ANALYSIS AND OPTIMIZATION OF CYCLONE SEPARATOR SYSTEMS IN A PFBC DEMONSTRATION PLANT M. A. Gil & C. Cortés Centro de Investigacién del Rendimiento de Centrales Eléctricas (CIRCE) c/ Maria de Luna, 3 50015 Zaragoza, SPAIN Telephone: (+34) 976 76 20 34 Fax: (+34) 976 73 20 78 E-Mail : tdyfqdb @posta.unizar.es D. Martinez Investigacion y desarrollo Escatrén PFBC power station Empresa Nacional de Electricidad, S. A. Escatron, Zaragoza (SPAIN) Telephone: (+34) 976 17 00 11 Fax: (434) 976 17 0271ANALYSIS AND OPTIMIZATION OF CYCLONE SEPARATOR SYSTEMS IN A PEBC DEMONSTRATION PLANT A. Gil & C. Cortés Centro de Investigacién del Rendimiento de Centrales Eléctricas (CIRCE) University of Zaragoza Zaragoza (SPAIN) D. Martinez Investigacion y Desarrollo Escatrn PFBC power station [Empresa Nacional de Electricidad, S.A. EscatrOn, Zaragoza (SPAIN) 1. Abstract This paper presents a pilot-scale gas cleaning system, designed and developed by CIRCE and ENDESA at Escatron PFBC power plant (Zaragoza, Spain). It is a cold-flow model of the two-stage filtration process performed in nine streams in the full-scale facility. Similarity criteria have been considered in the design, intending to reproduce the hydrodynamic behavior of the gas and particle phases. Operational parameters affecting cyclone performance, such as gas-particle ratio, gas flow rate, particle mean diameter and pressure drop have been examined, In PFBC cyclones, solids are removed (along with some quantity of transport gas) through a suction nozzle at the bottom of a dipleg. The performance of this device is scarcely reported in the open literature. Therefore, specific analysis dealing with flow patterns, vortex description and dipleg pressure drop have been made. The relationship between flow characteristics in the dipleg and separation efficiency and operating stability has been tentatively addressed. The study of the behavior of the pilot plant, including observation and measurement of flow patterns in the cyclone and dipleg, will allow charactérizing the performance of the full-scale ash removal system, establishing safe limits of operation, and testing design improvements. 2, Introduction An optimum performance of hot gas filtration systems in PFBC plants is essential in order to ‘maintain gas turbine working conditions and particle emissions. Entrainment of bed particles by combustion gases may lead to erosion and fouling in downstream equipment, and even to severe damage of gas turbine blades. In addition, fly ash can also produce corrosion, due to metal alkali compounds. Cyclone separator systems offer the best operating performance for removing particles from high temperature and pressure suspensions. Flue combustion gases from pressurized beds represent a harsh environment, dictating this as the only solution commercially available nowadays for PEBC plants. These systems are simple, compact and low cost, have no moving parts, can resist veryextreme conditions and have a relatively high collection efficiency. However, despite the experience accumulated as a result of their wide range of industrial applications, the design of cyclone separators is involved, because of their complex hydraulic behavior. Besides that, their major disadvantage is a low efficiency for particles generally below 5 jim of mean diameter. Consequently, cyclones are used as pre-cleaners, and are often arranged in series. In Escatrén PFBC power plant, the hot gas filtration equipment is a two-stage process performed in nine streams between the bed and the gas turbine. The cyclones are high efficiency, Van Tongeren's type, as shown in figure 1. The unit consist of a tangential inlet, a cylindrical body with a conical base, an axial outlet for clean gases and an outlet port for solid particles at the lower part. The main difference between standard and Escatrdn cyclones is the method of removing collected solids, which is based on pneumatic conveying. As it is shown in figure 2, the solid extraction bin is replaced by a dipleg similar to those found in catalytic cracking. At the bottom of the dipleg there is a suction nozzle, through which solids are evacuated along with some amount of transport gas. In contrast with other devices, such as a series of pressure-tight lockhoppers, the extraction of solids by pneumatic conveying improves cyclone efficiency and allows reliable handling and cooling of ash particles with low energetic costs. Ash and combustion gases exit the pressurized bed at nearly 800 °C and 11 bar. The gas and solid mass flow rates depend on load, coal and sorbent characteristics and other operating variables, in particular those related with fluidization. These parameters have a strong influence on the separation efficiency of the cyclones, and there are not well-established theories able to achieve an accurate and complete prediction. Operating experience at Escatrén adds other relevant phenomena that cannot be addressed from a theoretical standpoint, such as the effect of sintered deposits on separation efficiency and the existence of flow unsteadiness in the dipleg and suction nozzle affecting the capacity of the ash conveying lines. ‘Therefore, a thorough investigation must be based on experimental methods. The present project is conceived to characterize and improve the performance of the ash removal system of Escatrén power station through experimental studies in a pilot plant facility, Pressure and temperature conditions permit designing a cold-flow model that approximately preserves dynamic similarity. This offers the possibility of reproducing the actual behavior of the full-scale equipment, thus constituting an excellent framework for operational and design studies. 