Nuclear Engineering and Design 185 (1998) 153 – 172 Experimental analysis of heat transfer within the AP600 containment under postulated accident conditions Mark H. Anderson a, Luis E. Herranz b, Michael L. Corradini a,* b a Department of Engineering Physics, Uni6ersity of Wisconsin, 1500 Engineering Dri6e, Madison, WI 53706, USA Department of Nuclear Fission, Centre for Energy, En6ironment and Technology Research, CIEMAT, A6da. Complutense, 22, 28040 Madrid, Spain Received 29 July 1997; received in revised form 8 April 1998; accepted 1 May 1998 Abstract The new AP600 reactor designed by Westinghouse uses a passive safety system relying on heat removal by condensation to keep the containment within the design limits of pressure and temperature. Even though some research has been done so far in this regard, there are some uncertainties concerning the behavior of the system under postulated accident conditions. In this paper, steam condensation onto the internal surfaces of the AP600 containment walls has been investigated in two scaled vessels with similar aspect ratios to the actual AP600. The heat transfer degradation in the presence of noncondensable gas has been analyzed for different noncondensable mixtures of air and helium (hydrogen simulant). Molar fractions of noncondensables/steam ranged from (0.4– 4.0) and helium concentrations in the noncondensable mixture were 0 – 50% by volume. In addition, the effects of the bulk temperatures, the mass fraction of noncondensable/steam, the cold wall surface temperature, the pressure, noncondensable composition, and the inclination of the condensing surface were studied. It was found that the heat transfer coefficients ranged from 50 to 800 J s − 1 K − 1 m − 2 with the highest for high wall temperatures at high pressure and low noncondensable molar fractions. The effect of a light gas (helium) in the noncondensable mixture were found to be negligible for concentrations less than approximately 35 molar percent but could result in stratification at higher concentrations. The complete study gives a large and relatively complete data base on condensation within a scaled AP600 containment structure, providing an invaluable set of data against which to validate models. In addition, specific areas requiring further investigation are summarized. © 1998 Published by Elsevier Science S.A. All rights reserved. 1. Introduction Since the inception of the nuclear industry, nuclear safety has been one of the major issues for the nuclear fuel cycle. As a result, a large amount of work has been conducted and significant * Corresponding author. Tel.: + 1 608 2652001; fax: + 1 608 2626400; e-mail: corradini@engr.wisc.edu achievements have been realized, particularly in the area of understanding phenomena during accidents. This work has led to major improvements in the next generation reactor designs which incorporate new safety features. The improvement in these safety systems generally involves the use of natural forces to provide containment cooling and are referred to as passive containment cooling systems (PCCS). One of the reactor concepts that 0029-5493/98/$ - see front matter © 1998 Published by Elsevier Science S.A. All rights reserved. PII S0029-5493(98)00232-5 154 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 Fig. 1. Schematic of the AP600 PCCS configuration. The shaded section shows the portion of the containment modeled in the experiments. has been designed which implements PCCS and is under study is the Westinghouse’s AP600 pressurized water reactor. This design utilizes a unique system to maintain the containment atmosphere pressure and temperature within design limits (Spencer et al., 1993). Fig. 1 shows a schematic of the reactor containment. In the event of a postulated accident where high pressure cooling water escapes into the con- tainment, the pressure and temperature will increase as water flashes to steam. The steam will in turn start to condense on the steel containment vessel which is initially at ambient temperature. This results in an increase in the surface temperature of the steel wall. The heating of the steel containment wall causes air from outside, due to buoyancy forces, to be drawn in through an air baffle between the concrete containment and the M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 steel inner wall (not present in current reactors). This process along with the release of cooling water, by gravity from reservoirs situated above the containment, hold the wall temperature well below that of the internal bulk atmosphere. This temperature difference along with a concentration difference created by the condensation of steam in the presence of noncondensable gases, sets up a natural circulation flow pattern within containment. Steam condensation, enhanced by the turbulent natural convection, but inhibited by a noncondensable gas layer formed adjacent to the wall, should provide sufficient cooling to keep the ambient conditions within containment under safe structural limits. This cooling system performance needs to be tested by conducting experiments which measure the heat transfer rate from the containment bulk to the external atmosphere. Even though there has been extensive experimental and theoretical research in the area of condensation (Collier and Thome, 1994), much less work specifically addressed the condensation process at large scales, such as reactor containments. Uchida et al. (1965) and Tagami (1965) provided some of the pioneering work in this area, which has recently been corroborated by other investigators (Kataoka et al., 1992). This work has led to the development of correlations used in containment safety analysis (Corradini, 1984; Kataoka et al., 1992) that estimate the heat transfer coefficient (HTC) based on the ratio of Wnc/Wv. Recently, Green and Almenas (1996) carried out a peer review and analysis of both small and large scale experiments including CVTR (Carolina Virginia tube reactor) and the E-series of the HDR experiments, along with others. They concluded that the above mentioned correlations provide a too simplistic method of estimating the HTC, missing variables such as pressure, temperature, and bulk velocity which are of primary importance in condensing scenarios. With the advent of new PCCS cooling