Nuclear Engineering and Design 95 (1986) 275-283 North-Holland, Amsterdam SOME ASPECTS 275 OF TWO PHASE FLOW THROUGH POROUS MEDIA V.K. DHIR Mechanical, Aerospace and Nuclear Engineering Department, School of Engineering and Applied Science, University of California, Los Angeles. CA 90024, USA In this paper results of experiments conducted on two phase flow through porous layers formed of non heated glass particles (nominal diameter 3-19 mm) are reviewed. The porous layers were 30-70 cm deep and were formed in a 21 cm inside diameter plexiglass pipe. Experiments were conducted with water and air in both co and counter current modes. Data for the average void fraction and two phase friction pressure drop were taken. A flow regime map for two phase flow through porous media, a semi-theoretical model for void fraction, an empirical correlation for counter current flooding and empirical correlations for relative permeability multipliers under co-flow conditions are reported. The interaction between the overlying liquid layer and the porous layer under counter current flow conditions is discussed. 1. Introduction An understanding of the physical processes governing two phase flow and heat transfer characteristics of porous media is essential before models to assess the coolability of a degraded core can be implemented in system codes currently being used in light water reactor safety evaluation. For example, for a fixed pressure drop across the core, the flow rate of coolant undergoing phase change in the core will depend on the particle size and porosity characterizing the core and relative permeabilities of liquid and vapor phases. The relative permeabilities in turn will depend on the void fraction. If an impervious blockage forms in the lower portion of the core, the coolability of the core will depend on the counter current flooding limitations in the core. In the present work hydrodynamic characteristics of two phase flow through porous media are established. Specifically, data and semi-theoretical and empirical correlations for the void fraction, relative permeability multipliers and counter current flooding in porous layers composed of large diameter particles (3-19 mm) are reported. A two phase flow regime map for porous layers is also developed. • I 2. Experimental apparatus and procedure The experiments were designed to obtain pertinent data on two phase flow parameters for co and counter current flow through porous media. In all of the experi- ments, the porous layers were formed of glass particles while water and air were used as the continuous and discontinuous phases, respectively. A brief description of the experiments is given in the following text, but complete experimental details can be found in ref. [1]. 2.1. Experimental apparatus A schematic diagram of the experimental setup for co-flow experiments is shown in fig. 1. The basic components of the setup are: a plexiglass test section, a water reservoir, a main centrifugal pump, and a small centrifugal pump to fluidize the bed. The instrumentation consists of water manometers, air and water flow meters, and a pressure gage to measure the air back pressure through the flow meter. Solenoid valves are placed in the air and water inlet lines for rapid flow cutoff. The test section, a 21 cm inside diameter and 1.69 m long plexiglass tube, is placed on an inlet header with a perforated grid plate sandwiched in between. U p o n the grid plate a 3.175 mm inside diameter copper tubing wound in the form of a spiral is placed. The copper tubing is connected to an air supply line and has 0.794 mm diameter holes drilled on its top side to provide a uniform flow of air through the bed. Sixteen pressure taps are located on the test section. The pressure taps are placed 15.24 cm apart in the upper part of the test section but are located only 5.08 cm apart near the bottom. Several water outlets are placed on the test section diametrically across from the pressure taps. The water outlets are also plexiglass tubes with 22.2 mm inside diameter. 0 0 2 9 - 5 4 9 3 / 8 6 / $ 0 3 . 