Investigation of High-Rate and Pre-Heated Natural Gas Injection in The Blast Furnace

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

AISTech 2019 — Proceedings of the Iron & Steel Technology Conference

6–9 May 2019, Pittsburgh, Pa., USA


DOI 10.1000.377.040

Investigation of High-Rate and Pre-Heated Natural Gas Injection in the Blast Furnace

Tyamo Okosun1, Samuel Nielson1, John D’Alessio2, Mitchell Klaas2, Stuart J. Street3, Chenn Q. Zhou1
1
Center for Innovation through Visualization and Simulation (CIVS), Purdue University Northwest,
2200 169th Street, Hammond, IN 46323, USA
Phone: 1-219-989-2665
Email: czhou@pnw.edu
2
Stelco Lake Erie Works,
2330 Haldimand Road #3
Nanticoke, ON N0A 1L0
3
AK Steel Dearborn Works, 14661 Rotunda Dr., Dearborn, MI 48120

Keywords: Blast Furnace, Pulverized Coal, Natural Gas, Fuel Injection, Computational Fluid Dynamics, Modeling, Fluid
Dynamics, Combustion, RAFT

INTRODUCTION
The efficiency of the blast furnace (BF) process is greatly impacted by choices made from an operational standpoint, with
many factors influencing furnace production rate, fuel consumption, and stability. Ironmaking plants across the world
continue to endeavor to find new methods of decreasing operational expenditures and enhancing furnace performance. With
most blast furnaces operating using some form of auxiliary fuel injection, one area of focus for parameter adjustment in the
BF is related to the properties and injection rates of auxiliary fuels. In North America (NA) in particular, natural gas (NG)
has become popular as a tuyere-level injected fuel, in addition to, and sometimes in the place of, pulverized coal (PC)
injection. As the cost of NG injection is often lower than that of PCI (both present benefits in terms of coke savings) in North
America, methods of performing high-rate NG injection are of particular interest.
As a gas, NG injection presents additional benefits compared to PCI, in that fines buildup from ash or unburned injected fuels
will not occur. Unfortunately, the injection of NG in the blast furnace tuyere can also have adverse and limiting effects upon
furnace operation. Chief among these being the production of CO2 and H2O in the tuyere jet from NG combustion. These
gases enter the raceway region and take part in endothermic reactions with carbon, leading to a drop in raceway gas
temperatures. While the exact lower limit of raceway temperature, or Raceway Adiabatic Flame Temperature (RAFT), is
somewhat unclear, recent research has documented that most NA BFs operate in a range of flame temperatures above 2,020
K1,2. Generally, this quenching effect can be counteracted by increasing the amount of oxygen in the hot blast, however, this
will invariable result in decreasing furnace top temperatures and the potential for condensation. Together these two
constraints (RAFT and oxygen) form the boundaries of a stable operation window for BFs using NG injection, with a
theoretical upper limit of NG injection occurring somewhere around 150-160 kg/THM1,2.
Attempts to increase NG levels further without also adjusting other furnace operating parameters may introduce operational
stability concerns. With that in mind, a potential method to expand this operating window and allow for higher levels of NG
injection (and subsequently increased coke replacement with injected fuels) would be to pre-heat NG before injection into the
BF tuyere. Theoretically, this approach could provide supplemental heat energy to operate the blast furnace raceway at higher
injection levels without reducing the RAFT below safe values for stable operation. Additionally, higher temperature NG
would be injected at an increased velocity, potentially improving mixing and enhancing combustion.
This concept has been touched upon in limited public domain literature, but little has been put forth in the way of trials,
detailed studies, or implementation2. One study by Feshchenko et al. explored the benefits and implementation of a NG pre-
heating system designed for use at two blast furnaces (OJSC LMZ “Svobodny Sokol” and OJSC “Severstal”)3. This research
proposed that pre-heating injected NG could provide several benefits to the BF, including additional sensible heat, increased
NG combustion efficiency, increased hydrogen utilized (due to more uniform distribution in the BF through increased NG

© 2019 by the Association for Iron & Steel Technology. 383


velocity), among others. Mathematical modeling was performed on the tuyere region of the BF to examine the combustion of
NG, indicating that an increase in NG temperature from 0 °C to 300 °C reduced CH4 content at the tuyere outlet from 3.1% to
2.2%. It is indicated that this phenomena likely occurs due to the drop in NG density at increased temperature, leading to
higher NG speeds and improved mixing with hot blast (leading to more rapid combustion). Operating conditions were
recorded for both furnaces before and after implementation of this technique, resulting in the changes listed in Table I below.

Table I. Impacts of pre-heated NG injection at OJSC LMZ “Svobodny Sokol” and OJSC “Severstal”.
Svobodny Sokol Severstal
Baseline Pre-heat Baseline Pre-heat
Production (t/day) 1,334 1,391 3,303 3,207
NG Temp. (°C) 30 200 30 105
Blast Temp. (°C) 1,050 1,088 1,181 1,182
Coke rate (kg/thm) 497 480.4 420.2 411.1
Production % change + 4.15% - 2.9%
Coke rate % change - 3.3% - 2.2%