designs there has been a change in the potential condensing conditions within the containment. The results of which, significantly increases the temperature difference between the wall and the containment compared to those in the current reactors that have no external cooling. Also the absolute tem- 155 perature of the containment surfaces will be lower, affecting properties relevant in the condensation process. These issues and other differences in anticipated accident conditions, along with the increased importance of the condensation process of the PCCS, has prompted renewed interest in performing experiments. These new experiments should be aimed at confirming the PCCS capability to accomplish its goal and to fully characterize the condensation scenario outlined by the revised boundary conditions in the new systems. Westinghouse (Kennedy et al., 1994) has performed some large scale experiments designed to look at the entire cooling system to evaluate its performance and provide test data for license approval by the NRC. These large scale tests give general information on the pressure, temperature, and containment responses in the case of a postulated accident as a function of time, along with some information on the heat transfer rates, but lack some of the insight given by more closely controlled facilities on the effects of both primary and lower order variables. Some past experiments have been designed to look specifically at these effects at a smaller scale. Dehbi et al. (1991) conducted some experimental work on a 3.5 m long 0.038 m diameter tubular geometry with different pressures, mass fractions of vapor/noncondensables, and light noncondensable gases (helium as a hydrogen simulant). This led to the development of correlations that relied on these variables. Despite the valuable information provided by this data the geometry of the facility is a drawback due to the questionable non-uniform conditions in the vessel. Huhtiniemi and Corradini (1993) considered the effects of orientation and bulk velocity in a smaller rectangular facility. They improved on the cooling design of Dehbi, however, the small size required them to impose a forced velocity flow parallel to the cold wall so that one could achieve velocities similar to those anticipated in an actual containment accident. To bridge the gap between the large time varying simulation of an accident (Westinghouse) and the smaller separate effects studies (Dehbi et al., 1991; Huhtiniemi and Corradini, 1993), an experimental program which addresses conditions similar to those expected during an accident in the 156 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 Fig. 2. Key components in the test facility: test vessel, coolant supply system, and gas supply systems. AP600, at an indepth level has been conducted (Anderson, 1998). Specifically this experimental program was aimed at achieving a thorough understanding of the role of both major and minor variables on the heat transfer rate, along with providing a valuable data base to validate heat transfer models. Two experimental facilities have been constructed with similar features to the actual reactor such as a proper aspect ratio, similar surface finishes, and the ability to achieve prototypical accident conditions. Qualified experimental results have been gathered by demonstrating test reproducibility and by using redundant and accurate measurement techniques for the HTC. Particular care has been taken to follow a representative test protocol that allows an experimental test matrix of over 70 tests to span the expected range of conditions. 2. Experiment description 2.1. Test facility The test facility consists of three major components (Fig. 2): the test vessel, the coolant system, and the gas supply system. Two different test vessels have been used. One of them, called the ‘atmospheric test section’ was constructed to study the effects of temperature differences between the bulk atmosphere and the wall during steam– air and steam– air– helium condensation at M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 157 Fig. 3. Schematic of the pressure test vessel. Also shown are the probe hole locations, probes 2 – 8 were used for the mixture temperature. 158 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 atmospheric pressures and with transparent walls for visualization. The other, a ‘pressurized test section’, was designed to withstand more prototypical pressures and temperatures anticipated in the accident scenarios. The combined usage of both vessels along with a proper test matrix allowed a thorough understanding of the role of the major variables involved in the condensation process under realistic conditions and representative test protocols. The rectangular test vessels were both 1:12 scale radial slices of the AP600 containment from the operating deck to the top of the containment (Fig. 3), resulting in a test vessel with dimensions of 2.8 m tall, 1.7 m wide, and 0.32 m in depth (Anderson et al., 1998). The atmospheric test vessel had two O.91 m long condensing plates, one oriented vertically and one horizontally, in the top right hand corner of the facility. The remainder of the test section was made of transparent polycarbonate (Lexan™) sheets to allow visualization of the condensation process. The pressurized test vessel was constructed with steel side walls so that a design pressure of 4.0 bar absolute could be achieved, characteristic of actual accident conditions. The condensing plate in the pressurized test section was constructed in seven sections with various angles which resulted in a 2:1 semi-elliptical structure similar to the upper dome of the AP600 containment. In both test sections the aluminum condensing plate was thermally insulated from the rest of the test section to minimize any conduction from the side walls. There were access holes cut into the side of each test section and an injection flange placed on the access panel so that the injection system could be inserted 0.15 m above the bottom of the test section. After construction, the entire test vessel was insulated. The condensing plates consisted of 2024t351 aluminum 0.30 m wide and 0.038 m thick. The plates were prepared in a manner similar to the proposed procedure of the walls of the AP600 containment (WCAP, 1992). This consisted in sand blasting the plates and coating them with about 0.1 mm thick coating of inorganic zinc paint, which has a thermal conductivity of 0.0209 W cm − 1 K − 1. Each condensing plate was fitted with several 0.3048 × 0.1524 m coolant plates, similar to the construction used by Huhtiniemi and Corradini (1993), so that the coolant flow through each plate could be controlled and monitored separately by individual flow meters. A schematic of a typical cooling plate is shown in the blow up of Fig. 3. The aluminum condensing plates were held at a constant temperature by 12 and 14 individual cooling plates, for the atmospheric and pressurized sections, respectively. Coolant water, supplied by a Neslab HX-150 constant temperature water bath enters two coolant manifolds. The coolant flow rate is controlled by a series of independent needle valves (one for each plate) and measured with Dwyer RMC141 flow meters. After passing through the cooling plates, the coolant enters two exit manifolds and then returns to the coolant bath. The steam was supplied by a Sussman model ES-7L boiler and was injected into the test vessel through a uniform steam injection configuration. The 72 kW boiler which is able to produce 0.027 kg s − 1 of steam is equipped with a Mercoid Da 531 Bourdon tube pressure switch, which has a dead band of 3 psi so that the boiler pressure fluctuations are minimized resulting in a minimal variation of the steam temperature. The steam injection system produced a uniform distribution of steam across the bottom of the vessel at low velocities. It was constructed of 1 in. IPS manifold pipe with 19 evenly spaced nozzles (minimum opening of 4.8 mm) used to inject the steam into the test vessel. A small hole was also drilled in the bottom of the pipe to allow any condensed liquid to flow out of the tube. Steam was supplied on both ends of the pipe to create a more uniform injection from each nozzle. 2.2. Measurement techniques Two independent methods of determining the HTC’s were used. The first was a local heat flux measurement (HFM) using a linear array of thermocouples and the second was an area averaged heat transfer coefficient that was determined from a coolant energy balance (CEB). The heat transfer coefficient on the surface of the condensing plate can be determined by the heat flux through the plate and the temperature M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 159 Fig. 4. Schematic of the HFM probe design. difference between the gas bulk and the plate surface. hi = k(dTi /dx) Tb,i −Tw,i (1) The HFM’s were developed to measure both the temperature gradient and the surface temperature of the plate. A schematic of one is shown in Fig. 4. Each consists of a set of four E-type thermocouples encased in a 4.8 mm o.d. stainless steel tube. Four 1 mm precision holes were drilled in the side of the stainless steel sheath with a separation of 8 mm. Thermocouples were inserted into the hole and a silicon sealant was injected into the end of the tube to hold the thermocouples in place. The thermocouples were individually calibrated using a platinum thermometer to ensure a proper measurement of small heat fluxes. To determine the heat flux the HFM’s were placed in holes drilled through the cooling plates and into the aluminum condensing plate. A twodimensional analysis with ANSYS (Kohnke, 1995) showed that materials with higher thermal conductivities than air (such as silicon, for example) between the thermocouples and the aluminum would minimize the possible disruption of the temperature gradient due to the probes presence. Therefore, the small gap was filled with a high thermal conductive silicon grease. A linear least squares equation defining the temperature distribution of the plate was found from the temperature measurements and the positions of the thermocouples. The slope of this line is the temperature gradient in the aluminum plate and the y-intercept is the surface (wall) temperature. To find the temperature at the surface of the plate a linear temperature dependence is assumed: Ti (x) = dTi x +Tw,i dx (2) A similar equation governs the temperature distribution in the zinc coating. The existence of the coating was included in the analysis (however any contact resistance between the aluminum and the zinc coating was neglected). With knowledge of the gradient, surface temperature, and the bulk mixture temperatures the heat transfer coefficient could be calculated. The CEB method relies on the equation: hi = rlCpV: i DTcoolant,i Ai (Tb,i −Tw,i ) (3) The variables in the above equations were obtained by passing temperature controlled water through a flow meter and then through the coolant channels in the coolant plates. The temperature of the water was measured with E-type thermocouples at the inlet and subsequently at the 160 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 exit of the coolant plate. An energy balance of the liquid will yield the energy removed from the condensing plate and, thus, the heat flux associated with the area under the coolant plate could be found. The surface plate temperature was taken from the HFM measurements and the bulk mixture temperatures were measured. The test section was also equipped with several temperature probes to measure axial variations in the test section fluid temperature. Compression fittings with a minimum opening of 12.7 mm were used to allow easy installation of a variety of measurement probes. The probe holes numbered 2 – 8 next to the condensing plates (Fig. 3) were used as the mixture temperatures in the calculation of the HTC. These probes were located sufficiently outside of the thermal boundary layer but as close to the condensing plates as possible, considering the structural integrity of the side walls. This was done to ensure that the measured temperature was that of the bulk. The relative humidity (RH) could also be measured at any of the compression fitting positions using a Vaisala series HMP131Y humidity and temperature probe. It was found that the bulk was saturated everywhere to within the accuracy of the probe for all tests. This was also confirmed by gas sample measurements taken near the center of the test section approximately 0.76 m from each coolant plate and analyzed on a mass spectrometer. 