prediction of the blast furnace process by mathematical model

ISIJ International, Vol. 32 (1992). No. 4, pp. 481-488

Prediction of the Blast Furnace Process by a Mathematical Model

XuegongBl. Krister TORSSELL1)and Olle WIJK2)

Formerly Department of Process Metallurgy, The Royal Institute of Technology, Nowat Division of lronmaking. Wuhanlron and Steel University, 430081 Box 123, Wuhan.Hubei. People's Republic of China. 1) Formerly Department ofProcess Metallurgy, The Royal Institute of Technology. Nowat ABB POWDERMETAB, S-735 OO, Surahammar,Sweden. 2) Department of Process Metallurgy, The Royal Institute of Technology. S-1 OO44 Stockholm, Sweden.

(Received on September9. 1991; accepted in final form on January 24. 1992)

A prediction model has been developed based on the simulation model presented previously. Thesemodels are the two componentsof the KTHBlast Furnace Process Model. The concept of this predictionmodel is to use the samefundamental equations as in the simulation model and to use someoutput of asimulation of a blast furnace. Certain assumptions should be madefor an individual change in operationalconditions in order to build up the massbalance and heat balance submodelsfor the determination of theboundary conditions in a prediction. The ore to fuel ratio and the COutilization are adjustable parametersin the model. The furnace internal state as well as furnace productivity and fuel consumption in the last

iteration are considered to be the predicted results.

The prediction model has been designed for the following five cases: I) increased blast temperature, 2)

oxygen enrichment of the blast, 3) coai injection, 4) coal injection combinedwith oxygen enrichment and5) changedcoke quality. Moreover, this model can also be used for analysis of the thermal conditionsin a blast furnace whenan operational parameter, such as blast temperature, coke moisture and iron

content in the ore, fluctuates.

The predicted operationa] indices were comparedto the ones from industtial tests. The validity of the

KTHmodel is indicated by this comparison.

KEYWORDS:ironmaking; blast furnace process; prediction; mathematical modei; operational indices,

thermal conditions.

1. Introduction

A previous article by the authorsl) presents a sim-ulation model of the blast furnace process. Based onthis simulation model a prediction model has beendeveloped in order to predict the internal state of the

furnace as well as fuel consumption and furnace pro-ductivity whenan operational action is taken (a case).

This modelcan be used to analyse the thermal conditionsin the blast furnace whensomeoperational parametersfluctuate and to predict the thermal conditions as well

operational indices for someselected cases.Theprediction modelhas been programmedin ANSI

FORTRAN77 and inplemented in Nord-500 and Vaxminicomputers. By using the simulation results for aNordic blast furnace, prediction tests have been carried

out to makeanalysis of the thermal conditions in the

furnace for three simple cases and to makeprediction ofoperational indices for five cases. This paper presentsthe development of the prediction model and the

prediction results by the model.

2. Outline of the Prediction Model

In order to predict the internal state of the furnace aswell as fuel consumption and furnace productivity when

an operating action is taken, a prediction model hasbeen developed based on the simulation model. Thebasic concept of the prediction model is to utilize the

samefundamental equations as in the simulation mod-el and someoutput from a simulation of a furnace. Theoutput include: l) the correction factor for the rate

of the gaseous reduction of iron oxides, 2) the cor-rection factor for convective heat transfer between the

gas and condensed phases in the dropping zone, 3)

the ratio between the temperature of the condensedphases at the tuyere level and the fiame temperature,and 4) heat losses through the furnace wall per tonneof hot metal.

2.1. Division of the Furnace into Zonesand Reactions

The division of the furnace into zones in the predic-

tion model is the sameas in the simulation model, i,e.

the lumpy zone, the softening zone, the melting zoneand the dropping zone. The considered reactions arealso the same, i,e. reduction of iron ore by COand H2gases, solution loss reaction of coke, gasification of cokeby H20, decomposition of limestone, reaction between

COand steam, water evaporation, melting of reducediron and slag formation. The correction factor ~22 for

indirect reduction rate of iron ore in the softening andmelting zones is maintained constant. Heat transfer

481 C 1992 ISIJ

ISIJ International, Vol.

phenomenain the furnace are assumedunchanged.

