超声磨削装置结构设计【全套含有CAD图纸三维建模】
超声磨削装置结构设计【全套含有CAD图纸三维建模】,全套含有CAD图纸三维建模,超声,磨削,装置,结构设计,全套,含有,CAD,图纸,三维,建模
附录A
Dynamic Simulation of an Injection Molding
Author1, Author2, Author3.
Applied Thermal Engineering, 2016, 8(6):556-568.
Abstract: An integrated design method is discussed which thoroughly considers related parameters of the various subsystems in order to optimize the overall system that mainly consists of opto--mechanical structure CAD, CAE and the integrated information platform PDM. Based on the parameter drive of the virtual main model, the method focuses on the model transformation and data share among different design and analysis steps, and so the concurrent simulation and design optimization are carried out. As an example of application, the integrated design for a large-scale opto-mechanical structure is introduced, including optical design, structure design and analysis, which further validates the advantages of the method. Due to comprehensive consideration of the design and analysis process by CAD and CAE based on PDM, the integrated design well attains the structure optimization with high efficiency.
Keywords: integrated design; CAD/CAE; large-scale structure; optical instrument.
1. Introduction
The analysis of products integrating different technologies, e.g. mechanical, hydraulic and controls systems, becomes more and more feasible with the constant development of simulation software and more performing computer hardware. The combination of specialized software packages is possible and allows the simulation of so-called mechatronic systems. If in the past such tools were mainly used in the aeronautic and automobile industries, they now find their way into more common engineering applications. In this case,the dynamic characteristics of the clamp unit of an injection molding machine from HUSKY is investigated. For this purpose, the finite-element (FE) program ANSYS, the multi-body simulation (MBS) software ADAMS, the fluid power simulation software DSHplus and the controls design tool MATLAB/Simulink are used. Combining these different simulation tools and applying them to the damp unit, we can analyze and understand the dynamic behavior of the machine and the interaction between the different sub-systems. This is necessary to improve the performances, like reducing wear, cycle time or noise, or avoiding premature failure of parts.
The damp unit Is the mechanism that doses and opens the mold and keeps it effectively closed during the injection and the holding-pressure stages. The HUSKY OUADLOC' damp is a two-platen hydraulic damping system. The main components are the moving platen, the clamp base and the stationary platen with the four tie bars. The platen locking and the damping force are realized with the damp pistons, which are integrated in the moving platen. The damp pistons can be rotated by 45' to engage the tie bar teeth in order to lock the moving platen in its position. The main functions of the damp unit are actuated hydraulically;hydraulic pressure is applied to the damp pistons to generate the necessary damping force and two hydraulic cylinders are used for the displacement of the moving platen. [1]
Figure1: HUSKY QUADLOC clamp unit
The analysis focuses on two aspects: first, the quantification of the forces acting on the pads between the damp base and the foundation in order to foresee and prevent any creep of the machine during operation, and second, the optimization of the stroke command signal in order to reduce the overall cycle time. Therefore the simulation model Is limited to the moving platen stroke.
2． Simulation Models
The mechanical, hydraulic and controls systems are modeled in ADAMS, DSHplus and MATLAB/Simulink, respectively. ADAMS gives the possibility to Include non-standard phenomena by linking user-written FORTRAN or C subroutines In the model. DSHplus uses this feature to make a co-simulation between both programs possible. Besides, ADAMS disposes of a plug-In that allows the user to conned the MBS model with Simulink. Thus it is possible to link the three simulation tools and simulate the complete machine. There are two
possibilities for the computation: first, each program Integrates its own set of differential equations and exchanges the necessary parameters with the other ones, second, the three models are completely integrated and only the Slmulink solver Integrates the differential equations set.
Figure2:co-simulation of moving platen stroke
Additionally, the flexibility of mechanical parts can be included in ADAMS. We use ANSYS to generate the necessary FE models and to reduce these large models to a few degrees of freedom before being integrated Into ADAMS. The reduction method Is based on the component mode synthesis technique Introduced by Craig and Bampton.
