Velocity field uniformity analysis of an electrostatic precipitator.

Neimarlija, Namir ; Dzaferovic, Ejub

1. INTRODUCTION

The flow field characteristics of an electrostatic precipitator are very important for the particles collecting process inside precipitator box. Usually the problem of an insufficient efficiency of precipitator can be related to no uniformity of the velocity field distribution in the precipitation space. For these reasons, proper design of flow control devices within an ESP is a critical design function. The electrostatic precipitation theory considers that gas flow through the precipitation space should be uniform with respect to the standards, usually ICAC (Institute of Clean Air Companies) conditions required to be fulfilled (Dumont & Mudry, 2003). Modeling in some geometry ratio, usually performered under adiabatic and without voltage conditions, is still common method to obtain the necessary information. An alternative to this approach is the use of numerical methods and the mathematical modeling of the process what is our approach too.

Numerous papers on electrostatic precipitation phenomena exist ranging from the experimental and theoretical investigation electrical characteristics, in the past, to the complex numerical models of ESP's developed during last decade. The most of them research only some aspects of ESP's problem. They include also papers which deal with hydrodynamic problems in electrostatic precipitators. Some of them consider ambient air as media in numerical modeling of ESP (Dumont & Mudry, 2003) or simply offer the review of some ESP's problems (Parker, 1997). Several documents refer to the same ESP as the subject of this analysis: a numerical analysis an older version of the ESP (Teskeredzic & Dzaferovic, 1999), part of thesis (Neimarlija, 2006) and the test measurements data of the actual ESP (Internal documents, 1995-2003). The latter one contains data of the air velocity distribution analysis conducted on the considered actual ESP at Thermo Power Plant (TPP) Kakanj.

The actual electrostatic precipitator of unit 5 in Thermo Power Plant Kakanj consists of two parts (figure 1). Because both precipitators are equal, symmetric related to vertical plane which separates them, we chose one, left side. The main elements of precipitator are the precipitator box, inlet and outlet duct with perforated plates and turning vanes, dust hoppers, the electrode pairs, and so on.

[FIGURE 1 OMITTED]

The main data of the ESP are: number of the sections 4; the precipitator box width and height are 10.2 and 11.3 m, respectively; distance between the collecting electrodes is 400 mm; distance between the discharge electrodes is 495 mm; the flow rate of the gas is 543000 [Nm.sup.3]/h (both sides); the applied voltage is 50 kV.

For obtaining more realistic results, in case of numerical modeling ESP's flow pattern, it is better to present actual geometry as authentic as possible. But, because of real reasons (computer limits), it was necessary to introduce some approximations of the geometry. We neglected the hoppers and the discharge electrodes while the so called "sigma" collecting electrodes being observed as the plain plates (as smooth walls). However, even with these simplifications we can expect valuable and useful results.

2. NUMERICAL MODEL

Mathematical model consists of averaged equations of the mass and the momentum balance. The standard k-s model is used for turbulence modeling. Closed form of these equations is impossible to find for real configurations such as an ESP's complex geometry. The implemented finite volume method (Demirdzic, 1995) uses vectors and tensors expressed through their Cartesian components. Possibility of the use the unstructured mesh with polyhedral control volumes are very attractive properties of the numerical model easing creation numerical mesh of the complex geometry. Also, all dependent variables are stored at the cell center. Central differencing scheme is used for diffusive terms. For convection upwind difference scheme combined with central difference scheme are used. The segregated approach is used to solve the resulting set of coupled non-linear algebraic equations, for each dependent variable what leads to the set of decoupled linearized algebraic equations for each dependent variable. More information about the numerical method implemented here, especially about discretization schemes, verification and efficiency, can be found in (Demirdzic & Muzaferija, 1997). The very efficient computer code is built to perform the calculation.

3. RESULTS OF CALCULATION

In order to realize numerical calculation as first task it is necessary to perform discretization of the domain space. The discretization procedure resulted by dividing the domain space on 578170 control volumes with refined numerical mesh near location of perforated plates. The main support vertical beams are also defined with respect to their significant dimensions and so influence on the air flow current. The minor construction elements contained in the ESP are neglected.

