An integrated modelling framework to support ... - Semantic Scholar

4 downloads 259411 Views 1MB Size Report
Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199 ... A computer-based support tool has been developed to connect with the IDEF model, creating automatically a relational ...... J Mater Process Technol 2004;146:365–76.
ARTICLE IN PRESS

Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199 www.elsevier.com/locate/rcim

An integrated modelling framework to support manufacturing system diagnosis for continuous improvement J.C. Hernandez-Matias, A. Vizan, J. Perez-Garcia, J. Rios Mechanical and Manufacturing Engineering Department, E.T.S. Ingenieros Industriales,, Polytechnic University of Madrid (UPM), Jose Gutierrez Abascal 2, 28006 Madrid, Spain

Abstract This paper proposes an integrated modelling framework for the analysis of manufacturing systems that can increase the capacity of modelling tools for rapidly creating a structured database with multiple detail levels and thus obtain key performance indicators (KPIs) that highlight possible areas for improvement. The method combines five important concepts: hierarchical structure, quantitative/ qualitative analysis, data modelling, manufacturing database and performance indicators. It enables methods to build a full information model of the manufacturing system, from the shopfloor functional structure to the basic production activities (operations, transport, inspection, etc.). The proposed method is based on a modified IDEF model that stores all kind of quantitative and qualitative information. A computer-based support tool has been developed to connect with the IDEF model, creating automatically a relational database through a set of algorithms. This manufacturing datawarehouse is oriented towards obtaining a rapid global vision of the system through multiple indicators. The developed tool has been provided with different scorecard panels to make use of KPIs to decide the best actions for continuous improvement. To demonstrate and validate both the proposed method and the developed tools, a case study has been carried out for a complex manufacturing system. r 2006 Elsevier Ltd. All rights reserved. Keywords: IDEF; Modelling tools; Manufacturing improvement; Manufacturing datawarehouse; Key performance indicators

1. Introduction The high competitiveness of modern industry leads companies to a continuous refinement of their manufacturing processes. Time and motion studies and continuous quality improvement programs are very useful tools in the study of manufacturing systems. However, the high number of strategies, techniques and methods which can be implemented (JIT, TQC, TPM, SMED, QFD, etc.) make analysis of these systems difficult. The reasons are the complexity of the manufacturing system and the high number of implied factors. In many cases, the results obtained from conventional analysis are lacking in a detailed description of the system’s current state. The effort that implies the use of process analysis charts, data summary panels, modelling tools, check lists or the use of Corresponding author. Tel.: +34 913363124; fax: +34 913363003.

E-mail address: [email protected] (J.C. Hernandez-Matias). 0736-5845/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.rcim.2006.10.003

quality tools is wasted due to a lack of integration of this information in subsequent phases. Learning from the information structuring mechanisms provided by the system modelling and from the flexibility of the relational databases, this paper sets out a methodology for modelling manufacturing systems. This methodology allows a rapid analysis of the production and quality activities, and the creation of a data repository used in the evaluation of activities and in the exploitation of the system indicators. The developed methodology integrates the data acquisition cycle, graphical analysis and system evaluation in a single environment. The objective is to identify the activities without added value, the production capacity used and the technical–economical indicators of the system, mainly those related to productivity and costs. The application of this method implies it is oriented to supporting decision-making tasks in continuous improvement action planning that characterises new manufacturing strategies. The proposed method has been used as a conceptual base in the development of reference software architecture

ARTICLE IN PRESS 188

J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

and new software oriented to the analysis and improvement of manufacturing systems. The developed method, architecture and tools have been used for the study and evaluation of a complex production plant, so the results obtained have validated the proposed method. 2. Review of modelling methods to support manufacturing systems analysis Manufacturing analysis for continuous improvement is a technical area with high significance due to the increase of quality and flexibility requirements for end customers. There are many applications cases where manufacturing analysis through modelling methods is performed in order to take decisions [1]:

       

Diagnosis of a disorder (material, information or control flow). Restructuring a manufacturing process to improve its performance. Business Process Reengineering (BPR). Implementation of Enterprise Resource Planning (ERP), Manufacturing Execution Systems (MES) and Product Data Management (PDM). Tuning the organisation structure to face business change. Large-scale systems integration. Alignment or conformity to norms (ISO 9000, ISO 14000). Management decision (activity externalisation/internationalisation). To the above applications we must add the following:

  

Implementation of ABC cost accounting method. Implementation of JIT, TQM, TPM strategies. Diagnosis of wastes (capacity, resources or cost).

The diversity of existing techniques to support manufacturing systems analysis called for focusing the review of the state of art into five methods: IDEF,GRAI-Grid, simulation, Petri Nets and integrated modelling methods. IDEF methodology [2] is the main technique of Business Process Modelling (BPM) for redesigning processes to obtain sustained improvements in the outputs of manufacturing systems [3–5]. It is a descriptive method based on graphical and text description of functions, information and data. The flexibility of the method resides in its capacity for allowing the analysis of complex systems, where there is a need to study multiple levels of detail in a way the analyst can understand the system. Analysing the different IDEF approaches may the IDEF0 be the most widely used version in manufacturing analysis. It consists of a hierarchy of diagrams, text and glossary. The diagrams represent a set of system functions such as boxes (activities), objects interface (arrows) and information. The attachment point between arrows and boxes indicates

