MINIMIZING CONTACT STRESSES IN AN ELASTIC RING BY RESPONSE SURFACE OPTIMIZATION

Size: px
Start display at page:

Download "MINIMIZING CONTACT STRESSES IN AN ELASTIC RING BY RESPONSE SURFACE OPTIMIZATION"

Transcription

1 MINIMIZING CONTACT STRESSES IN AN ELASTIC RING BY RESPONSE SURFACE OPTIMIZATION Asim Rashid THESIS WORK 21 PRODUCT DEVELOPMENT AND MATERIALS ENGINEERING Postal Address: Visiting Address: Telephone: Box 126 Gjuterigatan Jönköping

2 MINIMIZING CONTACT STRESSES IN AN ELASTIC RING BY RESPONSE SURFACE OPTIMIZATION Asim Rashid This thesis work is performed at Jönköping Institute of Technology within the subject area Product Development and Materials Engineering. The work is a part of the master s degree. The author is responsible for the given opinions, conclusions and results. Supervisor: Niclas Strömberg Credit points: 3 ECTS credits Date: Archive number: Postal Address: Visiting Address: Telephone: Box 126 Gjuterigatan Jönköping

3 Abstract Abstract In this thesis an optimization routine is implemented to minimize the contact pressure. The routine is based on the successive response surface methodology and is applied to the problem of minimizing contact pressure between a snap ring and its housing. Design of Experiment techniques are used to decide the combination of design variables. A Python script is developed that performs the computer simulations in Abaqus. Maximum contact pressure information generated by Abaqus is imported in Matlab. Combination of design variables and results from Abaqus are used to build second order response surfaces by using least squares method. Then quadratic programming is used to determine the optimum of the response surface. This process is repeated according to the rules of successive response surface methodology until convergence is achieved. Two different geometries, polynomial and super elliptical curves are considered as the prospective candidates for improved design of snap ring. The results show that a polynomial curve is more suitable than a super elliptical curve. A polynomial curve gives a lower value of maximum contact pressure and better distribution over the whole length as compared to the super elliptical design. The optimization has been compared with a commercial software and shows promising results. Key Words Successive Response Surface methodology, SRSM, Contact Pressure, Contact Stress, Snap Ring, Optimization

4 Table of Contents Table of Contents 1 Introduction BACKGROUND PURPOSE AND AIMS DELIMITS OUTLINE Theoretical background RESPONSE SURFACE METHODOLOGY Performing Experiments Building Response Surface Optimization of Response Surface SUCCESSIVE RESPONSE SURFACE METHODOLOGY CONTACT MECHANICS Contact Algorithms SUPER ELLIPTICAL CURVE POLYNOMIAL CURVE Implementation SUPER ELLIPTICAL CURVE POLYNOMIAL CURVE OPTIMIZATION ROUTINE Results SUPER ELLIPTICAL BASE CURVE Design Variables Design Variables Design Variables Comparison with HyperStudy POLYNOMIAL BASE CURVE Comparison with HyperStudy Conclusion References Appendix... 5

5 Introduction 1 Introduction Snap rings are widely used in mechanical assemblies, both internal and external. When an internal snap ring is placed in position in the housing it exerts the pressure on the walls of housing. During service they suffer rapid wear at the open ends, due to higher contact pressure at these locations. Conventionally a circular arc with constant radius is used as base curve for snap ring geometry. In this thesis, super elliptical and polynomial curves are investigated as alternative to circular arc. Contact pressure calculations are combined with successive response surface methodology (SRSM) to determine the parameters of base curves for minimizing the contact pressure and hence wear. 1.1 Background Response surface methodology (RSM) was first introduced in Box and Wilson (1951). Since then it has been widely used for optimization problems. Successive Response surface methodology (SRSM) is the extension of RSM, Where the design space is gradually reduced. This gradual reduction in design space is presented in Stander and Craig (22). This technique has been implemented and improved by Gustafsson and Strömberg (28). The SRSM has then been used, for example, for shape optimization of castings (Gustafsson & Strömberg, 28) and optimization of sheet metal forming processes (Jansson, 25) Previous work to determine the optimal shape of a snap ring was done by Strömberg (27). In that work a polynomial curve was considered and an inhouse contact software was used. Neural networks with SRSM were used to find the optimal geometry. 1.2 Purpose and aims The aim is to determine the equations for super elliptical and polynomial curves that minimize the maximum contact pressure and compare the results for both geometries. The total length of all new proposed designs of snap ring must be equal to the length of original design with circular arc. Mathematically it can be expressed as min max s.t. (1) Here represents the collection of all design variables and represents the length of the original snap ring. Figure 1 shows geometry of the original snap ring in free state. 1

6 Introduction Figure 1: Original Geometry of Snap Ring Least squares method will be used to fit the second order response surface and optimization problem is solved by using Quadratic Programming. The solution obtained by Quadratic Programming is also compared with optimization performed in commercial software, HyperStudy, which is a part of the software package HyperWorks (HyperWorks, 29). In HyperStudy, method of moving least squares will be chosen to fit the response surface. 1.3 Delimits Only second order models will be used to build a response surface. Higher order models or the techniques like neural networks or Kriging model will not be considered for this work. 1.4 Outline The thesis is organized in several parts. In the current introductory section, the project background, aims and scope have been described. In section 2, a more thorough theoretical discussion on the main ideas and concepts behind the Response Surface Methodology, Successive Response Surface Methodology, Contact Mechanics and different types of base curves considered here, are discussed. Section 3 gives the details how the project was implemented. Which software were used and how the optimization routine works. In Section 4, the results achieved in this work are presented. In section 5, the conclusion of the performed work is summarized. In appendix, dimensions of the original snap ring and housing are given. 2

7 Theoretical background 2 Theoretical background This section will provide information about the theoretical knowledge on which the implementation of the project will be based. 2.1 Response Surface Methodology Response Surface Methodology (RSM) is a collection of statistical and mathematical techniques useful for developing, improving and optimizing processes. (Myers & Montgomery, 22, p. 1) Different techniques are used for optimization of a function. But in most practical problems the relation between response and design variables is not known. In such situations, several experiments or computer simulations are performed and relation (function) between independent variables and response is determined. Once the relation has been determined then the optimum of the original problem can be estimated by determining the optimum of this function. Basic Concepts Before going into the details of how response surface methodology is implemented, first a few basic concepts will be explained. Response There is always some criterion or measure to judge the performance of a product or system. This criterion is called response. Design Variable The variables that affect the response are called design variables, independent variables or factors. Response Surface The algebraic expression that describes the response as a function of the independent variables is known as response surface. In other words, response surface describes the relationship between the response and the independent variables. Figure 2, shows graphically the relationship between response and independent variables ( x 1,x 2 ) for the following function. y 1 5x 1x 8x 12x 3

8 Theoretical background x x Figure 2: Response Surface The graphical display is a very useful tool that elaborates the relationship between response and variables in a very convenient way. This graphical display of the response as a surface has led to the terminology of response surface. If a dependent variable y is a function of k independent variables then mathematically it is expressed as yfx,x,x..x (2) Such a relation is called the functional relation. For given values of independent variables, the function f indicates the corresponding value of y. Figure 3 shows functional relation graphically. Dependent Variable Independent Variable Figure 3: Example of Functional Relation But the function is not always known for most practical problems. In such situations, several experiments or computer simulations are performed and relation between independent variables and response is determined. In this situation it is called statistical relation and expressed as. 4

9 Theoretical background yfx,x,x..x ε (3) Here ε is the random error term. Figure 4 shows statistical relation along with responses graphically. Response Independent Variable Figure 4: Example of Statistical Relation A statistical relation, unlike a functional relation, is not perfect. This means that the curve (in case of single independent variable) does not, in general, passes through the data points. Instead it is a close fit to the data points. Implementation of RSM Implementation of RSM is a three step process. These steps are following 1) Performing Experiments 2) Building Response Surface 3) Optimization of Response Surface Performing Experiments To determine the statistical relation between independent variables and response, it is always required to perform sufficient number of experiments or computer simulations. Computer simulations have high computational cost for complex problems. Therefore it is desired to know about a problem much, with only a few simulations. DoE (Design of Experiments) is the method to choose the combinations of design variables to perform experiments or simulations. This is also called designed experiments or experimental design. DoE give maximum information using minimum runs and hence minimum cost. 5

10 Theoretical background Basic Terminology Before going into the details, first basic terminology will be explained. Variable Domain The set containing all the values between lower and upper limit of a variable is called variable domain. Level Level is defined as the value of a design variable considered for an experimental run. For two level designs, the levels are taken as the lower and upper limit values of a design variable. For three level designs, middle value of the variable domain, in addition to the lower and upper limit values are considered as the levels of variable. This is elaborated in Figure 5. Two Levels Three Levels Figure 5: Levels of a Variable Design Space Design space is composed of all possible combinations of values from the domains of variables involved in an experiment. Figure 6 elaborates the design space graphically. Max Variable 2 Design Space Min Min Variable 1 Max Figure 6: Design Space for two variables Coded Variable It is customary that the design variables are scaled such that the lower and upper limit of each variable is given by -1 and 1respectively. Let ξ represent a variable in natural units and x represent the variable in coded units, then a variable can be transformed by the following formula 6

11 Theoretical background /2 /2 (4) Design Point One run of an experiment or computer simulation can be represented by a single point in Design space. It is required to specify the levels of all variables to define a design point. Figure 7 elaborates the design point graphically Variable 2 Max Min Design Space Design Point Min Variable 1 Max Figure 7: Design Point for two variables Different Experimental Designs Different experimental designs are used depending upon the area of application. Only few will be presented here Factorial Experiments An experimental design is called factorial design if all the design variables involved are considered only for a specified number of levels. It contains all possible level combinations of design variables. Such a design consists of L k design points, here L denotes number of levels and k denotes the number of variables. The disadvantage of factorial experiments is that the number of experiments increases rapidly with the increase in number of design variables. Figure 8 shows a 2 k and 3 k designs. 2 k Factorial Design A design with lower and upper limit combinations of all design variables is called a full factorial design in two levels. 3 k Factorial Design A design with middle, lower and upper limit combinations of all design variables is called a full factorial design in three levels. 7

