PROCESSING FACTORS CONTRIBUTING TO GROWTH AND DECLINE IN THE
STEEL INDUSTRY
By
Michele Dufalla
SUBMITTED TO
THE DEPARTMENT OF MATERIALS SCIENCE AND ENGINEERING
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
BACHELOR OF SCIENCE IN MATERIALS SCIENCE AND ENGINEERING
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
JUNE 2007
© 2007 Michele Dufalla.
All rights reserved.
The authorhereby grants to MITpermission to reproduce
and to distributepublicly paper and electronic copies of
this thesis document in whole or in part in any medium now
known or hereafter created.
Signature of Author:
Departmdt of Materials Science and Engineering
11 May 2007
Certified by:
Tho&4s W. Eagar
Professor of Materials Engineering and Engineering Systems
Thesis Supervisor
Accepted by:
MASSACHL'SETTS INSTrIUTE
OF TECHNOLOGY
JUL i7 2008
LIBRARIES
-Caroline A. Ross
Professor of Materials Science and Engineering
Chair, Departmental Undergraduate Committee
Table of Contents
Abstract
3
Introduction
4
Figure 1
4
Figure2
5
Figure3
5
Furnace Analysis
6
Furnace Introduction
6
Furnace Data
7
Figure4
7
Figure5
8
Furnace Statistics
9
Figure6
10
Table 1
12
Table 2
12
Table 3
12
Casting Analysis
13
Casting Introduction
13
Casting Data
13
Figure 7
14
Casting Statistics
14
Table 4
15
Table 5
15
Conclusion
15
Works Cited
16
Acknowledgements
16
Appendices: Data Used
17
ABSTRACT
During the second half of the twentieth century, a technological shift occurred in
the steel industry. A different mix of refining and melting furnaces were used, with
increasing use being made of basic oxygen and electric arc furnaces as compared to the
basic open hearth. Additionally, continuous casting began to replace ingot casting. Iron
ore price, scrap steel price and electricity price were examined as predictor variables for
these technological shifts. For the furnace shift, iron ore price and scrap steel price
seemed to play a role, though much smaller than the role of time. For the casting shift,
only time seemed to be correlated.
INTRODUCTION
The steel industry underwent much change and upheaval over the second half of the
twentieth century. In the U.S., it has shifted from a dominant to less prominent industrial
position. For an industry overview, Figure 1 shows steel production in the U.S., while Figure
2 shows steel consumption in the U.S. Production peaks during the 1970s, only to fall during
the recession of the 1980s. By 2000, production had returned to approximately 1950s levels.
On the other hand, U.S. consumption follows the shape of the production curve with the
major distinction of a steady decline throughout the 1970s and 1980s and a large increase in
demand during the 1990s. It is important to consider, however, the fact that as casting
technology became more efficient, less raw steel was needed.1 Also to note, until around
1995 U.S. consumption was less than U.S. production. After this point, consumption jumped
to above production levels. This incidence underscores the growing role of imports in U.S.
steel consumption. Figure 3 shows the U.S. price of steel. It increases sharply throughout the
1970s, but decreases beyond the 1950s baseline before 2000.
160,000,000
C
120,000,000
(0 100,000,000
v
L)
2
Q.
80,000,000
4
~~~
140,000,000
Aa
d~
4
120,00,000
:
80,00,00 - 4Aj~
*
60,00,000~
*-·
~
**~
k4
4
-
60,000,000
A 40,000,000
*~~},
1940 1960
1950
0
1940
1950
1960
1970
1970
Year
Figure 1: Steel Production in the U.S. 2
1Written communication with Professor Eagar
from: Kelly, Thomas D. and Matos, Grecia
2Data
4
R.
,A~A
4
1980
1990
2000
201
1980
1990
2000
2010
i
· _ ___ ___
14U,1U,)UUU
Mr
I
A
S100,000,000
try \F~
~'AM
MyFy
4I0 40,000,000 0.000,0o
1940
1950
1960
1970
1980
~
V
~tr-
tryi~i~
1990
2000
2010
Year
Figure 2: Steel Consumption in the U.S.3
IlnM M
IL8/.1U -
~
*
~
i
1000.00
a
a
800.00
600.00
W
-
trA
&
~
A A
Vt~~>~t
"'
I--
A
400.00
200.000.00
1940
~t
f;}P2~pt~2
N~P
t*.r
1950
§ 7 tftr~~
p
F~F~yrNtr~
r~tr~
trt~rrrtr
1960
1970
trM~~V
1980
FF~
1990
F.r
*
2000
2010
Year
Figure 3: Steel Prices in the U.S. 4
During this time, the technology of steelmaking experienced a shift. The
distribution of refining furnaces changed." Additionally, a shift from ingot to continuous
casting also occurred.6 The type of furnace used impacts the ratio of scrap steel to iron
ore that can be used.7 Continuous casting creates less waste metal while using fewer
