Coinjection of Natural Gas and Pulverized Coal in the...

1

Coinjection of Natural Gas andPulverized Coal in the Blast Furnace at

High Levels: Field Test Results atUSS/Gary

J.C. Agarwal, F.C. Brown, D.L. Chin, G.S. StevensCharles River Associates Incorporated

200 Clarendon StreetBoston, Massachusetts 02116

Phone : 617-425-3000Fax : 617-425-3132

and

David M. Smith(former) Gas Research Institute

(current) Gas Technology Institute (GTI)1700 South Mont Prospect Road

Des Plaines, IllinoisPhone : 847-768-0971Fax : 947-768-0501

Key Words:

Coinjection of coal and natural gas, productivityincrease in blast furnace

INTRODUCTION

Supplemental fuels are injected at the tuyerelevel of blast furnaces to reduce coke consumptionand increase productivity. These fuels includenatural gas, coke oven gas, oil, tar, and coal. Theeconomic benefits derived from supplemental fuelinjection are of two types: 1) the reduction in costsof hot metal production arising primarily fromdecreased coke consumption, and 2) the value of theincreased production of hot metal — and steel —that can be sold. All blast furnaces in NorthAmerica inject supplemental fuel. Approximatelyone half of the hot metal produced in NorthAmerica is from blast furnaces that inject coal atlevels between 150 lb and 350 lb/ton of hot metal.Most of the rest of the hot metal is produced onblast furnaces injecting up to 310 lb of natural gasper ton hot metal.1 The balance is produced by blastfurnaces injecting oil or coke oven gas.

Ten years ago, about half of the furnaces inNorth America were injecting natural gas, but theaverage injection level was below 50 lb/THM. Theprevailing belief at the time was that it would benecessary to maintain a constant value of theraceway adiabatic flame temperature (RAFT) athigh-injection levels, so that very high blastenrichments would be required.2 Also, the extent ofproductivity improvement that could be obtained athigher gas injection levels had not beendemonstrated. To address these technical andeconomic concerns, the Gas Research Institute hassponsored a number of tests with injection levels of

1 Results of Ultra High Rates of Natural Gas into BlastFurnace at ACME Steel Co., Iron Making ConferenceProceedings, Iron & Steel Society, 1997.2 The Use of Natural Gas in the Blast Furnace Area – A WhitePaper, Gas Research Institute, February, 1988.

2

gas as high as 310 lb/THM in operating blastfurnaces.3 4 5 6

The productivity at the natural gas injectionlevel of 310 lb/THM was increased by 40% (to8.8 TPD/CCF) while the coke rate was reduced to641 lb/THM, and valuable insights were gained asto the requirements for practice changes appropriatefor differing burden permeabilities and for thedesigns of lances and tuyeres over a wide range ofinjection levels. Specific oxygen consumption inthese tests was typically in the range of 0.9 – 1.1lb/THM gas.

Increasing productivity was a goal in all ofthese tests, and was achieved by increasing blastenrichment as the level of natural gas injection wasincreased. In a subsequent experiment carried out atWCI Steel, natural gas injection levels wereincreased from 120 to 250 lb/THM at constantproductivity.7 These tests showed that whenproductivity increases are not required natural gascan be injected at high levels without acorresponding increase in blast enrichment: specificoxygen consumptions below 0.7 lb/lb natural gaswere obtained with stable furnace operation and acoke replacement ratio of about 1.1 at the highestinjection levels.

Of the 16 blast furnaces currently injectingcoal, only two consistently inject over 50 lb ofnatural gas/THM but the quantity of coal injected islimited because of limitations in the coalpreparation facilities. Very little data have been

3 Injection of Natural Gas in the Blast Furnace of High Rates:Field Experiments at Amico Steel Company, Gas ResearchInstitute (GRI-93/0353), 1993.4 Injection of Natural Gas with Blast Furnace at High Rates:Field Test Results at National Steel – Granite City, GasResearch Institute (GRI-95/0359), October 1995.5 Injection of Natural Gas in the Blast Furnace at High Rates:Field Test Results at Acme Steel Company, Gas ResearchInstitute (GRI-95/0358), October 1995.6 Injection of Natural Gas in the Blast Furnace at Very HighLevels: Field Test Results at Acme Steel Company, GasResearch Institute (GRI-97/0211), July 1997 and Iron MakingConference Proceedings 1998.7 Injection of Natural Gas at High Levels in the Blast Furnacewith Low Oxygen Consumption: Field Test Results at WCISteel, Gas Research Institute (GRI-98/0317), November 1998.

published on this practice. Thus, there is a need todevelop and document high-level coinjectionpractice.

OPERATING HISTORY AND TESTOBJECTIVES AT USS/GARY

This paper presents the results of a series offield tests sponsored by the Gas Research Instituteand managed by Charles River Associates on acommercial blast furnace, the No. 4 furnace atUSS/Gary, at gas coinjection levels up to125 lb/THM, or 2,800 scf/THM. The objective ofthe test program was to develop the processtechnology for high levels of coinjection to increaseproductivity and reduce coke consumption. A testplan for the experimental work was developed todefine the aim values and expected furnaceperformance at a baseline coal injection conditionand for natural gas coinjection levels ranging from25 lb/THM to 150 lb/THM. Throughout the testsUSS/Gary operated the furnace primarily to meetcommercial requirements for hot metal andsecondarily as a vehicle to obtain test data. A fulldescription of the planning of the tests is presentedelsewhere.8

United States Steel operates four blastfurnaces at its Gary works. The largest (No. 13) isbaseloaded and operates at constant conditions tothe extent possible. The three South Side furnaces(Nos. 4, 6, and 8) are used to manage swings in hotmetal demand.

All furnaces are served by a common coalpreparation system, and all are equipped withblowpipes that would accommodate coinjection ofoil. No. 13 is also equipped with a delivery systemfor natural gas to atomize coinjected oil. A numberof operating issues in the coal preparation systemand in the furnaces, however, limit the types of coalthat can be used and the maximum delivery rates tothe furnaces, however. The blast furnace area hasaccess to both high- and low-purity oxygen, but

8 Coinjection Of Natural Gas and Pulverized Coal in the BlastFurnace at High Levels: Field Test Results at USS/Gary, GasResearch Institute, GRI 98/0318, November 1998.

3

piping constraints limit the operator’s flexibility toobtain unlimited supplies of one type or the other.

On-site coke production is insufficient to meetthe total demand at Gary, so external coke isacquired from a variety of sources. Coke for No. 13is screened, but there is no blending facility and thecoke supply to the other furnaces can change on aday-to-day basis. The burden on No. 13 is carefullycontrolled, with the other furnaces taking up swingsin the supply of plant reverts, blended scrap, etc. Asa result, the operators of the South Side furnaces areforced to react continuously to changes inproduction requirements, burden quality, andinjectant rate and composition while trying tomaintain hot metal chemistry within acceptableranges.

Number 4 furnace operated at the baselineconditions of approximately 340 lb/THM of coaland less than 10 lb/THM of natural gas injection inSeptember and October 1997, and ramp-up tohigher gas injection levels was initiated inNovember after necessary upgrades had beencompleted. The coinjection levels had reached60 lb/THM gas and 325 lb/THM coal in mid-December when a hot furnace condition forced theoperators to reduce gas injection levels to restore theproper thermal balance. Levels of gas and coalinjection and the scrap charge on the burden variedover the rest of the winter and spring in reaction tochanges in production requirements andmaintenance problems, with aim values set inresponse to the operators’ perceptions of therequirements of the new practice. In early May1998, productivity improvement became the keyoperating objective because of the reline of No. 6;this objective remained in place through the end ofthe data acquisition period in August. The naturalgas and coal injection levels were held at about 100and 180 lb/THM respectively over this time frame,except for brief periods when the coal supply waslost. A maximum gas coinjection level of naturalgas of about 125 lb/THM was reached when thecoal injection level had been reduced to about175 lb/THM.

TEST SITE DESCRIPTION ANDOPERATING PRACTICES

This section describes the equipment in theblast furnace area, USS’s standard operatingprocedures, and the data collection procedures usedin the tests.

Furnace Description

USS/Gary No. 4 furnace is located in theGary, Indiana, plant and has a two bell top type witha two skip loading system. There is a single taphole with outlets for two torpedo cars via a tiltingspout. Hot metal is weighed at the furnace anddelivered by rail car to the BOF on site. Thefurnace details are listed below.

Furnace details

Furnace Name: No. 4

Hearth Diameter: 28 ft 10 in

Working Height: 82 ft

Number of Tuyeres: 20

Normal Top Pressure: 6.5 PSIG

Top Type: Two Bell System

Pressurizing Gas: Nitrogen

Top Gas Analyzer: ABB Extrel MassSpectrometer

Working Volume: 52,818 ft3

Tap Hole Single

Trough Design: Tilting Spout

Bosh Cooling: Channel

Type of BurdenDistribution:

Moveable Armor

Date of Last Blow-in: September 1996

Tuyere design - Eighteen of the tuyeres are6.5 inches in diameter, and #1 and #20 tuyeresadjacent to the tap-hole are 6 inches in diameter andported for addition of natural gas, since no coal isinjected over the tap-hole.

