alroll_afrc - American Combustion Inc

RECHERCHE ET DEVELOPPEMENTAmerican/Japanese Flame Research CommitteesInternational Energy Agency2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, HawaiiOXYCOMBUSTION FOR REHEAT FURNACES:MAJOR BENEFITS BASED ON ALROLL, A MATURE TECHNOLOGYO. Delabroy, O. Louédin, R. Tsiava, G. Le Gouefflec, P. BruchetAIR LIQUIDE, Francecontact : Bernard.zamuner@airliquide.comAbstract : The steel industry of the 21 st century is facing major challenges in continuously reducingenvironmental emissions while improving the economic viability of their processes. Reheat furnaces, due to thehigh-energy rate used in these processes, are no exception to this rule.Challenges encountered by the steelmakers are various: reheating the product (billet, slab, strip, etc.) may be thebottleneck of the production line, additional energy savings may be needed, stronger environmental target maybe aimed, etc. For all these issues, a solution exists based on furnace revamping using ALROLL, anoxycombustion technology. ALROLL appears to be a low investment cost, economically viable and veryflexible solution.Dedicated oxycombustion technology have been developed and validated on the field. Among that, side walloxyburners dedicated to reheat furnace conditions (large flue volume of combustion products crossing thefurnace, large furnace width), flat flame burners, NO x reduction techniques, etc.Beyond that mature technology, all the tools are available to design and propose a solution to common issues(0D, 1D modeling) as well as advanced issues like bending or specific product arrangement in the furnace (3Dmodeling). The specificity of all these models is that coupling is operated between the furnace and the product.The evolution of the product temperature distribution is obtained but also the product mechanical deformation aswell as its surface properties (scaling, decarburization).Examples of ALROLL field demonstration are detailed among them annealing-pickling lines, various reheatfurnaces (billet, bloom), in order to illustrate the variety of issues being solved in different steel processes.ALROLL oxycombustion technique solving the majority of furnace issues is leading to significant savings infossil energy consumption, reduction in pollutant emission (NO x and CO 2 ), while maintaining an excellentproduct quality even when the furnace performances are increased well over their nominal design.1. THE REHEAT FURNACE SITUATION IN STEEL MAKINGBoth oxygen steelmaking route and electric route are producing large section products such as blooms,billets or slabs (see Fig. 1). Section reduction is only achievable by hot rolling because of the forcesthat are required for such heavy product. Hot rolling is a combination of two major equipments: thereheat furnace and the rolling stands. The reheat furnace brings the product at the right temperature formilling operation (typically 1200° C).The reheated products are long products (billets or blooms) or flat products (slabs). Most of the recentreheat furnaces are walking beam furnaces where the beams lift all the products inside the furnace andmove them to the next position. This direct-fired heating process can occur both above and below theAFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 1/23

RECHERCHE ET DEVELOPPEMENTproduct line, using roof burners (to heat the vault that will then radiate to the products) or long flameburners (side walls or front). Fuel can be natural gas, coke oven gas or heavy oil. Low calorific valuefuel (eg: blast furnaces) are also used.Fig. 1 : Reheat furnaces in the steel making processIn the past years, steelmakers have made some developments and investments on the upstream anddownstream part of the process. However, reheat furnace have not been concerned by suchinvestments. Now, many steelmakers face a situation where there is a need in productivity increaseand the bottleneck of the process line is the reheat furnace.AIR LIQUIDE develops combustion technologies based on oxycombustion. ALROLL is dedicatedto reheat furnace and especially for retrofitting reheat furnaces that are bottlenecking. ALROLL isnow implemented in several industrial sites. This paper presents the advantages of this technology.Focus is made on three industrial references, based on different product issues and differentbottlenecking issues.In addition with productivity increase and bottlenecking, specific issues may exist which are alsosolved using ALROLL. Energy saving is one of these issues, especially for furnaces with no or witha low efficiency fumes energy recovering system (recuperators). Pollution is another one and is alsoaddressed by ALROLL. Because the high-energy rate used in these processes, these large facilitiesare concerned by environmental issues. Consequently, in every investment project concerning reheatfurnaces, meeting the environmental target is becoming a major issue.As reheat furnaces are the major targets for the spread of ALROLL technology, heat treatmentfurnaces are also targeted. In particular, all the furnaces equipped with a direct-fired zone mayencounter similar difficulties as in reheat furnace. As ALROLL for annealing-pickling line hasalready been discussed in other publications (see Bougault et al., 1998, 1999), an industrial referencefor a galvanizing line is discussed at the end of this paper.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 2/23