the interface type (input, control, output and mechanism). The generation of many levels of detail through the model diagram structure is one of the most important features of IDEF0 as a modelling technique. However, IDEF is a descriptive method based on the graphical and text description of functions, information and data that is often only of use as an aid to documenting and validating processes. It is static and qualitative, which is a drawback to the analysis of processes. There are more IDEF methodologies covering others capabilities. Among them, IDEF1x has been widely accepted by industry. It supports the development of a conceptual schema and semantic structure oriented towards the development of manufacturing databases [6]. However, it is oriented towards being used by information experts, not by technical personnel involved in continuous improvement. The most recent applications include the generation of knowledge management systems based on databases [7,8]. In these cases, IDEF is used to generate knowledge structure but only from a functional point of view. Its graphical display and simple notation make possible to cover any subject from any point of view to any desired degree of completeness. IDEF resolves specific problems in a manufacturing environment. However, a general manufacturing improvement analysis needs the formulation of a well-defined general manufacturing strategy For this purpose, other modelling methodologies such as GRAI-Grid [9–11] help to define the most important decisions in selected processes of the company and the information exchanged among the different functional areas. The main outcome of this model is the identification of improvement areas, which must be in line with the general strategic orientation. The decision centres identified in the GRAI-Grid are decomposed into detailed IDEF0 diagrams to be studied in a detailed way. There are other theoretical approaches to overcome the limitations of modelling using techniques such as Petri Nets or Fuzzy Logic. Petri Net modelling is a very popular and powerful method for modelling and systems analysis that exhibits parallelism, synchronisation, non-determinism and resource-sharing features [12]. Most studies on Petri Net application in process modelling focus on either information aspects or on functional aspects. Bosil et a al. [13] have evaluated the suitability of IDEF and IDEF3 in conjunction with Petri Nets for modelling processes. The latest developments try to integrate function, information, resource and organisation to support complex, dynamic and distributed processes. As regards Fuzzy Logic, Ma et al. [14] have developed a formal framework to provide extensions to IDEF1x to represent fuzzy information. This data model can be converted into a fuzzy relational database model in accordance with some transformation rules. However, the real application of Petri Nets and Fuzzy Logic to end-users is difficult due to the complexity of the techniques. The simulation technique has been used since the 1960s as a tool for investigating the underlying behaviour of

ARTICLE IN PRESS J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

many types of manufacturing systems, principally the study of queuing problems. Simulation is performed through the use of one of the many commercially available tools (Arena, Simprocess, Witness, etc.) [15,16]. The applicability of simulation is greater when it comes to solving specific problems in concrete sectors: production planning, bottlenecks, scheduling and the study of wastes. The simulation software provides user interfaces that allow significant reductions in programming time, but well-defined simulation model specifications are necessary. The use of modelling techniques enhances the quality of simulation models and in turn reduces the time needed to build them. This is the main reason for many researchers to focus their developments on obtaining a direct connection between the modelling tools and process simulators [17–19]. The major limit of simulation is in relation to integration into the real control process of production. It is very difficult to propose solutions and get easy feedback about the impact on the full production system [20]. In these cases, simulation tools use concepts that are too complex for most potential users. Therefore the use of simulation in companies remains irregular and limited. Every method, tool and methodology described above provides solutions to several problems in manufacturing systems. However, there is no single conceptual modelling method that can completely model a complex manufacturing system. As a result of the limitations of these techniques different integrated modelling methods have been developed. GIM Methodology [21] integrates three modelling methods: GRAI (to model decision systems), MERISE (to model information systems) and IDEF0 (to model physical systems). In this sense, SIM [22] composes two modelling methods:Data Flow Diagrams (DFDs) and GRAI-Grids. Neither method considers dynamic aspects of physical subsystems in the manufacturing environment. GI-SIM [23] integrate GRAI-Grid, IDEF0 and SIMAN. The method enhances the static model to include dynamic modelling. Zakarian [24] has developed an integrated framework based on IDEF, stream analysis approach and dynamic simulation. Currently, tendencies in manufacturing modelling are focused towards the use of UML language as complementary modelling method. UML is a modelling language that has been used to generate computer-executable models that encode key aspects of software engineering projects, but it can also be used for manufacturing process modelling. Kim et al. [25] provide a comprehensive review of IDEF techniques and UML. They identified similarities between IDEF objects and UML objects. They have observed that the combined development and reuse of IDEF and UML models has the potential of linking the description of the manufacturing system and the behaviour of the enterprise objects such as suppliers, products, machines, people, teams, etc. These techniques open up fresh opportunities for exploiting the data stored in a manufacturing model. In this sense, the use of extensible mark-up language (XML) as metadata standard is a useful way to create data

189

structures for many external applications that can be easily interpretable. This is the case of the XML-Based Modelling proposed by Huang et al. [26] who propose new models of knowledge using XML. As a conclusion to this review, we express our agreement with the following affirmations by Vernadat, Zakarian and Kusiak:





Most process modelling methodologies are based on informal notation, lack mathematical rigour, and are static and qualitative, thus difficult to be used for analysis [27]. Enterprise modelling is almost totally ignored by SMEs and there is still a long way to go before SMEs master these techniques on their own [28].