3 Data from: Kelly, Thomas D. and Matos, Grecia R.
4Data from: Kelly, Thomas D. and Matos, Grecia R.
5"50 Years of the ECSC Treaty: Coal and Steel Statistics."
6 "50 Years of the ECSC Treaty: Coal and Steel Statistics."
7"Electric Arc Furnace: Process Description.", "Basic Oxygen Furnace: Process Description."
processing steps.8 It is conceivable that changing scrap metal, iron ore and electricity
prices would have an effect on the rate that the industry used to shift to different
technologies.
FURNACE ANALYSIS
Introduction to Furnaces
Three types of furnaces used for steel production were examined. Electric furnaces
have the advantage of being able to accommodate a charge of as much as 100% scrap steel,
as well as the disadvantages associated with uncontrolled additive element management from
use of scrap. 9 Oxygen furnaces only accommodate charges of up to 30% scrap, but produce a
purer, higher quality product. 10 The category of "other furnaces" includes Thomas, Bessemer
1
and Siemens-Martin as well as other earlier methods."
The newer oxygen method started to be used commercially in the 1950s. 12 As can be
seen in the chart, by the mid-1980s this new method and electric furnaces had completely
overtaken the old. The superior efficiency of the new methods explains this changeover.
However, this efficiency of the new methods is related to lower energy usage and percentage
of scrap metal used. As such, it could be interpreted that the rate of change would be related
to changes in energy prices, as well as the changes in the prices of scrap steel and iron ore.
FurnaceData
8(("Continuous Casting: Process Description."
9 "Electric Arc Furnace: Process Description."
10 "Basic Oxygen Furnace: Process Description."
1""50 Years of the ECSC Treaty: Coal and Steel Statistics."
12 "50 Years of the ECSC Treaty: Coal
and Steel Statistics."
To examine this question, historic data from the European Coal and Steel Community
was used to determine the relative usage of each of the three furnace types. Data from the
United States Geological Survey was used to determine the constant dollar prices of scrap
steel and iron ore (Figure 4). Finally, data from the United States Energy Information
Administration was used to determine chained dollar' 3 electricity prices (Figure 5). The
period from 1952 to 2000 was studied.
US Scrap and Iron Ore Prices
US Scrap Steel Price
US Iron Ore Price Index
-
300.0 -
250.00-
1
Ta
200.00-
O 150o.ooo
50.00-
0.00
I
I
I
I
1952
1957
192
1967
~
I
I
1973
1977
1981
1996
l
l
1990
1995
i
199
2000
Year
Figure 4: US scrap and iron ore prices in constant 1998 US dollars' 4
'3 Chained dollars are an alternative to constant dollars designed to more accurately reflect the real value of
a dollar over time, developed by the EIA; from: "Chained Dollars."
14 Data from: "Iron and Steel Scrap Statistics." And "Iron Ore Statistics."
5
a)
U
o
Cl
a.,
o
aV)
_
o
in
9
4o
..-
1
2 '~7
1982
W7
.79.
1077
"S
HI
1I
'
1
1M
2ý
:¢
Year
Figure 5: US electricity prices in chained 2000 dollars' s
Several flaws in these data exist. Firstly, the data from the European Coal and
Steel Community (ECSC) was taken every five years. Often between periods, new
countries would join the community resulting in an increasing number of members as
time progresses. To partially alleviate this complication, each furnace type as a
percentage of all steel produced was used, rather than absolute values. Next, scrap steel,
iron ore and electricity prices were not available for Europe over the time period needed,
so those of the United States were used. The assumption is made that while the data will
not be exactly in alignment, the economies of the United States and Europe were
sufficiently open in the time period studied that the prices will be indicative of trends
experienced by both regions.
15 Data
from: "Annual Energy Review."
Finally, due to varying compositions of iron ore and differing processing
methods, a direct conversion factor between iron ore and scrap steel could not be found.