RECHERCHE ET DEVELOPPEMENT1.1. Principles of oxycombustionWhen combustion in air, oxygen is mixed with nitrogen that does not contribute to reaction. Thenitrogen is entirely evacuated in the flue gases produced by combustion that exit at a temperature closeto that of the furnace, representing a waste of energy.Oxycombustion consists in using the combustion with oxygen instead of air. By eliminating thenitrogen ballast in the air, it ensures energy optimization of the furnace without the need of any heatrecovery device. The geometrical configuration of the furnace is not changed in any way at all and theadaptation work is limited in terms of time and cost.The simulation in Fig. 2 describes the natural gas, air and oxygen volumes together with the flue gasesgenerated in a system (furnace temperature is 900°C) that transferred to the load an effective power of5 kWh. It demonstrates that the energy efficiency of oxycombustion is equivalent or even better thanthe "highly-preheated air" solution.Moreover, thermal transfer by oxycombustion is characterized by considerable localized transfer dueto high emissivity (considerable concentrations of CO 2 and H 2 O in the flames) and reduced flamevolume leading to first an enhanced capability to transfer its energy to the load and second an extragain in energy efficiency.Fig. 2 : Theoretical comparison between air and oxy combustionFinally, oxygen combustion is based upon the total elimination of nitrogen from the oxidizer: thisleads to a reduction in NO x concentration. Furnaces fired with 100% oxygen will obviously achievevery low NO x level. It has been demonstrated in the glass industry where more and more new furnacesare designed with only oxyfuel burners. This paper shows that NO x reductions are achieved in reheatfurnace, even when only part of the furnace is fired using oxycombustion.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 3/23

RECHERCHE ET DEVELOPPEMENT1.2. What alternative solution?As mentioned above, retrofitting projects are mainly motivated by a need in furnace energyoptimization. To improve the thermal transfer to the load, several alternative solutions could beconsidered.Extending the length of the furnace would consist in building an additional section and obviouslyrepresents a considerable investment and a long production stoppage.Improving the flue gases recovery device and preheating the combustion air at higher temperatureincludes the installation of a recovery system, the modification of the entire hot air circuit as well asthe installation of new preheated air burners that again represents a considerable investment andextended stoppage.Installing regenerative burners would imply installing twice as many burners and using themalternately. Several drawbacks exist, including the cost and maintenance rate of the burners,alternating operation that can be damaging the quality of reheating, flame stability during ignitionand/or power transition phases. Because the regenerative burners develop high temperaturecombustion in the presence of nitrogen, the NO x constraint is still a concern, even if technology hasbeen improved in the past few years.1.3. Bottlenecking reasons solved by ALROLLThe productivity is the major motivation for ALROLL installation. The main reasons of productionbottlenecking are illustrated below with a billet walking beam furnace. The identical reasons exist withpusher furnaces, walking hearth furnaces, for billet, bloom or slabs. The direct-fired zone of stripfurnaces are also concerned (see the reference for the galvanizing line).Detection of bottleneckingThe detection signal for productivity problem is generally the following. As the final targettemperature of the product is not achieved, the furnace operator demands an increase of the firingpower. This demand can be automatically generated through a furnace software controller. Facing thispower demand, the furnace system cannot answer positively: the furnace becomes the bottleneck. Theonly solution for the furnace operator is to pace down the given products.Typical limitations are shown on Fig. 3. For most of these limitations, a solution exists, based on theALROLL technology. Thermocouples of heating zones can be at the maximum authorizedtemperature. Furnace pressure may be at maximum, the firing power of dedicated zone may be at100% of the available firing power (due to limitations in air or fuel flow rates). The limitedtemperature at the recuperator entrance is a limitation that appears often on reheat furnaces.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 4/23

RECHERCHE ET DEVELOPPEMENTFig. 3 : Typical bottlenecking detection signals for a reheat furnaceSome other bottlenecking issues are correlated to the one discussed above: the limitations due to theproduct bending or to product thermal non-homogeneity. These signs show defaults in the heattransfer to the product. If the bending and / or the product homogeneity is not ameliorated when thefurnace power is increased, again, oxycombustion and its increased heat transfer efficiency is surelythe best technical solution available.Flexibility of ALROLL as an answer to furnace selective bottlenecking.It is important to note that generally, the conditions, which the furnace is the bottleneck for, are wellidentified. In particular, they appear only for certain type of product and/or certain charging conditions(cold / hot charging), for identified finished type products, etc. The bottleneck may not concern thetotality of the production. Heavy investment solutions that will address only a part of the totalproduction are of course a certain financial concern for the steelmaker. The ALROLL solution is notonly a low cost investment but also a flexible solution. When the furnace is not bottlenecking, theALROLL burners are not firing and the furnace works at it was before the ALROLL installation.2. ALROLL, A GLOBAL SOLUTION2.1. Burner technologiesALROLL burner: generic long flame burners.This flame technology is based on pipe-in-pipe injection of fuel gas and oxygen. It produces longcylindrical-shape flame, whose characteristics can be adjusted by simply changing fuel impulse troughgas injectors. ALROLL burners (Fig. 4) have a high flexibility in terms of power (30 to 150% ofnominal design). It is designed for various fuel and even dual fuel supply when necessary. Theimpulsion of natural gas and O 2 can be adjusted to control the flame length to the furnaceenvironment.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 5/23

