Available

Available online
online at
at www.sciencedirect.com
www.sciencedirect.com

ScienceDirect
Availableonline
Available onlineatatwww.sciencedirect.com
www.sciencedirect.com
Energy
Energy Procedia
Procedia 00
00 (2017)
(2017) 000–000
000–000

ScienceDirect
ScienceDirect
www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia

Energy
EnergyProcedia
Procedia120 (2017) 000–000
00 (2017) 454–461
www.elsevier.com/locate/procedia

INFUB - 11th European Conference on Industrial Furnaces and Boilers, INFUB-11

Furnace combustion and control renovation to improve the

The 15th International Symposium on District Heating and Cooling
productivity of a continuous annealing line
Assessing
Hai Wu theBenschop
a,*, Bertie van
a, feasibility bb
, Omar of Ben
using Driss the
bb heat demand-outdoor
, Ferry Frinkingaa, Ramon Speetsaa
temperatureResearch
function
Research &
forTataaSteel,
& Development,
Development, Tata
aa long-term
Steel, P.O.Box
P.O.Box 10.000,
district
10.000, 1970
1970 CA
heat
CA IJmuiden,
IJmuiden, The
demand forecast
The Netherlands
Netherlands
bb
Downstream
Downstream Operations,
Operations, Tata
Tata Steel,
Steel, P.O.Box
P.O.Box 10.000,
10.000, 1970
1970 CA
CA IJmuiden,
IJmuiden, The
The Netherlands
Netherlands
a,b,c
I. Andrić *, A. Pinaa, P. Ferrãoa, J. Fournierb., B. Lacarrièrec, O. Le Correc
a
IN+ Center for Innovation, Technology and Policy Research - Instituto Superior Técnico, Av. Rovisco Pais 1, 1049-001 Lisbon, Portugal
Abstract
Abstract
b
Veolia Recherche & Innovation, 291 Avenue Dreyfous Daniel, 78520 Limay, France
c
Département Systèmes Énergétiques et Environnement - IMT Atlantique, 4 rue Alfred Kastler, 44300 Nantes, France
The
The investigated
investigated annealing
annealing lineline was
was constructed
constructed inin late
late 1980s
1980s with
with aa design
design capacity
capacity ofof 350,000
350,000 t/y.
t/y. After
After many
many yearsyears of
of service
service
the
the annealing
annealing furnace
furnace was was inin an
an unsound
unsound state
state and
and renovated
renovated in in aa number
number of of steps
steps in
in three
three years
years time.
time. AA furnace
furnace survey
survey was
was
firstly
firstly conducted
conducted to to fingerprint
fingerprint thethe conditions
conditions ofof the
the combustion
combustion system
system and
and to
to identify
identify opportunities
opportunities for
for improvement
improvement with with aa
Abstract of investment. Burner tests were carried out to better understand the combustion process and identify the critical
quick
quick return
return of investment. Burner tests were carried out to better understand the combustion process and identify the critical
conditions
conditions for
for component
component design design and
and improvement.
improvement. Furnace
Furnace mathematical
mathematical model
model was
was developed
developed toto study
study the
the energy
energy balance
balance and
and
District heating
deficiency networks The are commonly addressed in the literature as one of the most effective solutions for decreasing the
deficiency of of the
the process.
process. The modelmodel was was further
further explored
explored in in the
the designing
designing and
and implementation
implementation of of aa new
new control
control automation
automation
greenhouse gas
system emissions fromControl.the building sector.results
These systems require high were
investments which are returned through the heat
system using
using Model
Model Predictive
Predictive Control. Production
Production results overover the
the recent
recent years
years were collected
collected and
and analyzed.
analyzed. The
The data
data after
after the
the
sales.
renovationDue to the changed climate conditions and building renovation policies, heat demand in the future could decrease,
renovation showed
showed some some significant
significant improvements
improvements with with aa reduced
reduced specific
specific energy
energy consumption
consumption below
below the
the benchmark
benchmark of of World
World
prolonging the investment
Steel return period.
Steel Association,
Association, aa 4% 4% increase
increase inin monthly
monthly production
production throughput,
throughput, over
over 10%
10% increase
increase inin production
production volume
volume andand anan advance
advance toto aa
The main
highly scope of
automated this paper is to assess the feasibility of using the heat demand – outdoor temperature function for heat demand
process.
highly automated process.
© forecast.
2017 TheThe districtPublished
Authors. of Alvalade,
by locatedLtd.
Elsevier in Lisbon (Portugal), was used as a case study. The district is consisted of 665
©
© 2017
2017 The
The Authors. Published
Authors. Published by by Elsevier
Elsevier Ltd.
Ltd.
buildings
Peer-review that vary
under in both construction period and typology. Three weather scenarios (low, medium, high) and three district
Peer-review
Peer-review under responsibility
responsibility of
of the organizing
of the
the organizing
committee
committee
committee of
of INFUB-11.
INFUB-11.
renovation under
scenariosresponsibility
were developed organizing
(shallow, intermediate, of deep).
INFUB-11 To estimate the error, obtained heat demand values were
compared
Keywords: with
Annealing;results from
Combustion; a dynamic
Modelling; heat demand
Furnace
Keywords: Annealing; Combustion; Modelling; Furnace control model,
control previously developed and validated by the authors.
The results showed that when only weather change is considered, the margin of error could be acceptable for some applications
(the error in annual demand was lower than 20% for all weather scenarios considered). However, after introducing renovation
1.scenarios,
1. the error value increased up to 59.5% (depending on the weather and renovation scenarios combination considered).
Introduction
Introduction
The value of slope coefficient increased on average within the range of 3.8% up to 8% per decade, that corresponds to the
decrease in the number of heating hours of 22-139h during the heating season (depending on the combination of weather and
Steel
Steel strip
strip goes through
through aa complex
goes considered). complex process
process including pickling, cold
cold rolling, annealing, temper rolling
rolling onand
renovation scenarios On the other hand, including pickling,
function intercept increased rolling, annealing,
for 7.8-12.7% temper
per decade (depending andthe
electrolytic
electrolytic tinning
tinning to
to become
become tinplate
tinplate before
before it
it is
is dispatched
dispatched to
to can
can manufacturers.
manufacturers. The
The annealing
annealing is
is aa crucial
crucial step
step to
to
coupled scenarios). The values suggested could be used to modify the function parameters for the scenarios considered, and
heat-treat
heat-treat
improve the the full
full hard
theaccuracy hard strip to
to produce
stripdemand
of heat produce the
the right
estimations.right mechanical
mechanical properties.
properties. Continuous
Continuous annealing
annealing is is aa very
very efficient
efficient