The above review focuses on the need to develop methodologies and tools oriented towards easily obtaining a manufacturing model with a global vision of the system through multiple indicators. This methodology must be user friendly if it is to be applied by any shop floor technical personnel with no experience in system information tools. 3. Modelling framework A modelling framework called Production and Quality Activity Model (PQAM) has been developed for modelling manufacturing systems by integrating hierarchy, database and performance indicators. By exploiting the final model, it is possible to identify wastes and areas for improvement. The prior requirements for the development of this method were:

 



  

Applicability to any manufacturing sector, specially SMEs. Design of a conceptual model with a hierarchical and multilevel approach. It must be supported by a more detailed classification of activities that enable the identification of quality costs and added values per activity. Adaptation of IDEF0 modelling to turn it into a quantitative and qualitative data model that represents and stores any information needed for subsequent analysis: times, performance rates, material flows, unit costs, improving characteristics, production activities, quality activities, capacity resources. Creation of interfaces and algorithms to turn the information stored in the activity model into a complete relational database of the production system. Development of tools exploiting the database that help to make decisions on continuous improvement actions. User interface oriented towards being used by manufacturing personnel and not just by system information experts.

ARTICLE IN PRESS 190

J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

MODELLING FRAMEWORK TO MANUFACTURING ANALYSIS PQAM Reference Information Model

Manufacturing Databa

Quantitative Information Model

Evaluation Methods Cuadro de Mando

Fig. 1. Modelling framework for manufacturing analysis.

The modelling framework (Fig. 1) is made up of four components: (1) a reference information model to be used as a reference for structuring and classifying the information, (2) quantitative and qualitative IDEF0 model to compile all the information from the manufacturing system, (3) a manufacturing datawarehouse to store all the information that is needed to diagnose the system and (4) evaluation methods for exploiting data and assisting decision-making on continuous improvement issues. 3.1. Reference information model A diagnosis of a manufacturing system is a complex problem due to the diversity of different and interdisciplinary concepts that must be considered. Some aspects to be taken into consideration must be: the heterogeneity of production flows, product space optimisation, resource capacity, production process organisation and the diversity of production activities. It means building a bridge between manufacturing concepts management and individual manufacturing activities. A multilevel decision hierarchy is required to link activities properties (attributes) with the company’s manufacturing strategy. Only analytic hierarchical process (AHP) has the capabilities to combine different types of information in a multilevel decision structure to get a full vision of the manufacturing system. In recent years, the use of analytic hierarchical process to represent manufacturing systems problems has been used intensively by researchers. We can refer to several topics: simulation models [20,29,30], ERP implementation [31], production planning [32–34], machine selection [35], advance manufacturing technology implementation [36], layout design [37] or cost estimation [38].

The objectives of manufacturing diagnosis advise using a decomposition of the global manufacturing system into a network of subsystems, each subsystem having a specific nature and distinct objectives. Basically, the proposed Reference Information Model has three hierarchical subsystems: (a) factory levels, (b) activities levels and (c) information objects (Fig. 2). (a) The factory levels are used to represent a full decomposition structure of the activities organisation, from the global system with various factories or cost centres to the low levels such as cells, groups or machines sections. (b) The second subsystem of activities levels defines a structure oriented towards classifying the manufacturing activities. From a diagnostic point of view, it is interesting to differentiate between production activities and quality activities. The reason for this first classification is related to modern industry’s need to differentiate quality and non-quality cost. The proposed activities levels are three: object class, activity class and activity type. The first classification level only contains production and quality activities. The second level (activity class) classifies the production activities into process and support activities, while the quality activities are divided into three types of classes: control, correction and support. The third level (activities type) reflects the most detailed level of definition for manufacturing activities by means of a set of 13 basic activities. These are: manual operation, machine operation, delay, production buffer, transport production, auxiliar operation, in process inspection, isolated inspection, manual correction, machine correction, quality buffer, transport quality, and auxilar quality. It offers a more detailed definition of manufacturing activities compared to the traditionally used classification that only establishes five types of activity: operation, waiting, transportation, storage and inspection. The benefits are evident if we think that these classifications allow a better analysis when we perform advanced analysis, filtering activities by different classifications. (c) The third subsystem’s mission is to associate quantitative or qualitative information to the activities. For this purpose five types of information objects have been established, each of them with a specific library of attributes. This means that each manufacturing activity has five types of information that must be incorporated in order to allow for subsequent analysis: activity data, material input, material output, and resource and improvements data. Each activity type has associated a set of specific attributes that have been developed. To design these attributes, we have considered the information that will be needed to calculate the main performance indicators of the system. Table 1, shows examples of attributes for the five information objects of the ‘‘machine operation’’ activity type, one of the 13

ARTICLE IN PRESS J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

191

MANUFACTURING INFORMATION MODEL PQAM Reference Information Model PQAM-IDEF Model

Factory m

Cost Center 1

... ... .

Cost Center p

Manufacturing Unit 1

... ... .

Manufacturing Unit q

Cell / Machine functional Group / Section Others levels

INFORMATION OBJECTS

Quality Activity

Auxiliar

Resource Data

Improvement Data

Process

Activity Data

Levels 0 to n-1

Act 1

Manufacturing routes or Quality routes Act 2 Act 4

... .... .

Auxiliar

Support

Activity type

ACTIVITIES LEVELS

Production Activity

Process

A

.......

Object Class

... ... .

.......

Activity Class

... ... .