RECHERCHE ET DEVELOPPEMENTFig. 4 : ALROLL burner -nominal power 1000 kW (left) / 200 kW (right)ALROLL FC burner: flat flame burner.This flame technology is based on separate injections of fuel gas and oxygen along two parallelplanes. It ensures a large and flat area of combustion. It is mainly characterized by a highly radiativepower and low speed of particules. ALROLL FC burner has also a high flexibility in terms ofpower. It is designed for various fuel and even dual fuel supply when necessary.The flame shape and a appropriate distribution of burners can easily replace a radiant vault zone ofburner. Actually, it is possible to maintain a plane zone of combustion at a medium distance betweenproducts and vault, without any heavy heating of refractory. The low flame temperature makes itpossible to be installed in cold zones or at the entry section of a furnace, when it is sufficientlyprotected from transverse flows (see § 4.2). Such a burner is particularly efficient and NOx-freebecause of a conjunction of low nitrogen content and a low flame temperature.For a ALROLL FC 1000 kW, Fig. 5 presents a iso-surface in the flow field of the fuel mass fraction(corresponding to X = 0.001) together with transversal temperature profiles at 1m, 2m and 3m fromthe burner tip, respectively. The results in Fig. 5 give a good indication of the extended length of theflame, as well as the flame flatness and thickness. This flame geometry leads to lower crowntemperatures and to increased heat transfer to the load surface.Fig. 5 : ALROLL FC 1000 kW : fuel mass fraction iso-surface, and T° profile in radial planes at1, 2 and 3 meters from the burner tipAFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 6/23

RECHERCHE ET DEVELOPPEMENTALROLL AOG burner : for flexibility burnersFor the processes that have a large variety of products (length, thickness, steel grade), oxyfuelapplication might be useful only in some certain conditions. In order to meet the highest economicperformance, AIR LIQUIDE has developed a dual-comburant burner (O 2 /air) that enables to switchfrom air-fuel combustion to oxygen-fuel combustion automatically on the same equipment.This burner is derived from the ALROLL burner family, fitted with an combustion air supply line.Its performances are similar to an ALROLL burner when oxygen is used. Air operationperformances proved to be as good as standard burner when substituted on an existing furnace (seeBougault et al., 1998, 1999).UV Cell and pilot flameWhen the process temperature can be lower than 750°C (EN 746.2, combustion norm in Europe),flame control and holding are required. All the burners mentioned above are optionally fitted with UVcell and pilot burner that are often required in metallurgical processes.Oscillating combustionIn addition to the previously described burners, AIR LIQUIDE offers experience and knowledge aboutthe innovative oscillating combustion technology (OCT). Oscillating combustion (Duchateau et al.,1996, Charon et al., 1996 and Philippe and Grosman, 1996) is a low-cost, patented, low-NOx, highefficiency technology that can be integrated in any combustion system and whose principle is based ona cyclic perturbation of the gas line. In oscillating combustion, the flow of fuel is oscillated around thestoichiometric value. This action produces alternating fuel-rich and fuel-lean zones within the flame,which produces less NOx than stoichiometric conditions.In the oscillating conditions, the flame is much longer, it is highly radiative and it has much lowertemperature peaks. It results in :- significant savings of fuel gas and oxygen (5 to 10%)- lower temperature of walls and vault temperature (-15 to -30°C)- dramatic reduction of NOx emission (down to 50%)Fig. 6 : Oscillating Combustion principle (left) andthermally insulated valves with by-pass piping manifold (right)AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 7/23

RECHERCHE ET DEVELOPPEMENTThis technique has been demonstrated on a glass furnace as part of a R&D Program developed at theAir Liquide Chicago Research Center and at the Institute of Gas Technology (IGT, now GTI). In1999, a field demonstration was carried out by Air Liquide, with DOE funding, in a borosilicate glassmelting furnace at the Johns Manville Cleburne plant.This technique is now available for steel reheating furnace, mainly for NO x reduction. OCTtechnology is designed to be easily retrofitted on existing combustion systems, including those firedwith ambient air, preheated air, enriched air, and oxygen.CombinationAll these technologies can be combinated in the same industrial heating equipment to achieve the bestperformance both for each zone of the furnace and for the overall process, while maintaining highstandard of energy saving and environment protection.2.2. ModelingAIR LIQUIDE has developed advanced modeling capabilities dedicated to steel reheating furnaces. Asmodeling of combustion chamber or furnace have been available for many years at AIR LIQUIDE, thekey issue was to take in account the specificity of reheat furnaces, which is the steel product. Inparticular, phenomena as scaling, decarburization, thermal homogeneity had to be calculated toestimate the influence of the ALROLL technology on existing reheat furnaces.0D modeling is available and will not be presented here. NO x modeling in reheat furnaces is alsoavailable and will not be presented here. Focus is made on 1D and 3D modeling and the coupling withthe steel product. See Till et al. (2001), for a review of recent AIR LIQUIDE modeling activities.Air Liquide “REHEAT 1D” softwareThe REHEAT1D software calculates the furnace atmosphere and temperature profiles, and thencouple those results with thermal conduction in the billet section. To insure good description of theradiation, a 2D description is used (fitted to oxycombustion, see Soufiani and Djavdan, 1994). In thebillet, the thermal conduction is solved in 2D. The input of the model are the power location (air oroxy) , the furnace geometry, the wall composition, the billet size and inlet temperature. The outputsare the temperature profile at any location in the billet, in the furnace.When adding a new oxyfuel zone in a furnace, the thermal behavior of the furnace will be modified. Itis necessary to modify the furnace operation. As the furnaces are usually operated using a controlsoftware (“Level 1” or “Level 2”), information concerning the oxyfuel zone is to be provided to thesoftware. REHEAT1D outputs are very effective to give a first input of what the temperaturemodifications can be inside the furnace allowing the estimation of the first temperature setpoints whenthe oxyfuel burners are firing.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 8/23