2.2. FundamentalEquations

The fundamental equations are also the sameas in

the simulation model except for the correction factor

~li for the heat losses in the differential equation for

gas temperature.

3. Determination of BoundaryConditions

3.1. UpperBoundaryConditions

The nonlinear equation system is the sameas in the

simulation model. However, two important parametersin the system, i.e. ore/fuel ratio and n*~ are calculated

from the operating data in the simulation model, whilethey are adjustable parameters in the prediction modelin order to satisfy the lower boundary conditions of the

Ordinary Differential Equation (ODE)system.

- composition of raw materials and products are keptconstant,the slag/hot metal ratio is constant,

- the limestonelhot metal ratio is maintained constant,

- the dust/hot metal ratio is constant, and

- nH, is kept constant.As for the blast, its volume is assumedto decrease to

such an extent that the racewaygas volume is maintainedconstant in the case of oxygen enrichment, and its

oxygen content must be increased in order to keep aconstant flame temperature in the combinedcase of coal

powder injection and oxygen enrichment.

Mass flowrate of the hot metal iron ore and coketogether with volumetric fiowrate and COcontent in the

top gas are output of the massbalance submodel for aset of ore/fuel ratio and n*. values. Accordingly, massflowrate of the slag, dust and limestone, C02and H20contents in the top gas, degree of direct reduction andindirect reduction of FeOare also calculated.

3.2. Lower BoundaryConditions

The following assumptions are madeto establish theheat balance submodelin the prediction model:

- temperature of the hot metal and slag is kept constant,

- heat losses are kept constant, and

- the ratio between the flame temperature and the

condensed phase temperature at the tuyere level is

kept constant.The upper boundary condition of gas temperature is

an output of the heat balance, and the temperature ofthe gas and condensed phases at the tuyere level areestimated iteratively by a half-way methodaccording toPehlke.2)

4. The Order of Calculation in the Prediction Model

The order of calculation is summarized in Fig. l.

First, initial value of ore/fuel ratio and n*. is preset anddata are read from the data base created in a simulation.

Massbalance and heat balance are madeto determinethe boundary conditions. The solution of the ODEsystem in the lumpy, softening and melting zones arethe sameas in the simulation. The iterative process in

simulation for searching the upper boundary condition

C 1992 ISIJ 482

32 (1992), No. 4

start

Read data f rorn the data base

setset

OIFl~

co

ratio

,(ass balance for- calculatin9 mass

- determinating fs

- calculating Yco'

of lowrate

, COO,

gYd and Y

of H~t

CO O,Z,g

H2

andoH2,g

cokeo

, ps

Heat

-T-T

balanceO

and,4

for

g, 4

calculating

gT

set ~11

Solvin ODE in the l um zone

no CO 12*g

yes

o P?.

Solvin ODE in the softenin zone

Solvin difference e s. in the meltin zone

set ~13

Solvin ODE in the dro In zone

no lT =Tg, 4

yes

?g

".

y..

w*it~ d*t*

E~d

F]g. l. Ftow chart of catcutation in the prediction model.

of gas temperature and for searching correction factor

~22 for the indirect reduction rate are no longer

necessary in the prediction.

There is only one parameter ~l3 for h~ that shouldbe searched in order to satisfy the lower boundarycondition of gas temperature whenthe ODEsystem is

solved in the dropping zone.Thedetermined ternperature of the condensedphases

at the tuyere level is supposed to be equal to its lowerboundary condition. If not, a newset of ore/fuel ratio

and n*. must be presented and the calculation repeated.