At Svobodny Sokol, a coke rate reduction of 16.6 kg/thm (3.3%) and a production rate increase of 57 tons/day (4.15%) were
observed at a NG temperature increase of 170 °C. At Severstal, a coke rate reduction of 9.1 kg/thm (2.2%) was observed at
lower production rates, corresponding to a NG temperature increase of 75 °C. While there is some variation in blast
temperature in the Svobodny Sokol furnace, the research did lay claim to observable benefits to pre-heated NG injection.
Additionally, it was noted that pre-heating altered the coke replacement ratio from 0.8 kgcoke/m3ng (1.2:1 kgcoke/kgng) to 0.95
kgcoke/m3ng (1.42:1 kgcoke/kgng).
Obviously, implementation of NG pre-heating in the field would require some capital expenditure, and as such, specific,
detailed studies are required to determine the potential benefits for operation under typical conditions. Operator experience
and industrial know-how can provide intuition for the feasibility of such new operational practices, however, to gain insight
into the intricacies of how such operating conditions can impact the BF, targeted research is required. Simulation of the BF
using computational fluid dynamics (CFD) has become one of the leading initial steps to exploring modifications and
optimizations for operation. CFD remains less expensive and faster than comparable experimentation on physical systems,
and computational modeling allows for a wider variety of conditions to be tested.
A number of researchers have tackled the complex task of exploring phenomena within the BF using CFD techniques, with
simulations developed ranging from greatly simplified scenarios to full-scale multi-physics models of the entire furnace4-12.
The Center for Innovation through Visualization and Simulation (CIVS) at Purdue University Northwest has performed a
range of studies on BF performance using both commercial and in-house CFD solvers. Studies have focused on fuel
combustion, burden distribution impacts, injection lance positioning, and more13-19. This paper will cover the first stages of
research on pre-heated NG injection in a North American BF and a number of the impacts such operation has upon
phenomena within the furnace.

NUMERICAL METHODOLOGY
Modeling of the blast furnace is performed using a multi-component methodology, targeted at specific parts of the furnace
system. It is worth noting here that all components of this CFD model and methodology have been previously validated
through focused studies and applied research on industry blast furnaces of varying sizes13-19. CFD modeling is used to
simulate the BF blowpipe, tuyere, injection lances, raceway, and shaft. The simulations are broken up into two primary
reaction zones, namely the raceway and shaft regions, each with their own computational model to calculate flow, heat
transfer, and reaction phenomena.
The raceway region is itself handled with component approaches. Within the blowpipe and tuyere region, phenomena are
calculated using a comprehensive Eulerian-based solver in the commercial CFD software ANSYS Fluent®. Fluid flow and
heat transfer are calculated based on the steady-state Navier-Stokes equations with the Semi-Implicit Method for Pressure
Linked Equations (SIMPLE) discretization scheme. Turbulence is handled using the k-ε turbulence model, with standard wall
functions representing behavior in the fluid boundary layers20. This model is widely implemented for its computational
efficiency and relative accuracy when properly applied to various geometries and flow conditions. Additionally, as the
furnace is radially symmetric, only a single tuyere is simulated, allowing for more rapid computation, as one slice can
represent conditions within the entire BF. Gas phase combustion is handled with the continuum method Eddy-Dissipation-
Concept (EDC) model21. Boundary conditions are defined based on industry recommendations and selected operating
conditions for any given case.

384 © 2019 by the Association for Iron & Steel Technology.


Beyond the tuyere tip, simulation of the raceway itself is handled with a coupled two-step approach. First, the raceway
envelope size and shape are calculated with a transient dual-phase Eulerian model. Coke, in this simulation, is treated as a
granular continuous phase, with properties representative of the solid bed. The model is again solved using ANSYS Fluent®,
and the results are used to generate a fixed distribution of coke in the raceway region. This fixed porosity distribution is then
used as the geometry for the combustion and reaction simulations of gas and coke in the second step of the raceway
simulation.
Flow, heat transfer, combustion, and other reactions within the raceway are calculated using a steady-state Eulerian multi-
fluid model built in an in-house CFD code. The details of this solver have been covered in previous publications13,18,22. Flow
inlet conditions (hot blast velocity, temperature, etc. incoming from the tuyere) are mapped directly from the tuyere &
blowpipe simulation and applied to the raceway combustion solver. As previously mentioned, the coke distribution in the
raceway region is held fixed for this solver, with the interior of the raceway as a region of high void fraction. Coke in the
porous media bed can interact with incoming gases in both combustion and other reactions to produce and consume gas
species and heat.
The impact of turbulence on the combustion rates of gaseous species is accounted for though the Eddy Breakup model and
Arrhenius chemical kinetics, with total reaction rates defined as the smaller of the two values23,24. Table II details the kinetic
constants for gaseous CO, H2, CH4, and carbon reactions. Constants are taken from literature as detailed in previous
publications13,18,22,25.

Table II. Reaction kinetics for the raceway combustion model.


Reaction A [1/s] B [m3/(kg*s)] Act. Energy E [J/mol]
N/A
CH4 + 2O2 → CO2 + 2H2O 1.6 x 1010 1.081 x 105
N/A
2CO + O2 → 2CO2 7.0 x 104 6.651 x 104
N/A
2H2 + O2 → 2H2O 5.4 x 102 1.255 x 105
C + O2 → CO2 1.225 x 103 N/A
9.977 x 104
2C + O2 → 2CO 1.813 x 103 N/A
1.089 x 105
C + CO2 → 2CO 7.351 x 103 N/A
1.380 x 105
C + H2O → CO + H2 1.650 x 105 N/A
1.420 x 105

