Nuclear Engineering and Design 101 (1987) 249-258 North-Holland, Amsterdam

249

THE EVALUATION ACCOMMODATION CONDUCTIVITY

OF TEMPERATURE JUMP DISTANCES AND THERMAL COEFFICIENTS FROM MEASUREMENTS OF THE THERMAL

OF UO 2 PACKED SPHERE BEDS **

R.O.A. HALL * and D.G. MARTIN

AERE Harwell, Oxfordshire OXll OR, UK A,

The results of thermal conductivity measurements on UO 2 80/~m diameter microsphere beds in helium, argon and krypton over a range of temperatures and pressures are reported. These and previous results on UO 2 microsphere beds in helium, krypton and nitrogen obtained at Oak Ridge National Laboratory (ORNL) were analysed in terms of a previously developed model which enables thermal accommodation coefficients (TAC) and temperature jump distances to be evaluated. Values derived from ORNL results tend to be in reasonable agreement with the corresponding, rather sparse data reported previously. In particular, in the case of helium, TAC's are in reasonable agreement with the direct measurements of Ullman et al. and exhibit a similar decrease with increasing temperature. However the TAC and temperature jump distance values derived from the I-Iarwell measurements are respectively larger and smaller than the comparable ORNL figures. Also, in the case of helium, the TAC increases with pressure in contrast with values obtained from the ORNL data which are pressure independent. These differences are attributed to the differing surface finishes that are achieved during the manufacturing processes employed in the two laboratories.

1. Introduction In a previous publication [1] a simple analytical model was developed to describe the thermal conductivity of powder, or packed sphere beds. It was based on a previously developed model due to Deissler and Eian [2], which has been employed fairly successfully in the past to describe the conductivity of various oxide sphere beds. Essentially these authors first calculated the thermal conductivity of two idealised structures, assuming heat flow lines are linear; the conductivity of a powder bed with a specific packing fraction value was then obtained by interpolation, using the two conductivity values corresponding to these two structures and also those for the gaseous and solid phase (i.e. relating to zero and 100% packing respectively). This simple model was elaborated [1] to allow for the fact that much of the heat conducted in the gaseous phase occurs in the small gaps close to points of contact between spheres by including a temperature j u m p distance, in a manner analogous to the treatment of the conduction of heat between the pelletted fuel and cladding of fuel pins * Chemistry Division. ** Theoretical Physics Division (on attachment from Materials Development Division).

during the course of their irradiation. In addition, the transmission of heat between microspheres by thermal radiation was included in the model; while this factor is taken into account in all the calculations that are presented below, it was found to make such a small contribution to the overall conductivity at temperatures which are of interest here that it may be safely ignored in our subsequent discussions. The appropriate analytical expressions relating to this model are given in ref. [1] and so need not be repeated here. Essentially they may be written as

k b = k g f ( 8 , m, q),

(1)

where 8 = k g / k s and m = g / D . kb, kg and k s are the thermal conductivities of the packed sphere bed, the gaseous and the solid phases, respectively, g the temperature j u m p distance and q the volume fraction occupied by the spheres. D is a sphere diameter which corresponds to the actual value in the case of a single component bed and to an average figure when the bed contains more than one size of sphere. It is important to note that in principle the model contains no fitting parameters, i.e. bed conductivities can be evaluated directly from the physical properties of the solid and gaseous components and the specifications of the sphere

250

R.0..4. Hall, D,G. Martin / Eualuation oJ temperature jump distance~

bed. in practice, because temperature j u m p distance values are not well known, this must be regarded as an adjustable parameter; on the other hand once the model has been established it may be regarded as a means whereby temperature j u m p distances may be derived from thermal conductivity measurements of packed sphere beds. One of us (R.O.A.H.) has measured the thermal conductivity of a variety of UO2 packed sphere beds at Oak Ridge National Laboratory (ORNL). Four types of bed, containing various combinations of 35, 300 and 1200 /~m diameter spheres were measured in helium, krypton and nitrogen over the temperature range 100-700 C. While most measurements were performed at atmospheric pressure, a few results were also obtained at pressures up to 0.6 MPa. It was found that a fit to better than 10% could be obtained between these results and the predictions of the model if it was assumed that the g values of helium, krypton and nitrogen at 700 K and 0.1 MPa are 10, 1.0 and 0.7 /~m, respectively; at other temperatures and pressures it was assumed that g (x T I 2 / P . Details of this comparison are given in ref. [1]. The fact that, for each gaseous environment a fit with the model is obtained for different packed sphere beds, temperatures and gas pressures, and also that the derived g values are in keeping with corresponding figures that have been obtained from other types of measurement, lends credence to the view that our simple model can yield meaningful g values from measurements of the thermal conductivity of packed sphere beds. Recently some further measurements on single component UO2 packed beds comprising 80 /xm diameter microspheres have been performed at Harwell. The purpose of this paper is to report these results (section 2) and attempt to fit them to the model in the same way as was done in ref. [1] (section 3). Because this proved not to be so straightforward we have re-analysed the results in terms of thermal accomodation coefficients (TAC's); we have also re-analysed the previously reported O R N L data in terms of T A C ' s (section 4). Finally the results from the two laboratories are compared and discussed in section 5.

