Você está na página 1de 11

Nuclear Engineering and Design 202 (2000) 219 229 www.elsevier.

com/locate/nucengdes

Passive containment cooling by natural air convection and thermal radiation after severe accidents
X. Cheng *, F.J. Erbacher, H.J. Neitzel
Institute of Nuclear and Energy Technology, Research Center Karlsruhe, Postfach 3640, D -76021 Karlsruhe, Germany Received 31 December 1999; received in revised form 7 July 2000; accepted 7 July 2000

Abstract One essential feature of composite containment is its potential to remove decay heat by natural air convection coupled with thermal radiation. Experimental and numerical investigations have been carried out to determine the coolability of such a passive cooling system and to study the contribution of thermal radiation to decay heat removal. A data base has been provided for validating advanced multi-dimensional computer codes and for developing physical models. It has been found that the passive containment cooling by natural air convection coupled with thermal radiation is a promising concept. Both the experimental data and the numerical results show that for intermediate and high wall emissivities, respectively, thermal radiation signicantly enhances the entire heat transfer, even at low temperatures of the containment wall. The FLUTAN code combined with the new radiation model developed has been proven to be an accurate and efcient numerical tool for investigating ow and heat transfer behaviour in the system considered. 2000 Elsevier Science B.V. All rights reserved.

1. Introduction Passive containment cooling by natural air convection and thermal radiation has been proposed in the past for several innovative reactor concepts, e.g. AP-600 (Hennies et al., 1989; Kennedy et al., 1994). A review of the experimental and analytical studies carried out shows the need for further experimental and theoretical research work concerning natural air convection and thermal radiation (Cheng and Mu ller, 1998). It became evident
* Corresponding author. Tel.: + 49-7247-824897; fax: + 497247-824837. E -mail address: xu.cheng@iket.fzk.de (X. Cheng).

that due to the complexity of the thermal-hydraulic processes involved in passive systems with their inherent small driving forces by temperature and pressure differences, experiments have to be carried out for each specic design. Validated computer codes must be developed to support the expected operational performance. The composite containment proposed by the Research Center Karlsruhe and the Technical University Karlsruhe is to cope with beyond-design basis accidents (Hennies et al., 1989; Eibl, 1994). It pursues the goal to restrict the consequences of severe core meltdown accidents to the reactor plant without any noticeable release of radioactivity impairing the public. Fig. 1 illus-

0029-5493/00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0029-5493(00)00366-6

220

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229

trates schematically the composite containment proposed. It consists of an inner steel shell of about 60 m diameter and 38 mm wall thickness and an outer reinforced concrete shell of approx. 2 m wall thickness. The annulus of approx. 80 cm radial gap width is bridged by longitudinal support ribs xed in the concrete shell. The ribs are placed on the circumference with approx. 50 cm spacing and transfer the load of the expanding and deecting steel containment to the reinforced concrete wall (composite containment) in a potential hydrogen detonation. With this concept the two individual containment shells of the present design remain essentially unchanged and the capability of withstanding higher loads is achieved by the composite structure which can cope with a maximum static pressure of 1.5 MPa. Such a containment concept has been investigated with respect to its feasibility (Eibl, 1994). One essential of this new containment concept is its potential to remove the decay heat by natural air convection coupled with thermal radiation in a passive way. In a core meltdown accident the decay heat is converted into steam by direct contact of the melt with the water. The steam produced condenses on the inner surface of the externally cooled containment shell. By reux of the condensate to the core catcher a passive selfcirculating steam/water ow is established. The increase in the temperature of the steel shell results in natural convection of air in the individual chimneys formed by the support ribs in the annular gap. Moreover, radiative heat transfer takes place between the steel shell, the support ribs and the concrete shell. Decay heat is thus removed by natural air convection coupled with thermal radiation to the ambient atmosphere in a passive way. All preliminary evaluations (Erbacher and Neitzel, 1992; Scholtyssek et al., 1992) carried out show that the composite containment concept with its passive containment cooling by natural air convection and thermal radiation is a very promising alternative for the next generation water-cooled reactors. However, additional experimental and theoretical research work needs to be done for the given conditions: large channel geometry and strong interaction between convective and radiative heat transfer. To investigate the

basic phenomena involved in natural convection coupled with thermal radiation, the separate-effects test program PASCO (acronym for PASsive COntainment cooling) has been carried out. In addition numerical calculations were made using the 3-D code FLUTAN (Shah et al., 1985; Gro tzbach and Cheng, 1996). The PASCO project was terminated by the end of 1999. The main results achieved are summarized below.

