FINAL REPORT T O NATIONAL COUNCIL FOR SOVIET AND EAST EUROPEAN RESEARC H

T I T LE : Energy Consumption and Analysis o f Optimal Interregional and Internationa l l Flows in the Soviet Iron and Stee Industry

AUTHOR : Dr . Craig Zu m Brunnen University of Washington

CONTRACTOR : Boston University

PRINCIPAL INVESTIGATOR : Dr . Jeffrey P . Osleeb Boston University

COUNCIL CONTRACT NUMBER : 624-7

DATE : June 30, 198 3

The work leading to this report was supported in whole or i n part from funds provided by the National Council for Sovie t and East European Research .

NCSEER March 8 3

EXECUTIV E SUMMARY

Industrialization in the Soviet Union has concentrated on the iro n and steel industry . Sometimes in the past emphasis has appeared to hav e been on increasing sheer physical output rather than on reducing cost s or on increasing efficiency . But attention is turning increasingly t o the relative costs of production at various places throughout th e country, as affected by the location and quality of iron ore, cokin g coal, and other raw materials, by transportation costs of such ra w materials to iron and steel centers, by costs of processing in relatio n to scale and alternative process, and by costs of transportation t o reach the markets both domestic and foreign . This study is an attempt to analyze the optimal locations for iro n and steel production and to predict probable shifts in location of th e industry and in the amount and processes of production at both present and future centers up to the year 1990 . A mathematical model was formulated to study the costs, production , locations, and flows in the Soviet iron and steel industry . This mode l incorporated the movement of energy, raw materials , intermediate products, and steel from ultimate sources of localized raw materials t o ultimate markets for steel throughout the entire Soviet Union . Th e study included flows in 1980 and projected flows to 1990 . It involve d 'some 1,200 constraint equations . and 18,000 decision variables . Th e model, called SISEM, a spatially disaggregrated mathematical model, is a multi-stage and multi-process formulation . In general it incorporate s upper and lower production constraints by process or commodity, by node , and by year . Detailed Soviet production figures, costs, technologica l coefficients (and their projections to 1990) were incorporated whereve r possible . The model includes domestic coking coal, imports of cokin g coal, coke manufacture, iron ore, ore enrichment, ore exports, scra p metal, pig iron, steel (in six basic variants of four processes), fina l domestic and export demands for pig iron and steel, and costs o f transportation joining all of them . The major findings and speculations are as follows . The optima l supply locations for coking coal exports tends to shift increasingl y eastward to the Kuzbas and beyond both for European as well as Pacifi c exports . Like the current actual pattern, the Ukraine dominates as th e source region for Soviet iron ore exports in both model years with th e Kola Peninsula being the secondary supplier in 1980 and sharing that rol e with the KMA in 1990 . Nonetheless, the Ukraine ' s absolute export flow s decline slightly while its relative export market share drops in th e model from 92,5 percent in 1980 to 82 percent in 1990 . Also, unlike th e actual current iron ore export flow pattern, the KMA does not optimall y serve any export markets in 1980 but does export 3 .0-million ton s through in 1990 . Ukrainian pig iron furnaces and those at Lipets k dominate the Soviet pig iron export markets in both years . Ukrainia n dominances in the steel export flows in 1980 are seriously challenged b y 1990 by two Urals competitors, Magnitogorsk and Chelyabinsk . The modelled completion of the BAM (Baykal-Amur Mainline railroad ) by 1990 was clearly a necessary condition to allow the proposed Chul ' ma n steel site to enter the optimal solution for 1990 as an active stee l node . Total transportation costs amount to only 5 .5 and 7 .1 per cent o f total costs for 1980 and 1990, respectively . The study has considerable bearing on technological change withi n the Soviet ferrous metallurgical industry . The model clearly indicate s that significant technological process shifts are warranted . Fo r

- 1 -

NCSEER March 8 3

example, the model satisfies steel demands while requiring 3 .7-millio n tons less pig iron in 1980 than the Soviets report . This situation i s attributable to the indicated switch to the production vast quantitie s of direct reduction steel for 1990 (22-million out of approximatel y 170-million tons) . In fact, the modelled pig iron production for 199 0 is still less than the reported actual tonnage for 1980 (105 .1-millio n versus 107 .3-million tons) . Krivoy Rog stands out as the clear plac e for any expanded pig iron production . Finally, none of the fiv e potential pi iron sites : Staryy Oskol, Barnaul, Tayshet, Svobodny y (981) or Tynda appears as an optimal choice . Several differences appear between the 1980 and 1990 stee l production patterns . First, all four of the proposed new stee l locations, Barnaul, Zhlobin, Chul ' man and Tayshet enter the optima l solution for 1990, the first with over 18-million tons, the second tw o with 5 .4-million and 5 .2-million tons, respectively, and Tayshet wit h 1 .3-million tons . More importantly, all of these new sites excep t Tayshet in East Siberia are shown as actively using the new direc t reduction process to produce from a low of 4 .2-million tons at Zhlobi n in Belorussia to a high of over 12 .6-million tons at Barnaul in Wes t Siberia . Second, whereas many locations are producing at capacity i n 1980, relaxed upper limits on steel plant capacities for 1990 result i n only two locations, Magnitogorsk and Krivoy Rog, operating at thei r upper limit . Several small sites fail to enter either solution, fewes t of all in 1990 . In effect, the model selects in favor of scal e economies, or in other words, the Soviets still appear to hav e unrealized potential savings to be garnered from scale economies . Third, because of the scrap supply problems the electric-ar c process using both pig iron and scrap does not produce steel in either year . Fourth, practically all of the steel produced by open-hearth furnaces i n 1980 used coke only . For 1990, however, the open-hearth furnaces ar e optimally shown as using much less coke and much more natural gas . The model in effect tested the internal consistency of the relevan t Soviet production and technological coefficient data . In this regar d the most significant finding is the scrap insufficiency problem . Scra p inputs appear as binding constraints in both runs . Simply stated usin g Soviet data one can not produce as much steel in 1980 as Sovie t statistics claim . Even allowing for the tonnage of ferro-alloys, lowe r input coefficients, etc ., a significant shortfall still would exist . The model ' s results imply that the Soviets are double counting stee l production by as much as 15 .55-million tons or 10 .5 per cent for 1980 b y reusing large quantities of on-site scrap . The scrap shortage i s dispersed throughout the Ukraine, Urals and West Siberia in 1980, bu t confined to the Urals in the 1990 run . Furthermore, other modellin g efforts indicate the magnitude of "on-site " scrap may be aroun d thirty-million tons . If true, then, the actual published Soviet crud e steel production data could overstate "useful " or net steel productio n by as much as nearly thirty-million tons, essentialy twice the quantit y suggested here . The investigation systematically assessed the impacts of projecte d changes in the availability and relative prices of energy and ra w material inputs both by type and region . Accordingly, coking coa l presents no " physical supply " problems in the model primarily becaus e coke is used in other industries beside steel and hence the model yield s large coking coal slack production values . On the other hand, th e escalating costs of Donbas coals make West Siberian coals marketabl e throughout much of the European U .S .S .R by 1990 . Although the model i s

- 2 -

NCSEER March 8 3

not constrained by crude iron ore or enrichment in either simulation, w e believe both are, in fact, already bottlenecks . There are some hints o f this in the 1990 run where all high-grade ore sites are forced t o operate at their maximum modelled capacities . The indicated changes i n optimal low-grade and enriched ore production suggests that the Soviet s should concentrate their expansion efforts in the KMA (Kursk Magneti c Anomaly), new deposits in northwest European , Kremenchug, alon g the BAM, and much less so at Krivoy Rog and Rudnyy . Over the modelle d decade both the KMA and Ukrainian ore deposits play a significantl y larger role in Urals iron and steel production . Neither electricity nor natural gas appear to be limiting factor s overall for the foreseeable future as both have vast slack capacities . On a regional scale, however, Europe could well need both mor e electricity and gas, while the Urals may need more West Siberian gas . The relative share of the total systems cost arising from each of thes e energy inputs increases substantially from 1980 to 1990, that of natura l gas tripling and that of electricity nearly doubling . Finally, the model indicates the possibility of very significan t average production costs savings of approximately 9 .20 rubles per ton o r 9 .8 per cent of the 1990 over the much more constrained 1980 model year .

- 3-

NCSEER March 198 3

Table of Content s

INTRODUCTION 1 BACKGROUND FOR THE STUDY 1 NATURE OF THIS INVESTIGATION 3

THE SISEM MODEL 7 THE OBJECTIVE FUNCTION 7 MODEL CONSTRAINTS 1 0 DESCRIPTION OF SUB-SYSTEM OBJECTIVES 1 2 SOLUTION PROCEDURE 1 3

DATA, LIMITATIONS, and ASSUMPTIONS 1 4 CARTOGRAPHIC DATA AND TRANSPORT NETWORKS 1 5 TRANSPORT COST DATA 1 5 RAW MATERIALS AND INTERMEDIATE PROCESSES 1 7 Coking Coal 1 7 Coke Manufacturing Data 1 9 Iron Ore Production Data 1 9 Iron Ore Enrichment Data 2 2 Scrap Metal Supply Data 2 4 Regional Energy Cost Data 25 Pig Iron Production and Cost Data 2 7 Steel Production Data 3 1 System Constraints Data 4 6 Final Demand Data 4 7

DISCUSSION OF THE RESULTS 5 4 INTRODUCTION 5 4 COKING COAL PRODUCTION AND FLOW PATTERNS 5 8 Coking Coal Production for 1980 5 8 Coking Coal Flow for 1980 59 Coking Coal Production for 1990 6 2 Coking Coal Flows for 1990 63 COKE MANUFACTURE AND FLOW PATTERNS 6 7 Coking Process for 1980 6 7 Projected Changes in Coke Manufacture between 1980 and 1990 6 9 Coke Flow Patterns for 1980 and 1990 7 0 IRON ORE PRODUCTION AND FLOW PATTERNS 7 6 Iron Ore Production for 1980 7 6 Projected Optimal Ore Production for 1990 7 8 Ore Enrichment 80 Changes in High-Grade Ore Output from 1980 to 1990 83 Changes in Low-Grade Ore Production from 1980 to 1990 83 Iron Ore Flow Patterns 8 5 Interpretation of Actual and Modelled Ore Production 90 SCRAP METAL SUPPLY AND FLOW PATTERNS 94 Scrap metal supply Problems 94 Scrap Flows97 & Dual Prices for 1980 and 1990 PATTERNS OF ELECTRICITY CONSUMPTION 103 NATURAL GAS CONSUMPTION PATTERNS 10 9 PIG IRON PRODUCTION AND FLOW PATTERNS 11 3 Pig Iron Production and Dual Activities for 1980 11 3 Pig Iron Dual Activities for 1980 11 4 Pig Iron Flows for 1980 11 6 Pig Iron Production Changes Projected to 1990 11 8 Changes in Pig Iron Dual Activities from 1980 to 1990 12 0

-i-

NCSEER March 8 3

Changes in Flow Patterns from 1980 to 1990 12 1 STEEL PRODUCTION AND FLOW PATTERNS 12 8 Introduction 12 8 Steel Production for 1980 12 9 Steel Dual Activities for 1980 13 3 Steel Distribution Patterns for 1980 13 5 Changes in the Geography of Steel to 1990 13 6 Changes in Dual Activities for Steel to 1990 13 8 Technological Shifts between 1980 and 1990 14 5 PRESENTATION OF OBJECTIVE FUNCTION COSTS 14 9

SUMMARY AND CONCLUSIONS 15 1

FOO TNOT ES 15 6

APPENDIX 1 160

APPENDIX 2 16 8

Table s

Table 1 . TRANSPORTATION COST DATA 1 6 Table 2 . COKING COAL PRODUCTION IN 1000 TONS 1 8 Table 3 . COKE MANUFACTURE IN 1000 TONS 2 0 Table 4 . IRON ORE PRODUCTION IN 1000 TONS 2 1 Table 5 . IRON ORE ENRICHMENT PROCESS 2 3 Table 6 . SCRAP SOURCES IN 1000 TONS 2 5 Table 7 . REGIONAL ELECTRICITY PRICES 2 6 Table 8 . REGIONAL NATURAL GAS PRICES 26 Table 9 . DATA FOR PIG IRON PRODUCTION FOR 1970 2 8 Table 10 . DATA FOR PIG IRON PRODUCTION FOR 1980 29 Table 11 . DATA FOR PIG IRON PRODUCTION FOR 1990 3 0 Table 12 . STEEL PRODUCTION DATA FOR 1970 3 4 Table 13 . STEEL PRODUCTION DATA FOR 1980 3 7 Table 14 . STEEL PRODUCTION DATA FOR 1990 40 Table 15 . GENERAL STEEL PROCESS COEFFICIENTS TABLE 44 Table 16 . STEEL PRODUCTION COSTS BY CAPACITY AND PROCESS 4 5 Table 17 . SYSTEM CONSTRAINTS BY COMMODITY BY YEAR 47 Table 18 . COKING COAL EXPORTS 4 9 Table 19 . IRON ORE EXPORTS IN 1000 TONS 49 Table 20 . ESTIMATED PIG IRON DEMAND BY REGION BY YEAR IN 1000 TONS 5 1 Table 21 . ESTIMATED STEEL DEMAND BY REGION BY YEAR IN 1000 TONS 5 3 Table 22 . MODELLED OPTIMAL COKING COAL PRODUCTION IN 1000 TONS 60 Table 23 . COKING COAL FLOWS IN 1000 TONS 6 1 Table 24 . MODELLED OPTIMAL COKE PRODUCTION IN 1000 TONS Table 25 . MODELLED OPTIMAL COKE FLOWS IN 1000 TONS 7 0 Table 26 . OPTIMAL IRON ORE PRODUCTION BY YEAR IN 1000 TONS 7 9 Table 27 . OPTIMAL IRON ORE ENRICHMENT BY YEAR IN 1000 IONS 8 1 Table 28 . CHANGES IN LOW-GRADE AND ENRICHED ORE IN 1000 TONS 84 Table 29 . MODELLED OPTIMAL IRON ORE FLOWS BY YEAR IN 1000 TONS 8 7 Table 30 . MODELLED OPTIMAL SCRAP IRON FLOWS BY YEAR IN 1000 TONS 98 Table 31 . OPTIMAL SCRAP DUAL ACTIVITIES IN RUBLES PER TON 10 3 Table 32, MODELLED OPTIMAL ELECTRICITY CONSUMPTION BY NODE BY YEAR—10 5 Table 33 . MODELLED OPTIMAL NATURAL GAS CONSUMPTION BY NODE BY YEAR, 11 0 Table 34 . MODELLED OPTIMAL PIG IRON PRODUCTION BY YEAR IN 1000 TONS 11 5 Table 35 . MODELLED OPTIMAL PIG IRON FLOWS BY YEAR IN 1000 TONS 122

NCSEER March 8 3

Tables Cont .:

Table 36 . MODELLED OPTIMAL STEEL PRODUCTION BY YEAR IN 1000 TONS 13 9 Table 37 . MODELLED OPTIMAL STEEE FLOWS BY YEAR IN 1000 TONS 14 2 Table 38 . OBJECTIVE FUNCTIONS AND SYSTEM COST DATA 150

Figure s

Figure 1 . FLOW DIAGRAM FOR IRON AND STEEL MODEL 9

Map s

MAP 1 . COKING COAL FLOWS FOR 1980 6 4 MAP 2 . COKING COAL FLOWS FOR 1990 65 MAP 3 . COKE FLOWS FOR 1980 7 4 MAP 4 . COKE FLOWS FOR 1990 7 5 MAP 5 . IRON ORE FLOWS FOR 1980 9 1 MAP 6 . IRON ORE FLOWS FOR 1990 9 2 MAP 7 . SCRAP METAL FLOWS FOR 1980 10 1 MAP 8 . SCRAP METAL FLOWS FOR 1990 10 2 MAP 9 . ELECTRICITY CONSUMPTION IN 1980 10 6 MAP 10 . ELECTRICITY CONSUMPTION IN 1990 10 7 MAP 11 . GAS CONSUMPTION IN 1980 11 1 MAP 12 . GAS CONSUMPTION IN 1990 11 2 MAP 13 . PIG IRON FLOWS FOR 1980 12 6 MAP 14 . PIG IRON FLOWS FOR 1990 12 7 MAP 15 . STEEL FLOWS TO FINAL DEMAND 1980 14 6 MAP 16 . STEEL FLOWS FOR 1990 14 7

--iii-

NCSEER March 83

INTRODUCTION BACKGROUND FOR THE STUDY :

Since the industrialization debates of the 1920's, Soviet economi c

development strategies can be fairly characterized as strongly favorin g

investment in the producer goods sectors of the economy and consequentl y

as slighting rather severely both the agricultural and consumer-servic e

sectors of the economy . Within this context ferrous metallurgy has lon g

represented the premier success story emanating from this "unbalanced "

development strategy . Hence, whereas the iron and steel industries o f

the industrialized West (with the possible exception of Japanes e

companies) have long since lost much of their romantic aura t o

petro-chemical and more innovative industries such as aerospace ,

electronics, and data processing ; the marriage between the Soviet iro n

and steel industry and the Soviet " Motherland " has remained unshaken, a t

least, until very recently . Stripping away its nationalist glamour, th e

iron and steel industry has in fact long been a most crucial sector o f

the Soviet economy . This situation can be interpreted as arisin g

concomitantly with Soviet autarkic development policies, regardless o f

whether one perceives these economic policies as having been

self-initiated or invoked in response to exogenous political, defense ,

and economic factors [1] .

Over the last fifteen or so years, however, the Soviets hav e

demonstrated some interest in relaxing their autarkic policies in favo r

of greater participation in trade with non-CMEA nations . In essence ,

Western observers discern two primary categories of motivating factor s

for this turn of events : (1) political and (2) economic . With som e

- 1 -

NCSEER March 8 3

qualification one can conceive of Soviet foreign trade wit h

(non-socialist) developing nations as being primarily politicall y

motivated and trade with the industrialized West as being primaril y

economically motivated . More specifically trade with the West ca n

plausibly be interpreted as attempts by the Soviet leadership to redres s

chronic problems of economic efficiency and labor productivity throug h

an infusion of modern technology for the lagging industrial sectors an d

to meet consumer demands for meat and other animal food products wit h

the purchases of wheat and corn, primarily for feed-grain usage .

Although the commodity structure of Soviet foreign trade ha s

evolved substantially over the years, trade statistics indicate tha t

even with CMEA countries the USSR still possesses characteristics mor e

commonly associated with those of a developing nation than a first-orde r

industrial power [2] . Expressed in another way, the recent pattern o f

import trade coupled with the ruble ' s lack of convertibility an d

American restrictions on Ex-Im credits has precipitated a severe balanc e

of payments problem for the Soviet Union . Hence, like a developin g

country the Soviet Union has been forced to export raw materials to the

West (as well as CMEA countries) to earn hard currency to offset he r

rising hard-currency trade deficit . For example, in 1977 fully 8 3

percent of all Soviet hard-currency earnings came from the export of ra w

materials . Petroleum exports dominated these hard-currency earning s

totaling over 50 percent [3] . The ability of the Soviets to maintai n

such petroleum export levels has been the subject of much discussion an d

debate [4] . On the other hand, no one seriously questions the vastnes s

of Soviet iron ore, coking coal, or natural gas reserves [5] . Of thes e

raw materials, the first two, especially iron ore, have long played a

role in Soviet exports, but mainly to CMEA countries [6] . Natural ga s

- 2-

NCSEER March 8 3

exports, including to Western Europe, during the eighties and beyond ar e

likely to increase rapidly, but nonetheless, still be inadequate t o

offset the decline in petroleum hard currency earnings [7] .

Furthermore, other recent research designed to assess the potentia l

impacts of various Soviet resource developments on the world economy ha s

strongly indicated that the future prospects for Soviet exports of bot h

iron ore and coking coal appear increasingly problematic [8] .

NATURE OF THIS INVESTIGATION :

This current research project is based on our earlier efforts a t

modeling optimal commodity flow patterns within Soviet ferrou s

metallurgy [9] . More specifically, within the limitations of the dat a

(1970 data only) and the simple linear programming model used, ou r

earlier research sought to cast some light on the following nin e

questions . (1) Assuming that the Soviets are optimizers, what would th e

commodity flow patterns for 1970 steel production and distribution loo k

like? (2) Assuming that the Soviets place highest priority upo n

producing steel for domestic consumption, which iron-ore reserves an d

coking-coal reserves would be utilized? (3) Are these " optimal " reserv e

locations easily accessible to or fairly remote from current o r

potential foreign markets? (4) What ferrous metallurgy locationa l

changes may be in the offing as a result of the increasingly expensiv e

underground Donbas coking coals compared to the cheaper, but mor e

remote, Kuznetsk coals? (5) What regional steel production changes ma y

be economically warranted as a result of the depletion of both th e

Urals' coking-coal deposits [10] and its iron-ore deposits [11] . (6 )

Similarly, what regional steel production changes might depend on th e

rapid development of the massive, but hydrogeologically problematic, KMA

- 3 -

NCSEER March 8 3

(Kursk Magnetic Anomaly) iron-ore deposits located in southern Europea n

Russia, seemingly well suited to serve both domestic and foreig n

markets? (7) Which, if any, of the new steel producing centers propose d

by the Soviets over the last couple of decades appear to be a t

economically viable locations? (8) Which, if any, of the stee l

producing nodes in 1970 are "uneconomical " within the context of a

systems analysis? In other words, does such an analysis suggest tha t

there are strong economic arguments for expanding the output of som e

steel complexes while reducing or even dismantling others {12] . (9 )

Finally, in what ways may these various model-indicated geographica l

changes be relevant to the future prospects of Soviet exports of iro n

ore, coking coal, pig iron, and steel .

In additional to being able to address all these questions for th e

year 1980 and of projecting them forward to 1990, this curren t

investigation incorporates a large number of significant extensions .

The chief purpose of this study has been to construct both a mor e

comprehensive time-series data base and a more realistic multi-proces s

analytical model, which could provide the capability of addressing a

larger number of quasi-dynamic resource policy issues being faced by th e

USSR and either directly or indirectly by her current and potentia l

trading partners through 1990 . Specifically, a reformulated systems

model has been designed to either incorporate and/or investigate th e

following eight broadly expressed Soviet natural resource managemen t

problems, questions, and issues :

(1) forecasts of the domestic resource base for the Soviet iron an d steel industry to 1990, including direct-shipping and enriched iro n ore, coking coal, coke, scrap, natural gas, and electricity ;

(2) forecasts of the Soviet export potential for iron ore an d coking coal to 1990 ;

(3) forecasts of the Soviet export potential for pig iron an d

-4-

NCSEER March 8 3

finished steel to 1990 ;

(4) forecasts of Soviet domestic iron and steel demands to 1990 ;

(5) impact of planned transport developments, such as the BAN an d other new transport linkages, on regional commodity flow patterns ;

(6) a modest assessment of the impact of new technologies , especially of direct reduction steel making, and of othe r technological changes on the spatial-temporal development of ra w material sites and on the siting of new and/or expanded capacitie s of iron and steel production ;

(7) an implicit series of checks on the internal consiste nc y of relevant Soviet production and technological coefficient data ; an d

(8) most importantly, an assessment of model-indicated "optimal " energy consumption patterns within Soviet ferrous metallurg y arising partially as a result of projected relative energy pric e changes both by fuel type and region .

Keeping these specific research questions in mind, the basic objectiv e

of this investigation is to determine what production and transportatio n

inputs will most economically produce and transport steel to supply th e

markets of the Soviet Union and her steel trading partners .

The analysis is multi-stage, that is it covers all stages fro m

production and transportation of coking coal, iron ore, enrichment o f

iron ore, supply of natural gas and electricity, production of pig iron ,

supply of scrap, production of steel (by several methods), and transpor t

of steel to the regional markets as actually distributed over the vas t

territory of the Soviet Union . Since the Soviet Union covers such a

large area (nearly three times that of the contiguous United States) ,

the specific location of resources, production facilities, and markets ,

and the transport costs of joining them, becomes particularil y

important . To make this analysis of production at each site, an d

possible changes, a linear programming model, specifically the Sovie t

Iron and Steel Energy Model (SISEM), was developed . This model i s

discussed more fully in Section 2, " The SISEM Model, " and the complet e

formulas utilized in this model are listed in Appendix 1 and Appendix 2 .

-5—

NCSEER March 8 3

Appendix 1 on pages 160-167 represents the formulas for the overal l

summation of all costs from production of raw materials at specifi c

points, transport to production facilities, processing costs at existin g

or planned mills, and transport to markets throughout all regions of th e

Soviet Union including a number of border locations used as surrogate s

for export demands . Subject to the approximately 1,200 constrain t

equations for model year 1980 (approximately 1,300 for 1990), the mode l

calculates different combinations of the approximately 18,000 choic e

variables in order to arrive at the least-cost combination . Appendix 2

on pages 168-169 represents the formulas for various sub-system costs .

The location, characteristics, costs, and capacities of all know n

facilities of each stage and component of the Soviet iron and stee l

industry (sometimes grouped for statistical processing) are tabulate d

and discussed in Section 3, " Data, Limitation, and Assumptions " on page s

14-54 . These data are necessary to supply the ingredients of the linea r

programming model, which then calculated the least-cost combination .

These data cover each production node for coking coal, coke production ,

iron ore and its enrichment, 'scrap metal, energy costs for electricit y

and natural gas, pig iron production, steel production, and distributio n

of both pig iron and finished steel to the regional domestic markets o f

the Soviet Union and surrogate export nodes .

The conclusions to be drawn as a result of the processing of thi s

vast body of data through the linear programming model are reported i n

Section 4, " Discussion of the Results, " on pages 54-151 . Thi s

discussion encompasses each of the stages of production and transpor t

from coking coal and iron ore, through iron and steel production, t o

distribution patterns of the finished steel both in 1980 and projected

patterns in 1990 .

-6-

NCSEER March 8 3

With this outline of the study in mind it is time to turn to a

fuller discussion of the mathematical programming model used in thi s

analysis and both its capabilities and limitations .

THE SISEM MODE L

THE OBJECTIVE FUNCTION :

The overall mathematical structure of the Soviet Iron and Stee l

Energy Model (SISEM) is presented in Appendix 1 . SISEM is a linea r

programming model designed to determine the optimal spatial allocatio n

of energy inputs (coal, natural gas and electricity) and raw materia l

inputs for the Soviet ferrous metallurgical industry . The model ha s

been designed to incorporate both the direct and indirect energy input s

required . The direct energy inputs are calculated as the energy neede d

to produce pig iron and steel . These inputs include coke, natural ga s

and electricity . At the same time the indirect energy inputs in th e

form of coal used to make coke and the electricity used in th e

production of enriched iron ore are also calculated .

The model is solved so as to minimize the total cost functio n

(variable Z in Appendix 1) . Accordingly, the model calculates the tota l

sum of extraction, processing, raw material (including energy), capital ,

and transport costs for the entire production system . The sum of th e

costs of the (1) optimal energy inputs by type, quantity, and location ;

(2) sources and quantities of direct shipping and enriched iron ore an d

scrap ; (3) the production activity levels of pig iron and steel plants ;

and (4) the transportation of these inputs and outputs represent th e

attainable least cost given the assumptions and data used to represen t

the system (refer to Appendix 1) .

Two types of decision variables are used to designate the differen t

-7-

NCSEER Marc h 83

choices made within the model . First, the flow decision variables hav e

the general structure Xorigin, destination representing the tons ,

kilowatt hours, or cubic meters transported from an origin node to a

destination node . Second, the Xn Ek process variables represent th e

n th type steel process selected at each of the i steel producin g

nodes .

The schematic model flow diagram (Figure 1) on the following pag e

does include all potential, but not detailed actual, flow lines linkin g

the various energy and raw material supplies from columns i and j to th e

various specific pig iron and steel processes shown under columns k an d

i, respectively . Whether or not a given flow will occur, of course ,

depends first of all on the different "input coefficients " which ar e

discussed in the ensuing data section of this report, and second of al l

on whether Dr not a given potential internodal flow enters the optima l

feasible model solution as a non-zero, basic variable . In other words ,

on whether or not a potential flow variable has a positive value in th e

optimal solution .

The steel supply nodes (column m in Figure 1) were introduced t o

reduce the number of potential final demand steel flows by a factor o f

approximately four . All quantities of steel produced by all n processe s

at a given mill i are "shipped " at zero cost to node m having the sam e

node number as the k th mill . Or more simply stated each mill k ha s

a corresponding node m to which all of is steel production i s

assigned . Thus, instead of having a total number of potential stee l

flows equal to the sum of all n's for all k ' s times d (the number o f

steel demand nodes) ; the number of potential steel flows is reduced to m

(i .e ., &) times d .

The following verbal description represents an example of possibl e

-8-

NCSEER March 8 3

commodity flow patterns calculated by the model (refer to Figure 1) .

Scrap metal from the i th scrap supply node is shipped to both th e

k th pig iron plant and the electric arc steel furnaces at th e

kth steel mill . Natural gas is shipped to the k th pig plant ,

but none is required by the electric steel furnaces at Z . Some of th e

pig iron production from the k th plant is shipped to the ar c

furnaces at steel mill Z and the rest is transported to the d th pi g

iron final demand node and to the e th export node . All of the iro n

ore produced at the i th low-grade ore site is enriched nearby at th e

jth enrichment complex . This enriched ore is then shipped partly t o

the k th pig iron plant, partly to the electric furnaces at Z, an d

the remainder is exported through node e . Electricity is consumed b y

the enrichment process at j, pig iron production at k, and stee l

production at k . Coking coal production at the i th site is shippe d

to the j th coking battery . The resulting coke is shipped primaril y

to the pig iron plant at node k and the remainder to an open heart h

furnace at different steel mill than the Zth one under discussio n

here . An alternative coking coal site and coke battery are calculate d

by the model as yielding system-wide production cost savings ove r

ith mine and j th coke battery just cited . Hence, they functio n

as the supplier of the modest amount of coke needed for the electri c

steel smelting at the Zth mill . Finally, all of the electric stee l

produced at Z is shipped to the d th final demand node . A myriad o f

other potential commodity and energy flow patterns could be elaborated .

MODEL CONSTRAINTS :

The constraints of SISEM (see Appendix 1) can be classified int o

six types :

-10-

NCSEER March 8 3

(1) coke plant mass balances ,

(2) iron ore enrichment plant mass balances ,

(3) pig iron plant mass balances ,

(4) steel plant mass balances ,

(5) final demands an d

(6) limits .

The coke plant mass balnce (Equation 1) ensures that the requisit e

quantity of coal is shipped to coking plants to satisfy the system-wid e

coke demand .

The iron ore enrichment plant mass balances (Equations 2 and 3 )

ensure that the required quantities of iron ore and electricity reac h

the enriching plants to satisfy the system-wide demand for enriched ore .

The pig iron plant mass balances (Equations 4-8) replicate th e

possible processes in the production of pig iron and make sure that th e

required quantities of coke, natural gas, electricity and iron or e

(either high-grade and/or enriched) are assembled at the pig iron plant s

to satisfy the system-wide final and intermediate demand for pig iron .

Steel plant mass balances (Equations 9-15) guarantee that th e

necessary quantities of the inputs needed for the n possible processe s

at each steel plant are assembled so that the system-wide final stee l

demand is satisfied . Each of the n processes requires different inpu t

combinations of iron ore, pig iron, scrap, electricity, natural gas an d

coke . Consequently, there is a different set of average input mas s

balance equations associated with each process . Furthermore, a s

discussed in the data section which follows, where we were able t o

obtain process and site-specific input coefficients, they were use d

instead of the average input coefficients by process (see Table 15, pag e

44,for the average coefficient values) . Additionally, it is assumed tha t

-11—

NCSEER March 8 3

each process produces a homogenous output, steel, and the amount o f

steel produced at all the plants can (must) satisfy the the system-wid e

steel demand .

The final demand constraints (Equations 16-19) require that th e

system-wide final demands for pig iron--both domestic and export, coa l

exports, iron ore exports and final steel demand--both domestic an d

export are satisfied . Since the model is demand driven, thes e

constraints assure that the inputs will flow from origins t o

destinations .

The final series of constraints (Equations 20-36) set limits fo r

the model . For example, equations 20-33 put upper and lower limits o n

the quantities of inputs, intermediate products or final products tha t

can flow from any origin node . Thus, all of these constraints ar e

site-specific . For example, no more coal can flow from a given site i n

model year 1980 than was actually produced there in 1980 . Similarly ,

the sum of the steel produced by all processes at a given mill can no t

exceed the upper limit on steel production at that given mill . Because ,

as will be mentioned in the data section, neither gas pipeline nor th e

electrical grid networks were modelled, only system-wide upper limit s

(Equations 34-35) on natural gas and electricity consumption wer e

utilized . Finally, system-wide upper and lower limits on stee l

production by process were incorporated (Equations 36-37) . Thes e

various limit quantities are derived from actual data for the 1970 an d

1980 runs and are estimated for the 1990 run .

DESCRIPTION OF SUB-SYSTEM OBJECTIVES :

The term "Z" measures the total cost of production inputs an d

transport inputs required to produce and transport Soviet iron and stee l

-12-

NCSEER March 8 3

(see Appendix 1, pp . 160-167) . Besides total cost, the objectiv e

function has been decomposed into 7 arguments each representin g

sub-system cost functions . The mathematical formulas for these 7 cos t

functions are enumerated in Appendix 2, pages 168-169 and are verball y

defined below :

1) COAL, measures the total cost of coal for both domestic iron an d steel use and export requirements .

2) ELECTRICITY, measures the total cost of electricity used in th e various stages of the Soviet iron and steel. industry

3) NATURAL GAS, measures the total cost of natural gas used in th e various stages of the Soviet iron and steel industry .

4) TOTAL ENERGY, measures the total cost of energy used in th e various stages of the Soviet iron and steel industry (exclude s cost of exported coking coal) .

5) NON-ENERGY, measures the total non-energy input and proces s costs in the Soviet iron and steel industry .

6) TRANSPORT (Domestic), measures the total " domestic" transpor t costs of the various inputs at all intermediate stages an d outputs of the Soviet iron and steel industry .

7) TRANSPORT (Export), measures the total " export " transport cost s incurred in order to ship the system ' s five export commoditie s to their respective " export nodes . "

SOLUTION PROCEDURE :

The linear form for the relationships developed in SISEM has a

number of advantages . Most importantly, all functions of SISEM ar e

continuous (non-integer) linear variables allowing for the linea r

programmming representation of the system to be solved with standar d

solution packages such as the IBM MPSX software . The current version o f

the system has been solved on the IPM 370/168 Computer at Bosto n

University . Second, the systems approach implicit in the model allow s

for the simultaneous determination of the optimal production activity b y

point (and process) and point-to-point flow values within the context o f

what is a highly geographically disaggregated, multi-stage ,

-13-

NCSEER March 8 3

multi-process, industry . As will be discussed in the next section, b y

incorporating capacity differentiated production cost functions, a n

attempt has even been made to embed production scale economies, an d

hence, finesse around the linear limitation on productin cost functions .

Obviously, a non-linear, dynamic model would be more ideal than th e

linear, static one used here . Unfortunately, we are unaware of an y

readily available non-linear software which could find stable solution s

to such a large mathematical programming problem as formulated by th e

SISEM data base .

Finally, the linear form facilitates performing post-optimalit y

sensitivity analysis for the solution . Since there is some degree o f

uncertainity surrounding many of the data inputs to the model, the rang e

of values that various parameters in the system can take on and stil l

keep the current solution optimal can be derived . This is also a n

option available on most standard linear programming software packages .

The SISEM program, when set up with the basic data described in th e

immediately following data section has approximately 1,200 (1,300 fo r

1990) functional constraints and 18,000 variables . The linear progra m

solves in approximately 7 to 11 minutes of CPU time .

This overview of the SISEM model should be sufficient for a n

understanding of the motivations behind the choice and collection of th e

pertinent raw data .

DATA, LIMITATIONS , an d ASSUMPTION

Several sets and types of raw data have been assembled for thi s

project . These data can be subsumed under six categories : (1 )

cartographic data and transport networks, (2) transport cost data, (3 )

raw material production and costs and intermediate raw material proces s

production and costs, (4) pig iron production and costs, (5) stee l

- 1 4 -

NCSEER March 8 3

production, processes and costs, and (6) final demands for iron ore ,

coking coal, pig iron, and steel .

CARTOGRAPHIC DATA AND TRANSPORT NETWORKS :

One of the earliest tasks of the project was to digitize th e

U .S .S .R . allowing for hierarchical separations at the national ,

republic, and oblast levels . Concomitantly, the Soviet railway system ,

inland and coastal waterways and selected pertinent surface roads wer e

encoded into a transportation system consisting ultimately of 984 node s

and their internodal distances [13] . Each node was given a n

identification number . These numbers of each place are used on the map s

and are listed along with their place names on the appropriate tables .

Transport links completed between 1970 and 1980 were added for the 198 0

model run and those projected for completion between 1980 and 1990 wer e

added for the 1990 computer run . These latter new developments include d

primarily some new linkages in the Urals, the Ukraine, Kazakhstan, th e

European North, West Siberia, and most importantly, the HAM . A

comparison between the 1980 and 1990 model runs, thus is potentiall y

able to shed some modest light on the hotly debated issue of th e

economic effectiveness of the Baikal-Amur-Mainline .

All of these transport nodes were then also digitized as part o f

the cartographic data base providing the capability of producin g

computer maps displaying different classes of nodal production data b y

proportionally sized symbols and three classses of linear node-to-nod e

commodity flows (see Maps 1 through 12) [14] .

TRANSPORT COST DATA :

Some modest attempt was made to incorporate both commodity an d

- 1 5-

NCSEER March 8 3

modal differentiated transport costs . Table 1 contains the ra w

transport cost data used in the analyses . Rather than using all six o f

these possible rates (two commodity classes times three modes) directly ,

they were approximated by first multiplying all water internoda l

distances by 0 .51 [15] and all surface road lin k

------

Table 1 . TRANSPORTATION COST DAT A

COMMODITY TRANSIT MODE COS T

IRON ORE, COAL, COKE RAILROAD 1 .75 RUBLES/1000TON-KM

SCRAP, PIG IRON, STEEL RAILROAD 2 .38 RUBLES/1000TON-KM

IRON ORE, COAL, COKE WATER 0 .90 RUBLES/1000TON-KM

SCRAP, PIG IRON, STEEL WATER 1 .20 RUBLES/1000TON-KM

IRON ORE, COAL, COKE SURFACE ROAD 3 .50 RUBLES/1000TON-KM

SCRAP, PIG IRON, STEEL SURFACE ROAD 4 .75 RUBLES/1000TON-K M

Sources : Transport i svyaz' SSSR (Moskva : Izdatel ' stvo Transport, 197 2 to 1979), various pages ; Kantorovich, L . V ., " Concerning Prices, Rates , and Economic Effectiveness, " Problems of Economics 19 :4-5-6 (August , Sept ., Oct ., 1976), pp . 214-217 .

distances by 2 .0 [16] and then multiplying all of the modified an d

unmodified (i .e ., rail links) internodal distances by 1 .75, the rai l

transport rate for ore, coal, and coke in rubles per 1000ton-km . These

resulting transformed data were then processed by a minimum distanc e

algorithm to yield a minimum-cost-distance matrix (in rubles/1000ton-km )

connecting all nodal pairs potentially having iron ore, coking coal, o r

coal flows between them . An identical procedure was used to produce a

transportation cost matrix for all potential flows of scrap, pig iro n

and steel using a multiplier of 2 .38 (rubles/1000 ton-km), the rail rat e

-16-

NCSEER March 8 3

for scrap, pig iron, and steel .