© 2017 The Authors. Published by Elsevier Ltd.

Peer-review under responsibility of the Scientific Committee of The 15th International Symposium on District Heating and
*
* Corresponding
Corresponding author.
Cooling. author. Tel.:
Tel.: +31-251-496-833.
+31-251-496-833.
E-mail address: [email protected]
E-mail address: [email protected]
Keywords: Heat demand; Forecast; Climate change
1876-6102 ©
1876-6102 © 2017
2017 The
The Authors.
Authors. Published
Published by
by Elsevier
Elsevier Ltd.
Ltd.
Peer-review
Peer-review under
under responsibility
responsibility of
of the
the organizing
organizing committee
committee of
of INFUB-11.
INFUB-11.

1876-6102 © 2017 The Authors. Published by Elsevier Ltd.

Peer-review under responsibility of the Scientific Committee of The 15th International Symposium on District Heating and Cooling.
1876-6102 © 2017 The Authors. Published by Elsevier Ltd.
Peer-review under responsibility of the organizing committee of INFUB-11
10.1016/j.egypro.2017.07.219
2 Hai WuHai Wu/ et
et al. al. / Energy
Energy Procedia
Procedia 120000–000
00 (2017) (2017) 454–461 455

process for this purpose, which conveys strip through a number of consecutive furnace sections providing the
required heat treatment in a continuous manner. The annealing processing line investigated in this paper is one of
the two continuous annealing facilities at Tata Steel’s IJmuiden site. The line was commissioned in late 1980s with a
design capacity of 350,000 t/y. The annealing process consists of heating, soaking, cooling, over-aging and final
cooling, as illustrated in Fig. 1. After the annealing, the strip is temper rolled or double reduced up to 30% to
produce various qualities.

