Towards Industrial Application Of High Efficiency Combustion

IFRF Combustion Journal Article Number 200704 , July 2007 ISSN 1562-479X

Towards industrial application of High Efficiency Combustion B.T. Burggraaf*, B. Lewis*, P.D.J. Hoppesteyn*, N. Fricker+, S. Santos+, B.K. Slim# * Corus RD&T + IFRF # Gasunie Engineering & Technology

Corresponding Author: B. T. Burggraaf Energy Optimisation Corus Group Wenckebachstraat 1 1951 JZ Velsen Noord Building 3F22 Noord-Holland THE NETHERLANDS. Tel: +44 (0) 782548196 Email: [email protected]

-2ABSTRACT In 2001 the IFRF commenced a programme of research aimed at providing information which industries could use to improve the thermal efficiency of their heating processes, reduce costs, NOx emissions and net CO2 emissions. This was the IFRF High Efficiency Combustion Research Programme. Following from developments at IJmuiden the research was based on the application of modern regenerative burners operating in the “flameless” mode. In the work described in this paper the IFRF cooperated with Corus RD&T and Gasunie Engineering & Technologies; the specific industrial application envisaged was steel slab-reheating, in particular, pusher type furnaces. The IFRF set up a 1MW semi-industrial scale test facility to simulate the reheating process equipped with two commercially available High Efficiency Combustion (HEC) burners. This paper gives details of the facility design and operation and the measurement capability with details of the results of both natural gas and coke oven gas firing campaigns. In order to investigate the application of the experience gained in the experimental investigations to a pusher type slab-reheating furnace, CFD modelling was developed based on a commercial code. Details of these developments and the application of the model are described. Finally there is a comparison between the operational data of a conventionally fired furnace and the predicted HEC fired furnace, with analysis of the gains to be made, leading to conclusions and the planning of a full scale application. 1

Introduction

The energy intensive industries, such as steel production, are becoming more and more interested in the High Temperature Air Combustion (HITAC) technology, due to the constantly rising price of energy and more stringent NOx emission legislation. An assessment of all high temperature processes within Corus integrated steel works showed that from an energy point of view the HITAC technology is most beneficial, in slab reheating furnaces and in particular the pusher type furnaces. HITAC involves the use of high air preheating to increase the thermal efficiency, combined with flameless combustion for NOx reduction.

-3In 2001, a research consortium comprising the IFRF Research Station, Corus Research Development & Technology and Gasunie Research commenced testing of HITAC technology, internally known as High Efficiency Combustion (HEC). The objective of this test was to enhance knowledge, gain experience and verify the potential benefits of the HEC technology for industrial application in general and steel slab-reheating furnaces in particular. The main areas of benefits to be assessed were: •

Increased thermal efficiency

•

Reduced NOx emissions

•

Increased productivity

These areas have been investigated using semi-industrial tests and CFD analyses. 2

Semi-industrial tests

Within the HEC research programme, experiments have been performed on a semi-industrial scale, with one set of commercially available HEC burners in a specially designed test furnace (see Figure 1 and Figure 2). The burners are equipped with honeycomb regenerators and are positioned facing each other in the furnace. Two-way valves are installed in the air/flue gas duct to allow the switching of the burners (i.e. if one burner is firing, the other burner acts as an exhaust for the combustion products).

X-axis

Figure 1. Front view of the HEC furnace at the IFRF research station (left) Figure 2. Cross section view of the applied regenerative burner (right)

-4-

A Programmable Logic Controller (PLC) is used to control the switching time of the valves, the pilot flames and the ignition mechanism. Eight water-cooled heat sinks are placed in the bottom of the furnace, to simulate the heat absorption process inside a reheating furnace. 2.1

Experimental set-up

The experiments were carried out using both natural gas and coke oven gas as fuel. Natural Gas