In addition to updating the Soviet transport network encoded durin g

previous research [17], this study extends our earlier work by includin g

commodity and modal differentiated transport rates plus actual an d

potential export nodes in the transport network . On the other hand, du e

to budget, data, and computational difficulties, attempts to explicitl y

model both the Soviet natural gas pipeline network and th e

electrical-grid system had to be abandoned . For similar reasons n o

attempts were made to embed transshipment or back-haul costs into th e

models .

RAW MATERIALS AND INTERMEDIATE PROCESSES :

The third major category of data encompasses raw materia l

production and cost data for three years 1970, 1980, and projections t o

1990 . These data include coking coal production and extractions costs ,

coke production and costs, iron-ore production and costs, ore-enrichmen t

and costs, scrap metal and costs, electricity costs, and natural ga s

costs .

Coking Coal :

A variety of sources were consulted to compile production and cos t

data for Soviet coking coal production aggregrated into twelve domesti c

sites plus a surrograte import node, Grodno, for imported Polish cokin g

coal for 1970 and 1980 . The Neryungri deposit along the route of th e

BAN was added for 1990 . Maximum and minimum production ranges were use d

to bracket production as accurately as possible to actual productio n

levels for 1970 and 1980, while for 1990 mimimum production levels wer e

reduced to 1 .0-million tons for eight coalfields to allow the model mor e

- 1 7 -

NCSEER March 8 3

flexibility in determining optimal patterns . Because not all of Soviet cok e

production is consumed by the iron and steel industry (e .g ., coke is used in

non-ferrous metallurgical smelting) and the 1970 minimum coking coa l

production constraints forced surplus coking coal output in the Ukraine i n

the 1970 preliminary computer run, the minimum production values for th e

Kamensk-Shakhtinskiy colliery was set at zero and that of th e Stakhanov

mining area was vastly reduced for the 1980 computer run .

Several sources were consulted to try to project realistically bot h

production and cost data for 1990 . For all three years some regional dat a

aggregation was used, especially for the southeastern Ukraine . Of note i s

the increasing relative cost and stagnating output of the underground mine d

Donetsk Basin coals relative to other deposits, especially those of th e

Karaganda and Kuzetsk Basins [18] . These data are listed in Table 2 .

------Table 2 . COKING COAL PRODUCTION IN 1000 TONS *

N PRODUCTION COST PRODUCTION COST PROJECTED COS T 0 PRODUCTIO N D 1970 RUB ./ 1980 RUB ./ 1990 RUB . / PLACE : E MIN MAX 1000T MIN MAX 1000T MIN MAX 1000T

VORKUTA 399 8966 8967 15500 12000 14615 17000 1000 15535 1800 0 ' 526 1450 1450 10000 1150 1518 12000 0 1518 1500 0 TKVARCHELI 209 250 325 13000 250 325 15000 250 325 1600 0 TKIBULI 215 250 325 17000 250 325 20000 250 325 2100 0 KARAGANDA 668 9480 9481 10900 10660 11220 13600 1000 14025 1500 0 KEMEROVO 739 6512 9768 9700 9768 14650 12000 1000 20350 1400 0 NOVOKUZNETSK 726 26050 30932 9700 30932 40700 12000 1000 44770 1400 0 PAVLOGRAD 813 3068 3069 15400 2655 3075 20000 1000 3075 2800 0 KAMENSK- 930 2516 2517 15400 0 2097 20000 1000 2450 2800 0 SHAKHTINSKI Y STAKHANOV 885 15678 15679 15400 1194 17825 20000 1000 17825 2800 0 DONETSK 284 34000 37563 15400 34950 38445 20000 1000 38445 2800 0 NERYUNGRI 963 NA NA NA NA NA NA 1000 9000 2000 0 URGAL 779 75 100 12000 75 150 12000 100 1500 1800 0 GRODNO(IMPT) 81 674 674 20400 600 1000 47800 600 4000 5000 0

*Assumes cleaning and concentrating at the minehead, thus coal productio n figures are for coal concentrate .

- 1 8-

NCSEER March 8 3

Table 2 . cont .

SOURCES : Leslie Dienes and Theodore Shabad, The Soviet Energy System (Washington, D .C . : V . H . Winston & Sons, 1979), pp . 119-124 . ; Theodor e Shabad, " News Notes," Soviet Geography, various issues ; Koks i khimiya , various issues ; Ugol ', various issues ; Narodnoye khozyayistvo SSSR , various issues ; V . A . Shelest, Regional'nyye energoekonomicheskiy e problemy SSSR (Moscow : Nedra,1975), pp . 112-131 ; USSR : Coal Industr y Problems and Prospects (Washington, D .C . : C .I .A ., 1980), pp . 1-18 ; Allan Rodgers, " Coking Coal Supply : Its Role in the Expansion of th e Soviet Steel Industry , " Economic Geography 40 :2 (April 1964), pp . 113-150 ; Robert W . Campbell, Basic Data on Soviet Energy Branches (Santa Monica : Rand Corporation, 1979), pp . 39-59 ; Robert Campbell, Soviet Energ y Balances (Santa Monica : Rand Corporation, 1978), pp . 7-9, 15-16, 40-42 , 83-89, 99-100 .

Coke Manufacturing Data :

Similarly a variety of sources were consulted (see Sources to Tabl e

3) to construct production ranges for Soviet coke production . A tota l

of twenty-two sites were used for 1970 and 1980, while two potentia l

coking locations at Chul ' man and Urgal were added for the 199 0

simulation run . Again, some aggregration of regional coke manufactur e

was done, but only in the Donetsk region . Because the coking coa l

production data we compiled incorporated both deposit specific cokin g

coal cleaning and concentrating factors, a uniform coefficient of 1 .5 1

was utilized to convert coking coal into coke . Since we were unable t o

locate any comprehensive regional or site specific coking process costs ,

the average cost (late 1970 ' s) of 9 .55 rubles per ton of coke was use d

for all coking sites for all three years .

Iron Ore Production Data :

Some modest data aggregation was also used when compiling the iro n

ore production data listed in Table 4 . For 1970 iron ore production

range data were attributed to a total of seventeen low-grade ore site s

and twelve high-grade or direct-shipping ore sites . For 1980, the new

- 1 9-

NCSEER March 8 3

------Table 3 . COKE MANUFACTURE IN 1000 TONS * ------PRODUCTION PRODUCTION PRODUCTIO N 1970 1980 199 0 PLACE : NODE : MIN MAX MIN MAX MIN MAX

LENINGRAD 24 100 500 100 500 100 50 0 CHEREPOVETS 836 6000 7000 6600 8000 6600 12000 MOSKVA 415 500 1000 500 2000 500 2000 KRIVOY ROG 172 1000 10000 6000 12000 8000 15000 DNEPROPETROVSK 811 200 5000 400 8000 400 1000 0 ZAPOROZH ' YE 179 200 5000 2000 6000 2000 750 0 YASINOVATAYA 291 1000 6400 7000 42500 7400 45000 KOMMUNARSK 884 200 4000 200 15000 200 1500 0 ZHDANOV 286 1000 10000 1000 10000 1000 1000 0 NOVOLIPETSK 995 1000 7200 1000 10000 1000 10000 KALININGRAD 59 100 500 100 1000 100 100 0 KHAR'KOV 298 100 1000 100 1000 100 100 0 526 960 961 775 1006 1000 100 0 NIZHNIY TAGIL 531 1000 6000 1000 10000 1000 10000 SVERDLOVSK 548 1000 6000 1000 10000 1000 10000 CHELYABINSK 581 1000 6000 1000 10000 1000 10000 MAGNITOGORSK 586 1000 10000 1000 15000 1000 10000 NOVOTROITSK 594 1000 6000 1000 10000 1000 10000 RUSTAVI 910 300 700 300 700 300 2000 KARAGANDA 668 2800 2800 3500 5000 3500 600 0 KEMEROVO 739 1000 2400 1400 3000 1400 300 0 NOVOKUZNETSK 726 1000 6000 9600 12000 5000 12000 CHUL'MAN 963 0 0 0 0 0 400 0 URGAL 779 0 0 0 0 0 400 0 ------*NOTE : A coefficient of 1 .51 was used to convert coking coal to coke . Constant price of coking process assumed to be 9 .55 rubles/ton .

SOURCES : Theodore Shabad, " News Notes, " Soviet Geography, variou s issues ; Koks i khimiya , 1978, No . 6, pp . 10-11 ; Leslie Dienes & Theodore Shabad, The Soviet Energy System (Washington, D .C . : V . H . Winstonugley & Sons, 1979), pp . 123-124 ; Effektivnost ' ispol ' zovaniy a (Moscow : Nedra, 1976), P . 100 ; Ugol , , various issues ; USSR : Coa l Industry Problems and Prospects (WASHINGTON, D .C . : C .I .A ., 1890), pp . 1-18 ; Narodnoye khozyaystvo SSSR , various issues ; Allan Rodgers , "Coking Coal Supply : Its Role in the Expansion of the Soviet Stee l Industry," Economic Geography 40 :2 (April 1964), pp . 113-150 ; Theodor e Shabad, Basic Industrial Resources of the U .S .S .R . (New York : Columbi a University Press, 1969), various pages ; Koks i khimiya, variou s issues ; Stal ' , various issues ; Robert W . Campbell, Basic Data o n Soviet Energy Branches (Santa Monica : Rand Corporation, 1979), pp . 39-59 ; Robert Campbell, Soviet Energy Balances (Santa Monica : Ran d Corporation, 1978), pp . 7-9, 15-16, 40-42, 83-89, 99-100 .

- 2 0-

March 8 : NCSEER

Table 4 . IRON ORE PRODUCTION IN 1000 TON S

N T* PRODUCTION COST PRODUCTION COST PRODUCTION COST q Y D P 1970 RUB ./ 1980 RUB ./ 1990 RUB . / PLACE : E E MIN MAX 1000T MIN MAX 1000T MIN MAX 1000 T 911 L 2300 2700 5000 0 7000 5700 1000 10000 6480 GOROLBAGODAT 530 L 8400 10300 4750 7000 8400 5400 1000 10000 648 0 ALAPAYEVO 541 L 400 500 4750 200 500 5400 100 5000 648 0 BAKAL 577 L 5000 5000 4750 4000 5890 5400 1000 6000 648 0 NOVORUDNAYA 595 L 2200 2200 4750 1000 3000 5400 1000 3000 648 0 TAYEZHNOYE 964 L 0 0 0 0 0 0 0 7000 5000 GARIN 774 L 0 0 0 0 0 0 0 2000 6250 UDA-SELEMDZHA 961 L 0 0 0 0 0 0 0 2000 6250 OLENOGORSK 370 L 8000 8000 3000 9500 12500 2885 1000 18000 3280 KOSTAMUKSA 361 L 0 0 0 0 0 0 325 8900 3400 PUDOZHGORSK 381 L 0 0 0 0 0 0 0 3000 3400 ZHELEZNOGORSK 976 L 300 1500 2400 5000 8000 2995 1000 31000 340 0 GUBKIN 891 L 3000 6000 2400 14000 16000 2995 1000 38000 3400 KREMENCHUG 302 L 2600 2600 2400 8500 11000 2580 1000 20000 297 0 KRIVOY ROG 172 L 63000 68000 2400 60000 61500 2580 1000 80000 297 0 KERCH' 187 L 4700 4700 2400 4700 5000 2580 1000 7500 297 0 DASHKESAN 243 L 1400 1400 3400 0 1100 3320 1000 1500 382 0 RUDNYY 873 L 12000 12100 3650 16650 20000 6080 1000 21000 730 0 KORSHUNOVSK 756 L 5000 5000 3800 6000 6400 4360 1000 8000 545 0 RUDNOGORSK 921 L 0 0 0 0 0 0 0 3000 545 0 ABAZA 730 L 1000 1400 3310 1000 1500 3800 1000 1500 4560 GORNAYA SHORIYA 727 L 3500 5500 3310 4000 8000 3800 1000 8000 456 0 TEYA 978 L 1400 1400 3400 1400 2000 4000 1000 2000 5000 KRASNOKAMENSK 979 L 0 0 0 500 1000 4000 500 2000 5000 IRBA 980 L 0 0 0 0 500 4000 500 2000 5000 ABAZA 729 H 0 1400 3310 0 1500 3800 0 1500 456 0 KRIVOY ROG 173 H 35000 40000 2400 45000 46000 2400 1000 40000 297 0 GUBKIN 460 H 6000 7000 2400 9000 12000 2400 1000 12000 340 0 ZHELEZNOGORSK 977 H 4500 5000 2400 8000 9000 2400 1000 9000 340 0 DNEPRORUDNOYE 180 H 100 100 2400 0 3000 2580 1000 5000 297 0 MAGNITOGORSK 586 H 9000 9000 1000 3000 4000 1110 0 4000 132 0 RAC-JAR 683 H 0 0 0 0 0 0 400 3000 816 0 TULA 433 H 200 980 2400 0 300 3690 0 50 4240 LIPETSK 955 H 200 1000 2400 0 300 3690 0 50 4240 RUDNYY 682 H 3000 3100 3650 3000 3500 6080 1000 3500 730 0 PESCHANSK 533 H 200 540 4750 200 540 5400 100 540 648 0 KARAZHAL 666 H 3000 3000 3650 0 4000 6080 1000 6000 730 0 TASHTAGOL 725 H 0 1500 3310 0 2000 3800 0 2000 456 0 *L= LOW-GRADE ORE, H= HIGH-GRADE OR E SOURCES : Theodore Shabad, Basic Industrial Resources of the USSR (New York Columbia University Press, 1969), p . 36 ; Theodore Shabad, "News Notes, " Soviet Geography , various issues ; Narodnoye khozyaystvo SSSR, variou s issues ; Paul Lydolph, Geography of the U .S .S .R ., 2nd ed . (New York : Joh n Wiley & Sons Inc ., 1970), p . 509 ; Gornyy zhurnal, various issues ; V . P . Novikov, "The Kursk Magnetic Anomaly--A Promising Iron-Ore Base for th e Iron and Steel Industry of the Urals , " Soviet Geography 10 :2 (FEB . 1969), PP . 70-7 1 Tony Misko, & Craig ZumBrunnen, Soviet Iron Ore : Its Domestic an d World Implications, " Soviet Natural Resources in the World Economy (Chicag o University of Chicago Press, forthcoming 1983) .

-21-

NCSEER March 8 3

West Siberian, low-grade, Irba ore deposit was allowed to potentiall y

enter the optimal solution, while the Krasnokamensk low-grade deposi t

was forced to enter the solution with a minimum of a half-million tons .

Irba was constrainted to enter the 1990 solution with a similar minimum

output . Likewise, the low-grade Kostamuksa deposit in Karelia wa s

required to enter with a minimum of 325,000 tons and the long delaye d

high-grade Kachar deposit in northwest Kazakhstan with a minimum o f

400,000 tons . Five additional low-grade ore deposits were potentiall y

allowed to enter the solution for 1990 : Tayezhnoye, Garin an d

Uda-Selemdzha in the Far East ; Rudnogorsk in East Siberia ; an d

Pudozhgorsk in Karelia (see Map 6, p . 92) . Minimum ore production a t

Dashkesan in Azerbaydzhan was inadvertantly set to zero rather tha n

one-million tons for 1980 ; and interestingly enough, entered the basi s

at the zero value . For similar reasons mentioned previously with regar d

to coking coal, the maximum lower range of production for deposits fo r

1990 were reduced to one-million tons . Production ranges reflec t

potential depletion or dwindling of output through time, especially a t

Magnitogorsk, Lipetsk and Tula . Finally, for deposits such as Krivo y

Rog, Rudnyy, and Zheleznogorsk which produce both direct-shipping an d

low-grade ores, output of each type was assigned to a separate transpor t

node for modelling simplicity . In all of these cases, however, the nod e

pairs were (are) in very close proximity .

Iron Ore Enrichment Data :

While high-grade ore may be used directly in iron and stee l

furnaces, low-grade ore must be concentrated and enriched first .

Accordingly, lower and upper bounds on ore benefication at twenty-fiv e

sites for all three years are listed in Table 5 . Because neither sit e

-22-

NCSEER. March 8 3

specific process costs nor electricity requirements were obtained ,

uniform values of 4 .28 rubles/ton and 12 kwh/ton of ore concentrate wer e

employed for all three time periods . Again, the maximum lower limit on

enriched ore output at any given ore concentrator was set at one-millio n

tons per annum for 1990 in order to have a less constrained projecte d

model run .

Table 5 . IRON ORE ENRICHMENT PROCESS*

PRODUCTION PRODUCTION PRODUCTION 1970 1980 199 0 PLACE : NODE : MIN MAX MIN MAX MIN MAX

KACHKANAR 911 2300 2700 0 7000 1000 10000 GOROLBAGODAT 530 8400 10300 7000 8400 1000 840 0 ALAPAYEVO 541 400 500 200 500 100 500 BAKAL 577 5000 5000 4000 5890 1000 600 0 NOVORUDNAYA 595 2200 2200 1000 3000 1000 300 0 TAYEZHNOYE 964 0 0 0 0 0 700 0 GARIN 774 0 0 0 0 0 200 0 UDA-SELEMDZHA 961 0 0 0 0 0 200 0 OLENOGORSK 370 8000 8000 0 11000 1000 20000 KOSTAMUKSA 361 0 0 0 0 325 890 0 PUDOZHGORSK 381 0 0 0 0 0 300 0 ZHELEZNOGORSK 976 600 1100 6000 8000 1000 2200 0 GUBKIN 891 5000 6000 12000 14000 1000 31000 KREMENCHUG 302 2600 2600 8500 11000 1000 2000 0 KRIVOY ROG 172 62000 68000 60000 61500 0 8000 0 KERCH' 187 4700 4700 4700 5000 1000 7500 DASHKESAN 243 1400 1400 0 1100 1100 150 0 RUDNYY 873 12000 12100 16650 20000 1000 21000 KORSHUNOVSK 756 5000 5000 6000 6400 1000 800 0 RUDNOGORSK 921 0 0 0 0 0 300 0 ABAZA 730 1000 1400 1000 1500 1000 150 0 GORNAYA SHORIYA 727 4500 4500 5000 8000 1000 8000 TEYA 978 1400 1400 1400 2000 1000 2000 KRASNOKAMENSK 979 0 0 500 1000 500 2000 IRBA 980 0 0 0 500 500 2000

*NOTE : Cost of enrichment process is assumed to be a constant value o f 42 .80 rubles/ton for all sites for all years . Electricity inpu t coefficient is assumed to be a constant value of 12 kwh/ton of enriche d ore .

SOURCES : Theodore Shabad, Basic Industrial Resources of the USSR (Ne w York : Columbia University Press, 1969), p . 37-44 ; Theodore Shabad , "News Notes , " Soviet Geography, various issues ; Narodnoye khozyaystv o

-23-

NCSEER March 8 3

Table 5 . continued :

SSSR, various issues ; Paul Lydolph, Geography of the U .S .S .R ., 2n d ed . (New York : John Wiley & Sons Inc ., 1970), p . 509 ; Gornyy zhurnal , various issues ; V . P . Novikov, " The Kursk Magnetic Anonaly—A Promisin g Iron-Ore Base for the Iron and Steel Industry of the Urals, " Sovie t Geography 10 :2 (Feb . 1969) . pp . 70-71 ; Tony Misko & Craig ZumBrunnen , "Soviet Iron Ore : Its Domestic and World implications, " Soviet Natura l Resources in the World Economy (Chicago : University of Chicago Press , forthcoming 1983) ; V . S . Siderova & A . A . Vadyukhina, " New Technolog y and the Location of the Iron and Steel Industry in the Eastern Portio n of the U .S .S .R .," Soviet Geography 18 :1 (Jan . 1977), p . 34 ; S . N . Shalayev, S . Ya . Arsenev & L . M . Peigin, "Puti povysheniy a effektivnosti zhelezorudnoy promyshlennosti v desyatoy pyatiletke, ' ' Gornyy zhurnal no . 3 (1976), p . 4 ; V . V . Strishkov, Mineral Industrie s of the U .S .S .R . (Washington, D .C . : U .S . Dept . of the Interior, Bureau o f Mines, 1979), pp . 2-5, 13-15 ; V . V . Strishkov, Mineral Industries o f the U .S .S .R . (Washington, D .C . : U .S . Dept . of the Interior, Bureau o f Mines, 1977), p . 9 ; Mineral Yearbook, Vol . III, Area Reports : International (Washington, D .C . : U .S . Government Printing Office, 1970 , 1971, 1972, 1975, 1979), various pages .

Scrap Metal Supply Data :

Total estimated Soviet consumption of scrap metal in the iron an d

steel industry was apportioned to twenty-one sources nodes (see Table 6 )

based on a more highly aggregrated version of the same procedure used t o

allocate final steel demand . The details of this procedure will b e

explained later during the discussion of the steel demand data .

Nonetheless, one very important research finding should be noted here .

Our preliminary highly constrained 1970 computer run had an infeasibl e

solution due to insufficient scrap inputs . This result occurred eve n

though we used reported total Soviet scrap inputs for 1970 of slightl y

over 47-million tons . Similar results would have occurred in both th e

1980 and 1990 runs had we not built in a fictitious scrap metal sourc e

node (node number 999) with a very high input cost of 99,999 rubles/100 0

tons . Furthermore, after our initial infeasible 1970 computer run, w e

added high-cost fictitious nodes 999 for all raw material an d

intermediate product inputs for both 1980 and 1990 in order to preclud e

- 2 4-

NCSEER March 8 3

any subsequent infeasibilities . Fortunately, however, scrap input s

proved to be the only problem commodity in all of the simulations . Th e

possible interpretaions of these infeasibilities constitute some of ou r

most significant research discoveries and will be discussed later .

Table 6 . SCRAP SOURCES IN 1000 TON S

PLACE NAME YEAR PRIC E AND REGION : NODE : 1970 1980 1990 R/1000 T

LENINGRAD (NORTHWEST) 24 2585 2505 2990 304 2 MOSCOW (CENTRAL) 415 5969 5097 5995 304 2 KUYBYSHEV (POVOLZH ' YE) 487 2303 3259 4095 419 5 GOR'KIY (VOLGA-VYATKA) 505 658 859 1050 304 2 RIGA (BALTIC) 43 1034 1161 1410 419 5 MINSK (BELORUSSIA) 84 1175 1396 1820 419 5 KRASNODAR ( N . CAUCASUS) 191 1833 1819 2200 419 5 KISHENEV (MOLDAVIA) 908 75 225 285 419 5 TBILISI (TRANSCAUCASUS) 222 1410 1819 1845 419 5 SVERDLOVSK (N . URALS) 549 2424 2107 2480 304 2 NOVOSIBIRSK (W . SIBERIA) 709 3290 3270 3905 355 0 IRKUTSK (E . SIBERIA) 858 893 977 1255 419 5 KHABAROVSK (FAR EAST) 782 893 873 1080 419 5 KARAGANDA (KAZAKHSTAN) 668 1363 1473 1840 355 0 TASHKENT (CENTRAL ASIA) 607 1175 1353 1705 419 5 VINNITSA (S .W . URKRAINE) 807 1457 1661 2085 419 5 DONETSK (DONETS-DNEPR) 284 9447 9105 10840 304 2 NIKOLAYEV (S . URKRAINE) 169 987 1026 1245 419 5 VORONEZH (CENT . CHERNOZEM) 459 2444 2491 3115 419 5 CHELYABINSK (S . URALS) 581 3384 3278 3870 304 2 PERM ' (W . URALS) 524 2209 2246 2720 304 2

Totals : 47008 48000 5599 0 ------SOURCES : Estimates computed by authors, prices based on data from Stee l in the USSR, various issues, scrap totals from Theodore Shabad, "New s Notes," Soviet Geography, various issues .

Regional Energy Cost Data :

Assessment of the potential impacts of changing energy costs b y

type and region were prime motivating factors in initiating thi s

investigation . Regional electrical costs for the three years ar e

- 2 5-

NCSEER March 8 3

enumerated in Table 7 and those for natural gas in Table 8, In th e

Table 7 . REGIONAL ELECTRICITY PRICES (Rubles/1000 kwh )

REGION : 1970 1980 199 0

CENTRAL RUSSIA 9 .78 14 .67 22 .0 1 BALTIC REPUBLICS 13 .84 20 .76 31 .1 4 NORTHWEST 13 .84 20 .76 28 .0 3 UKRAINE 8 .69 13 .04 19 .56 CENTRAL ASIA 12 .48 15 .60 21 .06 KAZAKHSTAN 6 .88 8 .60 10 .75 URALS 8 .69 11 .73 17 .60 EAST SIBERIA 6 .01 7 .51 10 .1 4 WEST SIBERIA 6 .01 7 .51 10 .1 4 BELORUSSIA 13 .84 20 .76 28 .03 CAUCASUS 11 .16 15 .07 20 .34 NORTH CAUSASUS 10 .04 13 .55 18 .29

Sources : Theodore Shabad & Leslie Dienes, The Soviet Energy System , various pages ; Judith Thornton, " Soviet Response to Changing Fuel Cost s and Availabilities : The Case of Electric Power," various pages ; Th e Soviet Economy in a Time of Change (Washington, D .C . : U .S . Governmen t Printing Office, 1979), various pages ; Leslie Dienes, " Energ y Conservation in the USSR, " Energy in Soviet Policy, pp . 101-119 ; various issues of Soviet technical journals .

Table 8 . REGIONAL NATURAL GAS PRICES (Rubles/1000 cubic meters )

Region : 1970 1980 199 0

CENTRAL RUSSIA 18 .00 26 .00 39 .00 BALTIC 18 .80 27 .00 40 .50 UKRAINE 19 .36 26 .00 39 .00 CENTRAL ASIA 9 .40 14 .00 21 .00 KAZAKHSTAN 12 .80 18 .00 27 .00 URALS 14 .50 21 .00 31 .50 EAST SIBERIA NA NA N A WEST SIBERIA NA 18 .00 27 .00 BELORUSSIA 18 .80 27 .00 40 .50 NORTHWEST 18 .80 27 .00 40 .50 CAUCASUS 18 .00 27 .00 40 .50 NORTH CAUCASUS 13 .00 19 .00 26 .00

Sources : Theodore Shabad & Leslie Dienes, The Soviet Energy System , various pages ; Judith Thornton, " Soviet Response to Changing Fuel Cost s and Availabilities : The Case of Electric Power," various pages ; Theodore Shabad, "News Notes" Soviet Geography, various issues ; Th e Soviet Economy in a Time of Change (Washington, D .C . : U .S . Governmen t Printing Office, 1979) various pages ; Leslie Dienes, " Energ y Conservation in the USSR, " Energy in Soviet Policy, pp . 101-119 .

- 2 6 -

NCSEER March 8 3

initial data tables for all commodities and/or processes utilizin g

electricity and/or gas, these estimated regional energy costs wer e

associated with any and all nodes within a given region potentiall y

using these energy supplies .

Pig Iron Production and Cost Data :

Pig iron production data used in the computer model runs for 1970 ,

1980 and 1990 are listed in Tables 9, 10, and 11 respectively . Excep t

for regional aggregation in the Ukraine and Urals, all data represen t

single blast furnace complexes . In addition to minimum and maximu m

production levels for each of the three years, the data include sit e

specific production costs . Here, as everywhere else in this study ,

production cost figures exclude the costs of major raw materials an d

energy inputs and their respective assembly costs for the given proces s

and include amortization costs on plants and equipment, labor costs, an d

other miscellaneous costs such as fluxing agents which have been assume d

to be ubiquitous [19] .

Foreshadowing Leontief ' s [20] criticisms of standardize d

theoretical production functions " . . .used to carry out a respectabl e

econometric study . .. " , we have endeavored to construct our model usin g

directly observable variables from detailed technological an d

engineering information contained in Soviet (and some Western) journals ,

especially Stal ' . Accordingly, Tables 9 through 11 include technica l

input coefficients for five commodities used in pig iron production : (1 )

coke, (2) iron ore, (3) natural gas, (4) electricity and (5) scra p

metal . Where site specific coefficients were not obtained averag e

values of 0 .6 tons of coke per ton of pig iron, 1 .786 tons of ore pe r

ton of pig, 60 cubic meters of gas per ton of pig, 30 kilowatt-hours pe r

ton of pig and 0 .012 tons of scrap per ton of pig were used .

- 2 7 -

NCSEER March b3

Table 9 . DATA FOR PIG IRON PRODUCTION FOR 1970*

PRODUCTION PROD . INPUTS/1000 TONS OF PIG IRO N PLACE : NODE : MIN MAX COSTS COKE ORE GAS ELEC SCRAP

RUSTAVI 910 780 780 38000 600 1788 60 30 1 2 KARAGANDA 668 1800 1800 38000 596 1753 52 30 1 2 NOVOKUZNETSK 726 7300 7300 28050 600 1786 60 30 1 2 KOMUNARSK 884 3300 3400 28370 600 1786 60 30 1 2 ZAPOROZH'YE 179 3900 3900 28370 510 1786 60 30 1 2 LIPETSK 955 4700 4700 28050 480 1792 60 30 1 2 CHEREPOVETSK 836 4000 4000 28050 463 1720 60 30 1 2 TULA 433 1200 2000 34000 600 1786 60 30 1 2 532 1000 2000 34000 600 1786 60 30 1 2 NIZHNIY TAGIL 531 5500 7000 28050 600 1786 60 30 1 2 528 1000 1000 34000 600 1786 60 30 1 2 CHELYABINSK 581 3000 4000 28370 600 1786 60 30 1 2 MAGNITOGORSK 586 10000 11000 17000 600 1786 60 30 1 2 SATKA 576 1100 2500 45300 600 1786 60 30 1 2 NOVOTROITSK 594 1600 2000 34000 600 1786 60 30 1 2 KRIVOY ROG 172 7000 10500 17000 505 1786 75 30 1 2 DNEPROPETROVSK 811 5000 7000 28370 544 1840 60 30 1 2 MAKEYEVKA 283 18400 18400 28050 600 1786 51 30 1 2 ------*NOTE : Production minimum and maximum in 1000 tons , production costs in rubles/1000 tons , coke in tons coke/1000 tons of pig iron , iron ore in tons of ore/1000 tons of pig iron , gas in 1000 cubic meters/1000 tons of pig iron , electricity in 1000 kilowatt hours/1000 tons of pig iron , and scrap in tons of scrap/1000 tons of pig iron .

SOURCES : A . T . Khrushchev, Geografiya promyshlennosti SSSR (Moscow : 1969), pp . 253-272 ; Theodore Shabad, Basic Industrial Resources of th e USSR (New York : Columbia University Press, 1969), pp . 35-44 ; Oxfor d Economic Atlas of the World, 4th ed . (Oxford : Oxford University Press , 1972), pp . 42-43 ; Atlas SSSR, 2nd ed . (Moscow : 1969), pp . 107 , 124-147 ; Theodore Shabad, "News Notes, " Soviet Geography, variou s issues ; Stal', various issues ; Steel in the USSR, various issues ; Minerals Yearbook, Vol . III, Area Reports : International (Washington , D .C ., : U.S. Government Printing Office, 1970, 1971, 1975, 1979), variou s pages ; V . V . Strishkov . Mineral Industries of the U .S .S .R . (Washington , D .C . : U .S . Department of the Interior, Bureau of Mines, 1979), pp . 2-15 ; Narodnoye khozyaystvo SSSR, various years and pages .

------

-28-

NCSEER March 8 3

------Table 10 . DATA FOR PIG IRON PRODUCTION FOR 1980 * ------PRODUCTION PROD . INPUTS/1000 TONS OF PIG IRON PLACE : NODE : MIN MAX COSTS COKE ORE GAS ELEC SCRA P

RUSTAVI 910 800 1000 38000 545 1695 85 30 1 2 KARAGANDA 668 0 4800 28050 533 1695 120 30 1 2 NOVOKUZNETSK 726 0 15000 28050 550 1695 85 30 1 2 KOMMUNARSK 884 4000 5000 19000 545 1695 85 30 1 2 ZAPOROZH'YE 179 4000 4200 28050 446 1695 85 30 1 2 LIPETSK 955 7500 10000 17000 459 1695 85 30 1 2 CHEREPOVETSK 836 5400 10000 19000 425 1652 85 30 1 2 TULA 433 1200 2000 34000 545 1695 85 30 1 2 SEROV 532 1000 2000 34000 545 1695 85 30 1 2 NIZHNIY TAGIL 531 8300 12000 19000 490 1695 88 30 1 2 CHUSOVOY 528 1000 1500 34000 545 1695 85 30 1 2 CHELYABINSK 581 4100 5000 28050 545 1695 85 30 1 2 MAGNITOGORSK 586 1400 14000 17000 490 1690 89 30 1 2 SATKA 576 1100 5000 45300 545 1695 85 30 1 2 NOVOTROITSK 594 1600 2000 34000 545 1695 85 30 1 2 KRIVOY ROG 172 12000 14000 17000 461 1695 145 30 1 2 DNEPROPETROVSK 811 6000 10500 28210 509 1695 66 30 1 2 MAKEYEVKA 283 0 20000 28210 545 1695 85 30 1 2 ------*NOTE : Production minimum and maximum in 1000 tons , production costs in rubles/1000 tons , coke in tons coke/1000 tons of pig iron , iron ore in tons of ore/1000 tons of pig iron , gas in 1000 cubic meters/1000 tons of pig iron , electricity in 1000 kilowatt hours/1000 tons of pig iron , and scrap in tons of scrap/1000 tons of pig iron .

SOURCES : A . T . Khrushchev, Geografiya promyshlennosti SSSR (Moscow : 1969), pp . 253-272 ; Theodore Shabad, Basic Industrial Resources of th e USSR (New York : Columbia University Press, 1969), pp . 35-44 ; Oxfor d Economic Atlas of the World, 4th ed . (Oxford : Oxford University Press , 1972), pp . 42-43 ; Atlas SSSR, 2nd ed . (Moscow : 1969), pp . 107 , 124-147 ; Theodore Shabad, " News Notes, " Soviet Geography, variou s issues ; Stal ', various issues ; Steel in the USSR, various issues ; Minerals Yearbook, Vol . III, Area Reports : International (Washington , D .C ., : U .S . Government Printing Office, 1970, 1971, 1975, 1979), variou s pages ; V . V . Strishkov, Mineral Industries of the U .S .S .R . (Washington , D .C . : U .S . Department of the Interior, Bureau of Mines, 1979), pp . 2-15 ; Narodnoye khozyaystvo SSSR, various years and pages .

- 2 9 -

NCSEER March 8 3

Table 11 . DATA FOR PIG IRON PRODUCTION FOR 1990 * ------PRODUCTION PROD . INPUTS/1000 TONS OF PIG IRO N PLACE : NODE : MIN MAX COSTS COKE ORE GAS ELEC SCRA P

RUSTAVI 910 800 2000 34000 509 1613 132 30 1 2 KARAGANDA 668 4800 15000 22000 509 1613 132 30 1 2 NOVOKUZNETSK 726 5000 20000 22000 509 1613 132 30 1 2 KOMMUNARSK 884 5000 20000 19000 509 1613 132 30 1 2 ZAPOROZH'YE 179 4000 15000 22000 509 1613 132 30 1 2 LIPETSK 955 4000 20000 17000 509 1613 132 30 1 2 CHEREPOVETSK 836 4000 20000 17000 509 1613 132 30 1 2 TULA 433 1200 5000 28050 509 1613 132 30 1 2 SEROV 532 1000 10000 19000 509 1613 132 30 1 2 NIZHNIY TAGIL 531 3000 20000 19000 509 1613 132 30 1 2 CHUSOVOY 528 1000 5000 28370 509 1613 132 30 1 2 CHELYABINSK 581 4100 10000 28050 509 1613 132 30 1 2 MAGNITOGORSK 586 15000 50000 17000 509 1613 132 30 1 2 SATKA 576 1100 7000 36000 509 1613 132 30 1 2 NOVOTROITSK 594 1600 8000 25000 506 1613 132 30 1 2 KRIVOY ROG 172 5000 25000 17000 509 1613 132 30 1 2 DNEPROPETROVSK 811 6000 20000 22000 509 1613 132 30 1 2 MAKEYEVKA 283 20000 50000 22000 509 1613 125 30 1 2 STARYY OSKOL 462 0 20000 28050 509 1613 125 30 1 2 BARNAUL 718 0 20000 28050 509 1613 125 30 1 2 TAYSHET 754 0 20000 28050 609 1613 0 30 1 2 SVOBODNYY 981 0 20000 28050 609 1613 0 30 1 2 TYNDA 962 0 20000 28050 609 1613 0 30 1 2 ------*NOTE : Production ranges in 1000 tons , production costs in rubles/1000 tons , coke in tons coke/1000 tons of pig iron , iron ore in tons of ore/1000 tons of pig iron , gas in 1000 cubic meters/1000 tons of pig iron , electricity in 1000 kilowatt hours/1000 tons of pig iron , and scrap in tons of scrap/1000 tons of pig iron .

SOURCES : A . T . Khrushchev, Geografiya promyshlennosti SSSR (Moscow : 1969), pp . 253-272 ; Theodore Shabad, Basic Industrial Resources of th e USSR (New York : Columbia University Press, 1969), pp . 35-44 ; Oxfor d Economic Atlas of the World, 4th ed . (Oxford : Oxford University Press , 1972), pp . 42-43 ; Atlas SSSR, 2nd ed . (Moscow : 1969), pp . 107 , 124-147 ; Theodore Shabad, "News Notes, " Soviet Geography, variou s issues ; Stal' , various issues ; Steel in the USSR, various issues ; Minerals Yearbook, Vol. III, Area Rep orts : international (Washington , D .C ., : U .S . Government Printing Office, 1970, 1971, 1975, 1979), variou s pages ; V . V . Strishkov, Mineral Industries of the U .S .S .R . (Washington, D .C . : U .S . Department of the Interior, Bureau of Mines , 1979), pp . 2-15 ; Narodnoye khozyaystvo SSSR, various years and pages .

- 3 0-

NCSEER March 8 3

Steel Production Data :

Steel production data for 1970 at 32 locations are listed in Tabl e

12 . While only one additional production site, Staryy Oskol, was adde d

to the system for 1980 (as shown in Table 13), the four potential site s

of Barnaul, Tayshet, Chul'man and Zhlobin were added for 1990 (see Tabl e

14) . A direct reduction steel mill is currently under construction a t

Staryy Oskol and originally should have been already commissioned b y

1980 [21], but was behind schedule . The other nodes represent some o f

the proposed locations for future integrated iron and steel centers mos t

frequently mentioned in the Soviet literature .

At any given node data for up to six different steel processes ar e

listed (refer to Table 15, p . 44) . These six include two major variant s

of electric arc furnaces--one using scrap only (Type A) and the othe r

using both pig iron and scrap inputs (Type T) . Similarly, two variant s

of open hearth processes are possible with respect to coke and natura l

gas usage--one using coke only (Type J) and the other using both cok e

and gas (Type H) . The final two processes are basic oxygen (Type B) an d

the new direct reduction technology (Type U) .

The total number of plants is less than the actual number o f

complexes in existence because the following regional aggegations wer e

performed in order to limit the size of the optimization problem withou t

seriously threatening to distort overall regional commodity productio n

and flow patterns . Kolpino represents small plants at both Kolpino an d

Leningrad . The small Elektrostal ' plant incorporates the small Mosco w

mill as well . Vyksa is subsummed under the adjacent Kulebaki mill .