5 0 © E l s e v i e r S c i e n c e P u b l i s h e r s B.V. ( N o r t h - H o l l a n d Physics P u b l i s h i n g D i v i s i o n ) ILK. Dhir / Two phase flow through porous media 276 Spiral Alumi Wa )ut obtained by knowing the difference between total void fraction and the active void fraction. The total void fraction in the particulate bed was measured by cutting off the gas and liquid flow rates and fluidizing the particulate bed with flow from the seconda~' pump. The secondary pump had an intake at the overlying liquid layer and a discharge at the bottom of the particulate bed. During fluidization, any gas trapped m the interstitials was released. After fluidization, the particulate bed was allowed to settle and the height of the overlying liquid layer was noted. 2.3. Data reduction The porosity of the particulate bed was determined by using the following relationship Mp ~=l-opA-- C-" Gate Valve Gate Valve Fig. l. Schematic diagram of the experimental apparatus for co-flow studies. 2.2. Procedure where Mp is the mass of the particles forming a bed of depth L, 0p is the material density of the particles, and A is the cross-sectional area of the bed. Knowing the hydrostatic head, Ap0, and height, H 0, of the free surface of the overlying liquid layer when gas is bubbling through the layer, the average void fraction, ~0, in the overlying liquid layer is determined as Pl -- A p ' / g ( Clean, dried preweighed particles of a given size were poured into the test section to a certain height. After formation of the bed, a predetermined flow rate of water was established through the porous layer. Air was gradually added to the test section. Air flow rate was increased in steps while keeping the water flow rate constant. At each flow rate, readings of the manometers connected to pressure taps placed along the bed were noted. Generally, a liquid layer overlaid the particulate bed. The gas leaving the particular bed bubbled through this layer. This height as well as the hydrostatic head of the layer were also noted for particular gas and liquid flow rates. Thereafter, the water outlet and air inlet were closed rapidly and the height acquired by the liquid in a given time was noted. The time was chosen such that any bubbles in transit will leave the particulate bed and the overlying liquid layer. This information was used to determine the average active void fraction in the particulate bed. During the course of the experiments, it was discovered that a certain number of bubbles got trapped in the interstitials of the particles. These bubbles did not move with the flow and constituted an inactive void fraction in the particulate bed. This void fraction was (1) G=-- " -- ~ ) (o,-og) (2) With the knowledge of the height, H I , the liquid layer acquired in time t after the air flow rate was cut off and making a volume balance, the average active void fraction. ~,c, in the particulate bed is calculated as j , t - ao ( Ho - L ) - ( H, - Ho ) aac (3) ( L = The average total void fraction in the particulate bed was calculated by knowing the difference between the overlying liquid layer heights H 0 and H z when both gas and liquid flow rates were maintained and when both flows were cut off the bed was fluidized and allowed to resettle, respectively. Making a volume balance, the total average void fraction, ~, in the particulate bed is written as = (H 0 - H2) - a0(H 0 - L) (4) ~L From eqs. (3) and (4), the average inactive void fraction, ain in the particulate bed can be found as ai =~-,~ac. (5) V.K. Dhir / Two phase flow through porous media Knowing the total pressure gradient in the particular bed, the frictional pressure gradient is calculated as dPf dP t dzz ]~ ~z I - ( l - ~ a c ) p l g - ~ a c o g g " Ogj2g, if the Reynolds member, Reg based on the gas properties is much greater than 8 6 ( 1 - e), the mean particle diameter is defined as (6) Only the active void fraction is used in the differential hydrostatic head term since the frictional pressure gradient through a particulate bed containing trapped air with no gas or liquid flow must equal zero. In the counter current experiments water was added at the top while air was injected at the bottom. A predetermined liquid flow rate was established and gas flow rate was gradually increased in steps. In the experiment, two phase air-water layer 10-40 cm depth existed over the bed. Readings of the water