1. Introduction
Steam condensation heat transfer is a common phenomenon in daily life and is widely used in the chemical industry, refrigeration, energy and other engineering processes. However, steam is always mixed with some gases that may not condense under different working conditions, while noncondensable gas can have a negative effect on condensation heat transfer [
1,
2,
3,
4].
The presence of noncondensable gases (such as air) in steam, even at very low levels, can significantly worsen heat transfer in applications [
4]. Taking nuclear power plant accidents as an example, the release of high-temperature and high-pressure gas from the primary system will result in a sudden increase in temperature and pressure in containment, thereby endangering the structural integrity of the containment. The condensation heat transfer of high-temperature and high-pressure steam on the containment wall is an important method to remove internal heat; however, the presence of any noncondensable gas, such as air, can greatly affect the steam condensation efficiency, lead to a deterioration of heat transfer performance and complicate the overall process [
2,
3]. Consequently, it is of great significance to analyze the factors affecting condensation so as to improve the calculation accuracy of condensation heat transfer.
The condensation process of steam mixed with noncondensable gas is very complicated, as it involves not only the exchange and transfer of mass, energy and momentum between steam and condensed liquid, but also the exchange and diffusion of energy and momentum between steam and noncondensable gas [
1]. So far, the research on this process mainly consists of experimental studies [
5,
6,
7,
8,
9,
10,
11,
12,
13,
14,
15] and theoretical analyses [
16,
17,
18,
19,
20,
21,
22,
23,
24,
25,
26,
27,
28].
In the experimental aspect, researchers have paid much attention to the heat transfer characteristics of condensation under various working conditions [
6,
7,
10,
14,
15], as well as heat transfer enhancement using variant means. They conducted many experiments and obtained different heat transfer correlations, like the commonly used Uchida [
7], Tagami [
8] and Dehbi [
9] models and so on, and correlations were made with the heat transfer rate via gas concentration, pressure, temperature, surface subcooling, and some other parameters. However, the main obstacle is that most correlations are not universally valid, and each empirical relation has a very strict application condition.
At the same time, a numerical calculation has become an important means to explain condensation phenomena by means of both lumped parameter codes and 3D CFD codes [
20,
21]; in contrast to the experiments, the numerical method can obtain more detailed information about the flow characteristics and gas concentration distribution. And many factors have also been considered to affect the condensation efficiency [
25,
26,
27,
28] like the geometrical parameter, thickness of film, suction factor and so on, but for pure theoretical numerical calculations, the calculation process is complicated and difficult to understand, and at the same time, too many iterations lead to a divergence of the calculation results.
In a nuclear power plant, the condensation process of steam with noncondensable gas on the surface of a heat structure is accompanied by an intense heat and mass transfer, which directly affects the spatial distribution of the temperature and pressure in containment, and then affects the safe operation of the integrated reactor [
4]. This paper analyzes and discusses this phenomenon, proposes a condensation calculation model, and applies this calculation model to the integral analysis code of the PISAA (Program Integrated for Severe Accident Analysis) for severe accidents. By conducting specific working conditions, the wall condensation model in the PISAA is compared and verified with the mainstream containment thermal hydraulic codes. Additionally, the calculation results of the model are validated by comparing the data with those of the Wisconsin condensation experiment [
15]. Finally, a brief sensitivity analysis is performed on the condensation heat transfer model, which further improves the accuracy of the condensation heat transfer calculation.
2. Condensation Heat Transfer Model
Figure 1 shows the condensation model of steam with noncondensable gas on the vertical wall. For the condensation process containing noncondensable gas, when the wall temperature is lower than the saturation temperature corresponding to the partial pressure of steam in the mixture, the saturated steam will condensate on the wall and form a liquid film. Meanwhile, the liquid film flows along the direction of gravity, and its flow state changes from laminar to turbulent as the liquid film thickness increases [
1,
14]. Due to the accumulation of noncondensable gas on the surface of the liquid film, when the mixture moves towards the interface, the aggregation of noncondensable gas molecules causes an increase in the gas’s partial pressure and forms the driving force of the reverse diffusion of noncondensable gas to the mainstream gas. While the steam condenses and its partial pressure decreases below that of the steam in the mainstream gas at the interface between the phases, this pressure differential promotes the diffusion of steam towards the interface, which is facilitated by the pressure gradient. The diffusion of steam towards the condensing surface and the noncondensable gases towards the mainstream gas maintain a dynamic equilibrium with a constant total pressure [
12]. Under the influence of two conditions, the state parameters at the phase interface cannot be determined.