Mechanical Model
The rgid-body model of the damp unit Is very simple and consists only of two parts: the moving platen and the stationary platen with damp base and tie bars (refer to figure 4). The damp base is fixed with linear spring-damper elements to the ground and the sliding of the moving platen is modeled with contact statements. Basically, the contact statement is a nonlinear spring-damper element where the force is proportional to the penetration depth x and the penetration velocity it :x
k and d are the contact stiffness and damping coefficients, respectively and a is an exponent that for numerical reasons should be chosen greater then 1. If there is no penetration, then no force is applied, otherwise the location of contact, the normals at the points of contact and the force acting between both parts are computed. The statement also includes a Coulomb friction model. The transition from the static to the dynamic friction coefficient Is based on the relative velocity of the two colliding geometries. Finally, the two stroke cylinder forces and a contact statement are defined between both platens. In order to have a more accurate mechanical model, also In view of a more realistic distribution of the forces on the damp-base pads, the different parts are included as flexible bodies. These flexible bodies are derived from FE models that are reduced before being imported. The reduction method Implemented In ADAMS is based on the component mode synthesis (CMS) technique, i.e. the deformation is written as a linear combination of mode shapes. In ADAMS, constraint modes and fixed-boundary normal modes are used to generate the component-made matrix 0. This approach is known as the Craig-Bampton method. In the following, the basic principles and steps of Integrating flexible bodies In ADAMS are shown; a more detailed theoretical presentation can be found in
First of all, a FE model Is created that is detailed enough to comedy represent the mode shapes of interest. The user has then to choose the nodes that serve as interface to the MBS model. The kinematic constraints or forces are applied to these boundary nodes; the remaining nodes are referred to as interior nodes.
By fixing the degrees of freedom (DOF) u, of the boundary nodes and solving an elgenproblem, we get the fixed-bounder, normal modes d>r. This normal mode set is usually truncated. The constraint modes dc are defined as the static deformation of the structure when a unit displacement Is applied to one DOF of a boundary node while the remaining DOF of the boundary nodes are restrained (Guyan reduction). Finally the component mode reduction matrix m Is defined by the normal modes set φand the constraint modes set d>r.
The relationship between the physical displacement coordinates u and the component generalized coordinates q is
With equation (2) the generally large number of physical DOF u is drastically reduced to few mixed physical and modal DOF q.
However, the Craig-Bampton modal basis q has certain disadvantages that make it sometimes difficult to directly use it in a multi-body simulation. The set of constraint modes contains the 6 rigid-body DOF that must be replaced by the large displacement DOF of the local body reference frame in ADAMS. They have to be removed and therefore the component-mode matrix Is transformed by solving the eigenproblem
where K and M are the reduced stiffness and mass matrix, respectively. The manipulation results In a modal basis where q - N4; N containing the elgenvectors from equation (3).
The last step Is a purely mathematical approach and does not further reduce the number of DOF. The new modal basis q has no direct physical meaning anymore but addresses the problems mentioned above
Table 1: First 13 eigenfequencies of the moving platen
The motion of a flexible body is derived from the same equation as for a rigid-body, I.e. Lagrange's equations. In order to calculate the kinetic and potential energy, the position and velocity of an arbitrary point on the flexible body Is expressed with the generalized coordinates.
M the generalized mass matrix depending on,
K the generalized stiffness matrix only depending on q
V the gravitational energy,
D the damping matrix defined using modal damping ratios Is
Q the kinematic constraint equations applied to the flexible body.
Adding flexible bodies to an ADAMS model is quite straightforward. Nevertheless, there are some limitations regarding forces and joints that can be defined to them. Especially the problem of a moving force on a flexible body.
I.e. moving platen sliding on clamp base, is an open Issue in multi-body dynamics. However, there are "standard" workarounds which work well and which have proven their usefulness.
The technique Implemented in our flexible-body model is based on the contact statement mentioned above. Basically it works as follows: for each of the selected nodes along the sliding path, a force is computed according to equation
That depends on the relative vertical position y and velocity y of the node to the moving platen.
However, the force is only activated when the node and the moving platen effectively overlap. In fad, It is weighted by a function that depends on the horizontal distance x between the node and the moving platen.
The force is ramped up from zero or ramped down to zero In order to guarantee a smooth application and to minimize any discontinuities.
No contact points and contact normals are computed. The distance and velocity of a node relative to the moving platen are taken In the global coordinate system and the contact and friction forces are always collinear with the coordinate system unit vectors.
Nevertheless, this approach gives acceptable results, as the deformation of the damp base is very small. A Coulomb friction force is applied in the same way.
Figure 3: moving platen sliding model
The main disadvantage is that a huge number of Interface nodes are needed to have any sound representation of the moving contact forces. Unfortunately, this gives a huge number of flexible-body DOF and therefore unacceptable computation times. Now, instead of defining these nodes as interface nodes, they remain interior nodes. From a purely theoretical point of view, the accuracy of the results is not guaranteed anymore when using interior nodes as interface nodes. However, choosing more normal modes can reduce the error. The comparison of a model using interior nodes for the moving platen sliding with one using interface nodes shows a very 9000 Compliance Of the results While having a much taster computation time. Therefore we used this model for the following simulations. Other methods were not tested but some of them are presented in (3) and (4).
