A virtual shop modeling system for industrial fabrication shops

https://doi.org/10.1016/j.simpat.2005.10.012Get rights and content

Abstract

Industrial fabrication is both a production system with a high product mix and a project-based industry. The accuracy of short-term project-based planning, such as estimating and project scheduling, is extremely important for a project’s success. This paper proposes an integrated modeling system that explicitly models the product mix in order to improve the planning accuracy. Automated process planning and a processing time estimation method were implemented through the integration of simulation with such modeling methods as computer aided process planning and artificial neural networks. Two case studies are presented to demonstrate the capability of the proposed system.

Introduction

Industrial construction includes a wide range of construction projects essential to our utilities and to basic industries, such as petroleum refineries, petrochemical plants, nuclear power plants, and off-shore oil/gas production facilities [4]. Generally, industrial construction projects involve the shop fabrication of steel and pipe, as well as the installation of these products in the construction field. The percentage of shop fabrication in construction has been increasing due to the fact that the production is better managed in the controlled shop environment than it is in the field. Good planning and control of the shop for a delay-free process is critical to the overall success of an industrial construction project.

Steel is an important building material for structural elements in structures, such as buildings and pipe racks. Pipe is also a major component carrying fluid and gas in many facilities, including petroleum refineries and chemical plants. Steel pieces and pipe spools are normally pre-fabricated in fabrication shops. These components either assemble into larger modules or ship directly to the site for installation. “Shop fabrication” refers to the production of steel pieces or pipe spools in a controlled shop environment through a series of cutting and detailing, fitting, welding, and surface processing according to the engineers’ design. Material handling and inspection activities also occur frequently during the fabrication process.

As material suppliers to the project-based construction industry, steel and pipe shops run on a project-by-project basis, which means each project has a relatively short and definite project duration. Traditionally, estimating and scheduling industrial fabrication projects is heavily relied on personal experiences of production engineers. Estimating and scheduling was based primarily on personal experience, information from component drawings, and knowledge of the status of the shop. However, given the complexity of the products and the working environment, as well as the many possible combinations of influencing factors, the human mind would not process the information required for an accurate analysis of such a production system. It is generally risky to make decisions based on “gut instinct” alone. Network-based tools such as CPM/PERT are used by fabricators for project planning and control at the project level. The shortcomings of these methods applied at the shop operational level arise due to several limitations, such as the incapability to model probabilistic branching, resources interaction, and production cycling [16]. Also, although researchers have introduced many analytical optimization-based scheduling algorithms [10], most of these algorithms are highly simplified and static in nature, which limits their direct applicability in managing dynamic industrial fabrication shops. The complex nature of the fabrication process, the industry’s growth, and the adoption of new fabrication technologies and materials require advanced and effective tools capable of thoroughly analyzing the fabrication process.

Simulation models can represent real-world systems at almost any level of detail in order to approximate the actual system. Its ability in modeling products, resources, activity interactions, queuing, and various uncertainties makes it especially suitable for modeling the industrial fabrication process. This paper proposes an approach to building virtual shop models for the purpose of planning the fabrication process. The developed modeling methodology and tools allow users to model both products and the fabrication process, by means of a coordinated use of product and process modeling, simulation, neural networks, and system integration techniques. Two case studies from two industrial fabricators based in Edmonton, Alberta, Canada are presented to demonstrate the capacity of the developed modeling system.

Section snippets

Problem statement

Simulation has been widely applied to modeling manufacturing and construction systems. Extensive literature reviews were conducted on simulation and its applications in a variety of industries, such as electronics manufacturing, shipbuilding, and bridge fabrication. Overviews of these applications are available in several publications, such as [3], [12]. Instead of enumerating these developments, this section first identifies the uniqueness of industrial fabrication. It then addresses the

Virtual shop modeling system architecture

The objective of this research is to provide users a comprehensive modeling tool to model the overall production process from design, process planning, to shop fabrication in a computer for industrial fabrication shop estimating and scheduling. The proposed virtual shop modeling system extends the functions of existing simulation system designs [3] in order to address the unique requirements generated in modeling the industrial fabrication process, namely high product mix and labor-intensive

Modeling products of industrial fabrication

In an industrial fabrication project, production engineers usually decompose the structure into smaller assemblies that are easier for fabrication. Each assembly contains a number of detail components. Components are the most basic elements in the fabrication process. An example of a component would include a section of pipe or steel. It is the “product” of the virtual shop modeling system. The product and its fabrication process are defined by the entity model.

Facility modeling system

The FMS consists of a customized fabrication facility modeling template and a general-purpose discrete-event simulation tool.