The steam in the gas mixture is transferred to the wall by means of mass diffusion, which requires it to pass through the highly concentrated noncondensable gas layer that gathers on the surface of the condensate film, and then condenses on the surface and releases latent heat. The heat transfer resistance of the whole steam condensation is mainly divided into three parts, and the steam transfers the heat to the surface of the liquid film by means of condensation and convection, respectively, that is, there is condensation thermal resistance and convection thermal resistance, and further heat passes through the liquid film thermal resistance and finally reaches the condensing wall.
Correspondingly, the total heat transfer coefficient is determined by the thermal resistance of the three aspects above: the liquid film heat transfer coefficient
hfilm determined by
Ti −
Tw, the convective heat transfer coefficient
hconv and the condensation heat transfer coefficient
hcond determined by
Tg − Ti. The energy conservation equation is as follows [
29]:
where:
q—total heat transfer flux;
htot—total heat transfer coefficient;
hfilm—film heat transfer coefficient;
hconv—convective heat transfer coefficient;
hcond—steam condensation heat transfer coefficient;
Tg—mixture gas temperature;
Ti—interface temperature;
Tw—surface temperature of the plate.
As can be seen from Equation (2), the total heat transfer coefficient htot is related to the liquid film heat transfer coefficient hfilm, convective heat transfer coefficient hconv and steam condensation heat transfer coefficient hcond, as well as the mixture gas temperature, interface temperature and surface temperature. So, in order to obtain the final total heat transfer coefficient, it is necessary to solve the hfilm, hconv and hcond one by one, and determine the interface temperature Ti.
For the film heat transfer coefficient
hfilm, generally, the Nusselt theory and the modified equation based on the Nusselt theory are widely used to calculate the heat transfer coefficient of liquid film. However, according to the Nusselt theory, the liquid film is assumed to be laminar flow; when the vertical surface is too long, the liquid film fully develops, and the flow in the liquid film gradually changes from laminar flow to wavy laminar flow and then to turbulence flow. Using the solution based on the Nusselt theory will simply produce a certain deviation, so the following formula is adopted for laminar flow when
Re < 30 [
29]:
Kutateladze’s relation is adopted for wavy laminar flow when 30 ≤
Re < 1800:
Labuntsov’s relation is adopted for turbulent condensate flow when
Re ≥ 1800:
where:
kl—thermal conductivity of the liquid;
L—height of the vertical plate;
—dynamic viscosity of the liquid;
—kinematic viscosity of the liquid;
g—gravitational acceleration;
Pr—Prandtl number.
Actually, the condensation process is cooled further to some average temperature between Ti and Tw, releasing more heat in the process. Therefore, the actual heat transfer will be larger. Rohsenow suggested that the cooling of the liquid below the saturation temperature can be accounted for the modified latent heat of vaporization , defined as , where is the specific heat of the liquid at the average film temperature.
After calculating
Re, the liquid film heat transfer coefficient
hfilm is finally calculated as follows:
For the convective heat transfer coefficient
hconv, according to the numerical correlation of the convective heat transfer similarity criterion without considering the convective heat transfer coefficient under the influence of normal mass transfer, it can be calculated as follows:
The
Nu number in the equation above refers to the relationship between turbulent natural convection and forced convection over a flat plate according to different convection forms:
For mixed convection, the following equation is adopted [
29]:
where
is the Nusselt number for the mixed flow, and
and
are the Nusselt number calculated via forced convection correlation and natural convection correlation under given conditions, respectively. A positive sign was taken when the two flow directions were the same, a negative sign was taken when they were opposite, and for an uncertain flow direction, the minimum value of
was adopted via conservative estimation. The exponent, n, is usually 3.
For the steam condensation heat transfer coefficient
hcond, the Kreith model based on the principle of heat/mass transfer analogy (HMTA) in the diffusion boundary layer was used for calculation [
13,
29]:
where:
—steam condensation rate per unit area;
k—thermal conductivity of mixture in diffusion boundary layer;
—steam diffusion coefficient;
—molar mass of steam;
—total mixture pressure;
—partial pressure of steam at phase interface;
—partial pressure of steam in main flow;
—mean partial pressure of noncondensable gas in gas phase;
Sc—Schmidt number.