Natural gas delivery and injection systems -Natural gas is delivered to the US Steel Gary plantat 200 PSIG in a 12-inch high-pressure supply line.The pressure is reduced at a metering station for

4

delivery to the blast furnace at 130 PSIG in a 6-inchfeed line. A 6-inch Vortex Shedding meter is usedas the primary measuring element for the natural gasflow control system. The 6-inch feed line isconnected to an 8-inch circle pipe that is fitted withtwenty 1 ¼-inch drop-down pipes that are fittedwith quarter-turn shut-off valves. Ten-foot runs offlexible hose connect the shut-off valves to checkvalves on stainless steel lances.

The lances are 54 inches long (except for #2and #19, which are 43 inches long) and enter theside of each blowpipe in a horizontal positionopposite to the point of insertion of the coal lances.The angle of entry is 27 degrees from the tuyerecenter line. The tip of the lance is normallypositioned 2 inches back from the tuyere/blowpipejoint where the blowpipe butts to the tuyere. Table Ishows the lance and flexible hose diameters usedduring the test, along with the range of flows andcircle pipe pressures.

Table I. Natural Gas Injection SystemConfigurations

LanceDiameter (in.)

Flexible HoseDiameter (in.)

NG FlowRate Range(MSCFM)

Circle PipePressure Range

(PSIG)

½ 1 ¼ 1,500-6,500 32-110

¾ 1 ¼ 4,000-9,000 45-95

Source: Charles River Associates, 1998.

Since the coal lance assemblies had beenlocated on the sides of the blowpipes with the mostaccess, positioning of the natural gas injectionlances was constrained by the furnace supportcolumns. As a result, the lances could only bemoved axially within a few inches of their nominalinsertion length.

Fire detection/suppression system - Twoflame detection thermocouples were placed in thereceiving hopper above the small bell, one in thecenterline next to the cables and one next to thehopper wall. When either of these thermocouplesdetects a temperature above 125°F, nitrogen isautomatically introduced between the bells. The N2

purge has been activated about once per monthduring the tests.

Top gas analysis - The top gas sample istaken after the demister before the clean gas goes tothe stoves. A 3/8-inch sample line runs about100 feet to the sample conditioning cabinet, whichdries and filters the gas down to 5 um. Then apump boosts pressure to deliver the gas another200 feet to the ABB Extrel Mass Spectrometer. Thetop gas analyzer is tested daily against fivecalibration gases and is recalibrated as needed.

Auxiliaries description - US Steel’s Garyplant has an on-site sinter plant that has the capacityto prepare about 15% of the metallic content of thecharge. An on-site coke plant provides most of thecoke used at Gary although coke from two externalsources was used during the trial. PCI operates afacility on-site to prepare and deliver coal to all fourfurnaces. The rate of coal injection for each furnaceis set at its control room.

Raw coal was fed to the coal preparationsystem as it became available from the in-plantstorage area. Since storage capacity for anyparticular type of coal was limited, and since thecoals were not blended prior to preparation, the typeof coal injected to the furnace could change fairlyfrequently and abruptly.

Oxygen supply - Both high- and low-purityoxygen are available from on-site Praxair plants.The oxygen content of the low-purity oxygen isabout 89-92% and 5,000 SCFM is available for BFNo. 4, but only when No. 13 furnace is operatingand taking low-purity oxygen. About 10,000 SCFMof high-purity oxygen is available at all times.

Stockhouse - Gary’s charging equipment is atwo-skip system fed by an automated stockhouse.The hoppers are filled automatically by a burden-charging program. Coke is the only materialscreened at the blast furnace. An on-linemeasurement of coke moisture (MOLA) gauge isused to measure coke moisture and track the amountof coke charged on a dry basis. Each skip has avolume of 330 ft3 and maximum weight of 34,000lb

Wind delivery system - Automaticcontrollers are used to set the wind at a constantvolume. The snort valve is used for wind ratechanges under upset conditions and for off blast

5

conditions. The blast moisture system has afeedback controller that manipulates steam additionto achieve the target gr/SCF.

Data acquisition system - An extensive real-time data system collects and stores all of the datarelated to the blast furnace operation. An Excelspreadsheet was used to average the data by day andsave the data into files by month.

Charging practice - The charging sequenceused throughout the tests was:

CC/CC/PPP/PPS\C/CC/CC/PPP/PPPS\C/

C = Coke, P = Pellets, S = Sinter,

/ = Large Bell Dump

Contaminated sinter, buckwheat coke, blendedscrap, and Auburn ore were added to the pellet skipsand trim was added to the sinter skip loads.

The coke-to-ore ratio was changed bychanging the amount of iron-bearing material per

charge; the amount of coke charged per skip washeld constant.

Reported burden analyses are shown inTable II, while Tables III and IV show typicalnatural gas and reported coal assays. The elementalassays of the burden constituents and coal arereported on a dry basis. When the compositions donot sum to 100%, the assays have been adjusted upor down on a pro ratio basis to bring the sum to100% for use in furnace material balancecalculations. The coal assays were performedbefore the onset of these tests and were notrepeated. The prepared coal delivered to the furnacecontains 1% or less moisture. Pellet and sintercompositions did not change materially over thecourse of these tests, but blended scrap andcontaminated sinter compositions were variable.The scrap originated from a variety of sources withgreatly different iron contents and degrees ofmetallization.

Table II. Reported Burden Analyses

Percent by WeightMaterial H2O C S P Mn SiO2 Al2O3 Fe MgO CaO Others(1) Total(2)

White Tag Coke 3.5 92.18 0.53 0.08 0.03 4.20 2.32 0.52 0.08 0.15 1.56 101.64

Wheeling Pitt Coke N.R. 91.66 0.6 0.01 0 4.23 2.37 0.55 0.07 0.28 1.81 102.47

Acme Coke N.R. 92.24 0.76 0.24 0.01 4.52 2.41 0.62 0.24 0.09 N.R. 102.0

Minntac Flux Pellet 3.5 0 0.11 0.01 0.12 4.26 0.17 63.04 1.10 3.52 0.04 98.15

Sinter 0.5 0.06 0.04 0.13 1.58 7.65 1.97 45.47 3.85 18.38 0.36 98.46

Auburn Ore 11.2 0.18 0.01 0.05 1.24 6.52 1.35 60.50 0.36 0.15 0.04 95.63

Contaminated Sinter N.R. 0.03 0.64 0.11 1.37 5.65 1.47 49.32 3.59 15.88 0.18 99.91

Trim 7.0 0 0.06 0.03 0.56 48.6 0.24 25.8 2.47 2.29 0.08 89.67

Blended Scrap 0.5 0 0.18 0.10 1.31 6.53 1.63 74.74 2.88 12.60 0.02 105.34

(1) Others include Ti, Zn, K2O, Na2O, H2 and N2.(2) Total includes oxygen not reported, dry basis.Source: USS/Gary

Table III. Typical Natural Gas AnalysisPercent by Volume

Constituent CO2 N2 CH4 C2H6 C3H8 C4H10 C5H12 H2

0.96 1.58 93.73 2.91 0.51 0.18 0.12 0

HHV = 1,020 Btu/cf at 14.65 PSIG, 60°F, and dry.Source: USS/Gary

6

Table IV. Reported Coal AnalysesPercent by Weight

Coal C H2 O2 N2 S Ash(1) Total(2) H2O HHV(3)

Corbin 79.16 3.27 5.98 1.61 0.98 7.71 98.71 7.0 14,140Premier Elkhorn 74.06 5.83 10.17 1.36 0.78 7.66 99.86 6.9 13,720Buckeye 79.04 5.40 6.43 1.59 0.90 6.40 99.76 5.5 14,350Pinnacle 86.60 4.22 1.75 1.22 0.83 5.10 99.72 6.5 14,950(1) Ash includes SiO2, Al2OS, CaO, MgO, P, Mn, Ti, K2O, Na2O, Cl2, Fe and their oxides.(2) Total on a dry basis.(3) BTU/lbSource: USS/Gary

7

Natural gas, pulverized coal, and oxygenflow rate control - Natural gas and oxygen flowrates are controlled by constant set-point feedbackcontrollers. The furnace injection rate for coal isspecified in the control room, with delivery to thefurnace surge hoppers controlled at the preparationfacility. If the blast pressure falls below 18 PSIG,the flows are automatically shut off to prevent ahighly enriched blast if the blowers shut down.

