Elsevier

Applied Thermal Engineering

Volume 110, 5 January 2017, Pages 1589-1608
Applied Thermal Engineering

Research Paper
A comprehensive performance investigation of cellulose evaporative cooling pad systems using predictive approaches

https://doi.org/10.1016/j.applthermaleng.2016.08.216Get rights and content

Highlights

  • Performance of evaporative pad was modeled using SCST approaches.

  • SCST models were used to predict the supply temperature and pressure drop.

  • It was shown that SCST are more accurate than analytical models.

  • Sensitivity analysis of evaporative pad was studied using the SCST model.

  • The effect of conditioned air recirculation was studied in hot areas.

Abstract

Developing the soft computing and statistical tools (SCST) for predicting the behavior pattern of the performance features of a cellulose evaporative cooling pad system was studied. Three soft computing and statistical tools- artificial neural network (ANN), genetic programming (GP), and multiple linear regression (MLR)- were used to predict the supply air temperature and pad pressure drop. The prediction abilities of obtained models were analyzed and compared with analytical models, and a comprehensive error analysis was conducted. It was found that the MLR and ANN models perform better than the other approaches for predicting the supply air temperature and the pad pressure drop, respectively. The obtained models had the accuracy of numerical models as well as the simplicity of analytical methods. Effects of inlet air conditions and pad characteristics on nine different system performance parameters like thermal comfort indices were also studied, comprehensively. It was found that the best values for pad thickness and specific contact area are the minimum values of them, which provide thermal comfort conditions (7 cm and 420 m2 m−3 for the investigated case respectively). Utilizing the direct evaporative cooling system with recirculation of a part of the cooled air in very hot and dry weather conditions was investigated and suggested as an alternative for conventional systems.

Introduction

In conjunction with the exacerbation of global energy shortage, increasing the cost of energy, and instinctively recognized environmental issues, interesting in the direct evaporative cooling (DEC) has revived, which offers a low-cost alternative to conventional direct expansion (DX) units. Being both energy-efficient and environmental-friend makes the DEC system significantly better than the mechanical vapor compression system [1]. Although the DEC systems are used widely in many non-humid areas of the world, these systems cannot be utilized in humid areas as the performance of these systems depends on the wet-bulb temperature of ambient air [2]. The investigation showed that in order to utilize the DEC, it is vital to consider the type of the conditioned space (office, home, or storage areas), and the wet-bulb temperature of the ambient air as the minimum achievable temperature.

DEC is the oldest and simplest form of evaporative cooling, in which the air contacts directly with the water. The most typical DECs utilize the pads, with large flat vertical air filters in their walls, which are embedded in metal cubes or plastic boxes. As shown in Fig. 1, the pads keep the moisture of the water and are wetted uniformly and fully by the water sprayed. They increase the contact area between air and water, hence, provide a higher mass transfer rate of water to the air stream. Both the cellulose and PVC pads as illustrated in Fig. 2 are the most common among various types of pads utilized in DECs. The water is supplied from the top of the media and trickled down, due to consisting of wettable porous material. The process air is pulled by unit’s fan within the coolers that is cooled and humidified in the channels through the pads. Finally, the air leaves the cooler to conditioned space as a washed and cooled air. Moreover, in order to regulate the leaving air states by users, many coolers utilize either two or three-speed fans. As you see the schematic of a drip-type DEC system in Fig. 3, the water uniformly distributed via gravity and capillarity and the water pump recirculates the falling water from the water basin. The process air not only does it contacts directly with the sprayed water, but is cooled and humidified by water evaporation [3], [4]. The thermal process on the psychrometric chart was depicted in Fig. 4.

The first attempts to study DECs were mainly focused on testing of their overall performance. The results of this testing provided technical guidance for local engineering applications. In recent years, the studies were mostly focused on the experimental and numerical analyses of these systems. These studies were reported by Camargo et al. [5], Wu et al. [4], [6], Malli et al. [7], Sheng and Agwu Nnanna [8], Franco et al. [9], and He et al. [10]. The models and correlations in the theoretical and numerical analyses were validated by the test results based on some reasonable assumptions and with those of experiments for the same evaporative cooler. Camargo et al. [5] developed a mathematical model for analyzing the DEC system, determining the effectiveness of saturation. Besides, they presented the experimental results of the tests performed on a DEC. Wu et al. [4] investigated the influences of the air frontal velocity and the thickness of pad module on the cooling efficiency of a DEC by proposing a simplified correlation based on the energy balance analysis of the air. The results indicated that the optimum frontal velocity of the pad module should be around 2.5 m s−1 for the required supply air volume. Moreover, they proposed that this recommended value can be used to decide the frontal area of pad modules in a DEC. Wu et al. [6] carried out a numerical simulation of a cross flow direct evaporative cooler, which the latent heat water evaporation was taken as a heat source in the energy equation. The models and methods were validated by comparing the numerical results with those of experiment for the same evaporative cooler. In their survey, the effects of the inlet frontal air velocity, pad thickness, inlet air dry-bulb and wet-bulb temperatures on the cooling efficiency of the evaporative cooler were studied. The results showed that the DEC with high-performance pad material may be well applied for evaporative coolers via reasonable choices for the inlet velocity and pad thickness. Malli et al. [7] presented a thermal study of two kinds of pads which were tested in a sub-sonic wind tunnel made from polyethylene. For several inlet air velocities, pressure drop, humidity variation, evaporated water and effectiveness had been investigated. Sheng and Agwu Nnanna [8] established an empirical equation between inlet air velocity and cooling efficiency of DEC. They determined that lower velocities and lower cooling water temperatures can lead to higher efficiencies. Franco et al. [9] carried out a comparative study between one evaporative cooling box and four pads whose geometry and thickness -manufactured by two different companies- were varied. It was demonstrated that the plastic packing in the cooling unit produced a pressure drop of 11.05 Pa at 2 m s−1, which was between 51.27% and 94.87% lower than that produced by the cellulose pads. Furthermore, they presented that the saturation efficiency of the evaporative cooling box was 20% higher than the saturation efficiency of the cellulose pads for the same flow. He et al. [10] presented an experimental analysis for one type of cellulose corrugated medium for the good insights regarding the performance of cellulose corrugated media. They measured the heat transfer coefficient, cooling efficiency and pressure drop across the medium with various thicknesses in a wind tunnel. It was found that in both larger medium thickness and higher air velocity, the higher pressure drop might occur.

