yield scrap shipments from 1980 to 2000 were obtained from literature. 7) Estimates of process yield scrap shipments from 1971 to 1979 assumed that the generation rates of process yield scraps for each end use were the same as those in 1980. Indirect imports and exports from 1971 to 2000 were also obtained from the interview survey. For China, the steel input by end use from 1949 to 2004 was estimated in the following manner. Steel production, imports and exports from 1949 to 2002 were based on a private communication.6) Steel consumption by end use from 1949 to 2002 was estimated by multiplying steel consumption by the rate of steel shipments according to end use, which was obtained from literature 10) for 1983 to 2000. The rates for the years that were not available in literature 10) were estimated using linear interpolation approximation. The rates for years pr...