Research Paper
Global sensitivity analysis of a generator-absorber heat exchange (GAX) system’s thermal performance with a hybrid energy source: An approach using artificial intelligence models

https://doi.org/10.1016/j.applthermaleng.2022.119363Get rights and content

Highlights

  • Temperature, heat and flow measurements of an experimental GAX cycle.

  • Evaluation of artificial intelligence (AI) techniques for multivariate modeling.

  • Artificial neural networks represent the GAX system with the best accuracy.

  • AI model coupled to the PAWN method for a global sensitivity analysis of GAX cycle.

  • COP is affected the most by the temperature measured at the generator inlet.

Abstract

Generator-absorber heat exchange (GAX) systems represent a promising alternative to substitute environmentally harmful refrigeration devices based on conventional vapor compression, as long as a proper analysis of thermal performance and the complex interactions of heat transfer that occur into GAX cycle is taken in consideration. In this research, a cooling process based on a GAX system that uses ammonia-water working fluid and a hybrid source (natural gas-solar) is studied to analyze the variables that affect the system’s thermal performance. The work’s novelty is the hybridization between artificial intelligence (AI) modeling and the global sensitivity analysis (GSA) developed with the PAWN method. Experimental data was obtained from a system with a cooling capacity of 10.5 kW (3 Ton), designed to work at heat source temperatures of 200 °C. The measured variables were the temperatures at generator, heat at evaporator, and working fluid volumetric flow. Three AI techniques (artificial neural networks, genetic programming, and support vector machines) were evaluated for modeling the thermodynamic cycle. Results obtained from the PAWN method applied to the artificial neural network, since it was the best AI model, indicates that the operational parameters with a greater impact in the system’s performance are the inlet temperature at the generator (30.7 %) and the heat measured at the evaporator for NH3 (27.4 %), for the first output COPNH3. For the second output COPH2O, the inlet temperature at the generator (32.5 %) and the and heat measured at the evaporator for H2O (26.7 %), have a greater impact for such output. The proposed IA-GSA methodology contributes to the development of operational decision-making related to instrumentation, operation performance, and corrective and/or preventive maintenance actions of GAX systems. The developed thermal performance model has potential for implementation in embedded systems (smart sensors) as a critical element in control and optimization strategies to improve the performance of these cycles.

Introduction

Humanity is at a stage where the need to find ways to reduce the emission of polluting gases is growing day by day, mainly in processes associated with the commercial [1], [2] and industrial [3], [4] sectors. As each country is responsible for its own emissions, each commercial and industrial sector is linked to a certain amount of polluting gases. It is estimated that worldwide, 15 % of the energy consumed is destined for the cooling of spaces at different scales [5]. In this context, refrigeration systems based on mechanical vapor compression (VCRS) currently represent one fifth of the energy consumed internationally. One of the main consequences of this percentage of energy consumption is the generation of large amounts of hydrochlorofluorocarbons (HCFCs), which deteriorate the ozone layer and contribute 10 % to the increase in global temperature [6], [7].

Since absorption systems consume less energy than VCRS systems and are capable of working with refrigerants that emit a lower amount of polluting gases into the atmosphere, they are considered as an attractive alternative to reduce the environmental impact associated with cooling systems. Additionally, different types of technologies based on renewable energies can be coupled to absorption systems, providing part of the energy demanded by the generator and thus contributing to the decarbonization of refrigeration systems [8], [9], [10].

Within the options of absorption refrigeration systems, generator-absorber heat exchange (GAX) cycles are considered one of the most prominent options currently on the market, due to their low cost and complexity, compared to other technologies such as single and multiple-effect absorption cycles [11]. The implementation of technologies that uses renewable energy sources on GAX cycles have made it possible to improve the coefficient of performance (COP) of the hybrid system compared to the traditional cycle [12], consequently reducing the amount of polluting particles emitted into the environment [13].

The main characteristic of the GAX system is the temperature difference presented between the absorber and the generator, which is achieved by a concentration gradient between the weak and a strong solution. The idea consists of obtaining high temperatures in the operational sections of the absorber. This eventually leads to a reduction of the external heat needed by the generator and improve the system’s efficiency, measured by the coefficient of performance (COP).

Different variations have been made on the GAX cycle in order to improve the system’s COP. For the simple GAX cycle, studied in this work, the absorber and generator are maintained at pressure and temperature conditions such that a temperature difference is created, which is directed to the generator, increasing the COP. In [14], [15], the competitive operating conditions associated with a hybrid GAX heat pump system were obtained, highlighting the low ambient temperature (7°C), abundant solar radiation (800W/m2) and operating temperature (180°C).

In [16], a two-stage GAX cycle with an absorption-reabsorption heat pump is developed employing an internal heat recovery system for sites where the ambient temperature is low for most time of the year. The results show that the maximum COP reached is 1,367, supplying heat at a temperature of 46.1°C; both quantities obtained at ambient temperatures above -15.6°C. In most cases, heat was obtained with a temperature above 42°C, ideal for heating buildings.