Coke Oven Gas

Lower calorific value

36.9 MJ/Nm

18.5 MJ/Nm3

Stoichiometric ratio

9.93 Nm3/Nm3

4.60 Nm3/Nm3

2028 ºC

2113 ºC

Adiabatic flame temp

3

Table 1. Typical properties of the applied fuels

The test programme consisted of firing Natural Gas and Coke Oven Gas flames at various thermal loads to prepare a database of experimental data to develop and validate Computational Fluid Dynamics (CFD) models. Detailed in-flame measurements of temperature and gas compositions were performed using water cooled probes and conventional gas analysis equipment. In addition Laser Doppler Anemometry (LDA) measurements were performed in natural gas flames at various thermal loads to measure local average velocities and turbulent intensities. During the trials the total heat flux to the heat sinks was monitored by measuring the temperature and flow of the water entering and exiting the heat sinks. The radiative heat flux to the heat sinks was monitored by eight ellipsoidal radiometers, which were located in the centre of each heat sink. 2.2

Experimental results

The experiments showed that the thermal efficiency of the test furnace is approximately 80%. The thermal efficiency of a pusher-type slab-reheating furnace is typically 50%. The thermal efficiency is defined as the ratio of the fuel heat input to that is extracted via the heat sinks. Thermal efficiency:

η thermal =

Q heat sinks Q in

-5The tests at different burner thermal loads showed that the thermal efficiency reduces as the load

Thermal efficiency [%]

increases (see Figure 3). 100 95 90 85 80 75 70 65 60 55 50 300

400

500

600 700 800 Thermal load [kW]

900

1000

Figure 3. Thermal efficiency as a function of the thermal load of the burners

The reduction in thermal efficiency is the result of two factors involving changes to the losses from the furnace. The furnace temperature increases with the increased thermal load and results in an increased casing loss. The increase in furnace temperature also results in an increase in the exhaust gas temperature after the regenerators, as shown in Figure 4. This is due to the fixed

Exhaust gas temperature [ºC]

program for the switching control for the burners and results in an increased stack loss. 200 190 180 170 160 150 140 130 120 110 100 300

400

500

600 700 800 Thermal load [kW]

900

1000

Figure 4. Exhaust gas temperature as a function of the thermal load of the burners

-6The semi-industrial experiments also confirmed the low NOx emission level of the HEC technology for both types of fuel in spite of the high combustion air temperature of around 1050ºC. The measured NOx emission levels are presented in Figure 5.

NOx [mg/Nm3] @ 3% O2

120 100

Natural Gas Coke Oven Gas

80 60 40 20 0 750

850

950 1050 1150 1250 Average furnace temperature [ºC]

1350

Figure 5. NOx emission as a function of the furnace temperature (nozzle diameter 15.5mm)

Figure 5 shows that the measured NOx values for natural gas are typically higher than for coke oven gas in spite of the higher adiabatic flame temperature of the coke oven gas. The higher NOx emission for the natural gas is explained by the fact that the flue gas entrainment is lower for the natural gas jet than for the coke oven gas jet. The reason for the lower flue gas entrainment is the lower velocity of the natural gas jet when the same nozzle diameter is applied for both types of fuel. At lower jet velocities the jet will entrain less flue gas and this will increase the overall NOx emission. The results presented in Figure 6 also show this phenomenon. The measured NOx emissions are typically higher for the larger nozzle diameter and therefore lower jet velocity.

-7-

NOx [mg/Nm3] @ 3% O2

120 100 80

21.8mm nozzle 15.5mm nozzle

60 40 20 0 750

850

950 1050 1150 1250 Average furnace temperature [ºC]

1350

Figure 6. NOx emission of Coke Oven Gas for different nozzle diameters

The downside of applying a smaller nozzle diameter is a higher CO emission. This phenomenon is demonstrated in Figure 7 and can, like the NOx formation, be attributed to the mixing behaviour of the fuel and the air in the furnace. In the case of CO emission, the reduced entrainment that reduced the NOx formation has the opposite effect and increases the level of the CO emission.

