Assessing response surface methodology for modelling air distribution in an experimental pig room to improve air inlet design based on computational fluid dynamics

doi:10.1016/j.compag.2017.08.009

Computers and Electronics in Agriculture

Volume 141, September 2017, Pages 292-301

https://doi.org/10.1016/j.compag.2017.08.009 Get rights and content

Highlights

•
RSM was evaluated for modelling of air distribution inside a pig room.
•
Air speeds in AOZ were modelled as functions of parameters of inlet configurations.
•
Ventilation rate and inlet angle had significant effects on air speed.

Abstract

The thermal comfort of pigs is strongly correlated with the air motion around the pigs. In a pig production building, it is the ventilation system that influences the indoor air distribution significantly. In the ventilation system design, air inlet is considered to be important on the air motion in room. In this study, the Response Surface Methodology (RSM) was evaluated to develop the prediction model of the air speed in animal occupied zone (AOZ). Three dimensional numerical simulations for an experimental pig room were conducted to estimate the air speeds within the AOZ with the inlet supplying air in different angles and airflow rates on three installation heights of inlets. The results showed the RSM was capable to model the air speed in AOZ. Based on sensitivity analysis, the initial air speed and angle significantly influenced the air speed at the AOZ level, while the installation height was a less significant parameter. The RSM models could be used for design and control of an optimal ventilation system to achieve a better environment for pigs.

Introduction

Increasing heat removal is an efficient way to reduce the heat stress and provide thermal comfort for the animals during hot seasons. Air velocity and distribution in the AOZ in livestock buildings has significant impact on the heat removal and consequently animal performance (van Wagenberg and de Leeuw, 2003). There are a number of studies showing that the heat removal, especially convective heat removal from animals, was highly correlated to the air speed in the buildings and higher air velocity will lead to higher convective heat transfer (Gebremedhin, 1987, Li et al., 2016b, Mitchell, 1976, Mitchell, 1985, Monteith and Unsworth, 2013, Wathes and Clark, 1981). In farm animal housing, the air distribution is highly linked with ventilation design. Therefore, it is essential to generate an effective distribution of fresh air within animal occupied space with desired air speed, to enable good thermal comfort for the animals during summer.

The inlet configurations are essential for control of the air motion and distribution in the building, and strongly affect the air speed in AOZ (Bjerg et al., 2002, Zhang et al., 2002). To reduce heat stress of animals, the inlet with jet supplying the air directly to the AOZ has been proved significant effect on the temperature distribution (Bjerg and Zhang, 2013). Additionally, the downward jet has also shown better performance on increasing the convective heat transfer from animal compared with conventional jet inlet with upward or horizontal initial air jet direction at the same flow rate from numerical simulation (Li et al., 2016a). However, further research on the relationship between air speed distribution in AOZ and the inlet parameters is still necessary for the optimal inlet design and control.

It is difficult to measure the real air speed in AOZ directly. Most of the studies in testing the ventilation system use empty rooms to make the measurement of air speeds in AOZ more easily (Bjerg et al., 2002, Randall, 1980, Zong et al., 2015). But the results of those studies can be different from the real condition, since the block effect of pigs are not considered. To measure the air speed in AOZ with pigs involved, van Wagenberg and de Leeuw (2003) use cages to protect the ultrasonic anemometers and place the anemometers in AOZ to measure the air speed. Although the air speed can be measured this way, the blocking from cage may still have effects on the air speed or turbulence intensity. With the development of the computational power of computers, research activity combining with CFD simulation on ventilation and indoor climate of the animal buildings is increasing, due to the lower cost on investigation setups and easier control of study parameters compared to conventional experimental method. Numbers of studies have been conducted on the ventilation study inside the livestock building using CFD method (Blanes-Vidal et al., 2008, Bustamante et al., 2015, Kwon et al., 2015, Lee et al., 2013, Mistriotis et al., 1997, Norton et al., 2009, Norton et al., 2010a, Rong et al., 2015, Wu et al., 2012). Animal models in real geometry are also used in some of CFD studies for environment parameters inside animal buildings to make the simulation results more reliable (Bjerg et al., 2008, Gebremedhin and Wu, 2005, Li et al., 2016b, Seo et al., 2012). Therefore, CFD method was adopted in this study to generate the air speed data in different configurations of the inlet.