12 Theoretical background X 2 X 2 X 1 X 1 Figure 8: 2 k and 3 K Factorial Designs Central Composite Design This is obtained by adding one center point and a set of star points along the axis to 2 k design. The star points are located at a distance from the origin. If 1, then the design is called circumscribed and if 1, then design is called inscribed as elaborated in Figure Circumscribed Inscribed Figure 9 : Central Composite Designs Box-Behnken design First all possible combination of variables pairs are identified. Then variables in these pairs are combined according to 2 2 factorial design and remaining variables are set to zero in each design point. Then a center point is added. More than one center points can be added if required. Figure 1 shows a Box-Behnken design for three variables. 8

13 Theoretical background Figure 1: Box-Behnken Design for three variables Choosing an Experimental Design Experimental design chosen for a given problem should minimize the number of runs without losing quality. Following are the general guidelines for choosing experimental design. 1) 2 k factorial design is appropriate for a first order response surface. 2) Central composite and Box-Behnken designs are appropriate for a second order response surface Building Response Surface After performing the desired number of experiments or computer simulations, next step is to determine the relation that describes the response as a function of the independent variables. The relation is an approximate model and statistical techniques are used to build it. This involves choosing a polynomial of suitable order. A first order model, for two variabl es is yβ β x β x ε (5) In this equation β s are unknown constants and need to be determined. β s are called regression coefficient s. A second order mod el, for two variables is yβ β x β x β x β x β x x ε (6) In generalized form first and second order polynomials can be written as Here K is the number of variables. K yβ β x ε K K K yβ β x β x β x x ε (7) 9

14 Theoretical background Selection of a suitable order of polynomial depends upon the complexity of the problem. When there is significant curvature in the responses, then first order model is inadequate. In such case second order model is reasonable choice. Figure 11 elaborates when higher order polynomials should be chosen. First Order Second Order Third Order Figure 11: Curvature and Order Of Polynomial The second order model is usually preferred as it is very flexible and can model quite complex processes. It is generally a good estimation of the true response surface. Estimation of the Regression Coefficients To estimate the values of regression coefficients in the polynomials, it is required to perform a minimum number of computer simulations. This number of computer simulations depends upon the number of variables and order of polynomial. Table 1 lists the minimum number of simulations required (Strömberg, 29). Order of model Minimum Number of Simulations (N min ) First Second K 1 K 1K 2 2 Table 1: Minimum Number of DoE Following three methods are used to estimate the regression coefficients. 1. Least Squares 2. Weighted Least Squares 3. Moving Least Squares 1

15 Theoretical background Leas t Squares Suppose that computer simulations are performed to compute the responses for at least n N experimental points. Let x denote the i th level of variable x and the responses be designated as y,y,y.y. Then the first order model in terms of the individual responses can be described as y β β x β x β x ε, i=1,2,3.n (8) This equation can be expressed in terms of all responses in matrix form as Where (9), β ε β, 1 ε 2, β ε The method of least squares estimates the β s so that the sum of the squares of the errors, ε, are minimized. Graphically it can be explained as minimizing the sum of the squares of the vertical deviations of responses from estimated model as shown in Figure 12. Response Value Response Estimated Model Independent Variable Figure 12: Deviation of Data Error in the matrix form can be expressed as Then sum of the squares of the errors can be expressed as (1) Lε (11) 11

16 Theoretical background According to method of least squares, the estimator of β is the b that minimizes the sum of squares of e rrors. This implies that L Where (12) b b 1 b The estimator b can be determined from the following equation So the fitted model is (13) ŶXb (14) Weighted Least Squares Residual can be defined as the difference between the Response value y i (obtained by computer simulation) and the corresponding fitted value ŷ i. Mathematically it can be expressed as ŷ (15) Magnitude of residual is represented by the vertical distance of the response value y i from the corresponding point on the fitted model as shown in Figure 13. ŷ i Response y i ŶXb Independent Variable Figure 13: Illustration of Residual Residuals are used to evaluate whether a fitted model is appropriate for the response data at hand. Figure 14 shows scatter plot of responses and fitted model. 12

17 Theoretical background It can be seen that the residuals increase with the increase in independent variable. Higher the value of the independent variable, the more spread out the residuals are. Response ŶX Independent Variable Figure 14: Scatter Plot and Fitted Model Responses with small spread are more reliable than those with large spreads. Weighted Least Squares (WLS) technique is used to take this relative reliability of responses into consideration. Responses with small spread are given more importance or weightage to determine the estimators of regression coefficients. In Weighted Least Squares technique, each response is given a weight and this weight is incorporated into the criterion for determining the estimators of regression coefficients. Lw ε If error variance (σ ) is known, then weight can be expressed as (Kutner, Nachtsheim & Neter, 24) The normal equation can be expressed as follows. (16) w 1 (17) (18) And the estimator b can be det ermined from following equation (19) W is the matrix containing the weights for all responses. 13

18 Theoretical background Moving Least Squares Moving least squares (MLS) method is a generalization of weighted least squares technique. Once a response surface has been built, response values can be computed for a new set of variables without running a computer simulation of original model. Let be a point where fitted model is evaluated and a DoE sampling point. Then the distance between these two points can be written as In MLS the weight is the function of this δ (2) w (21) The main difference between WLS and MLS is that the weights associated with individual experimental points remain constant in WLS but they are a function of δ in MLS. The weight associated with a particular experimental point, decays as the point moves away Examples of some commonly used functions are (Strömberg, 29) ω, ωδ exp δ 2α Because the weights are a function of, the estimators of regression coefficients are also dependent on. So (22) (23) Optimization of Response Surface Optimization can be defined as the process of achieving optimal level or state of something. Optimal state can be either the maximum or minimum of a function. Once a response surface has been built, response values can be computed for a new set of variables without running a computer simulation of original model. Optimum value for the process can also be determined by determining the optimum of response surface. If a second order model is used for a response surface, then we can easily compute the Stationary Point of a Response surface by differentiating the fitted model and equating it to zero. But additional information is needed to decide the nature of the stationary point i.e. if it is a point of maximum response, minimum response or saddle point. In case of saddle point, response is neither maximum nor minimum. Figure 15, Figure 16, and Figure 17 show these cases for two variables. 14

19 Theoretical background x 2 x 1 Figure 15: Stationary Point is Maximum Response x 2 x 1 Figure 16: Stationary Point is Minimum Response x 2 x 1 Figure 17: Stationary Point is Saddle Point Second order polynomial for k variables can be expressed as yβ β x β x β x β x β x β x x β x x β, x x ε (24) 15

20 Theoretical background Then the fitted model in matrix notation can be written as (Myers & Montgomery, 22). Where ŷ (25) x x 2, x b b 2, b /2 /2 /2. Stationary Point can be determined by differentiating the fitted model and equating it to zero. ŷ So By denoting the stationary point as x s The predicted response at the stationary point is 2 (26) (27) ŷ ŷ (28) Let λ1, λ1,.. λk be the eigenvalues of B. The signs of λ s determine the nature of stationary point. (Myers & Montgomery, 22) 1. If λ 1, λ 1,.. λ k are all negative, the stationary point is a pint of maximum response. 2. If λ 1, λ 1,.. λ k are all positive, the stationary point is a pint of minimum response. 3. If λ 1, λ 1,.. λ k are mixed in sign, the stationary point is a saddle point. 16

21 Theoretical background 2.2 Successive Response Surface Methodology As the response surface is an approximation of the true response of a product or process, the optimum point determined from a response surface is not necessarily the optimum of the actual process. So to determine the optimum with sufficient accuracy, additional DoE is performed and a new optimum is determined. The design space for this new DoE is centered around the previous optimum and smaller than initial design space. In this way successive DoE are performed around the optimums within a design space which is reduced in each successive DoE. This procedure is repeated several times until the following condition is satisfied (Strömberg, 29). (29) Here x denotes the optimum design point determined at k th iteration and y denotes the response calculated at this point. is a small number to check the convergence. This procedure of performing successive DoE and determining new improved optimums is referred as Successive Response Surface Methodology (SRSM). The varying design space is usually called Region of Interest (RoI). The principle of SRSM is presented in Figure 18. SRSM is an iterative method. Initially a DoE for a specified design space is performed, response surface is built and optimum point is determined from this surface. Then a new DoE is performed for a Design space that is smaller than initial design space and centered on the optimum point. This procedure narrows down the search for optimum point. The centering of new RoI around the optimum and its gradual reduction in size is usually called Panning and Zooming technique. This technique is very well presented by Gustafsson and Strömberg (28). 17

22 Theoretical background Opt 1 Opt 2 RoI 1 RoI 2 Variable 2 Opt 4 Opt 3 RoI 4 RoI 3 Variable 1 Figure 18: The Principle of Successive Response Surface Optimization 18

23 Theoretical background 2.3 Contact Mechanics When two solid bodies touch each other at one or more points, localized stresses are developed in the body and deformation takes place. Signorini s Contact Conditions Signorini s contact conditions are very important while studying the contact phenomenon. These conditions are as follows 1. A node on elastic body cannot penetrate the obstacle. 2. Normal contact force cannot be negative. 3. Normal contact force should be zero if the node is not in contact and greater than zero if contact establishes. To express these conditions mathematically, consider that an elastic body is deformed by an external force F and the deformation is restricted by a rigid body, which is fixed, as shown in Figure 19 Elastic Body Rigid Body Figure 19: An elastic body obstructed by a rigid body A set of nodes are defined on the boundary of elastic body where the two bodies are expected to contact each other. For each node a normal direction is determined. Now let denote the displacement of node and denote the initial normal distance between the node and rigid body. Then the first Signorini s contact condition can be expressed as.. (3) Let denote the normal contact force for a node, and then second condition can be expressed as 19