The flexible bodies are created in ANSYS. Simplified CAD geometries of both platens were imported In the FE program and meshed automatically with tetrahedral SOLID187 elements. The damp base was generated 'manually" with SHELL63 elements and the tie bars are modeled with BEAM4 elements. Stationary platen, clamp base and tie bars were put together b one assembly and the different components connected via spring-damper elements (COMBIN14). A macro for the computation of the modified Craig-Brampton basis is available In ANSYS. It allows the user to specify the Interface nodes and the number of normal modes. The resulting modal basis Is written to a file that has to be imported Into ADAMS.
The ANSYS macro automatically selects the six DOF of each interface node as u}. However, it is not imperative to select all the six DOF. This allows us to furthermore reduce the number of static modes and thus the number of flexible-body DOF. The macro has been changed accordingly to select only the effectively required DOF ua Finally, the moving platen has 29 flexible-body DOF and the stationary platen, clamp base and tie bars assembly has 95 flexible-body DOF.
Figure4:ADAMS flexible-body model
附录B
注塑机的动态模拟
Author1, Author2, Author3.
Applied Thermal Engineering, 2016, 8(6):556-568
摘要：讨论了充分考虑相关的参数是一个集成的设计方法为各子系统优化的整体系统，主要由光电—机械结构CAD，CAE与PDM集成信息平台。基于的虚拟模型的参数驱动的方法，侧重于模式的转型的设计和分析的步骤之间的数据共享，所以并行仿的设计进行优化。作为应用实例，综合设计介绍了一种大型光学机械结构，包括光学设计，结构设计与分析，进一步验证了该方法的优点。由于分析了基于PDM和CAD和CAE过程的设计，集成设计达到结构优化效率高。
关键词：集成设计：CAD / CAE：大规模的结构：光学仪器
1. 介绍
产品整体不同技术的分析，例如，机械、液压、控制系统，模拟软件可持续发展和更多计算机硬件操作变得越来越可行。结合特殊的软件程序包是可以和允许对所谓的机电一体化系统模拟的。如果在过去这些工具被应用与航空和汽车工业中，那么他们现在会找到更多应用工程技术的相同点。因此，被研究的是HUSKY注塑机夹紧装置的动态特性。ANSYS有限元（FE）程序，ADAMS多维模拟软件，DSHplus流体动力模拟软件和MATLAB仿真设计控制工具都被用于这个目。结合这些不同的模拟工具并应用它们于合模机构，我们就能分析和理解机器的动态行为和不同子系统之间的联系。这就需要去提高性能，像减少磨损、循环时间或噪音，或者避免部件早期的错误。
合模机构是一个开启与合上模具并能够在注塑和保压阶段有效的紧闭的机构。HUSKY QUADLOCTM合模机构是一个二板式液压合模系统。主要由动模板，合模工作台，定模板和四个拉杆构成。模板锁定和合模压力是通过动模板整合的合模栓塞实现的。合模栓塞可以旋转45度去啮合拉杆螺纹为了保证动模板在注塑位置上。合模机构的主要功能是驱动液压油：液体压力被施加在合模栓塞上去产生需要的合模压力并且两个油缸被用于动模板的移动[1]。
分析的焦点集中在两个方面：第一，压力作用在工作台之间的衬垫上和为了预见的基础并阻止机器在运转中爬行。第二，为了减少总体循环时间的最优化的喷射控制信号。因此，模拟模型被限制移动模版的注射。
2. 模拟模型