The relationship between the air speed in AOZ and the selected parameters of inlet configurations is important for both the inlet configuration study and the inlet control model development. To identify the relationship, statistical data-based models, i.e. the meta-model method, can be used (Wang and Shan, 2007). One of the most well-established and easy-to-use meta-modelling technique is the Response Surface Methodology (RSM) (Simpson et al., 2001). With a proper experimental design, RSM can generate high precision prediction model with reduced number of experiments compared with full factory design. RSM has been used in many research aspects in the field of building ventilation study. For instance, the ventilation rate in the natural ventilated building has been modelled based on the parameters of air speed, air direction, and opening size with RSM (Shen et al., 2012, Shen et al., 2013a, Shen et al., 2013b). Based on RSM, the indoor environment was also studied with different ventilation configurations (Ng et al., 2008, Norton et al., 2010b). All the studies showed RSM has strong potential to build the model in the ventilation study. Concerning the inlet design, the feasibility of RSM method is worthwhile to be investigated.

Thus, the objective of this study is to assess the feasibility of RSM method in developing the model for the average air speed inside the animal occupied zone (AOZ) as a function of the parameters of inlet configurations. The established RSM model will be verified based on simulation data in other setup. Finally, sensitivity analyses will be conducted to identify the key parameter in the developed RSM modelling for the inlet design.

Section snippets

Response surface methodology design

RSM is a combination of mathematical and statistical techniques for formulating the regression model between the selected variables and targeted responses (Anderson and Whitcomb, 2005). Normally, the RSM model can be solved by following four stages, i.e., design of experiment, conduction of the experiments, building the response surface model, and solution optimization.

The RSM model design is highly dependent on the parameters selected as model variables. Before the experiment design, to

Validation of the CFD model

The simulation results agreed with the experimental results well (Fig. 3). It indicated that the CFD model is applicable in the simulation of the air speed which is the main purpose in the study. Therefore, the CFD approach was adopted for the AOZ air speed model development in the following parts with pig models and different inlet designs.

Development of RSM model

Table 3 shows the simulation results of different cases following the experimental design. In general, the cases with downward inlets in 22.5° or 45° showed

Conclusion

In the current work, the RSM has been proven to be a powerful technique in analysing air speeds distribution with respect to various combinations of three design variables. Based on the least square method, first and second order models were used to fit on the simulation generated data in accordance with the Box-Behnken method. Based on the criteria of achieving the highest adjusted R² and predicted R², the RSM models in predicting average air speed were developed. Verified with the data

Acknowledgements

Thanks to the support from research grant of Innovation Fund Denmark/Advanced Technology Foundation, New hot climate ventilation system for poultry and pigs (j.nr. 17-2013-3). And the thanks extend to China Scholarship Council (CSC) for the scholarship (201306350135) of Hao Li Ph.D. study in Denmark.

Reference (39)

B. Bjerg et al.
Modeling of air inlets in CFD prediction of airflow in ventilated animal houses
Comput. Electron. Agric.
(2002)
V. Blanes-Vidal et al.
Application of computational fluid dynamics to the prediction of airflow in a mechanically ventilated commercial poultry building
Biosys. Eng.
(2008)
K. Gebremedhin et al.
Simulation of flow field of a ventilated and occupied animal space with different inlet and outlet conditions
J. Therm. Biol.
(2005)
K. Gebremedhin
Effect of animal orientation with respect to wind direction on convective heat-loss
Agric. For. Meteorol.
(1987)
K. Kwon et al.
Computational fluid dynamics analysis of the thermal distribution of animal occupied zones using the jet-drop-distance concept in a mechanically ventilated broiler house
Biosys. Eng.
(2015)
I. Lee et al.
The past, present and future of CFD for agro-environmental applications
Comput. Electron. Agric.
(2013)
H. Li et al.
Study on convective heat transfer from pig models by CFD in a virtual wind tunnel
Comput. Electron. Agric.
(2016)
H. Li et al.
Reliability of turbulence models and mesh types for CFD simulations of a mechanically ventilated pig house containing animals
Biosys. Eng.
(2017)
J. Mitchell
Heat-transfer from spheres and other animal forms
Biophys. J.
(1976)
M. Mitchell
Measurement of forced convective heat-transfer in birds - a wind-tunnel calorimeter
J. Therm. Biol.
(1985)