Radiation heat transfer is significant within the raceway region, and as such, attention must be paid to the modeling
techniques used. Radiation fluxes depend on several parameters such as material temperatures and emissivity. Various
radiation models are used in CFD modeling, applying both zone and flux methods26,27. It has been found that flux methods,
such as the Discrete-Ordinates (DO) model, tend to provide relatively accurate solutions relative to the necessary
computational costs. With this in mind, the DO model was used in the in-house raceway CFD solver.
The final stage of the raceway simulation methodology involves an iterative process. Gas-gas and gas-solid reactions in the
raceway alter temperatures and convert solids into the gaseous phase. These phenomena will impact the formation of the
raceway shape, and data on gas density changes due to temperature and gas mass generation from the combustion solver is
fed back into the raceway formation simulation. An updated coke bed porosity is generated, and the process is repeated until
a converged raceway shape is found.
The raceway region simulation results are used to generate inlet conditions for the bottom of the blast furnace shaft model.
This in-house CFD model tackles gas flow, chemical reactions, and burden distribution in the upper regions of the furnace.
Major functions include simulating burden distribution based on a charge matrix, solving gas and burden phase governing
equations, and solving phenomenological CFD sub-models28,30. It is also worth noting that the burden distribution can be
defined based on imported industry profiles, sourced from models calibrated for individual furnaces, for additional accuracy.
In this model, the steady-state ascending gas passes through a fixed porous region that represents a combination of solid and
liquid. The steady-state assumption is valid in this scenario, as the gas takes something on the order of seconds to pass from
the bottom of the furnace to the top, while the descent of the solid burden moves on a time scale of hours. The governing
equations solved for the gas and burden phases are detailed in previous publications and are omitted here for the sake of
brevity15,28-30.
The primary species included in the shaft model are CO, CO2, H2, H2O, and N2 in the gaseous phase, and FexOx Fe, CaO, and
MgO in the solid phase. Chemical reactions are primarily gas-solid, such as the indirect reduction of iron oxide by CO and
H2, coke gasification, and carbonate flux decomposition. Phase changes, including the evaporation of moisture in the charged
burden and the melting of iron and slag, are also considered. Heat and mass transfer is tracked between the gas and solid
phases, and a layered cohesive zone is modeled to capture gas flow through coke slits and the blockage of gas by the low
porosity ore layer. The upper boundary of the cohesive zone is defined by the ore softening temperature, while the lower

© 2019 by the Association for Iron & Steel Technology. 385


boundary is defined by the liquidus temperature. A mixing region between ore and coke layers is also tracked in the model to
take into account the decreased permeability of the burden in that region. This model was based upon one proposed by
Szekely and was validated in research detailed in previous publications28.

MODELING GEOMETRY & BOUNDARY CONDITIONS


Two industrial facilities form the basis for the computational studies in this paper, namely AK Steel’s Dearborn Works (DW)
and Stelco’s Lake Erie Works (LEW). A summary of the computational geometries utilized in this study are presented in this
section. It is worth noting that while simulations of the raceway regions for both furnaces were conducted and initial
simulations of the Stelco LEW BF shaft were undertaken, complex geometry (beyond furnace radius and tuyere diameter) in
those furnace regions are largely dependent on flow conditions and tuyere/blowpipe geometry for the raceway and furnace
burden distribution for the shaft. These specific details will be discussed along with the definition of typical operating
conditions for each furnace’s baseline simulation case.
The AK Steel Dearborn Works BF, located in Dearborn, MI, is a high-productivity furnace modernized and rebuilt in 2007
that currently utilized co-injection of both NG and PC. Current configuration allows for NG injection through a single port in
the upper portion of each tuyere. Injection of PC is performed using a single lance projecting through the side of the blowpipe
assembly, with nitrogen used as the PCI carrier gas. A short section of blowpipe, with surrounding refractory insulation,
leading into a single tuyere was modeled for the DW BF. This geometry is representative of all twenty tuyeres in operation
around the furnace. Figure 1 below provides a top and a side view of the tuyere & blowpipe region geometry as developed in
the commercial CFD solver ANSYS Fluent®.

Figure 1. AK Steel Dearborn Works blowpipe and tuyere geometry top view on section B-B (left) and side view on section
A-A (right).
The Stelco Lake Erie Works BF, located in Nanticoke, ON, is a medium sized BF that uses only NG as an injected fuel for
coke replacement, eschewing PCI. The furnace is currently configured to injected NG through a single lance inserted through
the blowpipe assembly. As with the DW furnace, a geometry consisting of a section of blowpipe leading into a single tuyere
was developed for the LEW BF simulation. Figure 2 below details a side view of the blowpipe & tuyere region, along with a
close-up view of the natural gas injection lance tip.

Figure 2. Stelco Lake Erie Works blowpipe and tuyere geometry side view (left) and detailed view of NG injection lance tip
(right).
In both of these designs, hot blast and injected fuels are supplied by defining input mass flow rates, an operating pressure is
defined based upon standard blast pressure, and heat losses through the blowpipe and tuyere are handled via assumptions. In
the blowpipe region, a thermal resistance is applied to represent the refractory, and a convection boundary condition handles
the heat loss to ambient air. In the tuyere region, a fixed heat flux, defined using a generic assumption based on discussions
with industrial partners, is utilized to represent the heat loss through the tuyere to the cooling water.
For each furnace, additional geometry was generated for the raceway region. This geometry is essentially a pie-shaped slice
of the furnace including a single tuyere and raceway. The boundaries of this domain are relatively simple, with the primary

386 © 2019 by the Association for Iron & Steel Technology.


differentiating factor between cases being the distribution of coke in the raceway region, which is defined by void fraction in
each computational cell within the domain. The size and shape of the raceway for each furnace under the conditions
simulated will be detailed later. Figure 3 shows a general overview of the raceway region geometry for the DW BF.

Figure 3. Structure of 3-D raceway domain geometry for the Dearborn Works BF.
The shaft region of the furnace was also modeled for the Stelco LEW BF. For the purposes of this study, this region is
assumed to be axisymmetric, with varying radii dependent on the furnace design. Geometry variations generated by the
position of the cohesive zone, the thickness of ore and coke layers, and other burdening parameters are taken into account via
varying cell definitions and porosity. Burden distributions for this scenario were provided by Stelco for conditions that
corresponded to the typical operation scenario selected for this study. Heat fluxes through the furnace wall/staves are handled
similarly to the tuyere and blowpipe regions.