The heating furnace is the critical part of the annealing process. The furnace has many radiant tube units installed
to provide the heating and furnace rolls to transport the strip. The furnace inside is filled with protective gas, a
mixture of nitrogen and hydrogen, to prevent the strip from oxidizing at elevated temperature. As the heat source,
every radiant tube unit is equipped with a burner on one end to fire the natural gas and a recuperator on the other end
to recover the waste heat to preheat the combustion air, as shown in Fig. 2. The radiant tube has a maximum service
temperature of 950°C as specified by the supplier, which is monitored by a thermocouple attached to the metal tube.
The tube is subjected to thermal degradation as the ‘normal’ process and also thermal shocks in emergency
situations when the furnace needs to be cooled down fast to avoid strip overheating and potential breaks [1].
Experiences show that the life time of the radiant tube can vary between 2 and 10 years depending on its service
temperature history. The availability of the tube units has a direct impact on the furnace capacity and performance,
and consequently the annealing process. Periodic inspection and maintenance are carried out to maintain the
installation condition. Furthermore, improvements are sought to meet the plant ambition to reduce the natural gas
consumption and hence CO2 emission, and further to maximize the line output in volume and quality in a
sustainable way.

Due to the economy recession back in 2008, large capital investment and furnace revamp incurring long
production stops were certainly out of question. The improvements largely based on the existing installation were
defined and executed in a number of consecutive steps: renovating the defected radiant tube units to improve the
installation availability, refining the combustion control to maximize the combustion efficiency, and implementing
an advanced control system to optimize the furnace operation and hence the productivity. The investigations and
obtained results are highlighted in the next sections.

Fig. 1. Continuous annealing line.

2. Furnace condition diagnosis

Apart from the tube failures as routinely seen, the inspection revealed unexpected damages on a number of
recuperators (Fig. 3) and sealing gaskets between the mounting flanges. Analysis led to believe that the deterioration
was primarily due to an enduring exposure to high temperature during service. The higher than normal temperature
can be triggered by the following and propagate by their combining effects.

 Insufficient insulation leads to a high heat flux to the structure of the recuperator.
 The supply of combustion fuel and air is out of balance with either too much fuel or too less air to the individual
burner units, which prolongs the combustion reaction beyond the radiant tube and further inside the recuperator.
456 Hai Wu et al. / Energy Procedia 120 (2017) 454–461
Hai Wu et al. / Energy Procedia 00 (2017) 000–000 3

 Air leakage through the flanges due to depleted or missing gaskets provides extra air for post-combustion
possibly taking place at the recuperator.
 Air leakage also distorts the normal air supply and transport through the outer shell of the recuperator and
weakens its cooling effect on the structure.
 Any misalignment and/or limited clearance for thermal expansion and contraction also contribute to the
mechanical failure eventfully.

Fig. 2. Schematic layout of radiant tube unit.

A combustion survey was conducted to inspect the flow balance and combustion conditions of all the burners
while in operation. Through the field sampling points at the individual burners, a number of process parameters
were checked with mobile instruments, including the fuel gas pressure before the burner (#1 in Fig. 2), the pressure
drop across the orifice in the air supply (#2), and the exhaust gas compositions (O2, CO, CO2) after the recuperator
(#3). Fig. 4 shows an overview of the exhaust gas analysis results of a controlled segment of burners. The burners
with either too low (< 2%) or too high (> 5%) O2 concentration were selected for further examination and flow re-
balancing if necessary. One would logically expect that the burners with a very low O2% in the exhaust gas also
produce an increasing level of CO, like the burners #A2, #B2 and #E3, which is an indication of incomplete
combustion due to the lack of air. The situation like the burners #B3, #C1 and #C2 exhibiting a high level of O2 and
CO coexisting in the exhaust gas leads to the suspicion of the aforementioned air leakage and structure deterioration.

Fig. 3. Damaged recuperator shell. Fig. 4. Field exhaust gas analyses of radiant tube burners.