Processing time modeling for manual operations

As mentioned previously, processing time for manual operations is affected by many factors and vary considerably in industrial fabrication shops. The traditional approach of randomly sampling from a statistical distribution does not reflect these factors in determining processing time. An artificial neural network (ANN) has been proposed by many researchers to model labor productivity due to the complex relationships between influencing factors and the resulted productivity, such as

Background

Researchers used the developed virtual shop modeling system to model steel fabrication shops for Waiward Steel Fabricators Inc. (WSF), an Edmonton-based steel fabricator. WSF needs a quantitative tool to help them to schedule their fabrication shops. Typical operations in a WSF fabrication shop include detailing, fitting, welding, surface preparation, and surface protection. In a steel fabrication project, the steel structure is normally decomposed into steel pieces, which are further

Background

The fabrication plant of KBR in Edmonton runs five fabrication shops. In order to improve the productivity of its production process, KBR tried to re-design the fabrication process according to lean manufacturing principles. KBR needs an analytical tool to help them evaluate the benefits and risks in applying the new system design.

Typical operations in a pipe spool fabrication shop include cutting, fitting, welding, QC checking, stress relief, hydro testing, and painting or other surface

Conclusion

Industrial fabrication features a high product mix and project-based production environment. The developed modeling system is intended as a platform for building virtual shop models for both long-term and short-term project planning purposes. It is capable of modeling high product mix and product uniqueness through the integration of simulation with CAD and CAPP systems. In this simulation system, processing time of manual operation is modeled and estimated by ANN models. Virtual shop models

Acknowledgments

This project was funded by the Natural Science and Engineering Research Council of Canada under grant number IRC—226956-99. The authors wish to thank Waiward Steel Fabricators Inc. and KBR (Edmonton) for their support.

References (21)

  • S.M. AbouRizk et al.

    Simplifying simulation modeling through integration with 3D CAD

    Journal of Construction Engineering and Management

    (2000)
  • S.M. AbouRizk, Y. Mohamed, Simphony—an integrated environment for construction simulation, in: Proceedings of the 2000...
  • J. Banks

    Handbook of Simulation

    (1998)
  • D.S. Barrie et al.

    Professional Construction Management, Including C.M., Design-Construct, and General Contracting

    (1992)
  • C.M. Bishop

    Neural Networks for Pattern Recognition

    (1995)
  • L.C. Chao et al.

    Estimating construction productivity: neural-network-based approach

    Journal of Computing in Civil Engineering

    (1994)
  • K. Crow

    Computer-Aided Process Planning

    (1992)
  • D. Hajjar, S.M. AbouRizk, Simphony: an environment for building special purpose construction simulation tools, in:...
  • D. Hajjar et al.

    Unified modeling methodology for construction simulation

    Journal of Construction Engineering and Management

    (2002)
  • W.J. Hopp et al.

    Factory Physics

    (1999)
There are more references available in the full text version of this article.

Cited by (12)

  • Random generation of industrial pipelines’ data using Markov chain model

    2018, Advanced Engineering Informatics
    Citation Excerpt :

    As a result, the product routing and the time required for its fabrication may vary widely according to the product features as well as its complexity [19]. Consequently, randomly generating inputs using a probability distribution describing the fabrication time without considering the geometrical randomness and complexity associated with the product itself, may result in the improper modeling of fabrication processes [18]. Hence, randomly generating combinatorial data capable of representing an entire product features such type of formation components and their properties along with its processing time is expected to improve the simulation model accuracy and provide additional flexibility for testing the efficiency of different models under different scenarios.

  • Credible interval estimation for fraction nonconforming: Analytical and numerical solutions

    2017, Automation in Construction
    Citation Excerpt :

    Pipe spool fabrication is crucial for the successful delivery of these industrial construction projects. Typically, pipe spools are built in a fabrication shop according to engineering designs and must be cut, fit, welded, and inspected [19]. Pipe spool fabrication is heavily dependent on welding, which must be sampled and inspected to ensure that welding quality requirements are met.

  • Automating measurement process to improve quality management for piping fabrication

    2015, Structures
    Citation Excerpt :

    In the case of photogrammetry, other factors may also affect the quality of the results, such as dark shadow and seasonal conditions (e.g. snow). Pipe spools are usually fabricated in fab shops (plants), with the process involving a series of activities based on the engineering designs: cutting, fitting, welding, etc. [54]. The production QA process is performed to ensure the fulfillment of the QA specifications and to find any possible defects.

View all citing articles on Scopus
View full text