RECHERCHE ET DEVELOPPEMENTFig. 7 : Principle of 1D and 3D modelingTo illustrate the performance and the reliability of the model, a comparison between fieldmeasurement using an optical pyrometer and the model results is shown below. The billet uppersurface temperature as well as the vault temperature are predicted and compared. The graph shows thatthe temperatures are well predicted and confirms that the model can be used as a first input withsignificant confidence.Fig. 8 : Comparison between 1D modeling and field measurement3D Simulation : ATHENA / SYSWELDAir Liquide 3D Model for Reheating application is based on the coupling between ATHENA, 3Dmodeling of turbulent combustion and fluid mechanics and SYSWELD, calculating the thermalconduction and mechanical deformation. The coupling principle is illustrated on the figure below.ATHENA is a in-house software developed by Air Liquide (initially for glass applications) whileSYSWELD is a commercial product used in particular by European steelmakers. ATHENA givesAFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 9/23

RECHERCHE ET DEVELOPPEMENTthe heat flux to the product and SYSWELD, using this flux as an initial condition, calculates thenew temperature profiles until converging is obtained.The coupling allows products (billet/slab) reheated on the hearth or on walking beams in the center ofthe furnace and is described on Fig. 9.Fig. 9 : Coupling between ATHENA and SYSWELDValidationTo illustrate the performance and the reliability of the 3D model, another between field measurementand the model results is performed. Field measurements are now performed on a slab reheat furnace,using an equipped slab. The slab upper surface temperature as well as lower surface temperature arepredicted and compared. Fig. 10 shows that the temperatures are well predicted and confirms that the3D model can be used with strong confidence.Temperature in K16001500140013001200110010009008007006005004003002001000Model data Experimental data0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34Position inside furnace in meterFig. 10 : 3D Modeling , validation with field data from a slab reheat furnaceAFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 10/23

RECHERCHE ET DEVELOPPEMENT2.3. Fine tuning and specific issuesThe 3D modeling of the furnace and the product is used only for advanced issues. In these issues, thethird dimension is needed to study correctly the identified technical limitations of the furnace.Examples are derived below from industrial installations of the ALROLL technology.Burner locationThe 3D-modeling tool can provide important information to optimize the burner configuration. Theburners location, nominal power can be adjusted to the particular flow pattern of the furnace.ALROLL burners have been characterized using the ALICE 2 MW pilot furnace located at the AIRLIQUIDE Claude Delorme Research Center, France (Dugué et al., 2000). Modeling of correspondingoxygen flames is then fitted and available for ATHENA modeling. Fig. 11 is an example of slabwalking beam furnace modeling.Fig. 11 : Example of 3D modeling of a complex geometry (slab reheat furnace)Temperature homogeneityThe thermal homogeneity of the product at the furnace exit is a major concern for steelmakers.Important temperature difference (top/bottom, top/heart) can lead to defects in final product quality, torolling mill stoppage due to cold spot incidents, billet bending etc. This issue can be studied using 3DAFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 11/23

RECHERCHE ET DEVELOPPEMENTmodeling. Fig. 12 shows typical result showing an improvement of billet temperature homogeneitywith ALROLL technology (see Delabroy et al., 2001).Air CombustionOxycombustionT millingT milling∆T sup/inf ~50°C∆T sup/inf ~25°CFig. 12 : Improvement of temperature homogeneity of a billet, using oxycombustion (constant pullrate) [color scale are identical for both pictures – 3D Modeling]Mechanical deformationThe maximum deformation is achieved at the maximum temperature difference. On Fig. 13, themaximum deformation and temperature difference are correlated and occurs around x=0.65. Reducingand controlling the temperature difference along the product heating history will directly control theproduct mechanical deformation and quality.1,00,04∆T / Tmax0,90,80,70,60,50,40,30,20,1DT / Tmaxvertical deformation0,030,030,020,020,010,01mechanical deformation (m)0,00,000,0 0,2 0,4 0,6 0,8 1,0X / XrefFig. 13 : Relation between temperature difference and mechanical deformation (3D modeling)AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 12/23

RECHERCHE ET DEVELOPPEMENTTypical results of the 3D modeling are shown on Fig. 14. The mechanical deformation and thetemperature of the product can be visualized on the same graph.Productway of travelFig. 14 : mechanical deformation of the billetScaling and decarburisationScaling thickness as well as decarburization thickness are also available as a post-processingcalculation. Tests have been performed on a thermobalance to select and adjust the scaling model. Theatmosphere tested were not only air combustion atmosphere, but also mixed atmosphere resulting fromALROLL implementation on a reheat furnaces.2.4. Process ControlIntegration of ALROLL control system on existing customer automation architectureOne important request from customers is to integrate information from new sensors values as well asburners control in their existing automation system with least amount of changes. AIR LIQUIDEpropose several solutions to integrate the additional sensors and actuators installed on the process,including thermocouples in the oxyfiring zone, controls for the new burners, etc. A new controlcabinet, including a PLC, is generally dedicated to the ALROLL system. It contains all logic andregulation parameters to safely operate ALROLL. One innovative architecture is based on a localEthernet network. This last architecture is cost-effective (20 to 40% reduction from a standard wiresolution), can easily be extended just by plugging-in new sensors on the Ethernet loop and opens theway to smart sensors / actuators. The supervision interface (HMI) can be independent or included inthe customer existing interface.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 13/23