Whenthermal condition in the furnace is to beanalysed, which might be termed as a simple prediction,

the ore/fuel ratio and COutilization in a simulation test

are used for determining the upper boundary conditions

and the iterative procedure to search for the appropriateore/fuel ratio is no more necessary. The calculated

temperature of the condensedphases at the tuyere level

does not satisfy its lower boundary condition because it

is determined by the assumption that hot metal andslag temperatures do not change. The hot metal andslag temperatures could be estimated by an iterative

procedure until the calculated temperature of the

condensedphases at the tuyere level is equal to its lower

boundary condition with the ore/fuel ratio and CO

ISIJ International, Vol, 32

utilization being maintained constant. However, it is

not necessary to do so for the purpose of analysingthe thermal condition when an operating parameterfluctuates

5. Analysis of Thermal Conditions in the Furnace

In practical operation someparameters such as blast

temperature, iron content in the ore and coke moisturefluctuate to someextent. It is interesting to analyze thethermal conditlon in the furnace when an operatingparameter fluctuates and no counteractive measuresaretaken accordingly because the fluctuation is either notdetected or neglected by operators.

5.1. Blast Temperature

An analysis has been conducted for 10 degreescentigrade of blast temperature. The test results areshownin Fig. 2. In all figures of this paper the dottedlines are for the base case. The results suggest that thethermal level of the furnace is higher due to the increasedheat input of the whole furnace.

o

5Ea'

~ 10

o~'

~o 15!Lthe'

~:'e

v' 20

- GasSolid -

(1992). No. 4

25

O 1000 1500 2000 2500500Temperature, oc

Fig. 2. Changein the thermal condition for lO'C increase ofblast temperature.

~~

a;

o~'

~:

oLLh

cU

u4J

c:~

O

5.2. CokeMoisture

An analysis has been carried out for I o/o increase ofcoke moisture, and the results are shownin Fig. 3. It is

obvious that the furnace thermal level becomeshigher,and that the temperature of the condensedphases at thetuyere level decreases. This is due to decreased input ofdry coke, and consequently decreased heat input of thewhole furnace.

5.3. Iron Content in the OreAn analysis has also been carried out for O.5olo

increase of iron content in the ore. The test results areshownin Fig. 4. It is clear that the thermal level of thefurnace becomeslower, and that the temperature of thecondensedphases at the tuyere level decreases. This is

due to the decreased coke consumption and heat inputof the whole furance.

Theanalysis above illustrate the importance to reducefluctuation of the operating parameters, especially cokemoisture and Fe content in the ores, in order to obtain

a stable operation.

6. Prediction of Operating Indices

It is very interesting to predict the internal state of thefurnace and the operating indices before an operatingaction is to be undertaken. At present, the model hasbeen built up to predict the effect of the following five

operating actions:

ACTIONI - increased blast temperature,ACTION2- oxygen enrichment of blast,

ACTION3- coal powder injection,

ACTION4combination of coal powder injection andoxygen enrichment, and

ACTION5- changedcoke quality.

6.1. Prediction Results

Predictions have been performed based on thesimulation results of BFA.1)

6. I.1.

Increased Blast TemperatureThree levels were chosen for the prediction test. The

5

lO

15

20

25

\\\\

~\x\1ll1Il

te;~~

t$~h~::\

Solid - \ - Gas:~-

\

O 1000 1500 2000 P_500500

Temperature, oc

Fig, 3. Changein the thermal condition for + I o/, increase ofcoke moisture.

O

5~e;s:'- IOuo+'v]

E 15OL

th

q,

~ 20

c::