According to the aforementioned survey of previous works, for accurately predicting the performance of DECs, it is obviously required to solve conventional mathematical models consisting of complex differential and analytical equations. Soft computing and statistical tools (SCST) not only do they save time and capital investment but as it was seen they are very accurate for modeling other types of evaporative coolers. Kiran and Rajput [11] predicted the performance of an IEC using soft computing approaches such as artificial neural network (ANN) fuzzy inference system (FIS) and adaptive neuro-fuzzy inference (ANFIS). According to the results, ANN was the most accurate SCST for prediction of IEC outlet temperature and efficiency which the corresponding value of the coefficient of determination (R2) were 0.9999 for both parameters. A cross-flow Maisotsenko (M-cycle) indirect evaporative cooler was modeled by Pandelidis and Anisimov [12] using a statistical approach called as response surface methodology (RSM). The satisfactory magnitude of statistical parameters including R2 was observed that meant this simplified model is a high accurate model for predicting the performance of the cooler. Sohani et al. [13] developed a statistical model in order to determine product air temperature of an M-cycle cross-flow indirect evaporative cooler by employing the group method of data handling (GMDH) type neural network. The results showed that the obtained model had a high value of R2 (0.999167) in the prediction of validation data. Then they employed their model to perform multi-objective optimization of the system in different applicable climates in the world. In another research Sohani et al. [14] selected the best SCST for prediction the performance of each part of desiccant enhanced evaporative (DEVAP) air conditioner among different models including ANN, GMDH, genetic programming (GP), multiple linear regression (MLR) and stepwise regression method (SRM). The results showed that SRM and GMDH were the best models for prediction of the dehumidifier (first part of the DEVAP system) and M-cycle counter-flow IEC (second part of the DEVAP system), respectively. They concluded that SRM model was the most accurate one if the second part was perforated counter M-cycle IEC. Also, it was found that it was possible to describe the performance of the DEVAP system precisely and easily. In addition to evaporative coolers, there has been a considerable interest among researchers to use of SCST for modeling the other air conditioning and energy systems and their equipment. Pacheo-Vega et al. carried out heat rate estimations of heat exchangers used for both refrigeration applications [15] and humid air–water heat exchangers [16] using correlations and ANN model. Hosoz et al. [17] predicted the performance of a cooling tower under a broad range of operating conditions. Şencan et al. [18] applied artificial neural networks (ANN) approach to predict the thermodynamic properties of an absorption system under limited experimental data and analytical functions. Amber et al. [19] developed genetic programming (GP) and Multiple linear regression (MLR) approaches to forecast electricity consumption of an administration building. Platon et al. [20] carried out the hourly prediction of a building’s electricity consumption using two artificial intelligence techniques (ANN) and case-based reasoning (CBR) to develop predictive models to facilitate generalization to other buildings. Therefore, these approaches are generally useful for systems whose behaviors are complex and dependent on various variables.