RECHERCHE ET DEVELOPPEMENTModification of existing Level 2 to take into account O 2 specificityTo control the heating curve for the products in reheat furnace, furnace engineering companies havedeveloped so-called “Level 2” controllers. The purpose of these controllers is to maintain an optimizedheating profile inside the furnace, whatever the heating history of the product. A key point of suchmodels is the representation of the thermal transfer inside the furnace. Thermal exchange occursbetween the vault, the product and the exhausts flowing in between – generated from air combustion.With the implementation of oxyfuel burners, ALROLL will modify the fumes composition and theassociated radiative properties. If the “Level 2”models are used with model coefficients that areoptimized for air combustion, the temperature predictions will be incorrect and may lead to someproduction troubles.AIR LIQUIDE has acquired the specific knowledge of thermal transfer in oxycombustion. Moreover,all the burners mentioned in the ALROLL technology have been extensively characterized (in termsof flame contour, radiative heat transfer, etc.). Consequently, whatever the Level 2 controller is, itsadaptation to ALROLL process is made possible, allowing optimized performances of the reheatfurnace.3. FREQUENTLY ASKED QUESTIONS ABOUT ALROLL3.1. Why the quality of the product is not changed (e.g. the scaling) ?The two main concerns about the product quality are the product scaling and the product thermalhomogeneity.ScalingScaling is the oxidation layer that is formed at the product surface. Representing a loss for thesteelmaker, scaling has to be kept as low as possible during the reheating process. Visual observationof the billets at the discharge zone of L.M.E. and SAM (see § 4.1 and 4.2) shows that less scaling ispresent. After more than a year of successful operation, L.M.E. operators have stated the followingconclusion: the scaling is not increased with oxycombusion. This can be explained: first, the furnacetemperature in the oxyfiring zone is lower than 900°C. It is known that in such conditions (even underoxidizing conditions) no scaling is formed (see Ormerod et al., 1997). In addition, when the oxygasburners are fired, the temperature setpoint of the soak zone is lowered from a few tens of degrees. Thepower and the temperature are therefore reduced in the particular zone where the maximum of thescaling is formed. Finally, when oxycombustion is coupled with a pull rate increase, the billet aredischarged faster, inducing a reduced time in the high temperature zones of the furnace, leading to theobserved decrease in scaling quantity.Additional advantages can be found in the literature. For instance, Wei F. shows that the scaling donein primary oxidizing conditions is more porous and is more easily removed.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 14/23

RECHERCHE ET DEVELOPPEMENTThermal homogeneityA good tracer for the thermal homogeneity of the product is the intensity of the first mill. A highintensity shows that some colder spots remain in the hearth of the billet, while lower intensity showsthat the product is homogeneous. At L.M.E., the maximum 1 st rolling or blooming intensity aredecreased by 10% when firing the oxyfuel burners, demonstrating an improvement in the thermalhomogeneity of the billets, even for periods when the nominal productivity is increased to themaximum output achievable by the mill (110 t/h).Modeling answersAnother way to address the product quality this phenomena is the REHEAT1D software. Thecalculation has been based on L.M.E. – like billet furnace (see Fig. 17). Three operating point areconsidered: a reference point (Case 1 : air combustion, 92 t/h), a conversion point (Case 2 : withALROLL, 92 t/h) and a productivity increase point (Case 3 : with ALROLL , 103 t/h). Table 1represents the main input / output of the case study.Case descriptionCase 1 Case 2 Case 3Air combustion With ALROLL With ALROLLReference case Conversion Productivity increasePull rate 92 t/h 92 t/h 103 t/hAir Firing power 31 MW 26.5 MW 31 MWOxy-burner power - 4 MW 4 MWFinal core T° 1114 °C 1115°C 1103 °CFinal ∆T 79°C 42°C 82°CTable 1 : Case studyScaling and thermal homogeneity of the product are displayed on Fig. 15. Concerning the scaling (Fig.15, left), it is shown that as the scaling only slightly increasing when the output is kept constant withALROLL (Case 2), the scaling is identical that the air reference when the production is increased(Case 3). Concerning the thermal homogeneity (Fig. 15, right), it is shown that using ALROLL, themaximum temperature difference and the final temperature difference are decreased for the sameproduction rate (Case 2). When the output is increased to 103 t/h (Case 3), the maximum temperaturedifference as well as the final temperature difference are identical to the one for the reference case.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 15/23