3. Cyclone performance Cyclones are, basically, centrifugal separators. They simply transform the inertia force of gas- particle flows to a centrifugal force by means of a vortex generated in the cyclone body. The particle- laden flow enters tangentially at the upper part and passes through the body describing the vortex. Particles are driven to the walls by centrifugal forces, loosing its momentum and falling down to the cyclone leg. In the lower section, the gas begins to flow radially inwards to the axis and spins upwards to the gas outlet duct. ‘The main variables describing cyclone performance are pressure drop, collection efficiency, and solids extraction capacity.3.1, PRESSURE Drop. Pressure drop through the cyclone determines the flow rate that will be attained for a given blowing capacity or total pressure difference. Pressure losses in cyclones are mainly a consequence of wo factors: 1) Wall friction 2) Rotational kinetic energy of the vortex The dominant contribution is the second, but, logically, it cannot be reduced, since it also accounts for the efficiency of the separation process. Main variables affecting cyclone pressure drop are geometry, solid-gas mass flow ratio, and inlet gas velocity and density. Based on dimensional analysis and experimental studies, many calculation formulae have been reported in the literature, although none of them can exactly predict the pressure drop for all cyclone designs. 3.2, COLLECTION EFFICIENCY The collection or separation efficiency is most properly defined for a given particle size Fractional efficiency curves have then the generic features shown in figure 3. Experience shows that collection efficiency of cyclone separators increases with 1) Increasing particle mean diameter and density 2) Increasing gas tangential velocity 3) Decreasing cyclone diameter 4) Increasing cyclone length 5) Extraction of gas along with solids through the cyclone leg Several theories have been developed to predict collection efficiency in cyclones through adequate correlation equations (1]-{5]. However, as in the case of pressure drop, there is not a model of general validity. Moreover, our case will possess uncommon features, arising from the flow inside the dipleg. For instance, itis a known fact that diplegs induce the gas vortex to run the full length of the cyclone, increasing its efficiency (6]. Consequences of the suction nozzle's driving the downward flow of gas and particles have been not reported in the literature. 3.3. CAPACITY OF THE ASH CONVEYING LINE Cyclone-dipleg hydrodynamics determines the solid-gas flow ratio inside the dipleg and through the ash suction nozzle. Escatrén experience shows that there are some blockages of the ash conveying lines that can be possibly attributed to a) abnormally high steady values of this parameter and b) dipleg flow instability. Conversely, high velocities in conveying lines may also occur, causing erosion of elbows and pipes. These aspects of the performance of PFBC cyclones remain largely unexplored. 4 Pilot plant design Values of density and viscosity of hot pressurized gas and elementary considerations on Pertinent dimensionless parameters suggest that a near-ambient pressure, cold-flow, scaled-down model will reproduce the hydrodynamic behavior of the full-scale ash removal system. This method of investigation has also the apparent advantages of low cost, low power requirements and relativelyeasy operation. In order to determine the scale of the model and to assess the reliability of the method, an analysis of dynamic similarity has been made, starting from what is known about the performance of this kind of systems, 41s CALING CRITERIA Qualitative models do exist for explaining cyclone efficiency and pressure drop, so that the relevant dimensionless numbers are known, On the contrary, flow patterns inside the dipleg can only be inferred, and the analysis should rely on approximate criteria, 4.1.1. COLLECTION EFFICIENCY ‘The collection of ash particles can be though as a result of the forces acting upon them: centrifugal, drag and gravitational. These can be expressed respectively as ad? ve Fo =p, 1 cE Pee a nd? Fp =Cp(Re)~ Ps Q) a Fo =f pak 8) In the last equation, only the difference of radial velocities has been taken into account. Particle and gas velocities are closely related. Therefore, an adequate measure can be the gas velocity, for instance, at the cyclone inlet. The radius of rotation of a particle is approximated by a geometric dimension, such as the cyclone diameter. The collection efficiency also depends on the solid concentration or mass loading ratio at the cyclone inlet, defined as the ratio of solid to gas mass flow rates, C, ig, It is usually observed that « high solid concentration improves collection efficiency, although it reduces the strength of the vortex, This can be attributed to particle agglomeration and to large particles’ sweeping out smaller ones as they move to the cyclone walls. For these reasons, and taking into account Eq.