CO [ppm] @ 3% O2

1200 1000 800

21.8mm nozzle 15.5mm nozzle

600 400 200 0 750

850 950 1050 1150 1250 Average furnace temperature [ºC]

Figure 7. CO emission of Coke Oven Gas for different nozzle diameters

1350

-8In addition to the low NOx emission levels, the test results also showed that the heat extracted from the test furnace is highly uniform. This is presented in Figure 8, which shows the extracted heat per heat sink at various thermal loads.

Extracted heat [kW]

150 946 kW

125

557 kW

353 kW

100 75 50 25 0 1

2

Burner A

3

4 5 6 Heat sink number

7

8 Burner B

Figure 8. Extracted heat per heat sink at various thermal loads (natural gas)

The uniform heat transfer to the heat sinks shows the "well-stirred" behaviour of the HEC technology within a furnace. 3

CFD analyses

CFD analyses have been performed to describe and understand the HEC process. Several turbulence and combustion models have been validated with the experimental data from the semi-industrial tests. In addition, CFD analyses have been performed to assess the feasibility of retrofitting a pusher type slab-reheating furnace with HEC burners. 3.1

Model validation

Commercially available turbulence models (the standard k-ε and Reynolds Stress Model) have been validated with isothermal data from the experiments. A comparison of the measured gas velocities and the CFD results are presented in Figure 9.

x-velocity [m/s]

-9-

20 18 16 14 12 10 8 6 4 2 0 -2

CFD calculation LDA measurement

0

0.2

0.4

0.6

0.8

1

y-position [m] Figure 9. Axial velocity [m/s] at x = 0.3 m from the burner (y=0 m is the centre axis of the air nozzle; y = 0.25 m is the centre axis of the fuel nozzle)

Figure 9 shows that the CFD model over-predicts the maximum velocity of the air jet (at y=0 m). This is explained by the fact that the air jet is tilted slightly upward, whereas the CFD model did not predict this. On the other hand the spreading of the air jet is predicted very well. This shows that the turbulence model predicts the diffusion of the axial velocity in the radial direction well; i.e. the turbulent viscosity is accurately predicted. The CFD model slightly under-predicts the maximum velocity of the gas jet (y = 0.25 m) and the spread of the jet is also slightly smaller in the CFD results. The Reynolds Stress Model has been applied to improve the prediction of the maximum velocity and spreading of the gas jet, but the results showed that this did not clearly improve the prediction. Therefore, the decision was made to perform the non-isothermal CFD analyses with the standard k-ε turbulence model. Since the average velocities and turbulent viscosity were well predicted by this model, and these parameters govern the turbulent mixing, there was good confidence for the CFD modelling of the combusting flow. Three combustion models were analysed in the non-isothermal CFD analyses. These were the PDF (Probability Density Function)-equilibrium model, PDF-flamelet model and the Eddy Break

- 10 Up model (two equations). The radiation has been modelled using the Discrete Ordinates radiation model. A comparison of the temperature prediction of the three combustion models and

Temperature [ºC]

the measurements is presented in Figure 10.

1500 1400 1300 1200 1100 1000 900 800 700 600 500

Measurements PDF equil PDF flamelet EBU 2 reac

0

0.2

0.4 0.6 y-position [m]

0.8

1

Figure 10. Temperature [K] 0.3 m from the burner (y=0 m is the centre axis of the air nozzle)

Figure 10 shows that the relatively flat temperature profile of the measurements is not predicted by any of the combustion models. In particular the PDF-equilibrium and the EBU models overpredict the temperature in the reaction zones. The PDF flamelet model predicts the low temperatures measured in these zones more accurately. Besides the high temperatures in the reaction zones, all models also under predict the temperature of the gas jet. The under predicted temperature of the gas jet results from two causes: •

The amount of flue gas entrainment in the gas jet is under predicted

•

The chemistry is not well predicted (lower temperature would mean less combustion reaction predicted).