That is, iron ore mined from some regions may have a much higher or lower recoverable
iron content than others.'i s Because of this variability, quantitative determination of the
actual cost of ore for each ton of steel produced is not possible. However, it is assumed
that price trends are more important than actual values, and so the data was left in its
original per 1000 ton state.
FurnaceStatistics
A stepwise linear regression was used for each furnace type against the year,
electricity price, scrap steel price and iron ore price. Please see Figure 6 for furnace
information. For the electric furnace, the best model included the year and scrap steel
price as predictor variables, with an R2 value of 0.992, adjusted to 0.990, and an ANOVA
significance of 0.000. Year contributes a standardized beta coefficient of 0.916. US scrap
steel price contributes a standardized beta coefficient of -0.125. For the oxygen furnace,
the best model included the year and iron ore price as predictor variables, with an R2
value of 0.836, adjusted to 0.789, and an ANOVA significance of 0.002. Year contributes
a standardized beta coefficient of 1.145. US iron ore price contributes a standardized beta
coefficient of 0.767. For the other furnaces, the best model included the year and iron ore
price as predictor variables, with an R value of 0.931, adjusted to 0.911, and an ANOVA
significance of 0.000. Year contributes a standardized beta coefficient of -1.196. US iron
ore price contributes a standardized beta coefficient of -0.567.
15
Kuck, Peter.
9
Percent of Crude Steel Production by Furnace
U Electric Furnace
U Oxygen Furnace
] other Furnace
OAM_-0.8000
0.6000-
o.0oo 00000-
i..,,.-
,-
--
i
r---T-
192 1957 1962 1907 1973
I
I
1977 1981 1M8 1900 199M
1998 2000
Year
Figure 6: Steel production by furnace type (percentage of ECSC) 16
Each model depended heavily, if not exclusively, on the mere passage of time. To
assess the effect of the focus contributors, the analysis was repeated to find the individual
R2 values of the predictor variables. For the electric furnace, considering only the year
results in an R2 of 0.982, adjusted to 0.980, with ANOVA significance of 0.000. Year
alone contributes a standardized beta coefficient of 0.991. Considering only scrap steel
price results in an R2 of 0.608, adjusted to 0.568, with ANOVA significance of 0.003 and
a standardized beta coefficient of -0.779. For the oxygen furnace, considering only the
year results in an R2 of 0.463, adjusted to 0.396, with ANOVA significance of 0.030.
16 Data from: "50 Years of the ECSC Treaty: Coal and Steel Statistics."
Year alone contributes a standardized beta coefficient of 0.681. Considering only iron ore
price results in an R2 of 0.011, adjusted to -0.088, with ANOVA significance of 0.747
and a standardized beta coefficient of 0.104. For the other furnaces, considering only the
year results in an R of 0.727, adjusted to 0.693, with ANOVA significance of 0.002.
Year alone contributes a standardized beta coefficient of -0.853. Considering only iron
ore price results in an R2 of 0.003, adjusted to -0.097, with ANOVA significance of 0.874
and a standardized beta coefficient of 0.051.
Due to the extreme activity of the 1970s, the analysis was repeated, using only the
data from 1967 to 1981. For the electric furnace, the best model included the iron ore
price as a predictor variable, with an R2 value of 0.931, adjusted to 0.897, and an
ANOVA significance of 0.035. US iron ore price contributes a standardized beta
coefficient of 0.965. For the oxygen furnace, the best model included only the year as a
predictor variable, with an R2 value of 0.934, adjusted to 0.901, and an ANOVA
significance of 0.034. Year contributes a standardized beta coefficient of 0.966. For the
other furnaces, the best model included only the year as a predictor variable, with an R2
value of 0.966, adjusted to 0.950, and an ANOVA significance of 0.017. Year contributes
a standardized beta coefficient of -0.983.
All of these complete models have ANOVA significance of less than 0.050,
suggesting validity in a difference of means not due to chance. 17 In the individual
component models, however, significance increases dramatically. However, the
predictive value of scrap metal prices and iron ore prices is debatable. While the two
indicators seem to play a part in several of the models, the dependence of the iron ore and
17 "One Way ANOVA."
scrap steel markets on the production of crude steel could be in reverse. The close ties
and dependence between the iron ore and scrap steel markets and the steel industry
presents a problem in determining which variables are independent and which are
dependent. Electricity price can safely be assumed to be independent, due to the many
other industries that purchase electricity. Additionally, additional independent acting
variables may act as controllers but remain unidentified. While coefficients are given,
their usefulness is limited due to the relative price and production, rather than absolute
price and production, approach. The furnace results are summarized in Tables 1, 2, and 3.