In this study, soft computing and statistical tools (SCST) including artificial neural networks (ANN) genetic programming (GP) and multiple linear regression model (MLR) have been considered to predict the overall performances of an evaporative cooling cellulose pad system in a simple and accurate way. The use of these modeling techniques in predicting the performance of evaporative cooling has been thoroughly discussed. Data for training, validating, and testing the models, have been extracted from numerical and experimental studies so as to predict the quantitative changes of two main performance features, supply air temperature, and pad pressure drop. The prediction ability of obtained models was analyzed and compared not only with experimental data, which were not being used for models development process but by error criterion including sum square error (SSE), mean square error (MSE) and coefficient of determination (R2). Based on the error analysis, the more accurate SCST models for prediction of supply air temperature and pad pressure drop were selected and an extensive and in-depth investigation of seven parameters (the inlet air temperature, the relative humidity, the air velocity, the pad thickness, the specific pad contact area and the pad incidence angles), which have substantial impacts on a broad range of performance criteria, was performed. This investigation was done by studying the effects of variation of these seven parameters on the product air temperature and relative humidity, wet bulb effectiveness, water consumption, and thermal comfort parameters. In Most researchers about DECs, only a few performance parameters were studied without considering other parameters such as provided thermal comfort condition. In these mentioned researches parametric study was usually dedicated to the only investigation of variation of the pad pressure drop and product air temperature or wet-bulb effectiveness but in the current paper, the effect of variation of all mentioned performance parameters was studied simultaneously. Additionally, the potential of using a fraction of cooled air for pre-cooling inlet air was analyzed and investigated for very hot and dry areas where conventional DECs (DECs without air pre-cooling) cannot bring thermal comfort.

Section snippets

Soft computing and statistical approaches

Three SCST, including artificial neural network (ANN), genetic programming (GP), and multiple linear regression (MLR) were employed throughout this paper.

ANN is a predictive model based on mimicking the learning process of biological nervous systems with high-speed computers. ANN is suitable for tasks involving incomplete data sets, fuzzy or incomplete information and for extraordinarily complex and ill-defined problems. By training or adjusting, the ANN model can learn how a specific input

Description of parameters

Carrying out an extensive and in-depth investigation was one of the main goals of this paper. In this regard, the potentially significant features of the pad, air, and water, which greatly influence the overall performances of the DEC system, were considered.

Development of models

The details about the development of ANN, GP, and MLR models were presented in this section. As previously mentioned in Section 2, the description of these methods and their mathematical approaches is beyond the scope of this survey. The complete description about ANN, GP and MLR was found in [14], [21], [22], [14], [23], [25], and [14], [24], [25], [26], respectively. The first step to developing these models is determining the dependent and independent variables. The supply air temperature

Supply air temperature

Table 4 compares three models with each other as well as experimental data [8]. It should be mentioned that these experimental data were those data that absolutely used for comparison purpose; thus, they are not those experimental data that were used in the development of the models. There is a satisfactory accommodation between the predictions and the actual results as presented in Table 4. It is evident that, generally, the errors in the MLR model are less than other models i.e. the MLR model

Conclusion

This paper presents three models, i.e. artificial neural network (ANN), genetic programming (GP) and multiple linear regression (MLR) - as soft computing and statistical tools (SCST) - for the predicting of the overall performances of a direct evaporative cooling cellulose pad system which seven independent variables have been used for development of the models. The supply air temperature and the pad pressure drop, as the two main performance features of the DEC system, were evaluated by the

References (44)

  • M. Hosoz et al.

    Performance prediction of a cooling tower using artificial neural network

    Energy Convers. Manage.

    (2007)
  • A. Şencan et al.

    Thermodynamic analysis of absorption systems using artificial neural network

    Renew. Energy

    (2006)
  • K.P. Amber et al.

    Electricity consumption forecasting models for administration buildings of the UK higher education sector

    Energy Build.

    (2015)
  • R. Platon et al.

    Hourly prediction of a building’s electricity consumption using case-based reasoning, artificial neural networks and principal component analysis

    Energy Build.

    (2015)
  • A. Şencan et al.

    A new approach using artificial neural networks for determination of the thermodynamic properties of fluid couples

    Energy Convers. Manage.

    (2005)
  • K. Amber et al.

    Electricity consumption forecasting models for administration buildings of the UK higher education sector

    Energy Build.

    (2015)
  • H. Hasani Balyani

    Acquiring the best cooling strategy based on thermal comfort and 3E analyses for small scale residential buildings at diverse climatic conditions

    Int. J. Refrig.

    (2015)
  • R. Hosseini et al.

    Performance improvement of gas turbines of Fars (Iran) combined cycle power plant by intake air cooling using a media evaporative cooler

    Energy Convers. Manage.

    (2007)
  • M. Barzegar

    Experimental evaluation of the performances of cellulosic pads made out of Kraft and NSSC corrugated papers as evaporative media

    Energy Convers. Manage.

    (2012)
  • S. He

    Pre-cooling with Munters media to improve the performance of Natural Draft Dry Cooling Towers

    Appl. Therm. Eng.

    (2013)
  • P. Rick Phillips

    Using direct evaporative + chilled water cooling

    ASHRAE J.

    (2009)
  • J. Camargo et al.

    An evaporative and desiccant cooling system for air conditioning in humid climates

    J. Braz. Soc. Mech. Sci. Eng.

    (2005)
  • Cited by (71)

    • Development of indirect evaporative cooler based on a finned heat pipe with a natural-fiber cooling pad

      2022, Heliyon
      Citation Excerpt :

      An in-depth investigation of cellulose cooling pads was carried out by Ali et al. Computational and statistical software (SCST) was used to predict the behavior patterns of the performance features of the cellulose evaporative cooler bearing system [24]. In 2016, Hou et al. has done a study on cooling media using cellulose pads [25].

    View all citing articles on Scopus
    View full text