It is expected that the effect of the entrainment is the main factor here, due to the under prediction of the axial velocity at 0.3 m downstream of the burner. Even without combustion, the model under predicted the entrainment of flue gas (see Figure 9). So the under prediction in temperature was expected.

- 11 Finally it is surprising to see that the temperature of the flue gas near the furnace wall is not well predicted. The over prediction of 150ºC is due to the fact that the applied thermal load was higher in the CFD analysis than in the experiments. To confirm this, another CFD analysis with the PDF-equilibrium model was performed with the same thermal load as during the experiments. These results are presented in Figure 11. These results clearly show that in this case the flue gas temperature is well predicted near the furnace

Temperature [ºC]

wall. 1500 1400 1300 1200 1100 1000 900 800 700 600 500

Measurements PDF equil

0

0.2

0.4 0.6 y-position [m]

0.8

1

Figure 11. Temperature [K] 0.3 m from the burner (y=0 m is the centre axis of the air nozzle) with a second set of boundary conditions

From these analyses the conclusion has been drawn that all analysed models qualitatively predict the air and fuel jet behaviour with the corresponding chemical reactions and diffusion processes well. Quantitatively there is room for improvement. 3.2

Full HEC implementation

Knowing the capabilities and limitations of the commercially available turbulence and combustion models, the feasibility of retrofitting an existing slab-reheating furnace with HEC technology has been investigated with CFD. Previous internal studies had shown that the largest benefits are expected for installing HEC technology in a pusher furnace. Therefore this type of furnace has been chosen for the study.

- 12 When assessing the number of burners that can be fitted in an existing slab-reheating furnace one has to take the following considerations into account: •

To maintain the current production capacity of the furnace, a larger number of HEC burners have to be installed. This is because only half of the burners are firing, while the other half extract flue gas.

•

That the skid post pitch determines the burner pitch in the bottom burner zones; otherwise several burners will be firing against a skid post.

•

That the jet diameter of the burner determines the burner pitch and vertical positions of the burners.

Based on these constraints, the conclusion was drawn that theoretically a maximum of 18 pairs of HEC burners could be fitted in the assessed pusher furnace. A schematic of this arrangement is presented in Figure 12.

Figure 12. Schematic of the pusher furnace equipped with 18 pairs of HEC burners

This shows that the HEC burners can only be fitted in the preheating and heating zones. The soaking zone will still be fired with conventional burners for three reasons: •

The heat input in the soaking zone is much smaller than the heat input in the other zones, so implementing HEC in the soaking zone will save a relatively small amount of energy.

•

The jet diameter is larger than the distance between the slab and the roof.

•

Side-firing the soaking zone will hamper controlling the temperature gradient across the length of the slab, which is important for the rolling performance.

- 13 -

When applying HEC burners with a thermal capacity of 5.8 MW each, a comparison can be made of the installed burner capacity in the original arrangement and the HEC arrangement. This comparison is presented in Table 2. Original

HEC

arrangement

arrangement

Preheating zone

82.6 MW

63.8 MW

Heating zone

53.4 MW

40.6 MW

Soaking zone

22.2 MW

22.2 MW

Total capacity

158.2 MW

126.6 MW

Table 2. Comparison of the thermal capacity (based on the fuel input) of the original arrangement and HEC arrangement

The table shows that the installed thermal capacity, based on the fuel input, is reduced by 20%. However a comparison of the total thermal input, based on the fuel and air input, shows that the total thermal capacity is reduced by just 5% (see Table 3) in case of the HEC arrangement. Original

HEC

arrangement

arrangement

104.0 MW

106.3 MW

Heating zone

67.2 MW

54.8 MW

Soaking zone

24.7 MW

24.7 MW

Total capacity

195.9 MW

185.8 MW

Preheating zone

Table 3. Comparison of the total installed thermal capacity of both arrangements (based on air and fuel input)

The semi-industrial experiments have demonstrated the "well-stirred" behaviour of the HEC technology (see paragraph 2.2). The well-stirred behaviour will increase the effectiveness of the heat transfer to the steel slabs in the furnace. The effectiveness of the heat transfer is defined as the ratio of the heat input of the fuel that is transferred to the slab and is a measure of the productivity. Effectiveness: ηe =

Q slab Q fuel

To confirm the statement that the heat transfer to the slabs with the HEC technology will be more effective, the reheating of the slabs has been analysed with CFD.