24 Theoretical background And the third condition can be expres sed as (31). (32) For all contact nodes, these equations can be expressed as o (33) In these equations N denotes the matrix containing for all contact nodes denotes the matrix containing for all nodes denotes the matrix containing for all nodes denotes the matrix containing for all nodes o denotes the Hadamard product. The last two conditions can also be derived by minimizing the potential energy in the elastic body while considering the no-penetration of rigid body as constraint. Mathematically it can be expressed as Here the potential energy can be describe d as min Π (34) Π (35) The solution to this problem is given by the following KKT conditions (Strömberg, 29).,, o (36) Here is a column vector containing Lagrange multiplier for each contact constraint. These are interpreted as the contact forces in normal direction. Normal distance can be represented as.. Signorini s contact conditions are illustrated in Figure 2 2

25 Theoretical background Contact Pressure Contact Clearance Figure 2: Signorini s Contact Conditions Contact Algorithms Following algorithms are usually used to solve the contact problems. 1. Penalty Method 2. Uzawa s method 3. Newton s method 4. Smooth approximation Penalty Method In penalty formulation, Signorini s contact conditions are replaced by following (Strömberg, 29). Here the notation used can be explain ed by (37) 2 r is a penalty coefficient. By letting penalty model resembles Signorini s contact conditions. But this does not work in practice. With this method the contact force is proportional to the penetration distance, so it always implies a certain penetration. Penalty method is illustrated in Figure 21 Figure 21: Illustration of Penalty Method 21

26 Theoretical background Uzawa s Method This method is described as follows (Strömberg, 29). 1. Let r > and a small number. Set and 2. Solve 3. Update the contact forces according to 4. If then stop otherwise let 1 and go to step 2 Newton s Method This method is described as follows (Strömberg, 29). First introduce, (38) Following search criterion can b e derived by Taylor expansion. Then Newton s approach can be described as (39) 1. Set and pick a small. Guess an initial state 2. Determine 3. Then find 4. If then stop otherwise let 1 and go to step 2 Smooth Approximation There is a drawback in the Newton method that Projection is not smooth. This is overcome in Smooth Approximation by replacing Signorini s contact conditions by a differentiable relation. 22

27 Theoretical background 2.4 Super Elliptical Curve Super elliptical curve can be described by the following equation. The symbols used are illustrated in Figure 22 1 (4) Y Ry r Rx X This can Figure also 22: be expressed Illustration as of Symbols used in Super Elliptical curve equation The super elliptical curve equation can also be expressed as (41) Length of Curve The length of curve between X = A and X= B can be found from 1 (42) Y X can be found be taking derivative of equatio n (41) By chain rule So dy dx dx dx dx dx dy. dy dy n dx dx (43) (44) / (45) 23

28 Theoretical background Instantaneous Radius Instantaneous radius, r can be dervied from super elliptical curve equation by expressing X and Y in terms of r. So super elliptical curve can be written as And it can be written as 1 (46) (47) 2.5 Polynomial Curve A curve expresed by a polynomial is called polynomial curve. A Polynomial curve where radius is the function of angle and order is four can be described by the following equation. The the symbols used are illustrate d in Figure 23 rrαc C αc α C α C α (48) Here C,C,C,C and C are coefficients. α γ r(α) Figure 23: Illustration of Symbols used in Polynomial curve equation Length of Curve The length of can be found from L (49) 24

29 Implementation 3 Implementation First of all existing geometry of the part is analyzed to see the contact pressure distribution along the length of outer edge of the ring. For this purpose commercial FE software Abaqus (Abaqus, 27) is used. Because the geometry of the ring is symmetric about horizontal axis, as shown in Figure 24, only the half of ring will be analyzed in Abaqus. It will reduce the computation time significantly. Figure 24: Symmetric Geometry of Snap Ring The cross-section of the ring is a full circle. It can be represented as a semi circle by using symmetry conditions. The cad model of snap ring modeled in Abaqus is shown in Figure 25. Figure 25: Snap Ring representation in Abaqus 25

30 Implementation The housing will be modeled as a rigid body, because it is much stiffer than the snap ring. It is represented by a single surface as shown in Figure 26 Figure 26: Representation of Housing as a Surface Augmented Lagrange method is selected to solve the contact problem between housing and ring. First order elements are used for meshing the ring as contact algorithm may not work properly with higher order elements. Figure 27 shows the contact pressure graph obtained by simulating in the Abaqus Figure 27: Contact pressure graph for 3D representation 26

31 Implementation For this simulation elements were generated for the snap ring and maximum contact pressure of 17 MPa was observed as shown in the graph. True representation of the parts in 3D is computationally expensive. So it was thought it could be time saving if 2D representation can be sufficient to represent the problem. Figure 28 shows the contact pressure graph obtained by simulating in the Abaqus for 2D representation of the problem. Figure 28: Contact pressure graph for 2D representation For this simulation 1864 elements were generated for the snap ring and maximum contact pressure of 24.6 MPa was observed as shown in the graph. By comparing the contact pressure graphs for 3D and 2D representations, it is obvious that although scale of magnitudes are different but the overall profile of the contact pressure is almost same of both represenatations. APython script is developed for this problem by recording macros in the Abaqus. A multi-point spline is used to build the base curve of snap ring, while recording macros. The spline curve is flexible enough to represent any geometry required for this project. The python script is modified so that it can read a text file containing coordinates of points of the base curve and build the spline according to this coordinate data. This script also performs the analysis in Abaqus and writes maximum contact pressure information for each simulation in a text file. A circular arc is used as base curve in the original design of the snap ring. By looking at the contact pressure graph, it is apparent that a curve with constant radius is not sufficient for obtaining the uniform contact pressure distribuition. To obtain the uniform contact pressure distribuition, two type of curves were considered i.e. these curves were used as base curve for creating the geometry of snap ring. These curves are as follows 1. Super Elliptical 2. Polynomial 27

32 Implementation 3.1 Super Elliptical Curve Figure 29 shows the snap ring in compressed state (deformed shape) after simulating in Abaqus. A Gap can be seen between the snap ring and housing in the right quarter. But, on the left quarter there is no gap between the ring and housing. Figure 29: Compressed state showing the gap The gap shown in Figure 29 implies that for super elliptical base curve, different parameters should be used for the right and left quarters of base curve. This can be expressed graphically as shown in Figure 3. Y Ry X L Rx 1 Rx 2 X Figure 3: Illustration of Symbols used in Super Elliptical curve equation 28

33 Implementation Here is the x-coordinate of end point of the curve. So the equation for left and right quarters can be written as 1 1 So optimization problem can be expressed as follows. (5) min ma x s.t. 1 1 (51) Here represents the length of original snap ring. For given values of,,, and, is computed from the constraint in (51). Then the coordinates for finite number of points at equal angular distances are computed by using equations in (5). 3.2 Polynomial Curve In the polynomial considered here, radius is the function of angle and order is three. The equation can be written as follows. The symbols used are illustrated in Figure 23. rα C C αc α C α (52) So optimization problem can be expressed as follows. min max s. t. C C C α C α α (53) For given values of C,C,C and C, is computed from the constraint in (53). The radius at finite number of points at equal angular distances is computed by using equation in (52). Then x and y coordinates are calculated by the radius and angle information at these points. Matlab coding was developed to generate the coordinates of base curves and export them as text file. To evaluate the integrals in (51) and (53), built-in command quad was used. 29

34 Implementation 3.3 Optimization Routine Optimization routine is written in Matlab. It is used to determine the parameters in the curve equations for optimum geometry and can be described as follows. Figure 31 shows the routine as a flowchart. 1. A suitable start point and initial range are specified. 2. Central Composite Design is used to decide the combination of design variables. For these set of design variables, coordinates of curves are generated in Matlab, while length constraint is implemented. These coordinates are then exported in a text file. 3. A python script reads the text file containing the coordinates and performs the analysis in Abaqus. Then the script writes maximum contact pressure information for each simulation in a text file. 4. The text file containing maximum pressure information is imported in Matlab. This information combined with the DoE variable combinations are used to build the quadratic response surface. Least squares method is used to fit the response surface. The optimum of this response surface is determined by using the quadratic programming. 5. If 1, where k is the iteration number, check the convergence according to the following criteria (Gustafsson & Strömberg, 28). Here denotes the optimum design point determined at k th iteration and RSM denotes the contact pressure calculated at this point by the response surface. FEA denotes contact pressure calculated by FEA simulation, at k th iteration, in Abaqus. If and then the optimum has converged and is the optimum point. If or then solution has not converged and further iteration is needed. Here is a small number to check the convergence. 6. Determine new RoI by Pan and Zoom technique as presented by Gustafsson and Strömberg (28) and go to step 2. 3

35 Implementation In this routine, tasks of building of response surface and computation of optimum are performed by developing the code in Matlab. This optimization is also compared with commercial software, HyperStudy. In that case, step 4 is modified such as DoE variable combinations and maximum contact pressure information is exported as a text file in a format that is readable to HyperStudy. HyperStudy is launched in batch mode from Matlab. Moving Least Squares method is chosen with second order approximation to perform optimization. After optimization, the optimum design point and optimum response are imported into Matlab. Other steps are same in both cases. Set Start Point and initial Range Setup a Design Of Experiments Solve with FEA Build Response Surface and Optimize Check Convergence No Pan and Zoom Yes Stop Figure 31: Flowchart for Optimization Routine Matlab provides the built-in functions to generate the DoE and solve quadratic programming problems. ccdesign function was used to generate central composite design and quadprog function was used to solve the quadratic programming problem. 31

