自行車腳蹬內(nèi)板沖壓工藝分析與模具設(shè)計(jì)、級(jí)進(jìn)模含13張CAD圖
自行車腳蹬內(nèi)板沖壓工藝分析與模具設(shè)計(jì)、級(jí)進(jìn)模含13張CAD圖,自行車,腳蹬,沖壓,工藝,分析,模具設(shè)計(jì),級(jí)進(jìn)模含,13,cad
1. Introduction
The function of the cooling system of a plastic injection mould is to provide thermal regulation in the injection moulding process. When the hot plastic melt enters into the mould impression, it cools down and solidi?es by dissipating heat through the cooling system. As the cooling phase generally accounts for about two-thirds of the total cycle time of the injection moulding process, ef?cient cooling is very important to the productivity of the process.The cooling system also plays an important role in the product quality. A cooling system that provides uniform cooling across the entire part ensures product quality by preventing differential shrinkage, internal stresses, and mould release problems. In addition to the functional aspects, the design of a cooling system should also consider the manufacturability of the system to control the cost of mould construction. The process of cooling system design is a complicated process and can be distinguished into three phases: preliminary design, layout design, and detail design. Although, CAD/CAM systems are widely used in the design of injection moulds, they are mainly limited to providing geometric modeling tools in the detail design phase. Specialized stand-alone or add-on software packages that provide interactive geometric modeling tools for designing various components or sub-systems of the mould structure are also commercially available. However, limited research works on automation tools that can play a more active role in the preliminary and layout design phases have been reported. In a previous research project, we developed a feature-based method which creates the preliminary design automatically [1,2]. Given a plastic part with a complex shape, the feature-based method decomposes the part into simpler shape features, called cooling features. Cooling sub-circuits are then generated automatically to provide the required cooling function for each recognized feature. In the present research, automation in the design process is extended to the layout design phase. Techniques are developed which generate the layout design automatically from the preliminary design by considering both the functional and manufacturing aspects of the cooling system
2. Related work
There are four major areas of research related to plastic injection mould cooling system, namely, computer-aided engineering (CAE) analysis, design optimization, new fabrication technology, and automatic design synthesis. Most of the early research work [3–7] focused on CAE analysis. After more than two decades of extensive research, commercial CAE packages such as MOLDFLOW and Moldex3D are now widely used in practice to analyze a given design. These CAE methods predict the temperature pro?le that changes with cooling time. Ef?ciency and quality can thus be estimated before the actual mold fabrication.While CAE methods are able to analyze a given design, they do not suggest design changes when problems are detected from the results of an analysis. Optimization methods [8–10] are reported which utilize the CAE analysis results to optimize a design. Given an initial cooling con?guration design, an objective function is formulated as a measure of the temperature uniformity and cooling ef?ciency. The objective function is expressed in terms of parameters relating to the con?guration of the cooling system and processing conditions. By integrating an optimization algorithm with a cooling analysis algorithm, the initial design can be ?ne-tuned to optimize the coolin system design
Recently, methods that build better cooling systems by using new fabrication technology have been reported. Instead of the conventional hole-drilling method to produce straight-line channels, Sachs et al. [11,12] reported a method that takes the advantage of solid freeform fabrication technology to produce conformal cooling channels. Such channels maintain a constant distance from the mould impression, so that accurate temperature control is possible, even for a part with a complex shape. It has been reported that a more uniform temperature distribution and better dimensional control of the moulded part can thereby be achieved. Sun et al. [13] used CNC milling to produce U-shape milled grooves for cooling channels. This technique is similar to the conformal cooling method, in that the channels are able to follow the shape of the mould impression. As with the conformal cooling method, an improvement in temperature control has also been reported
The focus of our work is in the automatic design synthesis of the cooling system, which is an area that has not been well investigated. In our previous work [1,2], a feature-based technique for automatic preliminary design generation was developed. The work reported in this paper concentrates on the automation of the layout design. The automatic layout design process is formulated as a heuristic search process. Design automation by heuristic search is a commonly used technique, and has been investigated by the author in the design automation of mechanical devices [14,15]. The heuristics that guide the layout design process are based on fuzzy evaluation of the cooling performance and manufacturability of candidate designs. Automatic manufacturability analysis in various application domains has been studied extensively, and Grupta et al. [16] reported a comprehensive survey. The evaluation method developed in this research was inspired by the method reported by Ong and Chew [17] in the manufacturability evaluation of machined parts and setup plans. The use of fuzzy logic in the evaluation was inspired by the successful applications of fuzzy logic in various aspects of injection mould design research, including parting direction determination [18] and mouldability analysis [19].
