Numerical simulation of water spray in natural draft dry cooling towers with a new nozzle representation approach

Highlights

• CFD investigation of real spray nozzles for inlet air pre cooling applications.

• Developments of a new adaptable hollow–cone spray representation method.

• Incorporation of experimentally measured spray characteristics into a CFD simulation.

• CFD spray model validation against wind tunnel measurements.

Abstract

Pre-cooling of inlet air with water spray is proposed for performance enhancement of natural draft dry cooling towers (NDDCTs) during high ambient temperature periods. Previous experiments showed promising results on cooling enhancement using spray cooling. In this study, a numerical Eulerian–Lagrangian 3-D model is used to simulate evaporating water sprays produced by real nozzles. A new adaptable method of hollow-cone spray representation in Eulerian–Lagrangian numerical models was developed to reproduce the real nozzle behaviour using experimentally measured initial spray characteristics and taking into account radial evolution of droplet size distribution and air/droplets momentum exchange. Experimental measurements from a wind tunnel test rig simulating NDDCTs inlet flow conditions have been performed for validation. Overall, a good agreement was obtained between numerical predictions and experimental measurements for the streamwise development of droplet size and velocity, and outlet air dry bulb temperature. An average deviation below 5.3% was achieved for all compared parameters. Moreover, the validated CFD model has provided insight into the experimental observations of local droplet velocity increase and higher air cooling in the lower region.

Introduction

This paper presents the development and validation of an Eulerian–Lagrangian water spray CFD model incorporating experimentally measured initial spray characteristics obtained from the wind tunnel measurement using a new nozzle representation method. A good nozzle model is crucial for designing an effective inlet air spraying system for performance enhancement of natural draft dry cooling towers (NDDCTs). Due to a number of reasons including water consumption restrictions, environmental regulations and flexibility of plant site selection [1], [2], air-cooled condensers in general, and NDDCTs in particular, are becoming the preferred choice for many power plants despite their higher capital costs and the reduced performance at high ambient air temperatures. The performance of dry cooling towers can be enhanced on hot days by various techniques [3], [4], [5], including water sprays.

In spray cooling, water is distributed into the inlet air to reduce the inlet air temperature by evaporative cooling prior to reaching the condensers. This increases the overall cycle efficiency and helps the plant recover some of the performance reduction caused by hot ambient temperatures. An effective water spray design needs to avoid non-uniform cooling distribution and incomplete evaporation of droplets. These issues can be avoided and an optimum design can be achieved only if the spray cooling mechanisms under these conditions are well understood.

Due to its simplicity, ease of operation and maintenance, and ease of operation and maintenance, spray cooling is becoming more popular due to its simplicity, low capital price, and ease of operation and maintenance [6]. Spray nozzles are used to help distribute water into the inlet airflow in order to provide a large contact surface area between air–water and to enhance mixing by producing very fine droplets. This offers higher evaporation rate and greater air cooling.

Detailed knowledge on the impact of physical parameters of two-phase flow in spray cooling systems is crucial for designing an effective spray cooling system. According to Wells [7], spray droplet size is a main parameter that impacts droplet transport and cooling efficiency. Spray cooling performance is also strongly influenced by air velocity [6], [8], [9]. Some of the other important parameters are: (a) nozzle cone angle, (b) water injection rate, (c) droplet velocity at nozzle exit, (d) injection direction, and (e) meteorological condition [8], [9], [10], [11], [12].

A number of experimental studies [1], [11], [13], [14] and computational fluid dynamics models [9], [12], [14], [15], [16], [17], [18] have been carried out on spray cooling performance. Spray is a two-phase flow phenomenon and experience several actions when injected into air including heat, mass and momentum transfer. Experimental analysis for such complex flow involving droplet dynamics is costly and challenging method. CFD is a good tool to analyse the two-phase flow. CFD offers the advantage that full-field local data can be obtained which can help understand the process, e.g. local increase in droplet flow velocity downstream the nozzle as discussed in section 5.2. In addition, it allows the control of physical parameters of the two-phase flow independently, which sometimes may not be possible with the available nozzles or very expensive and time consuming, e.g. small droplet size distribution and high flow rate. CFD also gives more control on boundary conditions. However, CFD simulation of water spray flow can be challenging due to strong coupling between the two phases and droplet evaporation. Moreover, exact spray initiation into CFD simulation is a challenging issue.

