Professional Documents
Culture Documents
Dutta 2001
Dutta 2001
This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact permissions@informs.org.
The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.
© 2001 INFORMS
INFORMS is the largest professional society in the world for professionals in the fields of operations research, management
science, and analytics.
For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org
A Survey of Mathematical Programming
Applications in Integrated Steel Plants
Goutam Dutta • Robert Fourer
Indian Institute of Management, Vastrapur, Ahmedabad 380015, India
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
M athematical programming techniques were used in the steel industry as early as 1958,
and many applications of optimization in steel production have been reported since
then. In this survey, we summarize published applications in the largest steel plants by type,
including national steel planning, product-mix optimization, blending, scheduling, set cov-
ering, and cutting stock.
(Steel; Applications; Mathematical Programming; Optimization)
phasis has been placed on the real-world implemen- continuous caster or poured into various ingot molds.
tation of the models. A brief description of an inte- The molten steel from other shops is cast into ingots.
grated steel plant is given in §2, prior to the survey In the primary rolling stage, ingots are shipped to
in §§3–9. the soaking pits where they are heated by a mixture
of gases to a uniform temperature, before being rolled
into blooms and slabs in the blooming mill. The
2. An Overview of an Integrated blooms are further rolled in the sheet bar and billet
Steel Plant mill into either sheet bars or billets. In the finishing
Figure 1 an iron- and steel-making plant as having rolling operation, the slabs, sheet bars, strip bars, and
four stages: iron making, steel making, primary roll- billets are the input materials to various finishing
ing, and finishing rolling. The output of each stage mills. The slabs are rolled in the plate mill into high
becomes the input to the following stage. In the iron- tensile and wear-resistant plates or ordinary mild
making stage, the blast furnaces are used to convert steel plates. The sheet bars are further rolled in the
iron ore, sinter, and other raw materials into molten sheet mills into high silicon, LPG (Liquid Petroleum
iron called hot metal. Hot metal is supplied to the gas), and galvanized sheets. The strip mill converts
steel melting shops where the process of steel making the strip bars into cold rolled or ordinary strips,
is either BOF (Basic Oxygen Furnace), OHF (Open which go either to the market or to the tube-making
Hearth Furnace), or EOF (Energy Optimizing Fur- plant.
nace). The molten steel from BOF is either sent to the Billets from the sheet bar and billet mill go either
to the conversion agents or to the merchant mill Stochastic Programming Model for Investment
where they are rolled into twisted bars, angles, and Planning in India
octagons. The blooms are further rolled into seamless Anandalingam (1987) discusses a stochastic program-
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
gothics (for seamless tube making) or into structurals ming model for investment planning in environments
in the medium and light structural mill. where demand projections and technological coeffi-
cients are not known with certainty. The model has
been used primarily for strategic planning rather than
3. National Steel-Planning Models operational planning. The usual programming for-
Before describing applications developed for specific mulation of an industrial process is extended to in-
integrated steel plants, we mention in this section corporate parameters and demand uncertainties by
several steel-planning models for national economies, modeling it as a stochastic linear program with sim-
using linear programming techniques. ple recourse (SLPR). The SLPR is solved using the less
National Steel-Planning Model in the United restrictive assumption that only the means and vari-
States ances of the stochastic entities (but not their distri-
Tsao and Day (1971) develop a process analysis mod- butions) are known. The methodology is applied to
el of production in the United States on a national the study of the steel industry in India with a novel
level. A technology matrix, which represents the tech- way of modeling investment and economies of scale.
nology structure, is estimated using engineering and
metallurgical information. This matrix together with
the detailed cost, sales, and revenue figures is then 4. Product-Mix Optimization
used in a linear programming model of short-run al- Models
locations of the steel industry as a whole. The linear In an integrated steel plant, the problem of determin-
programming model’s solution is obtained and com- ing the optimum production level at various stages is
pared with available industry statistics for each year of great practical importance. This is so because the
from 1955–1968. Although Tsao and Day claim to profit is sensitive to the product mix and not merely
have fairly good results, a later study by Nelson to the total volume of production. Because of the
(1971) reported that the model had an error in the complexity, suboptimal workable solutions are gen-
treatment of coking coal production. Nelson attempt- erally obtained by experience. Although these solu-
ed to correct this deficiency and presented a corre- tions, when implemented, achieve good plant utili-
lated matrix for this stage of production. zation, profits/revenue from these solutions are
considerably less than the potential profit/revenue
Mexican Steel Model that could have been accrued using the optimum
This study by Kendrick et al. (1984) develops two product mix. The optimum product mix changes
static models for production planning and one dy- from month to month and with the mill/furnace
namic model for investment analysis. The two static availability, and demand for the product in the mar-
models, formulated as linear programming models, ket. Pioneering work in this area by Fabian (1958) was
are mixed production and transportation problems. undertaken at Kaiser Steel Company, and since then
Inputs are prices of raw materials, operations and a number of applications in this area have been re-
shipments, demands, facility capacities, and input ported.
