The rising demand for high quality castings necessitates that vast amount of manufacturing knowledge be incorporated in manufacturing systems. Rotary furnace involves several critical parameters like flame temperature, preheat air temperature, rotational speed of the furnace, excess air percentage, melting time, fuel consumption and melting rate of the molten metal which should be controlled throughout the melting process. A complex relationship exists between these manufacturing parameters and hence there is a need to develop models which can capture this complex interrelationship and enable fast computation. In this paper the applicability and the relative effectiveness of Regression and Numerical technique for modeling and optimization of rotary furnace parameters have been investigated. The results obtained by these models are found to correlate well with the experimental data obtained from the Rotary Furnace.
Description of furnace The Rotary Furnace is very simple melting unit consisting mainly of a drum of required size having a cone on each side lined with refractory, fire bricks or ramming mortar generally having alumina as a constituent. This drum is placed on rollers so that they may be either locked or slowly rotated about their central axis. The rollers are driven by a small electric motor. At one end of the drum, a suitable burner is placed with appropriate blower system and combustion gases exit from other end. This drum or horizontal cylinder is flanked by two conical portions on both sides. One of the cones accommodate the burner whereas from the other cone hot flue gasses exit. Charging of the iron for melting is also done from this side. The cone on one side can accommodate different types of burners using the Light Diesel Oil (L.D.O.). The tap hole is located in the cylindrical wall halfway between the ends. This tap hole is used to take out the molten metal but it is kept closed during the melting of metal.
There are a number of variables controllable to varying degrees which affect the quality and composition of the outcoming molten metal. These variables, such as, revolutions per minute of the drum, melting time, fuel consumption and melting rate play significant role in determining the molten metal’s properties and should be controlled throughout the melting process. However, even an experienced operator may find it difficult to select the optimum input parameters which would yield ideal molten metal and often he may choose them by guessing which may not be effective and economical
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
Process of melting the charge is carried in following steps: (1) Preheating of oil and furnace (2) Charging (3) Melting (4) Tapping (5) Inoculation (6) Pouring
3. Experimental setup and data collection
The experiments have been conducted on self designed and developed furnace, installed at Harbhajan Singh Namdhari Enterprises, Industrial Estate, Nunihai, Agra as shown in Figure 1.In the experimentation, 200 kg of the charge is melted in the rotary furnace. A Circular burner is used for burning Light Diesel Oil (L.D.O.) as fuel. By rotation the major heat transfer takes place by radiation. Some heat transfer takes place by conduction when charge comes in contact with heated refractory, and part of it by convection, in initial stages. Initially furnace starts from room temperature, therefore more fuel is consumed in first heat and for later heats the fuel consumption is reduced. Numbers of experiments were conducted for excess air consumption and the observations for time and specific fuel consumed in melting, were recorded.
4. Experimental investigation (1)Operating furnace under existing conditions of operation
The furnace was operated at 2.0 rpm, as per existing conditions. The charge per heat is 200.0 kg. In first heat, as furnace started from room temperature, more air was required, the flame temperature, preheated air temperature, and melting rate were less, but time and fuel consumption were more. In subsequent heats, the air was reduced, flame temperature, preheated air temperature and melting rate increased whereas the time and fuel consumption reduced. Observations taken during the experiment are given in table1 (1 liter LDO =9.9047kwh)
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
It is clear that under existing conditions of operation the maximum flame temperature is 13500C, the specific fuel consumption (LDO) in melting is 0.415 liters/kg (415 liters/tone) which corresponds to energy consumption in melting 4110.45 kwh/tone As per TERI (The Energy Research Institute) norms the energy consumption in melting, for oil fired furnaces is 2221.60 kWh/tone. This energy consumption is 45.95% (app.46%) higher than TERI norms. The flame temperature depends on all other input parameters as shown in table 1.If flame temperature is increased the energy consumption automatically shall be reduced.