(1)-(3), the collection efficiency can be expressed as a function of the following variables: 11> Feng Ps Hg-s¥inyDes8sCs_in) (4) In order to identify dimensionless groups, this generic equation is rewritten in nondimensional form. For instance, selecting v, ix , py and d as independent variables, Ain y= jo Ed Pala De Pec, a Va’ Hyd 'Py’ ‘The first two independent groups are usually rearranged to read n= fer| Fr Stk 22 PC, ig © dp where Fr = gD,/v3, is the Froude number and Stk = p,d7vin/1,Dz is the Stokes number.1,2, PRESSURE DROP As a first approximation, pressure drop can be divided into two independent terms, one accounting for pressure losses due to a clean gas flow and other depending on solid concentration at the cyclone inlet. As explained above, the gas contribution is almost independent of viscosity, hence AP gay = for geometry. Py Vin) a Solid concentration inversely influence literature [7] is a slowing-down of the flow, and the subsequent decrease of the strength of the vortex inside the cyclone body. Taking both contributions into account, and expressing the result in s cyclone pressure drop. The reason pointed out in the nondimensional form, AP > Pe ¥n/2 fen(geometry,C, in) co) 4.1.3. DIPLEG FLOW There are no specific models of this flow, and a rigorous, full account of all the dimensional variables would be unpractical, since it would involve both cyclone and dipleg. Therefore, an approximate analysis, resorting to reasonable and simple models of the physical situation, is appropriate. Flow patterns inside the dipleg may be expected to possess the following features: — Particles slid down on the dipleg wall, following an helical path of variable pitch _— The gas flow generally follows two paths: near the wall, it moves downwards, with tangential and axial velocity components, whereas itis reversed at the leg core, forming an upward flow. ‘A partial and approximate description is the following equation, which can be obtained by applying a mechanical energy balance to a differential element of diameter Dy and height JH: IWroe = IWgin_gas + OWein_sotid ~ OW grav + OW pic_sotia + OW ric_gas (8) ap = ragvdv + aud ~(ty + 1 )QOH + righ, due hth ae (9b) Ps sing ** Dy 2 where the meaning of the different terms is: total energy required to maintain the movement (AWict), variations due to gas (AWein_gas) and solid (@Wkin_ sola) Kinetic energies, potential energy (IW,rav) and friction of particles (AWric_solid) and gas (OW;ric_gas). Dividing Eq, (9) by the dipleg gas flow rate and nondimensionalizing we obtain 2 AP AH hy (#4) +6,40% AO) v ghH |, AH =4C, 201+ 6) 857 +4, (10) 7/2 "'D, sinoe + Ca) + Ae eT The constant parameters appearing in this pressure drop equation reveal the relevant dimensionless groups. These are geometric ratios, particle concentration in the dipleg Csy, friction coefficient of the solids A; and friction coefficient of the gas Ag. According to the relative importance of the different contributions, the latter can be expected to have a negligible influence. Cyj will bea result of the hydrodynamic behavior of the system and its values could be reproduced at the pilot plant facility through adequate adjustment of the operating variables. Finally. corrections must beconsidered to account for differences in the particle friction coefficient, which depends on solid characteristics and dipleg material 4.2. PILOT PLANT DESIGN ‘A primary criteion was to preserve geometric similarity. As a result of practical considerations, all geometric dimensions were scaled-down by a geometric factor 1/5. With reference to Eqs (6), (8) and (10), the nondimensional parameters that were kept constant between the pilot plant and the full- scale installation are shown in table 1. Resulting operating conditions for the pilot-scale installation are summarized in table 2. Table 1 Dimensionless parameters Cyclone Dipleg Foude number, Fr = gD./v3, Froude number, Fr = gH/v, Inlet loading ratio, Cs in Loading ratio, Cy Stokes number, Stk = p,dvin/t1gD. Density ratio, p/py Variable Inlet Outlet Suction Nozzle kgls "0.189 0.1867 0.00233 rh, kgs 0.014 0.0028 0.01372 TC 20 20 20 P (gauge), bar 2.211 2.18 2.21 v, mis 13.42 16.52 0.122 G 0.076 0.00149 6.1 “5. Pilot plant description and layout Fig. 4 is a flow diagram of the pilot plant facility. It is made up of the following components: — Two air-pressure supplies: main (L1) and a ,secondary line (L9) for fluidization of the particle discharging hopper. — Pressurized storage tank (T1) and variable speed rotary valve connected to a Venturi nozzle. — Experimental