and output coefficients for each productive unit. Out-
puts are optimal product distributions. The dynamic Product-Mix Model at Kaiser Steel Company
model, formulated as a mixed-integer program, in- An integrated steel plant has a choice of the use of
corporates time factors and deals with the investment various materials and production processes. The eco-
issues in five time periods of three years each. The nomical usage rate of all materials is a function of a
inputs are similar to those in the static models but number of variables. Some of the most important var-
the output also includes investment decisions. iables are the market price of some materials, notably
various grades of steel scrap. This scrap price fluc- Models for Production Planning in the United
tuates, and therefore requires the periodic determi- Kingdom
nation of economical usage rate. The work of Fabian Lawrence and Flowerdew (1963) develop an econom-
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
(1958, 1967) is a cost-minimization linear program- ic model of steel production that focuses on the ap-
ming model that has four submodels: one for iron plication to the individual processes. A single-cost
making, one for steel making, and one each for shop model is constructed containing input and output
loading for rolling operations and finishing opera- variables, cost of variables and operations, relation-
tions. The models of various stages of production are ships between and restrictions on the variables, tech-
connected to form a ‘‘Master Model’’ of an integrated nical relationships, and flow restrictions. A simplex-
steel plant. The detailed formulation at each stage and type tableau is then constructed for a simplified
the principles of integration are also discussed in model, and the optimal solution is then computed.
these papers. The model considers all the techno-eco- Bandyopadhyay (1969) proposes a linear program-
nomical constraints like the capacity balance, material ming model that allocates different capacities be-
balance, product-dependent yield and thermal energy tween two processes for production planning, namely
balance (in the form of enthalpy balance). However, the basic oxygen furnace and the open hearth fur-
the oxygen balance and electrical energy balance are nace. The model is a cost-minimization model with
not discussed. all the technological and financial constraints. The
model can also predict the required operation level of
Large-Scale Database Model for the American Iron blast furnaces and lime-burning plants at different
and Steel Institute levels of total steel production.
Fourer (1997) presents a model which grew out of a
project to design an optimization package for steel German Model at Hoesch Siegerlandewerke
mill planning. Because this project was supported by Bielfield et al. (1986) at Hoesch Siegerlandewerke AG
the American Iron and Steel Institute (AISI) and not (HSW) in Germany have developed a set of account-
any particular steel company, it was based on a ge- ing matrices for budget planning. The company had
neric model. Any steel plant could customize the a revenue of one billion Deutsche Marks, and its main
model to its own operation simply by supplying its products were cold-rolled, hot-dip galvanized, elec-
own data. Users of this model would be concerned trogalvanized, and organic coated sheet steel. The
mainly with entering and maintaining their data and complexity of the steel company’s structure, opera-
with reporting the optimal production levels. The tion, and rapid environmental changes forced the
model is generic in nature and can be transported to HSW management to replace a manual system with
other similar industries. a computer-based strategic planning system with the
This work has been used in a number of steel objective of improving efficiency and performing
plants such as LTV, Dofasco, and Armco. Dofasco has mass calculations and cost accounting more efficient-
used this database-optimization software to generate ly. This is a linear programming model with multiple
models in excess of 1,000 variables, and Armco has objectives. These objectives may be maximizing rev-
developed an equivalent of this software in a spread- enue or minimizing total cost or cost per ton of steel
sheet (Excel) using the same solver but with a variety produced. The model has about 2,500 constraints and
of reports and diagrams customized to the company’s 3,000 structural variables.
requirements. In the LTV steel plant, it was suggested
to use this model in two plant-production and dis- Product-Mix Optimization Models in Indian Steel
tribution problems. Plants
On the basis of the above model, the importance of In India, the prices of half of all steel products were
inventories and the linkage between the time periods controlled by the Government from the 1950s until
was investigated by Hung (1991). See the later dis- 1991. In this environment, two interesting applica-
cussion of the AISI Inventory Model for details. tions of planning have been reported.