3. Overview of the method
In the preliminary design stage, the major issue to be addressed is the functional requirement, that is, the cooling requirement, of a given plastic part. The preliminary design speci?es the type (e.g. U-circuit, parallel channels, bubblers, cooling towers, etc.), size (e.g. channel length and diameter), and the approximate locations of the cooling elements that form the cooling sub-circuits. Each sub-circuit provides the cooling function that carries away the heat from a region of the part. Our previous research [1,2] shows that the preliminary design can be determined mainly from the geometric shape of the part, and a feature recognition approach for automating the design was developed. Given the preliminary design, complete cooling circuits are developed in the layout design phase by connecting the individual sub-circuits and cooling elements together. The layout design process considers the manufacturability and the feasibility of the physical realization of the cooling circuits.
In the current research, the investigation is focused on cooling systems that do not use parallel cooling and have only one cooling circuit. The major items to be addressed include: the location of the main cooling elements, the channels that interconnect the cooling elements and the sub-circuits, and the locations of the inlet and outlet of the cooling system. The design of each item is inter-related and the layout design process is divided into four major stages of operation。
1. The derivation of a graph structure that represents the preliminary design and modi?cation of the graph to facilitate subsequent operation.
2. The generation of candidate cooling circuits from thegraph structure.
3. The generation of layout designs from the candidate cooling circuits by developing tentative manufacturing plans.
4. The evaluation of the candidate layout designs with respect to cooling performance and manufacturability.
The ?rst stage of the design process is a preparation step, which serves to derive a representation of the preliminary design in a form that facilitates the subsequent operation. Starting from the second stage, a control structure is employed in the layout design process to control the transition of the process from one stage to the next, and the backtracking from a later stage to a previous stage. The design process is formulated as a searching process, whereby at each stage a candidate design is selected for processing. A search tree is devised to represent the design process, as illustrated in Fig. 1. Each node in the search tree represents either an operation or an output from an operation. As a selection at one stage may eventually lead to a dead-end at a later stage, backtracking is important in the design process. For example, one candidate cooling circuit may eventually lead to a layout design that has a very low rating in cooling performance. To search for an alternative solution, the design process backtracks to the second stage to investigate (i) an alternative cooling circuit from the same inlet; or (ii) an alternative inlet and generate a new cooling circuit, and then repeats the subsequent stages to develop an alternative design. Fig. 2(a) shows the preliminary design of the cooling system of an example part. For illustration purposes, only the core half of the mould is shown. Fig. 2(b) shows the ?nal layout design generated by the design process. Fig. 2(c) shows two of the alternatives generated in the second stage of the design process
4. Graph representation and operation on the preliminary design
Given a preliminary cooling system design in the form of a set of sub-circuits consisting of various type of cooling elements, a basic problem in generating the layout design is to identity appropriate connections within the sub-circuits (e.g. the interconnections between a set of parallel channels) and between adjacent sub-circuits (e.g. the connection between two U-circuits in two adjacent layers) so that they can be connected to form a complete cooling circuit. A graph-based technique is used to solve this problem, and this includes three major steps.
1. Devise a graph representation of the preliminary design.
2. Add extra nodes and edges to the graph so as to represent the various possible connections between and within the sub-circuits.
3. Employ a speci?c graph traversal method to ?nd paths that correspond to candidate cooling circuits.
Steps 1 and 2 are described in the following sub-sections, and step 3 will be described in the next section.
