Elsevier

Computers in Industry

Volume 58, Issues 8–9, December 2007, Pages 823-831
Computers in Industry

A concurrency control model for PDM systems

https://doi.org/10.1016/j.compind.2007.02.008Get rights and content

Abstract

In today's manufacturing environment, enterprises having work groups geographically dispersed are not uncommon. A product data management (PDM) system is therefore required for controlling the distribution and maintaining the integrity of the product data throughout its entire lifecycle; the efficiency of a PDM system is greatly affected by the concurrency control method it adopts. The paper proposes a concurrent control model for PDM that can also caters for version management and product architecture. The paper discusses how granularity and versioning are being embedded into a lock-based concurrency control model. The concurrent accessibility of an example product data is explained to illustrate the adjustability according to the actions taken by the users and the architecture of the corresponding entities.

Introduction

Product data management (PDM) systems have emerged over the last decade. These systems enable enterprises to conduct its business activities in a more efficient way via ingenious management of product information [1], [2]. The lack of communication among different product development stages often causes consistency problems in product lifecycle. The concept of concurrent engineering, integrated product and process development and others are introduced [3]. They are accompanied with the PDM system to manage all product-related data and provide data retrieval for product design and production. PDM system enhances collaborative work by online access and electronic interchange of product data [4]. The efficiency and quality of design and manufacturing processes can be greatly improved by product information sharing and visualization in the system. With the advent of the Internet- and web-based technologies, PDM systems can now be executed more effectively and efficiently. The development of web-based PDM system is essential for supporting collaborative design and manufacturing at geographically disperse sites [5], [6], [7]. The web-based PDM system not only facilitate the process of data exchanging and sharing but also increase the number of product data transaction.

Among the functions of the PDM system, concurrency control is essential to the checkout, release, obsolete, view, redlining and references. In order to provide the control over the data access, most PDM systems establish a set of access rules that determine what data can be accessed, in what mode and at what point in the product lifecycle. This is complicated when a part or drawing belongs to a certain subassembly. In such instance, the modification to the upper assembly can cause the part to be locked out for write access until the process is completed. Therefore, a concurrency control should be specifically designed with the consideration to the characteristics of PDM systems.

Design and manufacturing workflow can be streamlined by implementing suitable PDM system that manages all product-related data in an organized manner. One major function of PDM system is to maintain data integrity and to provide accurate data when required [8]. For example, building an assembly of a printer cartridge carriage (Fig. 1) involves the co-operation of many engineers from several departments. However, the large amount of interactions between the system and the users will give rise to concurrency problems, the lost update problem occurs when modifications are made in an uncontrollable manner is illustrated in Fig. 2. Let us denote Ti as the unit time i when a transaction T occurs, user A copies the assembly file of the carriage to his/her local workspace from the vault in T1. User A places the lid to the carriage, the content of the local copy of file J in user A's workspace is updated to J*. While user B copies file J to his/her workspace in T2, file J is replaced by user A's local file with content J* and user B inserts the cartridge to the carriage, hence the content of user B in his/her workspace is updated to J** in T3. Finally, file J in the vault, which has been previously updated by user A, is replaced by user B's local file with content J** in T4. Rather than updating the file based on user A's result, B is updating the file based on file J's original value; the work of user A (file J with content J*) is unintentionally erased by the later update. The carriage assembly file in the vault now has three cartridges and without the lid installed. To avoid inconsistency in the vault, concurrency control mechanism is required to prevent database updates performed by one user from interfering with retrievals and updates performed by another [6].

Concurrency control is well studied in traditional database management systems (DBMSs). However, there are relatively few studies that address the issues in PDM systems. In this paper, a new concurrency model is proposed to cater for version management and product architecture in addition to concurrent information sharing required by distributed PDM system.

Section snippets

Concurrency control

In this section, three techniques to concurrency control in traditional DBMSs will be reviewed. They are: two-phase locking [9], [10], timestamp ordering [10], [11], [12] and flow graph locking [13], [14].

Formal description of the model

The new concurrency control model should also integrate granularity locking for distributed PDM. For the former, the model considers each assembly be divided into different levels of subassembly. The subassembly consists of parts and documentations. The model also assumes that when a user updates an assembly, he/she checks out the lowest level subassembly rather than checking out a large assembly with many subassemblies in the writing process. However, this assumption is not applicable in the

Case study

The proposed concurrency control model has four classes namely, entity, project, part and assembly. The attributes of class entity include file name, description, redlining, r_version, version and other information of the document as shown in Fig. 9. Attribute redlining stores the visibility of redlining of the entity, attribute r_version stores the version of redlining and attribute version stores the version of the entity. Class part and assembly are children of entity; they inherit the

Conclusions

In product developments, product data are mainly composed of the information of assemblies, sub-assemblies and parts, which are often managed in distributed computing environments. PDM systems are often utilized for this purpose. When modifying the design of a product, a number of independent tasks may be performed to different components of products. However, locking all corresponding files and their descendants with one single lock limits the concurrency of a PDM system. The efficiency of a

Acknowledgements

The work described in this paper was fully supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. PolyU 5263/04E). Special thanks also to Viana Leung and Angie Lee for their contributions to this research work.

Edmond Chan is a PhD candidate in the Department of Industrial and Systems Engineering at The Hong Kong Polytechnic University. His current research interests include concurrent engineering and optimization of product data management system.

References (25)

  • A.A. Farrag et al.

    Towards a general concurrency control algorithm for database systems

    IEEE Transactions on Software Engineering

    (1987)
  • O.K.V. Li

    Performance models of timestamp ordering concurrency control algorithms in distributed databases

    IEEE Transactions on Computers

    (1987)
  • Cited by (21)

    • A meta-model for knowledge configuration management to support collaborative engineering

      2015, Computers in Industry
      Citation Excerpt :

      Several tools exist to manage product information or data in a design process. These tools are aimed at collaborative document management and sharing to ensure integrity of the product data used in a project [5]. Two types of tools are used in industry today.

    • Migrating product structure bill of materials Excel files to STEP PDM implementation

      2014, International Journal of Information Management
      Citation Excerpt :

      Product data management (PDM) has become one of the most important considerations for companies, especially in engineering and manufacturing industries. Design and manufacturing processes benefit significantly from improved PD sharing, and the usage and visualization possibilities offered by PDM (Chan and Yu, 2007; Kindrick, 2000). PDM stems from traditional engineering design activities that created product drawings and schematics on paper and using CAD tools to create parts lists (bill of material structures: BOM).

    • Concurrent material flow analysis by P3R-driven modeling and simulation in PLM

      2012, Computers in Industry
      Citation Excerpt :

      Liu and Xu reviewed various kinds of web-based PLM system such as SmarTeam®, Windchill and Teamcenter Engineering, with the analysis of system architecture [11]. Chan and Yu provided a new concurrency model for concurrent information sharing which is required by distributed PDM systems [12]. Also, Gunpinar and Han proposed the information sharing methodology for heterogeneous PDM system by using PLM-Services [13].

    View all citing articles on Scopus

    Edmond Chan is a PhD candidate in the Department of Industrial and Systems Engineering at The Hong Kong Polytechnic University. His current research interests include concurrent engineering and optimization of product data management system.

    Kai-Ming Yu is associate professor in the Department of Industrial and Systems Engineering at The Hong Kong Polytechnic University. He is also a member of the Society of Manufacturing Engineers. His investigations are related to CAD/CAM, CAE, PDM, reverse engineering and rapid prototyping technologies.

    View full text