Hierarchical IDEF-0 Model

Factory 1

Material Input Data

Material Output Data Data

Attributes IDEF Objects

FACTORY LEVELS

Manufacturing System

Act 3

Activity Type Activity Attributes

Attribute library

Fig. 2. Manufacturing information model (PQAM).

activities considered. This schema explains how the PQAM model is focused to detect wastes in relation with cost, time and production capacity. Waste being defined as ‘‘anything that adds cost without adding value’’ [40,41]. The proposed reference model is able to structure all resource organisation, production activities and quantitative-qualitative information in an integrated way. However, practical techniques are necessary to implement the concepts into a real model, which we will examine in following sections. 3.2. Quantitative and qualitative IDEF0 model The proposed conceptual model needs a method to model all information in an integrated way. The IDEF technique offers hierarchical capabilities to allow multiple levels of detail in order to get an abstract representation of a system. For this reason, a conceptual modification of classical IDEF0 methodology has been designed, in order to create the PQAM model including the qualitative and quantitative information. Basically, adaptations are

focused towards associating attributes to each IDEF object, and incorporating quantitative data about manufacturing variables (cycle time, cost, efficiency, etc.). The proposed structure requires the use of a BPM modelling tool with requirements such as support for the IDEF0 standard, possibility of associating attributes to each object in different formats (numerical, text, data list, etc.) and the ability to export the attributed information. The tool chosen was All Fusion Process Modeller (BPWin) [39] due to the fact that this tool permits the internal creation of attribute libraries using User Defined Properties (UDPs). This capability provides a possibility of associating quantitative or qualitative information to the objects by way of a series of libraries defined by the user. In a similar way, the customisation of the capabilities of Microsoft Visio with Visual Basic for applications and Object Model Methods is an alternative solution. It allows graphic objects to be designed with a set of free attributes that could be connected to any external database using ODBC or similar methods. The incorporation of attributes and the definition of templates could be considered as an improvement and a modification of the current standard IDEF methods.

ARTICLE IN PRESS 192

J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

Table 1 Examples of attributes for activity type ‘‘machine operation’’ Activity type: ‘‘machine operation’’ Attribute

Object

Datatype

Activity_ID Activity_description Setup_time Batch_size_setup Real_cycle_time Batch_size_operation Group_product_route Resource_ID Resource_description Num_resources Utilisation Type_resource Theoretical_capacity MTBF Operation_efficiency Standard_cycle_time Material_ID Material_description Product_BOM Level_BOM Type_route Material_ID Material_description Product_BOM Level_BOM Type_route Process_factor Improvement_ID Improvement_descript Situation Problem Indicator_type Indicator_value Indicator_meta

Activity Activity Activity Activity Activity Activity Activity Resource Resource Resource Resource Resource Resource Resource Resource Resource Input Input Input Input Input Output Output Output Output Output Output Improve Improve Improve Improve Improve Improve Improve

Int Text Double Double Double Double Int Int Text Int Percent Text_list Double Double Percent Double Int Text Text_list Int_list Text_list Int Text Text_list Int_list Text_list Double Int Text Text Text Text_list Double Double

From nodes 0 to n1, the adapted model represents the factory organisation through hierarchical general activities and the arrows between boxes represent the flow of general products in process (see Fig. 1). This means that the final model represents the evolution of the different parts included in the bill of materials (BOM). The last node defines the manufacturing activities to be studied in a detailed way. In these nodes, IDEF objects are used to model and group homogenous data on activities, products, resources and improvements. Each activity represents a production activity or a quality activity, according to the pattern exposed in the PQAM model. Input and output objects represent the different states of the material’s process. These objects include processes for rejected and reprocessed products. This means that connections among production activities will represent manufacturing routes and connections among quality activities will represent quality routes for non-conforming products. Resource objects exactly define any production element used in activities (machines, human labour, etc.), including auxiliary elements such as conveyors or warehouses. Improve-

ment objects reflect the inefficient process detected during the data capture phase in a real plant. The PQAM model defined different activity types and attributes as a way to classify and associate information to the model. With this purpose, an attribute dictionary has been developed for each one of 13 activity types proposed. The dictionary is divided into five types of attributes corresponding to each IDEF object:



   

Activity attributes: Information about the function that defines the production or quality activity (real cycle times, performances, buffer size, etc.). Real times are direct times obtained by direct timing methods. Input attributes: Identification of the products that are processed by the activity, including products reprocessed due to a lack of quality. Output attributes: Information relating to the identification of the products processed. Resources attributes: Information about manufacturing resources (direct and indirect). Improvement attributes: Information relating to inefficiencies detected in the activity, identification of the problem, description of effects and prior suggestions.

3.3. Manufacturing database Once the PQAM-IDEF0 model of the manufacturing system has been defined and the values assigned to the attributes extracted from the libraries, the modelling framework proposes the creation of a database exporting data from the information model. For this exchange purpose, standard files in plain text format containing data about activities, resources and routes have been specified to be generated by the IDEF0 modeller. The first standard file contains the description of the activity–resources–improvements, the processed product, the identification of the IDEF node number, and the values of all the associated attributes. A second file contains the processed products (inputs and outputs). In this way, it is assured that all the information is modelled: attributes, hierarchy of the PQAM model and the production and quality routes are made available for the creation of a database. This process is carried out by applying an algorithm that allows the creation and automatic loading of the database from the text files exported by the IDEF tool. Data Transformation Services (DTS) have been used for these tasks. However, a manufacturing database for analysing the system needs more information to diagnose the system and to be able to produce performance indicators. It is necessary to incorporate those data that have not been modelled, such as the production demand per product, rejected product factor, scrap factors, hourly costs (labour, machine, tools) or data about layout surfaces (machines, warehouse, etc.). For this reason, a subset of tables has been defined to store this information. The aim is to get a complete information model incorporating all the data

