Knowledge reuse in manufacturability analysis
Introduction
Substantial research effort has been pursued in knowledge-based systems to support manufacturability analysis in design [1]. These typically require the extensive capture and representation of knowledge related to both the product being designed and its manufacturing environment [2]. This is a time consuming and expensive undertaking, and so knowledge reuse is an important consideration when modelling enterprises.
A number of reference models and methodologies for modelling generic forms of enterprises are discussed in the research literature, e.g., RM-ODP [3], CIMOSA [4], and CommonKADS [5]. These describe a series of modelling stages that invariably include some form of information view, which structures knowledge and information according to a pre-defined class hierarchy.
Various hierarchies for product design and manufacturing environments have also been proposed, e.g., MOKA [6]; and “product and manufacturing models” [7].
Molina and Bell's manufacturing model describes an enterprise as an aggregation of facilities, i.e., stations, cells, shops, and factories while facilities are described in terms of resources performing processes, under the control of manufacturing strategies. An extension of this approach by Young et al. [8] is shown in Fig. 1 which extends the concept from representing an individual facility to the representation of enterprises. It also builds the model using a UML representation. Facilities representations cannot be constructed independently of product representations as they are used to manufacture products. One route to linking facility and product representations is illustrated in Fig. 2, building on the work of Zhao et al. [9].
More recent work on manufacturing facility representation has focussed on issues such as knowledge maintenance [10], knowledge sharing using ontologies [11], and representing global supply chains [12]. These more recent developments do not, however, provide application guidelines on how to maximise the reuse of knowledge classified using the proposed hierarchies.1 This paper provides these guidelines, and demonstrates their application through an industrial case example, based on the simplified jet engine combustor casing shown in Fig. 3.
The casing is manufactured by forging a series of metallic rings which are then welded together to form the rough shape of the component. The chamber is then machined, using a variety of turning, milling, and drilling processes to create the final object (and achieve the required tolerances).
This paper identifies three principles of knowledge reuse that can be applied to the representation of the chamber and its manufacturing environment. Firstly, knowledge (in the form of rules and constraints) needs to be separated from information; Secondly, knowledge needs to be classified according to the separate product and manufacturing hierarchies shown in Fig. 2.
Finally, different layers of manufacturing strategies (describing how resources perform processes) need to be applied. The following sections explain these principles in more detail.
Section snippets
Knowledge versus information
Before examining why knowledge and information should be classified separately it is worth describing what is actually meant by these two terms. Information exists when the relationships between data (i.e., numbers and symbols) are recognised within a specific context. In a geometric context for example, “5 cm from A” is recognised as a distance [13]. Knowledge; on the other hand, is information with added detail relating how it should be used or applied [14]. Knowledge may therefore include
Product/process separation
Separating product from process information is relatively straightforward (e.g., product characteristics such as diameters are clearly part of the product model). Knowledge classification may not, however, be quite so clear cut, and it is important to make the correct distinctions if reuse is to be maximised. Current literature describing manufacturing models provides basic classification hierarchies (see Fig. 1, Fig. 2), but does not provide clear guidelines on how to make this separation. The
Manufacturing strategies
The third reuse principle described by this paper is the classification (or layering) of manufacturing strategies. Strategies describe how resources perform processes, and form a central class in the facility representation described in Fig. 2.
As an example, a simplified component level strategy for the manufacture of part of a chamber ring is illustrated by the process schematic shown in Fig. 5. This illustrates how a range of processes can be brought together to specify a higher level
Experimental platform
The manufacturability analysis platform (MAP; see Fig. 7) has been used to test and evaluate the principles described above. Manufacturing strategies are expressed by as series of shared terms (implemented as Java methods). These support the description of product attributes, processes, and the relationships between feature requirements and processes. Models of standard features (e.g., cylinders and holes) can also be referenced by the shared methods, along with customised models for bespoke
Conclusions
A review of existing literature on the representation of manufacturing knowledge highlighted the need for application guidelines on the classification of knowledge for optimum reuse. This paper provides these guidelines by developing three principles for the improved reusability, i.e., the separation of information from rules and constraints, the separation of product knowledge from manufacturing process knowledge, and the layering of manufacturing strategies. Each of these concepts has been
Acknowledgements
This work is part of an ongoing research project entitled “Knowledge Representation and Reuse for Predictive Design and Manufacturing Evaluation”. This has been funded under EPSRC GR/R64483/01, and actively supported by Rolls Royce plc.
References (15)
- et al.
CIMOSA: enterprise engineering and integration
Comput Ind
(1999) - et al.
Ontology infrastructure for execution-oriented autonomous agents
Rob Auton Syst
(2004) Digital enterprise technology-defining perspectives and research priorities
Int J Comput Integr Manuf
(2003)- Young RIM, Guerra D, Gunendran G, Das B, Cochrane S, Cutting-Decelle A-F. Sharing manufacturing information and...
- ISO/IEC 10746-1. Reference model for open distributed...
- Schreiber G, Akkermans H, Anjewierden A, de Hoog R, Shadbolt N, Van de Velde W, Wielinga B. The CommonKADS methodology,...
- Oldham K, Kneebone S, Callot M, Murton A, Brimble R. in: Mårtensson N, Mackay R, Björgvinsson S. editors. Changing the...
Cited by (27)
Core domain ontology for joining processes to consolidate welding standards
2019, Robotics and Computer-Integrated ManufacturingCitation Excerpt :An Adaptive Holonic Control Architecture for Distributed Manufacturing Systems (ADACOR) ontology was designed in [22] to address the production planning and scheduling without looking into the specific domain of welding. In [23], a guideline was developed to improve the reusability of manufacturing knowledge for decision support systems, but it was also limited to machining processes. Similarly, the product ontology developed in [24] and the manufacturing ontology proposed in [25] were found to be generic as well.
How to reduce the negative impacts of knowledge heterogeneity in engineering design team: Exploring the role of knowledge reuse
2016, International Journal of Project ManagementCitation Excerpt :Markus proposed four distinct styles of knowledge reuse situations according to the knowledge reuser and the purpose of knowledge reuse, which involved shared work producers, shared work practitioners, expertise-seeking novices, and second knowledge miners (Markus, 2001). Cochrane et al. described three principles of reuse, i.e., the separation of information from knowledge, the separation of product knowledge from manufacturing process knowledge, and the correct classification of manufacturing knowledge (Cochrane et al., 2008). The design process reuse was classified into four types: full reuse, reuse after tiny revision, reuse after big revision, and reuse analogously (Yu et al., 2012).
On knowledge reuse for manufacturing systems design and planning: A semantic technology approach
2015, CIRP Journal of Manufacturing Science and TechnologyCitation Excerpt :The model followed an integrated approach including technological and business data, information and knowledge and it envisioned that the model would constitute the basis for the amelioration of CAPP solutions with knowledge management and advanced decision support features. A series of guidelines concerning the classification of manufacturing knowledge for its efficient reuse were proposed in [11]. The advent of the semantic technology has led to the development of knowledge management systems [12] leveraging ontological approaches for the knowledge representation, and reuse.
Recent development of knowledge-based systems, methods and tools for One-of-a-Kind Production
2011, Knowledge-Based SystemsCitation Excerpt :There are many activities in development process of OKP products, i.e., design, process planning, manufacturing, assembly, etc. Many KBSs have been developed in the recent decade to benefit these activities [13,46,30,111,84,55,27,4]. As the design stage involves fuzzy problems and a large amount of product knowledge, it is a crucial stage for the whole PD [54,64,109].