K. Ng et al.

Response surface models for CFD predictions of air diffusion performance index in a displacement ventilated office

Energy Build.

(2008)

T. Norton et al.

Assessing the ventilation effectiveness of naturally ventilated livestock buildings under wind dominated conditions using computational fluid dynamics

Biosys. Eng.

(2009)

T. Norton et al.

Improving the representation of thermal boundary conditions of livestock during CFD modelling of the indoor environment

Comput. Electron. Agric.

(2010)

T. Norton et al.

Optimising the ventilation configuration of naturally ventilated livestock buildings for improved indoor environmental homogeneity

Build. Environ.

(2010)

J. Randall

Selection of piggery ventilation systems and penning layouts based on the cooling effects of air speed and temperature

J. Agric. Eng. Res.

(1980)

L. Rong et al.

The effect of wind speed and direction and surrounding maize on hybrid ventilation in a dairy cow building in Denmark

Energy Build.

(2015)

L. Rong et al.

Summary of best guidelines and validation of CFD modeling in livestock buildings to ensure prediction quality

Comput. Electron. Agric.

(2016)

I. Seo et al.

Modelling of internal environmental conditions in a full-scale commercial pig house containing animals

Biosys. Eng.

(2012)

X. Shen et al.

Investigation of response surface methodology for modelling ventilation rate of a naturally ventilated building

Build. Environ.

(2012)

Cited by (19)