Test Plan Outline — Aim Values and DataAcquisition

Prior to these trials, No. 4 furnace hadoperated with gas injection through the nose of thetuyeres adjacent to the tap hole (at an overallaverage injection level of about 400–500 SCFM,5-10 lb/THM) and coal injection through lances onall other tuyeres. The test plan for these trials wasdesigned to develop and evaluate practices fornatural gas coinjection at levels up to 150 lb/THM.There were four major issues that needed to beresolved through execution of these trials:

• What would the required marginal oxygenconsumption be as the natural gas injectionlevel increased? Would it be necessary tomaintain a constant, high value for RAFT?

• What would the effect of increasing naturalgas injection levels on burden permeability be,especially if RAFT were allowed to drop?

• Would the extent of the decrease in thesolution loss reaction with increasing levels ofnatural gas injection be as high as with all gasinjection practice, or would it be lower as isthe case with all coal injection practice?

• Would the details of the design of the gasinjection lances (e.g., lance diameter) affectnatural gas and coal combustion, and therebyaffect furnace performance? Could the samelance design be used over the entire range ofinjection rates?

Because coinjection of large amounts ofnatural gas at high coal injection levels(> 250 lb/THM) was a new practice, a test plan wasdeveloped that increased gas injection levels inincrements of 25 lb/THM, with a data collection

plan that minimized the risk of extrapolating toconditions in which the issues described abovecould result in operability problems.

A key question was the amount of oxygen thatwould be required to combust the natural gas andburn out the coal sufficiently to prevent sootformation or char build-up that would decreasepermeability. Therefore, a sequence of aim valueswas prepared for the trials that would let theoperators initially “overshoot” the amount ofoxygen required, and then increase the natural gasinjection level up to the point where any adverseeffects could be observed.

As will be discussed later, the test plan anddata collection plan were altered, and data wereobtained at different coal injection rates and burdencompositions than were contemplated.

DATA ANALYSIS PROCEDURES ANDDATA ACQUISITION ISSUES

This section describes the data analysisprocedures used to analyze the informationcollected from the No. 4 furnace at Gary. Adescription of the blast furnace computationalprocedures used has been presented elsewhere (seereference 8).

Data sets that have met the described criteriaare considered “rationalized” and, unless otherwisenoted, are the results of a series of good days at thefurnace and are reported here.

Data Acquisition Issues

The operating data rationalization proceduresused for these tests produced a total of seven datapoints covering 35 days of operation in the two-month “base case” period with coal-only injectionon the furnace. Initial attempts to rationalize thedata according to CRA’s usual procedures were notsuccessful: no periods could be found that met allof the necessary criteria for satisfactoryrationalization. This led to a comprehensive andongoing review of the data acquisition system andprocedures used in these tests. Key results of thisreview and their impact on data rationalizationprocedures are discussed below.

8

Natural Gas Metering

With CRA’s usual procedures, no correctionfactor is applied to the values reported for furnacenatural gas consumption. The natural gas flow tothe tuyeres adjacent to the tap hole, normally at arate of about 400-600 SCFM, was added to the flowto the lances in the remaining tuyeres to obtain thetotal flow reported here.

Problems with the coal preparation systemduring the reline period resulted in a brief period ofoperation during which natural gas was the only fuelinjected at the tuyeres. The ordered injection levelwas about 150 lb/THM, well within the rangepracticed by many other furnaces whereperformance had been evaluated carefully. Reviewof the data from No. 4 at these conditions suggestedthat a scaling error had been introduced into thevortex meter transmitter, and that natural gas flowswere being underreported by about 20%. Therationale for this conclusion is discussed in detailelsewhere (see reference 8).

Coke Moisture

USS/Gary uses a Texas Instruments nucleargauge for an on-line measurement of coke moisture(MOLA), and the output of this instrument is usedin the data acquisition system to calculate the dryweight of coke in the charge. It was determinedearly in the tests that the calibrations of theinstrument tended to drift. Therefore, the cokemoisture values reported by the MOLA wereadjusted based on the moisture values reported fromthe lab assays performed on grab samples twice perweek.

Other Stockhouse Measurements

The scrap was obtained from a number ofsources, each with different total iron content anddegree of metallization. About 25% of theestimated composite scrap iron content ispurchased.

In house “scraps” are retrieved from a varietyof locations around the mill. Some are screened andblended, and the materials are transported to thescrap bins in the stockhouse. The average scrapcomposition shown in Table II is more in the nature

of an allowance for a “typical” mix than an accurateassay of the scrap actually charged at any time, as isthe estimate of 75% metallization of the ironcontent. Visual inspection of some of the materialsin the scrap bins suggests that they are far belowaverage in iron content and metallization at certaintimes.

The uncertainty in the iron content and itsmetallization is significant because misestimation ofthe assay affects the oxygen and carbon balances aswell as the iron balance.

Coal Injection

Raw coals are received, dried, and ground in acentral facility at Gary. In-plant stockpiles of rawcoal from any individual source are relatively smalland there is limited capability for coal blending, sothat the origin (and composition) of the coal fed tothe furnace could change abruptly, as frequently asevery two or three days.

The prepared coal is discharged tolockhoppers from which it is withdrawn at therequired rate and transferred to the furnace. Furnaceoperators stipulate the rate, but the actual control iseffected at the central coal preparation facility.USS/Gary can track the total amount of coaldelivered, which can be balanced against the sum ofall coals delivered to the operating furnaces as acheck on overall coal metering accuracy. While theaccuracy of the metering system for a single furnacecannot be checked in this manner, overallconsumption balance checks have been performedto the satisfaction of USS/Gary personnel. Checksare made by estimating changes to the in-plantinventory, which can be estimated to within ± 10%.

The coal compositions shown in Table IVwere obtained before these tests were initiated.Some variability is to be expected from shipment toshipment and over time but no data are available toquantify the variability during the test period.Uncertainty in the coal carbon, oxygen, andhydrogen composition or in the coal delivery rateintroduces uncertainty into the furnace materialbalances, especially in the carbon and hydrogenbalances. Where the amounts of carbon andhydrogen in the coal are understated, for example,

9

rationalization of the furnace balances will result inabnormally low CO:CO2 ratios and hydrogenutilization efficiencies. Where these parameters arefar from their expected values, and where othercorrection factors are also unusually large, theywould suggest that errors may have occurred in thecoal assay or that the coal injection rate had beenmisestimated.

These circumstances occurred, and wereidentified and evaluated in the description of theexperimental results that follows. The mostsignificant adjustment required was to the reportedcoal injection rate during periods of low-coalinjection. The logic behind the necessity to correctthe reported coal rate under these circumstances isthe same as described above for the correction to thenatural gas rate. The evaluations of furnaceperformance suggested that the coal injection ratewas being underreported by about 10% at injectionrates below about 16 tons/hour.

Top Gas Assay

For the first part of these tests, there were twoanalyzers available to measure top gas composition:an ABB Extrel and a P&E mass spectrometer. Bothinstruments were standardized with the samecalibration gases, but reported different assays. TheP&E instrument typically reported about 1.5% moreCO2 and 1.2% less CO than did the Extrel. Becauseof this difference, CRA rationalized furnaceperformance with both sets of assays as long as bothinstruments were in service. We found that thechoice of analyzer made essentially no difference inthe estimated values of key furnace operatingparameters. While the correction factors required tonormalize the data differed, the corrected top gas

assays were typically brought to within about 0.3%of each other.

Summary of Data Acquisition Issues

The effects of uncertainties in the scrap andcoal compositions and in the calibration of thenatural gas and coal injection meters are difficult toquantify. Therefore, the following discussions ofthe evaluation of rationalized data are based on theresults obtained by forcing closure of the furnaceelemental balances based on the informationobtained from the data retrieval system and onCRA’s judgment to adjust for errors or biasesbelieved to be incorporated in the raw data. Inaddition, summaries of furnace performance basedon normalized “as reported” data are presented forreference purposes.

EVALUATION OF RATIONALIZEDDATA

The operating data obtained from USS/Gary’sdata acquisition system were reviewed and checkedfor consistency. Data were rejected when thefurnace experienced transient conditions due toscheduled or unscheduled maintenancerequirements and other upsets in operatingconditions unrelated to the tests. As a result, steadystate data representing about 31 weeks of operationover the year that data were acquired were used toevaluate furnace performance. The rationalizeddata were used to estimate key furnace operatingparameters, which are summarized below. Thevalues of Table V are based on rationalizing datathat have been adjusted according to CRA’sjudgment of the extents of natural gas and coal ratemetering errors as described above. The values inTable VI are based on the data as reported.