To complete input data for the calculation purpose we need to know boundary conditions and physical properties of the media. The boundary conditions should be as closer as possible to the test conditions by reason for obtaining more comparable numerical values. The volumetric flow rate in amount of 108.3 [m.sup.3]/s is calculated on the basis of averaged velocity [??] = 0.94 m/s and known cross-section of the electrostatic precipitator. Hence, one is obtained the average velocity at the inlet of precipitator of 13.48 m/s. The air density ([rho] = 1.188 kg/[m.sup.3]) and the molecular viscosity ([mu] = 1.824 [10.sup.-5] Pas) of the air are considered as a constant. In the following presentation of results we are focused mainly on the velocity vectors at the inlet precipitator box plane.

In figure 2 are shown contours of air velocity through inlet plane, obtained by calculation and measuring. For the sake of easier comparison, the contour levels of numerical calculation are adjusted to contour levels of measure values. There is some similarity between contours of two velocity fields. We could notice from the contours in the figure 2 that calculation velocities exceed to some extent measured one what is mainly due to the boundary conditions. The influence of vertical central beams on velocity fields is visible on both contours.

For the left precipitator, the experimental analysis showed that the average velocity at inlet plane is 0.94 m/s, and deviation from average velocity is 38.27%, and at outlet plane average velocity is 0.78 m/s and deviation 21.92%. In relation with standard it is clear that these deviations are significant.

[FIGURE 2 OMITTED]

Three top rows from table 1 provide a flow distribution statistics for the ESP for the inlet and outlet plane. Obviously, ICAC 115% conditions are approximately satisfied only for outlet plane, while this deviation at inlet cross-section is above 14%. In second column of table 1 are given the participation percentages of velocities lower of 115% from [??], and in the next column the percentages velocities which value is between 115% and 140% of [??]. In third column are the rest of velocities, part which values exceed 140% of [??].

In three bottom rows from table 1 are given data about another parameter S (variation coefficient) defined as the ratio between standard deviation o and average velocity [??] times 100%. The numerical values of the parameter S are similar to those obtained with actual field measurements. For the inlet cross-section, numerical and measure values of S are 34.92% and S = 38.27%, respectively, while for outlet cross-section in numerical calculation S = 16.67%, and value on the measure basis is S = 21.92%. We see that numerical calculation gives some better values of parameter S. However, these values are also above prescribed value for parameter S of 15%.

The investigation of small velocities was also conducted. The percentage of air velocities higher of 60% and lower then 85% of [??] is about 20% in both planes. The fraction of smaller velocities (under 60% of [??]) is relatively high, near 14%, in case of inlet plane.

4. CONCLUSION

The results of numerical analysis confirmed that there is unsatisfactory uniformity of flow field inside considered actual electrostatic precipitator and the certain flow corrections have to be made. Numerical modeling has advantages because offers detailed data in each control volume what is very hard to obtain by test measurements or by physical model.

By other side, in case of electrostatic precipitators, there is a necessity for development new and/or improvement present numerical methods. The future investigation will go to the direction comprising some others phenomena.

5. REFERENCES

Demirdzic, I. (1995). Numerical Mathematics, Mechanical engineering faculty of Sarajevo

Demirdzic, I. & Muzaferija, S. (1997). Introduction to Computational Fluid Dynamics, Lecture notes, Mechanical Engineering Faculty, University of Sarajevo

Dumont, B. J. & Mudry R. G. (2003). Computational Fluid Dynamic Modeling of electrostatic precipitators, Electric Power Conference 2003, 05 March, 2003

Internal documents of TPP Kakanj (Unit 5), (1995-2003)

Neimarlija, N. (2006). Numerical modeling fly ash precipitation in electrostatic precipitator, Doctoral thesis, University in Sarajevo, Sarajevo, may 2006

Parker, K.R. (1997), Technological advances in high-efficiency particulate collection, Proc.Instn.Mech.Engrs. Vol.211, Part A

Teskeredzic, A. & Dzaferovic, E. (1999), Final Report for the project: High-performance scientific computing in fluid dynamics, (Ref.-No. GM 2048)

Table 1. Statistical parameters for hydrodynamic calculation.

 <+15% +(15-40)% >+40%

Inlet 62.94 22.21 14.84
Outlet 79.13 19.50 1.36

 [bar.v] [sigma] S (%)

Inlet 1.26 0.44 34.92
Outlet 0.96 0.16 16.67