25

\\\\

~~\

lliiIll

\\

~~~~\Solid - :t++- eas

\ :~

O 1000 1500 2000 2500500

Temperature, oc

Fig. 4. Changein the thermal condition for 0.50/, increase ofFe content in the ore,

483 @1992 ISIJ

ISIJ International, Vol. 32 (1992), No. 4

Table l. Predict~on results of BFA for mcreasedblast temperature

Blast temperature ('C)

O/C ratio

O/F ratio

Total heat input (GJ/THM)Cokeconsumption (kg/THM)Fuel consumption (kg/THM)Hot metal production (t/day)

A(coke), (kg/THM)Variation of coke consumption ("/,)

A(fuel), (kg/THM)Variation of fuel consumption (~/.)

Variation of hot metal production ("/.)

Blast volume (Nm3/THM)Gasvolume (Nm3/THM)Temperature of top gas at the

stock line ('C)

Position of the ending point ofmelting zone (m)

The volume of the dropping zone (m3)

l 0333.4423.4264.135

460, l462,3

2199

1109

l 617206

17,86

75,4

l 0833.483

3.4674. I07

454.8456.8

2262-5.3

- 1.2

-5.5

- I.2

+2.9

l 059

l 583

l82

l7.32

99.7

1133

3.519

3.504

4.090450. l452. l

2320

- i0.0

- 2.2

- 10.2

-2 2+ 5.5

l 033

l 553

l67

l7.24

103.2

l 183

3,551

3.535

4.085446. l448. i

2371

- 14.0

-3 l- 14,2

- 3, l+ 7,8

l 029

l 527

l59

17.l

l09.O

o

5Eq,

'- lO

o+'vl

~:15

o~e~e'v~ ZO

,:;

25

~\~\\~,\

\\~~

\\1Ll1l

~~

Solid - - Gas~\:

~

~:~O 500 100a 1500 2000 2500

Temperature, oc

Frg. 5. Predicted temperature distributions for the case of

100*C increased blast temperature.

prediction results are given in Table 1. The profile ofthe gas and condensedphase temperatures is illustrat-

ed in Fig. 5, and the profiles in the base case, i,e. the

simulation test, are also presented in the figure. Thepredicted effects of this operating action are shown in

Fig. 6.

The most interesting information obtained from the

prediction whenincreasing the blast temperature are:

- Both fuel consumption and coke consumption de-

creases with increasing blast temperature. This is

due to the substitution of higher blast enthalpy for

fuel combustion heat and decreased top gas enthalpy.

- Furnace productivity increases as a constant blast

volume is used and fuel consumption decreases.

- Direct reduction degree slightly increases due to in-

sufficient heating of iron ores in the lumpy zone.

- Total heat input to the furnace decreases slightly

which indicates that the heat carried into the furnace

by hot blast maybe employedmore efficiently thanthat supplied by the fuel.

Silicon transfer phenomenaare not taken into con-

(L,

~-

~e,

oIsG,

e,La,

c:,

5

4

3

2

l

lO

~5~~

8 *>,4J

>~,v6 ::~SO~CLaJ

urQ4 ,:!L::~

Lh1:,e,,1,

2 fT~

(V!L

u~:

o oo 50 200loo 150

Increased blast temperature oc

Fig. 6. Predicted effects ofincreased blast temperature.

sideration in the KTHmodel at present, but becausethe flame temperature and volumeof the dropping zoneincrease it is suggested that ['/oSi] in the hot metal mayincrease

.

The decrease in coke consumption predicted by the

KTHmodel is 2.2 o/o/lOO'C whenthe blast temperatureis increased from I 033 to I 133'C. This result wascomparedto data reported in the literature. Accordingto Schumacher3)a coke saving of about 4"/./100'C at

900'C, 3~/o/lOO'C at IOOO'Cand 2'/*/lOO'C at I 100'C

can be obtained for a furnace of high coke ratio oper-ation (635kg/THM, at a blast temperature of 850'C).

According to Wegman4)a coke saving of 2.3 ~/*/lOO'C

from I OOOto I 100'C can be achieved for a modernblast furnace. As to furnace productivity, an increase

of 5.50/0/lOO'C from I 033 to I 133'C was predicted

by the model. This figure was also comparedto oper-ational data. According to Schumacher3)an increase

of about 270/, by increasing the blast temperaturefrom 850 to 1250'C, i,e. 6.8~/o/lOO'C, can be obtain-

ed. It is interesting to see that the results predicted bythe KTHmodel agree reasonably well with these oper-

C 1992 ISIJ 484

ISIJ lnternational, Voi. 32 (1992), No. 4

Table 2. Prediction results of BFA for oxygen enrichment of blast.

Oxygenenrichment ("/*)

O/C ratio

O/F ratio

Total heat input (GJ/THM)Cokeconsumption (kg/THM)Fuel consumption (kg/THM)Hot metal production (t/day)

A(coke), (kg/THM)Variation of coke consumption ("/*)

A(fuel), (kg/THM)Variation of fuel consumption (~/*)

Variation of hot metal production ("/.)


stock line ('C)



0.00

3,442

3,4264, 135

460. l462.3

2199

l 109

l 617206

17.86

75.4

0,50

3,4353,4204. I15

461.O

463. 1

224l+0,9

+0,19+0,8+0,18

+ 1,9

1088

1597200

17.75

80.2

l .OO

3,425

3,4104, 102

462.4464.5

2279

+2,3

+O.49

+2,2

+0,48

+3,6

l 070

l 579

l97

I7.76

79.9

l .50

3.415

3.4004. 09l

463.8465.8

2316+ 3.7

+0.80

+3.5

+0.77

+5.3

l 053

l 562195

17.77

79.2

o

5EaJ

'- 10

vo+'

~ 15o~L~

aJ

uF:'e 20~'

c~

~.:\