10

Table V. Summary of Coinjection Tests at USS/Gary No. 4

Period

Process Parameters Units Baseline

HighSupplemental

Fuels RelineIntermediateGas and Coal

High Gas, LowCoal

BlastTemperatureMoistureDelivered WindSupplemented O2

AISI RAFT

°Fgr/SCF

MCF/THMlb/THM

°F

1,8958.9

30.3188

3,836

1,8846.0

31.0206

3,675

1,8977.9

29.1205

3,559

1,8946.2

28.6230

3,597

1,8926.7

29.3218

3,461Injectants

Natural GasCoalTuyere O2:C

lb/THMlb/THM

mole/mole

63391.10

493261.06

991891.39

972191.25

1251771.36

BurdenPelletsSinter, Ore, OthersScrapCokePermeability

lb/THMlb/THMlb/THMlb/THMm2/sec2

2,1057632946.7

6.16

2,2677421846086.44

2,3585092746477.84

2,3575522856167.16

2,3335342906326.78

ProductionHot MetalProductivityH.M. Temp/SDH.M. Si/SDH.M. S/SD

TPDTPD/CCF

°F%%

3,7247.05

2,677/310.74/0.17

0.049/0015

3,7577.12

2678/270.74/0.13

0.039/0.009

4,1847.92

2,665/310.58/0.16

0.049/0.015

4,0277.63

2,676/270.64/0.16

0.037/0.009

4,0457.66

2,657/320.60/0.16

0.039/0.011Operating Parameters

TCESolution LossBosh H2

H2Utilization

MMBtu/THMmole/THMmole/THM

%

0.709.3511.141.8

0.768.4117.249.2

0.727.3918.846.5

0.726.7218.851.6

0.716.1021.347.8

Table VI. Summary of Coinjection Tests at USS/Gary No. 4 Based on “As Reported” Data

Period

Process Parameters Units Baseline

HighSupplemental

Fuels RelineIntermediateGas and Coal

High Gas, LowCoal

BlastTemperatureMoistureDelivered WindSupplemented O2

AISI RAFT

°Fgr/SCF

MCF/THMlb/THM

°F

1,8958.9

30.3188

3,836

1,8846.0

30.9207

3,723

1,8977.9

28.0207

3,684

1,8946.2

28.2231

3,693

1,8926.7

27.8219

3,604Injectants

Natural GasCoalTuyere O2:C

lb/THMlb/THM

mole/mole

63391.10

433271.07

851731.51

832211.28

1061611.47

BurdenPelletsSinter, Ore, OthersScrapCokePermeability

lb/THMlb/THMlb/THMlb/THMm2/sec2

2,1057632946.7

6.16

2,2677421846156.40

2,3585092746537.28

2,3575522856246.91

2,3335342906346.19

ProductionHot MetalProductivityH.M. Temp/SDH.M. Si/SDH.M. S/SD

TPDTPD/CCF

°F%%

3,7247.05

2,677/310.74/0.17

0.049/0015

3,7467.09

2678/270.74/0.13

0.039/0.009

4,1557.87

2,665/310.58/0.16

0.049/0.015

3,9927.56

2,676/270.64/0.16

0.037/0.009

4,0437.64

2,657/320.60/0.16

0.039/0.011Operating Parameters

TCESolution LossBosh H2

H2Utilization

MMBtu/THMmole/THMmole/THM

%

0.709.3511.141.8

0.748.9415.447.0

0.747.3516.641.3

0.757.0517.247.6

0.736.0818.642.8

11

Effect of Uncertainties in Coal and ScrapCompositions

The hydrogen utilization efficiency during thebaseline period was in the expected range whenElkhorn coal was being injected, but was low whenthe average mix was two-thirds Corbin, one-thirdBuckeye. The calculated hydrogen utilizationefficiency for the hydrogen content assumed forCorbin coal suggests that the assay in Table IVunderestimates the hydrogen content of this coal.More reasonable utilization efficiencies are obtainedwith a coal hydrogen content of about 5.5%. This isalso consistent with the observations that, at 3.27%,the hydrogen content is rather low for a coal of thisvolatility and the total elemental and ashcompositions do not sum to 100%.

The sensitivity of the calculated furnaceperformance to the hydrogen content of the coalwould be lower at lower injection levels and theeffect would tend to be masked at higher natural gasinjection levels. However, knowledge of the coalchemistry is important because it affects not onlythe hydrogen utilization efficiency calculations butbosh gas hydrogen content (which is an importantcorrelating parameter), the extent of the solutionloss reaction, the calculated thermal profile in thehigh-temperature zones, and therefore the properaim values to set and the expected replacementratios for the coal and coinjected gas.

Lance Design Issues

Previous work on high rate gas injection hadshown that improper lance design could lead tooperating problems (see reference 6). There weretwo major concerns with respect to lance designs forthese tests:

• The high turndown in projected gas flow rate(7 times) could make achievingstable operations over the entire rangedifficult.

• The extent of the interaction, if any, betweenthe issuing gas and coal plumes, and thesubsequent effect on furnace operation wasunknown.

The potential turndown problem wasaddressed by using lances with different diameters:½-inch IPS Sc 40 lances were to be used at flowrates below about 4,000 SCFM, and ¾-inch IPSSc 40 lances were to be used at higher rates. Initialoperation was with the smaller lances, at gas flowrates of about 5,500 SCFM. The larger-diameterlances were in use when the estimated actual gasflow through the lances was about 10,400 SCFM.At this flow rate the calculated pressure drop fromthe circle pipe to the blast is about 52 PSIG, whileaverage measured pressure drop ranged from 54 to61 PSIG. The agreement between calculated andmeasured pressure drops is additional evidence ofthe appropriateness of the application of the 20%correction factor to the indicated natural gas flow.

The gas lances were designed to penetrate theblowpipe essentially the same distance as the coallances and to have the axes of the two lancesintersect. Thus, in the absence of the aerodynamiceffects of the blast, the gas and coal plumes wouldinteract with each other some distance into thetuyeres. The flow of the blast would tend toseparate the two plumes, however. Observation ofthe ashing patterns in the various tuyeres before theinitiation of coinjection showed clearly that a simpledescription of the coal plume flow patterns was notpossible. The coal lance positions and orientationswere obviously not uniform and ashing could occuron any quadrant of the tuyere nose. When the gaslances were installed it was obvious that theirorientations and positions were nonuniform as well.

During the initiation of coinjection testingwith gas flows through the lances below2,000 SCFM, there was very little indication of anyinteraction between the gas and coal plumes,although the operators claimed that there seemed tobe a reduction in the extent of ashing on sometuyeres. As the gas flow rate was increased,however, interaction could be observed on sometuyeres and by the time the gas flow rate exceeded3,000 SCFM several different types of interactionsbecame obvious. Minimal interaction was observedat the lower flow rates and when “misalignment” ofthe lances directed the gas plume over or under thecoal plume. In these cases one could observe the

12

gas plume appearing to gently redirect the coalplume up, down, or toward the center line,depending on the relative orientations of the flows.At higher gas flows, say 5,000 SCFM, the naturalgas plume would show a much more pronounced

tendency to redirect the coal plume, or even to“punch” right through the coal plume in some cases.Some types of interaction that were observed areshown in the sketches in Figure 1.

Figure 1. Some Types of Plume Interactions Observed During Coinjection Testing

Period 8: No Interaction (Low NGI)

Period 10: Slight Interaction

Period 10: Short PCI Lance

Period 8: Misalignment (Low NGI)

Period 24: Large Interaction (Low Coal)

Period 24: Down Draft of PCI (Low Coal)

PCI

NGI

PCINGI

PCINGI

PCI

NGI

PCINGI

PCINGI

Ashing

It was always observed that the gas plumeinfluenced the behavior of the coal plume, and notthe reverse. The reason for this is shown in the datain Table VII, which compare the momentum of theblast and the gas and coal plumes issuing from thelances at a variety of test conditions. During theramp-up period the momentum of the blast

exceeded both the coal and gas plume, which wouldretard mixing of the plumes. At a gas flow of about2,750 SCFM in the high-coal/low-gas period,however, the gas momentum exceeded that of theblast and therefore could penetrate the blast axiallymore effectively and begin to influence the coalplume. The estimated average velocities of the

13

blast, gas, and coal for this period are about 435,400, and 55 ft/sec while their average densities are

about 0.052, 0.14, and 2.6 lb/ft3, respectively.

Table VII Estimated Momentums of the Blast and Gas and Coal Plumes for Various Test Conditions

Coal andGas(1) Flows,

Lb/THM

BlastMomentum

ρρρρV2/2gc,PSI

GasMomentum(2)

ρρρρV2/2gc,PSI

CoalMomentum(3)

ρρρρV2/2gc,PSI Comments

307, 28 1.3 0.8 0.7 Ramp-up

333, 47 1.1 2.3 0.8 High coal, low gas

166, 88 1.3 8.1 0.3 Low coal, intermediate gas

218, 83 1.3 7.5 0.5 ½" lance

225, 74 1.3 2.4 0.5 ¾" lance

175, 116 1.3 4.7 0.3 Low coal, high gas

—, 164 1.4 9.0 — No coal, highest gas

(1) Gas flow through lances only.(2) Estimated at the lance tip prior to expansion.(3) Assumes homogeneous flow of coal and carrier gas.

