Whatever the plant, kiln configurations are process dependent and vary greatly. The required process temperatures differ considerably, and secondary air temperatures are highly variable as is the firing system employed. This also is often in combination with a wide range of fuel types.

Optimizing the energy consumption of the kiln involves both fossil fuel and electrical energy, and this article is principally concerned with the former. Optimization encompasses minimizing fuel consumption, unburnts, NOx, SO2, and clinker grinding energy.

Clinker with small crystals and sharp boundaries is easy to grind and gives the cement higher early strength. Crystal growth is influenced strongly by the heat transfer from the flame, favorable conditions being rapid heating from calcining to sintering temperature and sudden quench in the cooler to freeze the crystal structure. These conditions are produced by a flame with a high heat flux close to the burner nozzle. Flames with very flat heat transfer profiles give slow rates of heating and large crystals. The resultant clinker is harder to grind and produces cement with poor early strength.

To compensate and meet market requirements, the raw mix is sometimes adjusted, the kiln burnt harder, and the cement ground finer, thus increasing the energy consumption in both the kiln and the grinding mill.

The difference in energy consumption in the kiln and grinding mill between clinker produced by an optimized flame and that produced by a poor flame can be as much as 10%. A poor flame heat flux profile, therefore, imposes a high economic cost as well as a significant increase in atmospheric emissions.

Physical modeling of flames Despite the growth in computer modeling, physical modeling is still the most effective method for determining flame length and shape in rotary kilns. Acid/alkali modeling was developed by Sir William Hawthorne at MIT in the late 1930s and is used to model the combustion process in rotary kilns where fuel/air mixing determines the flame characteristics.

A physical model of the cooler, hood, and kiln is constructed to an appropriate scale in clear acrylic plastic. The fuel is represented by dilute caustic soda solution containing phenolphthalein indicator, while the combustion air is represented by dilute hydrochloric acid. The concentration of the alkali and the stoichiometric ratio of alkali to acid is chosen to represent the correct air/fuel requirement for the particular fuel. The flow of acid is adjusted to simulate different excess air levels, hence determining the relationship between flame length and excess air. The phenolphthalein becomes colorless at the boundary where the mixing is complete, thus the model flame envelope is defined by the colored region. The aerodynamics of the full-size system are reproduced on the physical model thus allowing an accurate simulation of the fuel/air mixing characteristics, and hence the flame length, under representative conditions.

These model results must be corrected since the model is run under isothermal conditions. However, in kilns, considerable changes in temperature usually occur as combustion takes place. This results in a reduction in the gas density and an increase in volume, giving a longer flame in the kiln than in the model. For most practical purposes, the model flame length has only to be corrected for the density changes.

NOx assessments The NOx formation in kiln flames is generally via both thermal and fuel routes (for coal, oil, and petroleum coke, all of which contain fuel nitrogen). Owing to the high flame temperatures that often occur above 2,000 degrees C, thermal NOx is generally the dominant mechanism and typically accounts for around 70% of the total NOx emission dependent on secondary air preheat temperature.

In gas-fired kilns, fuel NOx is absent, so all the NOx is thermal. However, the absence of the fuel variety in gas-fired kilns does not necessarily lead to a reduction in emissions, since gas-flame oxygen concentration and the residence time in the high-temperature zones influence the final thermal NOx emissions.

The formation of NOx is complex and still not a well understood process. Consequently, the modeling of this process is very difficult. Some of the models currently available are capable of predicting the trends in NOx formation with change in flame conditions and fuel type, but the accuracy is poor and sometimes little better than order of magnitude. Currently, the most reliable of methods of predicting NOx emissions from full-scale flames is by empirical scale up from test flames. Fuel and Combustion Technology (FCT) has achieved good results using the data from the test work undertaken by The International Flame Research Foundation for the CEMFLAM 1 Consortium.

In addition, for prediction of NOx in rotary kilns, FCT utilizes a customized version of the FACSIMLE kinetic package produced by AEA Technology. This version consists of a suite of closely related programs for the modeling of complex steady-state and time-dependent chemical reactions, including an extensive NOx modeling capability.

To allow for acceptable predictions to be made in industrial combustion processes, the company has modified the code to take account of gas temperature-time history and fuel/air mixing, which is generated from the associated physical and heat transfer modeling. To date, results have been encouraging, with predictions of emissions from an existing "dead-burned" dolomite kiln being within 10% of measured values. Further validation of this program over a broad range of combustion processes is currently being undertaken.

Validation of modeling It is one thing to produce an effective method and quite another to ensure that its predictions are correct and in agreement with experimental observations. Consequently, considerable effort has been made to validate these computer models. The method is to make detailed comparisons between predictions and experiments; to interpret whatever discrepancies are discovered in terms of computational inaccuracies, inadequacies of the assumptions and imprecisions of measurement; and then to implement improvements that result finally in the reduction of the discrepancies to acceptably small values.

In the real world Modeling can be used to solve problems with existing kilns, optimize the performance of existing kilns, assess the effect of fuel or other process changes in advance of the changes being made or optimize the design of a new plant.

Typically more than one modeling technique is used for a particular application because each technique provides only part of the answer. Within the kiln itself, acid/alkali modeling is used to simulate the combustion, while the zone method of heat transfer is used to predict heat transfer from the flame to the product.

For flash calciners, both techniques can be used together with CFD modeling of the particle trajectories and residence times. The major benefits are reduced costs and increased profits for the kiln operator with reduced environmental impact. The former is attributable to reduced fuel consumption, improved refractory life, and shorter downtime, with potentially greater sales resulting from longer production runs and improved product quality. The reduced emissions are the result of reduced flue gas volumes and less unburned fuel.