Fluent GB
Fluent GB
Dr Patrick MUSCAT, Doctor-Eng. of Marseilles University, dipl. engineer of ENSAE (Sup Aro) Manager of CFD Department, Fives Pillard, Marseille (France)
1/9
Rotary kiln for nickel In the plant concerned, a problem of ame shape and stability was encountered. CFD modeling was used to understand, then correct the problem:
The analysis showed that a bad combustion air velocity distribution in the kiln hood was probably the cause of the problem:
2/9
Several simulations have been carried out to dene the shape of devices necessary to optimize air ow:
We can see that once the air distribution has been deeply modied in the kiln hood, then the problem of the uncontrolled ame air root disappeared and the temperature prole improves:
Before optimisation
After optimisation
3/9
Studies on cement precalciner a) SNCR system on a rising duct downstream a rotary kiln The CFD is an essential tool for the engineering phase of a SNCR (NOx reduction) system applied to existing plants. Such an engineering process can be broken down as follows: First, further to site temperature measurements, we determine the appropriate reaction zone. Then, we use CFD for calculating the ue gas balance and simulate velocities and temperature proles in the reaction zone:
Temperature and velocity proles at kiln outlet (Fives Pillard Neutrinox SNCR system)
The second phase consists in modelling the droplets of (urea/water) or (ammonia/water) solution, and their vaporizing in the ue gas stream. The number of injectors, their location point and their droplet velocities may be optimized:
The target consists in covering the entire cross section of the ue gas duct with the injected urea / ammonia spray.
4/9
Higher proles
Depending on the CFD modelling conclusions, the urea (or ammonia) / water solution is injected through 6-10 atomizing guns installed on 1, 2, or 3 levels in the calciner. Usually two levels are sufcient. The rst level aims at covering the maximum surface with the longest resident time, while the second level aims at optimizing the reaction and efciency. The injectors optimization includes the choice of emulsion velocity (an atomizing uid is used, in general compressed air), the choice of orices, number and angle. The special atomizer is patented (n FR 0605355 of 06/15/2006). b) Another example of SNCR system on suspension preheaters application concerns a cement plant located in Vietnam:
Due to the high capacity of the cement line (5,000 tpd), it has been installed on two levels with six injectors each. Another example in Italy shows that data collected during the rst commissioning period obliged us to reconsider the hypothesis taken at the project stage, thus modifying the position of the injectors. They were moved downwards in the calciner chamber:
Precalciner (Italy): CFD model and location of urea injection levels 5/9
The rst injection level position corresponded to a too low residence time which was inducing low SNCR efciency. The CFD helped to redene the new injection level position with higher residence time leading to an increase of urea consumption efciency: The analysis of ue gas ow shape in the precalcinator showed that -as a matter of fact- a big recirculation zone was increasing the mixing time in a convenient temperature window:
Temperature prole
Velocity prole
With such a modication, an increase of about 10% of deNOx reaction efciency was reached, corresponding to a 10% urea consumption saving. c) Studies on precalciner burners A comparative study of the effect of primary air used on precalciner burners has been carried out, and allowed us to conclude that the primary air induces a reducing combustion zone around the burners which has an big effect on the temperature prole:
Further studies on precalciners are pending; they aim at looking deeper into the position of burner air inlets and to try to understand whether the precalciners operation can be improved in the future by simple modications.
6/9
Rotary cement kilns: comparison of ame shapes CFD simulations can be used intensively to analyse ow patterns around the burner. Cold CFD simulations allow to visualize and compare ow patterns. A rst example concerns the comparison of secondary air ow shape as a function of the burner tip position inside the kiln. It conrms the site results i.e. the penetration of secondary air inside the ame depends on the burner tip position:
This study (also) allows to prove that there is no recirculation zone around the burner ame: it is clear that no recirculation eddy can exist around the ame:
Another example concerns the effect of swirl with 2 different burner congurations (single annular outlet or double outlet). The comparison of the velocity prole at the burner outlet clearly shows that the double air outlet is better than a single air outlet since the ame stays under control and there no risk of ame opening (wide ame) leading to kiln shell overheating as well as sulphur volatilization.
The CFD modelling allows also to analyze the loss of velocities at the burner outlet. The simulations have been carried out with the same air fan pressure (250 mbar). It clearly appears that with a single primary air outlet there is a loss of velocity: The useful velocities at burner outlet are about 100 m/s, due to the enlargement of the burner cross section, while it remains at about 250 m/s when a separated second slot-designed:
yellow is # 250 m/s
Cold modelling of different burner congurations (axial primary air tip) show that in increasing the burner slot effect promotes hot secondary air suction into the ame, thus intensifying the ame:
AXIAL TIP WITHOUT SLOTS fully open closed AXIAL TIP WITH SLOTS fully open closed
Visualisation of secondary air introduction into the ame (the colour represents the temperature of the mix)
8/9
In order to conrm the above described cold modelling, a complete modelling in combustion has been achieved. The geometric size of the kiln is represented below:
Only the rst 50 m from the burner have been modelled due to computational considerations. The kiln rotation is taken into account. The details of the burner geometry have been taken into account. The fuel is coal. The secondary air temperature has been chosen at 800C. The clinker has been modelled as a solid entity in motion, and the heat exchange has been modelled. The simulations concerning the combustion of coal particles include various axial primary air injections for the same burner, thus several different SLOT EFFECTS for same operational conditions. The conclusion appears as follows:
CONCLUSION
The belief that CFD modelling cannot bring a real progress in combustion for rotary kilns or cement precalciners now appears completely obsolete. It is clear that comparisons based on different inlet [combustion/kiln fuel] parameters give precise data and allow to choose which conguration can improve a real situation. The site results obtained allowed to conrm that the provisions were sufciently reliable. Simulations with liquid uid ows do not hold any real interest since CFD modelling is now an efcient tool and the time available for modelling work is decreasing rapidly. Such a tool allows to design a combustion system associated with a real kiln which, as a matter of fact, may allow to simplify adjustments at site. Even more interesting is the fact that such a design can be tuned to the priority targets of the user, which may differ from one kiln to another (for example : clinker reactivity, high LSF raw material, limited NOx emission, high percentage of alternative solid fuels, use high ash low volatile coal, etc...). Such CFD studies are now a common way of working, and bring feedback from experience every day. This is why we hope that , thanks to such new tool, real progress on the calcined product quality or on the polluants emission (NOx, CO, and CO2.) will be more easily achievable.
9/9