36 Results 4 Results In this section, results obtained from the successive response surface optimization are described. It is important to mention that sometimes the results of iterations, even after convergence, have been presented for comparison purposes. 4.1 Super Elliptical Base Curve It can be concluded from the Figure 29 that for uniform contact pressure distribution. 2 2 (54) Therefore it is supposed that just by finding suitable values for these two variables, the target of uniform contact pressure with lower maximum magnitude can be reached. So an optimization study is performed for these two design variables and all others are kept constant Design Variables Following are chosen as the starting point for this successive response surface optimization (55) Figure 32 shows the graph for the maximum contact pressure at the location of optimum design point ( ) which is determined from the response surface at every iteration by quadratic programming. Solid line shows the contact pressure as computed from the response surface and dotted line shows the contact pressure as computed by the actual simulation in Abaqus at. During initial iterations, the contact pressure computed from response surface differs much from the actual simulation results in Abaqus. It can be expected due to a large design space in the beginning. But as the design space (or RoI) continues to reduce, the results predicted by the response surface come closer to the actual simulation results. 32

37 Results Maximum Contact Pressure[MPa] No. of Iterations Response Surface Simulation Result Figure 32: Maximum Contact Pressure at the optimum design point for 2 variables. Solid line shows results as computed by Quadratic Programming from response surface. Dotted line shows the results obtained by simulation in Abaqus at optimum design point. At iteration 13, convergence was achieved, at minimum contact pressure of 8.5 MPa with the following magnitudes of design variables (56) Figure 33 shows the graph of contact pressure for this optimum design point. Figure 33: Contact Pressure graph when 2 variables are considered for optimization 33

38 Results Figure 34 shows the snap ring in compressed state (deformed shape) for this optimum design point. A gap can be seen between the snap ring and housing. Figure 34: Compressed state showing the gap Although the maximum contact pressure has decreased to 8.5 MPa but still it is not distributed over the whole ring. Therefore it was decided to run the optimization study for three design variables, where the third variable will be the Ry Design Variables It was supposed that by increasing the value of Ry, gap can be eliminated or reduced. It was also supposed that new optimum for and will be near the optimums obtained while optimizing for 2 variables. So following are chosen as the starting point for this successive response surface optimization Ry.1 (57) Figure 35 shows the graph for the maximum contact pressure at the location of optimum design point ( ) at every iteration. During initial iterations, contact pressure computed from the response surface differs much from the actual simulation results in Abaqus. But as the design space reduces, the results predicted by the response surface come closer to the actual simulation results, but this behavior does not persist for further iterations. The difference keeps on changing. It can be concluded that the relation between the response and design variables is too nonlinear, both at global and local levels, that second order response surface cannot approximate it with sufficient accuracy. 34

39 Results Maximum Contact Pressure[MPa] No. of Iterations Response Surface Simulation Result Figure 35: Maximum Contact Pressure at the optimum design point for 3 variables. Solid line shows results as computed by Quadratic Programming from response surface. Dotted line shows the results obtained by simulation in Abaqus at optimum design point. As the response surface is unable to represent the true response, so it is necessary to look at the results of DoE at individual iterations. By looking at the results of DoE at iteration 8, lowest response of 6.57 MPa was observed for the following information of design variables (58) Figure 36 shows the graph for this design point. This design point is different from the optimum point calculated by the response surface. Figure 36: Contact Pressure graph when 3 variables are considered for optimization 35

40 Results Although maximum contact pressure has decreased to 6.57 MPa but it is not distributed over the whole ring and gap, between ring and housing, is still there. Therefore it was decided to run the optimization study for five design variables. The knowledge gained from the optimization studies for 2 and 3 variables will be utilized to decide the start point Design Variables It was supposed that by decreasing Rx 1 and Rx 2, gap can be eliminated or reduced. It was also supposed that new optimum for, and Ry will be near the optimums obtained while optimizing for 3 variables. So following are chosen as starting point for this successive response surface optimization (59) Figure 37 shows the graph for the maximum contact pressure at the location of optimum design point ( ) at every iteration. The contact pressure computed from response surface differs from the actual simulation results in Abaqus, throughout all the iterations. The difference between both keeps on changing, furthermore contact pressure computed from response surface shows negative values that is unrealistic. It can be concluded that the relation between the response and design variables becomes more and more nonlinear with the increase in design variables. Maximum Contact Pressure[MPa] No. of Iterations Response Surface Simulation Result Figure 37: Maximum Contact Pressure at the optimum design point for 5 variables. Solid line shows results as computed by Quadratic Programming from response surface. Dotted line shows the results obtained by simulation in Abaqus at optimum design point. 36

41 Results As the response surface is unable to represent the true response, so it is necessary to look at the results of DoE at individual iterations. By looking at the results of DoE at iteration 5, minimum contact pressure of 4.51 MPa was found for following information of design variables (6) Figure 38 shows the graph for this design point. The optimum design point which is computed from response surface has the response of 5.5 MPa. Figure 39 shows the graph for the design point shown in (6) when simulated for 3D representation, here the maximum contact pressure is MPa. Figure 38: Contact Pressure graph for 2D representation when 5 variables are considered for optimization 37

42 Results Figure 39: Contact Pressure graph for 3D representation, when 5 variables are considered for optimization Although maximum contact pressure has decreased significantly as compared to original design but still it is not distributed over the whole ring. Figure 4 shows the snap ring in compressed state (deformed shape) for this optimum design point. A gap can be seen between the snap ring and housing. Although the gap is considerably smaller than previous designs but it is still there. Figure 4: Compressed state of Snap Ring obtained in Abaqus 38

43 Results In general, FEA softwares are considered more reliable to compute the von Mises stresses as compared to the contact stresses. So another optimization study was performed with the objective of minimizing the von Mises stress instead of contact stress to see if it can propose a better design of snap ring. Figure 41 shows the result of this optimization study for the start point shown in (59). It was observed that the optimum designs proposed by minimizing of maximum von Mises stress problem are different from those of minimizing of maximum contact pressure problem. After five iterations, significant gaps between the ring and housing begin to appear. Maximum Von Mises Stress[MPa] No. of Iterations Response Surface Simulation Result Figure 41: Maximum von Mises Stress at the optimum design point for 5 variables. Solid line shows results as computed by Quadratic Programming from response surface. Dotted line shows the results obtained by simulation in Abaqus at optimum design point Comparison with HyperStudy Figure 42 shows the result for the start point shown in (59), while HyperStudy is used for optimization. Convergence is achieved at 5 th iteration and optimums computed during iterations 3, 4 and 5 from response surface confirm with the actual simulation results in Abaqus. Minimum pressure of 6.7 MPa is obtained at convergence. 39

44 Results Maximum Contact Pressure[MPa] No. of Iterations Response Surface Simulation Result Figure 42: Maximum Contact Pressure at the optimum design point for 5 variables. Solid line shows results as computed by HyperStudy. Dotted line shows the results obtained by simulation in Abaqus at optimum design point. By looking at the results of DoE at iteration 8, minimum contact pressure of 4.93 MPa was found for the following information of design variables (61) By comparing the graphs in Figure 42 and Figure 37, it can be concluded that HyperStudy performed better to attain convergence but the optimum found is not the real minimum. Figure 43 shows the result of another optimization study for the start point shown in (59), when HyperStudy is used for optimization but instead of maximum contact pressure, maximum von Mises stress is minimized. It was observed that the optimum designs proposed by minimizing of maximum von Mises stress and maximum contact pressure problem are different from the first iteration. 4

45 Results Maximum Von Mises Stress[MPa] No. of Iterations Response Surface Simulation Result Figure 43: Maximum von Mises stress at the optimum design point. Solid line shows results as computed by HyperStudy. Dotted line shows the results obtained by simulation in Abaqus at optimum design point. 41

46 Results 4.2 Polynomial Base Curve To determine the important variables in polynomial curve equation for response surface study, few simulations were performed where all coefficients were fixed and only was changed in the polynomial equation. It was found that does not have much influence on the contact pressure distribution over the snap ring although it changes the maximum contact pressure. So it was decided to keep its magnitude fixed at the original circular radius of the snap ring and consider the other three coefficients i.e., and as design variables during the successive response surface optimization. As compared to the super elliptical curve, it was difficult to guess the start point for design variables. So it was decided to run the optimization study over a large design space to explore it. During the investigation of the results of this study, it was found that there are some designs which give lower contact pressure but are not feasible i.e. they had large gaps between snap ring and housing in the compressed state. So another optimization study with the start point which would generate designs of snap ring that are most probably feasible is performed. The following was chosen as starting point for this successive response surface optimization. 1/1 25/1 2/1 (62) Figure 44 shows graph for the maximum contact pressure at the location of optimum design point ( ) at every iteration. It is evident that maximum contact pressure is decreasing rapidly in every iteration. For the initial five iterations, maximum contact pressure computed from the response surface and actual simulation results are nearly same but later they differ. It is interesting to note that when design space is large, quadratic response surface can calculate the response value with sufficient accuracy but as the response value reaches optimum, quadratic response surface cannot predict the response accurately. It can be concluded that when the response is near the optimum, the shape of true response surface becomes highly non-linear that a second order response surface approximation is not sufficient to represent the true response surface. 42