Dnepropetrovsk includes the Dneprodzerzhinsk mills . The Donetsk Oblas t

mills at Kramatorsk, Konstantinovka, Yenakiyevo, Makeyevka, Zhdanov (tw o

steel complexes) and Donetsk proper have been aggregated and assigned t o

-3 1 -

3 o March NCSEER

the Donetsk city node . Similarly, steel production in Voroshilovgra d

Oblast at Kadiyevka and Kommunarsk has been attributed to the latte r

node only . In the Urals the Beloretsk mill was lumped wit h

Magnitogorsk ; Asha, Zlatoust and Satka with Chelyabinsk ; Nizhni y

Salda, and with Nizhniy Tagil ; and wit h

Chusovoy . In Kazakhstan Temirtau's furnaces were combined with those o f

Karaganda . Finally in Kemerovo Oblast the newer West Siberian plant wa s

joined with the older Novokuznetsk steel complex located a mere sixtee n

kilometers to the southwest . With the exceptions of Dneprodzerzhinsk ,

Makeyevka, Zlatoust and Zhdanov, these deleted furnace installation s

have relatively small production capacities of one-million or les s

metric tons per year . Furthermore, even in these cases the assignment s

have been made to relatively close neighboring nodes .

These regional aggregations have been taken into account i n

determining the production cost data for these implicit multi-nod e

facilities . In general this has been achieved by using cost figure s

close to average costs for a given process .

The three tables 12 through 14 show two maximum productio n

constraints . The first, " NODE MAX, " refers to the upper limit on stee l

production at the given node by all processes combined . The second ,

"TYPE MAX," represents the upper limit of steel output by a give n

process at the given node . Because the "NODE MAX " constraints fairl y

closely reflect 1980 estimates of actual point source or regional stee l

production figures in the case of the aggregated nodes and the model i s

driven by mass balance equations where steel output must equal stee l

demand, nodal minimum production constraints are unnecessary for th e

most part . Nonetheless, there is some slack and range of uncertainit y

in the these maximum constraints .

- 3 2-

NCSEER March 8 3

In our earlier investigation nearly all of the small capacit y

plants dropped out of the set of producing plants even though th e

production costs used at all plants in 1970 were a constant average cos t

[22] . Exceptions were Kulebaki, Gor'kiy, Serov and Alapayevsk in on e

scenario and only Serov in a second run which included seven potentia l

plant locations . Since we were curious about likely opportunity cos t

penalties associated with such modest scale operations, we selecte d

Kolpino and Gor'kiy to have minimum nodal production levels . These tw o

were chosen because they represent two steel plant sites where w e

believe the Soviets would be unlikely to cease production . Althoug h

neither of these minimum constraint values is listed in the tables ,

Kolpino had a " nodal minimum " of 100,000 tons and Gor ' kiy 50,000 ton s

set for 1980 . For 1990 Kolpino ' s minimum was reduced to 50,000 tons an d

Gor'kiy ' s was left at 50,000 tons .

The last six columns in Tables 12 through 14 display coefficient s

for the six steel inputs explicitly modeled in this project : (1) coke ,

(2) natural gas, (3) scrap metal, (4) iron ore, (5) pig iron and (6 )

electricity . Wherever possible actual published Soviet coefficients fo r

particular processes at specific plants were used . More commonly ,

however, average coefficient values were used for both the various stee l

processes and pig iron production . These general input coefficien t

values are listed in Table 15 . A literature search through a decade ' s

worth of pertinent Soviet technical journals from 1970 through earl y

1981 suggested very little increased efficiency in input use except fo r

increasingly higher Fe—content in enriched iron ores . We thus believ e

little increased efficiency is likely through 1990 . Accordingly, w e

have not used any arbitrary "increased input efficiency" factors ove r

time . The empirical evidence in Soviet journals implies that futur e

-3 3 -

NCSEER March 8 3

Table 12 . STEEL PRODUCTION DATA FOR 1970* N T C E 0 Y 0 G 0 P L D P NODE TYPE PROD . K A R I E PLACE : E E MAX MAX COST B S SCRAP E G C

KOLPINO 027 A 2000 1500 33660 13 0 1058 0 0 759 027 T 1500 33660 13 0 669 0 391 75 9 LIEPAYA 046 J 440 440 35090 235 0 466 90 624 142 046 A 440 38200 13 0 1058 0 0 759 046 T 440 38200 13 0 669 0 391 759 MOGILEV 325 H 196 196 45290 13 158 466 90 624 14 2 325 J 196 k5290 235 0 466 90 624 14 2 325 B 196 48400 3 0 308 7 803 75 9 ELEKTROSTAL' 846 A 2000 2000 36500 13 0 1058 0 0 759 846 T 2000 36500 13 0 669 0 391 75 9 CHEREPOVETS 836 H 5500 5500 16260 13 114 466 90 624 14 2 836 J 5500 16260 161 0 466 90 624 14 2 836 B 5500 20890 3 0 308 7 803 15 2 TULA 433 H 1000 1000 33390 13 114 466 90 624 14 2 433 J 1000 33390 161 0 466 90 624 14 2 433 B 1000 36190 3 0 308 7 803 15 2 KULEBAKI 890 H 1200 1200 33390 13 158 466 90 624 9 890 J 1200 33390 235 0 466 90 624 9 890 A 1200 36500 13 0 1058 0 0 62 6 890 T 1200 36500 13 0 1058 0 0 62 6 GOR'KIY 505 A 1000 1000 36500 13 0 1058 0 0 75 9 505 T 1000 36500 13 0 669 0 391 75 9 LIPETSK 955 H 3000 3000 28720 13 114 466 90 624 14 2 955 J 3000 28720 161 0 466 90 624 14 2 955 A 3000 31830 13 0 1058 0 0 75 9 955 T 3000 31830 13 0 669 0 391 75 9 955 B 3000 31520 3 0 308 7 803 15 2 DNEPROPETROVSK 811 H 9200 8000 21560 13 114 466 90 624 14 2 811 J 5000 21560 161 0 466 90 624 14 2 811 A 5000 26500 13 0 1058 0 0 75 9 811 T 5000 26500 13 0 669 0 391 75 9 KRIVOY ROG 172 H 8800 8800 15680 13 114 466 90 624 14 2 172 J 8800 12160 161 0 466 90 624 14 2 172 B 8800 16794 3 0 308 7 803 14 2 ZAPOROZR'YE 179 H 4300 4300 21560 13 114 466 90 624 14 2 179 J 4300 21560 161 0 466 90 624 14 2 179 A 4300 26500 13 0 1058 0 0 759 179 T 4300 26500 13 0 669 0 391 75 9 179 B 4300 26190 3 0 308 7 803 15 2 DONETSK 284 H 18400 13400 21560 13 114 466 90 624 14 2 284 J 13400 21560 161 0 466 90 624 14 2 KOMMUNARSK 884 H 4300 4000 25060 13 114 466 90 624 14 2 884 J 4000 25060 161 0 466 90 624 14 2 884 B 4000 29690 3 0 308 7 803 15 2 TAGANROG 274 H 1600 1600 30550 13 158 466 90 624 14 2 274 J 1600 30550 235 0 466 90 624 14 2

- 3 4-

NCSEER March 8 3

Table 12 . continued : N T C E 0 Y 0 G 0 P L D P NODE TYPE PROD . K A R I E PLACE : E E MAX MAX COST E S SCRAP E G C

RUSTAVI 910 H 1400 1410 30550 13 114 466 90 624 14 2 910 J 1410 30550 161 0 466 90 624 14 2 SUMGAIT 244 H 730 730 33390 13 158 466 90 624 14 2 244 J 730 33390 235 0 466 90 624 14 2 BEKABAD 943 H 390 390 35090 13 158 466 90 624 14 2 943 J 390 35090 235 0 466 90 624 14 2 943 A 390 38200 13 0 1058 0 0 759 943 T 390 38200 13 0 669 0 391 75 9 VOLGOGRAD 265 A 1900 1900 33660 13 0 1058 0 0 75 9 265 T 1900 33660 13 0 669 0 391 75 9 KARAGANDA 668 J 2230 2230 30550 161 0 466 90 624 142 688 B 2230 33350 3 0 308 7 803 15 2 IZHEVSK 554 H 1500 1500 33390 13 158 466 90 624 9 554 J 1500 33390 235 0 466 90 624 9 554 A 1500 36500 13 0 1058 0 0 62 6 554 T 1500 36500 13 0 669 0 391 62 6 SEROV 532 H 2000 2000 30550 13 114 466 90 624 14 2 532 J 2000 30550 161 0 466 90 624 14 2 CHUSOVOY 528 H 3000 2000 30550 13 114 466 90 624 14 2 528 J 2000 30550 161 0 466 90 624 14 2 528 A 1000 36500 13 0 1058 0 0 62 6 528 T 1000 36500 13 0 669 0 391 626 NIZHNIY TAGIL 531 H 9600 9600 16260 13 114 466 90 624 14 2 531 J 9600 16260 161 0 466 90 624 14 2 531 B 9600 20890 3 0 308 7 803 15 2 CHELYABINSK 581 A 6800 2000 25500 13 0 1058 0 0 75 9 581 T 2000 26500 13 0 669 0 391 759 581 B 4800 26170 3 0 308 7 803 15 2 581 H 2000 30550 13 158 466 90 624 14 2 581 J 2000 30550 235 0 466 90 624 14 2 MAGNITOGORSK 586 H 13300 13300 12160 13 114 466 90 624 14 2 586 J 13300 12160 161 0 466 90 624 14 2 586 B 13300 16794 3 0 308 7 803 15 2 NOVOTROITSK 594 H 3000 3000 28720 13 114 466 90 624 14 2 594 J 3000 28720 161 0 466 90 624 14 2 594 B 3000 31520 3 0 308 7 803 152 NOVOKUZNETSK 726 H 6900 4500 21560 13 114 466 90 624 14 2 726 J 4500 21560 161 0 466 90 624 14 2 726 B 2400 26190 3 0 308 7 803 15 2 NOVOSIBIRSK 709 A 1000 1000 36500 13 0 1058 0 0 626 709 T 1000 36500 13 0 669 0 391 62 6 KRASNOYARSK 750 A 500 500 38200 13 0 1058 0 0 626 750 T 500 38200 13 0 669 0 391 626 PETROVSK- 951 A 500 500 38200 13 0 1058 0 0 62 6 ZABAYKAL ' SKIY 951 T 500 38200 13 0 669 0 391 626 KOMSOMOL'SK- 783 A 800 800 36500 13 0 1058 0 0 626 na-AMURE 783 T 800 36500 13 0 669 0 391 626

-35-

NCSEER March 8 3

Table 12 . continued : ------*DEFINITIONS : Type = type of steel process : A = electric arc using scrap only , T = electric arc using both pig iron and scrap , H = open hearth using gas and coke , J = open hearth using coke only , B = basic oxygen , and U = direct reduction ; NODE MAX = upper limit of steel production by sum of all processes a t a given node in 1000 tons ; TYPE MAX = upper limit of steel production by given process at give n node in 1000 tons ; PROD .COST = steel production costs in rubles,1000 tons ; COKE = coke input coefficient in kg of coke/ton of steel ; GAS = gas input coefficient in cubic meters of gas/tons of steel ; SCRAP = scrap input coefficient in kg of scrap/ton of steel ; ORE = iron ore input coefficent in kg of ore/ton of steel ; PIG = pig iron input coefficient in kg of pig/ton of steel ; ELEC = electricity input coefficient in KWH/ton of steel .

SOURCES : Theodore Shabad, Basic Industrial Resources of the USSR (Ne w York : Columbia University Press, 1969), pp . 35-44 ; Theodore Shabad , " News Notes, " Soviet Geography, several issues ; Atlas SSSR, 2nd ed . (Moscow :1969), pp . 107, 124-147 ; Oxford Eonomic Atlas of the World, 4t h edNarodnoye. (Oxford : Oxford University Press, 1972), pp . 42-43 ; khozyaystvo SSSR, various issues ; A . T . Khrushchev, Geografiy a promyshlennosti SSSR (Moscow : 1969), pp . 253-272 ; I . V . Komar , Ratsional ' noye ispol'zovaniye prirodnykh resursov i resursnyye tsikl y (Moscow : Nauka, 1975), pp . 123-153 ; D . I . Popov & R . N . Tikidzhiyev , Uluchsheniye ispol ' zovaniva osnovnykh proizvodstvennykh fondov v cherno v metallurgic (Moscow : Metallurgiya, 1968), pp . 4-95 ; Stal ', variou s issues ; Steel in the USSR, various issues ; Kenneth Warren, Worl d Steel :An Economic Geography (New York : Crane, Russak & Co, Inc ., 1975) , pp . 21-97, 243-265 .

------

-36-

NCSEER March 8 3

Table 13 . STEEL PRODUCTION DATA FOR 1980 *

N T C E 0 Y 0 G 0 P L D P NODE TYPE PROD . K A R I E PLACE : E E MAX MAX COST E S SCRAP E G C

KOLPINO 027 A 3500 3500 33660 13 0 1058 0 0 75 9 027 T 3500 33660 13 0 669 0 391 75 9 LIEPAYA 046 J 600 600 35090 235 0 466 90 624 14 2 046 A 600 38200 13 0 1058 0 0 75 9 046 T 600 38200 13 0 669 0 391 75 9 MOGILEV 325 H 600 300 45290 13 158 466 90 624 14 2 325 J 300 45290 235 0 466 90 624 14 2 325 T 300 48400 13 0 669 0 391 75 9 ELEKTROSTAL' 846 A 3000 1500 36500 13 0 1058 0 0 75 9 846 T 1500 36500 13 0 669 0 391 75 9 CHEREPOVETS 836 H 12000 12000 16260 13 11 466 90 624 14 2 836 J 12000 16260 161 0 466 90 624 14 2 836 B 12000 20890 3 0 308 7 803 15 2 TULA 433 H 2000 2000 33390 13 114 466 90 624 142 433 J 2000 33390 161 0 466 90 624 14 2 433 B 2000 36190 3 0 308 7 803 152 KULEBAKI 890 H 2000 1000 33390 13 158 466 90 624 9 890 J 1000 33390 235 0 466 90 624 9 890 A 1000 36500 13 0 1058 0 0 62 6 890 T 1000 36500 13 0 1058 0 0 62 6 GOR'KIY 505 A 2000 2000 36500 13 0 1058 0 0 75 9 505 T 2000 36500 13 0 669 0 391 75 9 505 B 2000 26190 3 0 308 7 803 15 2 LIPETSK 955 H 10000 10000 28720 13 114 466 90 624 14 2 955 J 10000 28720 161 0 466 90 624 142 955 A 10000 31830 13 0 1058 0 0 75 9 955 T 10000 31830 13 0 669 0 391 75 9 955 B 10000 31520 3 0 308 7 803 15 2 STARYY OSKOL 462 U 2500 2500 35080 300 412 0 1512 0 79 3 462 A 2500 26500 13 0 1058 0 0 75 9 462 T 2500 26500 13 0 669 0 391 75 9 462 B 2500 25080 3 0 308 7 813 152 DNEPROPETROVSK 811 H 8500 6000 21560 13 114 466 90 624 14 2 811 J 6000 21560 161 0 466 90 624 14 2 811 B 6000 26190 3 0 308 7 803 15 2 811 A 6000 26500 13 0 1058 0 0 75 9 811 T 6000 26500 13 0 669 0 391 75 9 KRIVOY ROG 172 H 13500 13500 15680 13 114 466 90 624 14 2 172 J 13500 12160 161 0 466 90 624 14 2 172 B 13500 16794 3 0 308 7 803 14 2 ZAPOROZH ' YE 179 H 5000 5000 21560 13 114 466 90 624 14 2 179 J 5000 21560 161 0 466 90 624 14 2 179 A 5000 26500 13 0 1058 0 0 75 9 179 T 5000 26500 13 0 669 0 391 759 179 B 5000 26190 3 0 308 7 803 15 2

- 37-

NCSEER March 8 3

Table 13 . continued :

N T C E 0 Y 0 G 0 P L D P NODE TYPE PROD, K A R I E PLACE : E E MAX MAX COST E S SCRAP E G C

DONETSK 284 H 24000 15000 21560 13 114 466 90 624 14 2 284 J 15000 21560 161 0 466 90 624 14 2 284 B 0 26190 3 0 308 7 803 15 2 KOMMUNARSK 884 H 5000 4000 25060 13 114 466 90 624 14 2 884 J 4000 25060 161 0 466 90 624 14 2 884 B 4000 29690 3 0 308 7 803 15 2 TAGANROG 274 H 2000 2000 30550 13 158 466 90 624 14 2 274 J 2000 30550 235 0 466 90 624 14 2 RUSTAVI 910 H 2000 2000 30550 13 114 466 90 624 14 2 910 J 2000 30550 161 0 466 90 624 14 2 910 B 2000 33350 3 0 308 7 803 15 2 SUMGAIT 244 H 870 870 33390 13 158 466 90 624 14 2 244 J 870 33390 235 0 466 90 624 14 2 BEKABAD 943 H 1100 1100 35090 13 158 466 90 624 14 2 943 J 1100 35090 235 0 466 90 624 14 2 943 A 1100 38200 13 0 1058 0 0 75 9 943 T 1100 38200 13 0 669 0 391 759 VOLGOGRAD 265 A 2500 2500 33660 13 0 1058 0 0 75 9 265 T 2500 33660 13 0 669 0 391 75 9 KARAGANDA 668 J 6000 6000 30550 161 0 466 90 624 142 688 B 6000 33350 3 0 308 7 803 15 2 IZHEVSK 554 H 1500 1000 33390 13 158 466 90 624 9 554 J 1000 33390 235 0 466 90 624 9 554 A 1000 36500 13 0 1058 0 0 62 6 554 T 1000 36500 13 0 669 0 391 62 6 SEROV 532 H 2000 2000 30550 13 114 466 90 624 14 2 532 J 2000 30550 161 0 466 90 624 14 2 532 B 0 33350 3 0 308 7 803 15 2 CHUSOVOY 528 H 3000 2000 30550 13 114 466 90 624 14 2 528 J 2000 30550 161 0 466 90 624 14 2 528 B 0 33350 3 0 308 7 803 15 2 NIZHNIY TAGIL 531 H 10800 7300 16260 13 114 466 90 624 14 2 531 J 7300 16260 161 0 466 90 624 14 2 531 B 7300 20890 3 0 308 7 803 152 CHELYABINSK 581 A 8700 6000 26500 13 0 1058 0 0 759 581 T 6000 26500 13 0 669 0 391 75 9 581 B 6700 25190 3 0 308 7 803 15 2 581 H 6000 30550 13 158 466 90 624 14 2 581 J 6000 30550 235 0 466 90 624 14 2 MAGNITOGORSK 586 H 16800 15800 12160 13 114 466 90 624 14 2 586 J 15800 12160 161 0 466 90 624 14 2 586 B 15800 16794 3 0 308 7 803 15 2 NOVOTROITSK 594 H 3000 3000 28720 13 114 466 90 624 14 2 594 J 3000 28720 161 0 466 90 624 14 2 594 B 3000 31520 3 0 308 7 803 15 2 NOVOKUZNETSK 726 H 11900 4500 21560 13 114 466 90 624 14 2 726 J 4500 21560 161 0 466 90 624 14 2 726 B 7400 26190 3 0 308 7 803 15 2

- 3 8-

NCSEER March 8 3

Table 13 . continued : ------N T C E 0 Y 0 G 0 P L D P NODE TYPE PROD . K A R I E PLACE : E E MAX MAX COST E S SCRAP E G C

NOVOSIBIRSK 709 A 4500 4500 36500 13 0 1058 0 0 62 6 709 T 4500 36500 13 0 669 0 391 62 6 KRASNOYARSK 750 A 1000 1000 38200 13 0 1058 0 0 62 6 750 T 1000 38200 13 0 669 0 391 62 6 750 B 1000 37890 3 0 308 7 803 1 9 PETROVSK- 951 A 1000 1000 38200 13 0 1058 0 0 62 6 ZABAYKAL'SKIY 951 T 1000 38200 13 0 669 0 391 62 6 KOMSOMOL ' SK- 783 A 1100 1100 36500 13 0 1058 0 0 62 6 na-AMURE 783 T 1100 36500 13 0 669 0 391 62 6 ------*DEFINITIONS : Type = type of steel process : A = electric arc using scrap only , T = electric arc using both pig iron and scrap , H = open hearth using gas and coke , J = open hearth using coke only , B = basic oxygen , and U = direct reduction ; NODE MAX = upper limit of steel production by sum of all processes a t a given node in 1000 tons ; TYPE MAX = upper limit of steel production by given process at give n node in 1000 tons ; PROD .COST = steel production costs in rubles/1000 tons ; COKE = coke input coefficient in kg of coke/ton of steel ; GAS = gas input coefficient in cubic meters of gas/tons of steel ; SCRAP = scrap input coefficient in kg of scrap/ton of steel ; ORE = iron ore input coefficent in kg of ore/ton of steel ; PIG = pig iron input coefficient in kg of pig/ton of steel ; ELEC = electricity input coefficient in KWH/ton of steel .

SOURCES : Theodore Shabad, Basic Industrial Resources of the USSR (Ne w York : Columbia University Press, 1969), pp . 35-44 ; Theodore Shabad , " News Notes, " Soviet Geography, several issues ; Atlas SSSR, 2nd ed . (Moscow :1969), pp. 107, 124-147 ; Oxford Eonomic Atlas of the World, 4t h ed . (Oxford : Oxford University Press, 1972), pp . 42-43 ; Narodnoy e khozyaystvo SSSR, various issues ; A . T . Khrushchev, Geografiya . promyshlennosti SSSR (Moscow : 1969), pp . 253-272 ; I . V . Komar , Ratsional ' noye ispol ' zovaniye prirodnykh resursov i resursnvye tsikl y (Moscow : Nauka, 1975), pp . 123-153 ; D . I . Popov & R . N . Tikidzhiyev , Uluchsheniye ispol ' zovaniya osnovnykh proizvodstvennykh fondov v cherno y metallurgic (Moscow : Metallurgiya, 1968), pp . 4-95 ; Stal', variou s issues ; Steel in the USSR, various issues ; Kenneth Warren, Worl d Steel : An Economic Geography (New York : Crane, Russak & Co, Inc ., 1975) , pp . 21-97, 243-265 .

------

-39-

NCSEER March 8 3

Table 14 . STEEL PRODUCTION DATA FOR 1990* N T C E O Y 0 G 0 P L D P NODE TYPE PROD . K A R I E PLACE : E E MAX MAX COST F S SCRAP E C C KOLPINO 027 A 30000 30000 33660 13 0 1058 0 0 759 027 T 30000 33660 13 0 669 0 391 75 9 LIEPAYA 046 H 750 750 35090 235 0 466 90 624 14 2 046 A 750 38200 13 0 1058 0 0 759 046 T 750 38200 13 0 669 0 391 75 9 MOGILEV 325 H 400 400 45290 13 158 466 90 624 142 325 J 400 45290 235 0 466 90 624 14 2 325 T 400 48400 13 0 669 0 391 75 9 FLFKTROSTAL' 846 A 4000 4000 36500 13 0 1058 0 0 759 846 T 2000 36500 13 0 669 0 391 75 9 CHEREPOVETS 836 H 30000 30000 16260 13 14 466 90 624 142 836 J 30000 16260 161 0 466 90 624 14 2 836 B 30000 20890 3 0 308 7 803 15 2 TULA 433 H 30000 30000 33390 13 114 466 90 624 14 2 433 J 30000 33390 161 0 466 90 624 14 2 433 B 30000 36190 3 0 308 7 803 15 2 KULEBAKI 890 H 30000 30000 33390 13 158 466 90 624 9 890 J 30000 33390 235 0 466 90 624 9 890 A 30000 36500 13 0 1058 0 0 62 6 890 T 30000 35500 13 0 1058 0 0 62 6 GOR'KIY 505 A 30000 30000 36500 13 0 1058 0 0 75 9 505 T 30000 36500 13 0 669 0 391 75 9 505 B 30000 26190 3 0 308 7 803 15 2 LIPETSK 955 H 30000 30000 28720 13 114 466 90 624 14 2 955 J 30000 28720 161 0 466 90 624 14 2 955 A 30000 31830 13 0 1058 0 0 759 955 T 30000 31830 13 0 669 0 391 75 9 955 B 30000 31520 3 0 308 7 803 15 2 STARYY OSKOL 462 U 30000 30000 35080 300 412 0 1512 0 79 3 462 A 30000 26500 13 0 1058 0 0 75 9 462 T 30000 26500 13 0 669 0 391 75 9 462 B 30000 25080 3 0 308 7 813 15 2 DNEPROPETROVSK 811 H 30000 30000 21560 13 114 466 90 624 14 2 811 J 30000 21560 161 0 466 90 624 14 2 811 B 30000 26190 3 0 308 7 803 15 2 811 A 30000 26500 13 0 1058 0 0 75 9 811 T 30000 26500 13 0 669 0 391 75 9 KRIVOY ROG 172 H 30000 30000 15680 13 114 466 90 624 14 2 172 J 30000 12160 161 0 466 90 624 14 2 172 B 30000 16794 3 0 308 7 803 14 2 ZAPOROZH'YE 179 H 30000 30000 21560 13 114 466 90 624 14 2 179 J 30000142 21560 161 0 . 466 90 62 4 179 A 30000 26500 13 0 1058 0 0 75 9 179 T 30000 26500 13 0 669 0 391 759 179 B 30000 26190 3 0 308 7 803 152

-40-

NCSEER March 8 3

Table 14 . continued :

N T C E 0 Y 0 G 0 P L D P NODE TYPE PROD . K A R I E PLACE : E E MAX MAX COST E S SCRAP E G C

DONETSK 284 H 30000 30000 21560 13 114 466 90 624 14 2 284 J 30000 21560 161 0 466 90 624 14 2 284 B 30000 26190 3 0 308 7 803 15 2 KOMMUNARSK 884 H 30000 30000 25060 13 114 466 90 624 14 2 884 J 30000 25060 161 0 466 90 624 142 884 B 30000 29690 3 0 308 7 803 15 2 TAGANROG 274 H 30000 30000 30550 13 158 466 90 62 4 142 274 J 30000 30550 235 0 466 90 624 14 2 RUSTAVI 910 H 1600 1600 30550 13 114 466 90 624 14 2 910 J 1600 30550 161 0 466 90 624 14 2 910 B 1600 33350 3 0 308 7 803 15 2 SUMGAIT 244 H 30000 30000 33390 13 158 466 90 624 14 2 244 J 30000 33390 235 0 466 90 624 142 BEKABAD 943 H 30000 30000 35090 13 158 466 90 624 142 943 J 30000 35090 235 0 466 90 624 14 2 943 A 30000 38200 13 0 1058 0 0 759 943 T 30000 38200 13 0 669 0 391 75 9 VOLGOGRAD 265 A 30000 30000 33660 13 0 1058 0 0 759 265 T 30000 33660 13 0 669 0 391 759 KARAGANDA 668 J 30000 30000 30550 161 0 466 90 624 142 688 B 30000 33350 3 0 308 7 803 15 2 IZHEVSK 554 H 30000 30000 33390 13 158 466 90 624 9 554 J 30000 33390 235 0 466 90 624 9 554 A 30000 36500 13 0 1058 0 0 626 554 T 30000 36500 13 0 669 0 391 626 SEROV 532 H 30000 30000 30550 13 114 466 90 624 142 532 J 30000 30550 161 0 466 90 624 14 2 532 B 30000 33350 3 0 308 7 803 15 2 CHUSOVOY 528 H 30000 30000 30550 13 114 466 90 624 14 2 528 J 30000 30550 161 0 466 90 624 14 2 528 B 30000 33350 3 0 308 7 803 15 2 NIZHNIY TAGIL 531 H 30000 30000 16260 13 114 466 90 624 14 2 531 J 30000 16260 161 0 466 90 624 14 2 531 B 30000 20890 3 0 308 7 803 15 2 CHELYABINSK 581 A 30000 30000 26500 13 0 1058 0 0 759 581 T 30000 26500 13 0 669 0 391 75 9 581 B 30000 26190 3 0 308 7 803 15 2 581 H 30000 16260 13 158 466 90 624 14 2 581 J 30000 16260 235 0 466 90 624 14 2 MAGNITOGORSK 586 H 30000 30000 12160 13 114 466 90 624 142 586 J 30000 12160 161 0 466 90 624 14 2 586 B 30000 16794 3 0 308 7 803 15 2 NOVOTROITSK 594 H 30000 30000 28720 13 114 466 90 624 14 2 594 J 30000 28720 161 0 466 90 624 14 2 594 B 30000 31520 3 0 308 7 803 15 2 NOVOKUZNETSK 726 H 30000 30000 21560 13 114 466 90 624 14 2 726 J 30000 21560 161 0 466 90 624 14 2 726 B 30000 26190 3 0 308 7 803 15 2

-4 1 -

NCSEER March 83

Table 14 . continued :

N T C E O Y 0 G 0 P L D P NODE TYPE PROD . K A R I E PLACE : E E MAX MAX COST E S SCRAP E G C

NOVOSIBIRSK 709 A 30000 30000 36500 13 0 1058 0 0 62 6 709 T 30000 36500 13 0 669 0 391 62 6 KRASNOYARSK 750 A 30000 30000 38200 13 0 1058 0 0 62 6 750 T 30000 38200 13 0 669 0 391 62 6 750 B 30000 37890 3 0 308 7 803 1 9 PETROVSK- 951 A 30000 30000 38%00 13 0 1058 0 0 62 6 ZABAYKAL'SLIY 951 T 30000 38200 13 0 669 0 391 62 6 KOMSOMOL,SK- 783 A 30000 30000 36500 13 0 1058 0 0 62 6 na-AMURE 783 T 30000 36500 13 0 669 0 391 62 6 BARNAUL 718 A 30000 30000 26500 13 0 1058 0 0 62 6 718 T 30000 26500 13 0 669 0 391 62 6 718 B 30000 26190 3 0 308 7 803 1 9 718 U 30000 25080 879 0 0 1512 6 660 TAYSHET 754 A 30000 30000 26500 13 0 1058 0 0 62 6 754 T 30000 26500 13 0 669 0 391 62 6 754 B 30000 26190 3 0 308 7 803 1 9 754 U 30000 25080 879 0 0 1512 0 660 CHUL ' MAN 963 A 30000 30000 26500 13 0 1058 0 0 62 6 963 T 30000 26500 13 0 669 0 391 62 6 963 B 30000 26190 3 0 308 7 803 1 9 963 U 30000 25080 300 412 0 1512 0 66 0 ZHLOBIN 88 A 30000 30000 26500 13 0 1058 0 0 62 6 88 T 30000 26500 13 0 669 0 391 62 6 88 U 30000 25080 300 412 0 1512 0 79 3 88 H 30000 21560 13 158 466 90 624 9 88 J 30000 21560 235 0 466 90 624 9

*DEFINITIONS : Type = type of steel process : A = electric arc using scrap only , T = electric arc using both pig iron and scrap , H = open hearth using gas and coke , J = open hearth using coke only , B = basic oxygen , and U = direct reduction ; NODE MAX = upper limit of steel production by sum of all processes a t a given node in 1000 tons ; TYPE MAX = upper limit of steel production by given process at give n node in 1000 tons ; PROD .COST = steel production costs in rubles/1000 tons ; COKE = coke input coefficient in kg of coke/ton of steel ; GAS = gas input coefficient in cubic meters of gas/tons of steel ; SCRAP = scrap input coefficient in kg of scrap/ton of steel ; ORE = iron ore input coefficent in kg of ore/ton of steel ; PIG = pig iron input coefficient in kg of pig/ton of steel ; ELEC = electricity input coefficient in KWH/ton of steel .

-4 2 -

NCSEER March 8 3

Table 14 . continued : ------SOURCES : Theodore Shabad, Basic Industrial Resources of the USSR (Ne w York : Columbia University Press, 1969), pp . 35-44 ; Theodore Shabad , " News Notes, " Soviet Geography, several issues ; Atlas SSSR 2nd ed . Narodnoye(Moscow :1969), pp . 107, 124-147 ; Oxford Eonomic Atlas of the World 4t h ed . (Oxford : Oxford University Press, 1972), pp . 42-43 ; — khozyaystvo SSSR, various issues ; A . T . Khrushchev, Geografiy a promyshlennosti SSSR (Moscow : 1969), pp . 253-272 ; I . V . Komar , Ratsional'noye ispol'zovaniye prirodnykh resursov i resursnyye tsikl y (Moscow : Nauka, 1975), pp . 123-153 ; D . I . Popov & R . N . Tikidzhiyev , Uluchsheniye ispol ' zovaniva osnovnykh proizvodstvennykh fondov v cherno v metallurgiiStal', various(Moscow : Metallurgiya, 1968), pp . 4-95 ; issues ; Steel in the USSR, various issues ; Kenneth Warren, Worl d Steel : An Economic Geographv (New York : Crane, Russak & Co, Inc ., 1975) , pp . 21-97, 2- .3-265 .

------

efficiency gains will be achieved by an evolving technological-proces s

mix rather than by significant overall improvements within a give n

process . This does not mean that no progress is being made a s

economically based resource substitutions are being experimented with a t

several metallurgical complexes . But, for example, some gas may b e

substituted for some coke . In fact, of course, it is exactly thes e

types of energy tradeoffs which have motivated this exploratory study .

It is precisely for this reason that we have tried to model no t

only different steel-making processes, but also, different variant s

within given processes . Table 15 thus includes four general open-heart h

process variants . While Types H (coke and gas) and J (coke only )

represent in general technologies with differing energy mixes ,

presumably the first H-type and the first J-type which are less energ y

intensive variants of their two respective types represent newer, mor e

technologically efficient facilities . Two direct reduction proces s

variants are included simply to take account of the fact that natura l

gas is not available at all potential direct reduction sites . Finally ,

the electricity coefficients for all processes is lower by 133 kwh/to n

at all sites lacking integrated rolling mills .

-4 3 -

NCSEER March 8 3

______------Table 15 . GENERAL STEEL PROCESS COEFFICIENTS TABLE* ------IRON PIG ELEC - COKE GAS SCRAP ORE IRON TRICITY PROCESS : TYPE (KG/T) (M /T) (KG/T) (KG/T) (KG/T) (KWH/T )

OPEN HEARTH H 13 114 466 90 624 14 2 OPEN HEARTH H 13 158 466 90 624 142

OPEN HEARTH J 161 0 466 90 624 142 OPEN HEARTH J 235 0 466 90 624 14 2

ELECTRIC ARC A 13 0 1058 0 0 75 9 ELECTRIC ARC T 13 0 669 0 391 75 9

BASIC OXYGEN B 3 0 308 7 803 152

DIRECT REDUCTION U 300 412 0 1512 0 79 3 DIRECT REDUCTION U 879 0 0 1512 0 79 3

PIG IRON PROD . P 409-600 50-132 12 1600-1800 NA 3 0

*Input coefficients are all in terms of input units per ton o f steel except for pig iron production which are expressed wit h respect to tons of pig iron .

SOURCES : Stal' (Jan, 1976), pp . 8-10 ; (Feb . 1976), pp . 131-133 ; (Ma y 1976), p . 232 ; (Nov . 1976), p . 594 ; (Feb . 1977), pp . 101-104 ; (Nov .1977), pp . 978-981 ; (Jan . 1978), pp . 7-8 ; (Sept . 1978, pp . 782-787 ; (Nov . 1978), pp . 982-986 ; (March 1979), pp . 167-172 ; (May 1979), pp . 328-331 ; Izvestiya vuz chernaya metallurgiya (Jan . 1976), pp . 36-38 ; (Feb . 1976), pp . 43-44 ; Steel in the USSR (Feb . 1976), p . 57 ; (Marc h 1976), p . 119 ; (April 1976), p . 171 ; (May 1976), p, 260 ; (July 1976) , pp . 349-351 ; (March 1977), pp . 129-130, 185-187 ; (May 1977), 249-254 , 300-305 ; (Feb . 1978), pp . 61-64 ; (March 1978), pp . 127-129 ; (Apri l 1978), pp . 189-190 ; (June 1978), pp . 307-308 ; (August 1978), p . 477 ; (Feb . 1979), pp . 61-62 ; (March 1979), pp . 155-156 ; (July 1979), p . 321 ; (August 1979), p . 382 ; (May 1980), p . 280 ;

------

Although non-linear, or in general economies of scale, producio n

cost functions can not be directly incorporated into a linea r

programming model ; nonetheless, they exist in reality and can not b e

ignored completely without seriously jeopardizing one ' s research

results . Essentially this modeling problem could have been addressed i n

a couple of ways, First, an attempt to use a non-linea r mathematical

-44-

NCSEER March 8 3

programming technique was quickly abandoned due to the lack of an y

software available to handle such a large, multistage optimizatio n

problem . Earlier efforts to develop a non-linear algorithm to solve a

mere " coal plus ore equals steel " problem for spatially disaggregate d

Soviet data yielded very sensitive, unstable results unless th e

geographical data base was collapsed into a quite modest number of node s

[19] . Therefore, a simpler, quasi-interactive approach was adopted a s

follows .

------Table 16 . STEEL PRODUCTION COSTS BY -CAPACITY AND PROCES S ------CAPACITY OPEN ELECTRIC BASIC DIREC T (MILLION HEARTH ARC OXYGEN REDUCTIO N TONS /YR) (RUB ./TON) (RUB ./TON) (RUB ./TON) (RUB ./TON )

7-10 12 .16 17,10 16 .79 15 .68 5-7 16 .26 21 .20 20 .89 19 .78 4-5 21 .56 26 .50 26 .19 25 .08 3-4 25 .06 30 .00 29 .69 28 .58 2-3 28 .72 31 .83 31 .52 30,4 1 1-2 30 .55 33 .66 33,35 32 .24 0 .5-1 33 .39 36 .50 36 .19 35 .08 0 .3-0 .5 35 .09 38 .20 37 .89 36 .78 0 .2-0 .3 37 .39 40 .50 40 .19 39 .08 0 .1-0 .2 45 .29 48 .40 48 .09 46 .98 ------SOURCES : V . Ya . Rom, " Geographical Problems in the Iron and Stee l Industry of the USSR," Soviet Geography 15 :3 (March 1964),p . 126 ; Stee l in the USSR (March 1977), pp . 185-187 ; (August 1978), p . 477 ; (Marc h 1979), pp . .155-156 ; Izvestiya VUZ . Chernaya metallurgiya (May 1977) , pp . 300-305 ; M . Kh . Gankina, V . A . Petruchika, & V . M . Fomina, eds . Perevozki gruzov (Moscow : Transport, 1972), p . 145 ; Kh . N . Gizatullin , " Zamykayushchiye zatraty na zhelezorudnyye resursy, " Ekonomika i matematicheskiye metody 14 :4 (July-August, 1978), pp . 700-708 .

------

For pig iron and steel processes production costs were initiall y

set both for 1970 and 1980 to reflect their actual production capacit y

(capacity by process in the case of steel) as closely as possible .

These costs, as a function of both capacity and process are listed i n

-4 5 -

NCSEER March 8 3

Table 16 . While this procedure seems reasonable enough for the mor e

highly constrained 1970 and 1980 computer runs, the 1990 productio n

costs are less satisfactory . For the projected year 1990 productio n

costs for each process at a given node were left the same as for 198 0

except for new or potential sites where production costs were assigne d

the average costs for a plant with a capacity of 4-to-5 million tons pe r

annum . This latter capacity ranges represents an average size Sovie t

steel complex .