manometers and overlying layer height were noted at each incremental air flow rate. The volume of air in the bed and overlying layer was found by simultaneously closing the water inlet and exit lines and the air inlet line. The effects of air compression on the air flow rate and non-instantaneous closure of the solenoid valves on overlying layer height drop were tested and accounted for. Experiments were terminated at air flow rates at which flooding was observed to occur. The onset of flooding in the bed was discerned by noticing a rapid rise in the height of the overlying layer and formation of large gas pocket in the lower portion of the bed. 3. Discussionof results In the following section the aspects of two phase flow through porous media considered are: (i) definition of mean particle diameter in mixture of particles, (ii) flow regime map, (iii) void fraction relationship, (iv) relative permeability multipliers, (v) fluidization of the porous layer, (vi) counter current flooding, (vii) coupling between overlying liquid layer and the porous layer. 3.1. Mean particle diameter Since in two phase flow, the gas/vapor and liquid superficial velocities can vary independently over a wide range, no generally applicable simple expression for the mean particle diameter in a mixture of various particle sizes is possible [1]. However in the limiting situations, simple expressions for the upper and lower values of the mean particle diameter can be obtained. For pljl z << 277 Xl i=1 Here jg and Jt are the superficial velocities of gas and liquid in the test section in the absence of the particles. The gas Reynolds number is defined as Reg = OgjgT)p/l~g, (8) and x i in eq. (7) denotes the volume fraction of particles of size i in the mixture. If however Reg << 86(1 c), the expression for the mean particle diameter is written as ] Op = 1/2 ~ Xi i=1 Op2t (9) Similarly for plj 2 > > pgjg2, eq. (7) is valid if Re l >> 86(1 - e) and eq. (9) if Re l << 86(1 - e), The liquid Reynolds number Re t is defined in the same way as the gas Reynolds number in eq. (8) except that now the liquid superficial velocity and liquid properties are used. In mixtures containing different fractions of particles varying in size from 1 to 19 mm and large range of superficial velocities of gas and liquid it is found that the observed friction pressure drops compare well with those predicted from the Kozeny-Carman equation when the mean particle diameter given by eq. (9) is used. 3.2. Flow regime map Visual observations of two phase flow through a bed of clear glass particles revealed a change of flow pattern from bubbly to slug to annular flow as the gas velocity was increased with a fixed liquid velocity. During bubbly flow, discrete bubbles were seen to move through the bed. These bubbles tended to be nearly spherical in shape, although they became slightly oblong at higher gas flow rates. Further increase in the gas flow rate caused the discrete bubbles to merge and form extremely oblong bubbles or slugs. These slugs often spanned the height of two to three particles. With further increase in the superficial gas velocity, the slugs merged and formed a continuous path for the gas. The liquid was seen to be pushed to the particle surface while the gas moved through the center of the pores. In beds composed of particles greater than 6 mm in diameter, the average bubble size in bubbly flow was seen to V.K. Dhir / Two phase flow through porous medm 278 remain constant at about 2.5 to 3.5 mm. For beds composed of particles in the size range 1.6 to 6 ram, the bubbles appeared to be larger than the pore size and were observed to engulf the particles as they moved upwards. Often the bubbles were seen to break up into one large and a few smaller bubbles after hitting a particle. The average bubble size was generally observed to be somewhat smaller than that observed in beds of larger particles. No evidence of channel formation in these layers was found as observed in beds of particles smaller than 1.6 mm [2]. Formation of an oscillating flow pattern prior to transition from slug to annular flow was observed in porous layers composed of 3.6 and 10 mm nominal diameter particles. In the oscillating flow the liquid was seen to travel