In calculation, it is assumed that the steam at the interface is saturated, and that film condensation occurs on the wall surface. After condensation, the liquid film is uniformly attached to the wall surface. At the same time, the thickness of the film can be expressed by combining the condensing mass quality and density of the condensed water and the heat transfer area of the wall surface.
After calculating the condensation rate, the heat transfer coefficient of condensation is further calculated as follows:
For the total heat transfer coefficient htot, since the calculations of hfilm, hconv and hcond all require the interface temperature Ti in order to obtain the total heat transfer coefficient htot, the liquid film heat transfer coefficient, convective heat transfer coefficient and steam condensation heat transfer coefficient need to be calculated successively by assuming the initial value of the liquid film surface temperature, and then a new Ti is obtained through the equations above, and an iteration is performed until the difference between the new and old interface temperatures Ti reaches an acceptable error; then, the final interface temperature Ti is obtained. After that, the total heat transfer coefficient htot can be obtained.
4. Parameter Sensitivity Analysis
After the verification and validation of the condensation model, it is necessary to carry out an appropriate analysis for the sensitivity parameters that affect the calculation results. There are many factors that affect the condensation heat transfer coefficient of a heat structure. The setting of the system parameters and the selection of the calculation models can both cause fluctuations in the results. In order to further investigate the accuracy and reliability of the model, this paper conducted a sensitivity analysis of the code by varying the geometric parameters of the system and calculating models, respectively.
4.1. Mesh Node Independence in Heat Structure
The node division of a heat structure affects the surface temperature to a certain extent. The denser the node division, the closer the node temperature to the real situation, which will then affect the physical property parameters of the fluid. For example, in the calculation of the convective heat transfer coefficient in Equations (8) and (9), the Re, Gr and Pr numbers all depend on the physical property parameters, such as the specific heat and viscosity, which may have a certain influence on the calculation results. Therefore, the number of nodes divided in the heat structure is analyzed as a sensitive parameter in this paper.
Table 3 and
Table 4 provide a detailed analysis of the influence of node numbers on the heat transfer coefficient of the condensing plate in different directions when they are divided. In this paper, the heat structure was divided into two, five and ten nodes along the heat transfer direction, respectively. After the calculation was completed, the comparison of the results revealed that the heat transfer coefficient of the heat structure’s wall was hardly influenced by the number of nodes, and the deviation of the calculation results for each case was within 5‰, which can be considered negligible. In fact, because the wall thickness of the condensing plate is small, the overall thermal conductivity and thermal resistance is small, which can make the nodes in the heat structure reach the thermal equilibrium state in a very short time; therefore, the differences caused by thermal diffusion can be completely ignored. Consequently, the number of nodes in the heat structure will not significantly affect the magnitude and range of the heat transfer coefficient.
4.2. Calculation Model for Mass Diffusivity
The physical parameters of water, steam, noncondensable gases and various materials are repeatedly used in the calculations, so the calculated values of these physical parameters affect the accuracy of the results to some extent. It can be seen from Equation (11) that the value of the mass diffusion coefficient DAB has an impact on the mass condensation rate. Different calculation models of the mass diffusion coefficient will cause a disturbance to the condensing amount of steam per unit area.
According to the principle of the heat/mass transfer analogy, combined with Equations (12) and (13), it can be seen that the condensing heat transfer rate is also further affected by DAB. Therefore, a sensitivity analysis of the mass diffusion coefficient is performed here.
When representing the mass diffusivity of steam to other mixed noncondensable gases, the Wilke and Lee model and the Fuller model were selected in this paper to calculate the mass diffusion rate of steam to a single noncondensable gas [
30], and after obtaining the binary mass diffusivity, the final calculation result was obtained using Wilke’s model for the mass diffusivity of steam to mixed gases.
The Wilke and Lee model is expressed as
The Fuller model is expressed as
where:
DAB—mass diffusion coefficient;
P—total gas pressure;
T—gas temperature;
M—molar mass.
Parameter values such as
σAB, Ω
D and
∑v can be obtained by looking up the parameter table [
30].
By selecting two different binary mass diffusivity calculation models, the final heat transfer coefficient obtained from the code was compared with the experimental data from the Wisconsin condensation heat transfer experiment, as shown in
Figure 4. The results show that although there is a significant difference between the condensation heat transfer coefficient calculated using the code and the experimental data, the results calculated using different mass diffusivity models are indeed different, and the deviation is about 5–10%. The results from the Wilke and Lee model were significantly closer to the experimental data than those from the Fuller model, indicating that physical parameters have an effect on the accuracy of the calculated results.