Had computer funds and time allowed, we had hoped to mak e

additional computer runs including sensitivity analyses . In addition t o

sensitivity analyses, it would be highly desirable to reset productio n

costs iteratively according to the optimal indicated output of th e

previous run(s) . Furthermore, it would be of interest to further asses s

the system-wide consequences of forcing all currently producing sites t o

operate in both the 1980 and 1990 runs . This could very simply be don e

by building in more minimum level production constraints than we did fo r

these current analyses, We hope to obtain additional funding so tha t

such sensitivity analyzes may be done, especially for the years 1980 an d

1990 .

System Constraints Data :

In addition to the disaggregated nodal production and/or suppl y

constraints on the various commodities and processes, a series o f

system-wide constraints were also imposed in order to force the model t o

reflect actual production and/or energy consumption levels within Sovie t

ferrous metallurgy . These constraints are shown in Table 17 . While th e

data for 1970 and 1980 represent actual published figures, those fo r

1990 are projections of the authors . The pig iron upper limit o f

-46-

NCSEER March 8 3

133-million tons annually is actually quite optimistic, whereas th e

range of steel production is quite conservative compared with Sovie t

projections . Recent stagnation and even reversals in Soviet iron ore ,

pig iron, and steel production, however, make us believe our estimate s

of 1990 steel output are possibly even still on the optimistic sid e

[24] . Because we doubt that the Soviets will retire any steel producin g

facilities during the eighties, it seems reasonable to set the lowe r

limit on 1990 steel production by process to the actual 1980 output b y

each of the four processes . Although the Staryy Oskol direct reductio n

plant was behind schedule and not producing in 1980, it certainly wil l

be operating at over one-million tons annually in 1990 .

------Table 17 . SYSTEM CONSTRAINTS BY COMMODITY BY YEA R ------1970 1980 199 0 COMMODITY : MIN MAX MIN MAX MIN MAX ------PIG IRON (1000 tons) NONE 85,933 NONE 115,000 NONE 133,00 0

STEEL (1000 tons) 115,900 115,900 147,900 147,900 169,000 171,80 0

OPEN HEARTH (1000 t .) 77,653 NONE 99,093 99,093 99,093 NON E ELEC . ARC (1000 t .) NONE 5,795 NONE 7,395 7,395 NON E BASIC OXY . (1000 t .) NONE 23,180 41,412 41,412 41,412 NON E DIRECT REDU .(1000 t .) NONE 0 NONE 1,000 1,000 NON E

ELECTRICITY (million NONE 70,700 NONE 131,000 NONE 170,000 kilowatt hrs ) NATURAL GAS (million NONE 27,246 NONE 43,259 NONE 65,00 0 cubic meters ) ------SOURCES : Narodnoye khozyaystvo SSSR v 1970 g. (Moscow : Statistika 1971) , pp . 190-191 ; Narodnoye khozyaystvo SSSR v 1979 (Moscow : Statistika , 1980), pp . 172 ; Leslie Dienes & Theodore Shabad, The Soviet Energy Syste m (Washington, D .C . : V . H . Winston & Sons, 1979), pp . 17-40, 224-225 ; Theodore Shabad, " News Notes, " Soviet Geography, various issues .

------

Final Demand Data :

Four types of final demands were collected : (1) coking coa l

-47- 3 March 8 NCSEER

exports, (2) iron ore exports, (3) pig iron, and (4) steel . While th e

first three commodities have a whole series of model determined spatia l

demands, here we are referring to fixed demands only . Foreign trad e

statistics were consulted to assemble Soviet coking coal expor t

quantities for 1970 and 1980 . These data were regionally grouped an d

attributed to four plausible Soviet border-export nodes (see Table 18) .

The increase in 1990 total export demand is predicated primarily o n

Japanese import contracts . An identical approach was taken to collect ,

total, and allocate Soviet iron ore export flow demands to fiv e

border-export nodes (see Table 19) . The 1990 ore export projection s

were based on an early Association of American Geographers-Nationa l

Science Foundation research project [25] .

Pig iron final demand is composed of foreign demands and domesti c

final demands such as for cast iron . Their sum plus pig iron used a s

inputs in the various steel processes equal total pig iron production .

Since total pig output, pig exports and the quantity of pig iro n

consumed in steel making in 1970 and 1980 are known, the differenc e

between total production minus exports and pig inputs in steel smeltin g

yields total domestic final demand for pig iron or 4 .206 million ton s

and 2 .0 million tons, respectively for 1970 and 1980 .

These quantities were attributed to regional final demand node s

proportional to published 1970 Soviet pig iron consumption data fo r

fifteen economic regions [26] . The pig iron consumption of the Centra l

Economic Regions was distributed to Moscow and Voronezh ; that of th e

Urals to Perm ' , Chelyabinsk and Sverdlovsk ; that of Povolzh ' ye t o

Gor ' kiy and Kuybyshev ; and that of the Ukraine to Vinnitsa, Nikolaye v and Donetsk proportional to regional population totals . Within each o f

the remaining economic regions all regional pig iron consumption wa s

-- 4 8-

NCSEER March 8 3

Table 18 . COKING COAL EXPORT S (x 1000 tons )

PLACE NAME : NODE* 1970 1980 199 0

RIGA 43 840 717 72 0 SEVASTOPOL ' 185 1008 1235 125 0 NAKHODKA 802 191 252 550 2 MOSTISTKA 107 2161 1796 197 5

TOTAL : 4200 4000 944 7

*Note : Nodes represent assumed export nodes based on regionall y aggregated export flows .

SOURCES : Vneshnyaya torgovlya SSSR za 1971 god (Moscow : Mezhdunarodnyy e Otnosheniya, 1972), pp . 27, 68, 114, 208, 259 ; Vneshnyaya torgovly a SSSR v 1975 g (Moscow : Statistika, 1976), pp .25, 68, 111, 153, 215 , 252, 269, 299 ; Vneshnyaya torgovlya SSSR v 1979 (Moscow : Statistika , 1980), various pages ; Lesie Dienes & Theodore Shabad, The Soviet Energ y System (Washington, D . C . : V . H . Winston & Sons, 1979), pp . 217-261 .

Table 19 . IRON ORE EXPORTS IN 1000 TON S ______PLACE NAME : NODE* 1970 1980 199 0

NAKHODKA 802 1258 650 140 0 MURMANSK 368 1390 3000 3500 SEVASTOPOL ' 185 6367 6888 8000 MOSTISKA 107 24095 27912 28100 RIGA 43 2990 2500 300 0

Total : 36100 40950 44000

*NOTE : Nodes represent assumed export nodes based on regionall y aggregated export flows .

SOURCES : Vneshnyaya torgovlya SSSR za 1971 god (Moscow : Mezhdunarodnyy e Otnosheniya, 1972), various pages ; Vneshnyaya torgovlya SSSR v 197 9 (Moscow : Statistia, 1980), various pages ; Theodore Shabad, " New s Notes, " Soviet Geography, various issues ; Tony Misko & Crai g ZumBrunnen, ;Soviet Iron Ore : Its Domestic and World Implications, " Soviet Resources in the World Economy, Robert Jensen, Theodore Shabad , and Arthur Wright, eds . (Chicago : University of Chicago Press , forthcoming 1983) .

-49—

NCSEER March 8 3

allocated to a single large urban node resulting in a total o f

twenty-one surrogate pig iron final demand centers (see Table 20) .

For 1980 and 1990 the same nodes were selected, but the fina l

demand assignment method allowed for reallocations proportional t o

regional population growth between 1970 and 1980 . The same growth rate s

were obtained for the reallocation of the projected domestic final pi g

iron demand to 1990 . Finally foreign trade statistics were consulted t o

arrive at regionally aggregated export flows .

The regional allocation of steel demand is probably the weakes t

series of the entire raw data base . Initially all Soviet cities equa l

to or greater than 65,000 inhabitants in 1970 were selected as potentia l

,steel demand nodes . Accordingly, the total 1970 Soviet steel productio n

of 115 .9 million tons less the 7 .0 million tons exported [27] or 108 . 9

million tons was assumed to be equivalent to total Soviet domestic stee l

demand . The following procedure was used to apportion this quantity t o

the selected group of 304 city nodes . First, good Soviet regional stee l

demand data for 1970 permitted the apportionment of the 108 .9 millo n

tons to nineteen economic regions [28] . Within each of these regions ,

the regional demand was assigned to individual nodes (i .e ., all citie s

> 65,000 in population within a given region) proportional to a city ' s

population size [29] weighted by the summed fraction of the city ' s labo r

force engaged in industry, construction and transportation [30] .

Unfortunately, we had to drastically collapse this geographica l

demand array for computational and cost reasons . Keeping this spatiall y

disaggregated demand set would have drastically increased the size o f

the problem from approximately 18,000 decision variables to an unwield y

number in excess of 135,000 . Preferring to view live trees rather tha n

4,500 pages of computer printouts for each run, we judiciously prune d

5 0-

NCSEER March 8 3

Table 20 . ESTIMATED PIG IRON DEMAND BY REGION BY YEAR IN 1000 TON S ------N T q Y DP PLACE OF PIG IRON DEM AND : E E* 1970 1980 199 0

LENINGRAD 24 D 181 83 103 MOSCOW 415 D 904 366 44 7 KUYBYSHEV 487 D 156 105 138 GOR'KIY 505 D 42 26 33 RIGA 43 D 55 28 35 MINSK 84 D 181 101 13 5 KRASNODAR 191 D 151 70 88 KISHENEV 908 D 8 4 5 TBILISI 222 D 109 67 7 0 SVERDLOVSK 549 D 135 58 7 0 NOVOSIBIRSK 709 D 248 119 14 5 IRKUTSK 858 D 50 27 35 KHABAROVSK 782 D 72 32 4 3 KARAGANDA 668 D 105 54 7 0 TASHKENT 607 D 88 48 6 3 VINNITSA 807 D 122 67 88 DONETSK 284 D 809 370 45 8 NIKOLAYEV 169 D 84 40 50 VORONEZH 459 D 374 180 23 3 CHELYABINSK 581 D 202 93 11 4 PERM' 524 D 130 62 7 7 VENTSPILS 45 F 78 70 7 5 BREST 92 F 3313 2750 310 0 BAKU 245 F 11 0 2 0 NAKHODA 802 F 272 245 25 0 SEVASTOPOL ' 185 F 1120 935 105 5

SUB-TOTAL DOMESTIC DEMAND : 4206 2000 250 0 SUB-TOTAL EXPORT DEMAND : 4794 4000 450 0

TOTAL PIG IRON FINAL DEMAND : 9000 6000 7000 ------*NOTE : Type D represent domestic demand nodes and type F export deman d nodes .

SOURCES : Pig iron demand allocated by authors based on regional dat a for steel consumption by N . D . Lelyukhina, Ekonomicheskaya effektivnost ' razmeshcheniya chernoy metallurgii (Moscow : Nauka, 1973), pp . 18-27 ; V . P . Yevstigneyey, Effektivnost ' pazmeshcheniya mashinostroyeniya v vostochnykh i zapadnykh rayonakh SSSR (Moscow : Nauka, 1972), pp . 214-215 ; and 1i . Kh . Gankina, V . A . Petruchika & V . M . Fomina, eds ., Perevozki gruzov (Moscow : Transport, 1972), pp . 119-147 ; and weighted b y population data from Narodnoye khozyaystvo SSSR v 197.(Moscow0 : Statistika, 1971), pp . 37-45 ; Andrew R . Bond & Paul E . Lydolph , " Soviet Population Change and City Growth 1970-1979 : A Preliminar y Report , " Appendix 2 (compiled by Theodore Shabad) Soviet Geography 20 : 8 (October 1979), pp . 481-488 ; export demand based on regionall y aggregrated trade data from various years of Vneshnyaya torgovlya SSSR ,

-5 1 -

NCSEER March 8 3

the demand set down to forty domestic nodes plus six foreign expor t

nodes by agglomerating the demands around large regional centers .

For 1980 and for 1990 the original total domestic demand wa s

initially again apportioned to the set of 304 nodes as follows . First ,

the point source demands derived for the 304 locations for 1970 serve d

as base level demands for 1980 . Second, these base level demand value s

were augmented proportional to each city ' s absolute gain in populatio n

compared with the total absolute population gain of the entire 30 4

cities between 1970 and 1980 multiplied times the absolute increase i n

assumed domestic steel demand, 31 .365 million tons (see Table 21) .

Third, these newly arrived at 304 total demands were then agglomerate d

into the forty domestic demand centers illustrated in Table 21 .

In iterative fashion the 304 disaggregated demand figures for 198 0

served as base values for the determination of the 1990 point sourc e

demands . In this instance an absolute increase of 21 .085 million ton s

of steel needed to be allocated . First, the absolute gain in populatio n

for each of the 304 cities from 1980 to 1990 was projected b y

multiplying each city ' s 1980 population by a number equal to th e

quotient of the sum of two times a given city ' s growth rate between 197 0

and 1980 plus the given city's growth rate between the 1959 and 197 0

censuses divided by three . Second, the absolute increase in stee l

demand from 1980 to 1990 was assigned to each of the 304 node s

proportional to each ' s relative gain in absolute population . These

calculated amounts were next added to each node's 1980 demand figures .

Finally, these augmented 304 demand values were collapsed again into 4 0

re gional demand foci .

Although less than wholly reliable, such an assignment procedur e

-52 -

NCSEER March 8 3

Table 21 . ESTIMATED STEEL DEMAND BY REGION BY YEAR IN 1000 TON S

N T q Y DP PLACE OF STEEL DEMAND : E E* 1970 1980 199 0

LENINGRAD 24 D 4403 5630 632 5 RIGA 43 D 2696 3427 393 4 VILNIUS 74 D 89 138 16 6 MINSK 84 D 2732 4112 505 8 L ' VOV 110 D 1084 1582 188 0 ODESSA 158 D 2731 3654 426 3 KRASNODAR KRAI 191 D 1155 1469 1700 YEREVAN 228 D 3277 4409 513 9 ORDZHONIKIDZE 250 D 1269 1655 192 2 VOLGOGRAD 265 D 2439 3027 363 7 ROSTOV 271 D 1836 2232 251 8 VOROSHILOVGRAD 278 D 2731 3190 356 8 DONETSK 284 D 7409 8571 949 2 KIEV 313 D 2669 3940 471 7 ARKHANGEL'SK 388 D 919 1155 132 1 MOSCOW 415 D 12497 15241 17067 VORONF7H 459 D 5374 7677 909 6 PERM ' 524 D 3476 4344 493 7 SVERDLOVSK 549 D 5667 7154 811 5 KAZAN' 560 D 8065 11359 13389 CHELYABINSK 581 D 4969 6098 680 9 MAGNITOGORSK 586 D 1359 1601 177 4 ORENBURG 596 D 1676 2204 252 4 TASHKENT 607 D 1898 2830 340 1 MARI 622 D 217 304 35 9 DUSHANBE 630 D 232 332 398 FERGANA 634 D 578 810 954 ALMA-ATA 653 D 828 1148 133 2 KOKCHETAV 692 D 2248 3026 350 8 NOVOSIBIRSK 709 D 3430 4164 470 5 SEMIPALATINSK 723 D 529 809 962 LENINSK-KUZNETSK 735 D 2526 2766 294 7 KRASNOYARSK 750 D 810 1058 124 1 BLAGOVESHCHENSK 775 D 1228 1493 173 4 YUZHNO-SAKHALINSK 788 D 350 537 643 VLADIVOSTOK 800 D 769 1015 118 5 DNEPROPETROVSK 811 D 11534 14432 16486 AKTYUBINSK 838 D 3287 474 59 1 IRKUTSK 858 D 769 1015 134 2 NORIL'SK 888 D 126 183 21 1 NAUSHKI (MONGOLIA) 761 F 12 30 35 TERMEZ (AFGHANISTAN) 629 F 54 155 11 0 VENTSPILS (BALTIC PORTS) 45 F 252 130 18 0 UZHGOROD (E . EUROPE) 113 F 4922 5400 610 5 NAKHODKA (PACIFIC) 802 F 100 100 15 0 SEVASTOPOL ' (BLACK SEA) 185 . F 1660 1820 205 5

-53-

NCSEER March 8 3

Table 21 . continued :

YEAR 1970 1980 199 0

SUB-TOTAL DOMESTIC DEMAND FOR STEEL : 108900 140265 16135 0 SUB-TOTAL FOREIGN DEMAND FOR STEEL : 7000 7635 863 5

TOTAL DEMAND FOR STEEL : 115900 147900 16998 5 ------*NOTE : Type D represent domestic demand nodes and type F expor t demand nodes .

SOURCES : Steel demand allocated by authors based on regional data fo r steel consumption by N . D . Lelyukhina, Ekonomicheskaya effektivnost ' razmeshcheniya chernov metallurgii (Moscow : Nauka, 1973), pp . 18-27 ; V . P . Yevstigneyev, Effektivnost ' pazmeshcheniya mashinostroyeniya v vostochnykh i zapadnykh rayonakh SSSR (Moscow : Nauka, 1972), pp . 214-215 ; and M . Rh . Gankina, V . A . Petruchika & V . M . Fomina, eds ., Perevozki gruzov (Moscow : Transport,1972), pp . 119-147 ; and weighted b y population data from Narodnoye khozyaystvo SSSR v 1970 (Moscow : Statistika, 1971), pp . 37-45 ; Andrew R . Bond & Paul E . Lydolph, "Sovie t Population Change and City Growth 1970-1979 : A Preliminary Report, " Appendix 2 (compiled by Theodore Shabad) Soviet Geography 20 :8 (Octobe r 1979), pp . 481-488 ; export demand based on regionally aggregrated trad e data from various years of Vneshnyaya torgovlya SSSR .

seems to us to be about the best estimating method one could hope fo r

given the dearth of actual point-source steel demand data for the Sovie t

Union . Fortunately the location of steel smelting is more sensitive t o

the pull of raw materials inputs than markets . The use of thes e

regional demand estimates thus seems justified given the broad genera l

geographical insights hoped for from the analyses .

Having systematically presented and discussed both the model an d

the data gathered for use in this study, it is time to turn to a

discussion and interpretation of the results of this investigation .

S DISCUSSION OF THE RESULT

INTRODUCTION :

Detailed results of the computer simulation runs are portrayed in a

-54--

NCSEER March 8 3

number of maps and tables . While the data base allowed for th e

simulation of the Soviet iron and steel industry for three time periods ,

only the results for the years 1980 and 1990 will be presented here fo r

two reasons . First, the 1970 run represented a highly constraine d

historical formulation of the Soviet system essentially designed t o

check on the overall feasibility of the model . As previously mentione d

the solution in the 1970 model year was in fact infeasible because o f

insufficient scrap inputs . Thus, in the 1970 run of data through th e

model steel production was short by approximately five million tons whe n

it came up against this binding scrap constraint . Unfortunately ,

inadequate funds were available to rerun the 1970 data set with revise d specifications .

Since this study was intended to focus more on the "possible" an d

"plausible " emerging patterns in the Soviet ferrous metals sector, thi s

unfortunate omission is far less troubling than it might have been .

Furthermore, this discovery allowed us to take the precaution o f

precluding other possible infeasible runs by reformulating the 1980 an d

1990 runs to incorporate fictitious data nodes (Node 999) for al l

possible resource inputs . Accordingly, and as cited previously, thes e

fictitious nodes had very high upper limits on their resource supplie s

as well as very high costs attached to them (99,999 rubles/1000 tons o f

any given commodity) . Since this cost is higher than any of th e

combined raw material, transportation and/or process costs, the mode l

will thus only utilize resources from Node 999 as a last resort afte r

using up all other available supplies of a given input . In effect thi s

procedure allowed us to make checks on the overall feasibility of th e

published Soviet technical coefficient data and geographical productio n

data upon which our SISEM model is based without having to discove r

_55_

NCSEER March 8 3

other possible infeasibilities in a costly interative fashion .

It bears repeating that these various production levels and flo w

patterns represent the optimal production levels and flow patterns a s

calculated by the SISEM model, and not the actual production and flo w

values for either 1980 or obviously 1990 . At the same time, however ,

the upper and lower bounds on the production of the various raw materia l

inputs, intermediate products, and final products as elaborated in th e

various data tables in the previous data section place very definite an d

real empirical limits on the range of values that these modele d

production levels and flows may take . For example, in 1980 the minimum

level of coking coal production at Vorkuta (node 399 on Maps 1 and 2 ,

hereafter the number in parentheses following a place name refers to th e

node number of the place on all of the appropriate maps) based on Sovie t

data was 12 .0-million metric tons, while the maximum production leve l

was 14 .615—million tons . Thus, the model is constrained to calculat e

Vorkuta ' s coking coal production for 1980 as being at least 12 .0—millio n

tons, but not more than 14 .615—million tons (refer to data in Table 2 ,

pp . 18—19) . Accordingly, calculated production levels for al l

commodities in the SISEM model will either be at a given productio n

node ' s " lower limit " (LL), its "upper limit" (UL), or will enter th e

basic solution at some value in between these bounds (formally referre d

to as being in the basis and feasible and labelled "BS " for "basi s

solution") . In other words, the calculated production levels and flow

patterns are equivalent to the system—wide production levels an d

commodity flows which satisify the overall system intermediate and fina l

demands at the least total cost, subject, of course, to the variou s

technological, cost, and geographical production data constraints a s

elaborated in the previous (data) section .

-56-

NCSEER March 8 3

Equally important, some clarification may be needed with respect to th e

dual activity (DUAL) figures included in many of the following tables o f

results . Mathematically, the DUAL is simply equal to the expressio n

-dZ/db i , where -dZ represents the change in the total system cost Z ,

and dbi represents a unit change of the resource bi . Stated i n

another more easily interpreted way, the dual variable of a give n

constrained resource indicates how much the objective function Z woul d

change with a unit change in the given resource, provided the curren t

optimal basic solution remains feasible . Non-zero dual values occu r

only for variables which in the optimal solution are bound either by a

lower or upper production or capacity limit . For example, in 1980 th e

modeled coking coal production at Vorkuta (399) is bound by the lowe r

limit, whereas that at Urgal (779) is confined by its upper limit . Th e

dual activity value of the former is a negative 24 .12 rubles per ton an d

that of the latter is a positive 6 .16 rubles per ton . Simply state d

this means that if the minimum output at Vorkuta could be reduced by on e

ton, the net result would be a total system cost savings of 24 .1 2

rubles . Conversely, if the minimum output at Vorkuta (399) were to b e

increased by one ton, then total system costs would necessarily increas e

by 24 .12 rubles . Analogously, a one ton reduction in the upper limit o f

Urgal ' s production would impose an additional total system cost of 6 .1 6

rubles, while an increase of one ton would yield an identical 6 .1 6

rubles total cost savings . Dual activity values are identical to shado w

prices or opportunity costs . For instance, the 150,000 ton upper limi t

on Urgal ' s coking coal production in 1980 constitutes a 6 .16 rubles/to n

opportunity cost .

Finally, because additional excess supply is of no value, al l

variables which enter the optimal solution at values less than thei r

-57-

NCSEER March 8 3

upper limits implicitly represent "variables " or resources with exces s

supply (or capacity), and hence, their shadow prices or dual activitie s

are zero . Such variables are called basis variables . In the SISEM

model several of the basis variables as well as their dual activities o r

prices are themselves zero valued . This situation merely reflects th e

fact that such large problems as the SISEM simulation very commonly ten d

to have degenerate solutions, or more simply stated, optimal feasibl e

solutions where all of the variables are merely non-negative ( > 0) a s

opposed to non-degenerate optimal feasible solutions where all th e

variables are positive (>0) .

Rather than discussing the results of each year chronologically ,

the discussion will proceed by sequentially comparing and contrastin g

the coal, coke, iron ore, scrap, pig iron, and steel production level s

and commodity flow patterns for both years at the same time . Then, some

mention will be made of overall cost calculations and of the seve n

separate cost functions presented in the model section previously .

COKING COAL PRODUCTION AND FLOW PATTERNS :

The optimal levels of coking coal production for 1980 and 1990 a s

calculated by the SISEM model are listed in Table 22 . The coal flow s

from production nodes to coal demand nodes, that is, coking ovens an d

surrogate regional export demand nodes are enumerated in Table 23 . Bot h

coal production levels and flow patterns are illustrated by Map 1 fo r

1980 and by Map 2 for 1990 .

Coking Coal Production for 1980 (Table 22 and Map 1) :

Because coking coal has uses other than as a reducing agent in iro n

and steel making, it comes as no sur prise that only one of the thirtee n

-58-

NCSEER March 8 3

coal production nodes, namely Urgal (779), is modelled as producing a t

its upper production capacity limit in 1980 (see Table 2, p . 18 & Tabl e

22, p . 60) . In fact, had we not drastically reduced the minimum

production level for Stakhanov (885) from 1970 to 1980 as previousl y

discussed, the model would have been forced by the minimum constrain t

values to produce substantial quantities of " surplus " coal in 1980 . Al l

of the domestic sites determined by the model to be producing at thei r

lower limits had negative dual activities of at least 15 .63 rubles pe r

ton on up to a high of 26 .64 rubles per ton for Donetsk (284) . Th e

model's results clearly indicate that the imported Polish coal throug h

Grodno (81) is the most uneconomic source of Soviet coking coal having a

negative dual activity value of 52 .94 rubles per ton . At the same tim e

a systems cost savings of 6 .16 rubles could result from each additiona l

ton of output from Urgal (779) in the Soviet Far East for export to th e

Pacific market . . The upshot of these two results is that according t o

this model the Soviets should reduce their coking coal imports fro m

Poland and expand production at Urgal (779) for eport to Japan .

Coking Coal Flow for 1980 (Table 23 and Map 1) :

Despite the long hauls Kuznetsk coal is able to penetrate into th e

European heartland with Novokuznetsk (726) supplying coking coal to cok e

batteries at Cherepovets (836) in northwest Russia and coal exports t o

Sevastopol ' (185) in 1980 . The more expensive underground mines in th e

Donetsk basin (nodes 813, 930, 885, and 284) are confined to supplyin g

the Ukraine proper except for export flows through Mostiska (107) an d

Riga (43) and small supplies for Leningrad (24) and Moskva's (415) cok e

ovens . Vorkuta (399) supplies Novolipetsk ' s (955) ovens and shares wit h

Stakhanov (885) the supply for exports from Riga (43) and wit h

-59-

NCSEER March 8 3

------Table 22 . MODELLED OPTIMAL COKING COAL PRODUCTION IN 1000 TONS * ------1980 1990 Projecte d

COAL SITE : NODE PROD . AT DUAL PROD . AT DUA L

VORKUTA 399 12000 LL -24 .12 15535 UL 5 .5 1 KIZEL' 526 1170 .2 BS 0 .00 1518 UL 9 .3 2 TKVARCHELI 209 250 LL -21 .28 325 UL 10 .1 5 TKIBULI 215 250 LL -26 .06 325 UL 5 .3 7 KARAGANDA 668 10660 LL -22 .12 14025 UL 7 .1 9 KEMEROVO 739 9768 LL -21 .85 20350 UL 6 .7 8 NOVOKUZNETSK 726 30932 LL -22 12 44770 UL 6 .5 1 PAVLOGRAD 813 2655 LL -26 .30 1000 LL -0 .1 5 KAMENSK- 930 0 BS 0 .00 1000 LL -0 .4 6 SHAKHTINSKI Y STAKHANOV 885 1194 LL -26 .58 1000 LL -0 .1 4 DONETSK 284 34950 LL -26 .64 9230 .7 BS 0 .00 NERYUNGRI 963 NA - - 2363 .8 BS 0 .00 URGAL 779 150 UL 6 .16 1500 UL 10 .24 GRODNO (IMPORT) 81 600 LL -52 .94 600 LL -23 .8 9

104579 .2 113542 . 5 ______*PROD . = coal production 1000 tons , BS = in basis and feasible , EQ = nonbasis, artifical or fixed , UL = nonbasis, activity at upper limit , LL = nonbasis, activity at lower limit , NA = not available for 1980 use, an d DUAL = dual activity, mathematically -dZ/db i (RUB ./T) .

Novokuznetsk (726) the supply of coking coal for Cherepovets ' (836 )

ovens . The westwardly flow component of the old Kuznetsk-Ural Combin e

is still clearly evident in 1980 (see Map 1, p . 64) . The small Georgia n

coking coal production travels a short distance from Tkvarcheli (209 )

and Tkibuli (215) to the coking facilities at Rustavi (910) . Oddly, i n

the optimal flow pattern the majority of the coal imports from Polan d

(via Grodno, node 81) end up being destined for re-export through Rig a

(43) . Overall the 1980 coal flows determined by the model fit very wel l

the actual patterns as gleaned from reading various Soviet publications .

-60-

NCSEER March 8 3

Table 23 . MODELLED OPTIMAL COKING COAL FLOWS IN 1000 TON S

N N T YEQR 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

VORKUTA 399 CHEREPOVETS 836 K 1775 .5 33 . 1 NOVOLIPETSK 955 K 10204 13887 . 6 LENINGRAD 24 K - 93 . 3 RIGA 43 E 20 .5 27 1 SEVASTOPOL ' 185 E - 125 0

KIZEL' 526 GUBAKHA 526 K 1170 .2 151 0 CHEREPOVETS 836 K - 8

TKVARCHELI 209 RUSTAVI 910 K 250 12 8 ZHDANOV 286 K - 1 0 YASINOVATAYA 291 K - 187

TKIBULI 215 RUSTAVI 910 K 250 32 5

KARAGANDA 668 KARAGANDA 668 K 7550 906 0 MAGNITOGORSK 586 K 1600 345 5 NOVOTROITSK 594 K 1510 151 0

KEMEROVO 739 KEMEROVO 739 K 4530 453 0 SVERDLOVSK 548 K 1551 .7 151 0 CHELYABINSK 581 K 3686 .3 151 0 LENINGRAD 24 K - 57 . 7 KRIVOY ROG 172 K - 289 7 DNEPROPETROVSK 811 K - 604 ZAPOROZH'YE 179 K - 2078 . 3 KHAR'KOV 298 K - 15 1 NIZHNIY TAGIL 531 K - 151 0 ZHDANOV 286 K - 150 0 NAKHODKA 802 E - 4002

NOVOKUZNETSK 726 NOVOKUZNETSK 726 K 18120 22650 CHEREPOVETS 836 K 8191 9924 . 9 MOSKVA 415 K - 75 5 KRIVCY ROG 172 K - 918 3 NIZHNIY TAGIL 531 K 1510 - CHELYABINSK 581 K 1881 .3 - MAGNITOGORSK 586 K 26 .6 282 . 1 MOSTISTKA 107 E - 197 5 SEVASTOPOL' 185 E 1101 .6 — NAKHODKA 802 E 102 -

PAVLOGRAD 813 KRIVOY ROC 172 K 255 - ZAPOROZH ' YE 179 K - 941 . 7 DNEPROPETROVSK 811 K 604 - YASINOVATAYA 291 K - 58 . 3 MOSTISTKA 107 E 1796 -

-61-

NCSEER March 8 3

Table 23 . Continued : ------N NT YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

KAMENSK- 930 YASINOVATAYA 291 K - 69 8 SHAKHTINSKIY KOMMUNARSK 884 K - 30 2

STAKHANOV 885 LENINGRAD 24 K 151 - KHAR'KOV 298 K 151 - MOSKVA 415 K 755 - YASINOVATAYA 291 K - 100 0 RIGA 43 E 3 .5 - SEVASTOPOL' 185 E 133 .5 -

DONETSK 284 YASINOVATAYA 291 K 21069 9230 . 7 KRIVOY ROG 172 K 8805 - ZAPOROZH ' YE 179 K 3020 - KOMMUNARSK 884 K- 302 ZHDANOV 286 K 1510 - RIGA 43 E 244 -

NERYUNGRI 963 CHUL'MAN 963 K NA 2363 . 8

URGAL 779 NAKHODKA 802 E 150 150 0

GRODNO (IMPORT) 81 KALININGRAD 59 K 151 15 1 RIGA 43 E 449 44 9

Sub-total for coking : 100579 .2 104095 . 5 Sub-total exports : 4000 .0 9447 . 0

Total : 104579 .2 113542 . 5 ------*E = flow of coking coal to export nodes , K = flow of coking coal to coke plants, an d NA = not available for 1980 use .

------

Coking Coal Production for 1990 (Table 22 and Map 2) :

As shown by Table 22, the projection indicates that by 199 0

slightly more than half of the coking coal production of the entir e

Soviet Union may come from the Kuzbas in Western Siberia deep in th e

interior of the country (Kemerovo and Novokuznetsk, nodes 739 and 726 ,

respectively, in Table 22 and on Map 2), with Karaganda (668) i n

Kazakhstan and Vorkuta (399) in the extreme northeast European USS R

-6 2 -

NCSEER March 8 3

being other major producers . In constrast Donetsk (284) in the Donba s

in the southern Ukraine, near the black Sea and the western boundary o f

the Soviet Union, long the leading producer, may reduce its coking coa l

production by about two-thirds, a very dramatic drop in a relativel y

short period . Remembering that the highest minimum production limit a t

any given node for 1990 coal production was reduced to one-million ton s

(see Table 2, p . 18) and that Donets Basin coals were anticipate d

realistically as becoming even more expensive in relative terms, th e

model's indicated dramatic contraction in Ukrainian coal output is no t

surprising, nor are the corresponding increases from Siberian mines .

Now eight of the sites are producing at their upper limits, while onl y

three sites, Pavlograd (813), Kamensk-Shakhtinskiy (930), and Stakhano v

(885)--all in the Donbas--plus the imports via Grodno (81) are at thei r

lower limits (see Table 22) . These four nodes are thus the only coa l

nodes in 1990 having negative dual activities . Except for import s

through Grodno (-23 .89 rubles/ton), however, these opportunity costs ar e

far more modest than in 1980 ranging from a minus 0 .14 to a minus 0 .46

rubles per additional ton of coal . Alternatively stated, only a

relatively modest sum of 0 .14 to 0 .46 rubles could be saved for each to n

reduction in the minimum production levels at these mining sites .

On the other hand, the potential cost savings on the upwardl y

constrained mines range from a positive 5 .37 to 10 .23 rubles pe r

additional ton produced (see Table 22) . The greatest potential saving s

would have accrued from expanded capacity at Kizel ' (526), Tkvarchel i

(209) and Urgal (779) . None of these represents a realistic possibilit y

for major expansion however as all have only modest reserves .

-63-

NCSEER March 8 3

Coking Coal Flows for 1990 (Table 23 and Map 2) :

Reducing the minimum output levels for 1990, of course, allowed th e

model more flexibility to determine optimal production and flo w

patterns . The results are quite graphic . For example, as shown b y

Table 23, the coking coal for Krivoy Rog (17), the great iron ore fiel d

and ferrous metallurgy center in the south-central Ukraine, largel y

supplied by nearby Donetsk (284) in the 1980 model run, is indicated a s

receiving its supplies of coking coal in 1990 from the distant Kuzba s

coalfields of Novokuznetsk (726) and Kemerovo (739) . In fact, the rang e

of Siberian coals especially from Kemerovo (739) and Novokuznetsk (726 )

has expanded to include practically all of European Russia outside th e

Transcaucasus . Even a cursory viewing of Maps 1 and 2 readily reveal s

these vast increases in heavy long-distance hauls of Siberian coals b y

1990 . Then, to), Vorkuta's (399) coals are also projected as being abl e

economically to withstand greater transport costs . The tether on Donba s

coal is shortened still further . In the eastern Trans-Baykal regio n

along a spur of the BAN the Neryungri (963) deposit is indicated a s

producing 2 .36 million tons, but only for local use at a projecte d

Chul ' man coking plant .

In general, the projected optimal coking coal production and flo w

patterns for 1990 suggest that large increases in long-haul transpor t

costs could well be less significant than relative increases in th e

extraction costs of deep, underground Donbas coals . These ver y

long-hauls represent a considerable burden on the railroad system .

Donbas production is thus unlikely to decline by 1990 as precipitousl y

as the model indicates ; nonetheless, to the extent our prices reflec t

the future situation, then clearly coking coal expansion efforts may b e

concentrated east of the Urals .

-6 6 -

NCSEER March 8 3

COKE MANUFACTURE AND FLOW PATTERNS :

Coking Process for 1980 (Table 24 and Map 3) :

The domestic destinations (i .e ., non-export nodes) of the cokin g

coal flows just discussed represent coke manufacturing facilities . Fo r

1980 there are twenty-two such nodes while two potential cokin g

locations, Chul'man (963) and Urgal (779) both in the extrem e

southeastern part of the USSR, are added to the set for 1990 (see Tabl e

24 and Maps 3 and 4) . As shown in Table 24 thirteen of these twenty-tw o

1980 coke ovens sites enter the optimal solution at their lower bound s

and only three at their upper bounds . Not surprisingly all of th e

latter three are east of the Urals, Karaganda (668) in Kazakhstan ,

Kemerovo (739) and Novokuznetsk (726) in Western Siberia . Three of th e

six Urals ' coking sites, Gubakha (526), Nizhniy Tagil (531) an d

Novotroitsk (594), enter at their lower limits . The other three ,

Sverdlovsk (548), Chelyabinsk (581) and Magnitogorsk (586), enter th e

basis, but all three with values much closer to their lower limi t

(one-million tons) than their upper limits (ten-million tons for th e

first two and fifteen-million for Magnitogorsk, see Table 3, p . 20) .

The three remaining unaccounted for sites are wide ranging Europea n

nodes which all enter the basis, Novolipetsk (955) in the Centra l

Region, Yasinovataya (291) in the Donbas and Rustavi (910) in Gruziya .

Overall coke production required for ferrous metallurgy needs for 198 0

of 66 .6-million tons, of course, is significantly less than the actua l

total Soviet coke production because other sectors of the econom y

besides ferrous metallurgy use coke .

Both the positive and negative dual activities or opportunity cost s

are much smaller for coke than for coal production . For example, if th e

upper bounds on the Karaganda site (668) were relaxed a cost savings o f

-67-

NCSEER March 8 3

Table 24 . MODELLED OPTIMAL COKE PRODUCTION IN 1000 TONS *

1980 1990 PROJECTE D

COKE PLANT : NODE PROD AT DUAL PROD AT DUAL

LENINGRAD 24 100 LL -6 .68 100 LL -6 .4 7 CHEREPOVETS 836 6600 LL -6 .06 6600 LL -4 .6 4 MOSKVA 415 500 LL -3 .47 500 LL -4 .7 1 KRIVOY ROG 172 6000 LL -3 .91 8000 LL -4,8 7 DNEPROPETROVSK 811 400 LJ . -C .25 400 LL -5 .2 5 ZAPOROZH'YE 179 2000 LL -0 .24 2000 LL -5 .6 2 YASINOVATAYA 291 13953 BS 0 .00 7400 LL -5 .9 9 KOMMUNARSK 884 200 LL -0 .10 200 LL -5,6 0 ZHDANOV 286 1000 LL -0 .36 1000 LL -6 .4 2 NOVOLIPETSK 955 6757 .6 BS 0 .00 9197 .1 BS 0 .00 KALININGRAD 59 100 LL -5 .18 100 LL -5 .1 9 KHAR'KOV 298 100 LL -0 .63 100 LL -5 .1 8 GUBAKHA 526 775 LL -0 .16 1000 LL -1,2 4 NIZHNIY TAGIL 531 1000 LL -0 .12 1000 LL -0 .6 6 SVERDLOVSK 548 1027 .6 BS 0 .00 1000 LL -0 .5 4 CHELYABINSK 581 3687 .2 BS 0 .00 1000 LL -0,05 MAGNITOGORSK 586 1077 .2 BS 0 .00 2474 .9 BS 0 .0 0 NOVOTROITSK 594 1000 LL -1 .05 1000 LL -1 .0 5 RUSTAVI 910 331 .1 BS 0 .00 300 LL -2 .7 3 KARAGANDA 668 5000 UL 1 .04 6000 UL 1 .04 KEMEROVO 739 3000 UL 1 .64 3000 UL 1 .4 8 NOVOKUZNETSK 726 12000 UL 1 .77 15000 UL 1 .8 9 CHUL'MAN 963 NA - - 1565 .4 BS 0 .0 0 URGAL 779 NA - - 0 BS 0 .00

Total : 66608 .7 68937 . 4 .-.-- .-.. ------*PROD = coke manufacture in 1000 tons , BS = in basis and feasibl e EQ = nonbasis, artifical or fixed , UL = nonbasis, activity at upper limit , LL = nonbasis, activity at lower limit , NA = not available for 1980 use, an d DUAL = dual activity, mathematically -dZ/db i (RUB ./T) .