ARSCN 0 1 ~ , , :

04----

~, o2 i -

20O

z,00 Temperature

50O C

'30O

Fig. 1. The thermal conductivity of an 80 /~m diameter bO 2 packed sphere bed in argon at 0.1 MPa as a function of temperature. Data points are experimental results, lines arc theoretical values obtained assuming two different values for the temperature jump distance.

hollow cylinders in which thermocouples are embedded; a central heater enables heat to flow from the inner to the outer cylinder. One improvement in design compared with the O R N L apparatus is that the l e n g t h / o u t e r diameter ratio of the bed is 10:1 rather than 5 : 1 , thereby implying that errors due to axial heat flow should be less significant. The UO2 microspheres were manufactured in Chemical Engineering Division at Harwell; their diameter was 82.9/zm and they had a density of 0.982 times the fully densified value. The fractional volume occupied by spheres in the packed beds was 0.646. To avoid hysteresis effects that were observed to

KRYPTON 01NPg

g=l~m

2. Experimental details and results

The apparatus that was employed to measure thermal conductivity is rather similar to the one employed at O R N L and described in ref. [1], and so will not be discussed in detail here. Essentially the packed sphere bed is contained in an annulus between two concentric
200 4OO Temperature ~q 600 8OO

Fig. 2. The thermal conductivity of an 80 /xm diameter UO a packed sphere bed in krypton at 0.1 MPa. Data points are experimental results, lines are theoretical values.

R.O.A. Hall, D.G. Martin /Evaluation of temperature jump distances

251

o8i
i

HELtUM 01MPo

..~.x. -=.~..~ ,,, .................................. g6 ~

~ x ~ ~ gUtlman * 0 8 jl w~ues
9=10jam U{imGn

........................................................

*4d'~

. . . . . . .

mental thermal conductivity values as a function of temperature in argon, krypton and helium at 0.1 MPa respectively. Table 1 shows the conductivity of two different beds in lielium at 300C at a number of gas pressures.

4!
I

3. Analysis of the results in terms of temperature jump distances

Fig. 1 shows the experimental thermal conductivity of the packed sphere bed in argon at 0.1 MPa as a function of temperature (data points) together with values obtained from theory, assuming two different values for the temperature j u m p distance (lines). (Here and hereafter all quoted g values relate to a temperature of 700 K and a gas pressure of 0.1 MPa; for other temperatures and pressures these values are assumed to scale as T 12//P.) It is evident that reasonable agreement between theory and experiment is obtained if g is assumed to be 0.6/~m; the line relating to a g value of 3 ~ m was included in fig. 1 to enable the sensitivity of calculated thermal conductivities with variations in the g value to be assessed. As a result the uncertainty in the above quoted g value is about 0.1/~m. Fig. 2 shows the analogous conductivity values relating to a bed in krypton, from which it will be apparent that the best fit with theory is obtained if g = 0.45 0.05 /~m. However this is not in accord with the analysis of the corresponding O R N L measurements from which, as mentioned above, a value of 1.0/xm was derived. The data points in fig. 3 represent the corresponding experimental measurements in helium at 0.1 MPa. It is apparent that using a g value of 10 /~m, the figure derived from the O R N L results, theory significantly under-predicts the experimental values, a better agreement occurring with g = 6.5 /xm. However it is also clear from fig. 3 that the variation of conductivity with temperature has not been reproduced particularly well. A better fit would be obtained if the thermal accommodation coefficient (TAC) decreases with temperature, rather than remains constant, as is implicit when a temperature exponent of 1.2 is taken for the temperature dependence of g (seen eq. (3) below). In fact Ullman et al. [3] have shown experimentally that for the h e l i u m - U O 2 system, values of the TAC decrease approximately linearly with temperature. When the g values derived from their T A C figures are employed in the calculations, the lower of the two dashed lines that are shown in fig. 3 is obtained. This clearly possesses the correct slope but predicts conductivity values that are somewhat lower than the experimental results. However