The study of spray behaviour injected into airflow has implications on many engineering applications. Several numerical investigations have been conducted on spray cooling systems to investigate the effect of physical parameters on spray cooling efficiency and droplet transport. They used available experimental spray characteristics with different initiation approach for their spray injection. A numerical investigation conducted by Tissot [12] on a small channel associated with refrigerating system has shown that there is a trade-off between air velocity and droplet size for better spray cooling efficiency due to the compromise effect of evaporation rate and momentum exchange. Chaker [19] numerically studied the effect of droplet size on cooling performance and efficiency of gas turbine inlet fogging using a single virtual injection. It was found by Chaker [19] that droplet size and residence time are the most important effective parameters on spray cooling efficiency. A computational fluid dynamics model was developed by Wang [15] to study mist transport at different fundamental geometries. Effect of droplet size on cooling performance was investigated. Actual drop size distribution was implemented into the numerical model with $D_{v90}=40\mu m$. However, no information was provided on how spray was injected. In addition, Montazeri et al. [9] performed a CFD analysis on cooling performance of a water spray system under differ physical conditions. Previous studies related to spray cooling systems used experimentally measured spray characteristics with different initiating technique for their spray injection. They showed that, within their tested range, inlet air velocity and droplet size distribution are two important parameters in the cooling performance of the system. However, no attempt was made in these studies reviewed so far related to spray cooling systems to accurately represent initial spray characteristics in the simulation, which requires more investigation.

Although CFD simulation for water spray performance is a powerful tool, it is a challenging task with respect to spray initialization into the Lagrangian modelling approaches [20]. While there have been improvements in spray characterization methods, it is very difficult to accurately predict initial spray characteristics with available breakup models [20]. Moreover, break-up models are not available for all nozzle types. One of the difficulties arises from the location where initial spray characteristics can be measured. Atomization or jet breakup takes place in a certain distance away from the nozzle exit depending on nozzle type and operating conditions. Experimental spray characterization is obtained as close as possible to the nozzle orifice at the breakup length (the distance between the nozzle exit and the location at which atomization starts). Therefore, the spray characteristic at the nozzle tip is not available to be incorporated into the numerical model as initial droplet size distribution and injection velocity.

The performance of spray cooling systems involving particle spray is determined by their initial injection parameters [6], [21]. Various numerical investigations showed the importance of the accuracy of the representation of the nozzle initial characteristics in the simulation in order to be validated with the experiments [22]. Spray performance has been evaluated numerically with different approaches for prescribing the initial spray characteristics. One approach was to use uniform droplet size distribution and velocity as initial spray characteristics in order to study the impact of different physical parameters on the system performance [6]. Another approach was to assume that the breakup process occurs straight at the nozzle tip and droplets are formed [22], [23]. With this assumption, the measured spray characteristics are averaged and extrapolated back to the nozzle location by adopting theoretical or empirical mathematical functions to prescribe the initial spray characteristics. Some studies used a single virtual injection back at the nozzle location. The initial spray characteristics were adjusted manually to match the predicted spray characteristics at the breakup length with measurements [24], [25], [26], [27]. A further approach was to inject droplet at a distance where droplets become more uniform across the spray plume and move at the same speed [28]. They employed an air jet to take into account the disregarded momentum exchange from droplets to air before the injection location. Some studies defined the initial spray characteristics with a single circle injection at the breakup region by averaging the measured spray characteristics [29].

An alternative approach is to define the initial spray characteristics at the breakup length with different circles to take into consideration the radial variation of the droplet size distribution within the spray plume [30]. While this approach has accounts for the radial variation of droplet size, the effect of droplet velocity on the airflow is not accounted for. This ignores the momentum exchange between the airflow and the water jet in the primary atomization region. Since there is no method available in the literature that accounts for radial evolution of size distribution in conjunction with momentum exchange at the primary atomization region, a new technique of nozzle representation in numerical simulation is introduced in this study without unduly increasing the complexity of the numerical model.

In the present work, an Eulerian–Lagrangian 3-D CFD model for evaporating water spray is developed and validated against experimental measurements obtained from a wind tunnel test rig under the same operating conditions as NDDCTs. Spray characteristics were measured under the same experiment condition using a Phase Doppler Particle Analyser (PDPA) and a high speed photography system. A new adaptable method of hollow-cone spray representation in the simulation is introduced to reproduce the real nozzle behaviour taking into account radial evolution of droplet size distribution and air/droplets momentum exchange at the primary atomization region. The present model is a good design tool to study the impact of physical parameters on spray cooling systems performances using the developed spray initialization approach.