2.3. Test protocol The scope of the experimental program encompassed the range of possible conditions that could occur in the AP600 containment after a hypothetTable 1 Ranges of main experimental variables Variable Range P (bar) Tb (°C) Tw (°C) XHe (%) RH (%) 1.0 – 3.0 60.0– 120.0 25.0– 100.0 0.0–30.0 100.0 Table 2 Typical experimental time sequence Time (min) Action items t=−210 t=−200 t=−190 t=−180 t= 0 t= 10 t= 30 Setting noncondensable conditions Recording initial conditions Operation of the coolant flow system Operation of the steam flow system Steady state conditions Data acquisition Completion of test ical accident. The range of the key variables are given in Table 1. The test protocol was kept as similar as possible to the sequence of events in a postulated accident. Table 2 summaries the major steps of the test procedure. Air – steam experiments required air at atmospheric conditions in the test section prior to steam injection. The steam– air– helium experiments conducted at 1 atm were carried out by mixing helium with air in a specific ratio by measuring the flow rate of both the air and helium with a Dwyer RMA122 and a Dwyer RMA8 flow meter, respectively. The mixing was done in a tee which was connected to the steam injection system. To obtain the appropriate mixture of noncondensables the test vessel was flushed with several volumes of the air– helium mixture. A gas sample was taken at the midsection of the vessel and analyzed on a mass spectrometer to ensure the correct noncondensable mixture was attained. A more prototypical introduction of helium was used in the pressurized facility. This was done by introducing steam into the facility which originally contained air. After a steady state at a given pressure was achieved, helium was slowly introduced at a known flow rate along with the steam until the proper fraction was attained and confirmed by mass spectrometer measurement. The desired temperature and pressure of the test section was set by allowing steam from the boiler to enter the test section through the steam injection system. The flow rate was adjusted manually until the appropriate conditions were reached. To perform atmospheric tests a valve to a suppression pool was opened, which allowed the mixture of steam and noncondensables to escape M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 161 Fig. 5. Local heat transfer coefficient measurement reproducibility for both test vessels (P= 1 bar, Tw = 30°C, Tb = 80°C, saturated conditions). from the test section. This valve was kept open until enough of the noncondensables had been released so that they remained saturated at a pressure of one atmosphere and the desired bulk temperature (the suppression pool was necessary so that during any slight bulk temperature fluctuations outside air would not enter the test section and change the noncondensable concentration). The coolant temperature and flow rate was also adjusted while the test section was reaching a steady state condition. This was done so that the surface temperature of the entire condensing plate was approximately constant and at the desired temperature. A quasi-steady state is achieved when the energy introduced by the steam is equal to the energy removed by the coolant water. This is determined when there are no significant changes in the temperatures of the coolant flow, condensing plate or the bulk test section. Once the steady state was achieved the thermocouple data was collected at a rate of 0.1 Hz. One hundred temperature measurements for each ther- mocouple were recorded. The RH was measured over the period of the data acquisition with the humidity probe and also a gas sample was taken and analyzed on a mass spectrometer to determine the fraction of vapor– air– helium within the test section at the time of the test. 3. Experiment results In each set of experiments there were 14 and 12 local HTC measurements conducted by both the CEB and HFM method for the pressurized and atmospheric test sections, respectively. The error in the local measurements was found to be approximately 10% for both the CEB and HFM, which was determined by the S.D. in the temperature measurements along with a propagation of errors for the different methods (Anderson, 1998). The agreement between the two methods was within the error band of the individual measurements, and the relative error between them was 162 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 approximately 15% for both the atmospheric and pressurized vessels. Several tests were conducted under the same conditions to verify the test reproducibility and this is shown in Fig. 5 for a few series of atmospheric tests. In this figure experimental values taken from both of the two independent test vessels are shown. As can be seen not only are the tests in the individual vessels reproducible but there is an outstanding similarity between the two facilities even with the differences in the geometry of the condensing plates. The slightly higher differences between the CEB and HFM measurements of the pressurized vessel as compared to the atmospheric vessel were attributed to an increase in the conduction heat transfer from the steel side walls and the connecting bolts to the aluminum condensing plates and the fact that the bulk and wall temperatures differed slightly between tests. In any case these differences are small enough to consider the heat flux measurements in both vessels to be in good agreement. In the atmospheric facility there was little variation in the local measurements, even between the horizontal and vertical plates. Such a similarity is surprising given the dramatically different behavior of the condensate depending upon the surface orientation. The vertical plate produced a uniform film of condensate, while on the horizontal plate ‘cosine shaped’ droplets of approximately 5 – 10 mm diameter were formed, similar to those described by Gerstmann and Griffith (1967) and Yanadori et al. (1985). The unexpectedly small increase of only 20% between the horizontal and the vertical surface is thought to be due to the presence of noncondensables which decrease the importance of the film structure because of the relatively high resistance of the noncondensable boundary layer. The increase that was observed is thought to be a