In addition to the experimental systems previously exposed, different computational studies have been developed to model and optimize the performance of hybrid GAX systems [17], [18]. In [17], as a modeling alternative for the complex thermodynamic performance equations, a surrogate model using artificial neural networks (ANN) is implemented, in order to obtain a relationship for the COP and the circulation ratio (f) as a function of the temperatures of the generator, evaporator, condenser and absorber, as well as the concentration of the weak and strong solutions. Using the mean squared error statistical coefficient (R2) as an evaluation instrument for the ANN model, values ​​of 0.9873 and 0.9996 are obtained for the COP and f, respectively, indicating a high reliability relationship of the model.

In [18] is presented a technical and financial evaluation of a cooling system with solar absorption. The impact that the size of the absorption subsystem and volume of the storage tank have on the cycle is analyzed with global optimization using genetic algorithms. The results indicate that the optimal configuration for the hybrid system consists of an absorption subsystem with a capacity of 0.04–0.06 kW/m2 and a storage tank with a volume of 40–60 L/m2. This optimized system has a payback period of 4.96 years and allows for annual energy savings of up to 68.8 kWh/m2.

Due to the fact that thermal refrigeration processes can present complex relationships between the study variables, the use of advanced computing techniques is often required to represent them. In this context, artificial intelligence models have been used to solve problems associated with different thermal systems, such as the exergetic analysis of VCRS [19], the study of operational parameters such as the operating temperature and the working fluid [20] and prediction of heating, ventilation and air conditioning (HVAC) performance [21], among others. According to a trend analysis, one of the main characteristics studied in refrigeration systems through artificial intelligence techniques is the COP [7].

Artificial intelligence models are suitable for studying the multiple stages that make up the GAX cycle, modeling the interaction between operational variables and system performance. In addition, the interest of carrying out the global sensitivity analysis lies in the possibility of determining the importance associated with each independent variable in relation to the output parameters studied. The use of these models is justified considering the complexity of the absorption, generation, evaporation and condensation sections that make up the GAX cycle as well as and any other refrigeration system, where multiple independent variables are measured in order to determine their association through a data-driven model [7].

As a summary, Table 1 shows the articles analyzed in this section with their most relevant characteristics, in which experimental or computational studies on GAX cycles are developed. The choice of these articles is based on the relevance that it has for the present work the obtaining of output variables for the GAX cycle, either experimentally or computationally, as well as the operational parameters measured in the different refrigeration systems. As can be seen in Table 1, most of the recent investigations of the GAX cycle presents a computational approach. Within the objectives set by each author, it is observed that the evaluation of the operational parameters and how they affect the system’s performance are the most prominent studies. Also, the combination of NH3/H2O has been one of the most studied working fluids for the GAX cycles.

However, some limitations have been found in the previous research reported in Table 1. In the experimental type of studies, they only establish correlations between a few operating parameters, such as the driving temperature, low-grade heat source temperature, supply water temperature, with outputs like the system’s performance and solar fraction [14], [22]. As for the computational studies, they mostly use methodologies created to optimize or improve certain performance parameters, without considering a previous analysis of variables, which can help to reduce the complexity of the studied problem [15], [23]. Sensitivity analysis applied to a data-based refrigeration model is a major gap of knowledge found in the literature review. This hybrid methodology can help to improve any optimization processes, by choosing only the variables with a higher relative influence regarding the studied output.

In the literature review, no works have been found that use Artificial Intelligence models (AI) to perform the computational representation of a GAX cycle, to later carry out a Global Sensitivity Analysis (GSA) of the operational parameters of the system. In this context, the novelty of this work consists in:

  • Creating a hybrid computational methodology, based on artificial intelligence techniques, capable of being applied to GAX systems.

  • Contributing to a better understanding of the coefficient of performance (COP) of a GAX system.

  • Integrating the AI models to a Global Sensitivity Analysis (GSA) through the PAWN method, with the objective of determine which are the operational variables of a GAX cycle that have a greater impact on the coefficient of performance (COP).

By considering such methodology, the results of this work are intended to be an auxiliary criterion for decision making, seeking to establish the main operational parameters that must be modified, in order to achieve an increase in the efficiency of the refrigeration system.

The manuscript is organized as follows: Section 2 describes the experimental system, the design parameters and the different stages that the working fluid goes through in the cycle. Section 3 presents the methodology to be followed, divided into three subsections: Processing and analysis of the experimental database, creation of artificial intelligence models and global sensitivity analysis using the PAWN method. Section 4 is integrated by the discussion of the results, dedicating a subsection for each stage of the methodology. Finally, Section 5 presents the conclusions of the study.