RESULTS AND DISCUSSION


AK Steel Dearborn Works Blast Furnace
The first set of analyses conducted in this study were aimed at exploring the impacts of shifting fuel injection rates in a co-
injection furnace (DW). The CFD blast furnace raceway model was used to examine the impacts of natural gas injection rates
on temperatures inside the raceway region. Comparisons were conducted on several cases, including a baseline case, a dual
lance design, two co-injection cases with varying ratios between PC and NG injection, and a case representing a scenario in
which all PC is replaced by NG (potential loss of PCI scenario). Previous studies on this furnace aimed at ascertaining the
influence of various co-injection parameters on injected fuel burnout16-19, while this study focuses primarily on raceway
temperatures.
The baseline operating parameters modeled in this study were a wind rate of 199,020 Nm3/hr, a hot blast temperature (HBT)
of 1,408 K, a hot blast oxygen enrichment of 13.5%, a PCI rate of 85 kg/MTHM, and a natural gas injection rate of 65
kg/MTHM (PCI/natural gas ratio of 1.31:1). The particle sizes of pulverized coal were based on data provided by AK Steel
Dearborn Works. Analyses of phenomena within the tuyere region and related impacts on coal combustion efficiency are
well documented in previous publications16-19. Figure 4 details the distribution of coke bed porosity under baseline operating
conditions. Additionally, it illustrates the selected ‘outer boundary’ of the raceway at a porosity value of 0.7 (70% void
fraction).

© 2019 by the Association for Iron & Steel Technology. 387


Figure 4. Side and top views of raceway porosity distribution (left) and raceway outer boundary (right).
Figure 5 shows a brief example of gas temperature and species distribution within the raceway region on the tuyere center-
plane. Pulverized coal combustion generates high temperature zones in the recirculation areas of the raceway, while the
injection pattern of natural gas leads to variations in temperature and species distribution. This blast furnace utilizes a tuyere
port to inject natural gas from the upper (inner) surface of the tuyere, and it is clear that this injection technique influences the
distribution of gases and temperature within the raceway.

Figure 5. Contours of gas temperature (left), CH4 mass fraction (middle), and H2O mass fraction (right) on the vertical tuyere
center-plane.
The CFD model can be used to predict an average value of raceway gas temperature after all CO2 and H2O has been
converted to CO and H2. This value serves as an analogue to the Raceway Adiabatic Flame Temperature (RAFT), obtained
by taking a mass-weighted average of all gas cells in the raceway region with less than 1% H2O, CO2, and O2 content. Under
these conditions, the predicted RAFT analogue is approximately 2,244 K. This corresponds well with the industrial value of
2,293 K for these conditions (2.1% error).
During operation, economic and operating factors can drive shifts from these baseline operating conditions, and to maintain
stable furnace operation, it is important to understand the potential differences between raceway conditions under these
operating parameters. In addition to the baseline case, two variations in co-injection parameters were simulated. Case 1 had a
PCI rate of 107.5 kg/MTHM and a natural gas injection rate of 47.5 kg/MTHM (PCI/NGI ratio of 2.26:1), and Case 2 had a
PCI rate of 67.5 kg/MHM PC and a natural gas injection rate of 77.5 kg/MTHM (PCI/NGI ratio of 0.87:1).
Case 1, with a higher PCI rate and a reduced natural gas injection rate, results in a RAFT analogue of 2,271 K. Case 2, with
an increased natural gas injection rate and a reduced PCI rate, results in a RAFT analogue of 2,229 K. Overall, this results in
a 1.2% increase in gas temperature in Case 1, and a 0.6% decrease in gas temperature in Case 2 when compared to the
baseline conditions. In general, the gas temperature and species distributions remain similar between these three cases, an
unsurprising result, given the relatively minor changes to operation.
Previous studies conducted on this blast furnace also examined the impacts of modifying the natural gas injection location17.
A dual-lance injection configuration was proposed in an effort to enhance mixing of pulverized coal and natural gas within
the tuyere and raceway. As determined through previous research, this increased mixing serves to enhance injected fuel
burnout within the raceway, potentially helping to reduce char buildup in the coke bed. The RAFT analogue in this dual

388 © 2019 by the Association for Iron & Steel Technology.


lance case was 2,288 K, an increase of roughly 2% from the baseline case. The increase in gas temperatures is primarily due
to the increase in pulverized coal burnout that occurs when using the dual lance injection apparatus.
The final scenario simulated for this study (Case 4) used only NG injection. This case was intended to represent a scenario
under which furnace operators are unable to utilize pulverized coal for a given period of time, for either economic or practical
reasons. To avoid increasing the furnace coke rate, additional NG was used to maintain the total injection rate. In this case,
natural gas was injected through a lance at 150 kg/MTHM, a rate near the upper end of potential operational feasibility.
Operating conditions were otherwise identical to the baseline scenario.
Figure 7 compares contours of gas temperature, CH4 mass fraction, and H2O mass fraction in the raceway region for Case 4.
The differences between the gas temperature distribution in a natural gas-only operating scenario and the standard co-
injection case are immediately apparent. High gas temperatures are primarily found in the tuyere jet, due to heat generated by
combustion of CH4. However, with no pulverized coal to increase temperatures further, the tuyere jet, composed of CO2,
H2O, and N2, contacts the coke bed and immediately begins to lose heat to the endothermic reactions of natural gas
combustion byproducts. In this scenario, the RAFT analogue is significantly lower than any of the previous cases, falling to
approximately 1,994 K, an 11.1 % decrease compared to the baseline.

Figure 7. Contours of gas temp. (left), CO2 mass fraction (middle), and H2O mass fraction (right) for Case 4.
Viewed together, these cases provide a unique view of the conditions within the blast furnace raceway region, as well as data
corresponding to the general quenching effect of natural gas upon the flame temperature. The simulations in this study clearly
illustrate the impacts of increased natural gas injection rates upon raceway gas temperatures. As shown in Figure 8,
increasing the injection rate of natural gas can significantly reduce gas temperatures, leading to a unique environment in the
blast furnace that must be carefully monitored by operators.