ARTICLE IN PRESS J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

Activity_class

Cost_System

Object_Class (FK,IE1) Activity_clas

Cost_System

External_data_IDEF_Activities

Id_modelo (IE1)

External_data_IDEF_Routes

193

MIX_Production

PKI_Activities

ID_modelo Product (FK,IE1)

Id_modelo Indicador1

Id_modelo (IE1)

Resource_type

Bottelnecks

Resources_List

Resource_type

ID_Model (FK) Resource (FK)

ID_MODEL (FK,IE1) Resource

Indicators_KPI_General Object_class

Activity_type

Object_class

Activity_type Object_Class (FK,IE1) Activity_class (FK)

Cost

Cost_type Enterprise Cost_type Company

Manufactitung_Areas Company (FK,IE1) Manufacturing_area

Id_model (IE1) Product

Resorce_type (FK)

Resources_use

Cost_elements Cost_element

Products

ID_MODEL (FK,FK,IE2) Activity (IE1) Resource (FK)

ID_Cost Cost_system (FK,IE3) Cost_element (FK,IE2) Cost_type (FK,IE4) Resource_type (FK,IE5) Activity_type (FK,IE1) Function

MIX_Quality

Indicators_KPI_Cost

ID_modelo Defect Product (FK,IE1)

Id_model Indicador3

ID_Model_PQAM ID_Model (IE2)

Activity

Company (FK,IE1)

ID_MODEL (FK,FK,IE3) Activity (FK,FK,IE1) Product

Improvements_Meta Id_improvement (FK,IE1) Id_task (FK) Id_meta

Activity_type (FK,IE2) Manufacturing_lines Company (FK,IE1) Manufacturing_area (FK) Manufacturing_line

Id_model (FK) Workplaces Company (FK,IE1) Manufcaturing_area (FK) Manufacturing_line (FK) Workplace

Routes

Improvements

Improvements_Planning

Circuito

ID_improvement

Id_Improvement (FK) Id_task

Id_model (FK,FK,IE1) Activity (FK)

Id_model (FK,FK,IE1)

Fig. 3. PQAM-IDEF1x reference.

necessary to a later evaluation of the system performance indicators. In order to draw up this information, a reference IDEF-1x model has been developed that can be used to create the full manufacturing data warehouse. Fig. 3 shows the proposed model containing PQAM data, external data and evaluation data. The structure only shows the entities, data relationships, indexes and foreign keys. 3.4. Evaluation methods The manufacturing database allows the chance to exploit the data to identify and calibrate the status of the system and identify improvements for the activities. A set of algorithms has been developed to generate evaluation data and the KPIs of the system. These algorithms use linear programming and dynamic simulation techniques in relation to time analysis, capacity planning, ABC costs and quality control. The programmed processes use the external data and the PQAM-IDEF information covering the following application areas:

     

Execution of static simulated production plan. Distribution of time–cost activities according to the PQAM model. Analysis of cycle times per line and bottlenecks. Manufacturing ABC costs analysis. Quality and non-quality costs. Target or minimum production costs.

 

Improvement potential, as a relation between actual cost vs. theoretical cost for each activity. Settlement of priority index for improvements according to costs approach.

The starting point to create this evaluation metadata has been the historical throughput analysis (external data). These analyses provide the knowledge of mix factors for calculating the quantity of product in progress in relation to the finished goods. Using these mix production ratios with a proposed throughput (standard demand) and the structure of PQAM products, it is possible to calculate the quantity to be processed in any activity (static production plan). The use of these data in combination with cycle times and takt times provides a feasible way to evaluate the production in progress, capacity utilisation and detailed ABC costs. An example of the internal process for the calculation of costs for a group of activities is shown in the following expression. It has been used to measure the costs of manufacturing and quality, and the factors of deviation with respect to the minimum theoretical cost: CEk ¼

n X m X

C ij  TCpj  N pj ,

i¼1 j¼1

where CEk is the cost element, k the activities group, i the manufacturing resource, j the basic activity, Cij the standard cost for resource i working in activity j, TCpj

ARTICLE IN PRESS J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

194

5. A case study

Table 2 Examples of attribute library Indicator

Parameters

Activity distribution Production distribution Quality distribution Resource type distribution Cost distribution P-Q Cost distribution P-M Cost distribution Q Capacity utilisation Machine surface factor Buffer surface factor Warehouse surface factor Lead time Cost factor improvement Production ratio improvement

Type production, time quality Process, auxiliary Process, auxiliary, correction Direct, support Production cost, quality cost Production cost, maintenance cost Quality, failure, reprocess cost Manufacturing time, standard time Machine surface, total surface Buffer surface, total surface Warehouse surface, total surface Product-progress, final throughput Target cost, profit improvements Target production, real production

the cycle time to process a unit of the product p in activity j, Npj the quantity of product p processed in activity j. The goal of these methods is to calculate indicators to design tools to decide the best actions for continuous improvement. In order to design the indicators we have taken several works as reference [42–45]. Table 2 shows examples of indicators exploiting the PQAM database, with a specific static simulated plan. 4. PQAM-Win tool The practical application of the proposed method has required the development of a specific tool called PQAMWin. This tool has been designed to offer the following capabilities:

    

Direct importation of the IDEF0 model to create a relational database. Definition of general external information: labour cost, cost system, allowed times, production target, etc. Definition of different production planning (mix plan) to be able to simulate the response of the system (capacity analysis). Access to information concerning the technical methods used to improve the manufacturing systems involved. Mechanisms to exploit the information and to diagnose the system, identifying the possibilities for improvement and planning the continuous improvement actions.