upwards in the bed in the form of discontinuous periodic waves. In the bed region between the two liquid wave fronts, annular gas flow was observed in the pores of the particles. The period of oscillation and the spacing between two wave fronts was found to increase with jg but decrease with j¢. In porous layers formed of particles 10 mm or larger in size, transition from slug to annular flow occurred in the absence of an oscillating flow. The boundaries of flow regimes are correlated using dimensionless superficial velocities, j/* and Js* defined as ..... h* =h g(p~_ o~)~d3 jg* , - [[ g(p,6Ps(! e ~pg) )-_ ~p¢3 ],/2 _jg . (10) (11) This non-dimensionalization results from the inertial term in the Kozeny-Carman equation [3]. Thus for flows in which viscous term dominates, a physical meaning to the above scheme of non-dimensionalizing the superficial velocities can not be advanced. Fig. 2 shows the flow regime map. The figure contains both co and counter flow data. It is interesting to note that for counter-current flow, the transition to the annular flow corresponds to the flooding limit for jg* < 0.13. For counter current flow with low jg* and high Jr*, the flow may go directly from bubbly to annular regime. i - - - r i ..... ' I Counter-Current i t % Cn -Current 0.3 / O 3ram [] 6mm 10mm 15mm Q 19mm / I/ / / t Jg Annular Flow 0.2 & ~ - " ~ ~/ Flooding Limit -J-J~ Slug Flow ).I Annular F ~ 0.8 Bubbly Flow i I 0.4 0.4 It" f I 0,8 Fig. 2. Flow regime map. volume fraction occupied by the inactive voids was found to be maximum for beds composed of 3 mm diameter particles and decreased as the particle size was increased or decreased. The inactive void fraction decreased with increase in superficial velocities of gas and liquid and varied between 0.2 and 0.1 for 3 mm diameter particles. However for particles of 6 mm diameter and larger the inactive void fraction was always found to be less than 0.05. The active void fraction, aac, is associated with the bubbles in transit and it is this void fraction on which most of the effort has been devoted to. The cross sectionally averaged active void fraction, ~ac, in a fixed bed is defined as volume occupied by active gas in a distance dz pore volume in a distance dz vg _A sdz A~ dz A~ dz As. (12) A~ The average pore velocities of the two phases in terms of the superficial velocities can be written as - Q, Jl (13) Js (14) h A~CI - ~CI , Jg A~C~ Qs ,C~ ' 3.3. Void fraction Visual observations showed that certain volume of gas got trapped in the interstitials of the particles. These trapped air bubbles did not move with the flow and thus constituted inactive voids in the porous layer. The where QI and Qg are the volume fluxes of liquid and gas respectively and Ct and Cs are the constants which account for the increase in the local velocity as the liquid has to flow along the periphery of the particles. From geometrical consideration, the constants can easily V.K. Dhir / Two phase flow through porous media be found to be 2/7r for flow over spherical particles. In single phase flow, generally a value of about 1 / v ~ has been used [3]. Combining eqs. (13) and (14) and substituting for C / and Cg, average pore velocity of the mixture is obtained as ), + )~ = ~( j, + j O / 2 , . (15) The local velocity of the gas phase averaged over the cross-section of the bed can be written as fig =;g/aac • velocity of the bubble can be written as u t = 2.9 02 sin 0. (19) The time, -r, taken by the bubble to travel a vertical distance equal to diameter of the particles can be found to be "r = (16) Dp dO f~-o, JO~ 2Ut " (20) Knowing, ~', the average terminal velocity in the vertical direction is written as After writing the identity fig ~-j + fig - j , Otac 279 (17) (jl+Jg) d- (2/'/7)~(Ug __)7) (18) In eq. (18), f g - j is simply the drift velocity [4]. The drift velocity can be related to the terminal velocity, fit, of a single bubble rising in porous layer saturated with liquid. At this point a distinction needs to be made between large and small particles. As mentioned earlier, in beds composed of particles larger than 6 mm in diameter the bubble size was independent of particle size. Bubbles generally moved along the periphery of the particles. Fig. 3 shows schematically typical path of a bubble. The local terminal 2, = Dp/l". (21) Substituting for ~" in eq. (21) from eq. (20) we get [Üg(Pl--Og)]l/4 COSO1 fit = - 2 . 