机械，液压，控制系统被分别的在ADAMS, DSHplus和MATLAB仿真中模拟。ADAMS可能包括由链接FORTRAN书面使用和C程序在模型中引起的不标准的现象。DSHplus利用这个特点在两个可能的程序上建立了一个co模拟。另外，ADAMS处理插件允许使用者用模拟链接MBS模型。因此，链接三个模拟工具和模拟整个机器是可能的。这里有两个可能估计：第一。各个整合它自己不平衡的部分并和其他的部分交换参数。第二，三个模型被完全的整合并只有模拟求解整合不同的因素部分。
图1 HUSKY QUADLOCTM合模机构
此外，柔性的机械部件可以包括在ADAMS中。我们用ANSYS产生必要的FE模式，并在被纳入ADAMS之前降低这些大型模型自由度。
减少自由度方法是基于引用Craig和bampton的组件式合成技术这是必要的。
图2 模型的联合仿真
2.力学模型
合模机构的刚性模型很简单,只分为两个部分: 移动模板和固定模板和合模工作台还有拉杆（参见图4）。合模工作台是固定在地面上装有线性弹簧-阻尼单元和滑动的动模板构成报表式的模型。基本上，报表式是一个穿透深度X和渗透速率X与压力成正比的非线性弹簧-阻尼单元K和d分别的是接触刚度和阻尼系数， e是一个指数，由于数值原因应选择大于1。如果没有渗透，那么就没有压力作用，否则就会有接触，正常应在接触点和作用力两部分之间计算。该报表还包括了一个库仑摩擦模型。从静态到动态的摩擦系数的变化，是基于相对速度的两个几何碰撞。最终，在两个模板之间定义了两个液压缸的压力和接触报表。为了有一个更准确的力学模型，同时，鉴于更为实际的分配作用在合模工作台上的力，他的不同部分被列为柔性机构。这些柔性机构是来自在被输入之前减少了的FE模式。这种在ADAMS里减少方法的实施是基于组件式频率合成器( CMS )中的技术，即变形是被看作是线性组合模式形状。在ADAMS里，约束模式和固定边界的正常模式是用来生成组件矩阵模式Ф。这种方法被称为Craig-Bampton方法。接下来，在ADAMS中基本的原则和柔性体积分的步骤被展示：更详细的理论介绍可以在[2]式中看到。一个FE模型的建立可以足够详细的正确表达出模型的重要性。操作者这时可以选择充当MBS模型的边界作为节点。被约束的运动和力被施加在这些边界节点上；其余的节点称为内部节点。由固定在边界节点上的自由度（DOF）和求解特征，我们得到固定边界常态模量，这个正常的设定模式通常是截断。约束模量被定义为当其余边界节点自由度都受约束时，一个元件施加在一个边界节点的一个自由度的静态结构的变形。最后组成还原矩阵的模式被解释为常态设定模式和约束设定模式。实际位移坐标u和组成广义坐标q之间的关系是
在方程（2）中，普遍的大量的自由度u被彻底的减少到只有少数混合现实和虚拟的自由度q。然而，Craig-Bampton的基础形态q也有一些毫无疑问的缺点,使有时在多维仿真中难以直接利用它。这套约束模式包含6个刚体自由度，必须取而代之在ADAMS中主体局部的大位移自由度。它们必须被移除，因此在求解方程中，结构模型的矩阵被转化。
其中k和m在矩阵中分别代表降低的刚度和质量。在基础模量上处理的结果是；N包含的特征在方程（3）中被表现。
最后一步,是一个纯粹的数学方法,并没有进一步减少自由度的数量。新的基础模量就没有了直接的物理意义了,而是针对上述的问题。
柔性体的运动模型是与刚性体来自同一方程，即拉格朗日方程。为了计算动能和势能，柔性体上任意点的位置和速度是由广义坐标表示的。
x，y，z，Ψ，θ，Φ，为局部参照系附加弹性体和描述六个刚体模式的坐标。最终形式的运动方程是
其中ξ 为广义坐标，
M 依赖于ξ的广义质量矩阵，
K 只取决于的广义刚度矩阵，
D 阻尼矩阵定义模态阻尼§，因此D是对角线，
Ψ 适用于柔性体运动学约束方程，
λ 拉格朗日乘数，
Q 广义应用的力。
另外，一个柔性体在ADAMS模式下相当的直截了当。不过，也有一些力量和关节限制，我们可以界定给它们。特别是在一个柔性体上摩擦力的问题，即动模板在工作台上滑动，是一个多维动力学的开放性问题。然而，它们制定的标准使得运行好并且被证明有用。
在我们的柔性模型中，技术的执行是基于上述的表诉。基本上它的模式如下:对于每一个沿滑动路径选定的节点，一个力根据作用在动模板上垂直位置和速度的关系的方程（1）进行计算。然而，力只是在当节点和动模板的重叠部分起作用时才作用。事实上，重量的函数是基于节点和动模板之间的水平距离。力从零斜线上升或斜线降低到零是为了保证应用能顺利和尽量减少任何间断。没有接触点和接触平均值计算。节点的距离和速度相对于动模板应采用总的坐标系统和适用于矢量单元坐标的接触和摩擦力。不过，这种做法是能够接受的结果，正如变形的合模基数很小。库仑摩擦力适用于同样的方法。
主要缺点是大量的分界节点都需要有充分代表性的移动接触力。不幸的是，这有太多的柔性体自由度和并且无法计算次数。现在，并不是把这些节点作为边界节点，他们仍是内部节点。从纯理论的角度看，当使用内部节点作为边界节点时结果的准确性无法保证。然而，更多的选择标准模式，可以减少误差。比较采用内节点的模型对对应采用表面节点的滑动模板会有一个非常一致的结果，同时还会加快计算速度。因此，我们使用这一模型进行模拟后，其他方法都没有测试过，但它们中的一些曾在上述中
图3 动模板弹性模型
-12-
收藏