The developed tool has been of different informationgrids and scorecard-panels. Information-grids show detailed results about productivity and cost in a grid format with registers. Filtering, grouping, ordering and summarising these registers with database tools will allow us to know information about the weak points of the system. The scorecard-panels are dashboards with direct performance indicators that can be quickly evaluated. Both tools are used to decide the best actions for continuous improvement.

In this section, we will discuss how to apply the PQAM model and associated tools to diagnose a real manufacturing system and to propose improvements. Fig. 4 shows a detailed plan for applying methodology and tools in a real case. Step 1: The IDEF-PQAM modelling 1. IDEF tool configuration: Creation of a full attributes dictionary using the proposed model. The first creation of a file pattern with these data allows their reutilisation in other IDEF models. 2. General structure definition: Firstly, the theoretical PQAM model is defined to level n1. The goal is to model the general structure of activities and the resources model. In this case, the Factory–Shopfloor–Manufacturing Line– Cell scheme is representative for continuous flow systems. 3. Basic activities of IDEF modelling: Definition of the basic activities types in all n-nodes. The inputs and output will define the product flow through the production activities (manufacturing routes) or through the quality activities (quality routes). The PQAM model needs to define direct activities and non-value added activities using the classification proposed (activity types). However, it is possible to define new types for specific production systems. 4. Data capture: A node-tree schema and list of modelled activities are exported from the IDEF tool. This information is used to create automatically a set of different sheetforms to capture data ion a real shopfloor. This information helps the analyst to study and compile the attributes values. Another data source is the manufacturing databases of ERP systems, ISO-9000 procedures, direct measures in workplaces (such as timing methods) or interviews with supervisors. The analytical revision of the AS-IS model permits to detect the first improvement possibilities. 5. Assigning attribute values: Data collected are associated to the different attributes in the IDEF model. Fig. 5 shows the way in which values have been assigned to the attributes for each IDEF object using the BPwin tool. The direct access to the library (or UPDs for the tool) allows the selection of the attribute and the definition of its value by the user. The available attributes for each activity have a direct relationship with the activity type and must be previously defined by the user. 6. Data export: The PQAM-IDEF model is exported through files in standard formats (csv, txt, xml). These files contain PQAM schema and value attributes. Step 2: The datawarehouse creation 7. Physical database creation: Using the IDEF1x specification proposed, it is possible to create a relational MS-SQL database. DTS have been used to create registers and generate the information infrastructure. These tasks have been embedded into the PQAM-Win tool. 8. External data configuration: Definition of external data not modelled in the PQAPM model, such as historical

ARTICLE IN PRESS J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

IDEF-PQAM Modelling

Datawarehouse creation

Manufacturing System Diagnosis

Continuous impovement Planning

Idef Tool Configuration

General Structure Definition

Physical database creation

External data configuration

Static simulation

Basic activities modelling

Managing information grids

Comparative analisys

PKI´s Analysis

Proyect Plan definition

Results measure

Standarizing

195

Data Capture

Attribute values

Data Export

General definition improvements

Priority index for improvements.

Detail definition improvements

Model redesign

Fig. 4. PQAM application plan.

Fig. 5. Attribute selection and configuration in a node-n.

production demand per product, rejected product factor, scrap factors, hourly costs (labour, machine, tools). Step 3: Manufacturing system diagnosis. 9. Static simulation: Using the historical demand, the mix ratios for any product are calculated. The execution of

static simulation for a generic production of finished goods (standard demand) generates values for manufacturing times, lead-time and labour productivity per activity. The structure of the PQAM model allows the data to be easily grouped using the qualitative attributes. Fig. 6 shows the

ARTICLE IN PRESS 196

J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

Fig. 6. Static simulation results.

Fig. 7. Information-grid.

evolution of cycle time along a subset of grouped activities for specific data of demand. 10. Managing information-grids: By using informationgrids in connection with activities, resources and costs, it is possible to identify the weak points of the system. Fig. 7 shows a resources grid monitor with different columns; resource description, quantity, activity, processed activities, real cycle time, standard cycle time, associated cost, and saturation. The grid monitor and offers tools for filtering and grouping and also offers methods for getting any summarised data using the qualitative attributes. 11. PKI’s analysis: The PQAM-Win tool has been developed with different scorecard panels showing calculated PKIs from the static simulation. Fig. 8 shows an example containing graphic distribution for: (a) activities

type, (b) resource utilisation, (c) machine capacity, (d) production-quality cost, (e) list of direct indicators (PKIs). 12. Comparative analysis: The data shown in the grids and scorecard panels are in relation to a static situation of the model for a single demand value and a specific set of activities. However, an effective diagnosis of the system means evaluating different versions of the model. In order to manage different analyses each study is classified by analysis ID number, analysis name and production ratio attributes. The possibility to define different alternatives of the model supported by the same database, such as with TO-BE activities and with different production plans, enables us to obtain comparatives that show the results for different states of the model (Fig. 9). For example, what if the demand varied, what if an activity was more responsive

ARTICLE IN PRESS J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

197

Fig. 8. KPI Scorecard panel.

