برنامه ریزی عملیات به کمک کامپیوتر برای خمش ورق های فلزی: حالت هنر

Purpose of this paper is to offer the reader an overview of recently performed and ongoing research related to process planning for sheet metal bending, thus providing a starting point for further exploration of this field. The scope of this review paper is limited to sheet metal bending as performed on numerically controlled press brakes, with special focus on air bending. Automatic process planning requires a good understanding of the material behaviour under process conditions. Therefore, some space has been reserved for an overview of bend modelling efforts and, directly linked to this, in-process measurement and adaptive control methods. Part representation and feature classification methods for bent sheet metal parts are also discussed. Sections are dedicated to the core problems of fully automated process planning in sheet metal bending: bend sequencing, collision detection, tolerance verification and tool selection. The state of the art review is completed with an overview of ergonomic analysis methods for process plan evaluation.

Sheet metal forming is one of the oldest manufacturing processes known to mankind [1], and bending can probably be considered its most basic variant. However, the numerous research contributions dedicated to sheet metal bending that have been published over the past decade, and the constant stream of announcements by R&D departments of machine constructors form strong indications that not all research challenges related to sheet metal bending have been exhausted. Purpose of this paper is to offer the reader an overview of recently performed and ongoing research related to process planning for sheet metal bending, thus providing a starting point for further exploration of this field. Sheet metal parts are typically produced by a sequence of bending operations. The bending process starts with a flat workpiece and ends up with a three-dimensional object of interconnected flanges. The bending operations are executed on bending machines – so-called press brakes – using various tools and holding devices (see Fig. 1). Tools consist of dies and punches of different shape and length. Usually, a machine can hold several tools at the same time, while a tool can be applied for different bends, too. Tool selection and operation sequencing is based, first of all, on geometric considerations so as to avoid interferences between the workpiece, the tool and the machine. Furthermore, the planner has to consider a number of issues concerning material properties, tolerances, ergonomics and cost factors. Full-size image (21 K) Fig. 1. The bending process and its resources. Figure options Main phases of the process of generating as well as executing process plans in sheet metal bending are shown in Fig. 2: definition of the planning problem, automatic plan generation and plan execution. Given a model of the part – including its dimensions, tolerances and material properties – and the description of the bending machine and tools, process planning is aimed at generating an executable sequence of bending operations, together with appropriate part set-up orientations, tools, gauging locations and punch displacements. At planning time, however, due to various simplifying assumptions, one can but approximate the final shape and tolerances of the part. Hence, adaptive processes are applied so as to manufacture what really has been required in the part design. This is the reason why physical bend models play a crucial role in each main phase of plan generation and execution. Full-size image (52 K) Fig. 2. Overview of sheet metal bending related research topics and their interrelationship, with indication of the corresponding section numbers. Figure options Details of automatic process planning are shown in Fig. 3. Planning is a “wicked” problem because its tasks mutually interact: there is no evident ordering of the decisions, and partial solutions cannot be simply combined into a final one. In the chosen overview structure, heart of the planning process is the sequencing of the bending operations. However, bend sequences have to be measured also in terms of hard reject criteria set by gauging location, collision detection and tolerance verification. On the other hand, feasible sequences can be evaluated and compared taking cost-efficiency of the tool set-up as well as ergonomic considerations into account. Full-size image (50 K) Fig. 3. Overview of research topics related to automatic process planning for bending and their interrelationships, with indication of the corresponding section numbers. Figure options This state of the art review is limited to sheet metal bending as performed on numerically controlled press brakes, with special focus on air bending that involves bending of parts on a V-shaped die with a punch. Swivel bending and wiper bending are explicitly excluded from the scope of this review. A detailed enumeration of the specific methods or algorithms described by the respective authors would be impractical within the scope of a single article. However, an attempt was made to summarise the most important results and to point out the relevance of the respective contributions. In order not to overload the text, joint works are referred to by means of the first author only. The research efforts, as described in this literature review, can largely be divided into two major categories. The first category concerns research related to the modelling of material behaviour under bending conditions and related measuring methods (Sections 2, 3 and 4). While the main focus of this review article is on process planning, it was considered useful also to cover these closely related topics which form the boundary conditions for the actual process planning methods and algorithms. A brief state of the art is sketched for bend modelling (Section 2) and on-line material testing (Section 3). In Section 4, the recent evolution in the field of adaptive control systems for bending is summarised. The second category contains contributions to the development of specific automatic process planning methods. In order to provide well structured input data for the described procedures, a number of authors have focused their efforts on part representation and classification: a summary is offered in Section 5. Bend sequencing methods are reported in Section 6, while the next three sections are dedicated to collision detection, tolerance verification and tool selection. Additionally, a number of research reports related to ergonomic aspects of process planning (Section 10) are mentioned. These publications contain more pragmatic contributions, which can also be implemented as stand alone software tools for industrial use. The interrelationships between the respective research topics covered in this paper are indicated in Fig. 2 and Fig. 3, where reference is made to the corresponding section numbers.

As can be concluded from this literature review, a variety of tasks, involved in process planning for sheet metal bending, span a rather broad and diverse research area. Significant contributions to the solution of a number of specific sheet metal bending-related problems have been reported over the last decade. Bend modelling and, complementary to this, the (in-process) measurement of material properties and adaptive control strategies received wide attention from both academic and industrial side. Bend sequencing has been intensively investigated in recent years. Some automatic sequence generators, capable of handling parts with a high complexity, have been reported. Although some of the underlying research projects are still ongoing, the reported results are already indicating a mature state of development in the covered subdomains. The results of these research efforts are becoming visible in a number of commercially available software solutions. Some of these packages come close to the objective of a fully automated process planning system for bent sheet metal parts. However, in order to seamlessly bridge the gap between design specification and production, the authors envision further research in areas like tolerance compliance verification, optimised automatic tool selection and fully automated in-process identification of material property models. For tolerance verification, fast methods, compatible with the practical time constraints of process planning, need to be developed in order to integrate a realistic, stochastic accuracy check in the bend sequencing procedures. Not only does verification need further attention, appropriate tolerance compliance oriented rules need to additionally be derived to efficiently steer bend sequencing. Target of these enhanced rule bases should be the identification of feasible and well-optimised bend sequences with a guaranteed accuracy level under polynomial time complexity conditions. For automatic tool selection crossing, the chasm between process and production planning forms an objective. Process planning has typically been regarded as an activity focused on a single production task. Well-optimised methods for tool selection and machine set-up generation should, however, aim for a global optimisation of the overall production performance indicators (total cost, total productivity and throughput time). Besides time and motion considerations linked to individual batches, the changeover costs between consecutive production tasks need to be taken into account in this respect. As for material modelling, for the time being, the vision of bending undocumented materials with unknown exact thicknesses has not yet been achieved. To reach full adaptive control, complementary work related to in-process modelling of the material behaviour is required. Only when a mature state of development is reached in real-time, in-process measurement of the relevant material parameters, process planning with approximate punch displacements can be continuously linked to the actual bending operations. The research objectives described here are not intended as an exhaustive list: they form but some of the challenges for further research, to which the authors hope to have contributed by facilitating an efficient access to the state of the art.