- 14 To reduce the computational time of the CFD analysis, only an analysis has been made of the heat transfer behaviour of the HEC technology in the heating zone of the pusher furnace. The CFD analyses have been performed using the standard k-ε turbulence model, PDF-equilibrium combustion model and the Discrete Ordinates radiation model. The heating zone with the conventional burners (original arrangement) has been analysed as a reference case. The boundary conditions were chosen such that the total thermal input (air and fuel) was the same as in the HEC case. The results of the CFD analysis of the reference case are presented in Figure 13.

Figure 13. Energy balance of the heating zone equipped with conventional burners

The energy balance shows that the recuperators recover approximately 5% of the heat input of the fuel and 47% of the heat input is transferred to the slab. The energy balance of the heating zone equipped with HEC burners is shown in Figure 14.

- 15 -

Figure 14 Energy balance of the heating zone equipped with HEC burners

The energy balance shows that the recovered amount of heat is approximately 50% of the heat input from the fuel. This is considerably higher than in the case of the original arrangement (see Figure 13) and shows the increased thermal efficiency of the HEC technology. In addition to the higher thermal efficiency, the energy balance also shows that the effectiveness of the heat transfer to the slab is higher. In the original arrangement 47% of the heat input is transferred to the slab, whilst in the HEC arrangement 72% is transferred to the slab. This shows the effect of the uniform heat transfer and confirms the increased productivity of the HEC technology. The uniformity of the heat transfer has also a down side, because it will also increase the wall losses. The latter is due to the fact that the average wall temperature is higher when applying HEC burners. The conclusion can be drawn that the CFD analyses of the heating zone in the original arrangement and the HEC arrangement confirm the increase in thermal-efficiency and higher productivity. The results also show that the lower installed burner capacity in the case of the HEC arrangement (as indicated in Table 3), will not result in a lower production capacity of the furnace.

- 16 4

•

Conclusions

The semi industrial tests with the HEC technology have confirmed the high thermal efficiency, relatively low NOx emissions and high uniformity of the heat transfer.

•

The CFD model validation showed that the commercially available models qualitatively predict the air and fuel jet behaviour with the corresponding chemical reactions and diffusion processes well. The models can still be improved quantitatively.

•

A feasibility study showed that it is theoretically possible to retrofit an existing pusher furnace with HEC technology.

•

CFD analyses of the heating zone have confirmed the increased thermal efficiency and increased productivity of the furnace.

5

Next step: full-scale industrial tests

As a result of the encouraging results of the semi-industrial tests and the CFD analyses carried out, Corus will undertake the next step, i.e. installing two pairs of HEC burners in an existing reheating furnace. The objective of the full-scale industrial test is to perform an endurance test of nine months. During the test period the following areas will be investigated: •

The lifetime of the regenerators and valves

•

The effect of the HEC technology on the product quality

The test is scheduled to start in the first half of 2006. Acknowledgements

The authors wish to thank Jonas Brobäck Adolfi+, Connie Ellul+ and Ferry Frinking* for their contribution to the semi-industrial tests at the IFRF research station. Further the authors wish to thank René Hekkens* and Marco Mancini (TU Clausthal) for their contribution to the CFD analyses. The research described in this paper was co funded by: •

The Dutch National Funding Agency SENTER/NOVEM

•

The International Flame Research Foundation's Members Research Programme.