Clearly, the gas plume had sufficientmomentum to influence the coal plume at gasinjection levels above 50 lb/THM with the ½"lances, or 75 lb/THM with the ¾" lances if thepositioning of the lances provided trajectories thatwould favor it. However, it is possible that thepresence of the gas in the tuyere/raceway couldinfluence the behavior of the coal even if the plumesdid not physically interact. For example, rapidcombustion of the gas could lead to preferentialconsumption of oxygen that would, in turn, retardcoal combustion and lead to incomplete charburnout and a decrease in furnace permeability orincreased soot formation in the top gas.

Previous modeling studies of gas combustion9

and the results of tests of gas injection with lancesof different diameters suggest that the extent ofpartial combustion of the gas within the tuyeres canand should be controlled. Regions of stable furnaceoperation were defined based on the mole ratio ofoxygen to gas in the blast and the ratio of the gas-to-blast momentum. Higher values of either parameterpromote combustion and prevent the occurrence of“twinkling” and partial closure of the tuyere that can

9 Direct Injection of Natural Gas in Blast Furnaces at HighRates — Tuyere/Lance Design, Gas Research Institute (GRI–90/0159), September 1992.

be caused by the cooling effect of high rates of flowof cold gas.

It is not clear how to develop such a simplemap for coinjection of coal and natural gas becausethe combustion rates and mechanisms are sodifferent.

Coal devolatilizes as it is heated, and thecomplex hydrocarbons released are less stable thannatural gas and so could be ignited at lowertemperatures. Turbulent mixing of the blast withthe gas and coal plumes determines the rates of bothheat and mass transfer and therefore the rates ofcombustion of the gas and of devolatilization andthen combustion of the coal.

EFFECT OF NATURAL GASINJECTION LEVEL

ON FURNACE PERFORMANCE

This section presents an analysis of theperformance of No. 4 furnace at Gary with coalonly, gas only, and coinjected gas and coal. Asdiscussed earlier, there were a number of concernsprior to the test regarding the most appropriate waysto set aim values and the effects of high-level gasinjection on furnace performance. Key issues to beresolved included:

14

• The appropriate level for RAFT at variousinjection levels;

• The extent to which increasing levels ofhydrogen would decrease the solution lossreaction;

• The effects that differing levels of fuelinjection would have on permeability;

• The productivity gains that might be obtainedat various levels of fuel injection;

• The replacement ratios that could be achievedwith natural gas; and

• The effects of coinjection on furnace stability,availability, and hot metal chemistry.

In the previous analyses of furnaceperformance with high levels of natural gasinjection, it has been possible to correlate changesin furnace performance with changes in either thelevel of gas injection or the bosh gas hydrogencontent, since these were the parameters that werevaried independently in the tests. Such a simplescheme is not possible here, however, because thecoal injection level and scrap charge to the burdenalso varied significantly throughout the tests.Changes in the coal injection level and scrap chargecan affect furnace performance to a greater extentthan the relatively small changes in gas injectionlevel effected in these tests, and this complicates theanalysis significantly.

Therefore, we have used the bosh gashydrogen content, in moles/THM, rather than theindividual or total level of supplemental fuelinjection, as a correlating parameter whereverappropriate. This parameter does reflect the effectsof changes in fuel mix and injection levelqualitatively since natural gas contains 25%hydrogen (or 2 moles/mole C) while the coals usedhere contain only 3-6% H2 (or about 0.4 mole/moleC). In addition, the data are displayed as part ofeither all-coal, all-gas, or high-, intermediate- orlow-coal injection level groups. Periods with coalinjection levels below 200 lb/THM are in the lowgroup, while periods with coal injection levelsabove 300 lb/THM are in the high group. Theeffects of changes in scrap charge or natural gas or

coal injection level within each group are thenaddressed as required.

Tuyere and Hearth Level Energy Control

Practice with high-level coal injection calledfor changing the supplemental oxygen rate by300 SCFM for each 1.5 TPH change in coalinjection rate, which is equivalent to a marginaloxygen consumption of about 0.5 lb/lb coal. Thechange in RAFT, then, would be determined largelyby the marginal amount of oxygen injected as thelevel of gas injection was changed. As shown inFigure 2 it proved possible to decrease the AISIRAFT by some 450°F during the tests, with theenergy balance RAFT (CRA RAFT) decreasing byabout 350°F.

While there is considerable overlap in thegroupings, the CRA RAFTs decreased withdecreasing coal injection levels (as shown by H2

content) in Figure 3. The decrease in RAFT withina group results from increasing levels of gasinjection, however. The RAFTs within theintermediate coal injection group drop by more than150°F, for example, as bosh hydrogen contentsincrease by about 6 mole/THM. The best line fitthrough these data gives a slope with a decrease ofabout 25°F RAFT per mole/THM increase in thehydrogen content, which is more than 20% greaterthan the drop in RAFT for all gas injection (seereference 6). The lowest RAFTs, around 3,460°F,were obtained at injection levels of about177 lb/THM coal and 125 lb/THM natural gas, andRAFTs were actually 40-60°F higher during the all-gas 170 lb/THM injection condition. Furnaceperformance was quite satisfactory under these lowRAFT conditions, although the operators believedthat the RAFT was actually about 3,600°F based onthe indicated coal and natural gas injection levels.

With the furnace properly balanced thermally,RAFT can be allowed to drop as increasing levels ofcoinjection increase the hydrogen content withoutcompromising hot metal temperature aim values asshown in Figure 4. With all-gas injection it hasbeen shown that the necessary furnace thermalbalance can be obtained by maintaining roughlyconstant values for the thermal-plus-chemical

15

energy contents of the hearth gases. This appears tobe the case with coinjected coal and gas as well, asshown in Figure 5.

Most of the data fall within range of 0.7-0.8MMBtu/THM for the thermal-plus-chemical energy,a range that is narrower than often found in furnacespracticing high-level gas injection. There does notappear to be any trend in the values with coal or gasinjection levels, but the two highest points are forgas-only injection. The furnace was simply beingrun “hotter” in the gas-only period than it had beenin the coinjection period immediately preceding it.

The fundamental reason why it is appropriateto maintain the thermal-plus-chemical energy levelconstant and allow the RAFT to decrease as boshhydrogen contents increase is that changinghydrogen contents change the reductionmechanisms in the furnace. Overall, the hydrogenutilization efficiency does not decrease as hydrogencontents increase, as shown in Figure 6. Asdiscussed earlier, the calculated utilizationefficiencies are sensitive to the coal compositionsand injection rate and natural gas injection rateassumed, and the biases in the top gas hydrogenanalysis. The scatter in the data in Figure 6 aretypical of the scatter obtained for gas-only injectingfurnaces where there is no uncertainty in the rate orcomposition of the injected fuel.

Since the hydrogen content is increasing andits utilization efficiency is constant, the extent of

indirect reduction in the bosh and stack must beincreasing as well. With a burden of constantcomposition, an increase in indirect reduction mustbe accompanied by a decrease in the energyintensive direct reduction reaction. That this doesindeed occur is shown in Figure 7, and it is thedecrease in the high temperature energy requiredthat permits the RAFT and physical hearth gastemperature to be decreased with increasing boshhydrogen contents.

The best fit of the change in the solution lossreaction with changes in hydrogen content gives aslope of about 0.4 mole/mole, which is about0.125 mole/mole higher than for an all-pellet burdenwith all-gas injection. The combined extent ofdirect plus indirect reduction, obtained by summingthe best fit slopes for each, show an apparentdecrease of about 0.1 mole/mole in the total amountof oxygen removed, but this is not related to thechanges in the levels of fuels injected. Rather, it isa consequence of changes in the amount of scrapcharged to the burden. While changes in the scrapcharge occurred under all injection conditions, mostof the practice was with scrap charges betweenabout 200 and 300 lb/THM. At the average scrapcomposition assumed, a change of 100 lb/THM inthe scrap charge would change the total reductionby about 1.5 mole/THM and, on average, there wasmore scrap on the burden when hydrogen contentswere higher than when they were lower.

