The Prediction of CHF at Near Critical Pressures by High Order Neural Network (HONN)

2010 ◽  
Author(s):  
Dariush Rostamifard ◽  
Mehdi Fallahnezhad ◽  
Salman Zaferanlouei ◽  
Saeid Setayeshi ◽  
Mohammad Hassan Moradi
Keyword(s):  
Neural Network ◽  
Heat Transfer ◽  
Local Heat Transfer ◽  
Local Heat ◽  
High Order ◽  
Superior Performance ◽  
Ann Model ◽  
Feed Forward Neural Network ◽  
Predicted Values ◽  
High Order Neural Network

The critical heat flux (CHF) condition is characterized by a sharp reduction of the local heat transfer coefficient that results from the replacement of liquid by vapor adjacent to the heat transfer surface. We used High Order Neural Network (HONN) as a stronger open box intelligent unit than traditional black box neural networks to predict of CHF at near critical pressures. The process of training and testing this model is done using a set of available published filed data. The CHF values predicted by the HONN model are satisfying compared with the measured data. The predicted values were also compared with those predicted using Feed Forward Neural Network (FNN) and available empirical equations that have been suggested in the literature. It was found that the HONN model with Root Mean Square (RMS) errors of 3.23% in critical pressures conditions has superior performance in predicting the CHF than the best accurate prediction of the methods. Results show by having an HONN model of our nonlinear input-output mapping, there are many advantageous than ANN model including faster running for new data, lesser RMS error and better fitting properties.

Empirical correlation study of dryout heat transfer at high pressure using high order neural network and feed forward neural network

2010 ◽  
Vol 47 (4) ◽  
pp. 439-448 ◽  
Author(s):  
Dariush Rostamifard ◽  
Mehdi Fallahnezhad ◽  
Salman Zaferanlouei ◽  
Saeed Setayeshi ◽  
Mohammad Hassan Moradi
Keyword(s):  
Neural Network ◽  
Heat Transfer ◽  
High Pressure ◽  
Correlation Study ◽  
High Order ◽  
Empirical Correlation ◽  
Feed Forward Neural Network ◽  
Feed Forward ◽  
High Order Neural Network

Development of High Order LES Solver for Heat Transfer Applications Based on the Open Source OpenFOAM Framework

2015 ◽  
Author(s):  
Luca Mangani ◽  
David Roos Launchbury ◽  
Ernesto Casartelli ◽  
Giulio Romanelli
Keyword(s):  
Heat Transfer ◽  
Open Source ◽  
Gas Turbines ◽  
Incompressible Flows ◽  
Stokes Equations ◽  
Turbulence Models ◽  
Local Heat Transfer ◽  
Local Heat ◽  
High Order ◽  
Computational Time

The computation of heat transfer phenomena in gas turbines plays a key role in the continuous quest to increase performance and life of both component and machine. In order to assess different cooling approaches computational fluid dynamics (CFD) is a fundamental tool. Until now the task has often been carried out with RANS simulations, mainly due to the relatively short computational time. The clear drawback of this approach is in terms of accuracy, especially in those situations where averaged turbulence-structures are not able to capture the flow physics, thus under or overestimating the local heat transfer. The present work shows the development of a new explicit high-order incompressible solver for time-dependent flows based on the open source C++ Toolbox OpenFOAM framework. As such, the solver is enabled to compute the spatially filtered Navier-Stokes equations applied in large eddy simulations for incompressible flows. An overview of the development methods is provided, presenting numerical and algorithmic details. The solver is verified using the method of manufactured solutions, and a series of numerical experiments is performed to show third-order accuracy in time and low temporal error levels. Typical cooling devices in turbomachinery applications are then investigated, such as the flow over a turbulator geometry involving heated walls and a film cooling application. The performance of various sub-grid-scale models are tested, such as static Smagorinsky, dynamic Lagrangian, dynamic one-equation turbulence models, dynamic Smagorinsky, WALE and sigma-model. Good results were obtained in all cases with variations among the individual models.

Arrays of Impinging Jets with Spent Fluid Removal through Vent Holes on the Target Surface—Part 1: Average Heat Transfer

1980 ◽  
Vol 102 (4) ◽  
pp. 994-999 ◽  
Author(s):  
B. R. Hollworth ◽  
L. Dagan
Keyword(s):  
Heat Transfer ◽  
Local Heat Transfer ◽  
Target Surface ◽  
Local Heat ◽  
Impinging Jets ◽  
Superior Performance ◽  
Jet Flows ◽  
Target Plate ◽  
Average Heat ◽  
Fluid Removal

Measurements of average convective heat transfer are reported for square arrays of impinging air jets. The target plate on which the jets impinge is perforated so that spent air is withdrawn through the plate rather than at one or more edges of the array, as is usually the case in such investigations. Jet holes and vent holes had the same diameters, but the spacing of the jet holes was twice that of the vent holes. This information is especially relevent to the design of hybrid cooling configurations, in which a surface is cooled by the combined mechanisms of impingement and transpiration. Tests were conducted for both inline arrangements (with a vent hole opposite each jet orifice) and for staggered arrangements; and the latter always yielded higher average heat transfer. The degradation of performance of inline arrays was most pronounced when the clearance between the jet orifice plate and the target plate was small. Under these conditions, a significant portion of each jet flows directly out through the opposing vent without “scrubbing” the target surface. Arrays with staggered vent holes yield heat transfer rates consistently higher (sometimes by as much as 35 percent) than the same jet array with edge venting. The authors attribute the superior performance of the former geometry to high local heat transfer due to boundary layer suction in the vicinities of the vent holes.

scholarly journals CORRELATIONS AND NEURAL NETWORKS AS MODELS FOR THE LOCAL HEAT TRANSFER COEFFICIENT ON CONDENSATION FOR R1234YF IN Ø 0.96MM DUCTS

2021 ◽  
Vol 20 (1) ◽  
pp. 79
Author(s):  
R. P. Mendes ◽  
J. G. Pabon ◽  
D. L. F. Pottie ◽  
L. V. S. Martins ◽  
L. Machado
Keyword(s):  
Neural Network ◽  
Heat Transfer ◽  
Heat Transfer Coefficient ◽  
Transfer Coefficient ◽  
Saturation Temperature ◽  
Local Heat Transfer ◽  
Local Heat ◽  
Local Heat Transfer Coefficient ◽  
Internal Diameter ◽  
The European Union

Due to global warming considerations, the European Union has banned the use of refrigerants with a GWP greater than 150 in new passenger cars (air-conditioning systems) and 750 for fluids used in residential heat exchangers starting on January 1, 2017 (E. UNION, 2006). In this sense, the R1234yf was developed which consists of a hydrofluorolefin derived from alkenes and commercialized with the name of Opteon YF. Given the need for research related to the use of this fluid, this work has the objective of comparing the data of the local heat transfer coefficient in condensation extracted from the work of Del Col et al. (2010) for flow in a mini channel of 0.96 mm internal diameter, with mass flux of 200, 300, 400, 600, 800 and 1000 kg·(m²·s)-1 at saturation temperature of 40ºC with ten different correlations from literature as well as one neural network. It is verified that among the correlations analyzed the one which best suited the experimental data was presented by Cavallini and Zecchin (1974), with MRD, MARD, and Accuracy values equal to 5.42%, 7.81%, and 96.96%, respectively. The neural network used as a prediction model presents values of MRD, MARD, and Accuracy equals to 2.53%, 3.66%, and 100%, respectively

Sign in / Sign up

Export Citation Format

Share Document