1 .04 rubles per ton of coke is indicated . The savings for Kemerovo an d

Novokuznetsk from similar relaxations would yield 1 .64 and 1 .77 ruble s

per ton, respectively . The greatest economic penalties for mimimum cok e

production levels occur in the central and northwest European parts o f

the country plus Krivoy Rog (172) (see Table 24) . Except for Krivoy Rog

(172) all coking sites in the Ukraine and Urals have cost penalties i n

-68-

NCSEER March 8 3

the range of only 0 .10 to 1 .05 rubles per ton associated with thei r

minimum production constraints .

Projected Changes in Coke Manufacture between 1980 and 1990 :

Considerable stability exists between the 1980 and 1990 productio n

and flow patterns . First, the same three eastern locations ar e

producing at their upper limits, Karaganda (668), Kemerovo (739) an d

Novokuznetsk (72b) . Second, all sites which were calculated a s

producing at their lower limits in 1980 are also in 1990 . Three

locations which were in the basis in 1980, Yasinovataya (291) in th e

Donbas and Chelyabinsk (581) and Sverdlovsk (548) in the Urals, becom e

nonbasis variables with modelled optimal production constrained at thei r

lower limits in 1990 . All three suffer absolute declines in cok e

manufacture, the first two by large quantities . The largest absolut e

increases occur at Krivoy Rog (172) (forced by the anticipated larg e

increase in the minimum production level there), Novolipetsk (955) ,

Magnitogorsk (586), Karaganda (668), Novokuznetsk (726) and th e

potentially viable new coke batteries situated at Chul ' man (963) . I n

the Transcaucasus Rustavi ' s (910) output drops slightly to its lowe r

bound .

While the dual activities indicated for Kazakhstan and the Kuzba s

are stable, most coking locations experience increasingly negative dua l

activities between 1980 and 1990 . Essentially these dual costs divid e

the country into distinct regions . All dual activity values for th e

Urals are low (-0 .05 to -1 .24 rubles/ton) whereas those in the rest o f

European Russia north of the Caucasus with the exception of Novolipets k

(955) are relatively high (-4 .64 to -6 .47 rubles/ton) . Rustavi ' s (910 )

value is intermediate at -2 .73 rubles/ton .

-69—

NCSEER March 8 3

Coke Flow Patterns for 1980 and 1990 (Table 25 and Maps 3 and 4) :

At the regional scale the flow patterns between 1980 and 1990 ar e

stable (refer to Maps 3 and 4) . Most changes that occur, of course, ar e

the result of model determined changes in the processes, outputs, an d

locations of coke consuming activities . The very modest absolut e

increase in total coke production of 2 .329-million (see Tables 24 or 25 )

compared with increases in steel output of slightly over 22-million ton s

over the decade in question clearly implies technological changes withi n

the production system .

Table 25 . MODELLED OPTIMAL COKE FLOWS IN 1000 TONS * ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTED ORIGIN : E DESTINATION : E E* 1980 199 0

LENINGRAD 24 KOLPINO 27 T 1 .3 0 . 7 ZHLOBIN 88 U - 83 . 9 ZHLOBIN 88 H - 415. TULA 433 P 98 .7 -

CHEREPOVETS 836 CHEREPOVETS 836 J 1932 - CHEREPOVETS 836 H - 306 . 6 CHEREPOVETS 836 B - 6 . 4 CHEREPOVETS 836 P 4237 .4 628 7 TULA 433 B 0 .4 - TULA 433 P 55 .3 - KRIVOY ROG 172 J 134 .9 - KULEBAKI 890 J 235 - GOR'KIY 505 B 6 -

MOSKVA 415 TULA 433 P 500 - ZHLOBIN 88 U - 499 . 9 GOR'KIY 505 B - 0 . 2

KRIVOY ROG 172 KRIVOY ROG 172 J 414 .9 - KRIVOY ROG 172 B - 90 KRIVOY ROG 172 P 5585 .1 791 0

DNEPROPETROVSK 811 DNEPROPETROVSK 811 J 400 - DNEPROPETROVSK 811 P - 40 0

-70-

NCSEER March 8 3

Table 25 . Continued :

N N T YEAR 0 0 Y D D P OPTIMAL PROJECTED ORIGIN : E DESTINATION : E E* 1980 199 0

ZAPOROZH'YE 179 ZAPOROZH'YE" 179 J 80 5 ZAPOROZH'YE 179 P 1195 200 0

YASINOVATAYA 291 DNEPROPETROVSK 811 J 566 — DNEPROPETROVSK 811 B 7 .5 — DNEPROPETROVSK 811 P 5344 .5 226 7 DONETSK 284 H 117 24 3 DONETSK 284 J 2415 — KRIVOY ROG 172 P 447 .1 — KOM UNARSK 884 J 644 — KOMMUNARSK 884 B 3 — KOMMUNARSK 884 P 2525 234 5 TAGANROG 274 J 80 .5 — MAKEYEVKA 283 P 1803 .4 254 5

KOMMUNARSK 884 KOMMUNARSK 884 P 200 200

ZHDANOV 286 KRIVOY ROG 172 P 321 .8 964 ZAPOROZH'YE 179 P 678 .2 3 6

NOVOLIPETSK 995 NOVOLIPETSK 955 B 30 — NOVOLIPETSK 955 P 4590 3733 . 2 STARYY OSKOL 462 U 300 — STARYY OSKOL 462 B 4 .5 27 . 7 KRIVOY ROG 172 J 1523 .8 — KRIVOY ROG 172 P — 385 1 SUMGAIT 244 J 204 .5 — RUSTAVI 910 P 104 .9 107 . 2 ZHLOBIN 88 U — 580 . 2 TULA 433 P — 610 . 8 DNEPROPETROVSK 811 P — 287

KALININGRAD 59 KRIVOY ROG 172 J 100 — ZHLOBIN 88 U — 100

KHAR'KOV 298 KRIVOY ROG 172 P 100 — DNEPROPETROVSK 811 P — 10 0

GUBAKHA 526 CHUSOVOY 528 J 322 - CHUSOVOY 528 P 453 509 SEROV 532 P - 49 1

NIZHNIY TAGIL 531 NIZHNIY TAGIL 531 B 12 .5 — NIZHNIY TAGIL 531 H — 101 . 1 NIZHNIY TAGIL 531 P 987,5 880 . 9 SEROV 532 P — 1 8

SVERDLOVSK 548 NIZHNIY TAGIL 531 P 935 .6 646, 1

-71-

NCSEER March 83

Table 25 . Continued :

N N T YEAR 0 0 Y D D F OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

SVERDLOVSK 548 CHUSOVOY 528 P 92 - SATKA 576 P - 353 . 9

CHELYABINSK 581 CHELYABINSK 581 J 811 .9 - CHELYABINSK 581 T 33 .2 - CHELYABINSK 581 B 8 .1 - CHELYABINSK 581 P 2234 .5 794 SATKA 576 P 599 .5 206

MAGNITOGORSK 586 MAGNITOGORSK 586 P 1077 .2 2474 . 9

NOVOTROITSK 594 NOVOTROITSK 594 B 9 - NOVOTPOITSK 594 P 872 809 . 6 MAGNITOGORSK 586 J 119 - MAGNITOGORSK 586 P - 190 . 4

RUSTAVI 910 RUSTAVI 910 P 331 .1 30 0

KARAGANDA 668 KARAGANDA 668 B 18 - KARAGANDA 668 P - 2443 . 2 MAGNITOGORSK 586 J 2424 .8 - MAGNITOGORSK 586 B 3 - MAGNITOGORSK 586 H - 390 MAGNITOGORSK 586 P 2554 .2 3166 . 8

KEMEROVO 739 NIZHNIY TAGIL 531 J 1066 .8 - NIZHNIY TAGIL 531 P 1366 .9 - NOVOSIBIRSK 709 T 21 .3 - SEROV 532 P 545 - CHELYABINSK 581 H - 232 CHELYABINSK 581 P - 1292 . 9 MAGNITOGORSK 586 P - 1475 . 1

NOVOKUZNETSK 726 NOVOKUZNETSK 726 J 724 .5 - NOVOKUZNETSK 726 B 22 .2 - NOVOKUZNETSK 726 P 5431 .4 254 5 PETROVSK - ZABAYKAL'SK 951 T 13 - KOMSOMOL ' SK - na-ANURE 783 T 14 .3 - BARNAUL 718 T - 78 . 3 BARNAUL 718 U - 11122 . 8 TAYSHET 754 T - 17 . 2 NIZHNIY TAGIL 531 P 2566 - MAGNITOGORSK 586 P 3228 .6 1236 . 8

CHUL ' MAN 963 CHUL'MAN 963 U - 1565 . 4

URGAL 779 NONE - - - -

-72-

NCSEER March 8 3

Table 25 . Continued :

Total coke flows : 66609 .8 68937 . 4

*H = open hearth process using mainly gas , J = open hearth process using no gas , A = electric furnaces, scrap only, no pig iron , T = electric furnaces, less scrap and some pig iron , B = basic oxygen process , U = direct reductio n steel process , P = pig iron process, an d NA = not available f or 1980 use .

-73-

NCSEER March 8 3

IRON ORE PRODUCTION AND FLOW PATTERNS :

Results of the two simulations runs for iron ore production and or e

flows are listed in Tables 26 through 29 and schematically illustrate d

by Maps 5 and 6 . For 1930 a total of nineteen possible low-grade or e

and twelve high-grade ore mining centers exist . Six new potentia l

low-grade workings and one high-grade site are added for 1990 .

Iron Ore Production for 1980 (Table 26 and Map 5) :

A distinct optimal production pattern for 1980 emerges from Tabl e

26 with respect to low-grade versus high-grade workings ; namely, th e

low-grade sites are shown in general as operating at their constraine d

minimum output levels, while the high-grade sites are much more likel y

to be calculated as producing at their constrained maximum capacit y

limits . For example, eight of the high-grade deposits are functionin g

at their maximum production levels in 1980 . These include the two Kurs k

Magnetic Anomaly (KMA) mining districts of Zheleznogorsk (977) an d

Gubkin (460) plus Tula (433) and Lipetsk (955) in central Europea n

Russia, Magnitogorsk (586) and Peschansk (533) in the Urals, Rudny y

(682) in northwest Kazakhstan and Tashtagol (725) in West Siberia .

Interestingly enough, the high-grade production at Krivoy Rog (173) ,

while at its modelled lower limit of 45-million tons has a very low dua l

value of -0 .08 rubles per ton . This very small negative opportunit y

cost suggests that, in fact, Krivoy Rog ' s actual 1980 production rang e

of 45 to 46-million tons of high-grade ore (see Table 4, p . 21), ma y

well be very close to optimal .

A similar argument can be made for all of the high-grade working s

for 1980 except Magnitogorsk (586) and Karazhal (666) . In other words ,

all of the high-grade sites have model indicated production levels at o r

- 7 6 -

NCSEER March 8 3

very close to their actual 1980 production figures except Karazhal (666 )

in Kazakhstan (zero model production versus three to four million ton s

actual output in 1980) and, at the same time, all have quite smal l

levels of dual activity except Magnitogorsk (586) which potentiall y

could yield the highest systems cost savings (5 .66 rubles per additiona l

ton of output) of any iron ore site . Unfortunately for the Soviets ,

expansion at Magnitogorsk (586) does not appear to be physicall y

possible because of " Magnetic Mountain ' s " nearly depleted high-grade or e

reserves .

In contrast to the high-grade sites none of the low-grade sites i s

at its upper limit in 1980 . While only one of the high-grade workings ,

Karazhal (666) in Kazakhstan fails to yield a positive output, three o f

the low-grade sites, Kachkanar (911) in the northern Urals, Dashkesa n

(243) in the Caucasus and Irba (980) in West Siberia are shown a s

optimally not producing any ore in 1980 . An additional twelve low-grad e

nodes are indicated as having their production constrained to their se t

minimum levels . These results are not surprising as the combination o f

extraction costs and necessary enrichment costs make the low-grade ore s

less competitive in general than the direct shipping or high-grade ores .

One site, Krasnokamensk (979) located east of the Kuzbas, enters as a

basis variable with a value equal to its upper limit ; however, thi s

seems clearly to be simply the result of the fact that its crude or e

output of one-million tons just equals the minimum constraint set on or e

enrichment at Krasnokamensk (979) for 1980 (see Table 5, p . 23) . Mor e

recent production data suggest that the actual ore capacity-output rang e

for Krasnokamensk (979) should have probably been set within the rang e

of 0 .5-million to 1 .0-million tons annually instead of the 1 .0-millio n

to 4 .28-million tons range actually used in the computer runs (Table 4 ,

- 7 7 -

NCSEER March 8 3

p . 21, contains this lower range rather than the higher one used in th e

1980 simulation) .

Based on the 1980 run expansion of the operations at the two ric h

KMA deposits near Gubkin (460) and Zheleznogorsk (977) seems mildl y

warranted (positive dual activities of 0 .74 and 0 .89 rubles in saving s

for each additional ton mined) . Such expansion, however, is ver y

problematic due to hydrogeologic problems associated with these dee p

underground mines . In a somewhat surprising contrast, the open-pit ,

low-grade mines at these two KMA deposits have the second and thir d

highest negative dual values (-3 .63 and -4 .28) of the entire ore-minin g

network for 1980 . This implies that no expansion is warranted and tha t

contraction would be desirable .

Projected Optimal Ore Production for 1990 (Table 26 and Map 6) :

The dichotomy between the production activity at low-grade an d

high-grade ore sites is even more evident in the 1990 optima l

projections . Probably the one most apparent pattern (see Table 26) i s

that all thirteen of the high-grade operations are indicated a s

functioning at their upper bounds including the long delayed, but no w

(i .e ., in 1990) operating Kachar (683) deposits . Furthermore, th e

positive opportunity costs imposed by these upper bounds have increase d

very significantly for all thirteen mining centers, ranging from at lo w

of 1 .32 rubles per ton for Kachar (683) to a high of 8 .68 rubles per to n

for Magnitogorsk (586) . In contrast only three of the low-grade sites ,

Abaza (730) and Gornaya Shoriya (727) in the Kuzbas and the propose d

Kostamuksa (361) mining center in Karelia, are calculated as working a t

their maximum sec capacities .

Cursory inspection indicates that the set of low-grade iron or e

- 7 8-

NCSEER March 83

------Table 26 . OPTIMAE IRON ORE PRODUCTION BY YEAR IN 1000 TONS * ------1980 1990 Projecte d

PLACE : NODE 1980 AT DUAL 1990 AT DUA L

KACHKANAR 911 L 0 BS 0 .00 1000 EL -1 .1 4 GOROLBAGODAT 530 L 7000 LL -3 .60 1000 LL -1 .0 5 ALAPAYEVO 541 L 200 LL -3 .62 100 BS 0 .0 0 BAKAL 577 L 4000 LL -3 .51 5993 .6 BS 0 .0 0 NOVORUDNAYA 595 L 1000 LL -3 .37 2580 .8 BS 0 .0 0 TAYEZHNOYE 964 L NA - - 2671 .7 BS 0 .0 0 GARIN 774 L NA - - 1400 BS 0 .0 0 UDA-SELEMDZHA 961 L NA - - 0 BS 0 .0 0 OLENOGORSK 370 L 9500 LL -4 .85 3500 BS 0 .0 0 KOSTAMUKSA 361 L NA - - 8900 UL 0 .2 5 PUDOZHGORSK 381 L NA - - 3000 BS 0 .0 0 ZHELEZNOGORSK 976 L 5000+ LL -3 .63 11086 .2 BS 0 .0 0 GUBKIN 891 L 14000+ LL -4 .28 23205 .9 BS 0 .0 0 KREMENCHUG 302 L 8945 .8 BS 0 .00 13354 .5 BS 0 .00 KRIVOY ROG 172 L 60000 LL -2 .78 46987 BS 0 .00 KERCH' 187 L 4700 LL -2 .33 7500 BS 0 .00 DASHKESAN 243 L 0 BS 0 .00 1290 .4 BS 0 .00 RUDNYY 873 L 16650 LL -3 .20 1000 LL -2 .3 1 KORSHUNOVSK 756 L 6000 BS 0 .00 1000 BS 0 .00 RUDNOGORSK 921 L NA - - 0 BS 0 .0 0 ABAZA 730 L 1000 LL -0 .09 1500 UL 0 .04 GORNAYA SHORIYA 727 L 5054 .2 BS 0 .00 8000 UL 0 .0 7 TEYA 978 L 1400 LL -0 .57 1286 .2 BS 0 .0 0 KRASNOKAMENSK 979 L 1000 BS 0 .00 500 LL -0 .9 4 IRBA 980 L 0 BS 0 .00 500 LL -1 .0 2 ABAZA 729 H 1433 .2 BS 0 .00 1500 UL 4 .2 4 KRIVOY ROG 173 H 45000 LL -0 .08 40000 UL 4 .4 9 GUBKIN 460 H 12000 UL 0 .74 12000 UL 4 .30 ZHELEZNOGORSK 977 H 9000 UL 0 .89 9000 UL 4 .46 DNEPRORUDNOYE 180 H 2254 .7 BS 0 .00 5000 UL 4 .74 MAGNITOGORSK 586 H 4000 UL 5 .66 4000 UL 8 .68 KACHAR 683 H NA - - 3000 UL 1 .3 2 TULA 433 H 300 UL 0 .17 50 UL 4 .1 9 LIPETSK 955 H 300 UL 0 .08 50 UL 4 .09 RUDNYY 682 H 3500 UL 0 .18 3500 UL 2 .1 9 PESCHANSK 533 H 540 UL 1 .00 540 UL 3 .4 7 KARAZHAL 666 H 0 BS 0 .00 6000 UL 2 .1 1 TASHTAGOL 725 H 2000 UL 0 .31 2000 UL 4 .9 1

Sub-total low-grade ore : 145450 .0 147356 . 3 Sub-total high-grade ore : 80327 .9 86640 . 0

Total ore production: 225777 .9 233996 . 3 ------*NOTE : L = low-grade ore ; H = high-grade ore ; BS = in basis an d feasible ; UL = nonbasis, activity at upper limit ; LE = nonbasis , activity at lower limit ; NA = not available for use in 1980 ; and DUAL = dual activity, mathematically -dZ,/db i (RUB ./T) . +NOTE : Gubkin produces 14000 tons total, but ships 1000 tons t o enrichment plant at Zheleznogorsk ; thus, Zheleznogorsk produces 500 0 tons of low-grade ore, but ships 6000 tons of enriched ore to fina l demand nodes .

- 7 9-

NCSEER March 8 3

centers are functioning much closer to optimality (i .e ., very small dua l

activities or opportunity costs) in 1990 than 1980. On the one hand ,

this is true ; but, on the other hand, it is a logical consequence o f

having reduced the highest minimum ore production level to one-millio n

tons annually for the 1990 simulation . As a result, instead of twelv e

ore nodes being constrainted to in effect " over produce " as in the 198 0

run, only five are forced to do so in 1990 . Of the six new or potentia l

low-grade ore nodes added for the 1990 run, five constitute part of th e

basis and the sixth, Kostamuksa (361) in Karelia as noted above enter s

the solution at its upper limit . Of these five, however, Uda-Selemdzh a

(961) in the Far East and Rudnogorsk (921) in Irkutsk Oblast nea r

Korshunovsk (or Zheleznogorsk-Ilimskiy) (756) optimally produce zer o

tons of ore .

Ore Enrichment (Table 27) :

Low-grade ores first have to be enriched before they may b e

substituted freely--in a physical sense--for rich ores to satisf y

export, pig iron and/or steel making input demands . Since enrichment i s

assumed to occur at all the low-grade mine sites, neither flows no r

transport costs are considered in the model for intermediate enrichmen t

processes in which all of the low-grade ore at a given mine site i s

enriched on-site . On the other hand, as mentioned previously (refer t o

p . 23), process and energy costs are incurred by enrichment . Also, i f

as a result of benefication, capacity constraints, and/or overall syste m

cost factors, some of a given node ' s low-grade output can not b e

processed locally, then, any and all necessary flows to other enrichmen t

facilities are subject to transport costs . The only instance of thi s

situation occurs in 1980 when the Zheleznogorsk (976) ore enrichmen t

combine receives one-million tons of ore from the Gubkin (891) mine i n

- 80-

NCSEER March 8 3

Table 27 . OPTIMAL IRON ORE ENRICHMENT BY YEAR IN 1000 TONS * ------1980 1990 Projecte d

PLACE : NODE : PROD . AT DUAL PROD . AT DUA L

KACHKANAR 911 0 BS 0 .00 1000 LL 0 .00 GOROLBAGODAT 530 7000 LL 0 .11 1000 BS 0 .0 0 ALAPAYEVO 541 200 BS 0 .00 100 BS 0 .00 BAKAL 577 4000 BS 0 .00 5993 .6 BS 0 .00 NOVORUDNAYA 595 1000 BS 0 .00 2580 .8 BS 0 .00 TAYEZHNOYE 964 NA - - 2671 .7 BS 0 .00 GARIN 774 NA - - 1400 BS 0 .00 UDA-SELEMDZHA 961 NA - - 0 BS 0 .00 OLENOGORSK 370 9500 BS 0 .00 3500 BS 0 .00 KOSTAMUKSA 361 NA - - 8900 UL 0 .2 2 PUDOZHGORSK 381 NA - - 3000 UL 0 .7 1 ZHELEZNOGORSK 976+ 6000 LL -0 .44 11086 .2 BS 0 .00 GUBKIN 891+ 13000 BS 0 .00 23205 .9 BS 0 .00 KREMENCHUG 302 8945 .8 BS 0 .00 13354 .5 BS 0 .00 KRIVOY ROG 172 60000 LL -1 .98 46987 BS 0 .00 KERCH' 187 4700 LL -2 .16 7500 UL 0 .1 7 DASHKESAN 243 0 BS 0 .00 1290 .4 BS 0 .00 RUDNYY 873 16650 LL -1 .09 1000 BS 0 .00 KORSHUNOVSK 756 6000 LL -7 .25 1000 LL -3 .3 7 RUDNOGORSK 921 NA - - 0 BS 0 .00 ABAZA 730 1000 LL -4 .40 1500 UL -0 .2 5 GORNAYA SHORIYA 727 5054 .2 BS 0 .00 8000 UL -0 .2 1 TEYA 978 1400 LL -3 .86 1286 .2 BS 0 .00 KRASNOKAMENSK 979 NA - - 500 LL -0 .1 4 IRBA 980 NA - - 500 LL 0 .00

Total low-grade ore : 145450 147356 . 3 ------*NOTE : BS = in basis and feasible ; UL = nonbasis, activity at uppe r limit ; LL = nonbasis, activity at lower limit ; NA = not available fo r use in 1980 ; and DUAL = dual activity, mathematically -dZ/db i (RUBLES/TON) .

+NOTE : Gubkin produces 14000 tons total, but ships 1000 tons t o enrichment plant at Zheleznogorsk ; thus, Zheleznogorsk produces 500 0 tons of low-grade ore, but ships 6000 tons of enriched ore to fina l demand nodes .

order to satisify its minimum production constraint of six-million tons .

This flow is further forced by the combination of the lower limits o n

low-grade ore production at Zheleznogorsk (976) and Gubki n

(891)(five-million and fourteen-million tons respectively, see Table 4 ,

-8 1 -

NCSEER March 8 3

p . 21) and the implicit relatively low demand for ore within Centra l

European Russia .

The model predicted enriched ore output by site and year ar e

enumerated in Table 27 . In 1980 the highest negative dual activitie s

are located in the east . Korshunovsk ' s (756) lower bounded productio n

level imposes the highest opportunity cost of -7 .25 rubles per ton .

Abaza (730), Teya (973), and Krasnokamensk (979) all located east of th e

Kuzbas follow with values of -4 .40, -3 .86 and -2 .71 rubles per ton ,

respectively . Kerch ' (187), Krivoy Rog (172), and Rudnyy (873) are al l

forced to produce above optimal levels in 1980, All of the othe r

eleven enrichment sites are producing at or close to optimal levels i n

1980, including the two complexes at Gorolbagodat (530)(a synthetic nam e

given to the nodal centroid of a group of ore operations in the Ural s

near Nizhniy Tagil) and Zheleznogorsk (976) which enter the solution a t

their lower limits . No ore benefaction facilities are operating a t

their set maximum capacity in 1980, while half of the eighteen plant s

are constrained by their lower limits . Similar to the pattern for crud e

ore production in 1990, the enrichment sub-system function s

significantly closer to the optimum possible with eighteen of the no w

expanded set of twenty-five enrichment nodes possessing zero opportunit y

costs . Except for Korshunovsk (756) (-3 .37 rubles/ton) and Pudozhgors k

(381) (0 .71 rubles/ton) the opportunity costs associated with th e

bounded production ranges of these crude ore processing installations i s

very small . But again, one must keep in mind that the relaxation of th e

lower production limits for both crude ore and enrichment has much to d o

with these lower opportunity figures for both crude and processed ore i n

1980 compared to 1990 .

-82-

NCSEER March 8 3

Changes in High-Grade Ore Output from 1980 to 1990 (Table 26) :

Overall production of high-grade ore by node between 1980 and 199 0

appears quite stable with four possible exceptions . First, Krivoy Rog' s

(173) production is reduced by five-million tons absolute simply as a

result of presumed declining readily available reserves of high-grad e

ore by 1990 . Second, Karazhal ' s (666) optimal calculated productio n

jumps from zero in 1980 to six-million tons in 1990 . Our estimate i s

that actual production here in 1980 was three-to-four million tons .

Third, Dneprorudnoye ' s (180) projected optimal output more than double s

to five-million tons in 1990 . Finally, presumed declining reserve s

modelled by lower maximum production levels accounts for the order o f

magnitude smaller absolute ore output reductions at Tula (433) an d

Lipetsk (955) . By way of review the high-grade nodes of Karazhal (666 )

(+6 million tons) and Kachar (683) (+3 million tons) both in Kazakhsta n

have a combined absolute increase in ore production equivalent to 109 . 5

per cent of the entire ore network ' s modelled 1980 to 1990 increase .

Changes in Low-Grade Ore Production from 1980 to 1990 (Table 28) :

Changes in the absolute levels of low-grade crude iron ore outpu t

and of the enrichment of iron ore at individual nodes, however, are muc h

more dramatic . They are illustrated below in descending order from ne t

losers to net gainers in Table 28 . For reasons discussed previously th e

two output columns (of output of low grade ore and of the enrichment o f

that ore) are identical except for Gubkin (891) and Zheleznogorsk (976) .

Rudnyy (873) in Kazakhstan and Krivoy Rog (172) in the Ukraine have th e

two biggest production declines followed by Olenogorsk (370) in the Kola

Peninsula, Corolbagodat (530) in the Urals and Korshunovsk (756) i n

Irkutsk Oblast . The largest indicated expansions occur in the KMA ,

Karelia, Kremenchug (302) and Kerch ' (187) in the Ukraine, Gornay a

-8 3 -

NCSEER March 8 3

Shoriya (727) in West Siberia and the new Tayezhnoye (964) deposit s

along the BAM . In sum the net change in the overall low-grade or e

------Table 28 . CHANGES IN LOW-GRADE AND ENRICHED ORE IN 1000 TONS ------1990 MINUS 1980 1990 MINUS 1980

PLACE : NODE : LOW-GRADE ORE ENRICHED OR E

RUDNYY 873 -15650 -15650 KRIVOY ROG 172 -13013 -1301 3 OLENOGORSK 370 -6000 -6000 GOROLBAGODAT 530 -6000 -6000 KORSHUNOVSK 756 -5000 -5000 KRASNOKAMENSK 979 -500 -500 TEYA 978 -113 .8 -113 . 8 ALAPAYEVO 541 -100 -100 UDA-SELEMDZHA 961 0* 0 RUDNOGORSK 921 0* 0 ABAZA 730 +500 +50 0 IRBA 980 +500 +50 0 KACHKANAR 911 +1000 +100 0 DASHKESAN 243 +1290 .4 +1290 . 4 GARIN 774 +1400 +140 0 NOVORUDNAYA 595 +1580 .8 +1580 . 6 BAKAL 577 +1993 .6 +1993 . 6 TAYEZHNOYE 964 +2671 .7 +2671 . 7 KERCH' 187 +2800 +280 0 GORNAYA SHORIYA 727 +2945 .8 +2945 . 8 PUDOZHGORSK 381 +3000 +300 0 KREMENCHUG 302 +4408 .7 +4408 . 7 ZHELEZNOGORSK 976 +6086 .2 +5086 . 2 KOSTAMUKSA 361 +8900 +890 0 GUBKIN 891 +9205 .9 +10205 . 9

Net change to totals : +1906 .3 +1906 . 3 ------*No production in either 1980 or 1990 .

------

production from the ten sites located in Siberia is a positiv e

2 .404-million tons . Accordingly, there was a net drop in tota l

production at the USSR's other fifteen low-grade nodes of 0 .497-millio n

tons with Kazakhstan (Rudnyy) dropping 15 .65-million tons, the Ukrain e

dropping by 5 .8-million tons, the Urals remaining stagnant (gain of a

mere 55,000 tons), the Transcaucasus (Dashkesan) increasing b y

-84-

NCSEER March 8 3

1 .3-million tons, Karelia gaining 5 .9-million tons, and the big gainer ,

the KMA, gaining 15 .3-million tons . It bears remembering tha t

production at the rich, but underground and problematic KMA sites wer e

confined by their upper capacity limits in both model years .

Iron Ore Flow Patterns (Table 29 and Maps 5 & 6) :

The modelled optimal iron ore flow patterns for 1980 and 1990 ar e

schematically illustrated by Map 5 and Map 6, respectively . Th e

calculated flow quantities are listed in Table 29 . The ore flow dat a

are the most complex of all the flows mapped as ore has four nodal set s

of potential destinations, namely, ore concentrating or enrichmen t

plants (low-grade ore only), export nodes, pig iron furnaces and/o r

steel furnaces .

The model simulated flows for 1980 appear quite close to actua l

flows with a few qualifications and exceptions . First, all but thre e

pig iron furnace nodes actually operating in 1980 are forced by thei r

minimum constraint values (see Table 10, p . 29) to produce ; and hence ,

require input flows of iron ore in the 1980 simulation run . The sam e

situation, however, is not true for steel production in 1980 (see Tabl e

13, pp . 37-39) . Accordingly, unless there were a very close correlatio n

between actual and model generated optimal steel production patterns ,

then one should not expect the modelled iron ore flows to map directl y

onto the actual flows .

Nonetheless, the simulated ore flows patterns match quite well wit h

most known actual ore flows in 1980 . For example, KMA ores beside s

serving the industrial needs of the European Russia hearthland als o

reach the Urals . On the other hand, no export demands are supplied b y

KMA ores in the model year 1980 where as KMA ores were actually exporte d

-8 5 -

NCSEER March 8 3

in 1980 . Ukrainian ores supply most of the export demands flowin g

westward, supply the Ukrainian iron and steel industry, but ship mor e

ore to the Urals than they actually did in 1980 . Instead of shippin g

all of its high phosphorous ores to appropriately designed Soviet stee l

mills along the north shore of the Azov Sea as it apparently actuall y

did (does) in 1980, Kerch ' (187) supplies over two-thirds of the expor t

demand at Sevastopol ' (185) for the 1980 computer run besides shippin g

ore to Taganrog ' s (274) steel mill . This difference between the actua l

and modelled flow patterns arises because, while we attempted to captur e

differences in Fe-contents in the ores in the model, we did not tak e

account of the geographical implications of different ore impurities .

Karazhal ' s (666) failure to enter the solution with a positiv e

production in 1980 partially accounts for the excess Ukrainian flows t o

the Urals and the modest flow of West Siberian ores from Tashtagol (725 )

to Karaganda (668) in the simulation .

The projected ore flows for 1990 (see Map 6) seem very plausible .

First, the KMA ores vastly expand their combined ore shipments to th e

Urals from 11 .53-million tons to 28 .52-million tons completely forcin g

Ukrainian ores out of the Urals and supplanting substantial quantitie s

of ores originating from northwest Kazakhstan in 1980 . Also, th e

high-grade ores from Zheleznogorsk (977) satisify the total expor t

demand of three-million tons assigned to Riga (43) . Despite th e

six-million tons of production from Karazhal (666) in 1990, Karaganda ' s

(668) iron and steel plants depend more on ore from West Siberia (node s

727, 730 & 978) than they currently do . In fact, the actual long-hau l

ore flowing into Karaganda (668) currently comes from northwes t

Kazakhstan (nodes 682 & 873) rather than West Siberia . It will b e

interesting to see whether or not these alternate source-demand pattern s

-86-

NCSEER March 8 3

------Table 29 . MODELLED OPTIMAL IRON ORE FLOWS BY YEAR IN 1000 TON S ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E* DESTINATION : E E* 1980 199 0

KACHKANAR 911N NIZHNIY TAGIL 531 H - 70 0 NIZHNIY TAGIL 531 P - 300

GOROLBAGODAT 530N CHUSOVOY 528 J 180 - CHUSOVOY 528 P 1695 - NIZHNIY TAGIL 531 P 3970 100 0 SEROV 532 P 1155 -

ALAPAYEVO 541N CHELYABINSK 581 P 200 - NIZHNIY TAGIL 531 P - 10 0

BAKAL 577N SATKA 576 P 1864 .5 1774 . 3 CHELYABINSK 581 P 1805 .7 4219 . 3 CHELYABINSK 581 J 310 .9 - CHELYABINSK 581 B 18 .8 -

NOVORUDNAYA 595N NOVOTROITSK 594 P 1000 2580 . 8

TAYEZHNOYE 964N CHUL ' MAN 963 U NA 2671 . 7

GARIN 774N NAKHODKA 802 E NA 140 0

UDA-SELEMDZHA 961N NONE -- NA -

OLENOGORSK 370N MURMANSK 368 E 3000 350 0 CHEREPOVETS 836 P 6500 -

KOSTAMUKSA 36IN CHEREPOVETS 836 P - 890 0

PUDOZHGORSK 381N CHEREPOVETS 836 P - 862 . 4 CHEREPOVETS 836 H - 2122 . 7 CHEREPOVETS 836 B - 14 . 9

ZHELEZNOGORSK 976N TULA 433 P 2034 800 NIZHNIY TAGIL 531 P 2531 .7 343 9 CHUSOVOY 528 P - 161 3 SEROV 532 P - 107 3 RUSTAVI 910 P 1356 - CHEREPOVETS 836 P - 4161 . 2 SUMGAIT 244 J 78 .3 -

GUBKIN 891N LIFETSK 955 P 12488 865 . 9 3 MAGNITOGORSK 586 P -22275. STARYY OSKOL 462 U 512 - STARYY OSKOL 462 B - 64 . 7

-8 7 -

NCSEER Marc h 83

Table 29 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E* DESTINATION : E E* 1980 199 0

KREMENCHUG 302N RIGA 43 E 2500 - MOSTISKA 107 E - 11090 CHEREPOVETS 836 P - 2989 .3 - NIZHNIY TAGIL 531 P 3323 .2 - NIZHNIY TAGIL 531 B 29 .2 - GOR'KIY 505 B 14 - KULEBAKI 89C J 90 - ZHLOBIN 88 H - 107 . 5 ZHLOBIN 88 U - 215 7

KRIVOY ROG 172N KRIVOY ROG 172 P 23730 4032 5 KRIVOY ROG 172 J 1215 - KRIVOY ROG 172 B - 21 0 ZAPOROZH'YE 179 P 7119 645 2 ZAPOROZH ' YE 179 J 450 - MOSTISKA 107 E 27486 -

KERCH ' 187N SEVASTOPOL ' 185 E 4669 .2 750 0 TAGANROG 274 J 30 .8 -

DASHKESAN 243N RUSTAVI 910 P - 1290 . 4

RUDNYY 873N CHELYABINSK 581 P 4943 .8 100 0 MAGNITOGORSK 586 P 11699 .2 - MAGNITOGORSK 586 B 7 -

KORSHUNOVSK 756N NAKHODKA 802 E 650 - NOVOKUZNESK 726 P 4893 .2 100 0 NOVOKUZNETSK 726 J 405 - NOVOKUZNETSK 726 B 51 .8 -

RUDNOGORSK 921N NONE - - NA -

ABAZA 730N NOVOKUZNETSK 726 P 1000 100 0 KARAGANDA 668 P - 500

GORNAYA SHORIYA 727N NOVOKUZNETSK 726 P 5054 .2 256 5 KARAGANDA 668 P - 956 . 2 BARNAUL 718 U - 4478 . 8

TEYA 978N NOVOKUZNETSK 726 P 1400 100 0 KARAGANDA 668 P - 286 . 2

KRASNOKAMENSK 979N NOVOKUZNETSK 726 P 1000 50 0

IRBA 980N NOVOKUZNETSK 726 P - 500

ABAZA. 729H NOVOKUZNETSK 726 P 1433 .2 150 0

-88-

NCSEER March 8 3

Table 29 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTED ORIGIN : E* DESTINATION : E E* 1980 199 0

KRIVOY ROG 173H MOSTISKA 107 E 426 17009 . 7 MAKEYEVKA 283 P 5573 356 5 DONETSK 284 H 810 1682 . 3 DONETSK 284 J 1350 - NIZHNIY TAGIL 531 P 1432 - NIZHNIY TAGIL 531 J 596 .3 - MAGNITOGORSK 586 P 5882 .8 - NOVOTROITSK 594 P 1712 - NOVOTROITSK 594 B 21 - DNEPROPETROVSK 811 P 17797 .5 967 8 DNEPROPETROVSK 811 J 540 - DNEPROPETROVSK 811 B 17 .5 - KOMUNARSK 884 P 8475 806 5 KOMMUNARSK 884 J 360 - KOMMUNARSK 884 B 7 -

GUBKIN 460H LIPETSK 955 P 4232 10914 . 4 TULA 433 P - 1085 . 6 TULA 433 B 1 .0 - STARYY OSKOL 462 B 10 .5 - CHEREPOVETS 836 P 6676 .5 - CHEREPOVETS 836 J 1080 -

ZHELEZNOGORSK 977H RIGA 43 E 3000 CHEREPOVETS 836 P - 5999 . 6 GOR'KIY 505 B - 0 . 4 NIZHNIY TAGIL 531 P 9000 -

DNEPRORUDNOYE 180H SEVASTOPOL' 185 E 2218 .8 500 MAKEYEVKA 283 P 35 .9 4500

MAGNITOGORSK 586H MAGNITOGORSK 586 P 2578 1300 MAGNITOGORSK 586 J 1422 - MAGNITOGORSK 586 H - 2700

KACHAR 683H CHELYABINSK 581 P NA 1393 . 7 CHELYABINSK 581 H NA 1606 . 3

TULA 433H TULA 433 P - 50 CHEREPOVETS 836 P 300 -

LIPETSK 955H LIPETSK 955 P 230 50 LIPETSK 955 B 70 -

RUDNYY 682H MAGNITOGORSK 586 P 3500 350 0

PESCHANSK 533H SEROV 532 P 540 54 0

-89-

NCSEER March 8 3

Table 29 . Continued :

N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E* DESTINATION : E E* 1980 199 0

KARAZHAL 666H KARAGANDA 668 P - 6000

TASHTAGOL 725H NOVOKUZNETSK 726 P 1958 - KARAGANDA 668 B 42 - BARNAUL 718 U - 2000

Total iron ore flows : 225779 .8 233996 . 3 ______*NOTES : Node numbers followed by an " N " represent enriched or e sites, while those followed by an " H " represent high-grade, direc t shipping ores . E = flow of iron ore to export nodes , P = flow of iron ore to pig iron plants , H & J = flow of iron ore to open hearth steel furnaces , B = flow of iron ore to basic oxygen steel furnaces, an d U - flow of iron ore to direct reduction steel furnaces . NA = ore nodes not available for use in 1980 .

emerge by 1990 .