2. Experimental apparatus The test program PASCO has been carried out at the Institute of Nuclear and Energy Technology (IKET) of the Research Center Karlsruhe (FZK). The general aim of this separate-effects test program is to investigate passive containment cooling by natural air convection coupled with thermal radiation. The PASCO test facility shown in Fig. 2 simulates one cooling channel in the annular gap of the composite containment proposed. The test section consists of a vertical rectangular channel of which one wall is electrically heated. The other three walls are thermally insulated from the ambient surrounding. The maximum channel cross-section is 0.5 1.0 m, the heated height is 8.0 m with four individually heatable zones. By changing the channel depth and the heated height, the effect of channel geometry on heat transfer will be studied. Two

Fig. 1. New containment concept Composite containment.

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229 Table 2 Measurement accuracy Temperature (C) Air velocity (%) Pressure (%) Heating power (%)

221

90.75 92.0 90.5 91.0

ent values of the heated wall temperature. During the experiment the heating power transferred from the heated wall is then obtained by subtracting the heat loss from the total heating power introduced to the heated wall. The temperature of the ambient air is measured at four different elevations. Moreover, the pressure at the channel inlet, the air humidity and the total heating power are recorded. The measurement accuracy of different parameters are summarized in Table 2. With this comprehensive instrumentation, experimental data are generated for the development of physical models and for the validation of multi-dimensional computer codes.
Fig. 2. PASCO test facility.

different values (0.4, 0.9) for wall emissivity are used, so that the inuence of thermal radiation on total heat transfer can be investigated. Table 1 summarizes the test parameters used. The test facility is equipped among others with approx. 170 thermocouples to measure the distribution of wall temperatures. Traversing probes for recording the air temperature and the air velocity are installed at ve different elevations. Cross-wise traversing probes at the inlet, in the mid-plane and at the outlet measure the temperature- and velocity distributions over each individual channel cross-section. Calibrations were carried out to determine the heat loss from the heated wall to the ambient surroundings at differTable 1 Test matrix Heated wall temperature, Th (C) Wall emissivity (m ) Heated height, H (m) Channel depth, L (m) 100175 0.4, 0.9 4.0, 8.0 0.251.0

Fig. 3. Air temperature (a) and air velocity (b) along the middle line (X = 0.25 m); Th = 150C, m = 0.9, L = 0.5 m.

222

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229

Fig. 4. Air velocity at the outlet cross-section and X = 0.25 m; : Th = 150C, m = 0.9, L = 1.0 m; : Th = 150C, m = 0.4, L = 1.0 m; : Th = 150C, m = 0.9, L = 0.5 m.

3. Experimental results Fig. 3 shows the air temperature and the air velocity along the mid-line (X = 0.25 m) versus the distance from the back wall (Y ) at different axial levels. Near the inlet cross-section (Z = 0 m), the air temperature distribution is uniform (about 20C) over the entire channel depth. At higher axial levels the temperature near the walls increases and the region affected by the higher wall temperature increases. Near the outlet cross-section, this affected region spreads over the entire cross-section, so that the air temperature in the central region is clearly higher than the inlet air temperature. The air velocity has a uniform distribution in the central region in the lower part of the channel. At higher axial levels, the air velocity near the walls increases, whereas it decreases in the central region. Fig. 4 shows the air velocity at the outlet crosssection along the mid-line (X = 0.25 m) versus the distance from the back wall (Y ) at different values of the channel depth L and the wall emissivity. By decreasing the wall emissivity from 0.9 down to 0.4, the effect of thermal radiation and consequently, the temperature on the back wall reduces. This leads to a reduction in the air velocity near the back wall. In the region near the heated wall, the ow condition remains nearly unchanged. Reducing the channel depth enhances the radiative heat exchange between the heated wall and the back wall, so that the air velocity increases.