CFD investigation on a novel pen partition-attached jet air supply for mitigating heat stress among lactating sows
2024, Computers and Electronics in Agriculture
To effectively mitigate heat stress among lactating sows, this study proposed a novel pen partition-attached jet air supply (PJAS) and compared its performance with commercial ceiling jet air supply (CJAS) by using computational fluid dynamics (CFD). Each PJAS unit, comprising a circular distribution duct with 5 slot orifices at each side, was installed at the pen partition between two adjacent pens to direct air towards lactating sows to enhance convective heat removal. The cooling performances of air supplier were primarily evaluated by assessing the convective heat flux from lying sows. The effect of jet angles induced by PJAS and CJAS on convective heat flux were investigated as well as the consequence of reducing the inlet dimension of CJAS. The results revealed that reducing the inlet dimension of CJAS doubled the speed of supplied air. However, it also limited the amount of air impinging on sow due to the narrowing of the inlet air jet. Consequently, the combined effect only marginally increased the convective heat flux when the inlet area of CJAS was reduced. By placing air supplier closer to lactating sows, the air jet induced by PJAS impinged on sows’ bodies directly and therefore significantly increased the convective heat removal. Compared to CJAS, PJAS required approximately 45% less ventilation rate to induce the same convective heat flux. When the ventilation rate was identical, PJAS could increase the convective heat flux by up to 27.6%. These results highlight PJAS as a promising energy-efficient localized cooling device for mitigating heat stress among lactating sows. To determine the optimal design and control strategy of PJAS based on industrial requirements, further research is still needed.
Simulation and experiment of air-duct heating equipment using electricity in a sheep barn using computational fluid dynamics
2024, Computers and Electronics in Agriculture
The temperature inside the sheep barn has a significant effect on sheep growth; therefore, providing a suitable thermal environment for the barn is essential. Considering the uneven heating problem of existing heating equipment in livestock barns, we designed an air-duct heating equipment using electricity to ensure a comfortable temperature for sheep with a suitable airflow rate. A computational fluid dynamics (CFD) method was used to simulate the temperature and velocity fields inside the sheep barn, check the heating effectiveness, and ensure uniformity. To determine the optimal opening direction and size of the heating equipment, simulation tests with various opening directions and sizes were conducted using the flow velocity and uniformity of the flow field as the key factors. The CFD simulation results showed that for the investigated sheep barn, a downward opening with diameter of 60 mm was the most effective for heating. Finally, in winter, a heating test was conducted during the decreasing, stable, and increasing periods of outdoor temperatures to verify the heating effectiveness in the experimental sheep barn. During the decreasing and stable periods of outdoor temperature, the indoor temperatures increased by 2 and 5 ℃, respectively, in 70 min. Conversely in the increasing period of outdoor temperature, the temperature inside the barn increased by 11 ℃ and continued to rise. Our results showed that the expected heating effect of sheep barns was achieved, which can provide a reference for the environmental regulation of livestock barns.
Applicability evaluation of innovative simplified methods of slatted floor in pig houses – A CFD study
2024, Computers and Electronics in Agriculture
The slatted floor system is often used in pig housing. Computational Fluid Dynamics (CFD) is a valuable technique for determining the airflow and gas dispersion within the pit headspace through the slatted floor. However, there is a practical issue related to modeling the thousands of small slot openings in real livestock buildings for CFD simulation. It is unpractical to simulate the slatted floor with geometry details due to the large mesh number and the limited computer capacity. In this study, several methods to simplify the model of the slatted floor were explored to reduce the computational cost. The study was based on CFD analysis supported by experimental validation. The slatted floor was modeled in four ways: direct geometry method (DGM), porous-region method (POM), and profiled-porous-region method (including PPOM-cell (PPOM-C) and PPOM-floor (PPOM-F)). To assess the feasibility of the simplification, the four modeling approaches were compared with the experimental data. The results showed that it’s feasible to model the slatted floor as POM or PPOM. The accuracy of POM was high, but it was greatly affected by the calculation of the resistance coefficient. PPOM-C has greatly reduced the amount of calculation, which could be used to simplify the large-scale pig house model whose flow fields and gas emissions below the slatted floor could be ignored. The simulation result of PPOM-F was the closest to DGM, however, due to the large number of meshes and complex setup, it was not suitable for practical simulation.
Estimation of ventilation rate in a piglet house considering ventilation system characteristics
2023, Biosystems Engineering