Dependent: Electric Furnace
Variable
RW
Year, Scrap Steel (1952-
Adjusted R2
Coefficient
ANOVA sig.
0.992 0.990
0.916, -0.125
0.000
0.982 0.980
0.608 0.568
0.931 0.897
0.990
-0.779
0.965
0.003
0.003
0.035
Coefficient
ANOVA sig.
2000)
Year (1952-2000)
Scrap Steel (1952-2000)
Iron Ore (1967-1981)
Table 1: Electric furnace model statistics
Dependent: Oxygen Furnace
Variable
R_
Year, Iron Ore (1952-2000)
Year (1952-2000)
Iron Ore (1952-2000)
Year (1967-1981)
Table 2: Oxygen furnace model
0.836 0.789
0.463 0.396
0.011 -0.088
0.934 0.901
statistics
1.145, 0.767
0.681
0.104
0.966
0.002
0.030
0.747
0.034
Dependent: Other Furnaces
Variable
R
Adjusted R
Coefficient
ANOVA sig.
Year, Iron Ore (1952-2000)
Year (1952-2000)
Iron Ore (1952-2000)
Year (1967-1981)
0.931
0.727
0.003
0.966
0.911
0.693
-0.097
0.950
-1.196, -0.567
-0.853
0.051
-0.983
0.000
0.002
0.874
0.017
Table 3: Other furnace model statistics
Adjusted RL
CASTING ANALYSIS
CastingIntroduction
In addition to the changes brought about by the shift in furnace type, casting
methods were also in flux during the tumultuous 1970s. Whereas ingot casting had been
the standard, continuous casting, which required fewer steps and resulted in less waste,
quickly became dominant. 1 While this inherent superiority explains the eventual
overtaking, the question of any side effects of iron ore, scrap metal and electricity prices
remains open.
CastingData
A similar approach to the furnace quandary is taken to this problem. Data from
the European Coal and Steel Community is again used (Figure 7). However, individual
countries' data were available for casting results so the original difficulty of a continually
changing sample population is no longer an issue. Germany was used as the country for
study, as it was the most industrially active of the initial members of the ECSC for this
time period. Due to the changing overall production level, however, percentages of toatal
production, ratherthan absolute values, are used. The same iron ore, scrap steel and
electricity data is used from the first problem. The same assumptions and restrictions
apply. Additionally, only the period from 1967 to 1981 was examined.
m"Continuous Casting: Process Description."
Percent of Steel Production by Processing
U Ingots
U Continuous Casting
1.000 -
F Liquid Steel
0.8000
-
0.6000 -
0.4000
-
0.2000 -
0.0000 -
7
L.
L.
I
I
I
I
0
1995 198200I
1952 1957 1982 1987 1973 1977 1981 1988 1990
Year
Figure 7: Percentage of total steel production by process (Germany) 19
CastingStatistics
A stepwise linear regression was used for each casting method against the year,
electricity price, scrap steel price and iron ore price. For ingot casting, the best model
included only the year as a predictor variable, with an R2 value of 0.978, adjusted to
0.968, and an ANOVA significance of 0.011. Year contributes a standardized beta
coefficient of -0.989. For continuous casting, the best model again included only the year
as a predictor variable, with an R2 value of 0.979, adjusted to 0.969, and an ANOVA
significance of 0.010. Year contributes a standardized beta coefficient of 0.990. These
19 Data from: "50 Years of the ECSC Treaty: Coal and Steel Statistics."
statistical results are summarized in Tables 4 and 5. Again, the coefficients are useful
only as a relative indicator, as absolute values were not used in the analysis.
Dependent: Ingot Casting
Variable
R2
Year (1952-2000)
0.978 0.968
Adjusted R
I Coefficient
ANOVA sig.
-0.989
0.011
Coefficient
0.990
ANOVA sig.
0.010
Table 4: Ingot casting model statistics
Dependent: Continuous Casting
Variable
Year (1952-2000)
R
I Adjusted R'
0.979 0.969
Table 5: Continuous casting statistics
While these two models are again valid due to the low significance values, energy
and material prices do not seem to have promoted or hindered the natural progression of
change between the two casting methods. The above restrictions regarding the
questionable independence of the predictor variables still applies, but in this case the
point appears to be moot.