A number of representative burners were selected and the same measurements were recorded for a prolonged
time using a data logger. From the production system, the furnace conditions were extracted to synchronize and
correlate with the measurement of the individual burners. Fig. 5(a) shows the data samples of the realized air-to-fuel
ratios or corresponding excess air levels of one selected burner over a period of time. Apparently, the scattered data
along with the varying flows are not consistent with and often much higher than the target level. It points to the
 Hai Wu et al. / Energy Procedia 120 (2017) 454–461 457
4 Hai Wu et al. / Energy Procedia 00 (2017) 000–000

issues in the basic combustion control of the segment to which the selected burner belongs. Investigations leads to
adjustments in the control and the improved results in Fig. 5(b), which contributes to the energy efficiency
improvement as further discussed in the next section.

Fig. 5. Measured excess air level against the target level: (a) before and (b) after improvement.

3. Combustion test and efficiency

Burner tests were carried out to answer those questions raised during the combustion survey. A full scale burner
and recuperator were brought to the test furnace located at the research center (Fig. 6). The test furnace is already
equipped with a radiant tube in the same dimension and two cooling panels alongside the tube to simulate the cold
strip in the annealing furnace. The following parameters were measured for a series of combustion tests at the
positions as indicated in Fig. 2.

 Combustion air and fuel gas air flows (#8, #9)

 Fuel gas pressure before the burner (#1)
 Air pressure drop across the flow orifice (#2)
 Air temperatures before and after the recuperator (#4, #5)
 Exhaust temperatures before and after the recuperator (#6, #3)
 Exhaust gas compositions (O2, CO, CO2) (#3)
 Recuperator exterior metal temperatures (#7)
 Pressure drop across the radiant tube unit (#4, #3)
 Furnace, radiant tube and cooling panel temperatures (#10, #11, #12). There are more thermocouples equally
distributed over the tube length (Fig. 6) and the panel surface to sample the local temperatures.

Fig. 6. Radiant tube test furnace. Fig. 7. Recuperator exterior critical metal temperature.
458 Hai Wu et al. / Energy Procedia 120 (2017) 454–461
Hai Wu et al. / Energy Procedia 00 (2017) 000–000 5

Fig. 7 shows the correlation between the recuperator exterior metal temperature and the test furnace temperature
obtained from the burner tests. A sustainable furnace temperature of 670°C (#B in Fig. 7) is reached under the 100%
burner capacity and an average cooling panel temperature of 320°C. The measured recuperator exterior temperature
is 440°C, below the material service temperature of about 500°C. In the actual production, the furnace and strip
temperature could reach 900°C and 650°C respectively, resulting in an increasing recuperator temperature up to
600°C by model extrapolation (#C in Fig. 7). The similar temperatures were obtained for a short period time in the
test furnace but it was not possible to sustain the condition. The information provides valuable insights for selecting
the suitable heat-resistant material for the outer shell of the recuperator, and the similar for the sealing gaskets.
Meantime more than 40 recuperators were replaced with the ones fabricated with the newly selected materials.

The burner tests in the test furnace provide opportunities for close observation and also easy access for
instrumentation. The test results are useful to create the reference to evaluate the field measurements from the
combustion survey. The same test setup was also used to study the options that could potentially improve the energy
efficiency of the radiant tube unit, like lowering the excess air level and placing the swirl inserts. The test results in
Fig. 8 shows that every 5% lowering in the excess air level towards the stoichiometric combustion results in an
approximate 1.2% increase in the energy efficiency. It is also observed that the burner starts to produce CO
measurable in the exhaust gas when the excess air level reaches below 5% or at the low firing capacity when the
excess air level is below 10%. This helps to define the new excess air target in Fig. 5(b) to further improve the
energy efficiency. The high emissivity swirl inserts placed inside the exhaust leg of the radiant tube improve the
convective and radiative heat transfer between the hot flue gas and the metal tube. As a result, it increases the local
tube temperature, by 27°C as measured in Fig. 9, and thus the radiation potential towards the furnace and the strip
given the same burner capacity. Other options to increase energy efficiency like the oxy-fuel combustion, the
regenerator heat recovery, the high temperature tube, etc. could be studied in the test furnace, but they were not
considered due to the circumstance as early mentioned.