9 02 in(tan½01) . From fig. 3, the angle 0, can be evaluated in terms of the bubble diameter D b and the particle diameter D 0. If we do so, eq. (22) becomes = 5 8[ ° g ( P l - pg) ]1/4 t 2t ;i [0 + vu/5.) × ln(1 + 2 D p / D b)] -1. PARTICLE (22) (23) It is interesting to note that as Dp ~ 0, eq. (23) reduces to the terminal velocity of a bubble rising in a pipe. However, it should be mentioned that as the particle size becomes small, the bubble dynamics is influenced by the relative size of the pores and eq. (23) is not directly applicable. The velocity ft is related to the drift velocity, fg - j , as fig - j = 7rf t / 2 . (24) The bubble size which remained nearly invariant in bubbly flow in beds of large diameter particles was found to be approximately given by D b = 1.1 g ( o l -o- Og) Db . (25) If a dimensionless particle size is defined as LIQUID Fig. 3. bcnematic diagram of the path of a bubble around large particles. Op, _-- Op g( Pl-o Pg) (26) 280 V.K. Dhir / Two phase flow through porous media substitution of eqs. (23), (24), (25) and (26) in eq. (181 yields an expression for the active void fraction as (1 - / J ~ . ) dPt • dz _ FD/ + /"Di .... pl~(1 - - i i ~ ) et dz )i-ti-z I~9) ~ac = jg/ ( jg + j, ) + 5.8, ×[(1 + )02 1.1/5;)ln(1 + 1.82~p)] "l t (27) It should be stressed that eq. (27) is applicable only when Dp >> 1.1. In fig. 4, the active void fraction predicted from eq. (26) is compared with the data. The solid lines show the predictions. In plotting the data, non-dimensionalized superficial velocities of gas and liquid are used. The plotted data are for particles with nominal diameters of 10, 15 and 19 ram. It is seen that the comparison between the predictions and the data is quite satisfactory noticing the magnitude of uncertainty in the measurements. The two phase frictional pressure gradient in a porous layer can be considered to result from the interracial and the particle drag. If the two phases are assumed to flow separately, the momentum equations for the co-flowing gas and liquid can be written respectively as rDg Adz I rDi A dz Psg~ac' I I dPf z (28) I 0.6 dP t dz FDg P g g ~ - ozg(1 -- ~ ) FDI -- A d z + A d z " (30) Defining relative permeability multipliers to account for the fractional area occupied by each phase and to provide a correction to the particle drag experienced by each phase if present alone, eq. (30) can be re-written after replacing Fog and FDt as dPf dz 3. 4. Relative permeability multipliers dP, a~¢ dz In writing the above equations the pressure gradient in the gas and liquid is assumed to be the same. The total pressure gradient is denoted by dPt/dz; whereas the particle drag on gas, the particle drag on liquid and the interfacial drag are denoted by FDg, FDt and FDI respectively. If the two equations are added, we can write the frictional pressure gradient as 150(1 - ~)2 K/~: Dp + . 1.75(1 - ( ) ~/c Dp 150(1-t)2 1.75(1__- c) 0 .2 g ~e3~p2 /xgJs + 7/;¢3~p gJg" (31) The measured two phase frictional pressure gradient data taken with Dp >t 4.5 mm and mac up to 0.6 have been used to develop correlations for the viscous multit pliers ~/' and rg' and the inertial multipliers ~/~ and ~/s of liquid and gas phases respectively. The dependence of these parameters on ~ac is found as ~ = 0.01~a ° ' + 0.02 ~ac + 0.57~a~ + 0 . 4 0 ~ 2 , (32) )); = 0.01~,c + 0.02fi~ + 0.57~, 3 + 0.40~, s, (33) ~ = 0.05(1 - ~.o)o..~ + 0.10(1 - ~ac) + 0.50(1 -- ~.c) 3 0.4 +0.35(1 - ~ac) 5, 2: "q) = 0 . 0 5 ( 1 +0.35(1 - ~.c)'. 0.2 £.E o ~ ( P ~ - P g ) I 0 -- mac ) q- 0 . 1 0 ( 1 P~ ] ~,.4 I I I I 5 10 15 20 . - - 25 Fig. 4. Comparison of void fraction predictions with the data for D_' > 2.4. (34) -- ~ a c ) 2 + 0 . 