I
c

__

I
800

__A

~oo
TemperGt ure eC

6oo

Fig. 3. The thermal conductivity of an 80 /*m diameter UO 2 packed sphere bed in helium at 0.1 MPa. Data points are experimental results, dotted lines assume that g cc T 1.2 in the model (i.e. the TAC is temperature independent) while dashed lines employ g values that are derived from Ullman's TAC measurements.

occur during thermal cycling, conductivity measurements of a specific bed were only made during its first increase in temperature. As a result measurements at a specific temperature were made consecutively in different gaseous environments, the temperature raised by an appropriate increment and the measuring procedure repeated. Measurements were performed in helium, argon and krypton at 0.1 MPa over the temperature range - 1 0 0 - 7 0 0 C . In the case of helium, measurements as a function of pressure up to 1.5 MPa were also obtained for a few limited temperatures. It is estimated that the point to point error in the results of a specific run was _+3%. The data points in figs. 1, 2 and 3 represent experi-

Table 1 The thermal conductivity of two packed sphere beds in helium at 300 C and a number of gas pressures Pressure (MPa) 0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 Thermal conductivity ( W / m / K ) Bed 1 0.82 0.99 1.1i I.I8 1.23 1.26 1.29 1.31 Bed 2 0.81 1.01 1.14 1.21 1.26 1.28 1.32 1.34

252

R.O.A. Hall, D.G. Martin / Evaluation of temperature jump distances

9RNL ~oto HE "~ 3 ~qc

it should be noted that Ullman's TAC values related only to measurements on helium atoms whose angles of incidence and emission were 45 o, whereas for the present application we are concerned with energy exchanges at all possible angles. N o w from Ullman's rather limited measurements at different angles, it appears that while values of the T A C will vary with angle, its dependence with temperature will be little affected. Accordingly it seems permissible to modify all of Ullman's T A C (and, in practice, the derived temperature j u m p distance) values by a constant factor. Fig. 3 then shows that reasonable agreement with experiment is obtained if Ullmans's g values are multiplied by a factor 0.8. Hitherto our analysis, both here and in ref. [1], has assumed that values of the TAC for a specific gas is independent of temperature and pressure. However since our discussion of the helium results casts doubt on this assumption, it seems appropriate to re-analyse both the O R N L and Harwell data in terms of T A C ' s in an attempt to obtain a more coherent overall picture of the various underlying mechanisms.

i ob~

I o~l

b............ ~ ........
"...

,,

.,
.

0 2 S E . . . . .

. . . . . . . . ............

x 3 component + 2 ~mponent 0 fine coarse Ultman

.......

U 200

, 5O0 C

i 800

400
Temperature

Fig. 4. TAC values derived from conductivity measurements at ORNL on four different UO 2 beds in helium at 0.1 MPa.

4. Analysis of present and previous results in terms of thermal accommodation coefficients

Before discussing the results obtained at O R N L which were reported in ref. [1], it is helpful if we list the various measurements that were performed in that particular series of experiments. These are given in table 2, which indicates that UO2 microspheres of three different diameters were available; combinations of these sizes produced four kinds of packed sphere bed, and measurements were made in three types of gaseous environment, namely helium, nitrogen and krypton. While most of these measurements were performed at 0.1 MPa, a few were also made at higher gas pressures; the range of temperature covered was - 1 0 0 - 7 0 0 C .

Further details of the characteristics of these microsphere beds will be found in ref. [1]. The analysis consisted first in finding the values of g in eq. (1) which gave conductivities that were in agreement with the various experimental values. These were then converted into T A C values, using the relation due to Kennard [4] 2-a a 1 ~ 1+y V
kg
i M

cvP

(2)

where j = g, a the TAC, y the ratio of specific heats at constant pressure and volume, R the gas constant, M the atomic or molecular weight, Cv the mass specific heat at constant volume and P the pressure. Note that in the case of many measurements (including the present ones) the resulting temperature j u m p distance refers to two gas-solid interfaces, whereas theoretical expressions only consider one. To emphasise this distinction we have introduced two variables, g and j , to

Table 2 Details of microsphere beds employed in measurements performed at ORNL Type of bed Composition by weight Volume fraction of bed occupied by microspheres 0.625 0.593 0.810 0.877 Gas environment employed in the measurements
He

Coarse Fine Two fraction Three fraction

100% coarse 100% fine 55% coarse + 45% fine 59% coarse + 20% medium + 21% fine

He He and N 2 He, N 2 and Kr

Coarse, medium and fine refer to UO 2 microspheres of diameter 1200, 300 and 35 tzm respectively.