4. Furnace model and control

A furnace model attracted interests for not only the offline studies but also online production control. The first-
principles model was developed because of its intuitive and generic nature [2]. The model is assembled as a matrix
of computational cells representing the physical structure of the radiant tube furnace, which basically consists of
three types of cells: a radiant tube cell, a transport roll cell, and an empty cell as illustrated in Fig. 10. Each cell
contains elements that represent a physical part of the furnace structure: the wall, the strip, the radiant tube, the gas
and/or the transport roll.

Fig. 8. Burner energy efficiencies at various excess air levels under Fig. 9. Effect of high emissivity swirl inserts on tube metal
furnace temperature 670°C. temperatures measured in test furnace.

The temperature states of the elements are computed based on numerical integration of the non-linear partial
derivative function in Equation (1). The partial net heat flow function qn of the element n is the balance of heat
 Hai Wu et al. / Energy Procedia 120 (2017) 454–461 459
6 Hai Wu et al. / Energy Procedia 00 (2017) 000–000

exchanges with the other elements in the cell through radiation, convection and conduction, as defined in Equation
(2). It depends on the temperature states T of all the elements and process parameters u such as the strip dimension,
the line speed and the furnace heating power. The mass mn is constant except for the strip element, and the specific
heat cp_n depends on the material and temperature of the element. The radiant tube element has an extra term to
compute, namely the internal combustion and associated heat transfer as illustrated in Fig. 2.

dTn qn (  , u ) (1)

dt m n (u )  c p _ n (  )

q wall (, u )  q wall

rad
(, u )  q wall
conv
( )  q wall
heatloss
()
q strip (, u )  q strip
rad
(, u )  q strip
conv
( ) (2)
q tube (  , u )  q rad
tube (, u )  q conv
tube ( )  q furnace
tube (, u )
 q (  )  q (  , u )
conv
wall
conv
strip
q gas ( , u )    
  qtube (  )  q roll ( ) 
conv conv

q roll (, u )  q roll

rad
(, u )  qroll
conv
( )  qroll
cond
(, u )

Fig. 10. Radiant tube furnace model discretization.

H exhaust (3)
  1
H air  H fuel  H combustion

 energy efficiency [-]

H exhaust heat content of waste gas [J]
H air heat content of cold air [J]
H fuel heat content of fuel gas [J] which is
negligible in this application
H combustion heat of combustion [J] based on lower
heating value of fuel gas

Fig. 11. Furnace energy balance in equilibrium.

The steady state solution, dTn  0 in Equation (1), is used to study the energy balance of the furnace. Fig. 11
dt
shows an example of the calculation results. From the breakdown, the average energy efficiency of the radiant tube
units or the combustion system as defined in Equation (3) can be deduced 1 – 28.0% = 72.0%. The energy efficiency
of the individual unit can largely vary between 55% and 80% as early discussed. The overall energy efficiency of
the furnace reads 60.3%, which means that 60.3% / 72.0% = 83.8% of the available radiation from the tubes goes to
the strip element and the other 16.2% goes to the furnace walls as heat loss, to the protective gas and the transport
rolls. The steady state calculation produces a benchmark for a specific production situation that reaches the
equilibrium.
 Hai Wu et al. / Energy Procedia 00 (2017) 000–000 7
460 Hai Wu et al. / Energy Procedia 120 (2017) 454–461

The furnace energy efficiency is furthermore influenced by many other parameters like the strip dimension, the
strip surface property, the line speed, the temperature target, etc. which change frequently in the real production
during the so-called strip transitions. Frequent strip transitions bring frequent disturbances onto the process and raise
thus a number of challenges to the furnace control: flexible to order changes, well-judged strip transition control,
high temperature accuracy and yet high throughput to maximize economical benefits, and energy conservation.
Production experiences show that the conventional feedback controller performs inadequately to meet up all these
demands and leaves a lot 'impossible' to operator. Manual interventions based on experiences sometimes help the
situation but the overall performance is lower than desired, resulting in not only high energy consumption, but also
unstable process, quality inconsistencies and rejects due to large temperature deviations especially during the strip
transitions.