5 0 ( 1 -- ~ a c ) 3 (35) Figs. 5 and 6 graphically show this dependence. All of the available data were correlated within +_20% with these parameters. At this point it should be emphasized that the above formulation is empirical and is not totally mechanistic. It is visually observed that the discontinuous phase V.K. Dhir / Two phase flow through porous media 1.0 I ~ I written as I / 0.8 K; 0.6 \,' K'Q, K' i -dPf/dz I = (1 - ~ac) pt + ~acPg' (36) (37) Knowing d P f / d z from eq. (31) and aac from eq. (27), the superficial velocities of gas and liquid at which a given bed will fluidize can be determined, After fluidization the pressure drop may decrease somewhat due to formation of channels as the bed reforms. / / O. 2 = g( pp - ~)(1 - , ) , where / 0.4 281 / 3.6. Counter current flooding 0 0.2 i .0 0.8 0.6 0.4 - Fig. 5. Dependence of x~ and Kg on %c. (gas) never directly interacts with the particles and as such it only experiences the interfacial drag. The relative permeability multipliers obtained from co flow data when used in a counter current flooding model give the correct trend but generally result in over-prediction of the flooding limits. To overcome this deficiency, interfacial drag needs to be modeled. 3.5. Fluidization A particulate bed will fluidize when the frictional pressure drop across the bed equals the pressure due to the effective weight of the bed. This condition can be 1,0 I I I I / III o.8 0.6 ' ~ _ 1"1'o,II'l 0.4 0.2 0 0:2 0.4 0.6 (~ac p - Fig. 6. Dependenceof ~1~and 7]g on a~. 0.8 .0 As stated earlier, the coolability of a degraded core in case an impervious blockage forms at the inlet, will depend on the rate at which water can flow into the core and steam can flow out. These flow rates in turn are determined by the countercurrent flooding limitation in the core. In the present work, the flooding limit in porous layers composed of uniform size particles varying in diameter from 3-19 mm and mixtures of these particles was determined by setting a given flow rate of water and gradually increasing the air flow rate. As the gas flow rate reached approximately 50% of the flooding limit, the overlying liquid layer height began increasing at a very slow rate, indicating liquid ejection from the bed as a result of the larger fraction of the pore volume occupied by gas. Increasing the gas velocity to near the flooding limit, the overlying layer height was seen to rise much more rapidly and the discontinuous (gas) phase tended to become continuous in small pockets within the bed. With increase in the gas flow rate slightly above the flooding point, the flow became annular and the gas formed a big pocket in the outlet header. The flooding limit, therefore, was observed to coincide with the transition to the annular flow regime in the range of liquid and gas superficial velocities studied. From the definition of the flooding limit stated above, it is known that the liquid flow through the bed is limited by the gas flow at the flooding point and cannot be further increased. Opening the water drainage valve, which usually increased the water drainage rate until a lower equilibrium height was reached, should have no effect on the water height at the flooding point. The experimental determination of the flooding limit was the gas velocity at which, upon opening the water drainage valve and thus drastically reducing the resistance to water flow downstream of the bed, the water layer height within the test section was seen to remain bqK. Dhir / Two phase flow through porous medm 282 1.0 i i ,\..~ C=0.875 0.8 "'\ ,."~ ~ 0.6 The dimensionless superficial velocities .h* and jg:" ha~c been defined in eqs. (10) and (tl). Figs. 7 and 8 sho~s comparison of the data and the correlation eq. (38). In these figures the chain line shows the correlation pr~,-~ posed by Wallis [4] based on the earlier data of several investigators. It is noted that the data for small particles with nominal diameter of 3 mm tends to be low. A plausible reason for this could be the different character of flow over smaller particles. It should also be pointed out that equation (38) may not hold in the limit of jg* -~ 0 or/;* -* O. i Dp O 3mm D 6mm O 10mm A 15mm O 19mm "~N~NN~ 0.4 0.2 C=0.775 " - - - / ' \ . \ ~ 0 , , 0.2 0.4 3.7. Coupling between the ooerlying liquid layer and the porous layer \,,\ , 0.6 0.8 1.0 .,1/2 Jg Fig. 7. Flooding data for beds composed of uniform size spherical particles. constant or increase slightly. Decrease or increase of the water height was indicated very accurately by pressure readings within the overlying layer. This method allowed determination of the flooding point within an uncertainty of + 15%. All of the counter current flooding data obtained with single size particles and mixtures of particles are correlated as ~.1/2 +j~l/2 = 0.875. 