~~~~

\\

\\1Lll

~b,

Solid -:~~_ Gas

~.\ ~

25 r~~r~~1=~O 500 1000 1500 2000 Z500Temperature oC

Fig. 7. Predicted temperature distributions for the case of

l "/~ oxygen enrichment of the blast.

cu~,

~L

aJ

ovaJ

cl,

!L

l.O

0.8

a.6

0.4

0.2

0.0

lO

8

6

4

2

OO 0.5 l.O 1.5 2.0

Oxyq.en enrichment of b~ast, ~Fig. 8. Predicted effects of oxygen enrichment of the blast.

i~~

::~

>~,

1:;

o~,1,

vfl:~

$:SL

Lhl:,cU

r~(1,

~u

ational data.

6. I.2. OxygenEnrichment of the BlastThree levels of oxygen enrichment of the blast were

chosen for prediction by the KTH-model. The results

are given in Fig. 7, Table 2and Fig. 8.

The following information obtained from the predic-tions is interesting:

- Both fuel consumptionandcokeconsumption increaseslightly. This is causedby the decreasedCOutilization,

n'o'

Furnace productivity increases due to the Increased

massflowrate of oxygen.

- The flame temperature and dropping zone volumeincrease comparedto the base case.It is found that the volume of the dropping zone

decreases with increasing of oxygen enrichment. Theaction of oxygen enrichment has two effects, i.e. in-

creasing the flame temperature and furnace productiv-ity. The former effect will lead to increased dropping

zone volume, which is shown by the results of the

computer tests of analysing the thermal condition for the

case of blast temperature fluctuation. According to acomputer analysis by the modelof the thermal conditionin the case of an increased furnace productivity with theblast volume being increased and the heat input beingconstant, the latter effect will lead to a decreases in the

volume of the dropping zone. It could be consideredthat the latter effect becomesstronger whenthe blast is

more oxygen enriched. As a result, the volume of the

dropping zone decreases when the amount of oxygenenrichment is higher.

It should be pointed out that the increased fuel

consumption, which is predicted in this study, is onlyvalid for the case whenthe raceway gas volume is keptconstant with Increasing oxygen content in the blast.

The reason is that the furnace productivity will increase,

and, consequently, that the COutilization will decreasein this case. If the furnace productivity is kept constantwhenthe blast is oxygen-enriched the fuel consumptionwill decrease due to increased COutilization.

6.1.3. Coal PowderInjectionSeveral levels of coal powderinjection werechosen for

485 C 1992 ISIJ

ISIJ International, Vol. 32 (1992). No, 4

Table 3. Prediction results of BFA for coal powder injection.

Massfiowrate of coal powder (kg/hr)

O/C ratio

O/F ratio

Total heat input (GJ/THM)Cokeconsumption (kg/THM)Coal consumption (kg/THM)Fuel consumption (kg/THM)Hot metal production (t/day)

A(coke), (kg/THM)Variation of coke consumption ("/.)

A(fuel), (kg/THM)Variation of fuel consumption ('/.)