16

Figure 2. AISI RAFT and CRA RAFT vs. Bosh Gas Hydrogen Content

10 12 14 16 18 20 22 24

Bosh Gas Hydrogen Content, lbmol/THM

3,400

3,500

3,600

3,700

3,800

3,900R

AF

T,

°F

AISI RAFTCRA RAFT

Source: Charles River Associates, 1998

Figure 3. CRA RAFT vs. Bosh Gas Hydrogen Content

10 12 14 16 18 20 22 24


3,400

3,450

3,500

3,550

3,600

3,650

3,700

3,750

3,800

CR

A R

AF

T, °F All Coal

High CoalInt. CoalLow CoalAll NGI


17

Figure 4. Hot Metal Temperature vs. AISI RAFT

ALL_PCI

HIGH_PCI

INT_PCI

LOW_PCI

ALL_NGI

AISI RAFT, °F

Ho

t M

eta

l Te

mp

era

ture

, °F

2630

2650

2670

2690

2710

2730

3,900 3,800 3,700 3,600 3,500 3,400


Figure 5. Thermal Condition vs. Bosh Gas Hydrogen Content

10 12 14 16 18 20 22 24


0.65

0.70

0.75

0.80

0.85

Therm

al &

Chem

ical E

nerg

y above

2,7

00°F

, M

MB

tu/T

HM

All CoalHigh CoalInt. CoalLow CoalAll NGI


18

Figure 6. Overall Hydrogen Utilization Efficiency vs. Bosh Gas Hydrogen Content

10 12 14 16 18 20 22 24


35

40

45

50

55

60

Hyd

rog

en

Util

iza

tion

Eff

icie

ncy

, %



Figure 7. Extents of Direct and Indirect Reduction vs. Bosh Gas Hydrogen Content at USS-Gary BF No. 4

10 12 14 16 18 20 22 24


0

10

20

30

40

Ext

en

ts o

f R

ed

uct

ion

, lb

mo

l/TH

M


EGC

Indirect


These analyses show that the aim values andthermal control for a furnace coinjecting natural gasand coal should be altered from values used with allcoal injection.

• The RAFT should be allowed to drop withincreasing levels of gas injection and bosh

hydrogen contents. The rate of decrease is atleast as high as for gas-only injection, at about25°F/mole/THM H2.

• This decrease is appropriate because the extentof the highly endothermic solution lossreaction decreases with increased bosh

19

hydrogen content. The rate of decrease inthese tests was higher than for gas-onlyinjection, at about 0.4 mole/mole H2, but thatmay have been influenced somewhat by thechanges in scrap charge made during the tests.

• Maintaining an essentially constant value forthe thermal-plus-chemical energy will provideproper thermal balance for the furnace as thelevels of coinjection change, even thoughRAFT is decreasing.

Furnace Productivity and Permeability

Increases in furnace productivity can beobtained by driving the furnace harder, i.e.,supplying more oxygen to it, or by reducing thespecific furnace energy requirements by puttingmetallics on the burden, or both. As discussedabove, the scrap charge on the burden was mostlybetween about 200 and 300 lb/THM through thetests, and its composition was variable. Furnaceproductivity data are shown in Figure 8 as afunction of the total amount of oxygen delivered tothe furnace. As expected, these productivities arehigher than for all-pellet burdens at the sameoxygen consumption because of high scrap charge.The slope of the best fit through the data shows amarginal productivity of about 3.4 TPD hotmetal/TPD oxygen supplied, which is quite high. Inabsolute terms, there was an increase of 460 TPD inproductivity, or 12%, with coinjection of naturalgas. While there is some overlap in the groupings,it is clear that the furnace took more oxygen withlow coal injection levels than at intermediate levels,and more at intermediate than at high levels or allcoal practice. As will be shown, there was someimprovement in furnace permeability as the level ofcoal injection decreased and the level of gascoinjection increased. However, increasing thelevel of coinjection also decreased the actualvolume of the bosh gases, because it wasaccompanied by higher enrichment and the

temperature of the gases was lower at the lowerRAFTs. This would have allowed the furnace toaccept more total oxygen even at a constant level ofpermeability.

Different levels of scrap charge would haveaffected both the absolute productivity and the slopeof the productivity–oxygen curve. Evaluation ofdata from gas injecting furnaces using wellcharacterized scraps on the burden showed that thecontribution to productivity was about 0.54TPD/CCF/100 lb/THM Fe in the scrap. Using thiscorrection factor, and normalizing to a blasttemperature of 1,850°F using a correction factor of0.3 TPD/CCF/100°F change in temperature, shiftsthe slope and location of the productivity curve asshown in Figure 9.

Here the slope has been reduced to about 2.65TPD production/TPD oxygen, but this is stillsignificantly higher than the average of about 2.0TPD/TPD obtained from furnaces injecting gasonly. However, the data on gas injection furnacesextend normalized productivity to 9 TPD/CCF andoxygen consumption to 3.8 TPD/CCF, far higherthan obtained here. In fact, the high range of thedata for coinjection in Figure 9 (7 and 2.9TPD/CCF, respectively) are right in the middle ofthe low range of data for gas injecting furnaces andthey are lower at the low end of productivity andoxygen consumption. Data at higher levels ofcoinjection and oxygen consumption would berequired to determine if the high slope shown herepersists to even higher levels of productivity.

There is reason to believe that the slope woulddecrease, however, because the relativecontributions of supplemental oxygen and wind tothe total oxygen supply would change at highproductivity. The sources of oxygen consumed bythe furnace during the tests are shown in Figure 10.

20

Figure 8. HM Production vs. Total Delivered Oxygen

2.3 2.4 2.5 2.6 2.7 2.8 2.9 3 3.1

Total Delivered Oxygen, TPD/CCF

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

HM

Pro

duct

ion,

TP

D/C

CF



21

Figure 9. Normalized HM Production vs. Total Delivered Oxygen

2.0 2.5 3.0 3.5 4.0

Total Delivered Oxygen, TPD/CCF

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

9.0

9.5

No

rma

lize

d H

M P

rod

uct

ion

, T

PD

/CC

F

All CoalHigh CoalInt. CoalLow CoalGary NGIOther NGI

Figure 10. Wind and Oxygen vs. HM Production

3,200 3,400 3,600 3,800 4,000 4,200 4,400

Hot Metal Production, TPD

0

500

1,000

1,500

2,000

Oxy

gen, T

PD All Coal

High CoalInt. CoalLow CoalAll NGI

Wind

Total

Supplemental


Higher productivity was achieved byproviding the furnace with both more supplementaloxygen and more wind, and the slopes of each werealmost exactly the same. This is in sharp contrast tothe situation with injection of gas only where

increased productivity is obtained with less windand much more supplemental oxygen. When gasinjecting furnaces are not being pushed forproduction, supplemental oxygen consumption athigh injection levels can be as low as 0.67 lb/lb gas

22

(see reference 7). When productivity is required,however, the supplemental oxygen consumption istypically in the range of 1 lb/lb, or about 8 lb ofoxygen per mole of contained hydrogen. Theconsumption of supplemental oxygen in these testsis shown in Figure 11 normalized to the bosh gashydrogen content, most of which is derived from thecoinjected coal and gas. While there is considerablescatter in the data, the best fit gives a marginalconsumption of oxygen of about 5.9 lb/mole H2.. The low coal injection data on average lie below thebest fit while the intermediate and high coal data lieabove it. These data show that total supplementaloxygen consumption was driven by coal injection,not gas injection, and the marginal consumption ofoxygen for gas was only of the order of 0.6 lb/lbgas. On a weight basis this is slightly higher thanthe amount of oxygen provided for coal combustion.However, this is only about 0.3 mole O2/mole CH4,which is far below the amount required forcomplete combustion. During the gas-onlyinjection periods supplemental oxygen consumptionwas very high, at more than 1.4 lb/lb, whichcontributed to the high thermal-plus-chemicalenergies and high hearth gas temperatures producedat that time.

The balance of supplemental oxygen and windthat is “optimum” for any mix of coinjected fuelsmay be constrained by the furnace productionrequirements and furnace permeability, and bothvaried significantly over the course of these tests.

The values of the furnace pressure drop perunit height divided by the average burden densityare shown in Figure 12, and the scatter in the dataare typical of furnaces with all-pellet burdensinjecting natural gas only. There does not appear tobe any dependence of pressure drop on either coalinjection level or coinjection level as measured bythe bosh hydrogen content. This is probablybecause operators would react to increasing blastpressures by reducing the coal and oxygen injectionrates, or by reducing the wind rate.

The permeability is lower than for all-pelletgas injecting furnaces, however, by as much as afactor of two. Of the variables examined, the totallevel of injection of supplemental fuels was the bestpredictor of furnace permeability as shown in Figure13.

As is the case with other correlations shown inthis chapter, the data can be grouped by coalinjection level. Permeabilities for all coal injectionpractice lie below the best fit through all of the data,and the slope of the permeability-injectionrelationship is high with about a 30% loss inpermeability per 100 lb/THM increase in coalinjection level. The high and intermediate coal ratecoinjection data straddle the best fit, and showpermeability losses of more than 15%/100 lb/THMincrease in injection level. These relationshipsbetween increasing permeability loss and increasingcoal injection level are consistent with theoperator’s previous experience with all coalinjection practice.

Some of the scatter in the data, and thegenerally low permeability values, is due to thepresence of the large amounts of scrap of variablequality and ore, burdening practice, and the effectsof the injected coals. The general increase inpermeability with increasing bosh gas hydrogencontents appears to be related to increases in thelevel of coinjected natural gas, however, which isconsistent with the observation that operators oftenmake of smoother burden descent upon introductionof natural gas.