consequence of the disruption of this gaseous boundary layer due to the departure of the droplets from the horizontal plate. Such an effect would increase the turbulence within the boundary layer and reduce the local thickness, decreasing the gas resistance to heat transfer. In the case of the pressurized vessel, with a smoother transition from horizontal to vertical no effect of orientation was seen, except in some circumstances where the plate attained a pure vertical position. Fig. 5 shows the local values as a function of the angle of the plate to the horizontal along with the plate number This observation can be explained by the formation of ‘rolling waves’. Fig. 6 shows a picture of the tracks of the liquid film. The top of the figure shows the horizontal plate and as you move down the picture the angle changes as indicated. The tracks of the droplets can be seen on the purely horizontal plate but as the angle changes to 5° from the horizontal the droplets roll across the surface and form ridges called rolling waves. With a further increase in the inclination the ridges become even more developed increasing in size until a drop departs from the ridge. This phenomena was also observed by Gerstmann and Griffith (1967) who saw the formation of similar ridges at angles greater than 2.75°. They observed that at even these small inclinations drop like disturbances are formed and they become elongated and grow in thickness as they move downstream until the drops fall from the interface. It is thought that the combination of the formation of the rolling waves, the steady droplet departure from the entire inclined surface, uniform turbulent convective velocities (no sharp discontinuities as in the atmospheric vessel) along with the high heat transfer resistance of the noncondensable layer resulted in the uniform HTC relatively independent of the film. The slight decrease in the HTC observed in some experiments at the purely vertical portion of the test vessel is due to the fact that there was enough liquid to form a complete film instead of the rolling waves and there was no droplet departure with a perpendicular component to the plate which would alter the noncondensable boundary layer. Since the individual local measurements were quite similar it was possible to average them without adding significant errors. This was done separately for the horizontal and vertical plates in the atmospheric test section due to the more extreme differences in the condensation modes and observed differences of 20%. All 14 plates for the pressurized test section were averaged together, since there did not seem to be a clear difference in the condensation mode. The S.D. in M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 163 Fig. 6. The condensate particle tracks within the pressurized vessel. The top of the figure shows the horizontal section and the formation of 5 – 10 mm droplets. As you move down the plate the angle changes causing ‘rolling waves’. M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 164 the average of the HTC measurements was slightly higher than the error in the local measurements (15– 20%), but this allowed an overall measure of the heat transfer rate in the entire test section enabling a simpler comparison with the models and a better understanding of the overall effects of the variables studied, therefore the remainder of the paper will deal primarily with the averages of the local measurements. The average value was also compared with an overall energy balance on the steam entering the test section and the condensate drained from the vessel and found to be within the error of the average HTC measurement for most tests, thus further validating the measurements and confirming that a steady state was achieved. The averaging of the HTC also allowed a comparison with some previous data. Dehbi et al. (1991) conducted some experiments of condensation in the presence of noncondensables on a 3.5 m long, 0.038 m diameter, vertical tube within a 4.5 m long, 0.45 m diameter, vessel. Although the temperature of the condensing surface changed significantly as a function of position, due to heating of the coolant as it passed through the condensing tube, it was possible to compare their average HTC measurements with the current investigation. Table 3 shows a sampling of data from Dehbi’s empirical correlation along with our data under various pressures and different noncondensable concentrations (Wnc/Wv). As can be seen the magnitude and trends are consistent between the two data bases. The highest discrepanTable 3 Comparison to Dehbi’s experimental results Experimental conditions Tb (°C) Tw (°C) P (bar) Wnc Wnc/Wv 70 30 1.0 0.78 3.52 73.7 33.7 1.5 0.83 5.02 93.5 53.5 2.0 0.71 2.42 106.2 66.2 2.5 0.61 1.59 115.7 75.7 3.0 0.54 1.19 420.3 326.4 22.3 584.0 428.1 26.7 Heat transfer coefficients (SI) hT,expt hT,Dehbi Error (%) 99.9 130.0 −30.0 131.8 115.8 12.1 275.5 221.8 19.5 cies were with low Wnc/Wv ratios (HTC’s approximately 25% lower) which may be due to the configuration of Dehbi’s apparatus. To create the steam they had a heated water pool at the bottom of the test section. The condensing tube was 0.3 m above the water level and the cold water entered at this location. This arrangement resulted in a distillation of the bulk atmosphere which caused higher noncondensable concentrations in the upper portion of the vessel. Since the degradation to heat transfer is not linear with noncondensable concentration and slight amounts in pure steam decrease the HTC significantly it is possible that at these low noncondensable concentrations the averaging of the HTC over the entire surface of the tube resulted in a low measurement. As the noncondensable concentration increased the effect on the HTC of more noncondensables was less important and thus the averaging could introduce negligible errors in their measurements. The discrepancy in the low pressure test (1 bar) is most likely a result of using this correlation outside of their actual data range. 