4.1. Graph representation
The graph that represents the preliminary design is initially constructed with a set of disjointed sub-graphs, such that each sub-circuit speci?ed in the preliminary design is represented by a sub-graph. Each edge in the sub-graph represents a cooling channel and each node represents an inlet, outlet, or a connection point between adjacent channels. The edges in the graph are labeled with one of the attributes: L-edge, C-edge, or X-edge. L-edge is used for edges that represent straight-line cooling channels (L-channel). C-edge is used for edges that represent cooling channels with complex shapes (C-Channel). X-edge is used for edges that represent channels, which serve to interconnect sub-circuits or cooling elements within a sub-circuit (X-channel). Their cooling effects, if any, are not considered and are not speci?ed in the preliminary design. An X-edge is created only during the layout design process. The cooling elements in each sub-circuit are represented as follows: for an individual cooling channel, two nodes connected by an L-edge are used; for a planar cooling element (such as aU-circuit orV-channel), a simple path consisting of an alternate sequence of nodes and L-edges is used; for an element with an inlet and outlet that are geometrically close to each other (such as bubblers and baf?es), or an element with a complex 3Dcooling channel (such as a cooling tower), two nodes connected by a C-edge are used.
4.2. Graph modi?cation
A cooling system that does not use parallel cooling and has only one cooling circuit corresponds to a simple path in the proposed graph representation. To search the graph for an appropriate simple path that connects all or most of the cooling elements speci?ed in the preliminary design, the graph should be a connected graph. However, the graph initially constructed to represent the preliminary design is not connected, because there is no connection between the sub-graphs that correspond to the individual sub-circuits, and a sub-circuit itself may not be represented by a connected component in the graph. Therefore, the initial graph has to be modi?ed by a set of operations so that the graph becomes connected.
4.2.1. Modi?cation of L-graphs
Those sub-graphs that contain only L-edges, which have the channels represented by these L-edges lying on the same plane, are labeled as L-graphs. Typical cooling elements captured by L-graphs include U-circuits, V-channels, and sets of parallel cooling channels lying on the same plane. Each L-graph is investigated and modi?ed, if necessary, to form a connected sub-graph with at least one cycle. The modi?cation is achieved by adding additional channels that connect the end points of the existing channels. Fig. 3(a) shows the resulting graph structure derived from a simple example of preliminary design given in Fig. 2(a). Notice that the graph structure consists of two connected sub-graphs, and that each sub-graph is composed of L-edges which represent the cooling elements speci?ed in the preliminary design, and X-edges which are added during graph modi?cation.
4.2.2. Merging of sub-graphs
The purpose of merging the sub-graphs is to connect the individual sub-graphs to form a connected graph. As each sub-graph represents a sub-circuit speci?ed in the preliminary design, the merging process should (i) avoid any major modi?cations in the sub-graphs; and (ii) use the shortest possible X-edge whenever an X-edge is needed to establish a connection. These ensure that the merging process does not cause signi?cation deviations from the preliminary design that is represented in the individual sub-graphs. The merging process consists of two stages. In the ?rst stage, only L-graphs are considered. Consider two L-graphs LGi and LGj lying on the planes PLi and PLj. They are merged if either (i) PLi and PLj are parallel and the geometric distance between them is less than a threshold value; or (ii) PLi and PLj are on the same plane, and there exists a pair of parallel channels, one from LGi and the other from LGj such that the geometric distance between them is less than a threshold value. In the former case, LGi and LGj are merged into a single sub-graph by adding nodes and edges according to the method described below. The corresponding channels represented in the original L-graphs are translated to the plane that is at equal distance from PLi and PLj. In the latter case, LGi and LGj are merged by replacing the pair of channels (and thus the corresponding edges in the sub-graphs) by a single channel that lies between the two channel
In the second stage, all sub-graphs are considered. Any two sub-graphs are merged if the distance between them is the smallest. The distance between two sub-graphs Gi and Gj is de?ned as the smallest distance among the distances between an edge of Gi and an edge of Gj. The distance between edge ei of Gi and edge ej of Gj is de?ned as the shortest geometric distance between the channels that are represented by ei and ej. The merging is executed iteratively until all sub-graphs are merged into a single graph.