In mechanically ventilated piglet houses, ventilation is considered the most important aspect that can enhance the rearing environment. However, the airflow rate of fans can be less than the design rate due to pressure losses caused by operating conditions. To ensure the accurate estimation of the airflow rate, pressure loss factors should be evaluated since they have a strong influence on ventilation. In this study, pressure losses were evaluated through field experiments with various inlet types (side window, ceiling slot), outlet types (side and chimney exhaust fan), and operating conditions (fan speed, inlet opening size). A fan airflow rate formula accounting for the characteristics of ventilation systems was developed based on the fan performance reported by the manufacturer (designed performance) and orifice theory. Compared with in-situ and design fan performance, the in-situ airflow rate reduced by an average of 7–35% due to pressure losses. The pressure loss of the chimney exhaust fan was higher than that of the side exhaust fan due to the long duct and the position of the fan. The ventilation rate for the ceiling slot inlet was prone to overestimation due to a higher resistance curve coefficient (0.4) than that of the side window inlet. The developed ventilation rate formula was validated using 360 cases and it showed high accuracy (side exhaust fan $R^{2} > 0.9947$ , chimney exhaust fan $R^{2} > 0.9947$ ). Thus, the evaluated pressure loss factors and the developed ventilation rate formula can be applied for the design of ventilation systems in piglet houses.
CFD study on the impacts of geometric models of lying pigs on resistance coefficients for porous media modelling of the animal occupied zone
2022, Biosystems Engineering
Citation Excerpt :
Li, Rong, Zong, and Zhang (2017) and Li et al. (2018) omitted small parts in pig models used for a simulation study of convective heat transfer between pigs and its surroundings. The convective heat transfer coefficients were correlated with the air speeds, and then the prediction model of air speed in the animal occupied zone was established by using simplified pig models (Li, Rong, Zong, and Zhang, 2017). However, the simplified pig models with small parts omitted would still generate high grid numbers causing a large amount of calculation or poor mesh quality could appear due to the irregular shape of pigs in order to reduce the grid numbers.
Computational fluid dynamics (CFD) has been widely used for studying the indoor environment of livestock houses. However, the detailed geometrical models of pigs in commercial pig production houses induce extremely high grid numbers which require huge computational capacity. Therefore, simplifying the pigs in the occupied zone into porous media model is an alternative. To use the porous media model, the resistance coefficients need to be determined. In this study, the influence of different geometric models of lying pigs under various body mass on resistance coefficients was investigated. Three simple geometric lying pig models (ellipsoid model, half ellipsoid model and half cylinder model) under different body masses (30 kg, 45 kg, 60 kg, 75 kg, 90 kg and 105 kg) were used to identify which type of simple geometric models could represent the simplified lying pig model by comparing the obtained resistance coefficients. Meanwhile, the relationship between the body mass and the resistance coefficients was also analysed. The results showed that the resistance coefficients were affected by different simple geometric models of pigs and the half ellipsoid model was the most suitable model to replace the simplified pig model. In addition, the pressure drop per unit length increased with the rise in body mass at the same air velocity and the increase was more significant at larger body mass with the same rise in body mass. The inertia resistance coefficients and body mass were highly correlated to each other. The resistance coefficients obtained in this study can be used in the indoor environment simulation of fattening pig buildings.
Investigating the feasibility of using computational fluid dynamics based response surface methodology and neural network to model the performance of the individualised ventilation in sow houses
2022, Biosystems Engineering
Microclimatic conditions affect the thermal comfort and health of sows kept in confined spaces. Individualised ventilation strategies (IV) have been shown to be more effective in providing homogeneous cooling effects and fresh air across the highly sectioned sow houses than conventional ventilation strategies. However, using computational fluid dynamics (CFD) simulations to establish the performance model of IV could be computationally intensive because the performance of IV relies on a complex interplay between several influential parameters. To efficiently model the performance of IV, we assessed the feasibility of using response surface methodology (RSM) in conjunction with Box–Behnken design (RSM-BBD) and neural networks (NN) as surrogate methods for CFD to reduce computational costs without compromising the modelling accuracy. Three influential parameters of air inlets (supply airspeed, supply angle, and nozzle height) were involved in the modelling processes. Skin convective heat transfer coefficient and local air age were modelled as performance indicators. Modelling results suggested that both RSM-BBD and NN could be employed to model the convective heat transfer coefficient of sows at significantly reduced computational costs. Compared to NN, RSM-BBD was less effective at capturing the cryptic underlying relationship between the inputs and outputs and was not a feasible option to model the air age. However, RSM-BBD only required approximately 16% of the test cases of NN despite a compromised model accuracy. The selection of which the surrogate methods for CFD is preferred depends on whether or not accuracy is the priority.

View all citing articles on Scopus

View full text

Assessing response surface methodology for modelling air distribution in an experimental pig room to improve air inlet design based on computational fluid dynamics

Highlights

Abstract

Introduction

Section snippets

Response surface methodology design

Validation of the CFD model

Development of RSM model

Conclusion

Acknowledgements

Comput. Electron. Agric.

Biosys. Eng.

J. Therm. Biol.

Agric. For. Meteorol.

Biosys. Eng.

Comput. Electron. Agric.

Comput. Electron. Agric.

Biosys. Eng.

Biophys. J.

J. Therm. Biol.

Energy Build.

Biosys. Eng.

Comput. Electron. Agric.

Build. Environ.

J. Agric. Eng. Res.

Energy Build.

Comput. Electron. Agric.

Biosys. Eng.

Build. Environ.