2400
2300
2200
Gas Temperature (K)

2100
2000
1900
1800
1700
1600
1500
1400
85-65 (PCI-NG)
0-150 (PCI-NG) 67.5-77.5 (PCI-NG) 85-65 (PCI-NG) 107.5-47.5 (PCI-NG)
kg/MTHM - Dual
kg/MTHM (Case 4) kg/MTHM (Case 2) kg/MTHM (Baseline) kg/MTHM (Case 1)
Lance (Case 3)
RAFT 1994 2229 2244 2271 2288

Figure 8. RAFT analogue for each the baseline case and Cases 1-4.
It is worth noting here that modeling for Case 4 predicts a RAFT analogue below the low-end RAFT levels mentioned earlier
(2,020 K), placing these conditions at the onset of potential instability. Stable plant operation was possible in similar
operating scenarios with lower production rates, but the impacts of continued operation in this vein for long periods is
currently unknown. With this phenomena confirmed in the computational model, the next step was to initiate an exploration
of NG pre-heating in the blast furnace. As previously mentioned, this approach could theoretically provide supplemental heat

© 2019 by the Association for Iron & Steel Technology. 389


energy to allow for operation of the blast furnace raceway at higher injection levels without leading to the reducing RAFT
observed in scenarios like Case 4.

Stelco Lake Erie Works Blast Furnace


With the impacts of a shift to heavy NG injection operation clearly delineated in the AK Steel simulation analysis, the
SMSVC project technical committee aimed to explore potential methods of counteracting associated difficulties. The Lake
Erie Works #1 BF operates with only NG injection as the auxiliary fuel, and as such, was presented as a valuable test bed for
analyzing the effectiveness of pre-heating NG for improving furnace operation. The LEW BF is roughly 35% larger than the
AK Steel DW BF, uses 24 tuyeres as opposed to 20, and, at the typical conditions for this study, operates at a slightly higher
production rate.
Previous studies have been conducted on this furnace under a variety of conditions, focusing on the impacts of NG injection
lance design on combustion and furnace operation15. Typical operating conditions have been updated since the time of that
study, and a new baseline case was required for this research. The updated baseline uses a wind rate of 270,000 Nm3/hr, a
HBT of 1,448 K, a hot blast oxygen enrichment of 29%, a hot blast moisture content of 23 g/Nm3, and a natural gas injection
rate of 95 kg/MTHM. Figure 9 contains an overview of gas temperature distributions in the tuyere & blowpipe region of the
LEW BF, while Figure 10 provides a representation of the NG combustion plume through a volumetric rendering of high
temperature combustion gases.

Figure 9. Contours of gas temperature in the LEW tuyere on a vertical center plane parallel to the blowpipe centerline (left), a
horizontal plane parallel to the lance and tuyere centerlines (center), and on four vertical planes perpendicular to the blowpipe
centerline (right).

Figure 10. Volume rendering of combustion gases above 2,000 K in the LEW tuyere under baseline conditions.
High temperature combustion gases are concentrated in the mixing region between injected NG and hot blast, with the
turbulent wake introduced by the physical flow disruption of the NG injection lance in the blowpipe flow pattern causing
some curling and rotation of the flow as the blast/NG mixture leaves the tuyere. The bored lance design functions similarly to
a straight pipe injection system, however, the ports around the lance tip allow for some additional mixing between NG and

390 © 2019 by the Association for Iron & Steel Technology.


hot blast, broadening the combustion plume slightly. This phenomena can be observed in Figure 18, particularly around the
lance tip itself, with small combustion plumes stemming from the ports on the upper side of the lance clearly visible. It is
worth noting that previously published studies have examined the impact of various lance designs on combustion
characteristics and flow patterns within the tuyere region, with a ‘fast mix’ lance providing an enhanced version of the effects
observed here15.
Figure 11 shows the distribution of coke bed porosity under baseline operating conditions for the LEW furnace. Additionally,
it illustrates the selected ‘outer boundary’ of the raceway at a porosity value of 0.7 (70% void fraction). This predicted
raceway boundary extends roughly one meter from the tuyere nose, and provides a low porosity region in which gas
recirculation and combustion will occur in the model.

Figure 11. Side and top views of raceway porosity distribution (left) and raceway outer boundary (right).
For the LEW BF baseline conditions here, the predicted RAFT analogue value is 2,114 K. Figure 12 includes contours of gas
temperature, CH4 mass fraction, and H2O mass fraction on vertical and horizontal cross-sections through the raceway region,
providing a snapshot of distributions. Significant left-right stratification exists in gas temperatures, CH4 distribution, and H2O
distribution, with an inverse relationship between H2O/CH4 content and gas temperature. This indicates that the NG injection
lance is pushing CH4 to the ‘left side’ (looking from the outside of the furnace in) of the raceway, leading to increased
endothermic reactions in that region and correspondingly lower temperatures. This seems to indicate that furnace operation in
the upper regions will experience periodic variations in gas temperature when moving in the tangential direction.

Figure 12. Contours of gas temperature (left), CH4 mass fraction (middle), and H2O mass fraction (right) in the LEW furnace
raceway region.

© 2019 by the Association for Iron & Steel Technology. 391


Figure 21. Contours of gas species distributions in the LEW raceway region, on both vertical and horizontal center planes.
Figure 13 shows additional species distribution patterns that correspond well with the gas temperatures observed in Figure
12. H2 production is highest to the left of the raceway region, as that is the direction in which most of the injected CH4 moves
as it travels down the tuyere jet. H2O production is highest in the combustion region of NG as well, and the water vapor then
reacts with carbon in the coke be to produce hydrogen gas. This results in some periodic variation in which the percentage of
the gas made up of H2 and CO respectively will fluctuate depending on tangential location within the furnace.
Distributions of gas species, temperature, and mass flow rate are exported along the upper boundary of the raceway model
and become input values for the BF shaft model. Results obtained from shaft model simulation include predictions of coke
rate, gas utilization, and more. Burden distributions with the furnace shaft were generated based on burden charge weights
and layer thicknesses provided by Stelco for the selected BF operating conditions. In addition, the moisture content in the
charged burden is assumed at 2.5% for ore pellets and 6% for coke, based on industry recommendations. These values can
vary in real-world scenarios, resulting in impacts to top temperature and top gas H2O content (which impacts H2 utilization
predictions in the CFD model). Figure 14 shows predictions of gas temperature and species distributions in the shaft region
for the baseline scenario.