Replacementratio of coal to cokeVariation of hot metal production ("/.)


stock line ('C)



0.00

3.4423,4264,135

460. l0.0

462.3

2199

l 109

l 617206

17.86

75.4

4OOO3.784

3.4104. i23

418.643

. 8464.5

2193

-41.5

- 9.03

+2.2

+0.5

0.949

- 0.3

1I12

1643226

17.89

67.9

8OOO4. 2013,3904. 127

377.088.2

467,3

2178

- 83, l- 18,07

+ 5.0

+1.l0.943

- I.O

1120

1674253

18,14

62.6

l2OOO4*7323.3704. 130

334.8

l33. l470, l

2163

- 125,3

- 27,24

+ 7.8

+ I.7

O.94l

- 1.7

l 128

l 705280

18,15

58.8

E~

e,~:

L,

o~E:

o~LhCIJ

u~:

+,

o

5

lO

15

20

25

+\\\\

\~!

\l11IllI

~~~~

\Solid - \\( Gas

\ ~:~

O 1000 1500 2000500Temperature, oc

Fig. 9. Predicted temperature distributions for the

133. Ikg/THMcoal powder injection.

2500

case of

o

5~aJ::~: lO

o+'

Eo 15~~h(v

vfl:i

~ 20

,:'

25

\\\\

\\1tIl1

~~

\Solid - - Gas\,

\ ~~

prediction. The volatile matter content in the coal is

26 "/o. The prediction results are given in Fig. 9 andTable 3.

The following information obtained from the predic-

tion is valuable:

- Cokeconsumption decreases.

- Fuel consumption slightly increases because n*.

decreases and top gas enthalpy increases. The re-

placement ratio of coal, i,e. kg(coke)/kg(coal), in this

example is between0.9 and I .O.

_[o/oSi] in the hot metal may drop because boththe flame temperature and dropping zone volumedecreases.Practical operation results of coal injection depend

not only on the coal properties but also on the efficiency

of furnace operation. This fact should be kept in mindwhenany comparison betweenmodel prediction results

and empirical results is made. Trials of coal injection

were conducted on BF4 of Hamborn, Thyssen Co.,

O lOaO 1500 2000 2500500Temperature, oC

Fig, lO. Predicted temperature distributions for the case of

l07.8kg/THM coal powder injection and oxygenenrichment to keep constant flame temperature,

West Germany1986.5) The coke ash content, naturalair blast and proportion of sinter and pellets in the

burden in the trials are similar to the conditions for the

prediction in this study. When136kg coal powder perton of hot metal was injected a coke consumption of

338kg/THMwas achieved in the trlals. According to

model test by the KTHmodel, coke saving is influenced

by the replacement ratio of coal to coke. Thehigher the

ratio, the bigger the coke saving. The replacement ratio

in the prediction is 0.94 comparedto 0.84 in the trials.

This could be the reason why the coke consumption in

the prediction is roughly the sameas the empirlcal onealthough 100'C Iower blast temperature was used for

the prediction.

6.1.4. Coal Injection Combinedwith OxygenEnrich-

mentPredictions for three levels of coal injection rates 54. I,

C 1992 ISIJ 486