An advanced furnace controller was developed as the answer [3]. It adopts the model predictive control (MPC)
technology which is considered a key solution [4,5,6] to the challenges and the limitations that the conventional
control has. The design makes use of the aforementioned furnace model, dTn  0 in Equation (1), to predict process
dt
dynamics in the near future, computes an optimisation process to yield in real time an optimal control sequences,
and integrates the furnace power and the line speed control to offer full automation. The optimisation is
mathematically described as a cost function and solved as a minimization problem for the future horizon. The
process parameters like temperature deviation, speed variation, fuel consumption, etc. are formulated as costs and
weighted distinctly. In this way, it introduces an explicit mechanism in the control automation to describe and fulfill
multiple production goals, something that the conventional controller is not capable of. Installation limits and
process constraints are also considered in the formulation of the optimisation problem.

The new controller is developed in a generic form and can be configured to accommodate the specific line
configurations and operational practices, as illustrated in Fig. 12. The design has been tested followed by
implementation on the investigated continuous annealing line in 2012. The results show a broad range of
improvements in full compliance with the design. Operators are relieved from the intensive manual operation and
can focus more on the supervisory tasks. Production teams are very pleased with the new controller.

Fig. 12. MPC-based controller design diagram.

5. Results and benefits

The renovation work was completed in three years. The production results over the last seven years were
collected and split into three equal periods for comparison: before, during and after the renovation.

Fig. 13 shows the progressing improvement in the natural gas consumption of the annealing process from the
period before the renovation (unit 1 as the reference), to an average 0.97 in a gradual descend as the renovation work
progresses, and to 0.93 after the renovation including the new furnace controller are fully implemented. The latest
 Hai Wu et al. / Energy Procedia 120 (2017) 454–461 461
8 Hai Wu et al. / Energy Procedia 00 (2017) 000–000

figures are consistently under the benchmark for the tinplate continuous annealing published by the World Steel
Association (formerly the International Iron and Steel Institute (IISI) until 2008) [7]. Other recorded improvements
include a 4% increase in the monthly maximum throughput primarily due to the implementation of the advanced
furnace controller, and over 10% increase in the production volume in which the furnace renovation plays an
essential role.

Fig. 13. Historical data of monthly specific energy consumptions.

6. Conclusions

The continuous annealing line was renovated to reduce the energy consumption and hence CO2 emission of the
heating furnace and to meet the increasing production demands of the business. Various improvements were
investigated through the field furnace survey, the laboratorial burner test, the detailed process modeling and the
model-based control application. The implementation in three planned phases led to the satisfactory results with an
improved furnace availability, a higher energy efficiency and a higher production throughput at the same time. The
development of the advanced MPC-based furnace controller and its application potential attracted more interests
from the other continuous processing lines in the company.

References

[1] Schmitz N, Hellenkamp M, Pfeifer H, Cresci E, Wünning J, Schönfelder M. Radiant tube life improvement for vertical galvanizing lines. Iron
and Steel Technology 2016;2:67-73.
[2] Wu H, Speets R, Heeremans F, Ben Driss O, Van Buren R. Non-linear model predictive control of throughput and strip temperature for
continuous annealing line. Ironmaking & Steelmaking 2015;42(8):570-578.
[3] Wu H, Speets R, Van Benschop B, Van Hasselt P, Van Der Kooij J. Advanced transitional system optimiser using model predictive control
for continuous annealing. Proc. 2nd European Steel Technology and Application Days (ESTAD) 2015.
[4] Lewis DL, Sandoz DJ, Norberg PO, Warwick K. The application of predictive control of steel annealing. Proc. Conf. CONTROL’94
1994;389:692-698.
[5] Toshitani N, Hasegawa A. Model-based control of strip temperature for the heating furnace in continuous annealing. IEEE Trans. Control
Syst. Technol. 1998;6(2):146-156.
[6] Bitschnau L, Jakubek S, Kozek M. Constrained model predictive control of a continuous annealing furnace. Proc. ASME Conf. on Dynamic
Systems and Control 2010;285-292.
[7] Cairns CJ. Energy Use in the Steel Industry. International Iron and Steel Institute, Committee on Technology; 1998.