4.3. Condensation Model
The condensation phenomenon of steam containing noncondensable gas plays an important role in the heat removal process of the containment in a nuclear power plant under an accident condition. As can be seen from Equation (2), the total heat transfer coefficient is affected by both the convective heat transfer coefficient and the condensation heat transfer coefficient, and there are certain differences in the calculation results obtained using the different condensation calculation models. So, it is important to select an accurate condensation model that is suitable for the development of a severe accident analysis code. In this paper, the most widely used experimental correlation models of Uchida [
7], Tagami [
8] and Dehbi [
9] are selected successively to compare with the experimental results and the calculated values of the PISAA code model.
where:
h—condensing heat transfer coefficient;
Wnc—noncondensable gas mass fraction; x
nc—noncondensable gas volume fraction;
L—characteristic length of condensing wall;
p—pressure;
Tg—gas temperature;
Tw—condensing wall temperature.
The comparison results are shown in the
Figure 5.
The comparison shows that the calculation results of the condensation model based on Fick’s law and the heat/mass transfer analogy principle chosen in this paper are closer to the experimental measurement data from the Wisconsin experiment. The deviation between the calculation results from the Tagami model and the experimental data is the largest, and the accuracy of the Dehbi model is slightly higher than that of Tagami and Uchida, but the deviation is much larger than that of the calculation model used in this paper.
To analyze the reason, the Uchida and Tagami models only consider the influence of the noncondensable gas fraction on the condensation heat transfer coefficient, and the Dehbi model is further improved, considering the influences of wall length, total pressure and wall subcooling degree on the condensation heat transfer coefficient, but the applicability range of the model is limited, and the factors affecting the mass and heat transfer of condensation, such as the temperature and composition of the gas in the main flow and the interface, are not considered in detail. The calculation model used in the PISAA code considers all the factors mentioned above; therefore, the calculation results are closer to the actual experimental measurements, and at the same time, simulate better.
The comparison results show that the condensation heat transfer model used in this paper, which is based on simplifying and integrating the mechanism and the experimental correlations, can effectively solve the limitations of the traditional empirical correlations model, and realize the accurate simulation of the condensation heat transfer of steam with noncondensable gas, and the calculation results are more reliable than the results calculated using the pure empirical correlations.
5. Conclusions
Based on the law of energy conservation and the thermal resistance relationship of each heat transfer process, combined with the mechanism and experimental correlations, this paper integrates a comprehensive calculation model of the condensation heat transfer of steam with noncondensable gas, including convection heat transfer, liquid film heat transfer and steam condensation heat transfer, and also considers the influence of various factors such as the condensation amount and flow patten classification in the liquid film on the process. The total heat transfer coefficient is obtained via the iterative method, the model is more applicable under different conditions and the calculated results are closer to the experimental data.
The condensation model is then used in an analysis code, PISAA, and the model is verified by comparing the calculation results with those from traditional containment analysis codes towards the same example, the change trend is basically consistent and the final steady-state values are only in a small variation range. And in the calculation and validation of the condensation heat transfer experiment in Wisconsin using the PISAA software, the average deviation of the heat transfer coefficient of the horizontal condensation heat transfer is 2.55%, and the maximum calculation deviation of the vertical condensing plate is about 29%, which is within the acceptable range of engineering. The reason for the large deviation may be that the simplified method of condensate film thickness in the vertical wall is not consistent with the actual situation, which can be further optimized and improved in the future.
Through the analysis of the sensitive parameters affecting the heat transfer coefficient, the division of the nodes on the condensing plate is less, that is, the temperature distribution has a little affect on the heat transfer coefficient. For the mass diffusivity calculation model, the maximum deviations between the calculated results obtained using the Wilke and Lee models and the Fuller model and the experimental values are 29% and 36%, respectively, and the physical property parameters affect the accuracy of the calculated results to some extent.
The deviations from different condensation models’ calculated results and the experimental data are much larger than that of the condensation model used in this paper, indicating that the model is scientific and reasonabl, and can be used for thermal hydraulic analysis in a nuclear power plant.
The prototypes of the calculation equations in this paper come from the existing experimental data and references, and are simplified according to the theory and practical situation, integrated and coupled together in the calculation code, PISAA, which solves the problem of calculating the condensation of high-temperature and high-pressure gas in the containment in a nuclear power plant under a severe accident scenario, and has great significance for the accident analysis and simulation of a nuclear power plant and the improvement in the nuclear power safety level.