47 Results Maximum Contact Pressure [MPa] No. of Iterations Response Surface Simulation Result Figure 44: Maximum Contact Pressure at the optimum design point. Solid line shows results as computed by Quadratic Programming from response surface. Dotted line shows the results obtained by simulation in Abaqus at optimum design point. During the study of DoE results at iteration 6 and 7 it was found that design points with the lower responses are scattered arbitrarily in the design space. It might be due to the presence of multiple stationary points in the design space. While looking at the DoE results for iteration 7, the following design point was found with lowest response value of 1.74 MPa. /1 75/1 55/1 (63) Figure 45 shows the graph of contact pressure for this design point. Figure 45: Contact Pressure graph 43

48 Results Although maximum contact pressure has reduced significantly for this design point as compared to the original design with circular arc as base curve, but still the pressure is not very well distributed over the whole ring, especially at ends. This requirement is difficult to be expressed in the form of a constraint and hence visual inspection of contact pressure graph is necessary. Therefore contact pressure graphs for some other design points with lower contact pressure during iterations 6, 7 and 8 were drawn. But none has pressure distributed over the whole length of the snap ring. To fulfill this requirement, some new design points were guessed by looking at the results for the design points where contact pressure was low. By little experimentation, the following design point was found with maximum contact pressure of 2.6 MPa and better distribution over the whole ring. 1 /1 65/1 3/1 (64) Figure 46 shows the graph for this new design point. Figure 46: Contact Pressure graph It was thought that a better design point with the lower maximum contact pressure and better distribution can be found in the vicinity of this design point. So to explore the design space around this design point, another successive response surface optimization was performed with this design point as starting point. The range about this start point was set to be very small. After a few iterations, following point was found with maximum contact pressure of 1.46 MPa and better distribution ove r the whole ring / / /1 (65) 44

49 Results Figure 47 shows the graph for this new improved design point. Figure 48 shows the graph for same point when simulated for 3D representation; here the maximum contact pressure is MPa. Figure 47: Contact Pressure graph for 2D representation Figure 48: Contact Pressure graph for 3D representation Figure 49 shows the result of another optimization study for the start point shown in (62) but instead of maximum contact pressure, maximum von Mises stress is minimized. It was observed that the optimum designs proposed by minimizing of maximum von Mises stress and maximum contact pressure problem are same until sixth iteration, and after that they begin to differ. After 1th iteration, unfeasible designs with large gap between the ring and housing appear. 45

Guidelines for proper use of Plate elements

Guidelines for proper use of Plate elements Guidelines for proper use of Plate elements In structural analysis using finite element method, the analysis model is created by dividing the entire structure into finite elements. This procedure is known

More information

CHAPTER 4. Numerical Models. descriptions of the boundary conditions, element types, validation, and the force

CHAPTER 4. Numerical Models. descriptions of the boundary conditions, element types, validation, and the force CHAPTER 4 Numerical Models This chapter presents the development of numerical models for sandwich beams/plates subjected to four-point bending and the hydromat test system. Detailed descriptions of the

More information

Revision of the SolidWorks Variable Pressure Simulation Tutorial J.E. Akin, Rice University, Mechanical Engineering. Introduction

Revision of the SolidWorks Variable Pressure Simulation Tutorial J.E. Akin, Rice University, Mechanical Engineering. Introduction Revision of the SolidWorks Variable Pressure Simulation Tutorial J.E. Akin, Rice University, Mechanical Engineering Introduction A SolidWorks simulation tutorial is just intended to illustrate where to

More information

Topology Optimization of Two Linear Elastic Bodies in Unilateral Contact

Topology Optimization of Two Linear Elastic Bodies in Unilateral Contact 2 nd International Conference on Engineering Optimization September 6-9, 2010, Lisbon, Portugal Topology Optimization of Two Linear Elastic Bodies in Unilateral Contact Niclas Strömberg Department of Mechanical

More information

Lagrange Multipliers and Problem Formulation

Lagrange Multipliers and Problem Formulation Lagrange Multipliers and Problem Formulation Steven J. Miller Department of Mathematics and Statistics Williams College Williamstown, MA 01267 Abstract The method of Lagrange Multipliers (and its generalizations)

More information

Truss structural configuration optimization using the linear extended interior penalty function method

Truss structural configuration optimization using the linear extended interior penalty function method ANZIAM J. 46 (E) pp.c1311 C1326, 2006 C1311 Truss structural configuration optimization using the linear extended interior penalty function method Wahyu Kuntjoro Jamaluddin Mahmud (Received 25 October

More information

Example 24 Spring-back

Example 24 Spring-back Example 24 Spring-back Summary The spring-back simulation of sheet metal bent into a hat-shape is studied. The problem is one of the famous tests from the Numisheet 93. As spring-back is generally a quasi-static

More information

A Short SVM (Support Vector Machine) Tutorial

A Short SVM (Support Vector Machine) Tutorial A Short SVM (Support Vector Machine) Tutorial j.p.lewis CGIT Lab / IMSC U. Southern California version 0.zz dec 004 This tutorial assumes you are familiar with linear algebra and equality-constrained optimization/lagrange

More information

D-Optimal Designs. Chapter 888. Introduction. D-Optimal Design Overview

D-Optimal Designs. Chapter 888. Introduction. D-Optimal Design Overview Chapter 888 Introduction This procedure generates D-optimal designs for multi-factor experiments with both quantitative and qualitative factors. The factors can have a mixed number of levels. For example,

More information

Subject of Investigation

Subject of Investigation Subject of Investigation A 3D Analysis of the Node to Node and Node to Surface Contact Phenomenon using Penalty Function and Lagrange Multipliers Constraints in Matlab Academic Advisor Professor José M.

More information

Design and Analysis of Multi-Factored Experiments

Design and Analysis of Multi-Factored Experiments Design and Analysis of Multi-Factored Experiments Response Surface Methodology (RSM) L. M. Lye DOE Course Introduction to Response Surface Methodology (RSM) Best and most comprehensive reference: R. H.

More information

ME Optimization of a Frame

ME Optimization of a Frame ME 475 - Optimization of a Frame Analysis Problem Statement: The following problem will be analyzed using Abaqus. 4 7 7 5,000 N 5,000 N 0,000 N 6 6 4 3 5 5 4 4 3 3 Figure. Full frame geometry and loading

More information

Engineering Effects of Boundary Conditions (Fixtures and Temperatures) J.E. Akin, Rice University, Mechanical Engineering

Engineering Effects of Boundary Conditions (Fixtures and Temperatures) J.E. Akin, Rice University, Mechanical Engineering Engineering Effects of Boundary Conditions (Fixtures and Temperatures) J.E. Akin, Rice University, Mechanical Engineering Here SolidWorks stress simulation tutorials will be re-visited to show how they

More information

Sliding Split Tube Telescope

Sliding Split Tube Telescope LESSON 15 Sliding Split Tube Telescope Objectives: Shell-to-shell contact -accounting for shell thickness. Creating boundary conditions and loads by way of rigid surfaces. Simulate large displacements,

More information

Edge and local feature detection - 2. Importance of edge detection in computer vision

Edge and local feature detection - 2. Importance of edge detection in computer vision Edge and local feature detection Gradient based edge detection Edge detection by function fitting Second derivative edge detectors Edge linking and the construction of the chain graph Edge and local feature

More information

CHAPTER 3 AN OVERVIEW OF DESIGN OF EXPERIMENTS AND RESPONSE SURFACE METHODOLOGY

CHAPTER 3 AN OVERVIEW OF DESIGN OF EXPERIMENTS AND RESPONSE SURFACE METHODOLOGY 23 CHAPTER 3 AN OVERVIEW OF DESIGN OF EXPERIMENTS AND RESPONSE SURFACE METHODOLOGY 3.1 DESIGN OF EXPERIMENTS Design of experiments is a systematic approach for investigation of a system or process. A series

More information

Topology Optimization and Analysis of Crane Hook Model

Topology Optimization and Analysis of Crane Hook Model RESEARCH ARTICLE Topology Optimization and Analysis of Crane Hook Model Thejomurthy M.C 1, D.S Ramakrishn 2 1 Dept. of Mechanical engineering, CIT, Gubbi, 572216, India 2 Dept. of Mechanical engineering,

More information

1.2 Numerical Solutions of Flow Problems

1.2 Numerical Solutions of Flow Problems 1.2 Numerical Solutions of Flow Problems DIFFERENTIAL EQUATIONS OF MOTION FOR A SIMPLIFIED FLOW PROBLEM Continuity equation for incompressible flow: 0 Momentum (Navier-Stokes) equations for a Newtonian

More information

ENGINEERING TRIPOS PART IIA FINITE ELEMENT METHOD

ENGINEERING TRIPOS PART IIA FINITE ELEMENT METHOD ENGINEERING TRIPOS PART IIA LOCATION: DPO EXPERIMENT 3D7 FINITE ELEMENT METHOD Those who have performed the 3C7 experiment should bring the write-up along to this laboratory Objectives Show that the accuracy

More information

3D Finite Element Software for Cracks. Version 3.2. Benchmarks and Validation

3D Finite Element Software for Cracks. Version 3.2. Benchmarks and Validation 3D Finite Element Software for Cracks Version 3.2 Benchmarks and Validation October 217 1965 57 th Court North, Suite 1 Boulder, CO 831 Main: (33) 415-1475 www.questintegrity.com http://www.questintegrity.com/software-products/feacrack

More information

An Optimization Procedure for. Springback Compensation using LS-OPT

An Optimization Procedure for. Springback Compensation using LS-OPT An Optimization Procedure for Springback Compensation using LS-OPT Nielen Stander, Mike Burger, Xinhai Zhu and Bradley Maker Livermore Software Technology Corporation, 7374 Las Positas Road, Livermore,

More information

SDC. Engineering Analysis with COSMOSWorks. Paul M. Kurowski Ph.D., P.Eng. SolidWorks 2003 / COSMOSWorks 2003