Fig. 5 illustrates the measured temperatures on the side wall and on the back wall at the middle elevation (Z = 3.8 m) for three tests with different values for the channel depth and wall emissivity. The temperature on the side wall decreases with increasing distance from the heated wall. Due to thermal radiation the temperature on all the unheated walls is much higher than the average air temperature at the same elevation (about 25C). These results emphasize the strong inuence of thermal radiation on total heat transfer. By reducing the channel depth the temperature on the back wall increases due to a larger view factor between the heated wall and the back wall. Decreasing the wall emissivity from 0.9 down to 0.4 leads to a reduction in the temperature of the unheated walls by more than 10C. Fig. 6 shows the measured heating power versus the heated wall temperature at different values of the channel depth and wall emissivity. It is found that the channel depth affects only slightly the heating power, whereas the heat power depends strongly on the temperature of the heated wall and on the wall emissivity. At high wall emissivity of 0.9, a total heat power of about 6.5 kW is removed at a heated wall temperature of 150C. This indicates that the passive containment cooling system with air natural convection and thermal radiation is a very promising concept. By reducing the wall emissivity from 0.9 down to 0.4, the total heating power decreases by more than 30%. Increasing the heated wall tem-

Fig. 5. Temperature prole on the side wall and the back wall at Z = 3.8 m; M019: L = 1.0 m, m = 0.9, Th = 150C; M040: L = 0.5 m, m = 0.9, Th = 150C; M068: L = 1.0 m, m = 0.4, Th = 150C.

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229

223

Fig. 6. Measured heating power versus the heated wall temperature at different conditions; : m = 0.9, L = 1.0 m; : m = 0.9, L = 0.5 m; : m = 0.4, L = 1.0 m.

perature from 100 to 175C leads to an increase in the heating power by about 150%.

the net radiative heat ux is equal to the convective heat ux. To calculate the radiative heat ux, a thermal radiation model has been developed with the following main features (Cheng et al., 1995): Net-radiation method for enclosures with grey and diffuse surfaces. Analytical methods for view factor computation. Macro-element methods for reducing numerical expenditure. The radiative heat transfer between diffuse and grey wall surfaces without participating uids can usually be computed by the net-radiation method for enclosures (Cheng and Mu ller, 1998). The net radiative heat power of the surface element Qi is computed by the following equation:

  
N

N QR,i E Ej = i % j,i mi mi mj J=1

4. Numerical simulations In addition to the experimental work, numerical simulation has been carried out to investigate the ow and heat transfer behaviour. The ow in the PASCO test channel is 3-D turbulent natural convection with a strong interaction between convective and radiative heat transfer. Numerical simulation of such a system requires a computer code of high capability. In the present study the FLUTAN code is used (Shah et al., 1985; Gro tzbach and Cheng, 1996) which is a 3-D nite-difference code for multi-component systems in Cartesian or in cylindrical coordinates. For the turbulence modeling the standard k m model is used with logarithmic wall functions for velocity- and temperature distribution near the wall. More details on the code and the physical models used can be found in Shah et al. (1985), Cheng et al. (1995) and Gro tzbach and Cheng (1996).

+ % (1 mj )j,i
J=1

  
QR, j mj

(1)

N is the total number of surface elements. Eq. (1) can be solved either directly or iteratively. The direct solution is exact and usually needs larger computing expenditure. An iterative solution, e.g. the GauSeidel iteration, guarantees the convergence. For intermediate or high emissivity only a few iterations are needed and a high accuracy is achieved. Nevertheless, at low emissivity the direct solution method is more efcient than the iterative method.

4.2. View factor


Generally, view factor can only be solved numerically. For ow channels in a Cartesian coordinate system where boundary walls are either parallel or perpendicular to each other, e.g. the PASCO ow channel, a analytical solution has been derived in the present work for computing the view factor between any two surface elements. Fig. 7 shows two different cases. To specify the dimensions of any two surface elements and their relative positions, seven geometric parameters are needed, indicated as a, b, c, d, e, f, and g in Fig. 7.