Furnace Fuel Consumption

Changing the amount and mix of fuelscoinjected would change the furnace fuel ratebecause coal and natural gas have differentreplacement rates for coke. The fuel rate would alsochange as the scrap charge to the burden, blastmoisture and blast temperature, and hot metalsilicon contents changed. The furnace coke and fuelrates obtained in these tests are shown as a functionof the bosh gas hydrogen content in Figure 14.

23

Figure 11. Supplemental Oxygen Use

10 12 14 16 18 20 22 24


140

160

180

200

220

240

260

Su

pp

lem

en

tal O

xyg

en

, lb

/TH

M



Figure 12. Pressure Drop per Unit Height Divided by Burden Density

10 12 14 16 18 20 22 24


0.20

0.25

0.30

0.35

0.40

0.45

0.50

Pre

ssure

Dro

p/H

eig

ht/B

urd

en D

ensi

ty, U

nitl

ess



24

Figure 13. Burden Permeability vs. Total Injectant Level

150 200 250 300 350 400 450

Total Injected Fuel, lb/THM (NGI+PCI)

5

6

7

8

9

10

Burd

en P

erm

eabili

ty, 1/K

', m

^2/s

^2



Figure 14. Unadjusted Coke and Fuel Rate vs. Bosh Gas Hydrogen Content

10 12 14 16 18 20 22 24


500

600

700

800

900

1,000

1,100

Fuel R

ate

, lb/T

HM


Coke Rate

Total Fuel Rate


Once again there is considerable scatter in thedata, with R2 values for the best fits to the fuel and

coke rates of only 0.28 and 0.12. The data aregrouped, however, with the low coal periods

25

showing lower and the high coal periods showinghigher than tread line fuel rates. This behavior isgenerally reversed on the coke rate trend line. Thecoke rate increased slightly as bosh hydrogencontent increased because of amount and mix ofinjected fuels changed. The total injection levelsdecreased from 330-400 lb/THM for the high-levelcoal periods to 310-340 lb/THM for theintermediate and 270-300 lb/THM for the low coalperiods. The total fuel rate decreased, however,because lower replacement ratio coal was beingdisplaced in the supplemental fuels mix by higherreplacement ratio natural gas. The natural gasinjection levels increased from 40-60 lb/THM in thehigh coal periods to 50-90 lb/THM and then to 100-120 lb/THM in the intermediate and low coalperiods.

The data in Figure 14 cannot be used todetermine the replacement ratio obtained with thecoinjected gas because of the changes in burdeningand coal injection levels that occurred at the sametime. To “back out” the effects of the injected gas it

is necessary to account for the effects that otherchanges in practice had on the coke rate. Based onanalysis of the performance of other furnaces, thecoke correction factors in Table VIII were applied tothe furnace parameters presented in Table VI tonormalize them to a common basis. The coke andfuel rates obtained by normalizing performance inthis way are shown as a function of the natural gasinjection level in Figure 15.

Table VIII. Coke Correction Factors used toNormalize Furnace Fuel and Coke Rates

FactorReference

Value Adjustment Factor

Coal 250 lb/THM 0.85 lb coke/lb coal

Scrap 250 lb/THM 0.20 lb coke/lbMetallic Iron

Blast Moisture 7 gr/scf 4 lb coke/gr/scf

HM Silicon 0.7% 10 lb coke/0.1% Si.

Figure 15. Adjusted Coke and Fuel Rate vs. Natural Gas Injection Level

0 50 100 150 200

Natural Gas Injection Level, lb/THM

500

600

700

800

900

1,000

1,100

Fu

el R

ate

, lb

/TH

M


Adjusted Coke Rate

Adjusted Total Fuel Rate

Note: Coke Rate Adjusted to 250 lb/THM PCI (0.85 lb Coke/lb Coal), 250 lb/THM Blended Scrap (20 lb Coke/100 lb Fe Met.) 7 gr/SCF blast moisture (4 lb coke/gr), and 0.70% HM Si (10 lb coke/0.1% Si)Source: Charles River Associates, 1998

26

The normalization procedure decreases theamount of scatter in the unnormalized data shown inFigure 15 and results in the expected trend ofslightly decreasing fuel rate and strongly decreasingcoke rate with increasing levels of natural gasinjection. The replacement ratio for coinjection ofnatural gas, the slope of the coke rate curve, isalmost 1.1 pound of coke displaced per pound ofgas injected, with an R2 value of the fit of 0.86. Thisvalue is somewhat sensitive to the magnitude of thecoke correction factors chosen. If the scrap qualityis lower than assumed, its coke correction factorwill be lower, but decreasing the correction factor to0.1 lb coke/lb Fe only increases the estimatedreplacement ratio for gas by 0.04 lb/lb Theestimated replacement ratio for gas is more sensitiveto the replacement ratio assumed for the coal: achange in coal’s replacement ratio by ±0.05 lb/lbwould result in a change in the ratio calculated forgas of ± 0.08 lb/lb These conclusions do notdepend on the reference point chosen to normalizethe data.

The data in Figure 15 represent the normalizedestimates of coal, coke, and natural gas rates afterapplying the correction factors to coal and gas ratesas discussed earlier. These data, in our judgment,best represent furnace performance, satisfyingmaterial and energy balances with the leastcorrection to “as-reported” data to give reasonableresults. Since the natural gas rates have beencorrected systematically by 20% and the coalinjection rates have been corrected upward by 10%at the lower range of rates, use of the “as-reported”data would have resulted in the calculation of higherreplacement ratios for gas. Using all as-reporteddata and the coke correction factors in Table VIIIgives a replacement ratio for gas of about 1.25 lb/lb

Given the limited span of injection ratescovered and the uncertainty in the coke and scrapcorrection factor used, the sense of the data is thatthe replacement ratio obtained for gas in these testswas at least 1.1 lb/lb

Hot Metal Quality

In Figure 4 it has been shown that the hotmetal temperature did not change significantly asthe level of gas coinjection increased and the RAFTdecreased. The standard deviations in the hot metaltemperatures were also unaffected by thecoinjection level, so by these measures hot metalquality was not affected by this practice.

With average coal sulfur contents in the rangeof 0.8-0.9%, a replacement ratio for coal of less than1.0, and coke sulfur contents of about 0.6%,decreasing the level of coal injection and increasingthe level of gas injection would decrease the sulfurload to the furnace. During these tests the sulfurlevel decreased from about 7.5 to about 5.5 lb/THMas the bosh hydrogen content increased from about12 to about 22 mole/THM.

The operators did not readjust the aim valuefor slag basicity during the tests and it remained inthe vicinity of 1.1. As a result of these changes, thehot metal sulfur content decreased, as expected,from about 0.05% in the base case periods to lessthan 0.04% in the low or no coal high gas injectionperiods, as shown in Figure 16. The decrease issignificant with respect to the magnitude of thestandard deviation in the averages, which weretypically about 0.011%.

The reduction behavior of silica underinjection or coinjection conditions is not wellunderstood. Overfueling the furnace increased thesilicon content in these tests even in the absence ofsignificant increases in hot metal temperature. Thedata in Figure 17 suggest that the hot metal siliconcontent decreased somewhat as the level of coalinjection decreased and the coinjected gas levelincreased, independent of changes in hot metaltemperature as shown in Figure 4. While there isconsiderable scatter in the data, the slope of thedecrease is large with respect to the standarddeviations of the averages, which are about 0.15%.

27

Figure 16. Hot Metal Sulfur Content vs. Bosh Gas Hydrogen Content

ALL_PCI

HIGH_PCI

INT_PCI

LO_PCI

ALL_NGI

Bosh Hydrogen Content, lbmol/THM

Ho

t M

eta

l Su

lfur

Co

nte

nt,

%

0.015

0.025

0.035

0.045

0.055

0.065

0.075

0.085

10 15 20 25


Figure 17. Hot Metal Silicon Content vs. Bosh Gas Hydrogen Content

ALL_PCI

HIGH_PCI

INT_PCI

LOW_PCI

ALL_NGI


Hot M

eta

l Sili

con, %

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

10 15 20 25


In addition to the usual statistical measures ofhot metal quality, the standard deviation in hot

metal silicon and sulfur contents and temperature,CRA calculates the cast-to-cast variability in the

28

thermal state of the hearth as measured by thechange in energy required to produce the amountand composition of the hot metal in each cast (seereference 4 for an extended description of thisparameter). A furnace that could produce hot metalwith identical compositions and temperatures onsuccessive casts would show a cast-to-castvariability, dQ/dt, of zero MMBtu/hr. The greaterthe changes in temperature and composition, thegreater the variability. The values of the cast-to-castvariability for the range of bosh hydrogen contentsin these tests are shown in Figure 18, together withsome data from the high-level injection tests atACME.