IS]J Internationai, Vol. 32 (1 992), No. 4

Table 4. Prediction results of BFA for coal powder injection and oxygen enrichment.

Massflowrate of coal powder (kg/hr) 5OOO 12 OOOO.OO

Oxygenenrichment ("/.) 0.250 0.5630.000

O/C ratio 3.877 4.4073.442

O/F ratio 3.408 3.3753.426

Total heat input (GJ/THM) 4. 100 4.0904, 135

Cokeconsumption (kg/THM) 408.6 359.5460. lCoal consumption (kg/THM) 54. I I07,80,0

Fuel consumption (kg/THM) 464.8 469.4462.3

Hot metal production (t/day) 2218 22252199A(coke), (kg/THM) - 51.5 - IO0.6

Variation of coke consumption (o/.) - I1.3 -21 .9

A(fuel), (kg/THM) + 2.5 + 7. lVariation of fuel consumption ("/o) +0.5 + I.5

Replacementratio of coal to coke 0.958 0.934Variation of hot metal production (o/.) +0.8 + 1.2

Blast volume (Nm3/THM) I 100 1096l 109

Gasvolume (Nm3/THM) 1635 1661l 617Temperature of top gas at the 224 250206

stock line ('C)

Position of the ending point of 17.94 18.0917,86

melting zone (m)The volume of the dropping zone (m3) 75.4 71.9 64.9

20OOO

1.463

6.2453.3953.998

253.721 1.0

466.7

2275

- 206.4

- 44.9

+4.4+0.95

0.978

+3.5l 072

l 677303

18.76

32,2

Table 5. Prediction results

quality.

of BF A for changed coke oi~

,~\

Starting temperature of 933.9solution loss ('C)

O/C ratio 3.431

O/F ratio 3.415

Total heat input (GJ/THM) 4, 136

Cokeconsumption (kg/THM) 461 .7

Fuel consumption (kg/THM) 463.8

Hot metal production (t/day) 2199.0

A(coke), (kg/THM) + I.6

Variation of coke +0.3

consumption ('/*)

A(fuel), (kg/THM) + 15Variation of fuel +0.3

consumption ('/.)

COutilization (-) 0.4854Blast volume (Nm3/THM) I 109.08

Gasvolume (Nm3/THM) 1619.80

Temperature of top gas at 200.l

the stock line ('C)

Thickness of lumpy zone (m) 12.93

Thickness of softening 3.02

zone (m)Thickness of melting zone (m) I.60

Position of the ending point 17.55

of melting zone (m)The volume of the dropping 72.4

zone (m3)

l OOO

3.442

3.4264. 135

460. 1462.3

2199.4

0.4908

l 108.86

l 617.0206.0

l3.05

2.83

l .59

17.48

75.4

l 066, l

3.4543.4384. 132

458.6460.7

2201 .2

-1.5-0.3

- 1.6

+0.3

0.4960

l 107.94

1613.5121 1.0

l3.93

2.32

l .30

17.55

72.5

l07.8 and 21 1kg/THM,which are correspond to 5OOO,

12000 and 20OOOkg/hr respectively, with appropriate

oxygencontents to keep the flame temperatures constant

were conducted. The results are given in Table 4 andFig. 10. The following information obtained from the

predictions is notable.

- Cokeconsumption decreases.

Fuel consumption increases slightly.

- Furnace productivity increases slightly.

- The raceway gas volume increases.

Theextent of all the changes for this combinedaction

case is less than that for pure coal powder injection.

~aJ

o+-

Eo~t~(vu,e~;

,3

Fig.

5

10

15

20

25

~L

~\hIL

1lllll

Solid -