3.1. Effects of bulk temperature 6ariations The majority of experiments designed to measure the effects of noncondensable concentration on the steam condensation were conducted under atmospheric conditions. This was done to remove the effects of pressure and allow a more thorough grasp of the inhibiting effect of the noncondensable boundary layer. Fig. 7 shows the effects of changing the bulk temperature from 60 to 90°C within the test section (at atmospheric pressure) while maintaining an approximately constant wall temperature of 30°C. As the bulk temperature rises, the heat transfer coefficient also increases. Such an increase is a direct consequence of the amount of steam in the vessel (Wair/Wv). Given the test protocol detailed in Section 2.3, a bulk temperature rise is achieved by injecting hot steam into the vessel until reaching the saturation conditions at the desired temperature. Then, as the wall temperature is kept constant, the driving force of the condensation, which is dependent upon the steam concentration difference between the gas bulk and the wall, is M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 165 Fig. 7. Evolution of the heat transfer coefficients with bulk temperature and steam mass fraction changes. As Tb increases the molar fraction of steam increases, resulting in higher HTC measurements. subsequently increased, resulting in more steam condensing onto the wall. In addition, the density difference between the bulk and the wall (rgi − rbg) becomes larger, since an increase in the amount of steam causes a smaller bulk gas density. This results in an enhancement of the natural convection motion of the gas, which in turn, and along with the increase in the concentration gradient, enhances the condensation heat transfer. 3.2. Effects of wall temperature 6ariations The wall temperature is an important variable in the AP600 PCCS because it is held at a low semi-constant temperature, as opposed to current reactor containment walls which are allowed to evolve freely with the bulk temperature (no external cooling). This new concept was not investigated in the data bases of Tagami or Uchida as a variable of the system. Thus, there is still a need to understand how this variable will effect the heat transfer rate. The driving potential is a function of both the density difference between the bulk and the interface and the concentration gradient. In this set of experiments the aim was to study how changes in the wall temperature, and thus changes in the density and steam concentration at the liquid gas interface, would alter the HTC. Fig. 8 shows the variation of the HTC with wall temperatures for both atmospheric and pressurized experiments. The atmospheric tests were conducted with a constant bulk temperature of 90°C and wall temperatures that varied from 30 to 80°C. The pressurized tests were conducted at a pressure of 3 bar with 1 atm of noncondensables (air) and a bulk temperature of 115.7°C with various wall temperatures. This figure shows that the HTC decreases with 166 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 Fig. 8. Evolution of the heat transfer coefficients with constant bulk and varying wall temperatures (Tb # 90°C for P= 1 bar, and Tb # 116°C for P= 3 bar). A decrease in the HTC is observed as the wall temperature is reduced since the increase in the heat flux is overwhelmed by the increase in DT. increasing temperature difference between the bulk and the wall (approximately 25% with a change in DT from 10 to 60°C), although the heat flux increases the increase in the subcooling, DT is larger than the increase in the heat flux, thus resulting in a lower HTC measurement. The trend is the same at both pressures although the curve for the higher pressures is elevated due to the enhancement of HTC with pressure. As the wall temperature is increased the concentration of steam at the wall will increase while remaining constant in the bulk. This will reduce the condensation driving force due to the concentration gradient (Xvb − Xvi). Along with a reduction in the concentration gradient there will be a reduction in the natural gas motion which is a function of the density difference between the bulk and the interface (rgi − rbg). This is due to a decrease in the density at the interface as a result of the increased temperature (ideal gas law). Since the density at the interface is closer to the bulk density there is a decrease in the convection driving force. Therefore, as the wall temperature increases the heat flux decrease substantially. The HTC which is directly proportional to the heat flux but inversely proportional to the temperature difference (Tb −Ti ) increases slightly with an increase in wall temperature because the reduction in the heat flux is overwhelmed by the large decrease in the DT. 3.3. Effects of increases in bulk pressure As mentioned in Section 2.3, the test protocol followed the events sequence during postulated accidents: initially there was 1 atm of air in the M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 167 Fig. 9. Evolution of the heat transfer coefficient with pressure. This increase is caused not only by the increase in pressure but also the increase in the molar fraction of the steam. vessel then steam was injected until the desired pressure was reached at a saturated state. As can be observed in Fig. 9, the HTC increases with pressure, showing a quasi-linear trend for pressure greater than 1.5 bar. This increase can be attributed to the rise in steam concentration that accompanies the increase in the system pressure. Namely, the condensation flux is enhanced by the enlargement of the gradient of steam concentration due to the fact that the steam content in the gas bulk is higher at higher pressures. This reinforcement of the steam concentration gradient is only partially compensated by the reduction that higher pressures cause on the diffusion coefficient (inversely proportional to pressure). With these effects one should add that the pressure rise indirectly involves a minimization of the noncondensable contribution to the gas molar fraction. 