To merge two sub-graphs Gi and Gj (not necessary L-graphs), an X-edge (and possibly an extra node) is added; this corresponds to adding a channel that connects the two sub-circuits represented by Gi and Gj. As the cooling effect of this channel has not been considered in the preliminary design stage, the shortest channel should be used to avoid any signi?cant changes in the cooling effect of the system. Therefore, the two end points of this new channel, one on each sub-circuit, should be placed at the locations where the minimum distance between Gi and Gj is obtained. If the minimum distance occurs between two parallel channels, two new channels (and thus two X-edges) are constructed. At least one of the end points of each channel is at an
existing end point in either sub-circuit. If the same minimum distance occurs between multiple pairs of parallel channels, X-edges are added to each pair by the same
method.
The connected graph G that is obtained from the merging process represents the various ways by which the cooling elements and sub-circuits can be connected. The connected graph for the preliminary design in Fig. 2(a) is shown in Fig. 3(b).
5. Generation of candidate cooling circuits
To ?nd the candidate cooling circuits from the connected graph G obtained from the previous step, an inlet is ?rst selected among the nodes in G. Heuristics can be devised for this selection according to the speci?c requirement of a particular moulding condition. A simple heuristic that corresponds to a common practice in cooling system design is to select the node that is closest to the gate position. Starting from the selected inlet node, a set of simple paths is obtained from G by traversing G using a speci?c recursive depth-?rst search algorithm. The search results in a search tree that captures all the simple paths that correspond to the candidate cooling circuits with the selected inlet.
A requirement for the layout design process is that the cooling circuit generated should connect as many cooling elements as possible. This requires that the simple paths found should be as long as possible. The notion of maximal path is thus introduced. Given a node n0 in G, a path that starts at n0 is maximal if it terminates at a node nN such that all nodes adjacent to nN are included in the path. In other words, a maximal path cannot be extended further in length without creating a loop. Fig. 4(a) shows a graph and Fig. 4(b) and (c) shows two of the maximal paths extracted from the graph that starts from node 2. A depth-?rst search algorithm is developed to retrieve all maximal paths from a given node. Given a graph G and a start node n0, the path searching algorithm generates a search tree T. For each maximal path in G that starts at n0, the searching algorithm constructs a corresponding path from the root to a leaf node in T. Fig. 4(d) shows the search tree generated from the graph shown in Fig. 4(a). The path highlighted in Fig. 4(d) corresponds to the maximal path in Fig. 4(c). The details of the algorithm are described below.
5.1. The path searching algorithm
Initially, a search tree T is constructed with a root that corresponds to the given start node n0. The searching algorithm is then invoked with graph G and start node n0. The searching algorithm ?rst retrieves all of the nodes ni that are adjacent to the input node n0 in G.If n0 does not have an adjacent node, then the searching algorithm terminates. Otherwise, the adjacent nodes ni are stored to the tree T as children of n0. The graph G is then copied to a new graph G1, and all of the edges n0–ni (i.e. the edges between n0 and all of its adjacent nodes ni) are removed from G1. The searching algorithm is then invoked recursively with each ni and a new graph G1. The recursive searching algorithm eventually terminates with a tree T containing all of the
maximal paths in G that start from n0. The pseudo-code below further illustrates the algorithm.
Algorithm Path-Search(G, n0)
Input: G—the input graph
n0—the input node
Global variable: T—the search tree
Begin
Retrieve nodes ni that are adjacent to n0 in G;
If no node is retrieved, Go to End;
Else
{
Store each ni to T as children of n0;
G1ZG;
Removed all edges n0–ni in G1;
For each ni
Path-Search(G1,ni);
}
End
5.2. Proof of the algorithm
It is important to prove that the search tree T generated by Path-Search(G, n0) actually contains all of the maximal paths that can be retrieved fro
收藏