RECHERCHE ET DEVELOPPEMENTFig. 15 : Scaling (left) and thermal homogeneity (right) of the case study3.2. Why NO x can be reduced using ALROLL ?Depending on country regulations, the NO x level may be measured as the specific NO x emission (kgNO x emitted per ton of reheated steel) or raw NO x emission (hourly NO x emission, in kg NO x / hr orkg NO x / Power unit). When speaking about NO x reduction, it is therefore important to precise whatkind of emission is concerned.Specific NO x reductionResults obtained at L.M.E. and detailed in Delabroy et al. (2000, 2001) are synthesized here. Themeasurements are made in the exhaust flue gases. The probe is located in the exhaust duct, right afterthe heat exchanger (recuperator). A specificity of the reheat furnace is the existence of importantamount of air leaks. The billet opening is one reason explaining this fact. For instance, measurementperformed at L.M.E. using helium (He) tracing shows typically 5000 Nm 3 /h of air leaks. This order ofmagnitude is confirmed anyway by the percentage of oxygen measured in the dry flue gases (5 to 6%O 2 ) at the probe location.Natural gas(Air Comb. Zone)Output(t/h)NOX mg/m3@8%O2NOX kg/ton% NOxReductionReference(Air combustion)2 oxyfuel burners(+2 MW)67% 77 172 0,07067% 93 182 0,060 -13%Natural gas(Air Comb. Zone)Output(t/h)NOX mg/m3@8%O2NOX kg/ton% NOxReductionReference(Air combustion)4 oxyfuel burners(+4 MW)96% 85 172 0,08997% 110 203 0,082 -8%Table 2 : Global NOx results showing NOx reduction by oxycombustionAFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 16/23

RECHERCHE ET DEVELOPPEMENTThe first remark about Table 2 concerns the “Reference” case (with only air combustion). The furnaceemission is 0.07 kg NO x per ton of reheated steel (corresponding to 172 mg/m 3 at 8% O 2 ) which iswell below the European regulations standards. The L.M.E. furnace is a well-tuned furnace concerningthe pollutant emissions. Despite this challenge, specific NOx can be achieved using oxycombustion.Two configurations are detailed on Table 2. On the top table, the furnace is fired at 67% of itsmaximum power. Then two (out of four) oxyfuel burners are started. A 20% increase in output isobserved and a 15% reduction of specific NO x emitted is observed. The second configuration (bottomtable) corresponds to the maximum furnace output and firing power (31 MW). Four oxyfuel burnersare then started, with another 20% production increase, together with an 8% reduction in NO xemission.The results shown in Table 2 confirm that ALROLL is a tool for specific NO x reduction as well asproductivity increase.Hourly NO x reductionIn addition with the hourly NO x reduction, the influence of the conversion percentage tooxycombustion on NO x reduction is addressed. At L.M.E. and SAM Neuves Maisons, a typical 10-15% of the total power is actually fired using oxyfuel burners (see § 4.1 and 4.2). This percentagecould clearly be increased and it is important to show its influence on the hourly NO x reduction.To have a meaningful comparison, emission data are needed at the same furnace pullrate. While suchmeasurements are not available at L.M.E. or SAM, some simulations have been performed for aconstant pull rate of 110 t/h, based on L.M.E. case study.With preheated air combustion, (even if additional air firing power is not available neither feasible atL.M.E.) the firing power requirement to achieve a 110 t/h pull rate is 41 MW (calculated using the AirLiquide 0D balance software). Table 3 shows that the NO x produced per MW are constant. A 41 MWpower would then give a 9.6 kg NO x / h for 110 t/h with air preheated combustion. At 110 t/h withALROLL, the field measurement is 9.0 kg NO x / h.. A potential 6% reduction is then achievedcompared to the air reference.Pull rate NO x measurement (mg/m 3 ) Power input Fumes volume NO x(t/h) (before dilution at 8% O 2 ) (MW) (Nm 3 /h) (kg / MW)77 199 23 27000 0,2385 199 33 38000 0,23Table 3 : NOx measurement at L.M.E. (the fumes volume has been estimated with the %O 2measured)In reality, the achieved reduction should even be greater: in the estimation of 41 MW, the hypothesisof constant efficiency has been made. It is known that the efficiency of a furnace is decreasing whenthe furnace is pushed well over its nominal power (e.g. L.M.E furnace at 110 t/h with air combustion).Also, this number would obviously increase if the percentage of conversion to oxycombustion isincreasing. To demonstrate that trend, an additional simulation is done, using the same hypothesis asAFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 17/23

