A CAD / CAE Integrated Optimization of Hot Runner System

Hot runner technology has been widely applied in the plastic injection mold industry. However, overheating of plastic melt inside the manifold plate will cause various defects in plastic parts. Nowadays, the design of hot runner systems is still mainly depended on the designers’ experiences and a time-consuming trial and error process is inevitable. An automatic optimization framework of hot runner systems is proposed in this paper. It integrates CAD/CAE/Optimization software to find an optimum design of manifolds. A skeleton model, which is a parameter repository that contains geometry and analysis parameter, is the core of this framework. It can effectively interoperate and propagate change between CAD model and FEA model. The ISIGHT platform adopted in this framework provides an efficient way to find the optimum layout of heaters.


Introduction
A Hot-runner system is used in plastic injection molds to transfer and inject molten plastic into the cavities of the mold.It is usually composed of a heated manifold plate and several heated nozzles.The plastic melt runs in the runner inside the manifold plate at a high temperature, around 240°C.Compared with traditional cold runner systems, hot runners can reduce plastic waste and shorten the cycle time because the plastic melt in the runner is not ejected with the plastic part when the mold is opened and therefore it is not necessary to wait until the runner freezes.Furthermore, hot runners make the design more flexible because they can carry the plastic melt to many places without worrying about that the plastic melt will be cooled down in the runners and cause the problems of under-filling.
Although hot-runner molds offer so many advantages as mentioned above, they also bring various defects such as dark spots, flow marks and gate marks if the temperature of the plastic melt is not well controlled [1].In engineering practice, manifold plates are usually heated by coiled tubular heaters to keep the plastic melt in the runners at a stable temperature.Hence, an optimum layout of heaters is required to achieve thermal homogeneity, which means uniform temperatures throughout the entire manifold plate.With the help of CAE, designers are now able to catch the potential issues on their designs and revise their designs before manufacturing.But a tedious loop of model-evaluate-remodel is still needed to find a satisfactory solution.
Efforts have been devoted to the automation and optimization of the injection mold design for the past two decades.Dimla et al. [2] have constructed a virtual model using I-DEAS and Moldflow to optimize cooling channels positioning.In his method, the candidate cooling channels layout plans were manually set and the modification of CAE and CAD model were also done by manual.Li et al. [3] used a graph traversal algorithm to generate possible candidate cooling circuits and a heuristic search algorithm was employed to find a satisfactory design.Ivan et al. [4,5] used genetic algorithms to optimize the family mold layout design, which is generally considered a demanding and experience-dependent task.The abovementioned works showed that the two most important issues in design optimizations are the optimization algorithms and the efficiency of interoperations between CAD and CAE.
The optimization of the injection mold design has no extra constrains on the selection of optimization algorithms.Besides heuristic search algorithm [3] and genetic algorithm [4][5][6] used in the previous systems, other algorithms such as Taguchi method, back-propagation neural networks [7] and sequential approximation optimization algorithm [8] can also be used.Several reported design optimization systems [8,9] were developed based on the commercial optimization platform ISIGHT.Because ISIGHT embeds a comprehensive selection of the most popular optimizers, it can be applied to a variety of design optimization problems and greatly shorten the development cycle.
The efficiency of interoperations between CAD and CAE model is another important issue to be considered in design optimization because the CAD and CAE need to transfer data in every iteration.In order to improve efficiency, various intermediate models are designed to integrate CAD and CAE data.Gujarathi et al. [10] proposed a CAD/CAE integration method using a common data model (CDM), which serves to record the driving design parameters and key constraints.Zhiyi Pan et al. [11] proposed a modeling mechanism named as CADWE (Computer-Aided Design-While-Engineering),which merges the two application models into a single pattern and can be manipulated synchronously.Byoung-Keon Park et al. [12] presented a sharable format, Practical post-Analysis Model(PAM), which allows the efficient sharing of FEA data in a collaborative design process.Other scholars [13][14][15][16] proposed different intermediate models.
A CAD/CAE integrated optimal framework for a hot runner system design is proposed in this paper.The framework is based on the ISIGHT optimization platform and a skeleton model for CAD/CAE integration.