392 © 2019 by the Association for Iron & Steel Technology.


Figure 14. Contours of gas temperature and gas species distribution, including cohesive zone location, in the LEW BF shaft
under baseline conditions.

Table III. CFD Predictions and Industry Expected Values for LEW Baseline Operation
CFD Prediction Industry Expected Values
Coke Rate 392 kg/thm ~390 kg/thm
CO Utilization 50.77% ~ 50%
H2 Utilization 51.46% ~ 50%

Avg. Top Gas Temperature 116.46 °C 100 – 110 °C

ΔP (Model bottom to top) 123 kPa N/A

In an effort to ascertain the impacts of increased injected NG temperature, six additional cases were simulated along with the
baseline scenario. These cases, selected based upon conditions proposed by project technical committee members, spanned a
range of NG pre-heat from zero to 300 K in 50 K intervals. In the baseline, the NG temperature is ambient temperature
(assumed to be 300 K), and the maximum pre-heat temperature was capped at 600 K to avoid potential cracking of CH4
within the NG lance. All other operating conditions were held constant in these cases to isolate the impacts of pre-heating on
RAFT, combustion characteristics, and shaft operation. For reference, a list of the pre-heat cases is presented in Table IV.

Table IV. NG Pre-heat Case Conditions


Wind Rate O2 % Hot Blast Temp. NG Rate NG Temp.
[Nm3/hr] [Mass Fraction] [K] [kg/THM] [K]
Baseline Case 270,000 31.4% 1448.15 95 300 (+ 0K)
Case 1 270,000 31.4% 1448.15 95 350 (+ 50K)
Case 2 270,000 31.4% 1448.15 95 400 (+ 100K)

Case 3 270,000 31.4% 1448.15 95 450 (+ 150K)

Case 4 270,000 31.4% 1448.15 95 500 (+ 200K)

Case 5 270,000 31.4% 1448.15 95 550 (+ 250K)

Case 6 270,000 31.4% 1448.15 95 600 (+ 300K)

The distribution of high-temperature gases are unsurprisingly similar between the cases both in the tuyere & blowpipe region
and the raceway. Flow rates, oxygen enrichment, and other operating conditions remain identical, and there is some variation
in NG velocity as the gas exits the lance due to the increased volume of higher temperature NG. Comparing Case 6 (300 K
NG pre-heat) to the baseline case, average gas temperature at the tuyere outlet is 1.4% higher and the average gas velocity is

© 2019 by the Association for Iron & Steel Technology. 393


1.8% higher. Gas temperature distributions in the tuyere region for the two extreme cases (baseline and case 6) are detailed in
Figure 15.

Figure 23. Contours of gas temperature in the tuyere region on a center plane and the tuyere outlet plane for the baseline case
(left) and case 6 (right).
Residence time within the tuyere is fairly low, and as such, the most significant effects of NG pre-heating can be more clearly
observed in the raceway region. While the location of high and low temperature zones remains similar between the cases, the
NG pre-heating has a visible impact on gas temperatures in the raceway region. A comparison of raceway temperature
distributions in the baseline case, Case 3 (150 K pre-heat), and Case 6 (300 K pre-heat) is shown in Figure 24.

Figure 24. Contours of gas temperature in the LEW furnace raceway region for the baseline case (left), Case 3 (middle), and
Case 6 (right).

Table V. NG Pre-heat Case RAFT Analogue Predictions


NG Temp. RAFT Analogue RAFT Inc. from RAFT Inc. from
[K] [K] Baseline [K] Baseline [%]
Baseline Case 300 (+ 0 K) 2,131 0 0.0%
Case 1 350 (+ 50 K) 2,144 13 0.6%
Case 2 400 (+ 100 K) 2,154 23 1.1%

Case 3 450 (+ 150 K) 2,166 35 1.6%

Case 4 500 (+ 200 K) 2,177 46 2.1%

Case 5 550 (+ 250 K) 2,195 64 2.9%

Case 6 600 (+ 300 K) 2,209 78 3.5%

394 © 2019 by the Association for Iron & Steel Technology.


Table V details the RAFT analogue values for each case in the NG pre-heating parametric study. CFD predictions indicate
that pre-heating NG by 50 K under these operating conditions leads to an increase of roughly 0.6% (12 K) in RAFT
analogue, with that increase remaining fairly linear across the range of 0 – 300 K. At the upper end of the range, the RAFT
analogue is increased by roughly 78 K. This would provide additional heat for reducing gases, and may produce similar
impacts to an increase in HBT. The increased temperature would theoretically provide additional headroom for higher NG
injection rates, as it would counteract some of the quenching effects observed under those conditions. This could in turn
reduce coke rates merely by way of supplying additional auxiliary fuel, though some previous studies have suggested that the
replacement ratio may actually increase when using pre-heated NG2,3. As previously mentioned, the “Svobody Sokol” BF
claimed an increase from a replacement ratio of 1.2:1 kgcoke/kgng to 1.42:1 kgcoke/kgng while using a 170 °C NG pre-heat.
Further research is underway to explore the impacts on the Stelco LEW furnace for typical conditions.
Potential fuels savings benefits are still being explored. Feshchenko et al.’s study indicated significant benefits at even
modest pre-heating levels (2.2-3.3% coke rate reductions)3. As the NG injection rate is held constant in this study, perhaps
the most direct method of ascertaining the impacts of NG pre-heating is via a comparison of RAFT analogue values to that of
a case with increased hot blast temperature, the effects of which on furnace operation are well known from industry rules of
thumb. An increase of 25 K in HBT would result in coke savings of 4.5 kg/mthm, and matching the RAFT analogue of this
case (essentially matching available heat) with a corresponding NG pre-heat case should yield a scenario with similar
impacts. Top temperatures could similarly be expected to increase by approximately 9 K for this 25 K HBT increase. During
periods of low top temperature this could mean that O2 injection could remain stable and additional coke would not be
necessary to increase top heat (resulting in overall better production rates in these conditions). Additionally, in periods of
stockline recovery where low top heats are encountered, it might be possible to use a system to temporarily increase NG
temperature in an effort to remedy such issues. The predicted ratio between the impacts of hot blast vs. the impacts of NG
pre-heating is the topic of continuing research.