NOx ( kg / h )RECHERCHE ET DEVELOPPEMENTabove (Fig. 5). Concerning the data for NO x generated by the oxyburners, we use the one measuredand presented in Table 2.Fig. 16 shows measurement data in plain symbols. Increasing the percentage of conversion to oxyfuelcombustion leads to additional reduction of hourly NO x (expressed in kg/h).10,09,08,07,06,0Measurement, aero combustionSimulation, aero combustionMeasurement, with oxyfuel burnersSimulation, with oxyfuel burnersair combustion10% O 2 conversionIncrease withoxyfuel conversion40% O 2 conversion5,04,03,060 70 80 90 100 110 120Pull rate (t/h)Fig. 16 : Reduction of hourly NOx increases with %O 2 conversionALROLL can reduce the NO x emission of a reheat furnace. If NO x emission is the major issue forthe concerned furnace, Oscillating Combustion Techniques (OCT) is available and has been alreadypresented. Oscillating combustion can be added to ALROLL oxy-burners installation and can beretrofitted to the exisiting air firing zones or can be added to the new oxy-firing zones.NO x reduction with Oscillating CombustionFigures and estimation of Fig. 16 opens the way for NO x reduction with ALROLL. Anothertechnical solution, already discussed in § 2.1, is the Oscillating Combustion Technology (OCT). WithOCT, 30 to 50% reduction in NO x is expected on both air burners and/or oxyburners. Even thetoughest NO x emission level can be achieved using Oscillating Combustion Technology on reheatfurnaces.3.3. Why ALROLL is cost effective ?ALROLL is based on the principle of improving heat transfer in an existing chamber withoutstructure and refractory modifications. In most of cases, investment can be as low as a 1/10 th oftraditional air revamping. It mostly consists in adding a new zone of oxyfuel burner zone orsubstituting an air burner zone. Process PLC and supervision are upgraded for the control and theregulation of this new burner zone.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 18/23

RECHERCHE ET DEVELOPPEMENTWhen running the process, fuel gas saving usually pays the cost of oxygen. Moreover, it is also highlyflexible because the application can be switched on or off upon process need. It doesn’t induce any useof natural gas or oxygen when it is not strictly necessary.In the previous industrial references, economics have shown the following facts :- the specific cost of utilities (NG, O2, aso) per ton of good billet remains stable- the reheating cost (fixed and proportional costs) are reduced by 10%- the total cost of rolling can be reduced by as much as 20% of standard operation withoutALROLL.The low investment of ALROLL has a dramatic effect on fixed costs. On one hand, the fixed costare almost reduced as much as the production is increased. On the other hand, when not usingALROLL because of production low heating requirement, regular operation doesn’t bear anyadditional investment.4. EXAMPLES AND REALIZATION4.1. Laminés Marchands Européens (L.M.E.)Laminés Marchands Européens is a French steelmaker of merchant bars. L.M.E. is a subsidiary ofBeltrame group from Italy and ARBED from Luxembourg. It produces 1.000.000 tons per year of rawsteel. It has three main works located in Trith Saint Léger (Northern France), Le Ruau (Belgium) andEsch-Schifflange (Luxembourg).The rolling capacity in Trith Saint Léger is currently 350 000 tons per year. L.M.E. is committed toincrease the annual output of this workshop up to 600.000 tons per year. The mill facility is currentlyproducing a large range of merchant steel bars, with a large range of final sections. In these conditions,it is especially difficult to have the reheating furnace and the rolling section perfectly adapted.The L.M.E. has favored the following strategy:- the construction of a new mill for low section production : + 150.000 t/y- the increase existing capacity for large section + 100.000 t/yL.M.E. have been operating ALROLL on the 92 t/h reheat furnace since January 2000. A 4-MWALROLL combustion zone is installed just before the preheating zone (see Fig. 17). The addition ofoxyfuel burners (representing up to 12% of the total furnace power) leads to a 20% productivityincrease (110 t/h). Another major consequence is the reduction of the production cost for L.M.E.Considering the operating costs of the reheating furnace plus the operating costs of the rolling mill,oxyfuel firing enables a 12% reduction of the total price per ton of finished steel bars (see Delabroy etal. 2000).AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 19/23

RECHERCHE ET DEVELOPPEMENTHeat exchangersoak main preheatAmbient airPreheated airrecuperationBillet outletOxyfiredzone2 x 2 ALROLL TM burners(1 MW nominal)Fig. 17 : ALROLL installation at L.M.E.Billet inlet4.2. SAM Neuves MaisonsSAM Neuves Maisons, RIVA group, is a producer of construction steel wire in Eastern France.Because of the proximity of the melting plant and the hot-rolling plant, billets are hot-charged in thereheating furnace. It enables a maximum production of 150 metric tons per hour. When the EAF andthe rolling mill do not have coordinated productions (e.g.: EAF maintenance, product size or grade,shutdown) products are cold-charged. Thus, the maximum production can be as low as 136 t/h becauseof the additional heat to provide to the product.A 6-MW ALROLL combustion zone is now installed at the charging section of the furnace. Whenthe temperature of the charged products lowers, oxyfuel zone starts in order to maintain the level ofproduction. According to the charging temperature, the temperature or power set point is adjusted bythe operator on a dedicated panel in the control room.Since October 2000, the maximum production of the rolling mill has been 150 t/h, whatever theupstream conditions are at the melting shop.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 20/23