During the past 15 years, India has been affected nical, economical, and environmental constraints
by an energy shortage. The crisis is most significant such as the balance of capacity, materials, energy, and
in the eastern part of India where the gap between oxygen. It is an optimization model of an integrated
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
supply and demand is greatest. The poor capacity steel plant with blast furnaces, steel-melting shops,
utilization of some power plants (which supply pow- and primary and finishing mills in a global energy-
er to the steel plant) makes the operation of energy- crisis environment or hot metal shortage situation.
consuming plants extremely difficult. In the operation The model has different objectives: maximizing profit
of a steel plant, some of the energy-consuming pro- contribution, minimizing cost or maximizing produc-
cessors (called essential loads) require a fixed amount tion; it has about 1,000 constraints and 1,000 vari-
of power and cannot be switched off, even in the ables. Its outputs are converted to a priority list of the
event of a power crisis. In this environment, the op- facilities to be switched off during the energy crisis.
erating manager of a plant has no other option but The round-the-clock implementation of the model has
to switch off those processors that are not essential improved the profitability of the steel plant signifi-
loads. Optimal allocation of electrical energy is thus cantly since 1986.
a very important decision for the management of the The Steel Authority of India Limited (SAIL), the
steel plant. largest steel company in India, is a multiproduct
Dutta et al. (1990), Dutta et al. (1994), and Sinha et company producing a wide range of products from
al. (1995) deal with the development and implemen- its five integrated steel plants at Bhilai, Bokaro, Dur-
tation of a mathematical model for optimal allocation gapur, Burnpur, and Rourkella. The salable outputs
of electrical energy in a plant of Tata Steel. The guid- from these plants can be divided into pig iron, semi-
ing principle of the model is that in the case of a finished steel, and finished steel. Another interesting
power shortage, power is allocated to those nones- option among these five steel plants is that of inter-
sential loads which have a higher profitability (based plant transfers. This arises because of the imbalances
on a mixed-integer linear programming model). Al- at various stages of production across SAIL steel
though a number of studies (Hunneault and Galiana plants. Sharma and Sinha (1991) describe an optimi-
zation model for determining the optimal product
1991, McCutcheon 1988) have reported the optimal
mix for the integrated steel plants of SAIL. The paper
use of power plants, such studies have addressed the
begins with a discussion of various issues relevant to
issue with a cost-minimization modeling approach
the choice of an optimum product mix in a steelmak-
for power-generating and distributing plants. Others
ing operation. Some planned applications of the mod-
have studied the most profitable use of an integrated
el are also discussed.
steel plant (Fabian 1958, Bielfield et al. 1986, Baker et
al. 1987) where the problem has been addressed as a Models of Production Planning in Zambia
cost-minimization or profit-maximization linear pro- Sashidhar and Achray (1991a) deal with the problem
gramming model. of production planning in a steel mill with the objec-
In the Tata Steel application, the steel plant has tive of maximizing capacity utilization. The model is
been modeled with a (contribution to) profit-maxi- formulated as a maximum flow problem in a multi-
mization objective, with energy as a limiting con- ple-activity network. The production is usually
straint. This is the pioneering attempt in India, where planned against customer orders and different cus-
the mathematical programming model has been im- tomers are assigned different priorities. The model
plemented not only for long-term strategic-planning takes into account the priorities assigned to the cus-
decisions, but also for short-term operating decisions. tomers and the order balance position. An algorithm
This use is not only in an integrated steel plant, but is presented for solving the multiple-activity network
also in an integrated steel plant vertically integrated formulation for production planning with the cus-
with a tube-manufacturing plant, which requires tomer priorities in a steel mill.
higher complexity. The model considers all the tech- In another paper, Sashidhar and Achray (1991b) dis-
cuss the problem of allocating the major components of a ‘‘blast furnace burdening problem.’’ The results ob-
process costs to various quantities of products pro- tained from Fabian (1967) enable a producer to de-
duced in a melting shop of an alloy and special steel- termine:
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
of alloys at the lowest price from a number of stocks Scheduling Model at LTV Steel
of scrap alloys of various composition and from un- In 1983, LTV Steel Company started up a twin-strand
alloyed metals is achieved through the use of a new continuous slab caster to convert molten steel to solid
algorithm. The method differs from normal linear pro-
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
compared to teeming (that is, casting by pouring The optimal design of production through a hot
molten steel into molds). This system annually strip mill is characterized by multiple and conflicting
saves over $1.95 million by reducing personnel and objectives. Jacobs et al. (1988) propose an optimiza-
tion model for this situation. Considering the hot
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
inventories and the linkage between the time periods The problem is to define the start time and the du-
was investigated by Hung (1991). Data for the plate ration of a production cycle in combination with a
mill and the batch-annealing process of Bethlehem, power schedule that meets the energy requirements
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
Armco, and LTV were used in an empirical study, of the different furnaces and a global power-supply
sponsored by the American Iron and Steel Institute. limit for the whole plant. The problem is formulated
Relations between the inventory level for plate mills as a combination of an optimization problem and an
and the batch-annealing process were determined by optimal control problem. The authors have proposed
least squares and least absolute deviation regressions. a two-level algorithm that shows 9% improvement in
A two-step procedure for production scheduling productivity on some test data.