CONCLUSION
The shift in furnaces is tied to changes in scrap steel and iron ore price. However,
due to the close ties between these industries, causality cannot be determined. This effect
is greatly overshadowed, however, by the passage of time, suggesting that other forces
were at work.
The shift from ingot to continuous casting was not tied to scrap steel, iron ore or
electricity prices. The passage of time did greatly influence this shift, which suggests that
the inherent improved efficiency of continuous casting may be an explanation.
Additionally, another unconsidered factor could be the driving force behind the
technological shift.
ACKNOWLEDGEMENTS
A large debt is owed to Professor Thomas Eagar, for the premise of this paper, as
well as informative discussion and assistance with statistics and editing.
WORKS CITED
"50 Years of the ECSC Treaty: Coal and Steel Statistics." Luxembourg: Office for
Official Publications of the European Communities, 2002. ISBN 92-894-3715-4.
European Communities, 2002. (Accessed through
http://epp.eurostat.ec.europa.eu/portal/page?_pageid=1073,46587259&_dad=portal&_sch
ema=PORTAL&p_product_code=KS-43-02-979 (9 May 2007).
"Annual Energy Review." EIA. Table 8.10.
http://www.eia.doe.gov/emeu/aer/pdf/pages/sec8_39.pdf (9 May 2007).
"Basic Oxygen Furnace: Process Description." METALS Advisor.
http://www.energysolutionscenter.org/HeatTreat/MetalsAdvisor/iron and_steel/processdescriptions/raw_metalspreparation/steelmaking/basicoxygenfurnace/basicoyxgenf
urnace_processdescription.htm (9 May 2007).
"Chained Dollars." EIA.
http://www.eia.doe.gov/emeu/consumptionbriefs/recs/heatingoil/chained_oil.html (9
May 2007).
"Continuous Casting: Process Description." METALS Advisor.
http://www.energysolutionscenter.org/HeatTreat/MetalsAdvisor/iron-and_steel/processdescriptions/raw_metalspreparation/steelmaking/primaryfinishing/continuous%20casti
ng/continuous_casting_process_description.htm (9 May 2007).
"Electric Arc Furnace: Process Description." METALS Advisor.
http://www.energysolutionscenter.org/HeatTreat/MetalsAdvisor/iron-and_steel/processdescriptions/raw metal s_preparation/steelmaking/electric_arc_furnace/electric_arc_furna
ceprocess-description.htm (9 May 2007).
"Iron and Steel Scrap Statistics." USGS.
http://minerals.usgs.gov/ds/2005/140/ironsteelscrap.pdf (9 May 2007).
"Iron Ore Statistics." USGS. http://minerals.usgs.gov/ds/2005/140/ironore.pdf (9 May
2007).
Kelly, Thomas D. and Matos, Grecia R. "Historical Statistics for Mineral and Material
Commodities in the United States." UCGS Minerals Information.
http://minerals.usgs.gov/ds/2005/140/ (9 May 2007).
Kuck, Peter. "Iron Ore Statistical Compendum." USGS. Table 12.
http://minerals.usgs.gov/minerals/pubs/commodity/iron_ore/stat/tbll2.txt (9 May 2007).
"One Way ANOVA." http://www.wellesley.edu/Psychology/Psych205/anova.html (9
May 2007).
APPENDICES
A - Excerpts from: "50 Years of the ECSC Treaty: Coal and Steel Statistics."
Luxembourg: Office for Official Publications of the European Communities, 2002. ISBN
92-894-3715-4. European Communities, 2002. (Accessed through
http://epp.eurostat.ec.europa.eu/portal/page?_pageid = 1073,46587259&_dadportal&_sch
ema=PORTAL&p_product_code=KS-43-02-979 (9 May 2007).
B - "Iron Ore Statistics." USGS. http://minerals.usgs.gov/ds/2005/140/ironore.pdf (9
May 2007).
C - "Iron and Steel Scrap Statistics." USGS.
http://minerals.usgs.gov/ds/2005/140/ironsteelscrap.pdf (9 May 2007).
D - "Annual Energy Review." EIA. Table 8.10.
http://www.eia.doe.gov/emeu/aer/pdf/pages/sec8_39.pdf (9 May 2007).