primary cyclone (T2), equipped with a dipleg made of methacrylate (T3). — Secondary cyclone with a solids-collecting bin (T4). — Two sedimentation chambers with fabric filters for complete collection of particles and cleaning of the exhaust air (TS and T6). — Control of air flow by flow meters and valves downstream the sedimentation chambers.6. Test program 61 CYCLONE PERFORMANCE CHARACTERIZATION, SIGNIFICANT PARAMETERS AND SAFE OPERATING LIMITS. The purpose of this activity is to demonstrate the similarity beoween the pilot plant and the full- size installation. Stable operating limits in terms of gas and particle flow rates will be also determined. Studies to be completed at this stage are the following. — Effects of gas flow rate through the suction nozzle on ash conveying system. The aim is to simulate cyclone dipleg performance when decreasing the flow of transport gas through the suction nozzle. This will simulate gas leakage in the ash conveying lines, which is supposed to be one of the underlying reasons for their plugging. The effects on cyclone efficiency will be also evaluated, in order to understand the implications of the operation under this abnormal condition — Effects of the ash flow collected by the cyclone and the solid-gas flow ratio. The purpose is to evaluate the consequences of gas and solid unbalance in the stability and collection efficiency of gas cleaning system. — Effects of deposit in different parts of the system. The geometry of observed deposits will be simulated, in order to evaluate their influence on cyclone efficiency, pressure drop, dipleg flow patterns and stability. 6.2. MODIFICATIONS TO IMPROVE GAS FILTRATION PERFORMANCE Modifications of geometry and cyclone internals, such as dipleg diameter, design and number of suction nozzles, static internals for improved efficiency, and others, will be tested with the objective of evaluating their effect on efficiency, pressure drop, flow stability and ash removal capacity. 7. Conclusion Based on dynamic similarity criteria, a pilot-scale facility simulating the hydrodynamic behavior of the ash removal system of Escatrén PEBC power plant has been designed. This model will serve to the purpose of studying the main operating parameters influencing cyclone performance, expressed as collection efficiency and pressure drop. Suction-driven flow patterns appearing in the dipleg will be also investigated, in order to ascertain the actual performance of this equipment. Conclusions derived from this project will help to plan operational and design modifications aimed at the improvement of cyclone performance, 8. Acknowledgments Original analysis of cyclone efficiency and dipleg flow are due to Mr. Roberto Guillén, presently at the Hewlett-Packard Company. Thank you very much. We would also like to thank all the staff at Escatrén power station for their help. Mr. Alfonso Ruiz Vergara, director of the plant, is specially acknowledged for their encouragement and for making possible this project. The support from Dr. Emilio Menéndez, head of the R&D department of ENDESA is also gratefully acknowledged.8. Nomenclature Co Cs Stk T « 4 uy v vin vr vr w drag coefficient particle concentration : particle concentration inside dipleg partice concetration at cyclone inlet particle mean diameter cyclone diameter dipleg diameter centrifugal force drag force gravitational force Froude number generic function of ... acceleration of gravity dipleg height gas mass flow rate solids mass flow rate pressure radial coordinate Reynolds number Stokes number temperature particle velocity particle radial velocity particle tangential velocity gas velocity gas inlet velociy gas radial velocity gas tangential velocity mechanical energy terms, as explained in the text, Greek symbols angle of sliding of solids on dipleg wall finite increment pressure drop pressure drop due to a clean gas flow separation efficiency gas friction coefficient solid friction coefficient dynamic viscosity of gas gas density particle density differential quantity9. References Ren Casal, J.M. Martinez Benet, 1989, “Célculo y Disefio de Ciclones” Ingenieria Quimica Leith, D. Mehta, 1973, “Cyclone Performance and Design’ Atmospheric Environment. Dirgo, D. Leith, Design of Cyclone Separators, in Encyclopedia of Fluid-Dynamics Vol 4 Dietz, 1981, “Collection Efficiency of Cyclone Separators” AIChE Journal, Vol 27, No. 3 API, 1975, Cyclone Separators, Manual on Disposal of Refinery Wastes, Volume On Atmospheric Emissions, Chap. 11. Crane, R.L., Barbaris, L.N., Behrouzi, P., 1992, “Particulate Behaviour in Cyclone Separators with Secondary Gas Extraction” J. Aerosol Science, Vol 23, Supp. 11, $765- $768. Beeckmans, J.M., Morin, B., 1987, “The Effect of Particulate Solids on Pressure Drop Across a Cyclone” Powder Technology, Vol. 52, pp. 227-232.SECTION Ama 1 Figure 1. Standard cyclone design | ia | | o Ss W 15 0 25 Panicle diameter, micrometers Figure 3,Fractional efficiency curve Figure 2.Escatron PFBC epclone(ashes: 1 kg/h Figure 4. Pi\SUISOET Sti HUTRADODE Gs Figure 5. Pilot plant in Escatron PFBC