The commencement of ore production at Tayezhnoye (964) by 199 0

probably will depend on a positive decision very soon to build a ne w

Siberian steel complex along the BAM .

Interpretation of Actual and Modelled Ore Production :

The most interesting aspect of the iron ore sub-system for bot h

years is the various possible interpretations which may be made of th e

total ore output . In 1980 actual total Soviet marketable iron or e

production amounted to a reported 244,713,000 tons [32] . The SISEM

model for 1980 requires only 225,777,900 tons in order to satisify al l

export, pig iron, and steel making demands (see Table 26, p . 79) . Th e

total for 1990 as calculated by the model amounts to 233,996,300 tons .

These discrepancies may be accounted for in a number of ways . First ,

-90-

NCSEER March 8 3

the Soviets may simply be overstating actual ore production . Second ,

they may be hoarding away unused iron ore . Third, both of thes e

phenomena may be occurring . The difference between actual an d

calculated ore demands, nineteen-million tons for 1980, seems somewha t

too great to be attributed to just these two above factors . Fourth, th e

various iron ore input coefficients used in the SISEM model may be to o

low . This interpretation appears to have much possible merit . Th e

inability of the Soviet economy to operate successfully within specifie d

technical norms is both well documented and pervasive . Thus, because w e

often had to resort to the use of Soviet " designed " technica l

coefficients as opposed to "actual " measured input norms at operatin g

pig iron and steel furnaces, the model could well understate th e

aggregrate iron ore input demand . Fifth, significant waste could simpl y

be happening between mines and/or enrichment combines and the blas t

furnaces [33] . Sixth, much of the disparity may, in fact, implicitl y

reflect sub-optimal production and commodity flow patterns within th e

actual Soviet ferrous metallurgical system .

Accordingly, then, to have a good basis for comparison it would b e

of much interest to rerun the model to more closely replicate actua l

1980 production patterns by introducing ever more narrowly specifie d

output ranges on pig iron and steel smelting by site and process . Ou r

best judgment is that the lower SISEM ore requirement results from al l

of the above listed factors . At the same time, evidence exist s

suggesting that ore output has already become a major limiting facto r

explaining the stagnation in Soviet iron and steel output in the pas t

few years [34] .

- 9 3 -

NCSEER March 8 3

SCRAP METAL SUPPLY AND FLOW PATTERNS :

Scrap Metal Supply Problems (Table 30 and Maps 7 & 8) :

Unlike iron ore supplies which do not constrain the system overall ,

scrap metal as noted previously does in fact constitute a bindin g

constraint or. the system overall for both 1980 and 1990 . As flows fro m

the last node in Table 30, fictitious node number 999, reveal there i s

insufficient scrap in 1980 to satisify the local input demands for iro n

and steel making at Krivoy Rog (172), Zaporozh ' ye (179), an d

Dnepropetrovsk (811) in the Ukraine ; Nizhniy Tagil (531), Serov (532) ,

Chelyabinsk (581), and Magnitogorsk (586) in the Urals ; and Novokuznetsk

(726) in West Siberia . These insufficiencies tally to 16 .4541-millio n

tons or 25 .5 per cent of the model ' s required scrap metal input . Fo r

1990 the excess scrap demand drops to slightly over 44 per cent of thi s

value or 7 .3113-million tons . Furthermore, in the 1990 scenario th e

unmet scrap input demands from only the two Urals integrate d

steel-making centers at Chelyabinsk (581) and Magnitogorsk (586) ar e

inducing flows from the fictitious scrap supply node 999 (see Table 30) .

The appearance of these unsatisified demands for scrap metal input s

and their various possible interpretations constitute perhaps the mos t

interesting discovery of the entire project . When reading the followin g

discussion one should keep in mind that a very considerable effort wa s

expended trying to incorporate both Soviet geographical raw materia l

supply and production data and technical input coefficients throughou t

the SISEM model . Having done this then one ' s curiosity is piqued b y

discovering that all three (including the 1970 historical run )

simulation runs are in fact infeasible due to insufficient scrap inputs !

Again, as noted previously, both the 1980 and 1990 computer simulatio n

runs were rendered feasible a priori by including fictitious supplie s

- 9 4 -

NCSEER March 8 3

(located at Node 999) for not only scrap metal but also for all ra w

material inputs and intermediate processes (i .e ., coking, enrichment an d

pig iron production) . By assigning very high " input costs " (99,99 9

rubles per 1,000-tons of a given input or process) these fictitiou s

quantities would only be called upon after all real supplies o r

production capacities were fully utilized . As it turns out the onl y

input or intermediate process which functions as an overall bindin g

constraint in both the 1980 and 1990 runs is scrap metal . How may thi s

input insufficiency be explained or interpreted ?

A number of simple possible explainations may be offered . First ,

the Soviets may merely have more scrap available than the data we foun d

state (see Table 6, p . 25) . Second, the scrap input coefficients w e

obtained for the various pig iron and steel smelting processes ma y

overstate scrap requirements .

On the other hand, the scrap insufficiency for 1980 can definitel y

not be blamed on the model trying to minimize systems costs by using a

steel process mix other than the actual one for 1980 . As discusse d

earlier, the model was constrained to replicate overall system stee l

production by process (see Table 17, p . 47) . Since it seemed mor e

plausible that the Soviets would expand production of cheaper processe s

rather than curtail production by more expensive processes, we only use d

minimum constraints on each steel process for 1990 (see Table 17, p .

47) . Accordingly, the fact that steel production was forced to increas e

by at leas' 21 .1 million tons between 1980 and 1990 coupled with th e

data in Table 30 showing that total scrap metal flows or usage decline d

slightly from 64 .5-million tons for 1980 to 63 .3-million tons for 199 0

gives clear foreshadowing that the model minimized costs in 1990 by

selecting a less scrap intensive technological process mix .

-95—

NCSEER March 8 3

A third possible partial explanation involves a consideration o f

the iron ore inputs, scrap inputs, pig iron poduction and ore and scra p

input coefficients simultaneously at least for the 1980 period . As wil l

be discussed shortly, the model could haye produced 3 .68-millio n

additional tons of pig iron in 1980 than it did . Remembering that th e

model yields approximately 18 .9-million tons less ore than the Soviet s

reportedly produced in 1980, this additional iron production would hav e

consumed about 6 .24-million more tons of iron ore . In turn, pig coul d

have been substituted for scrap in steel-making . On the one hand, th e

model clearly calculated such a solution as sub-optimal . On the othe r

hand, such a combination of factors would far from obviate the need fo r

"supplemental " scrap supplies !

Although all of the above factors may and probably do play a rol e

in accounting for the apparent failure of the data and SISEM model t o

yield feasible solutions, we strongly suspect that there i s

fundamentally a more sagacious answer . Very simply stated we believ e

that in effect the Soviets are using a significant level o f

double-counting in their steel production figures . Quite literall y

crude steel is being drawn off the bottom of furnaces and poured int o

the top again as on-site scrap . While such a practice is both pervasiv e

and economically rational everywhere in ferrous metallurgy, we strongl y

suspect that the Soviets are engaging in the practice to a far greate r

degree than is done in the West . In fact, upwards of one-half of al l

soviet scrap inputs may be generated on-site [35] !

The implications of this practice could be rather far reaching .

For example, if the entire infeasibility were to be explained by such a

double-counting process then the reported Soviet crude steel productio n

figure of 147 .9-million tons for 1980 could be overstating actua l

-96-

NCSEER March 8 3

off-site usable steel by a much as thirty million metric tons [36] .

Even in the optimal production pattern of the SISEM model the implici t

overstatement of usable steel for 1980 amounts to 15 .55 million tons o r

10 .5 per cent [37]! This in turn could have some major implications fo r

Western assessments of the military, industrial, and consumer sectors o f

the Soviet economy .

Scrap Flows & Dual Prices for 1980 and 1990 (Tables 30 & 31, Maps 7 & 8) :

The modelled optimal flows of scrap for 1980 and 1990 are portraye d

on Maps 7 and 8 . In both years the origin nodes for long-hauls of scra p

are essentially confined to scrap supply nodes east and south of th e

Urals . It is difficult to ascertain with any degree of confidenc e

whether or not such long distance flows in fact exist because the exces s

scrap demand forces all modelled scrap supplies to be fully utilized .

The dual activities of modelled scrap supply nodes are listed fro m

lowest to highest potential systems sayings (based on the 1980 run) i n

Table 31 . As a natural consequence of the model having more freedom t o

determine the optimal steel-making process-mix for 1990 than 1980, th e

positive opportunity costs (see Table 31) of the binding constraints a t

each node for 1990 are either equal to or less than the correspondin g

ones for 1980 . In both yeas the binding upper limits on scra p

availabilites at nodes in the Urals, the Donbas, Leningrad, Moscow an d

Gor ' kiy implicitly impose the highest economic penalties . Or ,

alternatively, additional supplies of scrap metal at these nodes woul d

yield the greatest per unit cost savings . On the other hand, th e

smallest implied economic forfeitures are attached to scrap nodes in th e

Far East, Central Asia, East Siberia, the Transcaucasus, the Baltic an d

Belorussia, none of which is a major producer of steel . Because th e

actual magnitudes of the dual values are predicated on th e

-97-

NCSEER March 8 3

------Table 30 . MODELLED OPTIMAL SCRAP IRON FLOWS BY YEAR IN 1000 TON S

N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

LENINGRAD 24 CHEREPOVETS 836 J 2505 - CHEREPOVETS 836 H - 299 0

MOSCOW 415 CHEREPOVETS 836 J 2112 .9 - CHEREPOVETS 836 H - 599 5 TULA 433 P 14 .4 - TULA 433 B 45 .4 - KULEBAKI 890 J 466 - LIPETSK 955 P 120 - LIPETSK 955 B 2338 .3 -

KUYBYSHEV 487 SATKA 576 P 13 .2 13 . 2 MAGNITOGORSK 586 P 168 - MAGNITOGORSK 586 J 3077 .8 - MAGNITOGORSK 586 H - 408 1

GOR'KIY 505 GOR'KIY 505 B 616 15 . 4 NIZHNIY TAGIL 531 P 143 .4 - NIZHNIY TAGIL 531 B 99 .6 - NIZHNIY TAGIL 531 H - 1034 . 6

RIGA 43 KOLPINO 27 T 66 .9 33 . 5 CHEREPOVETS 836 P 119 .7 - CHEREPOVETS 836 J 974 .4 - CHEREPOVETS 836 H - 1376 . 6

MINSK 84 KRIVOY ROG 172 J 1396 - CHEREPOVETS 836 P - 148 . 2 CHEREPOVETS 836 H - 463 . 5 CHEREPOVETS 836 B - 653 . 3 ZHLOBIN 88 H - 55 5

KRASNODAR 191 DONETSK 284 H 1819 - KRIVOY ROG 172 B - 208 0 KOMMUNARSK 884 P - 6 0 MAKEYEVKA 283 P - 6 0

KISHENEV 908 KRIVOY ROG 172 P - 28 5 KRIVOY ROG 172 J 225 -

TBIEISI 222 KRIVOY ROG 172 P - 1 5 KRIVOY ROG 172 B - 1748 . 4 RUSTAVI 910 P 9 .6 9 . 6 DNEPROPETROVSK 811 P - 7 2 SUMGAIT 244 J 405 .4 - TAGANROG 274 J 159 .6 - MAKEYEVKA 283 P 39 .7 -

-98-

NCSEER March 8 3

Table 30 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTED ORIGIN : E DESTINATION : E E* 1980 199 0

TBILISI 222 DONETSK 284 H 260 - KOMMUNARSK 884 P 60 - - .7 576 884 J KOMMUNARSK KOnMUNARSK 884 B 308 -

SVERDLOVSK 549 NIZHNIY TAGIL 531 J 2107 - CHELYABINSK 581 H - 248 0

NOVOSIBIRSK 709 NOVOSIBIRSK 709 T 1097 .3 - NOVOKUZNETSK 726 B 2172 .7 - CHELYABINSK 581 H - 1266 . 2 BARNAUL 718 T - 2638 . 8

IRKUTSK 858 NOVOKUZNETSK 726 P 118 .4 60 NOVOKUZNETSK 726 J 326 .7 - PETROVSK - ZABAYKAL'SKIY 951 T 531 .9 -

KHABAROVSK 782 PETROVSK - ZABAYKAL ' SKIY 951 T 137 .1 - KOMSOMOL ' SK - na-AMURE 783 T 735 .9 - BARNAUL 718 T - 19 5 TAYSHET 754 T - 88 5

KARAGANDA 668 KARAGANDA 668 P - 57 . 6 KARAGANDA 668 B 147 3 MAGNITOGORSK 586 H - 1782 . 4

TASHKENT 607 NOVOTROITSK 594 P 19 .2 19 . 2 NOVOTROITSK 594 B 924 - MAGNITOGORSK 586 J 34 .8 - MAGNITOGORSK 586 H - 1685 . 8 KARAGANDA 668 B 375 -

VINNITSA 807 KRIVOY ROG 172 J 1661 - KRIVOY ROG 172 B - 208 5

DONETSK 284 DONETSK 284 H 2115 8710 . 4 DONETSK 284 J 6990 - KRIVOY ROG 172 B - 2081 . 6 ZAPOROZH ' YE 179 P - 48

NIKOLAYEV 169 KRIVOY ROG 172 P 168 - KRIVOY ROG 172 J 858 - KRIVOY ROG 172 B - 124 5

VORONEZH 459 CHEREPOVETS 836 H - 165 . 9 TULA 433 P - 14 . 4

-99-

NCSEER March 8 3

Table 30 . Continued :

N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

VORONEZH 459 LIPETSK 955 P - 88 LIPETSK 955 B 741 .7 - STARYY OSKOL 462 B 462 2846 . 2 MAGNITOGORSK 586 P - 0 . 5 KOMMUNARSK 884 J 1287 .3 -

CHELYABINSK 581 CHELYABINSK 581 P 49 .2 - CHELYABINSK 581 J 1609 .9 - CHELYABINSK 581 T 1618 .9 - CHELYABINSK 581 H - 387 0

PERM' 524 CHUSOVOY 528 P 12 1 2 CHUSOVOY 528 J 932 - NIZHNIY TAGIL 531 P - 3 6 NIZHNIY TAGIL 531 J 115 .9 - NIZHNIY TAGIL 531 B 1186 .1 - NIZHNIY TAGIL 531 H - 2589 . 6 SEROV 532 P - 1 2 CHELYABINSK 581 H - 70 . 4

FICTITIOUS NODE 999 KRIVOY ROG 172 J 2151 - ZAPOROZH'YE 179 P 50 .4 - ZAPOROZH'YE 179 J 2330 - DNEPROPETROVSK 811 P 126 - DNEPROPETROVSK 811 J 2796 - DNEPROPETROSK 811 B 770 - NIZHNIY TAGIL 531 J 864 .7 - SEROV 532 P 12 - CHELYABINSK 581 P - 49 . 2 CHELYABINSK 581 H - 630 . 2 CHELYABINSK 581 T 90 .3 - CHELYABINSK 581 B 828 .6 - MAGNITOGORSK 586 P - 201 . 8 MAGNITOGORSK 586 J 4250 .2 - MAGNITOGORSK 586 B 308 - 0 643 - 586 H MAGNITOGORSK NOVOKUZNETSK 726 J 1770 .4 - NOVOKUZNETSK 726 B 106 .5 -

Total scrap flows : 64454 .1 63301 . 3 Total scrap supply : 48000 .0 55990 . 0

Total excess scrap used : 16454 .1 7311 . 3 ------*NOTE : P = flow of scrap metal to pig iron plants , A & T = flow of scrap metal to electric arc steel furnaces , H & J = flow of scrap metal to open hearth steel furnaces , and B = flow of scrap metal to basic oxygen steel furnaces .

- 1 00-

NCSEER March 83

------Table 31 . OPTIMAL SCRAP DUAL ACTIVITIES IN RUBLES PER TON * ------YEA R PLACE NAME OPTIMAL PROJECTE D AND REGION : NODE : 1980 199 0

KHABAROVSK (FAR EAST) 782 49 .04 43 .06 TASHKENT (CENTRAL ASIA) 607 53 .01 53 .0 1 IRKUTSK (E . SIBERIA) 858 53 .98 50 .39 TBILISI (TRANSCAUCASUS) 222 54 .35 51 .1 5 RIGA (BALTIC) 43 54 .90 54 .3 2 MINSK (BELORUSSIA) 84 55 .72 53 .7 7 KUYBYSHEV (POVOLZH'YE) 487 56 .00 56 .00 KRASNODAR ( N . CAUCASUS) 191 56 .52 53 .3 2 VINNITSA (S .W . URKRAINE) 807 56 .63 54 .2 1 VORONEZH (CENT . CHERNOZEM) 459 56 .75 53 .98 KISHENEV (MOLDAVIA) 908 56 .85 54 .4 3 NIKOLAYEV (S . URKRAINE) 169 57 .59 55 .1 7 KARAGANDA (KAZAKHSTAN) 668 62 .97 61 .7 3 NOVOSIBIRSK (W . SIBERIA) 709 63 .42 61 .1 0 GOR'KIY (VOLGA-VYATKA) 505 67 .35 66 .68 MOSCOW (CENTRAL) 415 67 .44 66 .86 LENINGRAD (NORTHWEST) 24 67 .83 67 .24 PERM' (W . URALS) 524 68 .72 68 .0 5 SVERDLOVSK (N . URALS) 549 69 .01 68 .7 5 DONETSK (DONETS-DNEPR) 284 69 .29 66 .0 2 CHELYABINSK (S . URALS) 581 69 .58 69 .5 8 ------*Dual activity indicates the change in the objective function fo r a unit change in a right-hand-side value, or expresse d mathematically = -dZ/dbi .

------

99,999 .00 rubles per thousand tons assigned to the supplies of scra p

from fictitious Node 999, only the relative differences and geographica l

pattern of the dual activities is of significance .

PATTERNS OF ELECTRICITY CONSUMPTION (Table 32 and Maps 9 & 10) :

One of the primary initial motivations for this investigation wa s

to explore the possible effects changing energy prices and suppl y

patterns could have on the pattern of Soviet steel production . Table 3 2

lists the consumption of electricity by node by year, Remember, ou r

effort to model the Soviet electrical transmission system was abandone d

-103-

NCSEER March 8 3

early . Nonetheless, the electricity costs used in the SISEM mode l

supposedly reflect differences in the delivered costs of electricity o n

a regional basis by year (see Table 7, p . 26) . Also, remember that th e

model requires electrical usage by three processes : 1) iron enrichment ,

2) pig iron production and 3) all steel processes . Accordingly, th e

total nodal electrical consumption by any and all of these thre e

processes is portrayed by Maps 9 and 10 .

The most surprising discovery is how little total electricity i s

used by the steel industry in the results of the model runs . Overal l

only about 29 .5-billion kilowatt hours are required for the 1980 run, o r

merely 22 .5 percent of the estimate of the upper limit of 131 .0-billio n

kwh on system-wide electricity consumption . One interpretation of thi s

huge discrepancy is that the Soviet production system is far fro m

optimal with respect to electrical use . We doubt this interpretato n

simply because except for the final or steel fabricating stage, th e

model is highly constrained to replicate the actual Soviet productio n

patterns for 198 0

Three other factors are likely to be more telling . First, th e

upper limit probably includes other processes or sub-sectors of Sovie t

metallurgy, including non-ferrous metals such as aluminum an d

ferro-alloys . Second, the Soviet technical coefficients may wel l

understate actual use . Third, clearly electricity is used for othe r

tasks in the system other than enrichment, and iron and steel smelting .

On the other hand, the input coefficients do take into accoun t

electricity consumed in the various processes associated with stee l

fabrication at integrated mills . Despite the bias against electric-ar c

steelmaking which will be noted later in the discussion and the fac t

that steel output between 1980 and 1990 is only slated to increase b y

-104-

NCSEER March 8 3

------Table 32 . MODELLED OPTIMAL ELECTRICITY CONSUMPTION BY NODE BY YEA R (in 1000 kilowatt hours ) ------PROJECTED CHANG E PLACE : NODE 1980 1990 1980-1990

BARNAUL 718 NA 12,121,408 12,121,40 8 CHUL'MAN 963 NA 3,443,956 3,443,956 ZHEOBIN 88 NA 3,351,628 3,351,628 KRIVOY ROG 172 3,057,000 5,573,844 2,516,844 CHEREPOVETS 836 2,003,040 4,042,122 2,039,08 2 MAGNITOGORSK 586 2,815,600 4,763,569 1, 9 47,996 9 TAYSHET 754 NA 828,126 828,12 6 STARYY OSKOL 462 1,021,000 1,404,611 383,61 1 GUBKIN 891 156,000 278,471 122,47 1 KOSTAMUKSA 361 NA 106,800 106,80 0 ZHELEZNOGORSK 976 72,000 133,034 61,03 4 KREMENCHUG 302 107,350 160,254 52,90 4 MAKEYEVKA 283 99,271 150,000 50,72 9 PUDOZHGORSK 381 NA 36,000 36,00 0 GORNAYA SHORIYA 727 60,650 96,000 35,35 0 KERCH' 187 56,400 90,000 33,60 0 TAYEZHNOYE 964 NA 32,060 32,06 0 BAKAL 577 48,000 71,923 23,92 3 NOVORUDNAYA 595 12,000 30,970 18,97 0 GARIN 774 NA 16,800 16,80 0 DASHKESAN 243 0 15,485 15,48 5 KACHKANAR 911 0 12,000 12,888 ABAZA 730 12,000 18,000 6 , IRBA 980 NA 6,000 6,000 SATKA 576 33,000 33,000 0 SEROV 532 30,000 30,000 0 RUSTAVI 910 24,000 24,000 0 TEYA 978 16,800 15,434 -1,36 6 KRASNOKAMENSK 979 12,000 6,000 -6,00 0 KULEBAKI 890 9,000 0 -9,00 0 TULA 433 58,412 36,000 -22,41 2 KOLPINO 27 75,900 37,950 -37,95 0 TAGANROG 274 48,621 0 -48,62 1 KORSHUNOVSK 756 72,000 12,000 -60,00 0 GOROLBAGODAT 530 84,000 12,000 -72,00 0 OLENOGORSK 370 140,000 42,000 -98,00 0 SUMGAIT 244 123,540 0 -123,54 0 RUDNYY 843 199,800 12,000 -187,80 0 CHUSOVOY 528 314,000 30,000 -284,00 0 GOR'KIY 505 304,000 7,600 -296,40 0 NOVOTROITSK 594 504,000 48,000 -456,00 0 PETROVSK-ZABAYKAL'SK . 951 626,000 0 -626,00 8 KOMSOMOL'SK -na-AMURE 783 688,600 0 -668,6 0 ZAPOROZH'YE 179 838,000 120,000 -718,00 0 KOMMUNARSK 884 870,000 150,000 -720,00 0 NIZHNIY TAGIE 531 1,933,871 1,194,369 -739,50 2 DONETSK 284 3,408,000 2,654,242 -753,75 8 KARAGANDA 668 912,000 144,000 -768,000 NOVOSIBIRSK 709 1,026,736 0 -1,026,73 6 DNEPROPETROVSK 811 1,547,000 180,000 -1,367,00 0 LIPETSK 955 1,820,000 220,032 -1,599,96 8 CHELYABINSK 581 2,162,359 283,626 -1,878,73 3 NOVOKUZNETSK 726 2,060,059 150,000 -1,910,05 9

Totals : 29,462,009 42,225,314 12,763,30 5 ------

-105-

NCSEER March 8 3

approximately 15 per cent, electricity usage jumps by 43 per cent (refe r

to bottom of Table 32) . Nonetheless, the 42 .2-billion kwh usage whic h

the model calculates for 1990 still represents slightly less than 25 pe r

cent of the " projected " upper limits on consumption by the metallurgica l

industries .

While we are hesitant to make any very precise predictions, it doe s

not appear to us as though electricity demands do or will place seriou s

constraints on aggregrate Soviet steel production any time soon . On a

regional perspective, however, net increases in electricity demand coul d

add to the current critically short electricity supply situation in th e

European U .S .S .R . The opposite or an excess supply situation currentl y

exists in West Siberia . The appearance of a large capacity direc t

reduction steel mill at Barnaul (718) by 1990, however, could alon e

significantly reduce surplus electrical generating capacity in th e

region .

It is of passing interest to mention the nodes which the SISEM

model indicates as " optimally " expanding their electrical consumption .

The last column in Table 32 and the shaded symbols on Map 10 illustrat e

these locations . The demand generated by the new Barnaul (718) direc t

reduction furnaces clearly overwhelms all other nodes in terms of it s

magnitude . The three other new steel sites, Chul ' man (963) along th e

BAM, Zhlobin (88) in southeastern Belorussia and Tayshet (754) in Eas t

Siberia all loom as first order electrical users . The other big gainer s

are Cherepovets (836), Staryy Oskol (462), Krivoy Rog (172), an d

Magnitogorsk (586), all centers judged by the model as suitable fo r

major expansions in their steelmaking capacities . In general th e

dynamic pattern of electrical use very closely mirrors the spatia l

dynamics of steel production . Accordingly, then, the nodes experiencin g

- 1 08-

NCSEER March 8 3

the greatest declines in electrical use for 1990, are simply those th e

model selects for curtailed steel production .

NATURAL GAS CONSUMPTION PATTERNS (Table 33 and Maps 11 & 12) :

With very few nuances what has been said above about electrica l

consumption can be repeated for natural gas consumption patterns (refe r

to Table 33 and Maps 11 & 12) . Total gas consumption is indicated a s

being very much below the model ' s upper bound . In percentage term s

" optimal " total gas consumption grows more than twice as fast a s

electrical demand between 1980 and 1990, 99 .5 per cent versus 43 pe r

cent . Unlike the case with electrical demand growth where all four ne w

steel mills are at or near the top in increased as well as absolut e

consumption, only Chul ' man (963) and Zhlobin (88) are major natural ga s

users . Similar to the electricity demand situation, Magnitogorsk (586 )

in the Urals and Cherepovets (836) in northwest European Russia vastl y

expand their gas use . The other nodes needing large boosts in their ga s

supplies are Chelyabinsk (581) in the Urals and Donetsk (284) in th e

Ukraine . Karaganda (668) along with Krivoy Rog (172) require an orde r

of magnitude smaller increments . All other sites are optimall y

projected as consuming less gas in 1990 than in 1980 primarily as a

result of technological shifts in steelmaking which will be discusse d

later .

While on a system-wide basis gas supplies do not appear to be a

problem, the situation could be more problematic on a regional basis .

First, as suggested by Map 12 more West Siberian gas supplies probably

will need to be allocated to both the Urals and the European U .S .S .R . i n

the next decade . The Soviets should have little difficulty in providin g

these additional gas supplies . Second, Yakutian gas pipelines ar e

- 1 09-

NCSEER March 8 3

------Table 33 . MODEELED OPTIMAL NATURAL GAS CONSUMPTION BY NODE BY YEA R (in 1000 cubic meters ) ------OPTIMAL PROJECTED CHANG E PLACE : NODE 1980 1990 1980-199 0

MAGNITOGORSK 586 1,246,000 3,953,850 2,707,85 0 CHELYABINSK 581 348,500 2,951,072 2,602,57 2 CHEREPOVETS 836 847,280 3,08z,016 2,236,73 6 CHUL'MAN 963 NA 2,149,864 2,149,86 4 ZHLOBIN 88 NA 1,923,934 1,923,93 4

DONETSK 284 1,026,000 2,130,870 1,104,87 0 KARAGANDA 668 0 153,600 153,600 KRIVOY ROG 172 672,000 800,000 128,00 0 RUSTAVI 910 68,000 25,600 -42,40 0 CHUSOVOY 528 85,000 32,000 -53,00 0

SEROV 532 85,000 32,000 -53,00 0 SATKA 576 93,500 35,200 -58,30 0 TULA 433 102,000 38,400 -63,60 0 NIZHNIY TAGIL 531 1,051,686 982,606 -69,08 0 NOVOTROITSK 594 136,000 51,200 -84,80 0

MAKEYEVKA 283 281,268 125,000 -156,26 8 ZAPOROZH'YE 179 357,000 128,000 -229,00 0 KOMUNARSK 884 425,000 160,000 -265,00 0 STARYY OSKOL 462 412,000 0 -412,00 0 DNEPROPETROVSK 811 693,000 192,000 -501,00 0

LIPETSK 955 850,000 230,401 -619,59 9 NOVOKUZNETSK 726 839,400 160,000 -679,40 0

Totals : 9,618,634 19,195,613 9,576,97 9

------

implicit in the model for the opening of the Chul ' man (963) stee l

complex . On the other hand, coke could be substituted (refer to Tabl e

15, p . 44) . In essence the Soviet metallurgical industry should remai n

in an excellent position with respect to natural gas inputs through th e

early 1990s at least .

- 1 1 0-

NCSEER March 8 3

PIG IRON PRODUCTION AND FLOW PATTERNS :

Pig Iron Production and Dual Activities for 1980 (Table 34 and Map 13) :

A total of eighteen possible centers for pig iron production exis t

for 1980 . As noted earlier in the discussion of the SISEM data set, som e

of these locations really represent regionally aggregrated productio n

nodes . In the aggregrate the model shows these production site s

yielding 103 .6-million metric tons of pig iron for 1980 . This quantit y

is approximately 3 .68-million tons less than the actual reported Sovie t

pig iron production for 1980 [38] . One interpretation of this modes t

disparity is that the Soviets are quite close to " optimal " in thei r

production of pig iron, that is, their actual production of pig iron i s

very close to the optimal or minimum as required by the SISEM model t o

satisify all pig iron demands at the minimum overall systems cost .

In the optimal solution for 1980 all sites except Karaganda (668 )

are producing pig iron (see Table 34, p . 115 and Map 13, p . 126) . Si x

of the eighteen plants haye their output constrained by their modelle d

upper limits (refer to Table 10, p . 29) . Four of the six nodes are i n

the Ukraine, namely, Krivoy Rog (172), Dnepropetrovsk (811), Kommunars k

(884), and Zaporozh ' ye (179) . A fifth site, the Urals gian t

Magnitogorsk (586) complex is tied at 14 .0-million tons with Krivoy Ro g

(172) as the largest producer . Lipetsk (955) in European Russi a

represents the sixth pig iron facility modelled as producing at its ful l

1980 capacity .

Seven of the producing plants, on the other hand, enter th e

solution at their lower limits . Of these seven, five are located in th e

Urals, Serov (532), Chusovoy (528), Chelyabinsk (581), Satka (576), an d

Novotroitsk (594) and the other two, Rustavi (910) and Tula (433), ar e

in the Caucasus and Central European Russia, respectively . Althoug h

- 1 1 3 -

NCSEER March 8 3

Karaganda (668) had a set lower limit of zero production for 1980, i t

actually enters the solution as a basis variable having a value of zer o

rather than as a nonbasis variable at its lower limit .

Finally, besides Karaganda (668), Makeyevka (283) in the Donbas ,

Cherepovets (836) in northwest European Russia, Novokuznetsk (726) i n

West Siberia, and Nizhniy Tagil (531) in the northern Urals enter a s

basis variables . The last three nodes, in fact, enter as large-scal e

producers ranging from slightly less than ten-million tons annually u p

to almost twelve-million tons .

Pig Iron Dual Activities for 1980 (Table 34) :

An inspection of the dual activities column for 1980 (see Table 34 ,

p . 115) reveals interesting insights into the overall performance of th e

pig iron stage within the entire Soviet ferrous metallurgical system .

First, of the six sites bound by their upper limits three hav e

sufficiently large positive dual values to suggest that added capacit y

might well be warranted at their locations . In other words, expande d

capacities at Krivoy Rog (172) (dual activity equal to 12 .3 8

rubles/ton), Lipetsk (955) (10 .45 rubles/ton) and Kommunarsk (884) (8 .98

rubles/ton) would yield the highest per unit system cost savings .

Interestingly enough, the Soviets have been expanding production at bot h

Krivoy Rog (172) and Lipetsk (955) fairly continuously over the las t

fifteen or so years . Potential system cost savings from added capacit y

at Magnitogorsk (586) is far more modest (1 .78 rubles/ton) . While th e

production at both Zaporozh ' ye (179) and Dnepropetrovsk (811) i s

constrained to their maximum capacities, these outputs are nonetheles s

close to optimal .

The marginal opportunity costs implicit in the pig iron from site s

- 1 1 4-

NCSEER March 8 3

------Table 34 . MODELLED OPTIMAL PIG IRON PRODUCTION BY YEAR IN 1000 TONS * ------OPTIMAL PROJECTE D 1980 199 0 PLACE : NODE PROD . AT DUAL PROD . AT DUA L

RUSTAVI 910 800 LL -16 .34 800 LL -22 .05 KARAGANDA 668 0 BS 0 .00 4800 LL -6 .3 5 NOVOKUZNETSK 726 9875 .3 BS 0 .00 5000 LL -5 .8 7 KOMMUNARSK 884 5000 UL 8 .98 5000 LL -2 .7 7 ZAPOROZH ' YE 179 4200 UL 0 .34 4000 LL -5 .5 3

LIPETSK 955 10000 UL 10 .45 7334 .4 BS 0 .00 CHEREPOVETS 836 9968 BS 0 .00 12351 .8 BS 0 .00 TULA 433 1200 LL -7 .67 1200 LL -10 .7 8 SEROV 532 1000 LL -15 .87 1000 LL -2 .65 NIZHNIY TAGIL 531 11951 BS 0 .00 3000 LL -1,8 5

CHUSOVOY 528 1000 LL -14 .95 1000 LL -11 .63 CHELYABINSK 581 4100 LE -8 .69 4100 LL -10 .5 4 MAGNITOGORSK 586 14000 UL 1 .78 16785 .6 BS 0 .00 SATKA 576 1100 LL -14 .71 1100 LL -19 .1 8 NOVOTROITSK 594 1600 LL -13 .44 1600 LL -8 .4 1

KRIVOY ROG 172 14000 UL 12 .38 25000 UL 0 .1 1 DNEPROPETROVSK 811 10500 UL 0 .78 6000 LL -5 .1 5 MAKEYEVKA 283 3309 BS 0 .00 5000 LL -5 .28 STARYY OSKOL 462 NA - - 0 BS 0 .00 BARNAUL 718 NA - - 0 BS 0 .00

TAYSHET 754 NA - - 0 BS 0 .0 0 SVOBODNYY 981 NA - - 0 BS 0 .0 0 TYNDA 962 NA - - 0 BS 0 .0 0

Total pig iron product : 103603 .3 105071 . 8 ------*NOTE : BS = in basis and feasible , UL = nonbasis activity at upper limit , LL = nonbasis, activity at lower limit , NA = site not available for pig iron production in 1980 , and DUAL = dual activity, mathematically -dZ/db i (in RUBLES/TON) .

------

forced to produce at least their minimum quantities are, in general ,

higher than those from the " under capacitated " sites . For instance, al l

of the negative dual activities exceed -7 .50 rubles per ton . Th e

largest per unit cost savings could be achieved by reducing the minimum

- 1 1 5-

NCSEER March 8 3

pig iron yields at Rustavi (910) (dual activity equal to -16 .34

rubles/ton) and at Serov (532) (-15 .87 rubles/ton), followed closely b y

Chusovoy (528) (-14 .95 rubles/ton), Satka (576) (-14 .71 rubles/ton), an d

Novotroitsk (-13 .44 rubles/ton) . Interestingly enough, these hig h

opportunity costs are predicated primarily on the high production cost s

per unit output at these relatively small capacity furnaces (refer t o

Table 10, p . 29) . Furthermore, with the possible exception o f

Chelyabinsk (581), the Soviets rationally have not been expanding th e

output at any of these locations which the model indicates as " ove r

producing " or producing at their lower limits .

Pig Iron Flows for 1980 (Table 35 and Map 13) :

Pig iron commodity flows haye several possible destinations .

First, as an intermediate product pig iron flows are required a t

open-hearth steel furnaces (types H & J), electric-arc steel furnace s

(type T only) and basic-oxygen steel furnaces (type B) . Second, as a

final product pig iron flows to both a set of five surrogate expor t

demand nodes and a set of twenty-one regionally aggregrated domesti c

final demand nodes . Several individual nodes in the Soviet iron an d

steel system generate more than one of the above types of pig iro n

demand .

While individual flow quantities from a given pig plant to each an d

every type of demand at a given node are listed in Table 35 (pp .

122-125), all pig iron flows from a given supply node to a given deman d

node are summed and represented on Map 13 (p . 126) as a singl e

flow-range link . In addition to the pig plants, all steel mills an d

export and domestic final demand nodes are shown on Map 13 . While Tabl e

35 distinguishes between export and domestic final demand nodes, a

- 1 1 6 -

NCSEER March 8 3

common octagon symbol is used for both types on the Map 13 .

Within the model none of the total of slightly over 37-million tons o f

pig iron smelted in the Ukraine leaves the Republic, except for expor t

flows channeled through Sevastopol ' (185) . In fact, Lipetsk (955) i n

Central European Russia supplies the southwest regional domestic fina l

demands situated at Vinnitsa (807) and Kishenev (908) . Within th e

European U .S .S .R . Cherepovets (836) located North of Moscow and Lipets k

(955) located south-southeast of Moscow clearly serve the geographicall y

widest ranging markets .

The only truly long-hauls occur in the eastern portion of the country .

In addition to supplying its own steel input demands, pig iron from th e

West Siberian Novokuznetsk (726) plant satisfies the export demand a t

Nakhodka (802), the domestic final demands at Khabarovsk (782) an d

Irkutsk (858), and steel-making input requirements a t

Komsomol ' sk-na-Amure (783) and Petrovsk-Zabaykal ' skiy (951) .

Pig iron centers in the Urals also have a fairly large market area .

