What does multiphysics mean

Parametric multiphysics simulation

Transcript

1 TECHNICAL REPORT Parametric multiphysics-multiphysics improve the understanding of the complex physical processes in several physical domains. As an example, the calculation of the current carrying capacity of an electrical connector taking into account the non-linear electrical-thermal-mechanical behavior is shown. The typical workflow through a sensitivity analysis and optimization of the design by means of statistical test planning (DoE Design of Experiments) up to the evaluation of safety is shown. This approach delivers detailed statements about the achievable results and limits of the design as well as possibilities of optimization. Figure 1 A multiphysics in Ansys Workbench including electromagnetic, thermal and mechanical 1 Introduction The more complex physical phenomena require more and more a consideration of the interaction between physical domains such as mechanics, temperature, fluid flow and electromagnetism. Taking into account the interactions between different domains plays an important role in obtaining reliable results and making the right decisions in virtual product development. Nowadays, s-software enables the coupling of different physical domains as a so-called multiphysics simulation for a deeper understanding of products. In order to use multiphysics as a driving factor for s-controlled product development, a second prerequisite is important: a fully parametric workflow should cover the entire process of CAD import, geometry preparation, material modeling, networking, boundary conditions, solution and result for everyone covered physical domains and all levels of detail. This parametric workflow enables sensitivity studies for a better understanding of the relationship between design parameters and results, the relationship between individual results and the meaning of certain design parameters. In addition, the optimization can be used to obtain the best design given various restrictions and competing objectives. In order to test the safety of a product under real production and usage conditions, the variation in the relevant input parameters such as material and geometric tolerances can be taken into account. This allows a reliable prediction of the failure probabilities under different load conditions. Author Dipl.-Ing. (FH) Christof Gebhardt, Cadfem Contact: Cadfem GmbH Marktplatz Grafing b. Munich Tel .: / Construction July / August 7 /

2 TECHNICAL REPORT Fig. 5 Electrical conductivity as a function of the contact pressure Fig. 3 Interaction of physical effects Fig. 4 Temperature-dependent plasticity Fig. 6 Validation result for the implemented contact property 2 Multiphysics of a connector A typical application example for multiphysics are electrical connectors that have high requirements in industrial applications have to meet. This is why high reliability, signal integrity, and optimal electrical, thermal and mechanical properties are the focus of the design of a connector. Figure 2 shows a connector for high current densities, which consists of a spring sleeve that is inserted into a casing. The current flow is transmitted via the connection through the contact between the pin and the sleeve and between the sleeve and the jacket. Figure 7 Multi-physical contact definition in Ansys Workbench with pressure-dependent properties Multi-physical field coupling The current carrying capacity of a connector is one of the most important properties for a predefined service life. Construction July / August 7 / 8-2014

3 SPECIALIST REPORT Fig. 8 must be guaranteed. Three important physical domains are involved in the current carrying capacity: 1. The electrical behavior of the connector with its resistance value. 2. Due to the flow of current and the electrical resistance, heat is generated in the copper material and on the contact pairs, which leads to a rise in temperature at the connector. 3. The rise in temperature affects the conductivity of the copper and its mechanical rigidity and shifts the yield point of the connector material. 4. The changed mechanical material properties influence the contact zone and the contact pressure, which leads to varying current paths and changed electrical and thermal contact resistances. This makes it clear that electrical, thermal, and mechanical domains necessarily interact in this system. A separate consideration of the individual domains could not take this interaction into account. For example, the temperature-dependent resistance or the temperature-dependent plasticity requires results from a thermal. The temperature dependence of the material properties is defined by the curves in Figure 4, which allow the solver to interpolate between the values. 2.2 Contact model For connector systems, the modeling of the contact situation is the key for a precise modeling of the energy transfer from one part to the other. The contact pressure in structural mechanics affects the electrical and thermal conductivity. This fact is approximated by a contact pressure-dependent electrical (Fig. 5) and thermal contact conductivity according to the square root law (according to R. Holm). The Ansys Parametric Design Language (APDL) was used to implement these specific structural-mechanical-thermal-electrical contact properties. This script language has control structures such as subroutines or loops and enables the user to implement their own models and functionalities in a simple manner and with a special focus on CAE applications. After the implementation of the newly defined pressure-dependent electrical and thermal contact conductivity, the sm model was first validated in order to ensure the correct functionality.Construction July / August 7 / network with copper sleeve (orange) between outer jacket and connector (gray) Workbench. One of the results calculated for validation is shown in Figure 6: As the force increases, the heat generation in the connector is reduced. The next step took place with the tested contact modeling script: the transfer of these functions to the workbench environment by ACT, the Ansys Customization Toolkit. This enables users to change and customize the standard Ansys Workbench environment for specific applications. The main advantages are the extended, application-related and extensively automated customer-specific functionalities, which support the user with a faster definition of his task and easier handling, so that even less experienced users can use the system. In addition, other software packages can be integrated, for example if a post-processing step is to be added upstream of the workbench process. 2.3 Model preparation The uses so-called multifield elements (Ansys 22x), which have all degrees of freedom in one element (Fig. 7). This results in a uniform network as shown in Figure 8. The network shows the initial penetration between the sleeve and the connector, which is eliminated in the first load step, with the plug connection closed, and thus depicts the assembly status. This method can be used when the friction during the mating process does not affect the relevant results. In a second step, the current flow is activated. For the electrical domain, the earth potential is specified on one end and the applied voltage or the flowing current on the opposite side. The assumption here is that radiation and convection do not play a decisive role in cooling the connector. Therefore, the heat conduction in the cable is modeled by a convective boundary condition with a high heat transfer rate. 65