was also proposed. It first assigns slabs to each plate
order and then sequences the rolling jobs. The slab
assignment was formulated as a linear programming 7. Set-Covering Applications
model with the objective of either maximizing yield, In this section, we discuss applications of the set-cov-
maximizing revenue, or maximizing profit. Both the ering approach in the area of assignment of slabs to
optimal slab assignment and the slab inventory mix orders, metallurgical grade assignment, and selecting
are determined by the slab-assignment model. The optimal ingot sizes. All studies in this section have
job-sequencing problem then finds a job sequence that been reported at facilities of Bethlehem Steel.
fulfills the operational constraints and also maximiz-
es plate quality. Optimal Ingot-Size Determination
After the installation of a new ingot mold striping
Dynamic Scheduling at Ensidesa Steel in Spain facility in 1984, Bethlehem Steel developed a two-
After building a new steel plant, Empresa National phase procedure for selecting optimal ingot dimen-
Siderurgica implemented automatic control in various sions, as reported in a series of publications (Vasko
production sections, giving the process computers 1984, Vasko and Wilson 1984a, 1984b, 1986, Vasko,
continuous and complete information throughout the Wolf, and Stott 1987, Vasko and Wolf 1988, Vasko et
production process. Making use of this information, al. 1989a). Previously, Bethlehem had been using
Diaz et al. (1991) developed an automatic coordinating about a dozen ingot mold sizes. Based on experience,
system for each facility in the plant. In this system, the it was established that any increase in the number of
operator selects a set of heats to produce and makes a distinct mold sizes would result in a significant in-
predetermined production scheme from various pre- crease in inventory and material-handling cost.
planned strategies. The system then arranges the heats The two-phase procedure is used for selecting the
accordingly and simulates the delay and the idle times optimal ingot dimensions and internal mold dimen-
that could occur if the operator chooses that scheme. sions. This procedure also incorporates research in
Unlike some American steel plants (where the se- yield improvement and a variety of metallurgical and
quences last for dozens of heats), the Spanish steel operational constraints. Only marginal improvement
plants have short sequences (six or seven sequences would have been possible if the old mold sizes had
per day). As the sequences are short, the objective is been retained. Phase I of the procedure generates fea-
to maximize the time the casters are producing slabs. sible ingot mold dimensions consistent with the con-
straints; Phase II then uses a set-covering approach to
Scheduling Model at a Canadian Steel Plant select, from the feasible sizes generated, the ingot di-
This work by Boukas et al. (1990) is an optimization mensions and ingot mold dimensions that minimize
model of productivity in a steel plant subject to glob- the number of distinct mold sizes required to produce
al energy constraints. The plant has four arc furnaces the finished products. On the basis of the results of
and three continuous-casting machines. In electric arc this model and trial mill tests, full production use of
furnaces, the allocation of energy, the fusion phase of new mold sizes influenced the entire plant operation
the total production cycle, is of critical importance. and resulted in annual savings of over $8 million.
the plates (customer service considerations), number. Once the solution of finding rectangular
and stock sizes is known, a second program is used to
(3) Cutting orders from a plate for as few distinct map the bill of materials onto plates of the chosen
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
customers as possible (logistical concerns). sizes. The practicality of the approach is demonstrat-
ed by generating a cutting plan for a real-world bill
Vasko et al. (1989) and Vasko (1989) present a for-
of materials having 392 distinct order sizes and over
mulation in a fuzzy environment that addresses these
7,700 order pieces.
concerns. A solution procedure is outlined and prac-
In a mill finishing a structural shape such as an I-
tical implementation at Bethlehem Steel’s Sparrows
beam, once the final product is produced it is cut
Point plant is described in Vasko et al. (1991). The
according to the customer’s order length. The actual
plant can produce narrow-width customer-plate or-
length may not be known precisely until just before
ders (typically 10 to 24 inches) efficiently when its 60-
cutting. Also, if the production rate of the mill is
inch plate mill is not operating. The heuristic proce-
higher than the cutting rate of the bars, then trying
dure is used to map these orders into mother plates
to generate cutting patterns with the number of cuts
for production in the 160-inch plate mill. This pro-
per bar close to the average number of cuts per bar
cedure was implemented as a module in the plant’s
will maximize primary saw (hotsaw) cutting and re-
production planning and control system and has
duce the number of cuts that have to be made at the
been used daily to generate mother-plate dimensions
secondary saw (coldsaw). Vasko et al. (1993b) discuss
and cutting patterns.
