Nonrealizability Problem With Quadrature Method of Moments in Wet-Steam Flows and Solution Techniques

Author(s):  
Ali Afzalifar ◽  
Teemu Turunen-Saaresti ◽  
Aki Grönman

The quadrature method of moments (QMOM) has recently attracted much attention in representing the size distribution of liquid droplets in wet-steam flows using the n-point Gaussian quadrature. However, solving transport equations of moments using high-order advection schemes is bound to corrupt the moment set, which is then termed as a nonrealizable moment set. The problem is that the failure and success of the Gaussian quadrature are unconditionally dependent on the realizability of the moment set. First, this article explains the nonrealizability problem with the QMOM. Second, it compares two solutions to preserve realizability of the moment sets. The first solution applies a so-called “quasi-high-order” advection scheme specifically proposed for the QMOM to preserve realizability. However, owing to the fact that wet-steam models are usually built on existing numerical solvers, in many cases modifying the available advection schemes is either impossible or not desired. Therefore, the second solution considers correction techniques directly applied to the nonrealizable moment sets instead of the advection scheme. These solutions are compared in terms of accuracy in representing the droplet size distribution. It is observed that a quasi-high-order scheme can be reliably applied to guarantee realizability. However, as with all the numerical models in an Eulerian reference frame, in general, its results are also sensitive to the grid resolution. In contrast, the corrections applied to moments either fail in identifying and correcting the invalid moment sets or distort the shape of the droplet size distribution after the correction.

Author(s):  
Ali Afzalifar ◽  
Teemu Turunen-Saaresti ◽  
Aki Grönman

The quadrature method of moments (QMOM) has recently attracted much attention in representing the size distribution of liquid droplets in wet-steam flows using the n-point Gaussian quadrature. However, solving transport equations of moments using high-order advection schemes is bound to corrupt the moment set, which is then termed a non-realizable moment set. The problem is that the failure and success of the Gaussian quadrature is unconditionally dependent on the realizability of the moment set. First, this article explains the non-realizability problem with the QMOM. Second, it compares two solutions to preserve realizability of the moment sets. The first solution applies a so-called “quasi-high-order” advection scheme specifically proposed for the QMOM to preserve realizability. However, owing to the fact that wet-steam models are usually built on existing numerical solvers, in many cases modifying the available advection schemes is either impossible or not desired. Therefore, the second solution considers correction techniques directly applied to the non-realizable moment sets instead of the advection scheme. These solutions are compared in terms of accuracy in representing the droplet size distribution. It is observed that a quasi-high-order scheme can be reliably applied to guarantee realizability. However, as with all numerical models in an Eulerian reference frame, in general its results are also sensitive to the grid resolution. In contrast, the corrections applied to moments either fail in identifying and correcting the invalid moment sets, or distorts the shape of the droplet size distribution after the correction.


2017 ◽  
Vol 140 (2) ◽  
Author(s):  
Ali Afzalifar ◽  
Teemu Turunen-Saaresti ◽  
Aki Grönman

This paper investigates the performance of moment-based methods and a monodispersed model (Mono) in predicting the droplet size distribution and behavior of wet-steam flows. The studied moment-based methods are a conventional method of moments (MOM) along with its enhanced version using Gaussian quadrature, namely the quadrature method of moments (QMOM). The comparisons of models are based on the results of an Eulerian–Lagrangian (E–L) method, as the benchmark calculations, providing the full spectrum of droplet size. In contrast, for the MOM, QMOM, and Mono an Eulerian reference frame is chosen to cast all the equations governing the phase transition and fluid motion. This choice of reference frame is essential to draw a meaningful comparison regarding complex flows in wet-steam turbines as the most important advantage of the moment-based methods is that the moment-transport equations can be conveniently solved in an Eulerian frame. Thus, the moment-based method can avoid the burdensome challenges in working with a Lagrangian framework for complicated flows. The main focus is on the accuracy of the QMOM and MOM in representing the water droplet size distribution. The comparisons between models are made for two supersonic low-pressure nozzle experiments reported in the literature. Results show that the QMOM, particularly inside the nucleation zone, predicts moments closer to those of the E–L method. Therefore, for the test case in which the nucleation is significant over a large proportion of the domain, the QMOM provides results in clearly better agreements with the E–L method in comparison with the MOM.