Fig. 9. Comparative analysis between KPI’s models.

establish the priority for implementation and seek optimum a benefit for the system. Step 4: Continuous improvement planning 15. Detailed definition of improvement actions: For each one of the prioritised improvement actions the different tasks to be carried out are defined, their durations and managers. The indicators that establish the current values and target values are defined too. 16. Project plan definition: The final planning of the actions is carried out exporting to a project management tool (MS Project in this case) the information of the database concerning tasks, durations, resources and priorities. 17. Results measurement: Periodically, the target values of the indicators are measured to verify the execution of the envisaged objectives and propose the complementary actions. 18. Standardising: The proposed improvements are standardised by means of documents written with the description of operational procedures and by spreading actions about their contents. 19. Model redesign: Continuous improvement is an iterative process. This process is carried out by upgrading the PQAM activities model to restart the process described. 6. Conclusions

or, more significantly, what if the activities flow was reconfigured in a different way? This is a way to provide dynamic response to the static model. 13. General definition of improvement actions: The final identification of the critical points in the system allows us to design and model new improvement actions that are added to the ones defined during the model evaluation. 14. Priority index for improvements: The knowledge of the economic effect of each improvement is used to

In the last 20 years, several methodologies, models and tools have been developed and applied to the analysis and optimisation of manufacturing systems in order to propose general improvements. Many of these aids make extensive use of techniques that stem from the field of systems engineering, such as data modelling, simulation, decisionmaking support, expert systems or reference models. Process modelling tools are based on informal notation,

ARTICLE IN PRESS 198

J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199

lack mathematical rigor, and are static and quantitative, and are therefore difficult to be used for analysis. Manufacturing analysis is in fact a complex task due to the difficulty involved in integrating the different types of data to be analysed—such as quality, time, costs, resource capacity, productivity, flexibility or improvements—within a single analysis environment. The extent of this problem, the lack of data integration and the widely divergent goals of these analyses are hindering the development of common, and standardised methods and tools. This paper proposes an integrated modelling framework for manufacturing analysis systems that can increase the capacity of modelling tools for rapidly creating a structured database, thereby obtaining key performance indicators (KPIs) that highlight possible areas for improvement. Manufacturing systems are complex arrangements of physical entities characterised by measurable parameters that must be recognised in order to evaluate the performance of the system. The proposed method is able to create a quantitative and qualitative information model using IDEF0. This model is the first stage for creating a full datawarehouse of the manufacturing activities. A specific decision-making support tool for managing performance indicators has been developed to use this data structure establishing a standard interface that can be used by any modeller and simulator. The flexibility of the proposed method and the use of an open architecture based on standards allow the integrated modelling framework to be applied in different manufacturing systems having a wide range of problems. Their application has an optimal implementation in manufacturing systems where there is a need to measure indicators in order to take decisions. Examples of these situation are systems with characteristics such as non-optimal layout, high buffers between machines, unbalanced production, overproduction, wastes of time (delays and transport) or high ratios of defective products. To summarise, the main application areas of the method are:







Rapid diagnosis of manufacturing systems, especially when the goal of analysis is to implement strategies for excellence like lean manufacturing or continuous improvement. The easy way to apply the method and the rapid return of results makes the system especially useful for SMEs due to the high cost of traditional system diagnosis consultant services. Design of a full database model of the productive system to be used to get KPIs. This information may be very useful for taking decisions about new equipment or the design of new productive systems. The flexibility of the method allows the possibility of importing data from different sources of information such as simulation software or spreadsheets. Non-quality and value analysis. Traditional cost models are not able to generate the right information for production managers about the non-quality cost or the added value of each manufacturing activity.

The work presented is part of a project currently under development in the Manufacturing Department of the Polytechnic University of Madrid in collaboration with SMEs belonging to the timber, aeronautical and automotive industries. The final aim is the proposal of a methodology and tools for the analysis of manufacturing systems that can be used by the maximum number of manufacturing enterprises from different industrial sectors.

References [1] Vernadat F. Tutorial on enterprise modelling. InterOP European Network of Excellence. 2004. /http://interop-noe.org/Events/past/ CAISE/tutorialEM/S [2] NIST. Integration definition for functional modeling. Federal Information Processing Standards Publications, FIPS PUB 183, NIST; 1993. [3] Arne B, Leo A. A systematic method for renewal of production technology. Sessions on optimization of manufacturing systems. CIRP Ann 1994;43(1):409–12. [4] Ki-Young J. Conceptual frame for development of optimized simulation-based scheduling systems. Exp Syst Appl 2000;18: 299–309. [5] Bruce S. BPR, tools add power to workflow automation. Industry insider report, 1997. [6] Jiang Z, Bing C, Edmun C, Zi D. Information modelling for manufacturing systems: a case study. Robotics Comput Integr Manuf 1996;12(3):217–25. [7] Howard L, Lewis H. The development of a database system to optimise manufacturing processes during design. J Mater Process Technol 2003;34:374–82. [8] Ping H, Yu-Cheng L. Developing an activity-based knowledge management system for contractors. Autom Constr 2004;13:781–802. [9] Doumeingts G. How to decentralise decision through GRAI model in production management. Comput Ind 1985;6:501–14. [10] Girard P, Doumeningts G. GRAI-Engineering: a method to model, design and engineering design departments. Int J Comput Integr Manuf 2004;17:716–32. [11] Dossou P, Chen D. Knowledge modelling for production system design in a computer aided GRAI environment. In: Seventh international conference on computer integrated manufacturing— intelligent manufacturing systems, May 2005, Poland. [12] Ming D, Frank C. Process modelling and analysis of manufacturing supply chain networks using object-oriented Petri nets. Robotics Comput Integr Manuf 2001;17:121–9. [13] Bosil J., Giaglis G, Hlupic V. IDEF diagrams and Petri Nets for Business Process Modelling. In: Proceedings of the 2000 winter simulation conference, p. 1676–83. [14] MA Z, Zhang W, Ma W. Extending IDEF1X to model fuzzy data. J Intell Manuf 2002;13:295–307. [15] Hlupic V, Paul R. Guidelines for selection of manufacturing simulation software. IEE Trans 1999;31:21–9. [16] Tillal E, Ray P. Evaluation of tools for modelling manufacturing systems design with multiple levels of detail. Int J Flexible Manuf Syst 2001;13(2):163–76. [17] Thompson W. Introduction to witness and linking to process mapping tools. In: Proceedings of the 1995 winter simulation conference, 1995. [18] Rojas F, Martı´ nez A. From process modelling to enactment and simulation. Department of Computer Science. University of Manchester. In: Proceedings of de European simulation multiconference, 1998. [19] Perera T, Liyanage K. Methodology for rapid identification and collection of input data in the simulation of the manufacturing systems. Simul Pract Theory 2000;7:645–56.