3.4. Effect of helium in the noncondensable mixture Experiments were conducted with helium gas mixed with air in molar concentrations of the total noncondensable ranging from 0 to 50%, which corresponds to total helium concentrations between 0 and 30%. In the atmospheric experiments the helium was added to the air prior to the injection of steam. Therefore if 30% of the noncondensable was originally helium this would remain constant for the various bulk temperatures, however the total helium molar fraction would decrease with an increase in the bulk temperature. Several experiments were done with concentrations ranging from 0 to 30% molar fraction of helium in the noncondensable and no noticeable effect was observed. Fig. 10 shows the HTC as a 168 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 Fig. 10. Heat transfer coefficients with varying helium concentrations (P= 1 bar, Tw = 30°C). function of Tb −Tw, for a constant wall temperature of 30°C and various bulk temperatures for helium concentrations in this range. As can be seen there are no significant effects on the HTC caused by the introduction of helium. These findings are consistent with those observed by Pernsteiner et al. (1992) and were thoroughly explained by Herranz et al. (1998) based on the compensating effects that helium has on the buoyancy driving force and on the diffusion coefficient of steam for helium concentrations up to about 35% or of the noncondensables. Experiments were also conducted with helium at more prototypical pressures with the same range of helium in the noncondensable mixture. These tests were performed by first attaining a steady state condition at 3 bar and a saturation temperature of 115.7°C. Helium was then slowly injected through a Dwyer RMA122 flow meter into the same line that the steam was entered through. This allowed the helium to be heated as it entered the vessel. This procedure is similar to how the hydrogen would be introduced in an accident scenario (ie. metal – water reactions in the core produce hydrogen which is carried along with steam into the containment). As the helium was injected it was necessary to reduce the steam flow rate so that the total flow rate was constant and maintained at a pressure of 3 bar. This resulted in an increase in the total noncondensable concentration and a reduction in the temperature of the vessel (saturated conditions were maintained). After a certain concentration of helium was injected a gas sample was taken to obtain a M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 169 Table 4 Helium in the noncondensable mixture at 3 bar Test Tbulk (°C) Twall (°C) PHe (bar) Pair (bar) Pv (bar) PT (bar) HFM CEB T5000 T5001 T5002 T5003 T5004 117.69 114.57 104.63 105.75 91.89 77.07 71.13 59.89 61.30 45.87 0.00 0.13 0.27 0.40 0.83 1.16 1.20 1.54 1.36 1.41 1.84 1.67 1.19 1.24 0.76 3.0 3.0 3.0 3.0 3.0 593.67 503.51 315.81 335.83 187.62 574.09 493.07 323.05 302.33 183.58 secondary measure of the helium molar fraction. Table 4 shows the results and key parameters of the experiments for pressures of 3 bar and concentrations below 40% molar fraction of the noncondensable. A few experiments were conducted with concentrations greater than 40% and it was found that as the noncondensable gas density approaches that of the steam, stratification can occur where a noncondensable gas layer is stable above a steam rich layer. Fig. 11 shows the bulk temperature profiles along with the heat transfer coefficients for the individual plates for a test with 53% molar fraction of helium and 47% molar fraction of air in the noncondensable gas mixture (Pair =0.147, PHe = 0.166, Pv =0.687) along with a test consisting of 100% air (Pair = 0.313, Pv = 0.687) as the noncondensable. Both tests were performed under the same conditions (e.g. steam flow rate and coolant flow rate) except for the replacement of the air with a mixture of air and helium. In this test the helium was injected after reaching steady state and taking data with the air– steam mixture. The helium was introduced along with the steam at a rate of 5.29e−4 m3 s − 1 until approximately 50% of the noncondensable was helium (air – steam was allowed to escape through the suppression pool maintaining a total pressure of 1 atm. The helium concentration was verified by mass spectrometer measurement at the time of the test). The temperature profile clearly shows the stratification of a noncondensable rich layer of gas above a steam rich region. This effect is also observed in the HTC measurements. Plates 1 – 8 which are in the upper portion of the test section show a dramatically lower HTC then plates 9 – 14 which are lower in elevation and are adjacent to the steam rich layer. The HTC’s for plates 9 – 14 with high helium concentrations are similar, quantitatively, to those with no helium present, however they seem to increase slightly with a decrease in the elevation of the plate. This stratification effect was seen at lower helium concentrations down to about 45% helium in the noncondensable mixture, however the pocket of the noncondensable rich layer is slightly reduced in size with the lower helium concentrations. It is also slightly less stable. This is due to the fact that as the helium concentration reaches this level it is approximately the same density as the steam but as steam is continually added the temperature of the steam rich region can increase and become less dense than the noncondensable. This along with the convective motion within the test section can result in a break down of the stratification layer allowing a more thoroughly mixed bulk atmosphere with approximately constant bulk temperatures and much more uniform HTC measurements. In summary, the amount of helium in the total mixture is not as important as the amount of helium in the noncondensables. If the fraction of helium in the noncondensable is high enough and the noncondensable mixture density becomes close to (or less than) the steam density, stratification is possible and was observed. 