SDC. Engineering Analysis with COSMOSWorks. Paul M. Kurowski Ph.D., P.Eng. SolidWorks 2003 / COSMOSWorks 2003 Engineering Analysis with COSMOSWorks SolidWorks 2003 / COSMOSWorks 2003 Paul M. Kurowski Ph.D., P.Eng. SDC PUBLICATIONS Design Generator, Inc. Schroff Development Corporation www.schroff.com www.schroff-europe.com

More information

Introduction to ANSYS DesignXplorer

Introduction to ANSYS DesignXplorer Lecture 4 14. 5 Release Introduction to ANSYS DesignXplorer 1 2013 ANSYS, Inc. September 27, 2013 s are functions of different nature where the output parameters are described in terms of the input parameters

More information

Chapter 1 Introduction

Chapter 1 Introduction Chapter 1 Introduction GTU Paper Analysis (New Syllabus) Sr. No. Questions 26/10/16 11/05/16 09/05/16 08/12/15 Theory 1. What is graphic standard? Explain different CAD standards. 2. Write Bresenham s

More information

An Efficient Sequential Approach for Simulation of Thermal Stresses in Disc Brakes

An Efficient Sequential Approach for Simulation of Thermal Stresses in Disc Brakes An Efficient Sequential Approach for Simulation of Thermal Stresses in Disc Brakes Asim Rashid 1, Niclas Strömberg 1 1 Jönköping University, SE-55111 Jönköping, Sweden Abstract In this paper an efficient

More information

Modelling Flat Spring Performance Using FEA

Modelling Flat Spring Performance Using FEA Modelling Flat Spring Performance Using FEA Blessing O Fatola, Patrick Keogh and Ben Hicks Department of Mechanical Engineering, University of Corresponding author bf223@bath.ac.uk Abstract. This paper

More information

SYSTEMS OF NONLINEAR EQUATIONS

SYSTEMS OF NONLINEAR EQUATIONS SYSTEMS OF NONLINEAR EQUATIONS Widely used in the mathematical modeling of real world phenomena. We introduce some numerical methods for their solution. For better intuition, we examine systems of two

More information

Theoretical Concepts of Machine Learning

Theoretical Concepts of Machine Learning Theoretical Concepts of Machine Learning Part 2 Institute of Bioinformatics Johannes Kepler University, Linz, Austria Outline 1 Introduction 2 Generalization Error 3 Maximum Likelihood 4 Noise Models 5

More information

The Finite Element Method for the Analysis of Non-Linear and Dynamic Systems. Prof. Dr. Eleni Chatzi, J.P. Escallo n Lecture December, 2013

The Finite Element Method for the Analysis of Non-Linear and Dynamic Systems. Prof. Dr. Eleni Chatzi, J.P. Escallo n Lecture December, 2013 The Finite Element Method for the Analysis of Non-Linear and Dynamic Systems Prof. Dr. Eleni Chatzi, J.P. Escallo n Lecture 11-17 December, 2013 Institute of Structural Engineering Method of Finite Elements

More information

ANSYS Workbench Guide

ANSYS Workbench Guide ANSYS Workbench Guide Introduction This document serves as a step-by-step guide for conducting a Finite Element Analysis (FEA) using ANSYS Workbench. It will cover the use of the simulation package through

More information

INTERIOR POINT METHOD BASED CONTACT ALGORITHM FOR STRUCTURAL ANALYSIS OF ELECTRONIC DEVICE MODELS

INTERIOR POINT METHOD BASED CONTACT ALGORITHM FOR STRUCTURAL ANALYSIS OF ELECTRONIC DEVICE MODELS 11th World Congress on Computational Mechanics (WCCM XI) 5th European Conference on Computational Mechanics (ECCM V) 6th European Conference on Computational Fluid Dynamics (ECFD VI) E. Oñate, J. Oliver

More information

FINITE ELEMENT MODELLING OF A TURBINE BLADE TO STUDY THE EFFECT OF MULTIPLE CRACKS USING MODAL PARAMETERS

FINITE ELEMENT MODELLING OF A TURBINE BLADE TO STUDY THE EFFECT OF MULTIPLE CRACKS USING MODAL PARAMETERS Journal of Engineering Science and Technology Vol. 11, No. 12 (2016) 1758-1770 School of Engineering, Taylor s University FINITE ELEMENT MODELLING OF A TURBINE BLADE TO STUDY THE EFFECT OF MULTIPLE CRACKS

More information

Finite Element Method. Chapter 7. Practical considerations in FEM modeling

Finite Element Method. Chapter 7. Practical considerations in FEM modeling Finite Element Method Chapter 7 Practical considerations in FEM modeling Finite Element Modeling General Consideration The following are some of the difficult tasks (or decisions) that face the engineer

More information

Stress analysis of toroidal shell

Stress analysis of toroidal shell Stress analysis of toroidal shell Cristian PURDEL*, Marcel STERE** *Corresponding author Department of Aerospace Structures INCAS - National Institute for Aerospace Research Elie Carafoli Bdul Iuliu Maniu

More information

Similar Pulley Wheel Description J.E. Akin, Rice University

Similar Pulley Wheel Description J.E. Akin, Rice University Similar Pulley Wheel Description J.E. Akin, Rice University The SolidWorks simulation tutorial on the analysis of an assembly suggested noting another type of boundary condition that is not illustrated

More information

Module 1 Lecture Notes 2. Optimization Problem and Model Formulation

Module 1 Lecture Notes 2. Optimization Problem and Model Formulation Optimization Methods: Introduction and Basic concepts 1 Module 1 Lecture Notes 2 Optimization Problem and Model Formulation Introduction In the previous lecture we studied the evolution of optimization

More information

Parametric. Practices. Patrick Cunningham. CAE Associates Inc. and ANSYS Inc. Proprietary 2012 CAE Associates Inc. and ANSYS Inc. All rights reserved.

Parametric. Practices. Patrick Cunningham. CAE Associates Inc. and ANSYS Inc. Proprietary 2012 CAE Associates Inc. and ANSYS Inc. All rights reserved. Parametric Modeling Best Practices Patrick Cunningham July, 2012 CAE Associates Inc. and ANSYS Inc. Proprietary 2012 CAE Associates Inc. and ANSYS Inc. All rights reserved. E-Learning Webinar Series This

More information

Challenge Problem 5 - The Solution Dynamic Characteristics of a Truss Structure

Challenge Problem 5 - The Solution Dynamic Characteristics of a Truss Structure Challenge Problem 5 - The Solution Dynamic Characteristics of a Truss Structure In the final year of his engineering degree course a student was introduced to finite element analysis and conducted an assessment

More information

Support Vector Machines.

Support Vector Machines. Support Vector Machines srihari@buffalo.edu SVM Discussion Overview 1. Overview of SVMs 2. Margin Geometry 3. SVM Optimization 4. Overlapping Distributions 5. Relationship to Logistic Regression 6. Dealing

More information

A Robust Optimum Response Surface Methodology based On MM-estimator

A Robust Optimum Response Surface Methodology based On MM-estimator A Robust Optimum Response Surface Methodology based On MM-estimator, 2 HABSHAH MIDI,, 2 MOHD SHAFIE MUSTAFA,, 2 ANWAR FITRIANTO Department of Mathematics, Faculty Science, University Putra Malaysia, 434,

More information

ES 128: Computer Assignment #4. Due in class on Monday, 12 April 2010

ES 128: Computer Assignment #4. Due in class on Monday, 12 April 2010 ES 128: Computer Assignment #4 Due in class on Monday, 12 April 2010 Task 1. Study an elastic-plastic indentation problem. This problem combines plasticity with contact mechanics and has many rich aspects.

More information

A nodal based evolutionary structural optimisation algorithm

A nodal based evolutionary structural optimisation algorithm Computer Aided Optimum Design in Engineering IX 55 A dal based evolutionary structural optimisation algorithm Y.-M. Chen 1, A. J. Keane 2 & C. Hsiao 1 1 ational Space Program Office (SPO), Taiwan 2 Computational

More information

OPTIMIZATION STRATEGIES AND STATISTICAL ANALYSIS FOR SPRINGBACK COMPENSATION IN SHEET METAL FORMING

OPTIMIZATION STRATEGIES AND STATISTICAL ANALYSIS FOR SPRINGBACK COMPENSATION IN SHEET METAL FORMING Optimization strategies and statistical analysis for springback compensation in sheet metal forming XIII International Conference on Computational Plasticity. Fundamentals and Applications COMPLAS XIII

More information

Recent advances in Metamodel of Optimal Prognosis. Lectures. Thomas Most & Johannes Will

Recent advances in Metamodel of Optimal Prognosis. Lectures. Thomas Most & Johannes Will Lectures Recent advances in Metamodel of Optimal Prognosis Thomas Most & Johannes Will presented at the Weimar Optimization and Stochastic Days 2010 Source: www.dynardo.de/en/library Recent advances in

More information

FE ANALYSES OF STABILITY OF SINGLE AND DOUBLE CORRUGATED BOARDS

FE ANALYSES OF STABILITY OF SINGLE AND DOUBLE CORRUGATED BOARDS Proceedings of ICAD26 FE ANALYSES OF STABILITY OF SINGLE AND DOUBLE CORRUGATED BOARDS ICAD-26-43 Enrico Armentani enrico.armentani@unina.it University of Naples P.le V. Tecchio, 8 8125 Naples Italy Francesco

More information

Revised Sheet Metal Simulation, J.E. Akin, Rice University

Revised Sheet Metal Simulation, J.E. Akin, Rice University Revised Sheet Metal Simulation, J.E. Akin, Rice University A SolidWorks simulation tutorial is just intended to illustrate where to find various icons that you would need in a real engineering analysis.

More information

AXIAL OF OF THE. M. W. Hyer. To mitigate the. Virginia. SUMMARY. the buckling. circumference, Because of their. could.