a branch-and-bound algorithm that generates high-
In another application, Vasko et al. (1992) discuss a
yield balanced cutting in real time based on the pre-
method that combines set-covering and cutting stock
cise length of the bar leaving the mill and arriving at
applications for improving Bethlehem Steel’s custom-
the saw.
er service. Some of the customer orders are slit from
master coils into a number of narrower and smaller Cutting Stock Applications in a German Steel
coils to fit specific manufacturing needs. To serve Plant
these customers, Bethlehem has developed a mathe- Pohl and Kaiser (1982) develop a cut-length optimi-
matical model that generates optimal coil widths and zation program for the computer-controlled Siege
slitting patterns. The model has the following objec- GeisWeid AG rolling mill. They describe a procedure
tives: for cutting the rolling strand lengths into marketable
lengths. The total rolling strand length is computed
(1) Minimize the number of slitter setups,
by comparison of volume and speed of billets, mer-
(2) Maximize the material utilization,
chant bars (after the first rolling block), and finished
(3) Generate minimum excess inventory, and
products. The speeds and lengths are determined by
(4) Generate minimum shortfall against forecast
measuring rollers in the front part and without con-
demand.
tact at the rear end of the mill. The cooling-bed
The linear program also generates coil widths that lengths are divided according to the optimization
optimally utilize the company’s facilities. This system computation and are conveyed under computer con-
is viewed by the customers as a value-added service trol to two cutting-off machines, which cut into mar-
provided by Bethlehem Steel. ketable finished lengths.
Vasko and Wolf (1994) address the problem of de-
termining what rectangular sizes should be stocked
to satisfy a bill of materials composed of smaller rect- 9. Other Applications
angles. They first generate a large number of stock The continuous-casting machine can be used to elim-
sizes ideally suited to the bill of materials; they then inate a number of processing steps associated with
use an uncapacitated facility-location algorithm to the traditional ingot/bloom-based production se-
consolidate the stock sizes down to an acceptable quence. However, a given continuous caster can pro-
duce only a small number of bloom thicknesses. This the following can be considered as potential areas for
creates a problem for selecting those continuous-cast- future work:
er configurations that would maximize utilization. (1) Simultaneous optimization of product-mix, in-
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
Vasko and Friedel (1982) present a dynamic program- ventory, and transportation problems over multiple
ming formulation that maximizes the cast bloom ton- periods. This would represent an extension of Fabian
nage that can be processed through one of the Beth- (1958) to the multiperiod case with inventory and
lehem Steel’s finishing mills. Without the aid of such transportation requirements as additional constraints.
a model, selecting the highest productivity would (2) Cutting stock optimization to maximize overall
have two conflicting considerations. The first factor is yield of multistage production processes. This would
that as the number of caster-produced bloom thick- go beyond most previous work on the cutting stock
nesses increases, the caster setup time and the con- problem, which has used single-stage models.
figuration complexity increases. The second factor is (3) Scheduling problems in the continuous caster.
that as the number of thicknesses decrease, the cast (4) Stochastic linear programming models where
tonnage processed through the finishing mill is re- not only the means and variances of the stochastic
duced, owing to reheating furnace and cooling-bed entities, but also their distributions, are known.
limitations. The model results were transmitted to the (5) Any research that increases the reliability and
plant management and were used in conjunction validity of the data. The success of mathematical pro-
with other information to determine the most eco- gramming models depends heavily on availability of
nomic caster configurations. relevant data. Often the desired data does not exist,
The Electro-Slag Remelting (ESR) process was de- or must be collected from multiple sources.
veloped for melting special alloys that were difficult Acknowledgments
to produce in conventional electrical arc furnaces. The authors gratefully acknowledge financial support from the
Gower et al. (1970) describe an application of dynam- American Iron and Steel Institute and an American steel company.
ic programming simulation to determine an ESR op-
Glossary
erating policy that is predicted to maximize cumu- (For a much more detailed glossary, see Everything You Always Wanted
lative profit over a number of stages. to Know About Steel . . . A Glossary of Terms and Concepts by Michelle
Applebaum, 具http://www.steel.org/learning/glossary/glossary.
htm典.)
10. Conclusion and Extensions Billets: Mostly square steel shapes in the range of 50mm ⫻
Although steel is a basic industry for the growth of 50mm to 125mm ⫻ 125mm. They may be semifinished or finished
a nation, relatively few applications of mathematical products depending on the customer. Blooms are rolled into billets.