R.O.A. Hall, D.G. Martin / Evaluation of temperature jump distances

0 R N L Data HELIUM 440C 0 r AERE Data HELIUM 300~C

253

x x

i o
o

X X

E <

~
x -tO 3 component 2 component : fine x

~o~

x x x

L
4 Pressure

I
8 NPa

I
12

I
4 Pressure MPo

I
8

Fig. 7. T A C values derived f r o m m e a s u r e m e n t s at Harwell in h e l i u m at 3 0 0 C .

Fig. 5. T A C values derived from m e a s u r e m e n t s at O R N L on different beds in h e l i u m at 440 C.

describe the temperature jump distance in these two cases. For a monatomic gas eq. (2) reduces to j = 0.2174 2 -a a MVr~ P-"

kg

(3)

Eq. (3) shows immediately that if the value of a is independent of temperature and pressure then j (x TI2/P, since, to a good approximation, k~o: T On. Fig. 4 shows the TAC values that were derived from the measurements at ORNL on the four different types of packed sphere beds in helium at 0.1 MPa. Specific values are represented by the data points through which, for each type of packed sphere bed, broken lines have

been drawn. The solid line indicates Ullman's [3] experimental results. It will be seen that around 400C TAC values derived from the coarse bed results are comparatively higher than those obtained from the other beds. However not too much significance should be attached to this difference since the associated uncertainties are an order of magnitude greater than those relating to the other beds. It is evident that the results of fig. 4 indicate a downward trend in TAC values with increasing temperature; also this trend and the actual TAC values are in fair agreement with UUman's [3] results. Fig. 5 shows the effect of pressure on the TAC at 440 o C, relating to beds in helium that were measured at ORNL. Results at other temperatures, which are not shown here, exhibit a similar behaviour. It is evident

AERE x

Data

HELIUM 01MPa

ORNL Data

KRYPTON 01 MPa { 3 component bed )

06

10

_~
*6

o4
.6

08

E E o <

~ o6

1
200

I
400 Temperature C.

I
600

J
800

0k

I
200

1
400 Temperature C

I
5O0

I
BO0

Fig. 6. T A C values derived from the Harwell m e a s u r e m e n t s on beds in h e l i u m at 0.1 MPa.

Fig. 8. T A C values derived f r o m the O R N L m e a s u r e m e n t s on a three c o m p o n e n t bed in krypton at 0.1 MPa.

254 F
[ AERE Cote

R.O.A. Hall, D.G. Martin / Et~aluation of temperature jump dtstam'e.s

KRYPTON , 1 MPo ,~R 3ate AR,GCN : 1 ~ 4 : ,

;J

g
"2-~x ~
x

E
X X K X

i i !
,3 8 ~- . . . . . . . . L .............. 200 L . . . . . . . . 400 Temperature o: 500

"

!
200 400 Temperature C

J_ . . . . . . . . . . . . . . .

~__ 800

Fig. 10. TAC values derived from Harwelt measurements in argon at 0.1 MPa. Figs. 8 and 9 show T A C values relating to krypton at 0.1 MPa that were obtained from respectively the O R N L and Harwell measurements. The rather sparse O R N L data does not indicate any very strong dependence of T A C values with temperature, while the Harwell results suggest a small decrease with increasing temperature. Note however that the Harwell T A C values are significantly higher than those derived from the O R N L results. Fig. 10 shows T A C results that were obtained from the O R N L measurements in nitrogen at 0.1 MPa. It is evident that T A C values differ somewhat for the two and three component beds, with a possibility that there is a decrease in this parameter with increasing temperature, Finally fig. 11 presents the results that were obtained from the Harwell measurements in argon at 0.l MPa. These are very similar to the ones obtained for a krypton environment and shown in fig. 9.

Fig. 9. TAC values derived from Harwell measurements in krypton at 0.1 MPa. that to a fair approximation T A C values are independent of the gas pressure. Fig. 6 shows the TAC values that were derived from the Harwell measurements on beds in helium at 0.1 MPa. It is evident that the temperature dependence is similar to that observed by Ullman [3], but that at a specific temperature the TAC is appreciably greater than those derived from the corresponding O R N L data or from Utlman's results. Fig. 7 shows the effect of gas pressure on TAC values in the case of conductivity measurements made in helium at Harwell. These particular results, which are given in table 1, were made at 300C; however similar trends are observed at other temperatures. It is evident that in contrast with the O R N L results of fig. 5, T A C values exhibit a definite increase with pressure.
I I
1

