Application of genetic algorithms in design and optimisation of multi‐stream plate–fin heat exchangers

This paper introduces two new approaches in the thermo‐hydraulic design of multi‐stream heat exchangers (MSHEs). In both approaches, geometrical aspects of the MSHE (e.g. exchanger dimensions, fin type, etc.) are optimised with a genetic algorithm (GA) using the Total Annual Cost (TAC) as an objective function. The first approach is capable of utilising the maximum allowable stream pressure drops and can result in minimal surface area requirements. In the second approach, all of the pressure drop values are considered as design variables and are therefore subject to optimisation. These approaches have been applied to two case studies taken from literature, and the results are compared to those arising from a currently used design method. In the first case study, application of the new approaches resulted in a smaller TAC than the current approach by 5.77% and 31.86%, respectively, and improvement in the second case was estimated to be 5% and 21.46%, correspondingly. The effect of different fins on an MSHE's TAC is discussed through application of GAs to the current approach. It is shown that correct selection of fin types reduces the TAC of the two case studies by 21.75% and 8.7%, respectively. © 2012 Canadian Society for Chemical Engineering     

Fatigue life estimation after crack repair in 6005 A-T6 aluminium alloy using the cold expansion hole technique

In this paper, the hole drilling (HD) and the cold expansion (CE) processes, which were used as a technique for crack repair, were investigated in order to estimate the beneficial effects on fatigue crack initiation (FCI). The FCI life is defined as the number of cycles to initiate a new crack of 0.2 mm on the surface of the specimen. Three hole radii and three degrees of cold expansion (DCE%) values were tested after a crack propagation period. Crack retardation after the CE process was observed. This phenomenon is due to two mechanisms: retardation owing to both geometric and mechanical effects, which is produced by the stress concentration at the drilled hole, and the large strain‐induced compressive residual stresses around the hole. In this report, the influence of the loading conditions was studied. For high values of the stress intensity factor range ΔK_ρ around the hole (based on the pseudo crack length a + ρ), the number of cycles corresponding to crack initiation N_i is low. At the edge of the hole, the maximum stress range can be approximated by the following formula: Δσ_max = 2ΔK_ρ /√πρ , where ρ is the hole radius and ΔK_ρ is the related stress intensity factor range.The FCI life extension, defined by the number of cycles corresponding to crack re‐initiation N_i , is related to the relative maximum stress range ratio R_σ = [(Δσ_max )/(Δσ_max )_th ] where (Δσ_max )_th is the value of the threshold maximum stress range obtained when N_i = 2 × 10⁶ cycles. The relationship between N_i and R_σ may be written as a power function.     

A shelf-space-dependent wholesale price when manufacturer and retailer brands compete

We propose a game-theoretic model involving the manufacturer of a national brand and a retailer selling her private label along with the national brand. The retailer can use either a differentiation strategy or an imitation strategy for offering her store brand. We consider two cases: the benchmark case, where both players have symmetric information and play a Nash game, and the incentive case, where the national brand’s manufacturer, acting as a leader, offers an incentive to the retailer in order to benefit from a larger proportion of the shelf space, which ultimately increases her own profit. By comparing both situations, we attempt to derive the conditions under which it is profitable for the manufacturer to implement such an incentive strategy and investigate if the results are idiosyncratic to the PL concept. These conditions are fourfold, and include the private label’s image, the price competition between the national brand and the private label, the transfer price level and the shelf-space allocated to the national brand in the benchmark case.     

Analysis of the distribution of thermal residual stresses in bonded composite repair of metallic aircraft structures

In this study, the distribution of the thermal residual stresses due to the adhesive curing in bonded composite repair is analysed using the finite element method. The computation of these stresses comprises all components of the structures: cracked plate, composite patch and adhesive layer. In addition, the influence of these residual stresses on the repair performance is highlighted by analysing their effect on the stress intensity factor at the crack tip. The obtained results show that the normal thermal stresses in the plate and the patch are important and the shear stresses are less significant. The level of the adhesive thermal stresses is relatively high. The presence of the thermal stresses increases the stress intensity factor at the crack tip, what reduce the repair performance.     