Blast Furnace: A facility that converts iron and other raw mate-
programming have been reported in comparison
rials to hot metal (liquid iron at a very high temperature). A typical
with other industries such as oil, airlines, and semi- blast furnace is about 30 m high and produces 500 to 10,000 tons
conductors. Also, very little work has been done in of hot metal per day.
the area of inventory control and manufacturing con- Blooms: Steel shapes that have a cross-section smaller than ingots
trol for steel plants. However, it is noteworthy that but larger than billets. They are square or slightly oblong, mostly
in the range of 150mm ⫻ 150mm to 300mm ⫻ 300mm. Ingots are
four applications (Jain et al. 1978, Box and Herbe
rolled into blooms.
1988, Vasko et al. 1989a, Sinha et al. 1995) have been Coils/Wire Rods: The smallest round sections of steel that can be
selected as finalists in the competition for the Franz produced by hot rolling. The sizes of rods vary from 5.5mm to
Edelman Award for Management Science Achieve- 12.7mm. Generally, rods are wound into coils of about 760mm in-
ment. This gives an indication of the potential finan- side diameter that weigh from 450 to 2,000 kilograms.
Continuous Caster: A facility between the basic oxygen furnaces
cial benefit of applying optimization techniques to the
and the rolling and finishing mills. It casts slabs and billets directly
problems of the steel industry. from the liquid metal, bypassing the ingot stage.
From the survey of different applications and our Heat: A batch of liquid steel, varying from about 50 tons to 300
personal experience in the modeling of steel plants, tons depending upon the technology and type of the blast furnace.
Hot Strip Mill: The rolling mill that reheats and rolls steel slabs steel production scheduling under a global energy constraint.
into hot bands, steel strips that are typically 0.10 inches thick and Ann. Oper. Res. 26 289–311.
50 to 60 inches wide. Box, R. E., D. G. Herbe, Jr. 1988. A scheduling model for LTV Steel’s
Ingot: Individual shapes cast by pouring liquid steel into indi- Cleveland works’ twin strand continuous slab caster. Interfaces
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
vidual molds. With continuous casters becoming more and more 18(1) 42–56.
common in steel making, ingots are tending to become obsolete. Diaz, A., L. Sancho, R. Garcia, J. Larrantea. 1991. A dynamic sched-
Ladle: A ceramic-lined open container used to transport and hold uling and control system in an ENSIDESA steel plant. Inter-
a heat of molten steel. faces 21(5) 53–62.
Mixer: A reservoir for storing and heating hot metal from a blast Dutta, G., G. P. Sinha, P. N. Roy. 1990. A product-mix optimizer for
furnace before it is sent to subsequent production steps. Its purpose an integrated steel plant. Abstracts Booklet, IFORS-90, 49–50.
is to maintain consistency in the composition, variation, and tem- , , , N. Mitter. 1994. A linear programming model for
perature of the hot metal. distribution of electrical energy in a steel plant. Internat. Trans.
Pig Iron: The metallic product of the blast furnace, containing Oper. Res. 1 19–30.
over 90% iron. Fabian, T. 1958. A linear programming model of integrated iron and
Pusher-Scraper: A machine used to transport raw materials like steel production. Management Sci. 4 415–449.
iron ore from one point to another. Some pushers are also used to . 1967. Blast furnace production planning—A linear program-
take coke out of the coke ovens. ming example. Management Sci. 14 B1–B27.
Rollers: The objects through which ingots are passed to produce Fourer, R. 1997. Database structures for mathematical programming
finished steel. Rollers are also used in other production steps to models. Decision Support Systems 20 317–344.
reduce the cross-sectional area of the product. Gercuk, M. Y. P. 1961. Linear programming in organization and
Slab: The intermediate product from a continuous caster or a planning of metallurgical production. Metallurgizdat 21–27.
roughing mill. It is always oblong in shape, mostly 50 mm to 230 Gower, R. C., W. C. Hahn, S. K. Tarby. 1970. Dynamic programming
mm thick and 610 mm to 1,250 mm wide. simulation of electroslag remelting process. J. Iron and Steel In-
Slag: The fusible material formed by the chemical reaction of a stitute. 208 1093–1099.
flux with gangue of an ore, with ash from a fuel, or with impurities Haessler, R. W., M. A. Vonderembse. 1979. A procedure for solving
oxidized during the refining of a metal. the master slab cutting stock problem in the steel industry.
Surface Quality: The presence or absence of flaws in the surface AIIE Trans. 11 160–165.
of a steel strip or sheet. The incidence of these flaws is extremely Hernandez, J. P., J. P. Proth. 1982. A good solution instead of an
sensitive to the process of steel making and/or slab casting or ingot optimal one. Interfaces 12(2) 37–42.
teeming. Hung, Y. 1991. Models for production and strategic planning in the
Teeming: The process by which molten steel is poured into ingot steel industry. Ph.D. dissertation, Department of Industrial En-
molds. gineering and Management Sciences, Northwestern University,
Thermal Chamber Lateral Deformation: Deformation of refrac- Evanston, IL.
tories due to high temperature. Hunneault, M., F. D. Galiana. 1991. A survey of optimal power flow
literature. IEEE Trans. Power Systems 6 762–767.