The bosh hydrogen contents in these testsranged from about 11 to 22 mole/THM, whereas therange in the tests at ACME was from about 4 to37 mole/THM because the practice there variedfrom no injection to an injection level of more than300 lb/THM gas. For injection levels where thebosh hydrogen contents were the same, there is nodifference between the cast-to-cast variability forthe two operations. The implication of this findingis that bosh hydrogen contents above 10 mole/THMor so are sufficient to reduce cast-to-cast variabilityto the same level regardless of burdening practice orthe level or composition of the injectants. This issomewhat surprising given the well knowndeleterious effects that high-level coal injection andpoor quality scrap have on permeability and burdenmovement.

ECONOMIC IMPLICATIONS OFCOINJECTION PRACTICE

Under circumstances in which the economicsof scale inherent in large coal preparation systemscan be utilized, injection of granular or pulverizedcoal at levels of 250–350 lb/THM can providesubstantial economic benefits in terms of cokesavings, so long as burdening practice can beadjusted to prevent productivity losses and limit theconsumption of supplemental oxygen. The lowestcosts will probably be obtained when the coalpreparation and injection systems are operated neartheir capacity limits, the proper coal has beenselected, and coke savings are maximized.

The coke replacement ratios that can beachieved by injected coals depends on furnaceoperating conditions, injection practice, andespecially on the composition of the coal. However,available data on medium to large sized furnacessuggest that replacement ratios are normally belowone pound coke displaced per pound coal injected,and that increasing the injection level of coaltypically results in decreases in burden permeabilitythat will either constrain production or require highlevels of blast enrichment.10 These tests haveshown that coinjection of natural gas at levels up to125 lb/THM can provide additional coke savingswith a high replacement ratio as well as increasedproductivity with relatively low marginalconsumption of supplemental oxygen whenadditional hot metal is required.

The economic impacts of coinjection practiceunder typical North American conditions areillustrated by the examples shown in Tables IXthrough XII.

The enrichment necessary in these examples ishigher than was required in the tests at Gary becausethe coal injection rate is constant and permeabilitymust be expected to decrease as the total amount offuel injected increases. The decrease shown in coalinjection level results from the increase in furnaceproductivity, not from a decrease in injection rate.

The decreases in coke rate resulted both fromreductions in the extent of the solution loss reactionand in the blast moisture level that are madepossible by the hydrogen content of the injected gas.The amount of top gas generated decreases as thegas injection level increases, but its energy contentincreases because of its higher hydrogen content.

When coinjecting gas at relatively low levels,costs are reduced in the furnace area because thereplacement ratio is high and relatively littleadditional enrichment is required. Whencoinjecting at high levels to obtain productivityincreases, the incremental costs of oxygen and gasexceed the savings due to decreased coke

10 The Impacts of High Rates of Fuel Injection on CokeReduction and Productivity Improvement in the BlastFurnace, Gas Research Institute (GRI-96/0226), June 1996.

29

consumption, but that is more than offset by the value of the incremental hot metal produced. Also,

Figure 18. Averages and Standard Deviations for Cast-to-Cast Variability

ALL_PCI

HIGH_PCI

INT_PCI

LOW_PCI

ALL_NGI

ACME

Bosh Hydrogen Content, lbmol/THM

dQ

/dt,

MM

Btu

/hr

0

5

10

15

20

25

30

35

40

4 6 8 10 12 14 16 18 20 22

Table IX. Compositions Assumed for Economic Analysis of Coinjection Practice

Material Composition, %

H2O Fe C S SiO2 AI2O3 CaO+MgO H2 O2

Pellets(1) 5.0 63.1 - 0.01 3.59 0.66 - 27.2

Sinter(1) 3.0 55.2 0.03 0.03 6.14 12.97 - 22.0

Coke 3.7 0.7 91.9 0.53 6.50 0.23 - -

Coal 1.0 -(2) 86.8 0.83 -(2) -(2) 4.23 1.76

(1) Burden is 60% pellets, 40% sinter.(2) Ash content 5.1%.

30

Table X. Furnace Parameters for Coinjection Economic Analysis

Case Base CaseCoke RateReduction

ProductivityIncrease

Parameter Units

Production

Productivity

TPD

TPD/CCF

7,000

7.77

7,000

7.78

7,400

8.22

Wind

Supplemental O2

MCF/THM

lb/THM

TPD

32.8

136

475

31.6

165

577

29.3

233

861

Blast Moisture

Blast Temperature

Gr/SCF

ºF

12

2,075

5

2,075

5

2,075

Coal Injection

Gas Injection

lb/THM

lb/THM

250

0

250

50

236

125

RAFT(1)

TCE(2)

ºF

MMBtu/THM

3,880

0.90

3,796

0.89

3,547

0.84

Coke Rate

Fuel Rate

lb/THM

lb/THM

729

979

658

958

600

961

Hot Metal S

Hot Metal Si

Hot Metal Temperature

%

%

ºF

0.0415

0.50

2,700

0.0395

0.50

2,700

0.0369

0.45

2,700

Top Gas Temperature

Top Gas CO/CO2

Top Gas HHV

Top Gas Energy

ºF

-

BTU/SCF

MMBtu/THM

310

1.11

89.7

4.66

333

1.07

92.2

4.69

358

1.12

104.7

5.28

Stove Heat

Blast Compression

MMBtu/THM

MMBtu/THM

2.09

1.55

2.04

1.51

1.95

1.44

(1) By energy balance with coke at 2900ºF; AISI RAFTs are higher.(2) Thermal plus chemical energy content of the hearth gases.

Table XI. Unit Costs for Economic Analysis of Coinjection Practice

Component Unit/Cost Component Unit Cost

Coke $120/ton Hot MetalDesulfurization

$0.05/0.001%

Coal(1) $60/ton Hot Metal Value(2) $30/ton

Natural Gas $3.0/MMBtu Top Gas Value $2.25/MMBtu

Supplemental O2 $26/ton Blast Moisture $4/Mlb

(1) Incremental cost of purchased coal plus preparation.(2) Incremental value.

31

Table XII. Cost Savings for Coinjection Practice

Cost Savings from Base Case, $/THM

Coinjection Case Coke Rate Reduction Productivity Increase

Cost Component

Furnace Area

Coal - 0.45

Gas (3.34) (8.34)

Coke 4.26 7.74

O2 (0.39) (1.30)

Total Area 0.53 (1.45)

Utilities and Other

Hot Metal Value - 1.62

Desulfurization 0.03 0.16

Blast Moisture 0.38 0.39

Stove Enrichment(1) 0.21 1.12

Total Utilities & Other 0.62 3.29

Net Top Gas Value 0.15 0.69

Total Savings and Credits(2) 1.15 – 1.30 1.84 – 2.53

(1) Reductions in gas consumption to fire stoves at 105 Btu/SCF.(2) Savings plus credits without and with value attributed to additional net top gas energy generated.

the total amount of coke consumed per day is lowestfor this case at about 2,200 TPD versus about 2,550TPD in the base case.

The coinjection practice described here resultsin additional cost savings and credits outside of thefurnace area due to reductions in desulfurizationcosts and in the energy requirements for blast steamgeneration and stove fuel enrichment. The lattersavings result from the increase in the heating valueof the top gas and are recovered even if there isexcess top gas available. Additional credits areavailable if the increased energy content of the topgas can be used elsewhere in the mill.

The extent to which coinjection of natural gaswith coal can reduce coke consumption and increasefurnace productivity will depend on the specificoperating practices and constraints on a givenfurnace. The economic benefits will depend onlocal unit costs and the amount of performanceimprovements that can be obtained. The savings

projected in Table XII are very significant, however,and show that coinjected gas can be a valuable toolto optimize the performance of furnaces injectingcoal.

Summary and Conclusions

Baseline and coinjection testing carried out onNo. 4 furnace at Gary over a year’s time showedthat:

• Increasing the level of coinjected gas whiledecreasing coal injection levels can increaseproductivity by more than 12%.

• Improvements in permeability allow thefurnace to take more wind so thatsupplemental oxygen injection levels onlyhave to be increased modestly.

• A coke replacement ratio for coinjected gas ofup to at least 1.1 lb/lb can be obtained, so thatlow furnace coke and fuel rates can beobtained at high productivity.

32

• RAFTs can be allowed to decreasesubstantially from the levels practiced for allcoal injection without compromising furnacestability or hot metal chemistry.

• Furnace SOPs appropriate for coal injectionmay have to be modified for successfulcoinjection practice.

• It is important to verify burden and fuel assaysand all furnace measuring instrumentation toobtain valid data and evaluate properly theeffects of changes in practice.

• Coal chemistry is very important and has alarge impact on the blast furnace performance.

Coinjection of Natural Gas and Pulverized Coal in the...

Documents

Transcript of Coinjection of Natural Gas and Pulverized Coal in the...