~~~t' r]as

~*

O 500 1000 15DO 2000 2500Temperature, oc

11. Predicted temperature distributions for the case ofincreased coke reactivity (66.1"C decreased starting

temperature of the solution loss reaction),

6.1 .5. ChangedCokeQualityThe starting temperature of the solution loss reaction

in the blast furnace is influenced by the coke quality.

Hatano et al.6) presented an expression that links thestarting temperature, the coke quality and the accumu-lated alkalis in the furnace. The starting temperatureincreases with increasing coke size and decreasing cokereactivity and alkali amountin the furnace. Twolevels

ofchangedstarting temperature, i.e. +66. I and - 66. l'Cwhich are comparable to - 100/0 and + 100/0 coke re-activity according to Hatano's expression by neglecting

the influence of coke size and accumulated alkali, wereselected for prediction. The test results are shown in

Table 5, Figs. Il and 12.

The following information obtained from the predic-

tion for higher coke quality is valuable:

- Both coke consumption and fuel consumption de-

creases slightly due to an increased COutilization.

487 C 1992 IS[J

o

ISIJ International. Vol. 32 (1 992),

~Q,

O~,

EO~~aJ

a:+,

C:~

5

10

15

20

25

No, 4

~~

:~

iL

\L

~~~

Solid - ~ - Gas

~

O 500 2000 2500lOOO 1500Temperature, oc

Frg. 12. Predicted temperature distributions for the case of

decreased coke reactivity (66.1"C increased starting

temperature of the solution loss reaction),

- Furnace productivity increases slightly.

It is noticed that the volume of the dropping zonedecreases whencoke quality changes. This observation

maybe explained as follows. Whencoke quality is highthe volume of the lurnpy zone becomesbigger whichresults in decreased dropping zone volume, in spite ofsmaller volume of the softening and melting zones (see

Table 5). In contrast, if coke quality is low, the volumeof the softening and melting zones becomebigger, whichleads to decreased dropping zone volume, although the

volume of the lumpy zone is smaller.

7. Concluding Remarks

A one-dimensional static model for simulation andprediction of the blast furnace process has beendeveloped at the Department of Process Metallurgy,

The Royal Institute of Technology (KTH), Sweden.The validity of the KTHmodel is supported by the

comparison of the predicted resultes for someopera-tional actions with industrial test data. This model canbe used for:

- obtaining a deeper understanding of the metallurgical

aspedts of the blast furnace process (research andpersonnel training), and

- prediction of the internal state in the blast furnace,

fuel consumption and furnace productivity whenblast

parameters and injection rates are changed(operationpolicy, managementand production planning). It is

possible to develop the model further so that predic-

tion can be madefor the change in other operating

parameters.These two functions are important for a successful

control of the blast furnace process.

Acknowledgements

This work is sponsored by the Swedish lronmasters'

Association (Jernkontoret). Thanksare sent to Professor

Sven Eketorp for his critical remarks during the

manuscript preparation. Techn Lic. Tor Borinder and

C 1992 ISIJ 488

Mr. Rutger Gyllenram are gratefully acknowledged fortheir help and fruitful discussions throughout the work.Thanks are also due to Mr. Anwar Jamil for his

assistance with someprogrammingwork, and to Mr.Carl-Lennart Axelsson for valuable discussions onprogramming.

NomenclatureC0~: COcontent of top gas (volume fraction)

CO~,g: C02content of top gas (volume fraction)

C0~,9: calculated C02 content of gas as the tem-perature of the solids is equal to the starting

temperature of the solution loss reaction (vol-

umefraction)f*o : degree of iron ore reduction at the stock line

(-)hp : coefficient for heat transfer between the gas

and condensedphases (kJ/m2 . h• 'C)h+. : coefficient for heat losses through the furnace

wall (kJ/m2 .h • 'C)H~,g : H2content of top gas (volume fraction)

R~: overall reaction rate of ore reduction by CO(kmol(CO)/m3(bed) • h)

Tgo : gas temperature at the stock line (K)Tgl : Iower boundary condition at the tuyere level

of gas temperature, i,e. the flame temperature(K)

Tg.4 : calculated gas temperature at the tuyere level

(K)Tsl : Iower boundary condition of the condensed

phase temperature at the tuyere level (K)T*,4 : calculated temperature of condensedphases at

the tuyere level (K)n.. : COutilization of gas (-)nH, : H2 utilization of gas (-)y*. : degree of FeOreduction by CO(-)yd : direct reduction degree of FeOby carbon (-)

yH2: degree of FeOreduction by H2 (-)~1i: correction factor for h* (-), i= 1, 2, 3 is for

the lumpy zone, the softening/melting zoneand the dropping zone, respectively.

~2i: correction factor for R~ (-), i= l, 2, 3 is for


~3i: correction factor for hp (-), i= l, 2, 3 is for


p~ : density of the condensedphases at the stockline (kg/m3)

REFERENCESl) X. Bi. K. Torssell and O. Wijk: ISIJ Int., 32 (1992), 470.2) R. D, Pehlke: Unit Processes of Extractive Metallurgy, Second

Printing, Elsevier Publishing Co., Inc., NewYork, (1975), 314.

3) H, Schumachef:Stah/ Eisen, 86 (1966), 313.

4) Eu. F. Wegmann:A Reference Book for Blast Furnace Op-erators, Trans. from Russian into English by V. Afanasyev,Mir Publishers, Moscow,(1984), 202.

5) M. Giuli. G. Hanniker, J. Koster. K. Kreibich, J. M. vanLangen, Y, de Lassat de Pressigny and A. Pcos: Proc. Europeanlron-making Cong, 1986 in Aachen, WestGermany,IV/2, 17.

6) M. Hatano, T. Miyazaki and Y. Iwanaga: SumitomoSea,'ch,

(1980), No. 23, l.

prediction of the blast furnace process by mathematical model

Documents