Gusel et. GP models have analysis dimension lower deviation compared to regression models. It is evident from the literature review that almost nothing has been reported on the prediction of the effect of the parameters such as die radius, punch radius, blank holder force and friction coefficient on wrinkling in aluminium AA by RSM and ANN.
In the previous study by these authors the influence of these variables was investigated [12], so in this paper, the effective range of each parameter is considered in the design of the tests. For each factor five levels are defined as summarized in Table 2.
In this paper, the central composite design CCD is employed for training data. This method is rotatable and created easily from 2k factorials.
The CCD for four parameters includes 25 samples that consist of 16 corner points, 8 axial points and 1 centre point. Table 3 shows the data set in a coded form.
These data are used by the Abaqus software so as to specify WH. The dimensions of blank, square cup and hydraulic press are considered as specified in the mentioned study [12] and Al , whose properties are presented in Fig.
Levels of input factors coded FEM R 45 0. Keymanesh Artificial Neural Network Fig. The output values obtained from the 3D simulation of the deep drawing process with FEM are presented in Table 3. RSM, which creates polynomial models for the available data, is given in Eq.
In Fig. The model fitted to the data is given in Eq. As Fig. WH Error Trial no. The performance of a neural network depends mainly on its weights of connection.
In general, each neuron can be connected with all other neurons. For manufacturing processes where no satisfactory analytic model exists or a low order empirical polynomial model is inappropriate, neural networks offer a good alternative approach [15].
Recently, many different neural network models have been developed. They include Perceptron, Hopfield and Hamming Network. There are several transfer functions such as Hardlim, Purelin and Logsig function utilized in neural network models. The neural network that is employed is the multilayer perceptron that learned by using the Bayesian Regularization Algorithm. The function of the output layer, Hardlim, is taken into account. The learning factors of the training network are presented in Table 5.
Initialize all of the weights of the links to random values. Present the input-desired output patterns one by one, updating the weights each time using Eq.
The variation of the MSE during the training is shown in Fig. The MSE target is achieved in epochs and the training is terminated. For calculating the sensitivity of each parameter in the ANN, the specified parameter has maximum and minimum quantity and other parameters have average quantity.
A minimum normal stress must be RT as shown in fig2. Fig4: Stresses acting on the force During deep drawing, metal in various transmission zone. The radial or hoop stress. In the part. Tearing is caused by excessive finite element analysis the structure is tensile stresses leading to fracture of the modeled by the assemblage of small pieces sheet while excessive compressive of structure, These pieces with simple stresses during forming cause wrinkling.
The Wrinkling is a wavy condition obtained in word "finite" distinguishes these pieces the blank while drawing due to unbalanced from infinitesimal elements used in compressive forces set up in the flange calculus. In the finite element analysis area. This can be achieved by using a FEA , the variation of the field variable blank holder force just above wrinkling on the element is approximated by the threshold from the beginning of the deep- simple functions, such as polynomials.
The drawing process. The wrinkling actual variation on the element is almost phenomenon in draw die forming has certainly more complicated, so FEA been considered as a major obstacle in provides an approximate solution. The RT limiting the ability of the die maker to produce good parts. In spite of its practical value of field variable and perhaps also its first derivatives are defined as unknowns IJE importance in sheet metal forming applications, the wrinkling instability at the nodes.
Solving a structural problem phenomenon has received very little by FEA involves following steps : attention in the industry. Much of the Learning about the problem; Preparing activity in the last two decades has been mathematical models; Discretizing the directed toward analyzing tensile model; Having the computer do instabilities in sheet metal forming calculations; Checking results Generally processes.
Less attention has been given an iteration is required over these steps. Metal technique which evaluates the behavior of forming is one of such area where a lot of components, equipments and structures for trials are required to arrive at the die various loading conditions.
It is a design to produce defect free parts. There computerized method for predicting how a is a body of literature that demonstrates real object will react to forces by mesh of finite element method can be used as a IJERTV2IS www.
If the finite element method is used properly in regards to reliable input data as well as correct modeling technique and appropriate analysis approach, then the results can be reliable enough to predict the outcome of drawing processes and thus direct the design of tools[18] Fig. A forming limit diagram, also known as a forming limit curve, is used in sheet metal Literature Review forming for predicting forming behavior of Literature review is an assignment of sheet metal.
The diagram attempts to previous task done by some authors and provide a graphical description of material collection of information or data from failure tests. In order to determine whether RT research papers published in journals to a given region has failed, a mechanical test progress our task.
It is a way through is performed. The mechanical test is IJE which we can find new ideas, concept. By repeating the mechanical test to generate a range of stress states, the P V R Reddy et al. Sheet friction coefficient on the limit strain of metal is marked with small circles aluminum AA in deep drawing ,stretched over a punch, and deformation is process using a finite element method, observed in failure areas.
FLD shows forming limit diagram are plotted for boundary between safe and failure various blank holding forces, die corner zones[1]. T et al. It is also observed that drawing steel sheet of 0. It is the ratio of maximum blank interaction with the blank holding force.
The blanks of progressively on CRDQ steel sheet of 1 mm thickness increasing diameters Do are drawn into and their effect on wrinkles and wall cups using a punch of diameter dp. The RT thickness distribution is analyzed by using maximum diameter of the blank D0max HyperForm software ,their analysis reveals just before the first defect occurs is used to IJE that as the blank holding force increases find the LDR value.
Additionally, considering minimizing the part complexity and any asymmetry can also help. Incorporating a multi-step drawing process offers a variety of advantages in preventing wrinkling in deep-drawn parts. Designing the blank geometry to minimize excess material can reduce the potential for wrinkling. The sheet metal blank has an inherent grain structure, so the stresses can vary depending on the design of the die and the orientation of the grain.
Adjusting the grain in an asymmetrical design to minimize the compound of grain stresses and the general stresses of the deep draw process is something to take into consideration. Surface conditions of each component can be tailored to improve overall performance.
Lubricants reduce the friction between the blank and the punch and die cavity and can be liquid wet or films dry. Generally, they are applied to the blank before drawing. Today, dry films are gaining acceptance because they reduce the need for part washing after fabrication.
While lubricants can facilitate the metal flow into the die cavity, consider increasing the blank holding force to account for the reduced friction. In the past, trial and error and operator experience optimized part and die design. Today, computer aided design and finite element modeling are used to create part and die designs and to simulate the deep drawing process, significantly reducing the costs of tooling and labor in the design process.
Guides Share:. Select From Over , Industrial Suppliers. Receive Daily Industry Updates. Search Over 6 Million Products.
0コメント