ORNL

DGto

NITROGEN

0 1 'APe

5. Discussion In ref. [1] and also in section 3 we have attempted to correlate experimental thermal conductivity values of packed sphere beds with a theoretical model in which the only adjustable parameter was the temperature j u m p distance, it being assumed that g ~ T I Z / P . As already noted this relation assumes that the corresponding T A C value is independent of the temperature and pressure. Table 3 lists the values of g that were obtained in this correlation and compares them with the results of other workers. It is evident that reasonable agreement exists between our results and those of other authors who measured gap eonductances between pelleted U O 2 and metallic surfaces, either in laboratory or during irradia-

2 c 10[--.2 i

x o < 8 x : 3 Component bed + 2 C o m p o n e n t bed x

I
200

I
&00 Tempernture

__
C

i
600

I___ 800

Fig. 10. TAC values derived from the ORNL measurements on two and three component beds in nitrogen at 0.1 MPa,

R.O.A. Hall, D.G. Martin / Evaluation of temperature jump distances' Table 3 Values of temperature jump distances at 700 K and 0.i MPa Gas Helium g value (p.m) 10 6.5 16 12.1 16.1 9.8 8.9 0.6 7.1
] .2

255

Notes (a) (b) (c) (d) (e) (f) (g) (b) (c)
(d)

Argon

1.8 Krypton 1.0 0.45 1.4 0.46 0.7

(e) (a) (b) (c) * (f) (a)

Nitrogen

Notes: (a) Hall and Martin's [1] ORNL results. (b) Hall and Martin's Harwell results reported here. (c) Ross and Stoute [5]. (d) Ross and Stoute's results corrected by W.R. Hobbs (AERE ttarwell, private communication). (e) Campbell et al. [6]. (f) Ainscough and Hobbs [7]. (g) Ainscough and Hobbs (unpublished results, 1982). * Hobbs's correction to the results of Ross and Stoute [5] for Kr is not available, but should decrease the figure of 1.4 by at least a factor 2. tion. g values obtained from the O R N L results tend to be in better agreement than those derived from the Harwell measurements; the reason for this will be considered below when the corresponding TAC values are discussed. Turning next to the T A C values, we note first that these are derived from eq. (2), an expression due to Kennard [4]. While almost certainly this is the most widely used equation relating temperature j u m p distance with TAC, it should be noted that slightly different expressions have been derived by other workers as a result of different treatments for the scattering processes, some of which are summarised by G a m i e r and Begej [8]. Naturally these will yield T A C values that are different from those obtained from the Kennard relationship. For example the expression due to Loyalka [9] gives T A C values which are typically 70-80% of the figures reported here. The thermal accommodation coefficient is a measure

of how, as a result of colliding with a surface, a stream of gas molecules is able to attain the temperature of that surface. There are at least two ways in which T A C values are expected to be higher than that obtained from an ideal surface of a specific solid. Firstly, if gas atoms are adsorbed on the surfaces; since such atoms normally possess atomic weights that are more comparable to that of the molecules comprising the gas compared than with the solid, more efficient transfer of energy will occur. Physically one would not expect the value of the TAC to exceed unity, in which case gas molecules that have collided with the surface will have attained the surface temperature. However, the theoretical work of Ku~der [10] has not completely ruled out this possibility. Secondly, if the surface is rough this will tend to decrease the value of the temperature j u m p distance which, if used in relations such as eq. (2), will result in an apparent increase in the corresponding T A C value. The reason why the temperature j u m p distance decreases with roughness may be appreciated by the following qualitative discussion: First we note that the existence of a temperature j u m p close to a wall when a temperature gradient occurs is a fundamental feature of the kinetic theory of gases. Qualitatively its existence may be pictured as being due to the fact that, at a specific point in the gas close to the wall, molecules arriving from the direction of the wall will, on average, have travelled a shorter distance since their last collision than those that come from the bulk of the gas. Consequently, if the temperature at this point is to remain constant with time, the temperature gradient must continually increase as the surface of the wall is approached. It is this enhanced temperature gradient which gives rise to the temperature j u m p (Kennard [4]). Note that if the T A C is less than unity this merely enhances the value of this j u m p distance. In the above qualitative treatment, molecules arriving from the wall at the point in consideration will subtend a solid angle which is dependent upon its distance from the wall. However, in the case of a rough surface this solid angle will be enhanced at sufficiently small distances, which implies that more efficient heat transfer from the wall will occur than if it were perfectly smooth. As a result the enhancement in the temperature gradient will be less and the g value correspondingly smaller. Thus, on the basis of this discussion there is no reason why, if g values obtained from rough surfaces are employed in expressions such as eq. (2), T A C values which are greater than unity cannot be derived. In conclusion it may be noted that using more theoretical arguments Ku~der [10] has also concluded that TAC values at rough surfaces may be greater than unity.