4.1. Radiation model


The interaction between the convective and the radiative heat transfer is expressed by the thermal boundary condition at the unheated wall, where

224

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229 Table 3 Parameters in Eq. (2) Xi Yi (g+e )/a g /a (cge )/a (cg )/a Zi +1 +1 i=2 i=3 i=4 (bf )/a (bfd )/a f /a 1 +1 1 1 +1 1 Si

The view factor for two parallel surface elements (Fig. 7a) is derived as follows:
12y A1 a2
4 4

1 Xj = % Zi % Sj Xj
1 + Y 2 i arctan 2 i=1 j=1
1 + Y 2 i + % Zi % Sj
i=1 4 j=1 4 4 4

1 2 % Zi % Sj ln(1 + X 2 j +Yi ) 4 i=1 j=1

! !

1 Yi Y
1 + X 2 j arctan 2 i
1 + X 2 j

"

" "
(2)

i=1

( f+d )/a

For two perpendicular surface elements (Fig. 7b), the view factor can be calculated by: 12y Ai a2
4 4

4.3. Macro -element method

= % Zi % Sj
i=1 4 j=1 4

1 2 (X Yj )ln(X 2 i + Yj ) 8 i

1 Xi
Yj Xi arctan (3) 2 i=1 j=1
Yj The parameters Xi, Yi, Zi and Si in Eqs. (2) and (3) are summarized in Tables 3 and 4. + % Zi % Sj

"  "

For computing ow conditions accurately, the three-dimensional computational domain has to be divided into sufciently small sub-zones. The ne discretization results in a large number of surface elements. If the radiation Eq. (1) has to be solved for all the surface elements, a large storage capacity and a huge computing time are needed. Therefore, the so-called macro-element method is introduced, to reduce storage need and computing time and to improve numerical efciency. The basic idea of the macro-element method is to combine a few surface elements (micro-elements) with identical wall emissivity to form one macro -element. The effective temperature of a macro-element is determined by the following equation:
Table 4 Parameters in Eq. (3) Xi i=1 i=2 i=3 i=4 ( f+d )/a (af )/a (afd )/a f /a Yi [g 2+(e+b )2]/a 2 [e 2+(g+c )2]/a 2 [g 2+e 2]/a 2 [(g+c )2+(e+b )2]/a 2 Zi +1 +1 +1 +1 1 1 1 1 Si

Fig. 7. Geometrical parameters of two surface elements in a Cartesian coordinate system.

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229

225

Fig. 9. Measured and calculated distribution of the air temperature and the air velocity at the channel outlet cross-section along the mid-line (X = 0.25 m): L = 1.0 m, m = 0.9, Th = 150C. Fig. 8. Measured and calculated prole of the air velocity at the channel outlet cross-section; Th = 150C, L = 1.0 m, H = 8.0 m, m = 0.9.

T4 I=

1 % A T 4 AI i  I i i

(4)

Here the subscripts I and i stand for macroand micro-elements, respectively. The net radiative heat ux of a micro-element is computed by qR,i = qR,I + ei eI (5)

where qR,I is the net radiative heat ux of the macro-element, and ei and eI are the emitting heat ux of the micro-element and of the macro-element, respectively. With the macro-element method developed, the computing time needed for solving the radiation Eq. (1) can be signicantly reduced, e.g. by a factor of 100, whereas its effect on numerical accuracy is negligibly small.

Fig. 9 compares the calculated with the measured distribution of the air temperature and the air velocity along the mid-line (X = 0.25 m) at the channel outlet as a function of the distance from the heated wall Y. The curves are the calculated results and the symbols are the experimental data. The temperature distribution is well reproduced by the FLUTAN code, whereas the FLUTAN code overpredicts the air velocity in the near wall region and underpredicts it in the central region. This discrepancy emphasizes the need for improving turbulence modeling in the FLUTAN code for the given conditions. Fig. 10 illustrates the measured and the calculated temperature on the side wall and on the back wall at the middle elevation. An excellent

4.4. Results and discussion


Fig. 8 shows an example of the calculated prole of the air velocity at the test channel outlet. In the region near the heated wall, the air velocity reaches its maximum value. It decreases rapidly by increasing the distance from the heated wall. Due to thermal radiation, the air velocity increases again by approaching the unheated walls. The minimum value of the air velocity appears in the central region.