ARTICLE IN PRESS J.C. Hernandez-Matias et al. / Robotics and Computer-Integrated Manufacturing 24 (2008) 187–199 [20] Habchi G, Berchet C. A model for manufacturing systems simulation with a control dimension. Simul Modell Pract Theory 2003;11:21–44. [21] Doumeingts G, Vallespir B, Chen D. Methodologies for designing CIM system: a survey. Comput Ind 1995;25:263–80. [22] Carrie A, Macintosh R. An assessment of GRAI Grids and their use in the Revisar: strathclyde integration method. Prod Plan Control 1997;8:106–13. [23] Al-Ahmari A, Ridway K. An integrated modelling method to support manufacturing systems analysis and design. Comput Ind 1999;38:225–38. [24] Zakarian A, Kusiak A. Process analysis and reengineering. Comput Ind Eng 2001;41:135–50. [25] Kim Ch, Weston R, Hodgson A, Lee K. The complementary use of IDEF and UML modelling approaches. Comput Ind 2003;50:35–56. [26] Chu-Che H, Tzu-Lian, Kusiak, A. XML-based modeling of corporate memory. IEEE Trans Syst Man Cybern 2005; 35: 629–40. [27] Zakarian A, Kusiak A. Analysis of process models. IEEE Trans Electron Packaging Manuf 2000;23:147:137. [28] Vernadat F. Enterprise modeling and integration (EMI): current status and research perspectives. Ann Rev Control 2002;26:15–25. [29] Qin SF, Harrison R, West AA, Wright DK. Development of a novel simulation modelling system for distributed manufacturing. Comput Ind 2004;54:69–81. [30] Zulch G, Jonsson U, Fischer J. Hierarchical simulation of complex production systems by coupling of models. Int J Prod Econ 2002;77:39–51. [31] Johnny KC, Ip WH, Lee TC. The development of an enterprise resources planning systems using a hierarchical design pyramid. J Intell Manuf 1998;9(5):385–99. [32] Hurtibise S, Olivier C, Gharbi A. Planning tools for managing the supply chain. Comput Ind Eng 2004;46:763–79.

199

[33] Yan H, Zhang X, Jiang M. Hierarchical production planning with demand constraints. Comput Ind Eng 2004;46:551–3. [34] Yan H. Hierarchical stochastic production planning for the highest business benefit. Robotics Comput Integr Manuf 2001;17:405–19. [35] Yurdakul M. AHP as a strategic decision-making tool to justify machine tool selection. J Mater Process Technol 2004;146:365–76. [36] Yusuff R, Yee K, Hashmi M. A preliminary study on the potential use of the analytical hierarchical process (AHP) to predict advanced manufacturing technology (AMT) implementation. Robotics Comp Integr Manuf 2001;17:421–7. [37] Yang T, Kuo C. A hierarchical AHP/DEA methodology for the facilities layout design problem. Eur J Oper Res 2003;147:128–36. [38] Koonce D, Judd R, Sormaz D, Masel D. A hierarchical cost estimation tool. Comput Ind 2003;50:293–302. [39] BPWin tutorial. Allfusion process modeller. Computer Associates International, Inc 2002. [40] Robinson A, Schroeder D. Detecting and eliminating invisible waste. Prod Inventory Manage J 1992;4Q:37–42. [41] Shingo S. Non-stock production: the Shingo system for continuous improvement. Cambridge: Productivity Press; 1988. [42] Shewchuk J, Moodie C. Definition and classification of manufacturing flexibility types and measures. Int J Flexible Manuf Syst 1998;10:325–49. [43] Browne J, Dubois D, Rathmill K, Sethi S, Stecke K. Classification of flexible manufacturing systems. FMS Mag 1984;April: 114–7. [44] Sethi A. Flexibility in manufacturing: a survey. Int J Flexible Manuf Syst 1990;2:289–328. [45] Munir A, Dhafer N. Establishing and improving manufacturing performance measures. Robotics Comput Integr Manuf 2002;18: 171–6.