4 SPECIALIST REPORT Fig. 11 Fig. 10 Iterative solution process of the non-linear multiphysics connector model Plastic expansion and electrical potential with voltage drop at the contact point due to contact resistance the voltage drop in the connector with the local current paths as well as the Joule heating and the resulting temperature distribution. The results help to get a better understanding of the physical behavior. Deformation, stresses, plastic strain, current density, rate of heat generation and voltage drop, temperature and heat flow show the engineer how the original design behaves. Due to the complex interaction of the design parameters, the number of effects, the positive, negative or lacking influence of the various parameters on the design, design optimization is a demanding task. 3 Sensitivity, optimization and robustness A sensitivity study with a subsequent optimization and robustness analysis was carried out in order to improve the design by minimizing the voltage drop and the insertion and extraction forces without exceeding the temperature of 115 C. Therefore, in the first step, the entire model was parameterized. This includes the geometry parameters that come from the CAD system, as well as parameters such as electrical resistance, yield strength, modulus of elasticity and other parameters. For each of these parameters, the variant space was defined by upper and lower limit values. With the optislang for Ansys software, the variant space was scanned by stochastic sampling with 100 designs based on an extended Latin hypercube method, which avoids an accumulation of design points in one area of ​​the design space. As the first result of the sensitivity study, the correlation matrix shows the input variables and their effects on selected results (Figure 13). It can be seen that the design parameters 2 to 6 are unimportant because they have no real influence on the results (correlation ~ 0). Parameter 1 (overlap of adapter sleeve and connector), parameter 7 (thickness of adapter sleeve), parameter 8 (angle) and parameter 9 (length of adapter sleeve) have a direct influence on the results (~ 1 or -1), which means that these are the important parameters. To achieve an optimal design with minimal voltage drop, minimal plug forces at a temperature of less than 115 C 66 construction July / August 7 / 8-2014

5 SPECIALIST REPORT, these values ​​are defined as goals and constraints. When performing the optimization, the optimum is achieved in 210 design steps within 90 seconds. The speed of this optimization results from the use of a meta model based on the sensitivities. This metamodel is a mathematical representation of the relationship between the design variables and the calculated results, as shown schematically in Figure 14 as a blue area. No further FEM analysis was performed during the optimization. Instead, the metamodel was used to find the optimal design. The essential success factor for such an approach is the quality of the metamodel itself. OptiSLang uses a unique technology called MoP (Metamodel of Optimal Prediction (mopcop.html), which provides the user with information about the prediction quality. The metamodel of optimal prediction is the most important tool in order to calculate only the minimum number of designs taking into account the required forecasting capability and thus to ensure maximum efficiency. The user has the option of checking the specifications and adapting the forecast quality to an individual relationship between speed and accuracy. The fact that two optimization goals are defined results in not just a single optimization result, but a series of Optima with the proposal for a specific design (here design no. 96). This set of Optima is defined by the so-called Pareto front. This front shows all designs with optimal fulfillment of the defined requirements. Each design of this front is one of several that meet both goals. A further improvement of one objective can only be achieved by worsening the other objective. This design decision can only be made by the developer himself, as he can best judge the goals. The verification of the optimum by an FE confirms the result of the optimization on the basis of the metamodel. Compared to the original design, the optimization results in the following improvements: Reduced force: Improvement of 44% (up to 75% if increased voltage drop is assumed) Reduced voltage drop: Improvement of 4% (up to 9% if higher force is tolerated Fig. 13 Correlation matrix, calculated by the sensitivity analysis As the last step in this analysis, the scatter of the design parameters was analyzed in order to ensure a high level of robustness under real operating conditions with varying loads and scatter in geometry and material properties (Fig. 15). The spread of the parameters was defined by stochastic information (e.g. normal distribution, mean value, standard deviation). For the variation in the thickness of the adapter sleeve as the most important design parameter, Figure 16 shows the relationship between the Figure 14 results (y) for the variants of the two design parameters (X 1, X 2) and the generated metamodel (blue) Figure 15 Definition of the variation of the individual Parameters for the robustness evaluation Scattering in the thickness and the critical parameter temperature. Scatter around the simulated result can also be seen. It can be seen that 0.6% of all designs exceed a temperature of 115 ° C. 4 Summary Analyzes with Ansys Multiphysics enable a better understanding of complex physical phenomena. Physicists and In- continued on page 74 Construction July / August 7 /

6 PROFESSIONAL REVIEW Mechatronics Continued from page 67 Fig. 16 The robustness analysis shows that 0.6% of all designs exceed 115 C in the physical domain (neglecting the interaction) and can hardly be assessed when concentrating on a single design. Statistical test planning (DoE), sensitivity analysis and optimization help developers to work more efficiently due to the powerful representation of the relationships and to take all relevant relationships into account. In addition, the robustness assessment provides valuable information on how reliable a product will be, taking into account variations in geometry, material, loads and other boundary conditions. Engineers use multiphysics to expand their experience and knowledge. Even for very specific requirements, extended s setups (extensions) can be implemented, which also allow sporadic users such as designers to use internal company knowledge, implemented through adapted workflows. In addition, the philosophy of Ansys as a parametric s-tool, together with optislang, is to enable a systematic variation for all physical sub-areas and all associated work steps. Especially in complex situations, with many physical effects and interactions, design changes are to be made taking into account only one literature [1] Hanke, M .: Parametric for System for Ansys, RDO Journal, [2] Will, J .: optislang tem 01 / 2013 Understanding, Ansys [3] Gebhardt, C .: Practice book FEM with An Converence & 31st CAD-FEM Users Meeting, sys Workbench, Carl 2013 Hanser Verlag, construction July / August 7 / 8-2014