4. Summary and final remarks An experimental investigation of the condensation phenomena anticipated to occur in a possible accident within Westinghouse’s AP600 nuclear reactor has been conducted. The in-depth study has 170 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 Fig. 11. The existence of stratification at high helium concentrations in the bulk noncondensable mixture (P = 1 bar). The upper portion of this figure shows the temperature distribution within the test vessel. This indicates a stratification of noncondensable gas for high helium concentrations. The lower portion of this figure confirms the existence of the stratification by showing the local heat transfer coefficients. M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 generated a valuable data base to validate models and has pointed out several interesting phenomena that should be considered when modeling condensation under this range of conditions. (1) At inclinations of greater than 5° from the horizontal droplets will start to roll and coalesce forming ridges called rolling waves. This along with a noncondensable layer results in a fairly uniform heat transfer rate across the surface which is slightly higher than a pure vertical surface. At angles less than 5° droplets can form changing the mode of condensation. The enhancement to the HTC is however of the second order, because the resistance to heat transfer due to the noncondensable boundary layer dominates making variations in the film resistance a secondary effect. (2) Differences in both the wall temperature and the bulk temperature effect the HTC and may be relevant in predicting the heat transfer rate. Therefore, if, as in the AP600, the primary heat removal is from the condensation of steam in the containment it is necessary to use a more complex model than the simple correlations which depend only on the ratio Wnc/Wv. (3) The increase in pressure causes an increase in the HTC, however it is necessary to realize that as the pressure changes there is a significant affect due to concentration changes that will also increase the HTC. These changes may perhaps be even more significant than the pressure enhancement itself. (4) The addition of light noncondensable gases has little effect on the HTC up to concentrations of approximately 40% molar fraction added to the air. At this point the density of the noncondensable is approximately the same as the steam and buoyancy forces (rgi − rgb) will tend to zero. At higher helium concentrations when the noncondensable mixture is less dense than the steam there can exist a stable pocket of noncondensables above the steam rich bulk which prohibits condensation and can result in as much as a 50% reduction in the heat transfer. Further work is under way to examine the effects of high concentrations of helium under both atmospheric and pressure conditions. The concentrations needed to effect the HTC and form stratification are above that which can exist in the bulk of the containment since they will be 171 above the detonation limit, however there are sub-compartments where these concentrations might exist. If one wishes to model condensation in these areas the amount of helium and level of stratification could be important. 5. Nomenclature A Cp D h k M Mw P RB q V t T W X x area specific heat diffusion coefficient heat transfer coefficient thermal conductivity reduced mass steam molecular weight pressure relative humidity heat flux volume flow rate time temperature mass fraction molar fraction axial coordinate Greek symbols D difference between appropriate quantities r density Subscripts a,b components in binary system air air b bulk bg bulk gas coolant cooling plate water conv convection cond condensation g gas gi gas at the liquid interface He helium i index for plates l liquid phase nc noncondensable gas at system temperatures T total w wall v vapor phase 172 M.H. Anderson et al. / Nuclear Engineering and Design 185 (1998) 153–172 References Anderson, M.H., 1998. Steam Condensation on Cold Walls of Advanced PWR Containments. PhD Thesis, University of Wisconsin, Madison, 1998. Anderson, M.H., Herranz, L.E., Corradini, M.L., 1998. Evaluation of condensation modeling based on mass transfer analogy. National Heat Transfer Conference, Baltimore, MD. Collier, J.G., Thome, J.R., 1994. Convective Boiling and Condensation, 3rd ed., chap. 10. Oxford University Press, Oxford, 1994. Corradini, M.L., 1984. Turbulent condensation on a cold wall in the presence of a noncondensable gas. Nucl. Technol. 64, 186 – 195. Dehbi, A.A., Golay, M.W., Kazimi, M.S., 1991. Condensation experiments in steam – air and steam – air-helium mixtures under turbulent natural convection. National Conference of Heat Transfer, AIChE Symp. Ser., pp. 19 – 28. Green, J., Almenas, K., 1996. An overview of the primary parameters and methods for determining condensation heat transfer to containment structures. Nucl. Saf. 37 (1), 26 – 47. Gerstmann, J., Griffith, P., 1967. Laminar film condensation of the underside of horizontal and inclined surfaces. Int. J. Heat Mass Transf. 10, 567 – 680. Herranz, L.E., Anderson, M.H., Corradini, M.L., 1998. The effect of light gases in noncondensable mixtures on condensation heat transfer. Int. Heat Transfer Conf., 1998Kyongju, Korea. Huhtiniemi, I.K., Corradini, M.L., 1993. Condensation in the presence of noncondensable gases. Nucl. Eng. Des. 141 pp. 429 – 446 (1993). . Kataoka, Y., Fukni, T., Hatamiya, S., Nakao, T., Naitoh, M., Sumida, I., 1992. Experiments on convection heat transfer along a vertical flat between pools with different temperatures. Nucl. Technol. 99, 386 – 396. Kennedy, M.D., Peters, F.E., Carelli, M.D., Pieczynski, A.T., 1994. Advanced PWR passive containment cooling system testing. Proc. ARS’94 Int. Topical Meet. on Advanced Reactor Safety, April 1994, vol. 1, pp. 249 – 256. Kohnke, P.E., 1995. ANSYS User’s Manual, Revision 5.2. ANSYS Inc., 1995. Pernsteiner, A., Huhtiniemi, I.K., Corradini, M.L., 1992. Condensation in the presence of noncondensable gases: effect of helium. Proc. NURETH-5 Conf. Salt Lake City, UT, USA. Sept. 21 – 24, 1992, pp. 1071 – 1078. Spencer, D.R., Woodcock, J., Wright, R.F., Gresham, J.A., 1993. Westinghouse— gothic comparison with passive containment cooling system tests scaling evaluation and analysis. ASME/JSME Nuclear Engineering Conf., 1, ASME. Tagami, T., 1965. Interim Report on Safety Assessments and Facilities Establishment Project for June 1965, No. 1. Japanese Atomic Energy Research Agency, unpublished work. Uchida, H., Oyama, A., Togo, Y., 1965. Evaluation of postincident cooling systems of light-water power reactors. Proc. Int. Conf. on Peaceful Uses of Atomic Energy, Geneva, Switzerland, 13, pp. 93 – 102. Class 2 WCAP, 1992. AP600 1/8th large scale passive containment cooling system heat transfer test baseline data report. WCAP-13566. Yanadori, M., Hijikata, K., Mori, Y., Uchida, M., 1985. Fundamental study of laminar film condensation heat transfer on a downward horizontal surface. Int. J. Heat Mass Transf. 28 (10), 1937 – 1944.