References Jacobs, L. T., J. R. Wright, A. E. Cobb. 1988. Optimal inter process
AISE Steel Foundation. 1998. The Making, Shaping, and Treating of Steel, steel production scheduling. Comput. Oper. Res. 15 497–507.
11th ed. AISE Steel Foundation, Pittsburgh, PA. Jain, S. K., K. L. Stott, E. G. Vasold. 1978. Order book balancing
Anandalingam, G. 1987. A stochastic programming model for in- using a combination of linear programming and heuristic tech-
vestment planning. Comput. Oper. Res. 14 521–536. nique. Interfaces 9(1) 55–67.
Baker, G. L., W. A. Clark, J. J. Frund, R. E. Wendell. 1987. Production Kendrick, D. A., A. Meeraus, J. Alatorre. 1984. The Planning of In-
planning and cost analysis on a microcomputer. Interfaces 17(4) vestment Programs in the Steel Industry, John Hopkins University
53–60. Press, Baltimore, MD.
Bandyopadhyay, R. 1969. Open hearth and basic oxygen furnaces: Lawrence, J. R., A. D. J. Flowerdew. 1963. Economic models for pro-
An allocation model for production planning. IEEE Trans. Sys- duction planning. Oper. Res. Quart. 14(1) 11–29.
tem Sci. Cybernetics 12 115–124. McCulloch, G. A., R. Bandopadhay. 1972. Application of operational
Beale, E. M. L., P. J. Coen, A. D. J. Flowerdew. 1965. Separable pro- research in production problems in the steel industry. Internat.
gramming applied to an ore purchasing problem. Appl. Statist. J. Production Res. 10 77–91.
14 89–101. McCutcheon, J. M. 1988. Long range electricity expansion planning
Bielfield, F. W., K. Walter, R. Wartmann. 1986. A computer based using a mixed integer linear programming model. Asia Pacific
strategic planning system for steel production. Interfaces 16(4) J. Oper. Res. 5 53–56.
41–46. Metzger, R. W., R. Schwarzreck. 1961. A linear programming ap-
Boukas, E. K., A. Haurie, F. Soumis. 1990. Hierarchical approach to plication to cupola charging. J. Indust. Eng. 12(2) 87–93.
Mihailor, M. O. A. 1961. Mathematical statistics and linear program- termining continuous caster configurations. IIE Trans. 14 38–
ming in ferrous metallurgical. Metallurgizdat 160. 43.
Muteanu, E., F. Rado. 1960. Calculation of most economical charges , G. R. Wilson. 1984a. Using facility location algorithms to solve
for pig iron smelter. Studii si Cercetari Stiintifice Mathematica, large scale set covering problems. Oper. Res. Lett. 3 85–90.
Downloaded from informs.org by [129.120.242.61] on 30 November 2014, at 05:40 . For personal use only, all rights reserved.
Academia Republici Populare Romine Fililia iasi, Institutul de , . 1984b. An efficient heuristic for large scale set covering
Mathematica Clju 11(1) 149–158. problems. Naval Res. Logist. Quart. 33 241–249.
Nelson, J. P. 1971. A note on economics of metallurgical coke pro- , . 1986. Hybrid heuristic for minimum cardinality set cov-
duction. Management Sci. 18 B237–B239. ering problem. Naval Res. Logist. Quart. 33 241–249.
Newhart, D. D., K. L. Stott, F. J. Vasko. 1993. Consolidating product , F. E. Wolf. 1988. Solving large set covering problems on a
sizes to minimize inventory levels for a multistage production personal computer. Comput. Oper. Res. 15 115–121.
, . 1989. A set covering approach to metallurgical grade
and distribution system. J. Oper. Res. Soc. 44 637–644.
assignment. Euro. J. Oper. Res. 38 27–34.
Pohl, B., W. Kaiser. 1982. Cut length optimization by volume com-
, . 1994. A practical approach for determining rectangular
parison for a merchant mill (Schnittlaengenoptmierung uber
stock sizes. J. Oper. Res. Soc. 45 281–286.