Determination of adsorption operating conditions in dynamic mode on basis of batch study: Application for Dimethylphthalate elimination on activated carbon prepared from Arundo donax

This study aims to determine the operating conditions to implement the Dimethylphthalate removal using an activated carbon prepared from Arundo donax, carbonized at 358?°C during 13?min. To achieve this objective, the study is conducted in batch and dynamic mode. Several kinetic models are applied, namely pseudo-first order, pseudo-second order, intragranular, and Bangham models. The pseudo-second-order model fits the data perfectly, the estimated regression coefficients >0.999. The intragranular diffusion takes place in two stages. The two-parameter (Langmuir, Freundlich, Temkin, and Dubinin-Radushkevich) and three-parameter models (Redlich-Peterson, Sips, and Toth) are applied to model the equilibrium isotherms. The nonlinear regression methodology based on the error functions (hybrid fractional error function, Marquardt’s percent standard deviation, average relative error, Sum of the absolute errors) is applied. The HYBRID fits properly the data showing that the Temkin model gives the best fitting (R²_adj = 0.992), and the Dubinin-Radushkevich model the worst (R²_adj = 0.793). The thermodynamic study shows that the adsorption occurs according to a physical process. The DMP desorption is more effective with a 5% NaOH solution. In dynamic mode, the runs are conducted in fixed bed column. The effect of the bed height, the DMP initial concentration, and the flow rate on the breakthrough curves is investigated, then these breakthrough curves are modeled using the Thomas and Bed Depth Service Time models. The regeneration of the exhausted Arundo donax activated carbon is performed in a column, after 5 cycles, the breakthrough time decreases from 65.3?to 26.8?h, however, the exhaustion time varies less rapidly.     

Adaptive neural network control based on neural observer for quadrotor unmanned aerial vehicle

This paper proposes an adaptive neural network control with neural state’s observer for quadrotor. The adaptive approach is used to solve the dynamics uncertainty problem of the controller. To perform the control, a Single Hidden Layer Neural Network (SHLNN) is used. Based on the structure of Sliding Mode Observer (SMO), a new neural observer is proposed to estimate the states. The aim of this work is to propose an observer insensitive to the measurement noise. The stability proof of global system is made by Lyapunov direct method. The adaptation laws of both artificial neural networks (ANNs) are derived from Lyapunov theory. The proposed controller is validated by simulation on the quadrotor under measurement noise conditions. A comparative study with SMO is made to highlight the performances of the proposed neural observer.     

Robust regression fusion of GMM-UBM and GMM-SVM normalized scores using G729 bit-stream for speaker recognition over IP

A novel approach, based on robust regression with normalized score fusion (namely Normalized Scores following Robust Regression Fusion: NSRRF), is proposed for enhancement of speaker recognition over IP networks, which can be used both in Network Speaker Recognition (NSR) and Distributed Speaker Recognition (DSR) systems. In this framework, it is basically assumed that the speech must be encoded by G729 coder in client side, and then, transmitted at a server side, where the ASR systems are located. The Universal Background Gaussian Mixture Model (GMM-UBM) and Gaussian Supervector (GMM-SVM) with normalized scores are used for speaker recognition. In this work, Mel Frequency Cepstral Coefficient (MFCC) and Linear Prediction Cepstral Coefficient (LPCC), both of these features are derived from Line Spectral Pairs (LSP) extracted from G729 bit-stream over IP, constitute the features vectors. Experimental results, conducted with the LIA SpkDet system based on the ALIZE platform3 using ARADIGITS database, have shown in first that the proposed method using features extracted directly from G729 bit-stream reduces significantly the error rate and outperforms the baseline system in ASR over IP based on the resynthesized (reconstructed) speech obtained from the G729 decoder. In addition, the obtained results show that the proposed approach, based on scores normalization following robust regression fusion technique, achieves the best result and outperform the conventional ASR over IP network.     

Channeling in solid Xe, Ar and Ne direct dark matter detectors

The channeling of the ion recoiling after a collision with a WIMP changes the ionization signal in direct detection experiments, producing a larger scintillation or ionization signal than otherwise expected. We give estimates of the fraction of channeled recoiling ions in solid Xe, Ar and Ne crystals using analytic models produced since the 1960s and 1970s to describe channeling and blocking effects.