Serov ' s (532) requirement to produce at least one-million tons account s

for the fact that this northern Urals facility, rather than Novokuznets k

(726), supplies Novosibirsk (709) . Urals' pig iron production from

Magnitogorsk (586) supplies the Central Asian final iron demand centere d

at Tashkent (607) . Then too, it seems quite reasonable that Urals ' pi g

iron would supply Gor ' kiy (505) and Kuybyshev (487) as the modeled flow s

indicate . The flows of Urals iron into Kazakhstan, however, is rathe r

surprising . As noted above Karaganda ' s (668) minimum constraint for

1980 was set to zero and as a result, this Kazakhstan site does no t

yield any production in 1980 . It seems likely that if some of the othe r

sites that have minimum output levels ranging upward from one-millio n

metric tons had had lower or zero minimum levels, then Karaganda (668 )

- 1 1 7-

NCSEER March 8 3

would have entered the solution with significant output . Such was (is )

not the case, and thus, Nizhniy Tagil (531) and Magnitogorsk (586 )

supply Karaganda ' s (668) final pig iron demand and steel input needs .

The only other apparent surprise involves Rustavi (910) in th e

Transcaucasus . In actuality, we presume that Rustavi ' s (910) pi g

production remains totally within the three Transcaucasus republic s

unlike the model which has Rustavi (910) supplying final iron demand a t

Krasnodar (191) in the North Caucasus and steel-making needs at Taganro g

(274) along the north shore of the Azov Sea as well as the regiona l

final demand centered at Tbilisi (222) and pig iron inputs for Sumgait ' s

(244) open-hearth steel furnaces .

Pig Iron Production Changes Projected to 1990 (Table 34 and Map 14) :

Between 1980 and 1990 an additional five potential pig iro n

production nodes were incorporated : Staryy Oskol (462) in the KMA

district, Barnaul (718) in West Siberia, Tayshet (754) in East Siberia ,

Svobodnyy (981) in the Far East, and Tynda (962) along the BAM (se e

Table 34, p . 11 .5 and Map 14, p . 127) . Furthermore, except for thes e

five potentially new facilities, all sites were forced to produce a t

least 800,000 tons of pig iron annually and where appropriate both lowe r

and upper production limits, production costs, and various technica l

input coefficients were altered as well (compare Tables 10 and 11, pp .

29-30) .

Despite the absolute increase in steel production of slightly ove r

22-million tons from the 1980 to the 1990 simulation, pig iro n

production is only required to expand by 1 .47-million tons as a resul t

of technological shifts in the production of steel as will be discusse d

latter . Although the system is thus quite stable with respect t o

- 1 1 8-

NCSEER March 8 3

overall pig iron production between these two time periods, severa l

nodes experience significant changes . Before discussing these change s

it should be noted that all five of the potential new sites enter th e

solution as basis variables, but all with production levels equal t o

zero ! The model results thus suggest that none of the five propose d

new pig iron production sites is economically rational .

By far the biggest indicated increase in pig iron occurs at Krivo y

Rog (172) . In both years this Ukrainian metallurgical giant ' s output i s

confined to its upper limit . However, the 1990 upper bound o f

25-million tons is 11 .0-million tons larger than that of 1980 .

Interestingly enough from 1980 to 1990 Krivoy Rog ' s (172) dual activit y

drops precipitiously from a marginal savings of 12 .38 rubles pe r

additional ton to a trivial 11 kopecks per marginal ton, respectively .

Makeyevka (283), also in the Ukraine, is forced by the change d

constraints to expand its output by approximately 1 .7-million tons .

The Ukraine and the Urals each have one large scale loser .

Dnepropetrovsk ' s (811) output falls from its 1980 upper limit o f

10 .5-million tons to its lower limit in 1990 or 6 .0-million metric tons .

The greatest negative change, however, occurs in the Urals at Nizhni y

Tagil (531) where pig iron output drops from a basis solution o f

11 .95-million tons in 1980 to its lower limit of 3 .0-million tons i n

1920 . While only suffering a loss of 200,000 tons from 1980 to 1990 ,

Zaporozh ' ye ' s (179) furnaces along the Dnepr River in the Ukraine shif t

from being constrained by their upper limit in 1980 to being bound b y

their lower one in 1990 .

Two major examples of intra-regional nodal production shifts occur .

For example, within the Central European Heartland the reduction of pi g

iron output at Lipetsk (955) (-2 .67-million tons) is nearly made up fo r

- 1 1 9-

NCSEER March 8 3

by expansion at Cherepovets (836) (+2 .38-million tons) . The other shif t

involves Karaganda and the Kuzbas . Because of the added minimu m

constraint on Karaganda (668), it is now forced to produce at a level o f

4 .8-million tons . This forced " over production " comes at margina l

additional cost of 6 .35 rubles per ton . A corresponding reduction of

4 .8-million tons or nearly fifty percent in Kuzbas pig iron output i s

the result primarily of the elimination of steel input demands a t

Novokuznetsk (726), Komsomol ' sk-na-Amure (783) and Petrovsk -

Zabaykal ' skiy (951) . In turn, these mills have been forced to ceas e

steel production in the 1990 model year because of the entry of Barnau l

(718) in West Siberia, Tayshet (754) in East Siberia, and Chul'man (963 )

along the BAM as steel centers by 1990 . Production at all other site s

remains unchanged from 1980 to 1990 .

Changes in Pig Iron Dual Activities from 1980 to 1990 (Table 34) :

The sign of the dual activity changes for some of the sites fro m

1980 to 1990 (see Table 34, p . 115) . Nonetheless, by comparing th e

absolute magnitude of the dual costs between the two periods, one find s

that ten of the producing sites are farther from optimality in 1990 tha n

1980, seven sites are closer to optimal production while the ouput a t

Cherepovets (836) is optimal in both periods . The most uneconomi c

production site in both years is Rustavi (910) in the Georgian S .S .R .,

followed closely in 1990 by the small Urals plant at Satka (576) . To

the north, Serov ' s (532) constant lower bounded production is far close r

to optimal in 1990 (-2 .65 rub ./ton) than 1980 (-15 .87 rub ./ton) .

Whereas overall there were six plants producing at their upper limits i n

1980, only Krivoy Rog (172) does so in 1990 . This situation arises, o f

course, chiefly as a result of the general increase in the uppe r level

- 1 2 0-

NCSEER March 8 3

production limits for the 1990 simulation (again, compare Table 10, p .

29 and Table 11, p . 30) . The number of pig iron facilities producing a t

their lower limits doubles from seven in 1980 to fourteen in 1990 .

Although the number of pig plants functioning at their optimal level s

increases from five in 1980 to eight in 1990, this is due to the fac t

that all five of the proposed sites enter the solution as optimall y

having zero output . Hence, of the plants having a positive output i n

1990 only three are basis variables have zero dual activities . On th e

other hand, these three, Magnitogorsk (586) in the Urals, Cherepovet s

(836) in the northwest, and Lipetsk (955) in the European center are al l

major producers, second only to Krivoy rog (172) .

Changes in Flow Patterns from 1980 to 1990 (Table 35 and Maps 13 & 14) :

Because the total calculated pig iron production only expands b y

slightly less than 1 .5-million tons from 1980 to 1990 (see bottom o f

Table 35, p . 125), one ' s first impression might be that the flo w

patterns would be very stable . This impression must be tempered ,

however, by the fact that a number of sites more or less " traded "

portions of their production between the two periods as discusse d

previously . For example, by not having to supply the " defunct " stee l

mills at Komsomol ' sk-na-Amure (783) and Petrovsk-Zabaykal 'skiy (951) ,

Novokuznetsk (726) (in effect the loser to Karaganda) adds links to th e

new steel mills at Tayshet (754) and Barnaul (718) and reaches a lon g

distant tentacle to the Urals helping to supply the inputs for th e

open-hearth steel furnaces at Chelyabinsk (581) . Concomitantly, thi s

West Siberian iron center captures Novosibirsk ' s (709) final deman d

market from Serov (532) while maintaining its original three fina l

demand markets in the Far East (see Map . 14, p . 127) .

- 1 2 1 -

NCSEER March 8 3

------Table 35 . MODELLED OPTIMAL PIG IRON FLOWS BY YEAR IN 1000 TON S ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

RUSTAVI 910 KRASNODAR i91 D 70 88 TBILISI 222 D 67 7 0 TASHKENT 607 D - 6 3 BAKU 245 E - 2 0 SUMGAIT 244 J 542 .9 - TAGANROG 274 J 120 .1 - DONETSK 284 H - 559

KARAGANDA 668 KARAGANDA 668 D - 7 0 CHELYABINSK 581 H - 4395 . 6 MAGNITOGORSK 586 H - 334 . 4

NOVOKUZNETSK 726 NOVOSIBIRSK 709 D - 14 5 IRKUTSK 858 D 27 3 5 CHELYABINSK 581 D - 11 4 KHABAROVSK 782 D 32 4 3 NAKHODKA 802 E 245 250 NOVOKUZNETSK 726 J 2808 - NOVOKUZNETSK 726 B 5942 .2 - PETROVSK - ZABAYKAL'SKIY 951 T 391 - KOMSOMOL ' SK - na-AMURE 783 T 430 .1 - TAYSHET 754 T - 517 . 2 BARNAUL 718 T - 2354 . 7 CHELYABINSK 581 H - 1541 . 1

KOMMUNARSK 884 KOMMUNARSK 884 H 2496 - KOM IUNARSK 884 J 803 - DONETSK 284 H 1701 500 0

ZAPOROZH'YE 179 NIKOLAYEV 169 D - 50 SEVASTOPOL' 185 E - 105 5 ZAPOROZH ' YE 179 J 3120 - DONETSK 284 J 1080 - DONETSK 284 H - 1562 . 7 STARYY OSKOL 462 B - 1332 . 3

LIPETSK 955 LENINGRAD 24 D - 10 3 RIGA 43 D - 35 MINK 84 D - 56 . 8 KUYBYSHEV 487 D - 13 8 GOR'KIY 505 D - 3 3 VORONEZH 459 D - 23 3 VINNITSA 807 D 67 -

- 1 2 2 -

NCSEER83 M ar ch

Table 35 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

LIPETSK 955 KISHENEV 908 D 4 - BREST 92 E 2750 3026 . 2 VENTSPILS 45 E - 7 5 LIPETSK 955 B 7179 - STARYY OSKOL 462 B - 258 . 7 GOR'KIY 505 B - 40 . 2 CHEREPOVETS 836 H - 3335 . 5

CHEREPOVETS 836 LENINGRAD 24 D 83 - RIGA 43 D 28 - MINSK 84 D 101 - MOSCOW 415 D 366 - VORONEZH 459 D 180 - VENTSPILS 45 E 70 - KOLPINO 27 T 39 .1 - CHEREPOVETS 836 J 7488 - CHEREPOVETS 836 H - 10648 . 4 CHEREPOVETS 836 B - 1703 . 3 KULEBAKI 890 J 624 - STARYY OSKOL 462 B 137 .9 - LIPETSK 955 B 851 -

TULA 433 MOSCOW 415 D - 44 7 TULA 433 B 118 .4 - STARYY OSKOL 462 B 1081 .6 - KOLPINO 27 T - 19 . 5 CHEREPOVETS 836 H - 733 . 5

SEROV 532 NOVOSIBIRSK 709 D 119 - NOVOSIBIRSK 709 T 641 .3 - GOR'KIY 505 B 239 .7 - NIZHNIY TAGIL 531 H - 100 0

NIZHNIY TAGIL 531 GOR'KIY 505 D 26 - SVERDLOVSK 549 D 58 - KARAGANDA 668 D 54 - GOR'KIY 505 B 1366 .3 - KARAGANDA 668 B 2337 .2 - CHUSOVOY 528 J 310 - NIZHNIY TAGIL 531 J 4134 .5 - NIZHNIY TAGIL 531 B 3351 .9 - NIZHNIY TAG IL 531 H - 3000 CHELYABINSK 581 B 313 .1 -

CHUSOVOY 528 PERM ' 524 D 62 7 7 SVERDLOVSK 549 D - 7 0 CHUSOVOY 528 J 938 - NIZHNIY TAGIL 531 H - 85 3

- 1 2 3 -

NCSEER March 8 3

Table 35 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

CHELYABINSK 581 CHELYABINSK 581 D 93 - CHELYABINSK 581 H - 4100 CHELYABINSK 581 J 2155 .8 - CHELYABINSK 581 T 998 .9 - CHELYABINSK 581 B 852 .3 -

MAGNITOGORSK 586 TASHKENT 607 D 48 - MAJNITOGORSK 586 H 9859 .2 16785 . 6 MAGNITOGORSK 586 B 803 - NOVOTROITSK 594 B 809 - KARAGANDA 668 B 2480 .8 -

SATKA 576 KUYBYSHEV 487 D 105 - CHELYABINSK 581 B 995 - CHELYABINSK 581 H - 110 0

NOVOTROITRSK 594 NOVOTROITSK 594 B 1600 - MAGNITOGORSK 586 H - 160 0

KRIVOY ROG 172 NIKOLAYEV 169 D 40 - DONETSK 284 D 370 - VINNITSA 807 D - 88 KISHENEV 908 D - 5 BREST 92 E - 73 . 8 SEVASTOPOL' 185 E 935 - KRIVOY R00 172 J 8424 - KRIVOY ROG 172 B - 24090 DONETSK 284 J 316 - DONETSK 284 H 3915 - ZHLOBIN 88 H - 743 . 2

DNEPROPETREOVSK 811 MINSK 84 D - 78 . 2 DNEPROPETROVSK 811 J 3744 - DNEPROPETROVSK 811 B 2007 .5 - DONETSK 284 J 4748 .5 - STARYY OSKOL 462 B - 5921 . 8

MAKEYEVKA 283 DONETSK 284 D - 458 DONETSK 284 J 3215 .5 - DONETSK 284 H - 454 2 TAGANROG 274 J 93 .5 -

STARYY OSKOL 462 NONE -- NA -

BARNAUL 718 NONE -- NA -

TAYSHET 754 NONE -- NA -

-124-

NCSEER March 8 3

Table 35 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN : E DESTINATION : E E* 1980 199 0

SVOBODNYY 981 NONE - - NA -

TYNDA 962 NONE - - NA ------Total pig iron flows : 103603 .3 105071 . 8 ------*D = flow of pig iron to domestic final demand nodes , E = flow of pig iron to export nodes , T = flow of pig iron to electric arc steel furnaces , H & J = flow of pig iron to open hearth steel furnaces , B = flow of pig iron to basic oxygen steel furnaces, an d NA = not available for pig iron production in 1980 .

------

In 1990 the market range of Urals ' pig iron contracts drasticall y

compared with 1980 . For example, trans-Caspian shipments from Rustav i

(910) displace Magnitogorsk ' s (586) tie with the Central Asian fina l

demand market centered at Tashkent (607) . No longer requiring Urals '

pig plants to supply its needs in 1990, Karaganda (668) reverses th e

commodity flow pattern and supplies both Magnitogorsk (586) an d

Chelyabinsk (581) with iron . Iron flows originating from Lipetsk (955 )

replace ones previously emanating from the Ural ' s in both the Kuybyshe v

(487) and Gor'kiy (505) final demand markets .

Three changes are evident in the European part of the country .

First, despite a very substantial increase in pig output at Cherepovet s

(836), this northwest metallurgical complex ceases shipments of pig iro n

to other nodes in 1990 . This is chiefly due to vastly increased on-sit e

pig iron input demands by its expanded steel production in 1990 .

Second, as a consequence, Lipetsk (955) not only picks up Cherepovets '

(836) former markets in the northwest portion of European Russia, bu t

also, ships pig iron to Cherepovets ' (836) open-hearth furnaces . Tul a

-125-

83 March NCSEER

(433) assists Lipetsk (955) in supplying Cherepovets' (836) furnaces an d

satisfies Moscow's (415) final iron demand and Kolpino ' s (27 )

electric-arc steel furnaces . Third, Ukrainian pig plants extend thei r

market ranges in 1990 compared with 1980 . Krivoy Rog (172) now eithe r

fulfills or shares the final and export demands of the southwester n

nodes previously supplied by Lipetsk (955) exclusively . Finally ,

Dnepropetrovsk (811) and Zaporozh ' ye (179 establish pig iron flow link s

with the new steel furnaces at Staryy Oskol (462) .

STEEL PRODUCTION AND FLOW PATTERNS :

Introduction :

The last manufactured commodity to be discussed, of course, i s

steel . The SISEM algorithm does not differentiate between types o f

steel . As previously explained, however, this homogeneous end-produc t

may be produced by four different major steel-making processes (i .e .,

sets of input requirements, see Table 15, p . 44) : (1) electric arc (two

variants), (2) open hearth (two variants), (3) basic oxygen (BoF) an d

(4) direct reduction [39] . Remember also that each of the various stee l

producing nodes have both lower (zero tons per year in most instances )

and upper bounds on steel production by process type as well as an uppe r

limit on the combined output by all processes at a given node in a give n

model year . The steel output in both 1980 and 1990 is transported to a

total of six export nodes and forty domestic final demand nodes .

It should be noted that while Table 36 (pp . 139-141) which follow s

does include all possible steel smelting locations (remember some of th e

nodes represent regional plant aggregates), it does not repeat all o f

the possible processes at each site as enumerated in the data Tables 1 3

and 14 (refer to pp . 37-43) . Table 36 (pp . 139-141) only list s

- 12 8 -

NCSEER March 8 3

processes which have positive production at a given site in either 1980 ,

1990 or both periods . In cases where the model indicates that a give n

process at a given node only produces steel in one of the two mode l

years, dashes are placed in the respective year ' s production, an d

accordingly, dual activity columns . Due to the degeneracy situatio n

explained earlier, all processes not listed at a given node have bot h

zero production and zero dual activities in both scenario years .

Steel Production for 1980 (Table 36 and Map 15) :

The results from the thirty-three possible steel production center s

are listed in Table 36 (pp . 139-141) from highest to lowest (i .e ., zero )

production for 1980 and illustrated by Map 15 (p . 146) . Overall th e

modelled optimal steel production for 1980 offers little in the way o f

surprises as both the open-hearth and basic-oxygen processes wer e

constrained to yield 99 .093-million and 41 .412-million tons ,

respectively (refer to Table 17, p . 47) . These figures represent th e

actual quantities produced by each process in 1980 . Four outcomes ,

however, are of interest .

First, remembering that Types J and H are two variants of th e

open-hearth process with the former requiring coke only and the latte r

both coke and gas, the energy production and cost data appear to be suc h

that Type J is strongly favored over Type H in 1980 . In fact, only a t

Donetsk (284) where Type J was limited to 15 .0-million tons annuall y

does process Type H enter the solution with a positive outpu t

(9 .0-million tons) . This sub-process selection bias by the SISEM mode l

is implicitly an overall cost savings one and accounts in large part fo r

the low total natural gas usage as was discussed previously .

Second, electric-arc furnaces yield one-million tons below thei r

- 1 2 9 -

NCSEER March 8 3

estimated actual output in 1980 . This shortfall was made possibl e

because we only set an upper limit on steel smelted in electric-ar c

furnaces in 1980 . This maximum was (is) equal to the actua l

7 .395-million tons of "electric steel " made in the U .S .S .R . in 1980 . A s

discussed below direct reduction steel in effect displaces one-millio n

tons of electric-arc steel .

Third, the Staryy Oskol (462) site where the Soviets are currentl y

building a direct reduction steel complex does, in fact, enter th e

model ' s optimal solution producing steel by both the new direc t

reduction process and by the basic oxygen technique . On the other hand ,

neither type of electric-arc process option embedded in the model a t

Staryy Oskol (462) yields any output . Just as our earlier work [40 ]

provided ex post facto spatial-economic vindication of the Soviets '

decision to undertake steel manufacture in the KMA ore district a t

Staryy Oskol (462) so does this current work, but even more convincingl y

because of the added complexity and empirical realism of the SISE M

model .

Fourth, the shortage of scrap discussed previously biases the mode l

completely against electric steelmaking variant Type A which uses scra p

only in favor of the less scrap intensive Type T, which uses both scra p

and pig iron .

Fifteen of the set of thirty-three possible steel producing centers ar e

indicated as having their operations confined by their 1980 noda l

production limits . Of this number, however, only one, Kolpino (27), i s

limited by a lower nodal bound . The other fourteen sites are producin g

at their 1980 nodal upper bounds . Furthermore, half of these plants ,

namely, Dnepropetrovsk (811), Krivoy Rog (172), Zaporozh ' ye (179) ,

Donetsk (284), and Kommunarsk (884) in the Ukraine, Magnitogorsk (586 )

in the Urals, and Novokuznetsk (726) in the Kuzbas, are smelting stee l

- 1 30-

NCSEER March 8 3

at their respective upper limits for process J (open-hearth using coke ,

but no gas) .

Six other steel complexes have output from one of their steelmakin g

processes constrained by its upper limit . Kulebaki (890) an aggregrat e

node east o Moscow including nearby Vyksa, Sumgait (244) along th e

Caspian Sea north of Baku, and Chusovoy (528) in the Urals all come u p

against the upper bounds on their open-hearth (Type J) furnaces .

Gor ' kiy ' s (505) basic oxygen process (Type B) is performing at its uppe r

limit and Petrovsk-Zabaykal ' skiy (951) and Komsomol ' sk-na-Amure ' s (783 )

electric-arc (Type T) furnaces are operating at " full capacity . " A t

Gor ' kiy (505), Sumgait (244), Petrovsk-Zabaykal ' skiy (951) an d

Komsomol ' sk-na-Amure (783) the process upper limits are the same as th e

respective nodal limits . Nonetheless, unlike Zaporozh ' ye (179) wher e

both process and nodal upper limits are encountered, the noda l

production figures at these three complexes are indicated as optima l

with each constituting a basis variable, and hence, having a zero dua l

value .

The reverse situation, that is, nodal production limits formin g

binding constraints with a particular single process producing all o f

the nodal output, yet entering as a basis variable, occurs a t

Cherepovets (836), Lipetsk (955), Karaganda (668), and Novotroits k

(594) . This may be interpreted as it being cheapest for these fou r

complexes to operate at their nodal capacity by concomitantly onl y

producing steel by the single cheapest process option available at eac h

respective node . This is only possible because at these nodes at leas t

one of the potential proceses has an upper limit equal to the noda l

limit . On the other hand, at Dnepropetrovsk (811) and Donets k

(284)--which, for example, both produce steel at their nodal limits--n o

- 1 3 1 -

NCSEER March 8 3

single steel process has an upper limit equal to the nodal maximum .

Consequently, both sites encounter process as well as nodal uppe r

limits .

Only three locations enter the solution for 1980 having positiv e

production that is neither constrained by process nor nodal productio n

limits . All three represent small or relatively small capacit y

facilities and are widely dispersed geographically . For instance, Tul a

(433) is in the old European metallurgical hearth south of Moscow .

Taganrog (274) is along the shores of the Azov Sea south of the Donba s

proper . Finally, Novosibirsk (709) is the commercial, scientific ,

educational, and largest urban center of Siberia .

Nine of the sites are shown as optimally producing no steel . Non e

of these mills--with the possible exception of Volgograd (265)--actuall y

produced above 2 .0-million tons in 1980 . However, in the model two o f

the sites had nodal maxima limits set above this value : Volgograd (265 )

at 2 .5-million tons and Elektrostal' (846), an aggregated node includin g

Moscow, at 3 .0-million tons . On the one hand, one could argue that th e

failure of these small mills to enter into production is simply a resul t

of the higher per unit production costs associated with them (refer t o

Table 16, p . 45) . On the other hand, in our earlier and simple r

modeling efforts for 1970 Soviet steel production in which a unifor m

average production cost of 35 rubles per ton was used at all stee l

plants [41], all but six of the twenty-three small mills (i .e ., thos e

having an actual 1970 production of 2 .0-million tons or less) failed t o

enter the optimal solution with any positive steel producion . Becaus e

some sites in this current study involve regional aggregation, there ar e

only fourteen sites with upper limits of 2 .0-million tons or less . O f

these fourteen seven fail to enter the optimnal solution with productio n

- 1 3 2 -

NCSEER March 8 3

above zero tons per year . Overall, then, the evidence strongly suggest s

that these seven sites are sub-optimal production locations fo r

additional spatial-systems cost factors other than merely their hig h

unit production costs .

Steel Dual Activities for 1980 (Table 36) :

Dual activities can be used to identify where plant capacity shoul d

be expanded and to a lesser extend where production should be curtaile d

from the perspective of systems cost savings . Curtailment is mor e

difficult to justify on the basis of (negative) dual activities becaus e

we are not confident enough in our cost data to say it is necessaril y

cheaper to abandon existing facilities in favor of building new ones .

Our published Soviet production cost data (refer to Tables 13-14, pp .

37-43 and Table 16, p . 45) do supposedly include amortization payments .

Some would argue that for existing plants only variable costs should b e

included . Our approach, however, must be thought of as a long-ter m

planning approach with a long-term time horizon . Thus, even thoug h

capital in an already existing plant is sunk, there is also th e

requirement to maintain the facility and invest in new capital t o

upgrade the facility to incorporate modern techniques . For the 199 0

runs we have assumed new technological fixes are possible for the olde r

plants and this assumption is consistent with this view . This appear s

to the case in the Unitd States where in many cases old plants are shu t

down in favor of new plants even under conditions of excess supply .

Here it is a question of whether retrofitting is more expensive tha n

building an entirely new facility . We are merely assuming that th e

costs are the same . With this theoretical discussion in mind we ca n

turn to a evaluation of the 1980 dual activities for steel .

- 1 3 3-

March 83 NCSEER

The greatest potential per unit costs savings would accrue b y

expansion at four already large steel complexes . Added capacities i n

the Ukraine at Krivoy Rog (172) with a potential savings of 18 .2 0

rubles/ton, in northwest European Russia at Cherepovets (836) wit h

potential savings of 15 .49 rubles/ton, and in the Urals at Magnitogors k

(586) with a potential savings of 17 .56 rubles/ton and at Nizhniy Tagi l

(531) with a potential savings of 14 .46 rubles/ton are suggested mos t

strongly . These steel centers represent four of the six location s

calculated by the model for 1980 as producing over 10 .0-million ton s

(refer to Table 36, pp . 139-141) . The other two, Donetsk (284) an d

Novokuznetsk (726) have positive dual activities only about one-third a s

large . Additional smaller potential savings ranging from 2 .76 ruble s

per ton to 8 .92 rubles per ton could be achieved by expansion at th e

following sites : Lipetsk (955) and Staryy Oskol (462) in the Europea n

heartland, Dnepropetrovsk (811), Zaporozh ' ye (179), and Kommunarsk (884 )

in the Ukraine, and Novotroitsk (594) in the Urals .

Because we only imposed a few minimum production constraints abov e

zero tons per annum on the steel plants only one, Kolpino (27), has a

negative dual activity (-3 .11 rubles/ton) . Obviously, since several o f

the sites in both runs yield no production, inclusion of such lowe r

limits could have undoubtedly resulted in some fairly high negative dua l

values . Our reasons for including only a few such constraints is tha t

we believed that our nodal upper limits for 1980, which we tried to mak e

as close as we could to the known actual 1980 figures, would go a lon g

way toward making the lower limits somewhat redundant .

Finally, all other sites are shown as producing at their optimum o r

would yield less than one ruble per additional ton in potential syste m

cost savings .

- 1 34-

NCSEER March 8 3

Steel Distribution Patterns for 1980 (Table 37 and Map 15) :

The 1980 steel flow patterns are enumerated in Table 37 (pp .

142-145) and illustrated schematically by Map 15 (p . 146) . As Map 1 5

clearly shows Novokuznetsk (726) in the Kuzbas has the largest number o f

market linkages at ten and the widest geographical range of any stee l

center . The central Kazakhstan iron and steel center of Karaganda (668 )

supplies all of Central Asia except for Alma-Ata . Of the Urals ' stee l

mills Nizhniy Tagil (531) and Magnitogorsk (586) both have Europea n

markets including Moscow (415) for the former and Vilnius (74), Mins k

(84), L ' vov (110) and Kazan ' (560) for the latter . In the souther n

Urals Novotroitsk (594) shares the Volgograd (265) market with Donets k

(284) and Kommunarsk (884) . As expected Cherepovets (836) serves th e

northern Baltic and northwest European markets . A total of six center s

serve Moscow (415), the most divided steel market . As expected Krivo y

Rog (172) (13 .5-million tons) supplies the western and southwester n

Ukraine and some export demands . Like Novokuznetsk (726) the singl e

largest steel producing center in the 1980 model results, the aggregrat e

node representing Donetsk Oblast (284) (24 .0-million tons), has a radia l

steel distribution pattern sending steel to eight different destination s

besides satisifying its own high final demand . With the exceptions o f

the small regional Komsomol ' sk-na-Amure (783) and Petrovsk-Zabaykal ' ski y

(951) mills which serve small regional demands in the Far East, al l

other producing centers have very limited market ranges .

The shadow prices at the final demand nodes range from a low o f

120 .99 rubles per ton at Leninsk-Kuznetsk (735) in West Siberia to a

high of 134 .49 rubles per ton at the Pacific export node of Nakhodk a

(802) . In other words, the maximum regional difference in the modelle d

-l .35-

optimal marginal delivered cost of steel in 1980 is 13 .50 rubles per ton

or about 11 .2 per cent .

Changes in the Geography of Steel to 1990 (Tables 36 & 37 and Maps 15 & 16) :

Changes besides costs and production ranges on all the variou s

inputs between the 1980 and 1990 runs should be mentioned . First, wit h

few exceptions (refer to Table 14, pp . 40-43) an upper limit o f

30 .0-million tons per annum was imposed on all steel nodes and on eac h

given process at a given node in the 1990 simulation . Similar to th e

1980 run, very few nodes, and no potential producing processes at an y

node, were given a lower limit other than zero tons per annum . Second ,

four potential steel sites were added : Zhlobin (88) in Belorussia ,

Barnaul (718) in West Siberia, Tayshet (754) in East Siberia, an d

Chul ' man (963) along the BAM roughly midway between Lake Baykal and th e

small steel mill at Komsomol'sk-na-Amure (783) . Third, unlike the 198 0

run where the total system-wide output from both open-hearth an d

basic-oxygen processes were given fixed values, none of the fou r

processes was given a fixed production requirement . Instead all wer e

given minimum production levels only (refer to Table 17, p . 47) . As

discussed previously these production floors were set equal to th e

actual reported 1980 steel output by process . Overall steel productio n

was bracketed between 169 .0-million and 171 .8-million tons . Obviously ,

these constraints are unnecessary as the system will only produce enoug h

steel to satisify all of the final demands or 169,985,000 tons (se e

Table 21, pp . 53-54) .

The most notable change between 1980 and 1990 is the increase i n

the average quantity of steel smelted by producing mills accompanied b y

the reduction in the number of active nodes (see Table 36, pp . 139-141) .

-136-

NCSEER March 8 3

Of equal interest is the fact that all of the four proposed sites ente r

into production ranging from slightly over l .3-million to ove r

18 .6-million tons . Whereas twenty-four of the thirty-three centers i n

1980 were active, only thirteen of the enlarged set of thirty-seve n

sites are indicated as active in 1990 . The nine sites which fail t o

produce in 1980 also fail to do so in 1990 . An additional fifteen o f

the active mills in 1980 drop out of production in 1990 . Eight of thes e

consist of small mills none having optimal 1980 steel output s exceeding

2 .0-million tons . Their productive disappearance is thus not terribl y

surprising given the expanded upper limits by process and node .

Some of the other seven " non-producing " complexes call fo r

discussion (refer to Tables 36 and 37, pp . 139-145, and Maps 15 and 16 ,

pp . 146-147) . Novokuznetsk (726) is the largest of the 1980 producer s

(11 .9-million tons) to cease operations . The combined production from

the new mills at Barnaul (718), Tayshet (754) and Chul ' man (963 )

eliminates the need for what must now be more expensive Novokuznets k

(726) steel . Expansion of production at Cherepovets (836), Magnitogors k

(586), and Staryy Oskol (462) squeezes out Lipetsk (955) . The more tha n

doubling of Krivoy Rog ' s (172) steel output forces both Dnepropetrovs k

(811) and Zaporozh ' ye (179) out of operation . Magnitogorsk (586) an d

Donetsk 284) replace Kommunarsk (884) in the Volgograd (265) an d

Voroshilovgrad (278) markets . Novotroitsk (594) loses all of it s

markets to Magnitogorsk (586) . Finally, had they not been confined t o

miminum outputs of 50,000 tons both Gor'kiy (505) and Kolpino (27) woul d

surely have had zero output as well .

Unlike 1980 when there were fourteen plants functioning at thei r

upper limits, the vastly increased upper nodal limit of 30 .0-millio n

tons results in only two sites operating at full capacity in 1990 .

-137-

NCSEER March 8 3

Interestingly enough, they are two current actual Soviet iron and stee l

production giants, Krivoy Rog 9172) and Magnitogorsk (586) . Compared t o

1980 modelled production they rank first and third in the absolut e

magnitudes of their increased steel tonnage of all the plants i n

existence in 1980, 16 .5-million tons and 13 .2-million tons ,

respectively . The second greatest indicated expansion occurs a t

Cherepovets (836), 13 .7-million tons . All of these absolute increase s

in levels of production, however, are less than the new production o f

nearly 18 .7-million tons at Barnaul (718) in West Siberia . Chelyabins k

(581) in the central Urals and Staryy Oskol (462) are also larg e

gainers .

Besides the outright disappearance of steel flows from seven significan t

1980 production centers, two centers suffer substantial contractions .

In the Ukraine Donetsk (284) drops from 24 .0-million to slightly les s

than 18 .7-million tons . In the northern Urals Nizhniy Tagil ' s (531 )

1990 output of 7 .78-million tons is approximately 3 .0-million tons belo w

its 1980 figure .

Changes in Dual Activities for Steel to 1990 (Table 36) :

Obviously, with the less constrained nature of the 1990 mode l

formulation, especially as a result of the expanded upper bounds, dua l

activities for 1990 have nearly been eliminated . In fact, production b y

all steel processes at all locations is optimal (i .e ., zero dua l

activity) . At the same time, only four sites operate at constraine d

limits with respect to total nodal steel production . Accordingly ,

further reduction of steel output at the small Gor ' kiy (505) facilit y

would yield a modest 2 .42 rubles savings per ton . Similar action a t

Kolpino (27) would save a substanitial marginal amount 18 .75 rubles pe r

-138-

NCSEER March 8 3

------Table 36 . MODELLED OPTIMAL STEEL PRODUCTION BY YEAR IN 1000 TON S ------N T YEAR YEA R 0 Y 1980 199 0 D P OPTIMAL PROJECTE D PLACE : E E* PROD . AT* DUAL PROD . AT* DUAL

DONETSK 284 S 24000 UL 5 .59 18691 .8 BS 0 .0 0 H 9000 BS 0 .00 18691 .8 BS 0 .0 0 J 15000 UL 3 .03 ---

MAGNITOGORSK 586 S 16800 UI 17 .56 30000 UL 3 .2 7 J 15800 UL 0 .09 - - - H - - - 30000 BS 0 .0 0 B 1000 BS 0 .00 - - -

KRIVOY ROG 172 S 13500 UL 18 .20 30000 UL 8 .9 0 J 13500 UL 0 .73 --- B --- 30000 BS 0 .0 0

CHEREPOVETS 836 S 12000 UL 15 .49 25706 .8 BS 0 .0 0 J 12000 BS 0 .00 --- H - - - 23585 .6 BS 0 .00 B -- - 2121 .1 BS 0 .0 0

NOVOKUZNETSK 726 S 11900 UL 5 .44 0 BS 0 .00 J 4500 UL 1 .30 - - - B 7400 BS 0 .00 - - -

NIZHNIY TAGIL 531 S 10800 UL 14 .46 7777 .2 BS 0 .0 0 J 6625 .8 BS 0 .00 --- B 4174 .2 BS 0 .00 - - - H - - - 7777 .2 BS 0 .0 0

LIPETSK 955 S 10000 UL 3 .16 0 BS 0 .0 0 B 10000 BS 0 .00 ---

CHELYABINSK 581 S 8700 UL 0 .97 17847 .3 BS 0 .0 0 J 3454 .8 BS 0 .00 --- H - - - 17847 .3 BS 0 .0 0 T 2554 .8 BS 0 .00 - - - B 2690 .4 BS 0 .00 - - -

DNEPROPETROVSK 811 S 8500 UL 8 .74 0 BS 0 .0 0 J 6000 UL 0 .95 - - - B 2500 BS 0 .00 - - -

KARAGANDA 668 S 6000 UL 0 .33 0 BS 0 .0 0 B 6000 BS 0 .00 - - -

ZAPOROZH'YE 179 S 5000 UL 8 .44 0 BS 0 .00 J 5000 UL 0 .94 - - -

KOMMUNARSK 884 S 5000 UL 4 .37 0 BS 0 .00

-139-

Table 36 . Continued :

------N T YEAR YEA R q Y 1980 1990 D P OPTIMAL PROJECTE D PLACE : E E* PROD . AT* DUAL PROD . AT* DUA L

KOKMUNARSK 884 J 4000 UL l .05 - - - B 1000 BS 0 .00 - - -

NOVOTROITSK 594 S 3000 UL 2 .76 0 BS 0 .00 B 3000 BS 0 .00 - - -

STARYY OSKOE 462 S 2500 UL 8 .92 9240 .9 BS 0 .00 U 1000 BS 0 .00 - - - B 1500 BS 0 .00 9240 .9 BS 0 .00

GOR ' KIY 505 S 2000 BS 0 .00 50 LL -2 .4 2 B 2000 UL 10 .02 50 BS 0 .00

CHUSOVOY 528 S 2000 BS 0 .00 0 BS 0 .00 J 2000 UL 0 .55 - - -

NOVOSIBIRSK 709 S 1640 .2 BS 0 .00 0 BS 0 .00 T 1640 .2 BS 0 .00 - - -

KOMSOMOL'SK- 783 S 1100 BS 0 .00 0 BS 0 .00 na-AMURE T 1100 UL 9 .79 - - -

KULEBAKI 890 S 1000 BS 0 .00 0 BS 0 .00 J 1000 UL 1 .08 - - -

PETROVSK- 951 S 1000 BS 0 .00 0 BS 0 .00 ZABAYKAL ' SKIY T 1000 UL 1 .30 - - -

SUMGAIT 244 S 870 BS 0 .00 0 BS 0 .0 0 J 870 UL l .01 - - -

TAGANROG 274 S 342 .4 BS 0 .00 0 BS 0 .00 J 342 .4 ES 0 .00 - - -

TULA 433 S 147 .5 BS 0 .00 0 BS 0 .00 B 147 .5 BS 0 .00 - - -

KOLPINO 27 S 100 LL -3 .11 50 LL -18 .7 3 T 100 BS 0 .00 50 BS 0 .00

ELEKTROSTAL' 846 S 0 BS 0 .00 0 BS 0 .00

VOLGOGRAD 265 S 0 BS 0 .00 0 BS 0 .00

LIEPAYA 46 S 0 BS 0 .00 0 BS 0 .00

MOGILEV 325 S 0 BS 0 .00 0 BS 0 .00

RUSTAVI 910 S 0 BS 0 .00 0 BS 0 .00

-140-

NCSEER March 8 3

Table 36 . Continued : ------N T YEAR YEA R q Y 1980 199 0 D P OPTIMAL PROJECTE D PLACE : E E* PROD . AT* DUAL PROD . AT* DUA L

BEKABAD 943 S 0 BS 0 .00 0 BS 0 .00

IZHEVSK 554 S 0 BS 0 .00 0 BS 0 .00

SEROV 532 S 0 BS 0 .00 0 BS 0 .00

KRASNOYARSK 750 S 0 BS 0 .00 0 BS 0 .0 0

BARNAUL 718 S NA NA 18676 BS 0 .0 0 T NA NA 6022 .1 BS 0 .0 0 U NA NA 12653 .9 BS 0 .0 0

ZHLOBIN 88 S NA NA 5404 BS 0 .0 0 H NA NA 1191 BS 0 .0 0 U NA NA 4213 BS 0 .0 0

CHUL'MAN 963 S NA NA 5218 .l BS 0 .0 0 U NA NA 5218 .1 BS 0 .0 0

TAYSHET 754 S NA NA 1322 .9 BS 0 .0 0 T NA NA 1322 .9 BS 0 .0 0

Sub-total processes A & T : 6395 BS 0 .00 7395 LL -45 .30 Sub-total processes H & J : 99093 EQ 22 .35 99093 LL -28 .2 2 Sub-total process B : 41412 EQ 28 .01 41412 EL -22 .4 5 Sub-total process U : 1000 UL 58 .38 22085 BS 0 .00

Total all processes : 147900 16998 5 ------*NOTES : S = total steel production by all processes at node , A = electric arc using scrap only , T = electric arc using both pig iron and scrap , H = open hearth using gas and coke , J = open hearth using coke only , B = basic oxygen , U = direct reduction ,

BS = in basis and feasible , UL = nonbasis, activity at upper limit , LL = nonbasis, activity at lower limit , and NA = node not applicable fo steel production in 1980 .