256

R.O.A. Hall, D.G. Martin / Evaluation qf temperature lump distancE:v

Table 4 Values of the thermal accommodation coefficient at room temperature Gas Helium
-

TAC value 0.37 0.62 0.2 0.35 0.38, 0.50 0.36 0.77 1.20 0.84 0.80, 0.88 0.66 - 0.82 1.25 0.71 - 1.0 0.81 0.73

Reference Present work, ORNL results Present work, Harwell results GarNer and Begej [8] Thomas and Loyalka [11] Kennard [4] Ullman et al. [3] Lassmann [12] Present work, Harwell results Thomas and Loyalka [11] Kennard [41 Lassmann [12] Present work, ORNL results Present work, Harwell results Lassmann [12] Present work, ORNL results Kennard [4] Lasmann [12]

Argon

Krypton

Nitrogen

Table 4 compares the T A C values derived here with previously published figures. Because most other workers quote values at around room temperature the data presented in figs. 4, 6 and 8-11 has been extrapolated to that temperature. However Ullman et al. [3] and Garnier and Begej [8] give T A C values for helium as a function of temperature; the comparison with Ullman's data has already been made in section 4. G a m i e r and Begej obtained T A C values which are comparable to Ullman's around 500C, but somewhat lower at room temperature due to a tendency for them to increase with temperature. Considering next the references that are quoted in table 4, it should be noted that Thomas and Loyalka's [11] measurements were performed at very low pressures ( - 0.03 to 0.003 Torr) on a highly polished or fused surface of a UO2 sphere, Kennard's [4] values are some figures taken from his book and refer to a platinum surface, while values attributed to Lassmann [12] are taken from a review. In the case of helium, it is evident that the figure derived from the O R N L results is in accord with data given by Thomas and Loyalka, Kennard and Ullman. G a m i e r and Begej's value appears to be on the low side, while Lassmann's is more in keeping with the value derived from the Harwell results. For argon, The T A C value derived from the Harwell results is significantly higher than those quoted by other workers. D a t a relating to krypton appear to be somewhat sparse; however,

the Lassmann recommendation is in fair agreement with the O R N L data. Finally the nitrogen figure obtained from the O R N L results is somewhat higher than the other values quoted in table 4. The overall conclusion is that given the variability and sparseness of data available in the literature, moderate agreement with figures derived from the O R N L results has been achieved; however in the case of the Harwell results the correlation is less satisfactory. In all cases T A C values relating to measurements at 0.1 MPa appear to decrease with increasing temperature, as was also the case with Ullman's [3] observations. One possible explanation for this trend is that as the temperature is raised, gas atoms residing on the surface of the U O 2 will be progressively desorbed. However it should be noted that for a perfectly clean surface theoretical work due to Jackson and Haworth [13] suggests that a small increase in the TAC occurs w~th increasing temperature (not more than ~ 0.03 over the range considered here), so clearly this is swamped by gas desorption or other effects that lower the value of the T A C with increasing temperature. It is suggested that the reason why consistently higher T A C values are derived from conductivity measurements at Harwell compared with those obtained from the O R N L results is because surfaces of the Ha_r-well spheres are significantly rougher. Unfortunately neither the manufacturing details relating to the O R N L spheres nor the results of their microscopic examination are available to us; however, from what is known in general terms of the respective preparative techniques, one might expect that surfaces of the spheres manufactured at O R N L would be smoother (P.F. Wace, A E R E , Harwelk private communication). As far as the Harwell spheres are concerned it is worth noting that the final annealing temperature they underwent during manufacture was 1 4 0 0 - 1 4 5 0 C which, compared with most methods for manufacturing solid UO2, is on the low side; also metallographic examination suggests that the spheres exhibit surface roughnesses of up to - 2~m. Thus there is circumstantial evidence to suggest that the surfaces of spheres manufactured at Harwell are not as smooth as those made at O R N L , and perhaps also most solid U O , that is manufactured by conventional methods for the nuclear industry. The proposed differences between the surfaces of spheres manufactured at O R N L and Harwell can also be employed to account for the effect of helium pressure on the T A C (figs. 5 and 7). The increase in T A C with pressure in the case of the Harwell spheres may be accounted for if it is supposed that the comparatively low f i n n annealing temperature during manufacture

R.O.A. Hall, D.G. Martin / Evaluation of temperaturejump distances

will enable active sites to remain on the surface, thereby enabling helium atoms to be progressively adsorbed with increasing pressure. By contrast the TAC relating to the ORNL spheres is independent of pressure because adsorption of helium on the smoother surface cannot occur significantly.