Fig. 10. Temperature distribution on the back wall and on the side wall: Th = 150C; L = 1.0 m, H = 8.0 m, m = 0.9.

226

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229

heated wall by radiation than that transferred directly by convection, even at low values of the heated wall temperature.

Fig. 11. Comparison of the measured with the calculated heating power.

agreement between the experimental data and the numerical results has been found. The strong effect of thermal radiation on the total heat transfer is also demonstrated by the FLUTAN code. Fig. 11 compares the measured heat power with the calculated results for different test conditions. It is seen that for all test conditions the deviation between the measured and the calculated heat power is less than 10%. Further analysis shows that the average heat ux at each individual heated plate can also be well reproduced by the FLUTAN code. Fig. 12 presents the calculated convective and the radiative heat ux at the heated wall. It is clearly seen that in the entrance region the convective heat ux is high. Nevertheless, this entrance effect is restricted to the rst 2 m height. After this region the convective heat ux remains nearly constant. The radiative heat ux is maximum at the inlet and the outlet regions. This is due to the radiation heat transfer to the surroundings. On average, the radiative heat ux is still higher than the convective heat ux. A systematic analysis of the effect of thermal radiation on the total heat transfer can be found in Cheng and Mu ller (1998). Fig. 13 shows the radiative heat power and the convective heat power at different values of the heated wall temperature. The calculation is made for the conditions: channel depth 1.0 m, wall emissivity 0.9 and air inlet temperature 20C. It can be seen that more heat is transferred from the

Fig. 12. Calculated convective and radiative heat ux at the heated wall; Th = 150C, L = 1.0 m, H = 8.0 m, m = 0.9.

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229

227

Fig. 14. Calculated air velocity in the containment cooling channel; Th = 150C, H = 50 m, m = 0.9. Fig. 13. Heat power transferred from the heated wall by radiation and by convection; L = 1.0 m, m = 0.9, Tin = 20C.

5. Estimation for reactor containment The results presented above show that the FLUTAN code has been well validated for the conditions of the passive containment cooling system considered. For the containment calculation made in the present study it is assumed that the ow channels in the containment gap are vertical, rectangular channels which are thermal-hydraulically disconnected from each other. The cross-section and the height of each channel are 0.5 1.0 m and 50 m, respectively. An air lter is located at the outlet of each cooling channel. The pressure drop characteristics of the lter is expressed by (Ruedinger and Ensinger, 1983): DP = 370u + 21u 2 (6)

and the air velocity is about 0.5 m s 1. A decay heat of about 23 kW (Fig. 15) can be removed in each individual cooling channel. The average heat ux at the steel containment shell is about 0.9 kW m 2. A total decay heat of about 7.2 MW can, therefore, be removed by this passive cooling system. Furthermore, it is found that for the conditions considered, the heat power transferred by thermal radiation is about 50% higher than that transferred by convection (Fig. 15).

6. Conclusions For future nuclear power plants, the consequences of severe accidents must be limited to the near vicinity of the plant. To meet this new safety goal, the use of new innovative features, e.g. passive decay heat removal, is required in advanced water-cooled reactors to ensure containment integrity. One essential of the composite

The pressure drop coefcient of the lter is dened as: p= 2DP 740 42 = + zu 2 zu z (7)

Figs. 14 and 15 show the calculated air velocity and the transferred heat power versus the pressure drop coefcient of the air lter for the following conditions: temperature of the steel shell containment is 150C and the wall emissivity is 0.9. It can be seen that without the air lter, an air velocity of about 4.0 m s 1 is obtained. It decreases rapidly with increasing the pressure drop coefcient of the lter. At the operating conditions, i.e. the intersecting point of both lines in Fig. 14, the pressure drop coefcient of the lter is about 1300

Fig. 15. Removed decay heat in one cooling channel; Th = 150C, H = 50 m, m = 0.9.