Author(s):  
Ali Afzalifar ◽  
Teemu Turunen-Saaresti ◽  
Aki Grönman

The method of moments offers an efficient way to preserve the essence of particle size distribution, which is required in many engineering problems such as modelling wet-steam flows. However, in the context of the finite volume method, high-order transport algorithms are not guaranteed to preserve the moment space, resulting in so-called ‘non-realisable’ moment sets. Non-realisability poses a serious obstacle to the quadrature-based moment methods, since no size distribution can be identified for a non-realisable moment set and the moment-transport equations cannot be closed. On the other hand, in the case of conventional method of moments, closures to the moment-transport equations are directly calculated from the moments themselves; as such, non-realisability may not be a problem. This article describes an investigation of the effects of the non-realisability problem on the flow conditions and moment distributions obtained by the conventional method of moments through several one-dimensional test cases involving systems that exhibited similar characteristics to low-pressure wet-steam flows. The predictions of pressures and mean droplet sizes were not considerably disturbed due to non-realisability in any of the test cases. However, in one case that was characterised by strong temporal and spatial gradients, non-realisability did undermine the accuracy of the predictions of measures for the underlying size distributions, including the standard deviation and skewness.


2017 ◽  
Vol 2 (1) ◽  
pp. 1-10
Author(s):  
HUSSEIN WHEEB MASHI

     This research is to find a wetness using the (laser beam) optical Forward Scattering Method (F.S.M.) which is applicable to calculate the wetness in a convergent-divergent De-Laval steam nozzle operating (1.3 Mach number) which may be works at wet steam with pressure (1) bar and temperature (373) K .The light source of He-Ne laser of wave light (λ=0.632) µm was used to prediction the wetness in nozzle a wet steam flow. Both  droplet diameter of water (Dr) and relation of intensity of light S= /  are assumed to be (Dr=10,30,50,70,100 µm) and (S = 0.9,.0.8,0.7,0.6,0.5 )  respectively. From the relation of light intensity pattern of many diameter droplet of water (Dr) and different droplet size distribution N (Dr), the MATLAB program can calculate the light attenuation coefficient (Ks) consequently. The increase of the droplet size distribution, N (Dr), leads to decrease the values of (S) and (Ks). The increase of the droplet diameter causes increase of the scattered light, and the minimum value of scattering light is with (Dr= 10µm) for the tested samples. The wetness of steam (yₒ) = (1,3,557,10) % ,which depend on the [Dr, N(Dr)] in the scattered zone  can be determined easily by the MATLAB program. The radius of droplet water in two – phase can be adversely calculated by using the research output that then the concentrations or wetness is previously specified.                                                                                        


Author(s):  
Rutger H. A. IJzermans ◽  
Rob Hagmeijer ◽  
Ryan S. R. Sidin

Condensing flows can be found in a large variety of industrial machinery such as steam turbines and supersonic gas conditioners. In many of these applications, it is very important to predict the droplet size distributions accurately. In the present research, the droplet size distribution in condensing flows is investigated numerically. We consider condensing flows with droplets that nucleate and grow, but do not slip with respect to the surrounding gas phase. To compute the coupling between the condensed phase and the carrier flow, one could solve the general dynamic equation and the fluid dynamics equations simultaneously. In order to reduce the overall computational effort of this procedure by roughly an order of magnitude, we use an alternative procedure, in which the general dynamic equation is initially replaced by moment equations complemented with a closure assumption. This closure assumption is based on Hill’s approximation of the droplet growth law. The method thus obtained, the so-called Method of Moments, is assumed to approximately accommodate the thermodynamic effects of condensation, such as the temperature, pressure and velocity field of the carrier flow. We use the Method of Moments as a basis for the calculation of the droplet size distribution function. We propose to solve the general dynamic equation a posteriori along a number of selected fluid trajectories, keeping the flow field fixed. This procedure, called Phase Path Analysis [1], leads to accurate size distribution estimates, at a far lower computational cost than solving the general dynamic equation and the fluid dynamics equations simultaneously. In the present paper, we investigate the effect of a variation in the liquid mass density on the droplet size distribution, using the proposed method. In case of a varying liquid mass density, both the equation for the dropltet growth rate and the moment equations are modified. This modified form coincides with the usual form of the moment equations in the event that the variation in liquid density is negligible. This research is relevant for condensation in flows where large temperature differences may occur which lead to significant variations in the liquid mass density. We show that the implementation of a variable liquid mass density in the Method of Moments and the Phase Path Analysis results in a higher extremum in the droplet size distribution, whereas the skewed shape of the distribution function is nearly similar to that obtained in the constant liquid density case.


2006 ◽  
Vol 16 (6) ◽  
pp. 673-686 ◽  
Author(s):  
Laszlo E. Kollar ◽  
Masoud Farzaneh ◽  
Anatolij R. Karev

Sign in / Sign up

Export Citation Format

Share Document