257

6. Summary and conclusions (1) In a previous paper [1] a simple model was developed which enables temperature jump distances (g values) to be derived from measurements of the thermal conductivity of packed sphere beds. It was applied to conductivity values relating to a number of types of UO 2 packed sphere beds in a variety of gaseous environments that were measured at ORNL, and found to give g values which were in accord with those obtained using other experimental techniques. The temperature and pressure dependence of the g values derived in this manner assumed implicitly that values of the associated thermal accommodation coefficients (TAC) are independent of these two particular parameters. (2) A similar analysis was performed on some recent experimental thermal conductivity values that were obtained at Harwell on packed beds comprising 80 Ixm diameter spheres of UO2. While a reasonable fit between experiment and theory was achieved when the gas phase consisted of krypton or argon, this was not possible in the case of helium if the TAC is assumed to be independent of temperature a n d / o r pressure. Accordingly these conductivity results, together with those obtained previously at ORNL, were re-analysed in terms of the TAC values at various temperatures and pressures that are derived from the model. (3) TAC values derived from the ORNL results were in reasonable agreement with the rather sparse data relating to UO2 that is reported in the literature. In particular, in the case of a helium environment a decrease in TAC with temperature was found which was in keeping with Ullman et al.'s [3] observations; in addition the TAC was found to be independent of pressure to a reasonable approximation. By contrast TAC values derived from the Harwell data were consistently higher than comparable figures obtained from the ORNL results and, in the case of helium, exhibited an increase with gas pressure. (4) Differences in the TAC values between those obtained from respectively the ORNL and Harwell measurements are attributed to differing manufacturing procedures in the two laboratories. In particular there is

evidence to suggest that the Harwell manufacturing route produced spheres with rougher surfaces, with a greater capability of gas adsorption, and this would result in comparatively higher TAC values. (5) In view of the better agreement between TAC values derived from the ORNL data and previously published figures, together with the evidence that the Harwell manufacturing route produces surfaces that are less typical of commercially manufactured UO2, the ORNL results are to be preferred wherever possible in recommending data for use in fuel pin modelling calculations. On the basis of the present work (see figs. 4 and 5) the TAC of helium is independent of pressure, its value being given by the relation =0.371-1.1410 40, (4)

where 0 is the temperature (C). Eq. (4) is valid over the range 100-800C and gives TAC values to an accuracy of _+0.04. At 700 K and 0.1 MPa eqs. (2) and (4) indicate that g = 10.6 _+ 1.6 /zm. (Note that all g values quoted below relate similarly to 700 K and 0.1 MPa and that in those cases when TAC values are independent of the temperature and pressure g scales as T12/P.) Eq. (4) may be compared with Ullman et al.'s [3] TAC values (which relate to only one specific angle of incidence and emission rather than averaged over all angles), which are given in analytical form by the relation [8] = 0.362 - 2.3 10-40. (5)

Eq. (5) gives a g value of 13.5/xm. On the basis of the rather sparse data obtained for krypton (fig. 8) it is proposed that a = 0.8 _+0.1 and is independent of temperature and pressure. This yields a g value of 0.94 _+ 0.20 /xm. Similarly from the sparse and somwhat scattered data of fig. 10, a temperature and pressure independent value of 0.85 _+ 0.15 is proposed for the TAC of nitrogen, giving a g value of 0.85 _+0.27/,m. In the case of argon only Harwell data are available, so this is expected to yield TAC and g values which are respectively on the high and low side. Our results give a TAC value of 1.14 _+ 0.05 and a g value of 0.59 _+0.06 /~m. Finally, while the following data are not recommended, it is nevertheless of interest to note the TAC and g values that are obtained from the Harwell data for those gaseous species for which data from ORNL are also available. In the case of helium the TAC is given, to an accuracy of _+0.02, by the relation c~= 0.613 - 3.33 10-40, (6)