Overview of the framework
Conventionally, CAD models only contain geometric information and do not consider the information for analysis, such as load conditions, boundary conditions, key constrains et al.Therefore, it cannot meet the needs of interoperations and change propagation between CAD model and FEA.Furthermore, although the CAD and analysis models represent the same object, it is time-consuming to prepare a qualified CAE model from the CAD geometry.To meet the requirements on the efficiency of CAD/ CAE data exchange in optimization computation, a CAD/CAE integrated optimization framework is proposed.
As shown in Figure 1, the framework includes the following components: -The skeleton model is a collection of design semantic parameters required to build geometry model, finite element model and to conduct engineering analysis.-A CAD model can be generated based on the geometric information of the skeleton model.It has more details than the skeleton model and is a complete 3D model.-Analysis process description (APD) file is automatically generated by a userdeveloped routine, which transforms the geometry and analysis information in the skeleton model to an ASCII text file according with the specification of ANSYS parametric design language (APDL).-A CAE support environment is used to parse the APD file, construct geometry, mesh, create boundary conditions and then solve the problem.-An optimization tool will evaluate the simulation results.If the results meet the requirements, the whole process will be completed.Otherwise, it will adjust the parameters in the skeleton model and another cycle begins until the goal is achieved.

Skeleton Model
• Parametric geometry

Proposed design process for hot runner design
A typical hot runner manifold is composed of a plate and several coiled tubular heater are embedded into it.The heaters will heat up the plate and keep the plastic melted inside the manifold at a stable temperature.The optimization of hot runner system is to find a suitable layout of heaters that makes the temperature distribution through the manifold plate as uniform as possible.
The skeleton model of a hot runner contains parametric geometry information and FEA information.The parametric geometry information includes the sketch of the plate outline, the thickness of the plate, the section dimensions, size and position of the heaters.As show in Figure 2, the size of a heater is expressed by the length L and the width W. The position point of a heater is the center of the rectangle.Analysis information in a skeleton model includes material properties, meshing information, and boundary conditions.For example, the material properties of the plate are: the material density (7800 kg/m 3 ), the coefficient of thermal conductivity (70 W/m•°C) and the specific heat (448 J/kg•°C).Meshing information includes element type, meshing method and so on.The boundary conditions for thermal equilibrium analysis include the heat transfer coefficient, the heat generation rate and the air temperature.
A detailed sketch for a manifold can be constructed from the skeleton model by adding some extra features, such as the fillets and bulges.Then by using the preset parameters including the thickness of the manifold and the section dimension of heater, a 3D manifold part can be generated as shown in Figure 2.This 3D CAD model is same with traditional feature-based CAD model.
The APD file is automatically created from the skeleton model by a user-developed routine.An APD file mainly includes a geometry section and a analysis section.All of the geometries for analysis are represented in the geometry section.For this case, the outline sketch of the manifold is expressed as curves and the heaters are expressed as parametric blocks.Take the middle heater as an example, the APD description is as follows: As shown in Figure 1, ANSYS will then import the generated APD file and run simulation.Guided by ISIGHT, an optimum solution will be found after a large number of iteration.In each iteration, the size, position and heating rate of the heaters in the APD file is within the searching space.Then these optimized parameters are used to drive the skeleton model and the CAD models.
The above optimization process is governed by ISIGHT.ISIGHT supports many optimization algorithms, including Evol (Evolutionary Optimization), MIGA (Multi-Island Genetic Algorithm), ASA (Adaptive Simulated Annealing).The Optimization interface, as shown in Figure 3, allows the user to specify design variables, constraints, and objectives in ISIGHT.In this paper, Evol algorithm was selected because it is suitable for non-linear and discontinuous problems.
The Optimization model is defined as following: The design variables of this problem are the heater size parameters (L1,L2,W1,W2) and the heat generation (P), which control heat flow rate per unit volume.The Object function: f(X) is the temperature variance calculated from some selected nodes along the runners.During the optimization process, the design variables are adjusted to obtain a minimum temperature variance.

An example of optimization
Take the manifold plate in Figure 2 as an example.The initial and optimized values of design variables are listed in Table 1.These initial values are decided according to experience.An initial APD file is created from user skeleton model.Then the design variables, constraints, objective function and the APD file are input into ISGHIT.4 shows the temperature distribution in the manifold before and after optimization.The temperature range before optimization is from 249 to 264°C,while this range is narrowed to from 242 to 244.Finally, the optimized parameters can be used to drive the skeleton model.The 3D detailed design of Manifold can also be updated.

Conclusion
In this paper, a framework of optimum design and its application for hot runner system are proposed.In this framework, a skeleton model acts as bridge connecting CAD, CAE and Optimization software.The design cycle can be considerably shortened because the parametric skeleton model has much higher efficiency in exchange data with CAE model than the traditional CAD model.The automatic optimization process is helpful for hot runner mold designers to quickly find a satisfactory layout of heaters.

Figure 2 .
Figure 2. Design process with skeleton model

Figure
Figure4shows the temperature distribution in the manifold before and after optimization.The temperature range before optimization is from 249 to 264°C,while this range is narrowed to from 242 to 244.Finally, the optimized parameters can be used to drive the skeleton model.The 3D detailed design of Manifold can also be updated.

Figure 4 .
Figure 4.The temperature distribution in the manifold before and after optimization

Table 1 .
Optimization results