Volumenvergliech fuer eine Stabstahlstrassre). Stahl und Eisen
, , J. A. Pflugrad. 1991. An efficient heuristic for planning
102 573–575.
mother plate requirements at Bethlehem Steel. Interfaces 21(2)
Rao, P. P. C., R. Singh, V. Rao, R. P. Mohanty. 1993. Applications of
38–49.
operational research techniques in the steel industry: A clas-
, , K. L. Stott. 1987. Optimal selection of ingot sizes via set
sificatory review. J. Shah and A. Tripathy, eds. Operational Re-
covering. Oper. Res. 35 346–352.
search in the Indian Steel Industry. Wiley Eastern Limited, New , , . 1989. Solution to fuzzy two-dimensional cutting
Delhi, India pp. 1–15. stock problem. Fuzzy Sets and Systems 29 259–275.
Sarma, G. V. 1995. Application of lexicographical goal programming , M. L. Cregger, D. D. Newhart, K. L. Stott. 1993b. A real time
to solve a product mix problem in a large steel manufacturing one-dimensional cutting stock algorithm for balance cutting
unit—A case study. OPSEARCH 32 55–78. patterns. Oper. Res. Lett. 14 275–282.
Sashidhar, B., K. K. Achray. 1991a. Multiple arc network of produc- , , K. L. Stott, L. R. Woodyatt. 1994a. Assigning slabs to
tion planning in steel mill. Internat. J. Production Econom. 22 orders—An example of appropriate model formulation. Com-
189–193. put. Indust. Engin. 26 797–800.
, . 1991b. Applications of OR techniques in cost allocation , J. A. McNamara, D. D. Newhart, F. E. Wolf. 1994b. A practical
of major operational materials in an alloy and special steel solution to a cargo loading problem at Bethlehem Steel. J. Oper.
manufacturing unit. Internat. J. Production Econom. 22 195–202. Res. Soc. 45 1285–1292.
Shah, J., A. Tripathy. 1993. Operational Research in the Indian Steel , F. E. Wolf, K. L. Stott, O. Ehrsam. 1992. Bethlehem Steel com-
Industry. Wiley Eastern Limited, New Delhi, India. bines cutting stock and set covering to enhance customer ser-
Sharma, A., S. K. Sinha. 1991. Product mix optimization: A case vice. Math. Comput. Model. 16 9–17.
study of integrated steel plants of SAIL. OPSEARCH 28 188– , , , J. W. Scherier. 1989a. Selecting optimal ingot size
201. for Bethlehem Steel. Interfaces 19(1) 68–83.
Sinha, G. P., B. S. Chandrasekaran, N. Mitter, G. Dutta, S. B. Singh, A. , , , L. R. Woodyatt. 1989b. A fuzzy approach to op-
R. Choudhury, P. N. Roy. 1995. Strategic and operational man- timal metallurgical grade assignment. G. W. Evans, W. Kar-
agement with optimization at Tata Steel. Interfaces 25(1) 6–19. wowski, M. R. Wilhelm, eds. Applications of Fuzzy Set Method-
ologies for Industrial Engineering. Elsevier, Amsterdam, The
Stott, K. L., Jr., B. W. Douglas. 1981. A model based decision support
Netherlands 285–299.
system for planning and scheduling ocean borne transporta-
, , , . 1993a. Adapting branch and bound for real
tion. Interfaces 11(4) 1–10.
world scheduling problems. J. Oper. Res. Soc. 44 483–490.
Tarabar, K. 1963. An example of the use of linear programming
Vonderembse, M. A., R. W. Haessler. 1982. A mathematical pro-
model in ferrous industry (in Serbo Croatian, English sum-
gramming approach to schedule master slab casters in the steel
mary). Automatika 4(5–6) 311–314.
industry. Management Sci. 28 1450–1461.
Tokuyama, H., Nobuyki. 1981. Cutting stock problems in the iron
Westerberg, C., B. Bjorklund, E. Hultman. 1977. An application of
and steel industry. Proc. 9th Triennial Conf. IFORS. Hamburg, mixed integer linear programming in a steel mill. Interfaces
Germany 809–823. 7(2) 39–43.
Tsao, C. S., R. H. Day. 1971. A process analysis model for the U.S Woodyatt, L. R., K. L. Stott, F. E. Wolf, F. J. Vasko. 1992. Using fuzzy
steel industry. Management Sci. 17 B588–B608. sets to optimally assign metallurgical grades to steel. J. Metals
Vasko, F. J. 1989. A computational improvement to Wang’s two-di- 44(2) 28–31.
mensional cutting stock algorithm. Comput. Indust. Engin. 16 , , , . 1993. An application combining set covering
109–115. and fuzzy sets to optimally assign metallurgical grades to cus-
, D. C. Friedel. 1982. A dynamic programming model for de- tomer orders. Fuzzy Sets and Systems 53 15–26.