258

R.O.A. tIall, D.G. Martin / E~aluation of temperature lump distam'c~

giving a g value of 6.6 + 0.4 #m. Note that strictly speaking eq. (6) relates only to a pressure of 0.1 MPa. Because of the variations in values of the T A C with pressure o b t a i n e d at a n u m b e r of temperatures (e.g., see fig. 7), n a m e l y a drop followed b y a steady rise, n o detailed proposal for the pressure d e p e n d e n c e is given. Instead, it is suggested that at pressures up to 1 M P a eq. (6) should be adopted, with the proviso that at pressures o t h e r t h a n close to 0.1 M P a the error in the value of c, should be increased to +0.15. In the case of k r y p t o n the Harwell d a t a yields a t e m p e r a t u r e i n d e p e n d e n t value for the T A C of 1.15 + 0.05 at 0.1 M P a pressure, yielding a g value of 0.46 _+ 0.05 p,m. Note that a l t h o u g h one might argue that in all cases T A C values at 0.1 M P a pressure a p p e a r to decrease with increasing t e m p e r a t u r e this is only manifest in the r e c o m m e n d a t i o n s relating to helium. F o r other gases, t e m p e r a t u r e i n v a r i a n l T A C values have been chosen because of the sparseness a n d scatter in the data. (6) The present work has emphasised the importance of m a k i n g m e a s u r e m e n t s of the T A C and g value o n U O 2 whose surface morphology is typical of commercially m a n u f a c t u r e d material if these are to be employed meaningfully in the modelling of reactor fuel pins. It also raises the question of h o w meaningful are T A C and g values that are o b t a i n e d on u n i r r a d i a t e d fuel d u r i n g the i r r a d i a t i o n of such pins, w h e n the surface m o r p h o l o g y at the f u e l - c l a d gap is expected to be modified.

References
[1] R.O A. Hall and I).G. Martin, Thc thermal conductivity of powder beds. A model, somc measurements on UO, vibro-compacted microspheres and their correlation, .L Nucl. Mater. 101 11981) 172-183. [2] R.G Deissler and C.S. Eiam lnvcstigation of effectixc thermal conductivhies of powders, U S. National Advisory Committee for Aeronautics Research Memorandum NACA RM E52C05 11952). [3} A. Ullman, R. Achai~,a and D.R Olander, 7hernial accomodation coefficients of inert gases on stainless steel and UO-,. J. Nucl. Mater. 51 11974) 277-279. [41 E.H. Kennard, Kinetic Theo W of (iases (McGraw-Hill. New York, 1938) pp. 311 -327. 151 A.M. Ross and R.L. Stoute, Heat transfer coefficient bet'aecn UO: and zircatoy-2, Atomic Energy of Canada Ltd. Report AECL-1552 11962). [6] F.R. Campbell, L.R. Bourque, R. Deshaies, Iq. Sills and M.J.F. Notley, In-reactor measurements of fuel-to-sheath heat transfer coefficients between UO 2 and stainles> steel, Atomic Energy of Canada Ltd. Report AECI.-5400 (1977) [7] J.B. Ainscough and W.R. Hobbs. The effects of ga~ corn position and pressure on the thermal conductancc of UO: zircaloy interfaces under irradiation, in: Fabricatior~ of Water Reactor Fuel Elements (IAEA. Vienna. 1979) pp. 23 37. [8] ].E Gamier and S. Begej, Ex-reactor determination of the thernml gap and contact conductance between uranium dioxide: zirealoy-4 interfaces. Stage 1: Low' gas pressure, Pacific Northwest Report NUREO/CR-0330 (1979). [9] SK. Loyalka, A model for gap conductance in nuclear fuel rods, Nucl. Tech. 57 (1982) 220-227. [10] 1. Ku.~der, Phenomenology of gas-surface accomodation, in: Rarefied Gas Dynamics IX, VoI. Ii, eds. M. Becker and M. Fiebig (AFVLR, (}~ttingem 19741 pp. E.I-1 E.121. [11] L.B. Thomas and S.K Loyalka, Determination of thern-mi accommodation coefficients of inert gases on a surface of vitreous UOe at - 35C, Nucl. Teeh. 59 11982) 63 69 [12] K. Lassinann, Zum W~irmedurchgang im Bereich zwischen H[dle und Brennstoff eines Brennstabes. W~.rme-und Stoff'flbertragung 11 (1979) Appendix pp. 1-18. [13] J.M. Jackson and A. Haworth, Exchange of energy' between inert gas atoms and a solid surface, Proe. Ro,;. Soc 142 (1933~ 447-456.

Acknowledgements
O n e of the authors (R.O.A.H.) wishes to acknowledge the laboratory facilities offered him at Oak Ridge N a t i o n a l L a b o r a t o r y which e n a b l e d d a t a used in the p r e s e n t analysis to b e measured. A c k n o w l e d g e m e n t is also m a d e to Mr. P.F. Wace of Chemical Engineering Division, Harwell for supplying the U O 2 spheres that were employed in the conductivity m e a s u r e m e n t s performed at Harwell.