1.0 0.8 I (38) I I 1 Dp (ram) ~ o6,10 1200 i I "\'~"~ ' ~ . i i i I tx 3'6'10'15'1~ 800 LAYER " ~--P=-.~oop, dZ 0 t I --~ C=0.875 C=0.775 ~z" s II O V E R L Y I N G POROUS BED E 0.2 ,I T D 6,15 A 6,10,15,19 '\..\~ J~/20"60.4 The flooding correlation given by eq. (38) represents the limiting mean velocities in the bed in the absence of any extraneous effects. The presence of an overlying liquid layer alters the hydrodynamic conditions in the top portion of the bed adjacent to this layer. Since the pores in the bed occupy only a fraction of the total bed cross-section, the void fraction must decrease as the gas leaves the particulate bed and enters the overlying liquid layer. The reduction in void fraction will occur gradually over a portion of the bed. This reduction in the void fraction is equivalent to an acceleration of the gas and a reduction in the particle drag on the liquid. As a result of this adjustment of flow in the upper portion of the bed, the absolute value of the pressure gradient increases from that in the lower portion of the bed to that in the overlying liquid layer. The variation in pressure within the top portion of the bed at the flooding limit is given in fig. 9 for a bed composed of 10 mm nominal -.475 pig ~,l dZ \. I J I 0.2 0.4 0.6 \~. N 0.8 1.0 I 0 I 400 I I 800 I 1200 1600 .,1/2 lg Fig. 8. Flooding data for beds composed of mixtures of various size particles. Z (ram) Fig. 9. Variation of pressure near the top of particulate bed at flooding limit. V.K. Dhir / Two phase flow through porous media diameter particles. In this case, all the gas was injected at the bottom of the bed. It is noted that at about 90 mm below the top of the bed the pressure gradient starts to deviate from that in the lower portion of the bed. The region over which this deviation occurs is termed as the coupling layer and the depth of this layer measured from the top of the bed is denoted as h c. Data obtained with other liquid and gas superficial velocities showed a variation of the coupling depth between 65 and 90 mm. The coupling height was found to decrease with decrease in particle size. For beds in which all the gas is injected at the bottom and the bed depth is greater than the depth of the coupling layer, coupling between the overlying liquid layer and the bed does not affect the maximum possible liquid and gas velocities through the bed. However for shallow beds or beds in which gas or vapor is injected axially as is the case in beds of volumetrically heated particles, the presence of coupling between the overlying liquid layer and the bed can significantly increase the flooding limit. 4. Conclusions 1. For certain range of flow rates of gas and liquid, it is possible to obtain simple expressions for the mean particle diameter in porous layers containing mixtures of several size particles. 283 2. Two phase flow regimes similar to those in tubes have been observed to exist in porous layers of large diameter particles. 3. Both inactive and active voids exist in porous layers. The active void fraction depends on the particle size and porosity. Correlation based on drift flux approach have been developed for -Dp ' > 2.4. 4. For two phase friction pressure drop under co flow conditions, the relative permeability multipliers have been found to depend on the active void fraction. 5. Under counter current flow conditions the coupling between the overlying liquid layer and the bed can influence the flooding conditions especially if the beds are shallow a n d / o r gas is injected axially. References [1] J.S. Marshall and V.K. Dhir, Hydrodynamics of counter current two phase flow through porous media, NUREG/ CR-3995 (1984). [2] A.S. Naik and V.K. Dhir, Forced flow evaporative cooling of a volumetrically heated porous layer, Int. J. Heat Mass Transf. 25 (1982). [3] M. Leva, Fluidization (McGraw Hill Book Co., New York, 1959). [4] N. Zuber and J.A. Findlay, Average volumetric concentration in two phase flow systems, J. Heat Transf. (1965). [5] G.W. Wallis, One Dimensional Two Phase Flow (McGraw Hill Book Co., New York, 1969).