6. Further Experimental Investigations
Effect of reducing combustion volumeand oxygen enrichment of less than theoretically required air, preheated, using compact heat exchanger, rotating furnace at optimal rotational speed, on Flame temperature and Energy consumption
If the combustion volume is more than more fuel and time shall be required for reaching to a certain temperature. Hence it is thought to optimize the combustion volume by reducing the amount of air and supplying oxygen externally. Several experiments were conducted, gradually reducing air to its theoretical requirement and even lesser in steps of 5.0 to 10.0% and supplying oxygen externally in steps of 1.0 to 2.0 % and its effect on flame temperature, time, fuel, melting rate, and fuel consumption was studied. The effect was significant only when air was reduced approximately to 75.0% of its theoretical requirement and approx 7.0% oxygen was supplied externally. The series of experimental investigations conducted are shown in following sections.
6.1. Experimental investigations
(2)–Effect of 6.9 %oxygen enrichment of75.375.4% of theoretically required air preheated up to 460.00C, using compact heat exchanger, rotating furnace at optimal rotational speed of 1.0 rpm, on flame temperature, time, fuel, melting rate, and specific fuel consumption Numbers of experiments were conducted, rotating furnace at optimal rotational speed 1.0 rpm, with 6.9% oxygen enrichment of 75.375.4% of theoretically required air, and using compact heat exchanger, preheating LDO to 700C. The effect of above on flame temperature, time,
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
fuel, melting rate, specific fuel consumption (lit/kg) and energy consumption is shown in table.2
Table 2: Effect of 6.9% oxygen enrichment of 75.375.4% of theoretically required air preheated up to 460.00C, using compact heat exchanger, rotating furnace at optimal
rotational speed of 1.0 rpm, on flame temperature, time, fuel, melting rate, and specific fuel consumption.
On basis of optimal input parameters the specific fuel consumption has reduced from 0.415 liters/kg (415 liters/tone) to 0.260liters /kg (260 liters/tone).The reduction in specific fuel consumption is 155 liters/tone which corresponds to reduction in energy consumption of 1535.22 kwh/tone i.e., by 37.349%. The energy consumption measured in terms of specific fuel consumption has reduced significantly
8. Modeling of rotary furnace
1. ANN method has been adopted for modeling The inputs parameters are
2. Charge –fixed 200 kg 3. Fuel (liters)
4. Preheated air volume (combustion volume) 5. Oxygen consumption (oxygen supplied for combustion)
6. Time required to melt charge of 200 kg (time/heat) 7. Flame temperature
8. Preheated air temperature. These parameters are to be optimtsed for minimum specific fuel consumption. A.N.N. Model of Furnace is shown in figure 2.
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
The rotary furnace datas, used to train and test the model have been extracted from actual experiments conducted on a 200 kg rotary furnace as given in table 2. A feed forward modeling neural network was constructed to model the rotary furnace data as given in MATLAB. The network used seven inputs, two hidden layers of seven neurons (nodes) in addition to one node output layer as shown in fig 4.The network as mentioned above has yielded comparatively better results. The network could predict output parameter with about 12% error. All input and output data’s are scaled so that they are confined to subinterval of 01.in this case each input or output parameter P is normalized as Pn before being applied to the neural network. The results of tests are shown in following graphs .the conclusion from these modeling experiments is therefore that the tested network appears to constitute a workable model for modeling of rotary furnace parameters. The discrete structure is shown in figure 3
Figure 3: Discrete structure of ANN network modeling
The output (specific fuel consumption) as given by the above programme is 0.2800lit/kg. The epoch error curve and feed forward model is shown in figs 4 and fig 5 respectively.