CONCLUSION
Comprehensive 3D CFD modeling of the blast furnace tuyere, blowpipe, raceway, and shaft was performed to examine the
impacts of high-rate and pre-heated natural gas injection into the blast furnace tuyere. This study encompassed analyses of
blast furnaces at two industrial plants (AK Steel Dearborn Works and Stelco Lake Erie Works), with various scenarios
simulated at each, including loss of PCI, altered PC/NG co-injection ratios, and a range of pre-heat temperatures for injected
NG. The results of these cases confirmed that there is a clear inverse correlation between the rate of NG injection and
raceway gas temperatures, aligning with expectations. Simulation of scenarios under which all fuel in a co-injection furnace
was replaced with NG (and injection was maintained at the same effective rate) predicted raceway gas temperatures below
the low-end of reported industry values for stable operation furnace. Plant operation under reduced production rate conditions
revealed no initial problems, however, the impacts of running the furnace at such levels for extended periods, even under
reduced production rates, are unknown.
CFD modeling of pre-heated NG injection predicted that an increase of 50 K in NG temperature would result in a
corresponding 0.6% (12 K) in raceway gas temperature, and that increase remains fairly linear across a range of 0 – 300 K
NG pre-heat (with a 3.5% increase observed at a 300 K pre-heat). This increase in temperature should theoretically allow for
higher NG injection rates, as it would serve to counteract the quenching effect observed as more NG is put into the raceway.
Additionally, under higher NG injection rates, the impact of pre-heating is expected to be more significant, corresponding to
the increased sensible heat flow. It is also worth noting that heating the injected NG could help to increase furnace top heat.
Research work on this topic area is currently ongoing within the SMSVC, and immediate next steps will focus on
ascertaining how pre-heated NG injection compares to increased hot blast temperature scenarios and simulating the pre-
heated NG injection cases through the shaft region of the furnace to examine variations in top temperature, coke rates, and
gas utilizations. Additionally, a second set of raceway simulations will be completed to confirm whether or not a stable
RAFT can be maintained at higher NG injection rates through the use of NG pre-heating. These simulations will endeavor to
determine the feasibility of using pre-heated NG injection to widen the stable window for BF operation by providing a
potential counteraction to the negative impacts of scenarios including high NG injection rates and high moisture content
burden.

ACKNOWLEDGEMENTS
The authors acknowledge the SMSVC blast furnace project technical committee for providing operating conditions,
geometry, and validation data. The authors would also like to thank the industrial contributors and management from both

© 2019 by the Association for Iron & Steel Technology. 395


Stelco and AK Steel for their input to and continued support of the project. Finally, the authors would like to thank the CIVS
team for their support and input to the simulation research work.

DISCLAIMER
Any views or opinions presented in this report do not necessarily represent those of AK Steel Corporation. AK Steel accepts
no liability for the content of this report and for consequences of any action taken on the basis of information contained in
this report.
The material in this paper is intended for general information only. Any use of this material in relation to any specific
application should be based on independent examination and verification of its unrestricted availability for such use and a
determination of suitability for the application by professionally qualified personnel. No license under any patents or other
proprietary interests is implied by the publication of this paper. Those making use of or relying upon the material assume all
risks and liability arising from such use or reliance.

REFERENCES
1. M. Geerdes, R. Chaigneau, I. Kurunov, O. Lingiardi, J. Ricketts, “Modern Blast Furnace Iornmaking: An Introduction,”
IOS Press, 2015, pp. 70-75.
2. P.C. Pistorius, J. Gibson, M. Jampani, “Natural Gas Utilization in Blast Furnace Ironmaking: Tuyère Injection, Shaft
Injection and Prereduction.” Applications of Process Engineering Principles in Materials Processing, Energy and
Environmental Technologies, Part of The Minerals, Metals & Materials Series, 2017, pp. 283-292.
3. S. A. Feshchenko, V. I. Pleshkov, V. N. Loginov, I. F. Kurunov, “Synergetic Effect of Natural Gas Pre-heating Prior to
its Injection into a Blast Furnace,” AISTech 2008 Conf. Proc., 2008, 6 pgs.
4. M. Zhen, Z. Zhou, A. B. Yu, Y. Shen, “CFD-DEM Simulation of Raceway Formation in an Ironmaking Blast Furnace,”
Powder Technology, Vol. 314, 2017, pp. 542-549.
5. C.P. Yeh, S.W. Du, C.H. Tsai, R.J. Yang, “Numerical Analysis of Flow and Combustion Behavior in Tuyere and
Raceway of Blast Furnace Fueled with Pulverized Coal and Recycled Top Gas,” Energy, Vol. 42, Iss. 1, 2012, pp. 233-
240.
6. A. Babich, D. Senk, H. W. Gudenau, “An Outline of the Process,” Ironmaking, Verlag Stahleisen GmbH, Dusseldorf,
2016, pp. 180-185.
7. Burke P. D. and J. M. Burgess, (1989), “A coupled gas and solid flow, heat transfer and chemical reaction rate model
for the ironmaking blast furnace”, Proc. Ironmaking Conf., 48, 773-781.
8. A. Vuokila, O. Mattila, R. L. Keiski, E. Muurinen, “CFD Study on the Heavy Oil Lance Positioning in the Blast
Furnace Tuyere to Improve Combustion,” ISIJ International, Vol. 57, No. 11, 2017, pp. 1911-1920.
9. P.R. Austin, H. Nogami, J. Yagi, “A Mathematical Model of Four Phase Motion and Heat Transfer in the Blast
Furnace”, ISIJ International, 1997, vol. 37, pp. 458–467.
10. K. M. Komiyama, B. Guo, H. Zughbi, P. Zulli, A. Yu, “Improved CFD Model to Predict Flow and Temperature
Distributions in a Blast Furnace Hearth,” Metallurgical and Materials Transactions B, Vol. 45 B, 2014, pp. 1895-1914
11. S. Natsui, S. Ueda, H. Nogami, J. Kano, R. Inoue, T. Ariyama “Analysis of Non-Uniform Gas Flow in Blast Furnace
Based on DEM-CFD Combined Model,” Steel Research International, 2011, No. 8, pp. 964-971.
12. T. Kon, S. Natsui, S. Matsuhashi, S. Ueda, R. Inoue, T. Ariyama, “Influence of Cohesive Zone Thickness on Gas Flow
in Blast Furnace Analyzed by DEM-CFD Model Considering Low Coke Operation,” Steel Research International, 84,
2013, No. 11, pp. 1146-1156
13. D. Fu, D. Zheng, C.Q. Zhou, J. D’Alessio, K.J. Ferron, and Y. Zhao, “Parametric Studies on PCI Performances,”
Proceedings of the ASME/JSME 2011 8th Thermal Engineering Joint Conference, Honolulu, Hawaii, United States,
Paper no. AJTEC2011-44608, 2011.
14. Y. Chen, D. Fu, C. Q. Zhou, “Numerical Simulation of the Co-Injection of Natural Gas and Pulverized Coal in Blast
Furnace,” AISTech Conf. Proc., 2013, pp. 573-580.