AXIAL OF OF THE. M. W. Hyer. To mitigate the. Virginia. SUMMARY. the buckling. circumference, Because of their. could. IMPROVEMENT OF THE AXIAL BUCKLING CAPACITY OF COMPOSITE ELLIPTICAL CYLINDRICAL SHELLS M. W. Hyer Department of Engineering Science and Mechanics (0219) Virginia Polytechnic Institute and State University

More information

f xx (x, y) = 6 + 6x f xy (x, y) = 0 f yy (x, y) = y In general, the quantity that we re interested in is

f xx (x, y) = 6 + 6x f xy (x, y) = 0 f yy (x, y) = y In general, the quantity that we re interested in is 1. Let f(x, y) = 5 + 3x 2 + 3y 2 + 2y 3 + x 3. (a) Final all critical points of f. (b) Use the second derivatives test to classify the critical points you found in (a) as a local maximum, local minimum,

More information

Design optimization of C Frame of Hydraulic Press Machine

Design optimization of C Frame of Hydraulic Press Machine IOSR Journal of Computer Engineering (IOSR-JCE) e-issn: 2278-0661,p-ISSN: 2278-8727 PP 79-89 www.iosrjournals.org Design optimization of C Frame of Hydraulic Press Machine Ameet B. Hatapakki 1, U D. Gulhane

More information

3 Nonlinear Regression

3 Nonlinear Regression CSC 4 / CSC D / CSC C 3 Sometimes linear models are not sufficient to capture the real-world phenomena, and thus nonlinear models are necessary. In regression, all such models will have the same basic

More information

Linear Methods for Regression and Shrinkage Methods

Linear Methods for Regression and Shrinkage Methods Linear Methods for Regression and Shrinkage Methods Reference: The Elements of Statistical Learning, by T. Hastie, R. Tibshirani, J. Friedman, Springer 1 Linear Regression Models Least Squares Input vectors

More information

Finite Element Analysis Prof. Dr. B. N. Rao Department of Civil Engineering Indian Institute of Technology, Madras. Lecture - 36

Finite Element Analysis Prof. Dr. B. N. Rao Department of Civil Engineering Indian Institute of Technology, Madras. Lecture - 36 Finite Element Analysis Prof. Dr. B. N. Rao Department of Civil Engineering Indian Institute of Technology, Madras Lecture - 36 In last class, we have derived element equations for two d elasticity problems

More information

PATCH TEST OF HEXAHEDRAL ELEMENT

PATCH TEST OF HEXAHEDRAL ELEMENT Annual Report of ADVENTURE Project ADV-99- (999) PATCH TEST OF HEXAHEDRAL ELEMENT Yoshikazu ISHIHARA * and Hirohisa NOGUCHI * * Mitsubishi Research Institute, Inc. e-mail: y-ishi@mri.co.jp * Department

More information

Abstract. Introduction:

Abstract. Introduction: Abstract This project analyzed a lifecycle test fixture for stress under generic test loading. The maximum stress is expected to occur near the shrink fit pin on the lever arm. The model was constructed

More information

DESIGN OF EXPERIMENTS and ROBUST DESIGN

DESIGN OF EXPERIMENTS and ROBUST DESIGN DESIGN OF EXPERIMENTS and ROBUST DESIGN Problems in design and production environments often require experiments to find a solution. Design of experiments are a collection of statistical methods that,

More information

Faculty of Science and Technology MASTER S THESIS

Faculty of Science and Technology MASTER S THESIS Faculty of Science and Technology MASTER S THESIS Study program/ Specialization: Petroleum Engineering/Reservoir Engineering Writer : Ibnu Hafidz Arief Faculty supervisor: Prof.Dr. Hans Spring semester,

More information

( ) = Y ˆ. Calibration Definition A model is calibrated if its predictions are right on average: ave(response Predicted value) = Predicted value.

( ) = Y ˆ. Calibration Definition A model is calibrated if its predictions are right on average: ave(response Predicted value) = Predicted value. Calibration OVERVIEW... 2 INTRODUCTION... 2 CALIBRATION... 3 ANOTHER REASON FOR CALIBRATION... 4 CHECKING THE CALIBRATION OF A REGRESSION... 5 CALIBRATION IN SIMPLE REGRESSION (DISPLAY.JMP)... 5 TESTING

More information

Quarter Symmetry Tank Stress (Draft 4 Oct 24 06)

Quarter Symmetry Tank Stress (Draft 4 Oct 24 06) Quarter Symmetry Tank Stress (Draft 4 Oct 24 06) Introduction You need to carry out the stress analysis of an outdoor water tank. Since it has quarter symmetry you start by building only one-fourth of

More information

Themes in the Texas CCRS - Mathematics

Themes in the Texas CCRS - Mathematics 1. Compare real numbers. a. Classify numbers as natural, whole, integers, rational, irrational, real, imaginary, &/or complex. b. Use and apply the relative magnitude of real numbers by using inequality

More information

Through Process Modelling of Self-Piercing Riveting

Through Process Modelling of Self-Piercing Riveting 8 th International LS-DYNA User Conference Metal Forming (2) Through Process Modelling of Self-Piercing Riveting Porcaro, R. 1, Hanssen, A.G. 1,2, Langseth, M. 1, Aalberg, A. 1 1 Structural Impact Laboratory

More information

Curriki Geometry Glossary

Curriki Geometry Glossary Curriki Geometry Glossary The following terms are used throughout the Curriki Geometry projects and represent the core vocabulary and concepts that students should know to meet Common Core State Standards.

More information

CCSSM Curriculum Analysis Project Tool 1 Interpreting Functions in Grades 9-12

CCSSM Curriculum Analysis Project Tool 1 Interpreting Functions in Grades 9-12 Tool 1: Standards for Mathematical ent: Interpreting Functions CCSSM Curriculum Analysis Project Tool 1 Interpreting Functions in Grades 9-12 Name of Reviewer School/District Date Name of Curriculum Materials:

More information

Machine Learning for Signal Processing Lecture 4: Optimization

Machine Learning for Signal Processing Lecture 4: Optimization Machine Learning for Signal Processing Lecture 4: Optimization 13 Sep 2015 Instructor: Bhiksha Raj (slides largely by Najim Dehak, JHU) 11-755/18-797 1 Index 1. The problem of optimization 2. Direct optimization

More information

Automation of Static and Dynamic FEA Analysis of Bottomhole Assemblies

Automation of Static and Dynamic FEA Analysis of Bottomhole Assemblies Automation of Static and Dynamic FEA Analysis of Bottomhole Assemblies Nader E. Abedrabbo, Lev Ring & Raju Gandikota 1 Weatherford International 11909 Spencer Road, Houston, TX nader.abedrabbo@weatherford.com

More information

ALGEBRA II A CURRICULUM OUTLINE

ALGEBRA II A CURRICULUM OUTLINE ALGEBRA II A CURRICULUM OUTLINE 2013-2014 OVERVIEW: 1. Linear Equations and Inequalities 2. Polynomial Expressions and Equations 3. Rational Expressions and Equations 4. Radical Expressions and Equations

More information

Support Vector Machines

Support Vector Machines Support Vector Machines SVM Discussion Overview. Importance of SVMs. Overview of Mathematical Techniques Employed 3. Margin Geometry 4. SVM Training Methodology 5. Overlapping Distributions 6. Dealing

More information

TOPOLOGY OPTIMIZATION OF ELASTOMER DAMPING DEVICES FOR STRUCTURAL VIBRATION REDUCTION

TOPOLOGY OPTIMIZATION OF ELASTOMER DAMPING DEVICES FOR STRUCTURAL VIBRATION REDUCTION 6th European Conference on Computational Mechanics (ECCM 6) 7th European Conference on Computational Fluid Dynamics (ECFD 7) 1115 June 2018, Glasgow, UK TOPOLOGY OPTIMIZATION OF ELASTOMER DAMPING DEVICES

More information

Voluntary State Curriculum Algebra II

Voluntary State Curriculum Algebra II Algebra II Goal 1: Integration into Broader Knowledge The student will develop, analyze, communicate, and apply models to real-world situations using the language of mathematics and appropriate technology.

More information

Appendix A: Mesh Nonlinear Adaptivity. ANSYS Mechanical Introduction to Structural Nonlinearities

Appendix A: Mesh Nonlinear Adaptivity. ANSYS Mechanical Introduction to Structural Nonlinearities Appendix A: Mesh Nonlinear Adaptivity 16.0 Release ANSYS Mechanical Introduction to Structural Nonlinearities 1 2015 ANSYS, Inc. Mesh Nonlinear Adaptivity Introduction to Mesh Nonlinear Adaptivity Understanding

More information

7 Fractions. Number Sense and Numeration Measurement Geometry and Spatial Sense Patterning and Algebra Data Management and Probability

7 Fractions. Number Sense and Numeration Measurement Geometry and Spatial Sense Patterning and Algebra Data Management and Probability 7 Fractions GRADE 7 FRACTIONS continue to develop proficiency by using fractions in mental strategies and in selecting and justifying use; develop proficiency in adding and subtracting simple fractions;

More information

TWO-DIMENSIONAL PROBLEM OF THE THEORY OF ELASTICITY. INVESTIGATION OF STRESS CONCENTRATION FACTORS.

TWO-DIMENSIONAL PROBLEM OF THE THEORY OF ELASTICITY. INVESTIGATION OF STRESS CONCENTRATION FACTORS. Ex_1_2D Plate.doc 1 TWO-DIMENSIONAL PROBLEM OF THE THEORY OF ELASTICITY. INVESTIGATION OF STRESS CONCENTRATION FACTORS. 1. INTRODUCTION Two-dimensional problem of the theory of elasticity is a particular

More information

This blog addresses the question: how do we determine the intersection of two circles in the Cartesian plane?