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
Figure 4: Epoch error curve Figure 5: feed forward model
9. Test running of model
The model so created has been checked for its validity by test running it. The model has been created as per actual experimental input parameters at sn 2 of table 2
For actual experimental input parameters sn 6 of table2. The input parameters selected are
p6= [458, 1754, 30.5, 52, 393.44, 36.60 ,426.7], the output is 0.260.The results are shown in Fig. 6
. Figure 6 : Test running of model for experimental data 6
10. Optimization
Engineering optimization is a rigorous mathematical approach for identifying a set of designed alternatives and selecting the best within that set, for optimal value of desired output. Optimization can be defined as the process of finding the conditions that gives the minimum or maximum value of a function. On basis of model created the optimal input parameters identified for optimum specific fuel (energy) consumption are given in table 3
Table 3: The optimal input parameters for optimum specific fuel (energy) consumption
Rp m
Preheat ed air
Flam e
Time min
Fuel liter
Meltin g
Oxygen Consum
Preheate d air
Specific fuel
Energy consum
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
Simulation is the process of designing a model of a real system and conducting experiments with this model for purpose of understanding the behavior of operation of the system. ANN Artificial neural networks has been employed for simulation The programme was run with above input parameters. The graphical representations and results of simulation of effect of various input parameters on specific fuel consumption is shown in following figures and tables
11.1 Effect of preheated air temperature on specific fuel consumption
The simulation of effect of preheated air temperature, in range of 400500 0 C, on specific fuel has been carried out and shown in fig7.specific fuel consumption is in liter/kg The results are presented in table 4
Figure 7: The effect of preheated air temperature (400500 0 c) on specific fuel consumption
Table 4: The effect of preheated air temperature (400500 0 c) on specific fuel consumption
11.2 Effect of flame temperature on specific fuel consumption
The simulation of effect of flame temperature, in range of, 17001800 0 C on specific fuel has been carried out and shown in fig 8. The results are presented in table 5
SN 1 2 3 4 5 6 7 8 9 10 11 Preheated Air temp. 0 C
11.3 Effect of time/heat on specific fuel consumption
The simulation of effect of time/ heat range 2835 minutes on specific fuel has been carried out and shown in fig.10 The results are presented in table 6
Figure 9: The effect of Time/heat (2835min.) on specific fuel consumption
Table 6: The effect of Time/heat (2835min.) on specific fuel consumption sn 1 2 3 4 5 6 7 8 9 10 Time/heat min. 28 29 30 30.5 31 31.5 32 33 34 35 Specific Fuel lit/kg
11.4 Effect of fuel/heat on specific fuel consumption
The simulation of effect of fuel/ heat range 5058 liters on specific fuel has been carried out and shown in fig.11. The results are presented in table 7
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
11.5 Effect of melting rate on specific fuel consumption
The simulation of effect of melting rate range (340400kg/hr).on specific fuel has been carried out and shown in fig.11. The results are presented in table 8
Figure 11: The effect of melting rate ((340400 kg/hr) on specific fuel consumption
Table 8: The effect of melting rate ((340400 kg/hr) on specific fuel consumption
11.6 Effect of oxygen consumption/heat on specific fuel consumption
The simulation of effect of oxygen consumption/heat range (3542m 3 /heat) on specific fuel has been carried out and shown in fig.12.The results are presented in table 9.
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
11.7 Effect of preheated air volume/heat on specific fuel consumption
The simulation of effect of preheated air volume/heat range (400500m 3 /heat) on specific fuel has been carried out and shown in fig13. The results are presented in table 10
Figure 13: The effect of preheated air volume/heat(400500m 3 /heat) on specific fuel consumption
Table 10: The effect of preheated air volume/heat (400500m 3 /heat) on specific fuel consumption
sn 1 2 3 4 5 6 7 8 9 10 11 Preheated Air volume m 3
The optimal values of input parameters on basis of simulation using ANN method is given in table11
Table11: The optimal values of input parameters on basis of simulation using ANN
parameter Preheated air temp 0 C
Flame temp 0 C
Time min
Fuel liter
Melting rate kg/hr
Oxygen Cons.m 3
Preheated air volume.m 3
Optimal value 470.0 1790.0 28 50 400 35.0 400
On basis of these optimal values of input parameters the programme was again trained. Best training performance curve is shown in fig14 and régression curve in fig.15.The simulated output is 0.2502 lit/kg.
Figure 14: Best training performance curve Figure 15: régression curve
13. The another set of input parameters
The theoretical (infeasible) l values of input parameters on basis of simulation using ANN method is given in table12
Table 12: Theoretical (infeasible) l values of input parameters on basis of simulation using ANN
parameter Preheated air temp 0 C
Flame temp 0 C
Time min
Fuel liter
Melting rate kg/hr
Oxygen Cons.m 3
Preheated air volume.m 3
Optimal value 500.0 1800.0 30.00 51.0 400.00 37.5 400.0
On basis of this (infeasible) optimal set of input parameters the programme was again trained. Best training performance curve is shown in fig. 16 and régression curve in fig 17.The simulated output is 0.240 lit/kg
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
Modeling, Optimization and simulation of input parameters have been carried out using artificial neural network. The valid model is created which can further be used for simulation.