Section snippets

System description

The experimental facility to obtain the 10.5 kW GAX system’s performance data is described in Fig. 1. It is integrated by a generator, condenser, evaporator, and absorber that operates with pair ammonia-water (NH3-H2O) mixture as working fluid; with the elements designed to be air-cooled. The generator and the absorber are divided into smaller sections in order to increase the amount of heat received by the working fluid and therefore reducing the external heat demanded (generator) and the

Methodology

Fig. 4 illustrates the numerical approach applied in this study, divided in three parts. In the first stage, a working database (summarized in Table 3) is created based on several tests performed on the experimental system. The numerical information is correlated to evaluate the strength of the relationship between each pair of variables. In this stage, the variables are also normalized to align them in the same range and facilitate computational models training. In the second stage, the

Correlation analysis

Before using the independent variables as inputs for the creation of the artificial intelligence models, a correlation analysis was carried out for the GAX system. As it can be seen in Fig. 7, the temperature at the inlet of the generator (Tg,in) has a strong inverse correlation (r=-0.76) with the flow of the working fluid (Fv); this means that, when the flow slowed down, the temperature measured at the entrance of the generator tend to increased. Also, the highest direct correlation (r=0.66)

Conclusions and future work

In this study, a global sensitivity analysis was applied to the best prediction model of the coefficient of performance associated with a GAX experimental system, obtained from the comparison between artificial intelligence techniques ANN, SVM and GP. The input variables of such models where the operational parameters of the system: Temperature inlet (Tg,in) and outlet (Tg,out) at the generator, volumetric flow of the working fluid (Fv), and the heat measured at the evaporator for NH3 (Qev,NH3)

Declaration of Competing Interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Victor Cardoso Fernández reports financial support was provided by National Council on Science and Technology.

Acknowledgement

The author V. Cardoso-Fernández is grateful to the financial support of CONACYT to pursue a postgraduate degree in Universidad Autónoma de Yucatán under the following quantitative details CVU: 1006703, scholarship number: 808252. The authors are grateful with Dr. Roberto Best y Brown, who led the GAX system construction and testing project, and with the Renewable Energy Institute of UNAM, where these tests were carried out.

References (60)

  • E. Dai et al.

    Experimental investigation on a GAX based absorption heat pump driven by hybrid liquefied petroleum gas and solar energy

    Sol. Energy

    (2018)
  • E. Dai et al.

    Theoretical and experimental investigation on a GAX-Based NH3-H2O absorption heat pump driven by parabolic trough solar collector

    Sol. Energy

    (2020)
  • T. Jia et al.

    Working domains of a novel solar-assisted GAX-based two-stage absorption-resorption heat pump with multiple internal heat recovery for space heating

    Energy Convers. Manage.

    (2020)
  • A. Şencan

    Performance of ammonia-water refrigeration systems using artificial neural networks

    Renewable Energy

    (2007)
  • Y. Xu et al.

    Assessment and optimization of solar absorption-subcooled compression hybrid cooling system for cold storage

    Appl. Therm. Eng.

    (2020)
  • J.U. Ahamed et al.

    A review on exergy analysis of vapor compression refrigeration system

    Renew. Sustain. Energy Rev.

    (2011)
  • S. Anand et al.

    Simulation studies of refrigeration cycles: A review

    Renew. Sustain. Energy Rev.

    (2013)
  • J. Wang et al.

    Experimental investigation on NH3–H2O generator-absorber heat exchange (GAX) absorption heat pump

    Energy

    (2019)
  • K. Mohammadi et al.

    Thermo-economic analysis of a novel hybrid multigeneration system based on an integrated triple effect refrigeration system for production of power and refrigeration

    J. Cleaner Prod.

    (2019)
  • C. Liu et al.

    Working domains of a hybrid absorption-compression heat pump for industrial applications

    Energy Convers. Manage.

    (2019)
  • W. Wu et al.

    A novel hybrid-energy heat pump with refrigerant injection: Performance characterization and injection optimization

    Energy Convers. Manage.

    (2020)
  • S.M.S. Mahmoudi et al.

    A novel combination of absorption heat transformer and refrigeration for cogenerating cooling and distilled water: Thermoeconomic optimization

    Renew. Energy

    (2022)
  • Y. Ai et al.

    Numerical analysis of the influence of molten pool instability on the weld formation during the high speed fiber laser welding

    Int. J. Heat Mass Transf.

    (2020)
  • Y. Ai et al.

    The investigation of molten pool dynamic behaviors during the “∞” shaped oscillating laser welding of aluminum alloy

    Int. J. Therm. Sci.

    (2022)
  • R. Karimi et al.

    Back propagation artificial neural network and central composite design modeling of operational parameter impact for sunset yellow and azur (II) adsorption onto MWCNT and MWCNT-Pd-NPs: Isotherm and kinetic study

    Chemometr. Intell. Lab. Syst.

    (2016)
  • A. Mossalam et al.

    Using artificial neural networks (ANN) in projects monitoring dashboards’ formulation

    HBRC J.

    (2018)
  • A. Khosravi et al.

    An artificial intelligence approach for thermodynamic modeling of geothermal based-organic Rankine cycle equipped with solar system

    Geothermics

    (2019)
  • A.H. Elsheikh et al.

    Modeling of solar energy systems using artificial neural network: A comprehensive review

    Sol. Energy

    (2019)
  • A. Zendehboudi et al.

    Application of support vector machine models for forecasting solar and wind energy resources: A review

    J. Cleaner Prod.

    (2018)
  • O. Kisi et al.

    A survey of water level fluctuation predicting in Urmia Lake using support vector machine with firefly algorithm

    Appl. Math. Comput.

    (2015)
  • Cited by (9)

    View all citing articles on Scopus
    View full text