228

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229

containment is its potential to remove the decay heat by turbulent natural convection of air coupled with thermal radiation. A thorough bibliographic survey underlines the deciency in experimental as well as in theoretical studies on turbulent natural convection heat transfer coupled with thermal radiation in a 3-d vertical channel with asymmetrical heating. In the framework of the PASCO programme, experimental and numerical investigations have been carried out to determine the coolability limit and to study the ow and heat transfer behaviour in such a passive cooling system. The following specic main conclusions can be drawn: Passive containment cooling by natural air convection and thermal radiation offers a promising safety feature for future pressurized water reactors. A broad and detailed data base has been generated for the validation of computer codes and for the development of physical models. At intermediate and high wall emissivities thermal radiation contributes signicantly to the total heat transfer by natural air convection, even at low temperatures of the containment wall. A radiation model with high numerical efciency has been developed. Analytical equations have been derived for calculating the view factor between any two surface elements. The so-called macro-elements method has been developed to improve numerical efciency. The FLUTAN code combined with the radiation model developed has been proven to be an accurate and efcient numerical tool to investigate the ow and heat transfer behaviour in a passive containment cooling system.

Appendix A. Nomenclature

A E e H L DP Qr qr T Th u m p

surface area, m2 emitting heat power, W emitting heat ux, W m2 channel height, m channel depth, m pressure drop, Pa net radiative heat power, W net radiative heat ux, W m2 temperature, C heated wall temperature, C velocity, m s1 wall emissivity, pressure drop coefcient,

References
Cheng, X., Mu ller, U., 1998. Turbulent natural convection coupled with thermal radiation in large vertical channels with asymmetric heating. Int. J. Heat Mass Transf. 41 (12), 1681 1692. Cheng, X., Erbacher, F.J., Neitzel, H.J., 1995. Thermal radiation in a passive containment cooling system by natural air convection, in: International Symposium on Radiative Heat Transfer, Kusadasi, Turkey, August 14 18. Eibl, J., 1994. Zur bautechnischen Machbarkeit eines alternativen Containments fu r Druck-wasserreaktoren Stufe 3, KfK 5366, August. Erbacher, F.J., Neitzel, H.J., 1992. Passive containment cooling by natural air convection for next generation light water reactors, in: NURETH-5, Salt Lake City, UT, September 21 24. Gro tzbach, G., Cheng, X., 1996. Development progress of the FLUTAN code for modelling heat transfer in LWR-systems (in German), in: Proceedings of the Annual Meeting on Nuclear Technology 96, Mannheim, pp. 168 171. Hennies, H.H., Kessler, G., Eibl, J., 1989. Improved containment concept for future pressurized water reactors, in: International Workshop on Safety of Nuclear Installations of the Next Generation and Beyond, Chicago, IL, August 28 31. Kennedy, M.D., et al., 1994. Advanced PWR passive containment cooling system testing, in: International Topical Meeting on Advanced Reactors Safety, Pittsburgh, PA, April 17 21. Ruedinger, V., Ensinger, U., 1983. Investigation into the Behaviour of HEPA-Filters at High Temperature, Air Humidity and Elevated Differential Pressures, KfK 3350, Research Center Karlsruhe.

Acknowledgements This work was executed under the multi-partners research contract DABASCO (contract No. FI4S-CT96-0042) co-nanced by the European Commission under the Euratom specic Nuclear Fission Safety programme 1994 1998.

X. Cheng et al. / Nuclear Engineering and Design 202 (2000) 219229 Scholtyssek, W., Alsmeyer, H., Erbacher, F.J., 1992. Decay heat removal after a PWR core meltdown accident, in: International Conference on Design and Safety of Advanced Nuclear Power Plants (ANP 92), Tokyo, Japan, October 25 29.

229

Shah, V.L., et al., 1985. COMMIX-1B: A three-dimensional transient single-phase computer program for thermal hydraulic analysis of single and multicomponent systems, NUREG/CR-4348, Vol. 1 and Vol. 2.

Você também pode gostar