- 1 4 1 -

NCSEER March 8 3

Table 37 . MODELLED OPTIMAL STEEL FLOWS BY YEAR IN 1000 TON S ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN DESTINATION : E E* 1980 199 0

DONETSK 284 DONETSK 284 D 8571 949 2 ODESSA 158 D 486 - KRASNODAR KRAI 191 D 1469 170 0 YEREVAN 228 D 3196 .6 1413 . 8 VOLGOGRAD 265 D 895 - ROSTOV-na-DONU 271 D 2232 251 8 VOROSHILOVGRAD 278 D - 356 8 VORONEZH 459 D 4398 .4 - DNEPROPETROVSK 811 D 93 2 SEVASTOPOL' 185 F 1820 -

MAGNITOGORSK 586 MAGNITOGORSK 586 D 1601 177 4 VILNIUS 74 D 138 - MINSK 84 D 3112 - L ' VOV 110 D 590 188 0 ORZHONIKIDZE 250 D - 192 2 VOLGOGRAD 265 D - 363 7 VORONEZH 459 D - 3580 . 3 KAZAN' 560 D 11359 13389 ORENBURG 596 D - 252 4 AKTYUBINSK 838 D - 59 1 UZHGOROD 113 F - 702, 7

KRIVOY ROG 172 L ' VOV 110 D 992 - ODESSA 158 D 3168 426 3 KIEV 313 D 3940 471 7 DNEPROPETROVSK 811 D - 1648 6 UZHGOROD 113 F 5400 247 9 SEVASTOPOL' 185 F - 205 5

CHEREPOVETS 836 LENINGRAD 24 D 5630 627 5 RIGA 43 D 3427 393 4 ARKHANGEL ' SK 388 D 1155 132 1 MOSCOW 415 D 1758 14176 . 8 VENTSPILS 45 F 30 -

NOVOKUZNETSK 726 ALMA-ATA 653 D 1148 - KOKCHETAV 692 D 1322 .2 - NOVOSIBIRSK 709 D 2523 .8 - SEMIPALATINSK 723 -D 809 LENIN'SK-KUZN'ETSK 735 D 2766 - KRASNOYARSK 750 D 1058 - BLAGOVESHCHENSK 775 D 1045 - IRKUTSK 858 D 1015 - NORIL ' SK 888 D 183 - NAUSHKI 761 F 30 -

- 1 4 2 -

NCSEER March 83

Table 37 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN E DESTINATION : E E* 1980 199 0

NIZHNIY TAGIL 531 MOSCOW 415 D 3769 .2 2840 . 2 PERM ' 524 D 2344 493 7 SVERDLOVSK 549 D 4686 .8 -

LIPETSK 955 ORDZHONIKIDZE 250 D 1655 - MOSCOW 415 D 7566 .4 - VORONEZH 459,, D 778 .6 -

CHEEYABINSK 581 CHELYABINSK 581 D 6098 680 9 SVERDLOVSK 549 D 2467 .2 811 5 KOKCHETAV 692 D 134 .8 - UZHGOROD 113 F - 2923 . 3

DNEPROPETROVSK 811 DNEPROPETROVSK 811 D 8500 -

KARAGANDA 668 TASHKENT 607 D 2830 - MARI 622 D 304 - DUSHANBE 630 D 332 - FERGANA 634 D 810 - KOKCHETAV 692 D 1569 - TERMEZ 629 F 155 -

ZAPOROZH ' YE 179 DNEPROPETROVSK 811 D 5000 -

KOMMUNARSK 884 VOLGOGRAD 265 D 1810 - VOROSHILOVGRAD 278 D 3190 -

NOVOTROITSK 594 VOLGOGRAD 265 D 322 - ORENBURG 596 D 2204 - AKTYUBINSK 838 D 474 -

STARYY OSKOL 462 YEREVAN 228 D - 3725 . 2 VORONEZH 459 D 2500 5515 . 7

GOR ' KIY 505 MOSCOW 415 D 2000 5 0

CHUSOVOY 528 PERM' 524 D 2000 -

NOVOSIBIRSK 709 NOVOSIBIRSK 709 D 1640 .2 -

KOMSOMOL ' SK -- 783 YUZHNO-SAKHALINSK 788 D 537 - na-AMURE VLADIVOSTOK 800 D 563 -

KULEBAKI 890 MINSK 84 D 1000 -

PETROVSK- 951 BLAGOVESHCHENSK 775 D 448 - ZABAYKAL ' SKIY VLADIVOSTOK 800 D 452 - NAKHODKA 802 F 10 0

- 1 4 3 -

NCSEER March 8 3

Table 37 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN E DESTINATION : E E* 1980 199 0

SUMGAIT 244 YEREVAN 228 D 870 --

TAGANROG 274 YEREVAN 228 D 342 .4 -

TULA 433 MOSCOW 415 D 147 .4 -

KOLPINO 27 LENINGRAD 24 D - 5 0 VENTSPILS 45 F 100 -

ELEKTROSTAL' 846 NONE -- - -

VOLGOGRAD 265 NONE - - - -

LIEPAYA 46 NONE - - - -

MOGILEV 325 NONE -- - -

RUSTAVI 910 NONE -- - -

BEKABAD 943 NONE - - - -

IZHEVSK 554 NONE -- - -

SEROV 532 NONE -- - -

KRASNOYARSK 750 NONE -- - -

BARNAUL 718 TASHKENT 607 D NA 340 1 MARI 622 D NA 35 9 DUSHANBE 630 D NA 39 8 FERGANA 634 D NA 954 ALMA-ATA 653 D NA 133 2 KOKCHETAV 692 D NA 350 8 NOVOSIBIRSK 709 D NA 470 5 SEMIPALATINSK 723 D NA 96 2 LENINSK -KUZNETSK 735 D NA 294 7 TERMEZ 629 F NA 11 0

ZHLOBIN 88 VILNIUS 74 D NA 166 MINSK 84. D NA 505 8 VENTSPILS 45 F NA 180

CHUL'MAN 963 KRASNOYARSK 750 D NA 129 . 1 BLAGOVESHCHENSK 775 D NA 173 4 YUZHNO -SAKHAEINSK 788 D NA 64 3 VLADIVOSTOK 800 D NA 118 5 IRKUTSK 858 D NA 134 2 NAUSHKI 761 F NA 35 NAKHODKA 802 F NA 150

- 1 44 -

NCSEER March 8 3

Table 37 . Continued : ------N N T YEA R 0 0 Y D D P OPTIMAL PROJECTE D ORIGIN E DESTINATION : E E* 1980 199 0

TAYSHET 754 KRASNOYARSK 750 D NA 1111 . 9 NORIL'SK 888 D NA 21 1

Tonal steel flows : 147900 16998 5 ------*D = flow of steel to domestic final demand nodes , F = flow of steel to export nodes . NA = not applicable for 1980, i .e ., potential steel sites for 1990 .

ton, but again, total constrained output at this northwest European mil l

is very small only equalling 50,000 tons . Marginal expansion a t

Magnitogorsk (586) would yield a potential system cost savings of 3 .2 7

rubles per ton while that at Krivoy Rog (172) would be nearly thre e

times as cost effective (8 .90 rubles/ton) . The upshot of the les s

constrained 1990 model year is that substantial economies of scale coul d

continue to be realized by expansion at Magnitogorsk (586), Krivoy Ro g

(172), Cherepovets (836), Chelyabinsk (581), and Staryy Oskol (462 )

combined with the addition of the new mills at Zhlobin (88), Barnau l

(718), Tayshet (754) and Chul ' man (963) . Caution must be advised thoug h

as to the use of the model ' s numerical results as accurate indicators o f

the actual absolute magnitude of proposed expansions because, as note d

above, much of the indicated expansion comes at the expense of alread y

existing mills . Accordingly, since the Soviets have a history o f

equating the productive life of capital facilites with their physical a s

opposed to economic life, we do not foresee them closing many, if any ,

of the mills shown by the model as ceasing production in 1990 . As suc h

then, the indicated optimal absolute increases in production woul d

constitute essentially excess supplies of steel .

- 1 4 5 -

NCSEER March 8 3

Technological Shifts between 1980 and 1990 (Table 36) :

Some very interesting changes occur with respect to technologica l

shifts between the 1980 and 1990 optimal computer scenarios . First ,

open-hearth, basic-oxygen and electric-arc steel totals are not abov e

the minimum requirement set for each for the 1990 run or thei r

respective 1980 maximum production levels (refer to Table 17, p . 47, an d

bottom of Table 36, p . 141) . Nonetheess, though, electric-arc furnace s

do produce one-million tons more in 1990 simulation than in 1980 one .

Second, the major shift is that three of the four new sites produc e

steel by the direct reduction method . Their combined output by this ne w

process amounts to over 21 .0-million tons, all but one-million tons o f

the modelled system-wide production gains from 1980 to 1990 . Third ,

although open-hearth steel production remained constant, a complet e

change over to variant Type H, which uses both coke and gas, i s

indicated . This pattern is just the opposite of 1980 . This switch i n

the input requirements for open-hearth steel clearly seems to be a

manifestation of rising coking coal costs compared with those of natura l

gas . Similar to 1980 the scrap metal supply problems forces the mode l

to manufacture all electric-arc steels by Type T, which uses both pi g

iron and scrap rather than scrap only .

Finally, the bottom of Table 36 (p . 141) lists the dual activitie s

for steel by process for both 1980 and 1990 . These dual values clearl y

support the trend toward the direct reduction technology and away fro m

electricity intensive electric-arc steels in particular . In both year s

the basic oxygen process is indicated as being more cost effective tha n

open-hearth processes . In other words, there is a constant and clea r

marginal cost effectiveness pattern by process in the two period s

- 1 48-

NCSEER March 8 3

ranging as follows from cheapest to most expensive : direct reduction ,

basic oxygen, open hearth, and electric-arc . In general, Soviet

planning decisions over the past ten to twenty years have clearl y

recognized these marginal cost relationships .

PRESENTATION OF OBJECTIVE FUNCTION COSTS (Table 38) :

The value of the objective function, the seven sub-objectives, an d

average cost figures are listed in Table 38 (refer to p . 13 and Appendi x

2, pp . 168-169 for definitions of the objective functions) . The tota l

minimized system cost (variable Z in Appendix 1, p . 160) amounts t o

13 .9-billion rubles for 1980 and a slightly higher 14 .4-billion fo r

1990 . The very substantial increases in regional energy and raw materia l

input prices used in the model, appear to be more than offset by th e

savings inherent in the relaxed production constraints and technologica l

process shifts which occur for 1990 . As a consequence, the averag e

total cost for steel production actually drops from 94 .14 rubles per to n

in 1980 to 84 .94 rubles per ton in 1990 .

Presumably because of either overall cheaper or less inpu t

intensive steel manufacture, the non-energy and non-transport costs o r

raw material and process costs (TOTAL NON-ENERGY) actually decline in a n

absolute sense by slightly less than 800-million rubles . Concomitantly ,

of course, their relative share of total costs drops from 78 .2 per cen t

to 69 .9 per cent .

All other production factors or sub-objectives increase bot h

absolutely and relatively . The smallest increase accrues to expor t

transportation costs . The absolute increases in coal costs and domesti c

transportation costs are quite similar, 216 .6-million rubles versu s

207 .5-million rubles, respectively . As

- 1 49-

NCSEER March 8 3

------Table 38 . OBJECTIVE FUNCTIONS AND SYSTEM COST DAT A ------1980 199 0

OBJECTIVE COST PERCENT COST PERCEN T FUNCTION : in Rubles of Total in Rubles of Tota l

COAL 1,666,628,400 12 .0 1,883,214,716 13 . 0

ELECTRICITY 367,101,060 2 .6 733,588,622 5 . 0

NATURAE GAS 228,514,864 1 .6 710,053,36 4 . 9

TOTAL ENERGY* 2,180,650,424* 15 .7* 3,166,351,174* 21 .9*

TOTAL NON-ENERGY 10,891,443,126 78 .2 10,092,297,922 69 . 9

TRANSPORT (Domestic) 667,689,336 4 .8 875,214,562 6 . 1

TRANSPORT (Export) 102,107,597 0 .7 144,174,824 1 . 0

TOTAL COST 13,923,484,383 100 .0 14,438,544,483 100 . 0 (Min Z value) )

Ave . Steel Cost : 94 .14 rub ./ton 84 .94 rub ./to n

*Excludes the cost of exported coking coal included in the COAL variable ; therefore, COAL + ELECTRICITY + NATURAL GAS sum to a number greater tha n TOTAL ENERGY .

Note : For fuller explanations of objective functions refer to Appendix 2 , pp . 168-169 .

------

expected the largest absolute and relative cost increases are registere d

by the energy inputs, led by natural gas . Overall, coal (excluding th e

cost of coking coal exports) , natural gas, and electricity represent a n

increasing fraction of the total steel production bill, rising from 15 . 7

per cent in 1980 to 21 .9 per cent in 1990 (TOTAL ENERGY) . Lastly, whil e

natural gas costs triple as a relative share of the total bill an d

electricity's share nearly doubles, coking coal's share, again ,

excluding the cost of exported coking coal, is nearly static rising fro m

11 .5 per cent to 12 per cent . The relative share of the total syste m

cost attributable to transport costs increases from 5 .5 percent in 1980

-150-

NCSEER March 8 3

to 7 .1 percent in 1990 . Clearly, then, only the total raw material an d

process costs are modelled as decreasing both relatively and absolutely .

This situation arises despite the general increases in raw materia l

extraction costs . It is evident that the model economized by realizin g

greater economies of scale, switching to cheaper processes, and changin g

production locations .

SUMMARY AND CONCLUSION S

In the introductory section of this investigation eight question s

or issues were raised (refer to pp . 4-5) . All of these eight have been

addressed .

The first four questions pertained to a series of forecasts .

Accordingly, several sets of forecasts have been enumerated and

incorporated into the raw data base of the SISEM model . Estimate s

and/or projections of the domestic resource base of Soviet high-grad e

and low-grade iron ore, coking coal, coke, scrap, natural gas an d

electricity supplies have been developed and used (question 1) .

Similarly, the Soviet export potential for coking coal, iron ore, pi g

iron, and steel has been forecast (questions 2 & 3) . Also, the domesti c

final demands for pig iron and steel have been developed (question 4) .

As illustrated by Table 23 (pp . 61-62) the optimal supply locations fo r

coking coal exports tends to shift increasingly eastward to the Kuzba s

and beyond for both European and Pacific exports . Like the curren t

actual pattern, the Ukraine dominates as the source region for Sovie t

iron ore exports in both model years . The Kola Peninsula becomes th e

secondary supplier in 1980 and shares that role with the Kursk Magneti c

Anomaly (KMA) in 1990 . Nonetheless, the Ukraine ' s absolute export flow s

decline slightly while its relative export market share in the mode l

- 1 5 1 -

NCSEER March 8 3

drops from 92 .5 percent in 1980 to 82 percent in 1990 . Also, unlike th e

actual current iron ore export flow pattern, the KM does not optimall y

serve any export markets in 1980 but does export 3 .0-million ton s

through Riga in 1990 . Ukrainian pig iron furnaces and those at Lipets k

dominate the Soviet pig iron export markets in both years . Ukrainia n

dominance in the steel export flows in 1980 are seriously challenged b y

1990 by two Urals competitors, Magnitogorsk and Chelyabinsk .

The fifth question dealt with the role of planned or propose d

transportion developments . Clearly, the modelled completion of th e

Baykal-Amur-Mainline railroad (BAM) by 1990 was a necessary condition t o

allow the proposed Chul ' man (963) steel site to enter the optima l

solution for 1990 as an active steel node . Transportation costs both i n

absolute and relative amounts are projected to rise but amount to onl y

5 .5 and 7 .1 per cent of total costs for 1980 and 1990, respectively .

The sixth issue focused on the potential impact of technologica l

change within the Soviet ferrous metallurgical industry . The mode l

clearly indicates that significant technological process shifts ar e

warranted . For example, the model satisfies steel demands whil e

requiring 3 .7-million tons less pig iron in 1980 than the Soviet s

report . This disparity could be interpreted as indicating the order o f

magnitude of sub-optimality in the Soviet production system . In fact ,

due to the indicated switch to vast quantities of direct reduction stee l

for 1990 (22-million out of approximately 170-million tons), th e

modelled pig iron production for 1990 is still less than the reporte d

actual tonnage for 1980 (105 .1-million versus 107 .3-million tons) .

Krivoy Rog (172) stands out as the clear place for any expanded pig iro n

production . Finally, none of the five potential pig iron sites : Stary y

Oskol (462), Barnaul (718), Tayshet (754), Svobodnyy (981) or Tynd a

- 1 5 2 -

NCSEER March 8 3

(962) appears as an optimal choice .

Several differences appear between the 1980 and 1990 stee l

production patterns . First, all four of the proposed new stee l

locations, Barnaul (718), Zhlobin (88), Chul ' man (963) and Tayshet (754 )

enter the optimal solution for 1990, the first with over 18-millio n

tons, the second two with 5 .4-million and 5 .2-million tons ,

respectively ; and Tayshet with 1 .3-million tons . More importantly, al l

of these new sites except Tayshet (754) are shown as actively using th e

new direct reduction process to produce from a low of 4 .2-million ton s

at Zhlobin (88) to a high of over 12 .6-million tons at Barnaul (718) .

Second, whereas many locations are producing at capacity in 1980, th e

relaxed upper limits on steel plant capacities for 1990 result in onl y

two locations, Magnitogorsk (586) and Krivoy Rog (172) operating a t

their upper limit . Several small sites fail to enter either solution ,

fewest of all in 1990 . In effect, the model selects in favor of scal e

economies, or in other words, the Soviets still appear to hav e

unrealized potential savings to be garnered from scale economies .

Third, because of the scrap supply problems the electric-arc proces s

(Type A) using scrap only does not produce steel in either year .

Fourth, practically all of the steel produced by open-hearth furnaces i n

1980 used coke only . For 1990, however, the open-hearth furnaces ar e

optimally shown as using much less coke and much more natural gas .

The seventh issue referred to a check on the internal consistenc y

of the relevant Soviet production and technological coefficient data .

In this regard the most significant findings is the the scra p

insufficiency problem . Scrap inputs appear as binding constraints i n

both runs . Simply stated, using Soviet data, one can not produce a s

much steel in 1980 as Soviet statistics claim . Even allowing for th e

- 1 5 3 -

NCSEER March 8 3

tonnage of ferro-alloys, lower input coefficients, etc ., a significan t

shortfall still would exist . The model ' s results imply that the Soviet s

are double counting steel production by as much as 15 .55-million tons o r

10 .5 per cent for 1980 by reusing large quantities of on-site scrap .

The scrap shortage is dispersed throughout the Ukraine, Urals, and Wes t

Siberia in 1980, but confined to the Urals in the 1990 run .

Furthermore, other modelling efforts indicate the magnitude of " on-site "

scrap may be around thirty-million tons . If true, then, the actua l

published Soviet crude steel production data could overstate " useful " o r

net steel production by as much as nearly thirty-million tons ,

essentialy twice the quantity suggested here! [42 ]

The eighth and final question was an assessment of the impacts o f

projected changes in the availability and relative prices of energy an d

raw material inputs both by type and region . Accordingly, coking coa l

presents no " physical supply " problems in the model primarily becaus e

coke is used in other industries beside steel and hence the model yield s

large coking coal slack values . On the other hand, the escalating cost s

of Donbas coals make West Siberian coals marketable throughout much o f

the European U .S .S .R by 1990 . Although the model is not constrained b y

crude iron ore or enrichment in either simulation, we believe both are ,

in fact, already bottlenecks . There are some hints of this in the 199 0

run where all high-grade ore sites are forced to operate at thei r

maximum modelled capacities . The indicated changes in optimal low-grad e

and enriched ore production suggests that the Soviets could economize b y

concentrating their expansion efforts in the KMA, new deposits i n

northwest European Russia, Kremenchug, along the DAM, and much less s o

at Krivoy Rog and Rudnyy . Over the modelled decade both the KMA an d

Ukrainian ore deposits play a significantly larger role in Urals iro n

- 1 5 4-

NCSEER March 8 3

and steel production .

Neither electricity nor natural gas appear to be limiting factor s

overall for the foreseeable future as both have vast slack capacities .

On a regional scale, however, Europe could well need both mor e

electricity and gas, while the Urals may need more West Siberian gas .

The relative share of the total systems cost arising from each of thes e

energy inputs increases substantially from 1980 to 1990, that of natura l

gas tripling and that of electricity nearly doubling .

Finally, the model indicates the possibility of very significan t

average production costs savings of approximately 9 .20 rubles per ton o r

9 .8 per cent of the 1990 over the much more constrained 1980 model year .

- 1 5 5 -

NCSEER March 8 3

FOOTNOTE S

[1] For a concise review of the Soviet debate over interdependenc e versus autarky see Marshall I . Goldman, " Autarchy o r Intergration--the USSR and the World Economy, " in U .S . Congress , JointPerspective Economic Committee, Soviet Economy in a New (Washington, D .C . : U .S . Government Printing Office, 1976), pp . 81-96 ; and Marshall I . Goldman, "The Changing Role of Ra w Material Exports and Soviet Foreign Trade , " Discussion Paper No . 8 , Association of American Geographers Project on Soviet Natura l Resources in the World Economy (Washington, D .C . : AAG, 1979), pp . 44-52 and under same title in Soviet Economy in a Time of Chang e (Washington, D .C ., U .S . Government Printing Office, 1979), pp - 179-181 .

[2] For example, see Goldman, " The Changing Role of Raw Materia l Exports . . ., " op . cit ., footnote 1, pp . l-44 ; and Edward H . Hewett, " Soviet Primary Product Exports to CMEA and the West , " Discussion Paper No . 9, Association of American Geographers , Project on Soviet Natural Resources in the World Econom y (Washingotn, D .C . : AAG, 1979), pp . 1-46 .

[3] Goldman, " The Changing Role of Raw Material Exports .. ., " op . cit ., footnote 1 .

[4] For example, see Goldman, " The Changing Role of Raw Materia l Exports .. ., " op . cit ., footnote 1, especially pp 30-43 ; Lesli e Dienes and Theodore Shabad, The Soviet Ener gy System (Washington , D .C . : Scripta Publishing Co ., 1979), pp . 217-294 ; Leslie Dienes , " Soviet Energy Policy and the Hydrocarbons, " Discussion Paper No . 2, Association of American Geographers, Project on Soviet Resource s in the World Economy (Washington, D .C . :, AAG, 1978) ; Lesli e Dienes, " The Soviet Union : Energy Crunch Ahead?, " Problems of Communism, Vol . 26, No . 5, Sept .-Oct ., 1977, pp . 41-60 an d " Soviet Energy Resources and Prospects, " Current History, Marc h 1978 ; Robert Campbell, "Implications for the Soviet Economy o f Soviet Energy Prospects, " The ACES Bulletin, Spring 1978, p p . 37-52 ; Robert Campbell, Soviet Energy R &D : Goals, Planning an d Organization (Rand Corporation : R-2253-DOE, May 1978) ; Rober t Campbell, Soviet Energy Balances (Rand Corporation : R-2257-DOE , October 1978) ; Arthur W . Wright, " The Soviet Union in World Energ y Markets, " in Edward W . Erichson and Leonard Waserman, eds ., Th e Energy Question : An International Failure of Policy, Vol . I (Toronto : University of Toronto Press, 1974) ; U .S ., C .I .A . , Prospects for Soviet Oil Production, Er-77-10270, Washington , April 1977 ; U .S ., C .I .A ., Prospects for Soviet Oil Production :A Supplementary Analysis, ER-77-10425, June 1977 ; and The Sovie t Oil Situation : An Evaluation of CIA Analyses of Soviet Oi l Production, Staff Report of the Senate Select Committee o n Intelligence, United States Senate (Washington, D .C . : U .S . Government Printing Office, 1978, 15 pp .) .

[5] For reserve figures of KMA (/Kursk Magnetic Anomaly), see : Ye . I . Kapitonov, " The Kursk magnetic Anomaly and Its Development , " Sovie t Geography : Review and Translation, Vol . 4, No . 5, May 1963, pp .

- 1 56-

NCSEER March 8 3

10-15 and V . P . Korostik, " Comparative Analysis of the Iron-Or e Basins of Krivoy Rog and the Kursk Magnetic Anomaly, " Sovie t Geography : Review and Translation, Vol . XX, No . 4, April 1979 , pp . 225-230prirodnykh; and G . I . Martsinkevich, Ispo ' zovaniy e resursov i okhrana prirody (Minsk : BGU, 1977), p . 64 . For exampl e of coal and especially coking coal reserves, see Dienes and Shabad , The Soviet Energy System,op . cit ., footnote 4, pp . 106 , 121-124 ; and lain F . Elliot, The Soviet Energy BALANCE (New York : Praeger Publishers, 1974), pp . 122-175 .

[6] Hewett,2, "Soviet Primary Product Export .. ., " op . cit ., footnot e p . 55 ; and Theodore Shabad, " Soviet Re g ional Policy and CMEA Integration, " Soviet Geography : Review and Translation, Vol . XX , No . 4, April 1979, pp . 237 and 240-244 ,

[7] Dienes and Shabad,The Soviet Energy System, op . cit . , footnote 4, pp . 225-228 .

[8] Craig ZumBrunnen, Jeffrey Osleeb and Charles Morrow-Jones , " Modeling the Optimal Commodity Flows within Soviet Ferrou s Metallurgy_ : Implications for Soviet Iron Ore and Coking Coa l Exports, " Discussion Pa per No . IA, Association of America n Geographers Project on Soviet Natural Resources in the Worl d Economy (Washington, D .C .: AAG, 1979, 57 pages) and Tony Misko an d Craig ZumBrunnen, " Soviet Iron Ore : Its Domestic and Worl d Implications, " Soviet Resources in the World Economy (Univ . o f Chicago Press, forthcoming, 1983) .

[9] ZumBrunnen Osleeb and Jones, "Modeling the Optimal Commodit y Flows .. ., " op . cit ., footnote 8 .

[10] Dienes and Shabad, The Soviet Energy System, op, cit ., footnot e 4, p . 123 .

[11] Pravda, April 3, 1978, p . 2 .

[12] Because our earlier model utilized only average steel productio n costs ratherr than using a non-linear production function combine d with the fact that this Sovet value (35 rubles/ton) probably doe s not incorporate all amortization costs of these hypothetical ne w plants, the model predicted closures, expansions, contractions an d additions must be considered suggestive rather than definitive .

[13] Primary sources for this transport system were Skhema zheleznyk h dorog SSSR (Moskva : SSSR Ministerstvo putey soobshcheniya, 1972 ) and Atlas SSSR (Moskva : Glavnoye upravleniye geodezii i kartografi i pri Sovete Ministrov SSSR, 1969), pp . 102-103 . Surface roads wer e included initially only to link cities over 65,000 in 1970 no t having railroad connections .

[14] While the plotting software we developed could plot up to 1 6 classes of five different symbols, for visual clarity we opted t o use only four classes at most for any given commodity . An Apple I I microcomputer linked to a Watanabe plotter was used to plot Maps 1 through 16 .

- 1 5 7 -

NCSEER March 8 3

[15] Average of the ratios 0 .9/l .75 and 1 .20/2 .38, water rate/rail rat e for ore-coal-coke and water rate/rail rate for scrap, pig an d steel, respectively .

[16] Average of the ratios 3 .50/1 .75 and 4 .75/2 .38, surface/rail rat e for ore-coal-coke and surface/rail rate for scrap-pig iron-steel , respectively .

[17] ZumBrunnen, Osleeb & Jones, "Modeling the Optimal Commodit y Flow .. ., " op . cit ., footnote 8 .

[18] Cost data primarily derived from USSR : Coal Industry Problems an d Prospects (Washington, D .C . : C .I .A ., 1980), pp . 5-20 .

[19] For the argument justifying the ubiquitousness and/or minor cos t differential significance of these raw materials see Allan Rodgers , "Industrial Inertia : A Major Factor in Location of the Stee l Industry in the U .S ., " Geogrphical Review, Vol . 42 (Januar y 1952) : p . 57 .

[20] Wassily Leontief, " Academic Economics," Science 217 :4555 (9 Jul y 1982), pp . 104 and 107 .

[21] Shabad, " Soviet Regional Policy and CMEA Integration, " Sovie t Geography : Review and Translation, Vol . XX, No . 4 (April 1979) : p . 243 ; PRAVDA, January 31, 1975 .

[22] ZumBrunnen, Osleeb & Jones, " Modeling the Optimal Commodit y Flow . . ., " op . cit ., footnote 8, pp . 24-25 .

[23] For a discussion of our rationale for including capital as well a s variable costs for all processes at all sites for all years refe r to the first paragraph under the " Steel Dual Activities for 1980 " heading on page 133 in the text .

[24] For example, see Theodore Shabad, " News Notes, " Soviet Geography : Review and Translation, April issues over the last several year s and Misko & ZumBrunnen, "Soviet Iron Ore . . ., " op . cit ., footnot e 8 .

[25] Misko & ZumBunnen, " Soviet Iron Ore . . ., " op . cit ., footnote 8 .

[26] N. D. Lelyukhina, Ekonomicheskiya effektivnost' razmeshcheniy a chernoy metallurgii (Moskva : " Nauka, " 1973), pp . 18, 19 & 22 .

[27] Narodnoye khozyaystvo SSSR v 1970 g . (Moskva : "Statistika, " 1971) , pp . 191 and 616 .

[28] Lelyukhina, Ekonomicheskiya effektivnost ' .. ., op, cit . , footnote 25, p . 22 .

[29] For city sizes, see Narodnoye khozyaystvo SSSR v 1970 op . cit ., footnote 26, pp . 35-45 .

[30] For urban employment data by sectors, see Chauncy Harris, Cities o f the Soviet Union (Chicago : Rand McNally and Co ., 1970), pp . 70-78 .

- 1 58 -

NCSEER March 8 3

[31] The cost of each computer run including both the program t o generate teh problem form the various data tables as well as th e MPSX algorithm ranged between $750.00 and S900 .00 .

[32] Narodnoye khozyaystvo SSSR y 1980 g . (Moskva : "Statistika, " 1981) , p . 159 .

[33] "Uchest ' v planakh, " Sotsialisticheskaya industriya (Oct . 1, 1978) , p . 2 ; " Stal ' noy profil ' , " Pravda (Jan . 22, 1979), p . 2 ; an d "Slovo o metalle, " Sotsialisticheskaya industriya (Jan . 19, 1979) , p . 2 . These references were provided thanks to Professor Gregor y Grossman .

[34] Misko & ZumBrunnen, "Soviet Iron Ore . . ., " op . cit ., footnote 8 . Also, see various April issues of Soviet Geography : Review & Translation over the last few years .

[35] Phone conversation with Dr . Lee W . Bettis .

[36] Ibid .

[37] 16,454,000 excess scrap demand/1.058 tons of scrap/ton of steel by process A equals 15,552,079 tons of Type A steel ; 15 .55-millio n tons/147 .9-million tons times 100 equals 10 .5% .

[38] Narodnoye khozyaystvo SSSR v 1980 g., on . cit ., footnote 31, p . 158 .

[39] Note : remember that both types A & T represent variant s electric-arc furnaces, H & J different variants of open-heart h furnaces . Also, the coefficients for some processes at severa l locations are site-specific .

[40] ZumBrunnen, Osleeb & Jones, " Modeling the Optimal Commodit y Flow .. ., " op . cit ., footnote 8, pp . 24-25 .

[41] Ibid .

[42] Phone conservation with Dr . Lee W . Bettis .

-159- APPENDI

X 1 SOVIET IRON AND STEEL ENERGY MODEL (SISEM) Subject to : Coke Plant Mass Balance s

Coal Inpu t

Enrichment Plant Mass Balance s

Enrichment Plant Mas s Balances Iron Inpu t

Electricity Inpu t

Pig Iron Plant Mass Balance s

Coke Inpu t

Electricity Inpu t

Natural Gas Inpu t

Iron Ore Inpu t

Scrap Metal Input Steel Plant Mass Balance s Iron 0re Inpu t

Pig Iron Inpu t

Scrap Inpu t

Electricity Inpu t

Natural Gas Inpu t

Coke Inpu t

Total Plant Stee l Productio n

Final Demand s Pig Iron Deman d

Coal Deman d

Iron 0re Deman d

Steel Demand NCSEER March 8 3

Limit Constraint s Upper limit on steel productio n by process by plan t

Lower limit on steel production by process by plan t

Upper limit on steel productio n by plan t

Upper limit on pig iro n production by plan t

Lower limit on pig iro n production by plan t

Upper limit iron or e production by min e

Lower limit iron or e production by min e

Upper limit coal production by min e

Lower limit coal productio n by min e

Scrap upper limit by sourc e

Upper limit on cok e production by battery

Lower limit on cok e production by battery

Upper limit on enriched iro n ore by plan t

Lower limit on enriched iro n ore by plan t

Upper limit on electricit y for entire system

Upper limit on natural ga s for entire syste m

Upper limit on process fo r system

Lower limit on proces s for system

-163 - NCSEER March 8 3

Variable Definitions :

Quantity of iron ore from mine i shipped to enriching plant j

Quantity of iron ore from mine i shipped to pig iron plant k

Quantity of iron ore from mine i shipped to steel plant k, , process n Quantity of iron ore from mine i shipped to export node e

Quantity of coking coal from mine i shipped to coking plant j

Quantity of coking coal from mine i shipped to export node e

Quantity of enriched ore from enrichment plant j shipped t o pig iron plant k Quantity of enriched ore from enrichment plant j shipped t o steel plant Q, process n Quantity of enriched ore from enrichment plant j shipped t o export node e Quantity of coke from coking plant j shipped to pig plant k

Quantity of coke from coking plant j shipped to steel plant i , process n Quantity of pig iron from plant k shipped to steel plant k , process n Quantity of pig iron from plant k shipped to final domesti c demand node d Quantity of pig iron from plant k shipped to final expor t demand node e Quantity of electricity consumed by enrichment plant j

Quantity of electricity consumed by pig iron plant k

Quantity of electricity consumed by steel plant i, process n

Quantity of natural gas consumed by pig iron plant k

Quantity of-natural gas consumed by steel plant Q, process n

Quantity cf scrap metal from source i shipped to pig iron plant k

Quantity of scrap metal from source i shipped to steel plant 9 , process n Quantity of steel from steel plant 2,, process n, shipped to stee l supply node m (note : equivalent to total steel produced by plant Q )

-164 NCSEER March 8 3

Quantity of steel from supply node m shipped to final domesti c demand node d Quantity of steel from supply node m shipped to export node e

Cost of iron ore at mine i

Cost of coal at mine i

Cost of enriched iron at plant j

Cost of coke at plant j

Cost of pig iron at plant k

Cost of scrap steel at source i

Cost of electricit y

Cost of natural ga s

Cost of steel from process n at plant x

Transport cost of iron ore from mine i to enriching plant j

Transport cost of iron ore from mine i to pig iron plant k

Transport cost of iron ore from mine i to steel plant Q , process n Transport cost of iron ore from mine to final demand e (export ) Transport cost of enriched iron from enriching plant j t o pig iron plant k Transport cost of enriched iron from enriching plant j t o steel mill Q process n Transport cost of enriched iron from enriching plant j t o final demand e (export ) Transport cost of coal from mine i to coking plant j

Transport cost of coal from mine i to final demand e (export )

Transport cost of coke from coking plant j to pig iron plant k Transport cost of coke from coking plant j to steel mill 2 process n

- 1 65-

NCSEER March 8 3

Transport cost of pig iron from pig iron plant k to stee l mill Q process n Transport cost of pig iron from pig iron plant k to fina l domestic demand d Transport cost of pig iron from pig iron plant k to fina l export demand e Transport cost of scrap from scrap source i to pig iro n plant k Transport cost of scrap from scrap source i to steel mill Q process i u Transport cost of steel from steel supply node m to fina l domestic demand d Transport cost of steel from steel supply node m to fina l domestic demand e Final pig iron domestic demand at node d

Final pig iron export demand at node e

Final steel domestic demand at node d

Final steel export demand at node e

Final coking coal export demand at node e

Final iron ore export demand at node e

Upper limit on steel production by plant and proces s

Lower limit on steel production by plant and proces s

Upper limit on steel production by all processes at mill 2

Upper limit on pig iron production at plant k

Lower limit on pig iron production at plant k

Upper limit on iron ore production at mine i

Lower limit on iron ore production at mine i

Upper limit on coal production at mine i

Lower limit on coal production at mine i

Upper limit on scrap from site i

Upper limit on coke production at plant j

Lower limit on coke production at plant j

-166-

NCSEER March 8 3

Upper limit on enriched iron ore produced at plant j

Lower limit on enriched iron ore produced at plant j

Upper limit for steel produced by the n th process for th e system Lower limit for steel produced by the n th process for th e system Upper limit of electricity availabl e

Upper limit of natural gas availabl e

The input coefficient for coal at the j th coking plan t

The input coefficient for iron ore at the j th enrichin g plan t The input coefficient for electricity at the j th enrichin g plan t The input coefficient for coke at the k th pig iron plan t

The input coefficient for electricity at the kth pig iro n plan t The input coefficient for natural gas at the k th pig iro n plan t The input coefficient for iron ore at the kth pig iro n plan t The input coefficient for scrap at the kth pig iron plan t

The input coefficient for coke at the Q th steel mil l n proces s

The input coefficient for electricity at the Qth steel mil l n proces s ith The input coefficient for natural gas at the steel mil l n proces s th The input coefficient for iron ore at the Qth steel mill n proces s t h The input coefficient for scrap at the Q `h steel mill n proces s ith t h The input coefficient for pig iron at the steel mill n process

- 1 6 7 -

NCSEER March 8 3

- 1 6 8 -- NCSEER March 198 3

- 1 6 9