RECHERCHE ET DEVELOPPEMENTTop soakingzoneTop heatingzoneTop preaheatingZone2 x 2 ALROLL TM FC burners(1 MW nominal)Dis c h arg ezo neChargezo neproductsBottomzoneFume exhaustFig. 18 : ALROLL installation at SAM Neuves Maisons4.3. SOLLAC StrasbourgThe site of SOLLAC Strasbourg is part of the USINOR Group and it is especially involved in thefinishing operations of carbon steel strip coils. SOLLAC Strasbourg site has a painting line, agalvanizing line and a large packaging facility.The galvanizing line is producing a wide range of production with thickness and width. The output ofthe line is mostly determined by the product dimensions and grade. Production can be limited byvarious factors, such as heating power, cooling capacity, maximum speed or other mechanicallimitation. These limitations never occur simultaneously. Each limitation is generally connected withone of the product characteristic. Therefore, SOLLAC is committed in a program of upgrading parts ofthe line to achieve a global increase of production and performances. Such program involvesincreasing temporarly the heating capacity at the entry section, so-called direct-fired non-ox zone. Itconsists of chamber of hot-air burners that are fired over the incoming strip, in order to bring itstemperature up to 700°C.A 1-MW ALROLL technology burner zone has been installed upstream of existing burners whenconsidering the strip way of travel. When high-thickness or large strips are processed, the zoneautomatically switches on and brings the additional heat that is necessary to the strip. The oxyfueltechnology enables an heating section in the non-ox chamber without affecting significantly the fluegastemperature out the chamber. Such a zone would have been impossible to run through aircombustion burners.SOLLAC is operating this application on regular operation since March 2001 and has already reacheda 7-to-10-% production increase for the limiting products. Additional gains are expected when coolingsystem is upgraded this year.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 21/23

RECHERCHE ET DEVELOPPEMENTTowards recuperatorRecuperatortemperaturealarmPost-CombustionZone temperature alarm710° 790°460°Non-ox(Direct-fired)«Annealing cell »Electrically heatedJet-cooler2 x 2 ALROLL TM burners(200 kW nominal)Fig. 19 : ALROLL installation at SOLLAC StrasbourgGalvaning bathAknowledgmentsThe authors wish to thanks our partners companies and especially the following people for their strongtechnical involvement in the ALROLL projects mentioned in this paper.: M. Barbotin and Lebrun atL.M.E. (Trith St Léger, France), M. Pochopien, Ruzin and Sauvageot at SAM (Neuves Maisons,France) and M. Gauvenet and Haaser at SOLLAC (Strasbourg, France)M. F. Delacroix from ADEME (France) which supported the ALROLL installation at L.M.E. is alsogratefully acknowledged.AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 22/23

RECHERCHE ET DEVELOPPEMENTReferencesBougault T., Desbrosses S., Slonski Ph., Crespo D., Pouchelet JP, Borissoff T. and Le Gouefflec G.(1999) “Increased Performance of annealing-pickling lines for stainless strip by oxycombustion”American Iron and Steel Engineers Annual Conference, Cleveland, USABougault T., Desbrosses S., Slonski Ph., Crespo D., Pouchelet JP, Borissoff T. and Le Gouefflec G(1998) “Improvement in the performance of an annealing-pickling lines for stainless strip byoxycombustion”, International Steel Making Days, ATS, Paris, FranceCharon O. , Jouvaud D. and Genies B (1993)., “Pulsated O2/Fuel Flame as a new technique for LowNOx Emission” Combustion Sci. and Tech., vol 90, pp 1-12.Delabroy O., Le Gouefflec G., Lebrun C., Barbotin A. and Cervi R., (2000) “PerformancesEnhancement of reheating furnaces using oxycombustion”, American Iron and Steel Engineers AnnualConference, Chicago, USADelabroy O., Le Gouefflec G., Lebrun C., Barbotin A. and Cervi R., (2001) “Oxycombustion for NO xreduction in Reheat Furnace”, NOxCONF, Paris, FranceDugué J., Von Drasek W, Samaniego JM, Charon O. and Oguro T. (1998) “Advanced combustionfacilities and diagnostics” AFRC / JFRC International Symposium, Maui, Hawaii.Duchateau, E., Philippe, L., Grosman, R., Wagner, J., Abbasi, H. and Rue, D., (1996) “Oxy-GasOscillating Combustion Offers Lower NO x and Higher Efficiency”, 1996 AFRC Spring Member’sTechnical Meeting, Orlando, FLOrmerod R.C., Becker H.A., Grandmaison E.W., Pollard A. and Sobiesak A., (1997) “Effects ofprocess variables on scale formation in steel reheating”, The Canadian Journal of ChemicalEngineering, Vol 75 pp 402-413, 1997Soufiani A., Djavdan, E. (1994) “A comparison between weighted sum of gray gases and statisticalnarrow-band radiation models for combustion application”, Combust. and Flame, Vol. 97, pp 240-250Philippe L. and Grosman R. (1996) “Oscillating Combustion Meets Tough Challenges. “ Natural GasApplications in IndustryTill M., Marin O., Louédin O. and Labégorre B. (2001) “Numerical simulation of industrialprocesses, coupling combustion chamber and load calculation”, AFRC / JFRC / IEA JointInternational Combustion Symposium, Kauai, Hawaii.Wei, F. A study of oxidation behaviour of various steels in simulated reheating furnace atmosphere,CSC China Steel Tec. Report, N°2, pp 8-14AFRC / JFRC / IEA 2001 Joint International Combustion SymposiumSeptember 9-12, Kauai, Hawaii 23/23