396 © 2019 by the Association for Iron & Steel Technology.


15. A. K. Silaen, T. Okosun, Y. Chen, B. Wu, J. Zhao, Y. Zhao, J. D’Alessio, J. Capo, C. Q. Zhou, “Investigation of High
Rate Natural Gas Injection through Various Lance Designs in a Blast Furnace,” AISTech Conf. Proc., 2015, pp. 1536-
1549.
16. T. Okosun, S.Street, Y.Chen, J.Zhao, B.Wu, C.Q.Zhou, “Investigation of Co-Injection of Natural Gas and Pulverized
Coal in a Blast Furnace”, AISTech 2015 Conf. Proc., 2015, pp. 1581-1594.
17. T. Okosun, S. J. Street, J. Zhao, B. Wu, C. Q. Zhou, “Investigation of Dual Lance Designs for Pulverized Coal and
Natural Gas Co-Injection,” AISTech 2016 Conf. Proc., 2016, pp. 581-594.
18. T. Okosun, S. J. Street, J. Zhao, B. Wu, C. Q. Zhou, “Influence of Conveyance Methods for Pulverized Coal Injection
in a Blast Furnace,” Ironmaking and Steelmaking, Vol. 44, Iss. 7, 2017, pp. 513-525.
19. T. Okosun, X. Liu, A. K. Silaen, D. Barker, D. P. Dybzinksi, C. Q. Zhou, “Effects of Blast Furnace Auxiliary Fuel
Injection Conditions and Design Parameters on Combustion Characteristics and Injection Lance Wear,” AISTech 2017
Conf. Proc., 2017, 11 pgs.
20. B. Launder, D. Spalding, “Lectures in Mathematical Models of Turbulence,” New York, NY: Academic Press; 1972.
21. B. Magnussen, “On the Structure of Turbulence and a Generalized Eddy Dissipation Concept for Chemical Reaction in
Turbulent Flow,” Proc. of the 19th Aerospace Sciences Meeting of the American Institute of Aeronautics and
Astronautics, 1981.
22. W. Walker, M. Gu, J. D’Alessio, N. Macfadyen, C. Q. Zhou, “Methodology for the Numerical Simulation of Natural
Gas, Coal, and Coke Combustion in a Blast Furnace,” Proc. of the 2008 ASME Summer Heat Transfer Conf, 2008, pp.
215-223.
1. D. Spalding, “Mixing and Chemical Reaction in Steady Confined Turbulent Flames,” Symposium (International) on
Combustion, Vol. 13, Iss. 1, 1971, pp. 649-657.
2. B. Magnussen, B. Hjertager, “On Mathematical Modeling of Turbulent Combustion with Special Emphasis on Soot
Formation and Combustion,” Symposium (International) on Combustion, Vol. 16, Iss. 1, 1977, pp. 719-729.
3. L. Zhou, “Combustion Theory and Chemical Fluid Dynamics,” Science Press, 1986.
4. K. Bueters, J. Cogoli, W. Habelt, “Performance Prediction of Tangentially Fired Utility Furnaces by a Computer
Model,” Proc. of the 15th Symposium (International) on Combustion, 1974.
5. A. Lowe, I. Steward, T. Wall, “The Measurement and Interpretation of Radiation from Fly Ash Particles in Large
Pulverized Coal Flames,” Proc. of the 17th Symposium (International) on Combustion, 1978.
6. D. Fu, F. Huang, F. Tian, C. Q. Zhou, “Burden Descending and Redistribution in a Blast Furnace”, Proc. Of AISTech
2010, 2010.
7. C. Q. Zhou, “Minimization of Blast Furnace Fuel Rate by Optimizing Burden and Gas Distribution,” Final Technical
Report to U.S. Department of Energy (DOE), 2012.
8. D. Fu, Y. Chen, C. Q. Zhou, “Mathematical Modeling of Blast Furnace Burden Distribution with Non-Uniform
Descending Speed,” Applied Mathematical Modelling, Vol. 39, 2015, pp. 7554-7567.

© 2019 by the Association for Iron & Steel Technology. 397

You might also like