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
Regression techniques have been used to develop a mathematical relation between the specific fuel consumption and all other input parameters. The effect of individual input parameter on specific fuel consumption has been evaluated and presented in form of graphs and tables.
On basis of actual experimental set of input parameters, simulated specific fuel consumption, is 0.264 lit/kg, and energy consumption is 2.6148 kwh/kg, which is 1.538% more than the actual experimental one.
The simulation has also given another better set of feasible input set of parameters, on basis of which simulated specific fuel consumption, is 0.2502 lit/kg, and energy consumption is 2.4781 kwh/kg, which is 3.769 % less than the actual experimental one.
It has further explored the possibility of another set of input parameters. The simulated specific fuel consumption is 0.240 lit/kg, and energy consumption is 2.377 kwh/kg which is 7.692% less than the actual experimental one. This set of input parameters, is only theoretical and practically not feasible. Under existing conditions of experimental investigations Preheated air temp.of 500 0 C and flame temp.of 1800 0 c can not be achieved .
The Artificial neural networks, is capable of modeling, optimizing and simulation of LDO fired rotary furnace. This can be used for any size of rotary furnace. The maximum variations between simulated and experimental specific fuel consumption is +1.538% and between feasible and actual experimental one is 3.769 %. These both variations lie within permissible limit of ±10%, hence are acceptable. It also establishes the authenticity of experimental set up.
16. References
1. Levis W.W. (1947):“Variables affecting carbon control in cupola” transactions of AFS 55, pp 626632.
2. Pehlke R.D (1963): “Thermo chemical model of computer prediction of cupola performance” transactions of AFS 1963, l71, pp 580587
3. Landefeld C.F.and Katz S (1976): “A dual stream model of carbon pick up based on carbon activity” Cast Metals, 3(4), pp.163171
4. Sahajiwala V, Pehlke R.D (1992): “Experimental investigations and mathematical modeling of carbon transport in a cupola” transactions of AFS, 100, pp. 343352.
5. StanekV, KatzS, Landefeld C.F, Bauer M.E, (1999): “The AFS cupola process model a computer tool for foundries” Modern Casting, 13(5), pp 4143
6. Karunakar D.B. and.Dutta G.L (2002): “Modeling of Cupola furnace parameters using Artificial Neural Networks” Indian foundry journal, 48(5), pp. 2939.
7. Syamal M.C (2004): “Thermo chemical model for computer prediction of cupola performance” M.Tech. Thesis, Mech. Engg. Deptt. IIT Kharagpur,
INTERNATIONAL JOURNAL OF APPLIED ENGINEERING RESEARCH, DINDIGUL Volume 2, No 1, 2011
8. Jain R.K, Gupta B.D (2008): “Mathematical modeling of critical parameters of rotary furnace” Thirteenth annual and first International conference of Gwalior academy of mathematical sciences, 1013 th January Agra
9. Jain. R.K., Gupta B,D (2008): “ Mathematical Computation of flame Temperature and its effect on performance of LDO fired Rotary Furnace” Thirteenth Annual and First International Conference of Gwalior Academy of Mathematical Sciences 1013 th January Agra
10. Jain R.K, Singh R, (2008): “Modeling and optimization of rotary furnace parameters using Regression & Numerical Techniques” 68 th world Foundry Congress, Feb 710, Chennai
11. Jain R.K, SinghR, (2008):“Modeling and optimization of critical parameters of rotary furnace using computational techniques(Excel solver)” Indian foundry journal, 54,(.3) pp. 2834
12. Jain R.K, Gupta B.D.,Singh Ranjit (2009):“Effect of oxygen enrichment & reducing air energy consumption , emission level, and performance of LDO fired Rotary Furnace ”Indian Foundry Journal 55 (12) pp. 3034