This blog addresses the question: how do we determine the intersection of two circles in the Cartesian plane? Intersecting Circles This blog addresses the question: how do we determine the intersection of two circles in the Cartesian plane? This is a problem that a programmer might have to solve, for example,

More information

CHAPTER 6 EXPERIMENTAL AND FINITE ELEMENT SIMULATION STUDIES OF SUPERPLASTIC BOX FORMING

CHAPTER 6 EXPERIMENTAL AND FINITE ELEMENT SIMULATION STUDIES OF SUPERPLASTIC BOX FORMING 113 CHAPTER 6 EXPERIMENTAL AND FINITE ELEMENT SIMULATION STUDIES OF SUPERPLASTIC BOX FORMING 6.1 INTRODUCTION Superplastic properties are exhibited only under a narrow range of strain rates. Hence, it

More information

3 Nonlinear Regression

3 Nonlinear Regression 3 Linear models are often insufficient to capture the real-world phenomena. That is, the relation between the inputs and the outputs we want to be able to predict are not linear. As a consequence, nonlinear

More information

2: Static analysis of a plate

2: Static analysis of a plate 2: Static analysis of a plate Topics covered Project description Using SolidWorks Simulation interface Linear static analysis with solid elements Finding reaction forces Controlling discretization errors

More information

13.1. Functions of Several Variables. Introduction to Functions of Several Variables. Functions of Several Variables. Objectives. Example 1 Solution

13.1. Functions of Several Variables. Introduction to Functions of Several Variables. Functions of Several Variables. Objectives. Example 1 Solution 13 Functions of Several Variables 13.1 Introduction to Functions of Several Variables Copyright Cengage Learning. All rights reserved. Copyright Cengage Learning. All rights reserved. Objectives Understand

More information

KEYWORDS Morphing, CAE workflow, Optimization, Automation, DOE, Regression, CFD, FEM, Python

KEYWORDS Morphing, CAE workflow, Optimization, Automation, DOE, Regression, CFD, FEM, Python DESIGN OPTIMIZATION WITH ANSA MORPH 1 Tobias Eidevåg *, 1 David Tarazona Ramos *, 1 Mohammad El-Alti 1 Alten AB, Sweden KEYWORDS Morphing, CAE workflow, Optimization, Automation, DOE, Regression, CFD,

More information

Webinar Parameter Identification with optislang. Dynardo GmbH

Webinar Parameter Identification with optislang. Dynardo GmbH Webinar Parameter Identification with optislang Dynardo GmbH 1 Outline Theoretical background Process Integration Sensitivity analysis Least squares minimization Example: Identification of material parameters

More information

Integrated Mathematics I Performance Level Descriptors

Integrated Mathematics I Performance Level Descriptors Limited A student performing at the Limited Level demonstrates a minimal command of Ohio s Learning Standards for Integrated Mathematics I. A student at this level has an emerging ability to demonstrate

More information

Mathematics for chemical engineers

Mathematics for chemical engineers Mathematics for chemical engineers Drahoslava Janovská Department of mathematics Winter semester 2013-2014 Numerical solution of ordinary differential equations Initial value problem Outline 1 Introduction

More information

SETTLEMENT OF A CIRCULAR FOOTING ON SAND

SETTLEMENT OF A CIRCULAR FOOTING ON SAND 1 SETTLEMENT OF A CIRCULAR FOOTING ON SAND In this chapter a first application is considered, namely the settlement of a circular foundation footing on sand. This is the first step in becoming familiar

More information

Parameter Estimation in Differential Equations: A Numerical Study of Shooting Methods

Parameter Estimation in Differential Equations: A Numerical Study of Shooting Methods Parameter Estimation in Differential Equations: A Numerical Study of Shooting Methods Franz Hamilton Faculty Advisor: Dr Timothy Sauer January 5, 2011 Abstract Differential equation modeling is central

More information

Lecture 8. Divided Differences,Least-Squares Approximations. Ceng375 Numerical Computations at December 9, 2010

Lecture 8. Divided Differences,Least-Squares Approximations. Ceng375 Numerical Computations at December 9, 2010 Lecture 8, Ceng375 Numerical Computations at December 9, 2010 Computer Engineering Department Çankaya University 8.1 Contents 1 2 3 8.2 : These provide a more efficient way to construct an interpolating

More information

Programming, numerics and optimization

Programming, numerics and optimization Programming, numerics and optimization Lecture C-4: Constrained optimization Łukasz Jankowski ljank@ippt.pan.pl Institute of Fundamental Technological Research Room 4.32, Phone +22.8261281 ext. 428 June

More information

Optimizations and Lagrange Multiplier Method

Optimizations and Lagrange Multiplier Method Introduction Applications Goal and Objectives Reflection Questions Once an objective of any real world application is well specified as a function of its control variables, which may subject to a certain

More information

Driven Cavity Example

Driven Cavity Example BMAppendixI.qxd 11/14/12 6:55 PM Page I-1 I CFD Driven Cavity Example I.1 Problem One of the classic benchmarks in CFD is the driven cavity problem. Consider steady, incompressible, viscous flow in a square

More information

Calibrating an Overhead Video Camera

Calibrating an Overhead Video Camera Calibrating an Overhead Video Camera Raul Rojas Freie Universität Berlin, Takustraße 9, 495 Berlin, Germany http://www.fu-fighters.de Abstract. In this section we discuss how to calibrate an overhead video

More information

Parameterization of triangular meshes

Parameterization of triangular meshes Parameterization of triangular meshes Michael S. Floater November 10, 2009 Triangular meshes are often used to represent surfaces, at least initially, one reason being that meshes are relatively easy to

More information

Some Aspects for the Simulation of a Non-Linear Problem with Plasticity and Contact

Some Aspects for the Simulation of a Non-Linear Problem with Plasticity and Contact Some Aspects for the Simulation of a Non-Linear Problem with Plasticity and Contact Eduardo Luís Gaertner Marcos Giovani Dropa de Bortoli EMBRACO S.A. Abstract A linear elastic model is often not appropriate

More information

Support Vector Machines

Support Vector Machines Support Vector Machines . Importance of SVM SVM is a discriminative method that brings together:. computational learning theory. previously known methods in linear discriminant functions 3. optimization

More information

Multi-Disciplinary Design of an Aircraft Landing Gear with Altair HyperWorks

Multi-Disciplinary Design of an Aircraft Landing Gear with Altair HyperWorks Multi-Disciplinary Design of an Aircraft Landing Gear with Altair HyperWorks Altair Engineering, October 2008 Introduction Task: To design an aircraft landing gear that meets design requirements of several

More information

Approximation Methods in Optimization

Approximation Methods in Optimization Approximation Methods in Optimization The basic idea is that if you have a function that is noisy and possibly expensive to evaluate, then that function can be sampled at a few points and a fit of it created.

More information

ANSYS User s Group Non-Linear Adaptive Meshing (NLAD)

ANSYS User s Group Non-Linear Adaptive Meshing (NLAD) 19.2 Release ANSYS User s Group Non-Linear Adaptive Meshing (NLAD) Sriraghav Sridharan Application Engineer, ANSYS Inc Sriraghav.Sridharan@ansys.com 1 2017 ANSYS, Inc. October 10, 2018 Topics Background

More information

Predicting Tumour Location by Modelling the Deformation of the Breast using Nonlinear Elasticity

Predicting Tumour Location by Modelling the Deformation of the Breast using Nonlinear Elasticity Predicting Tumour Location by Modelling the Deformation of the Breast using Nonlinear Elasticity November 8th, 2006 Outline Motivation Motivation Motivation for Modelling Breast Deformation Mesh Generation

More information

Recent Developments in the Design and Optimization of Constant Force Electrical Contacts

Recent Developments in the Design and Optimization of Constant Force Electrical Contacts Recent Developments in the Design and Optimization of Constant orce Electrical Contacts John C. Meaders * Stephen P. Harston Christopher A. Mattson Brigham Young University Provo, UT, 84602, USA Abstract

More information

3-D Numerical Simulation of Direct Aluminum Extrusion and Die Deformation

3-D Numerical Simulation of Direct Aluminum Extrusion and Die Deformation 3-D Numerical Simulation of Direct Aluminum Extrusion and Die Deformation ABSTRACT W.A.Assaad, University of Twente Enschede, The Netherlands H.J.M. Geijselaers, University of Twente Enschede, The Netherlands

More information

Auto-Extraction of Modelica Code from Finite Element Analysis or Measurement Data

Auto-Extraction of Modelica Code from Finite Element Analysis or Measurement Data Auto-Extraction of Modelica Code from Finite Element Analysis or Measurement Data The-Quan Pham, Alfred Kamusella, Holger Neubert OptiY ek, Aschaffenburg Germany, e-mail: pham@optiyeu Technische Universität

More information

Chapter 7 Practical Considerations in Modeling. Chapter 7 Practical Considerations in Modeling

Chapter 7 Practical Considerations in Modeling. Chapter 7 Practical Considerations in Modeling CIVL 7/8117 1/43 Chapter 7 Learning Objectives To present concepts that should be considered when modeling for a situation by the finite element method, such as aspect ratio, symmetry, natural subdivisions,

More information

Support Vector Machines.

Support Vector Machines. Support Vector Machines srihari@buffalo.edu SVM Discussion Overview. Importance of SVMs. Overview of Mathematical Techniques Employed 3. Margin Geometry 4. SVM Training Methodology 5. Overlapping Distributions

More information

MEI Desmos Tasks for AS Pure

MEI Desmos Tasks for AS Pure Task 1: Coordinate Geometry Intersection of a line and a curve 1. Add a quadratic curve, e.g. y = x² 4x + 1 2. Add a line, e.g. y = x 3 3. Select the points of intersection of the line and the curve. What

More information