Selection of lean manufacturing systems using the PROMETHEE

The Authors

G. Anand, Mechanical Engineering Group, Birla Institute of Technology & Science, Pilani, India

Rambabu Kodali, Mechanical Engineering Group, Birla Institute of Technology & Science, Pilani, India

Acknowledgements

The authors would like to acknowledge the anonymous reviewers for their valuable comments and suggestions. Thanks are also due to Dr Srikanta Routroy, Assistant Professor, Mechanical Engineering Group, BITS, Pilani for his timely help.

Abstract

Purpose – In recent years, many manufacturing companies are attempting to implement lean manufacturing systems (LMS) as an effective manufacturing strategy to survive in a highly competitive market. Such a process of selecting a suitable manufacturing system is highly complex and strategic in nature. The paper aims to how companies make a strategic decision of selecting LMS as part of their manufacturing strategy, and on what basis such strategic decisions are made by the managers.

Design/methodology/approach – A case study of a small- and medium-sized enterprise is presented, in which the managers are contemplating on implementing either computer integrated manufacturing systems (CIMS) or LMS. To supplement the decision-making process, a multi-criteria decision making (MCDM) model, namely, the preference ranking organisation method for enrichment evaluations (PROMETHEE) is used to analyse how it will impact the stakeholders of the organisation, and the benefits gained.

Findings – An extensive analysis of PROMETHEE model revealed that LMS was the best for the given circumstances of the case.

Research limitations/implications – The same problem can be extended by incorporating the constraints (such as financial, technical, social) of the organisation by utilising an extended version of PROMETHEE called the PROMETHEE V. Since, a single case study approach has been utilised, the findings cannot be generalized for any other industry.

Practical limitations/implications – The methodology of PROMETHEE and its algorithm has been demonstrated in a detailed way and it is believed that it will be useful for managers to apply such MCDM tools to supplement their decision-making efforts.

Originality/value – According to the authors’ knowledge there is no paper in the literature, which discusses the application of PROMETHEE in making a strategic decision of implementing LMS as a part of an organisation's manufacturing strategy.

Article Type:

Research paper

Keyword(s):

Lean production; Decision making; Strategic choices.

Journal:

Journal of Modelling in Management

Volume:

Number:

Year:

2008

pp:

40-70

Copyright ©

Emerald Group Publishing Limited

ISSN:

1746-5664

1 Introduction

In recent years, many organisations worldwide, especially in the manufacturing sector have implemented either manufacturing philosophies and practices like lean manufacturing systems (LMS), total quality management (TQM), six-sigma (SS), total productive maintenance (TPM), etc. or the technically sophisticated manufacturing systems like flexible manufacturing systems (FMS), computer integrated manufacturing systems (CIMS), etc. Even in the Indian manufacturing sector, the situation remains the same. For example, Seth and Gupta (2005) discussed about the application of value stream mapping (VSM) for productivity improvement of an Indian company and reported about the gain in production output per person and reduction of work-in-process (WIP) and finished goods inventory. Similarly, Sharma et al. (2006) presented a case study of an Indian company, in which TPM was implemented in a semi-automated cell. They found that TPM not only leads to increase in efficiency and effectiveness of manufacturing systems measured in terms of overall equipment effectiveness (OEE) index, but also prepares the plant to meet the challenges put forward by globally competing economies to achieve world class manufacturing status. Antony et al. (2005) explained about the application of SS-based methodology (i.e. DMAIC: define-measure-analyse-improve-control) in eliminating an engine-overheating problem in an Indian automotive company, which resulted in a reduction of the jamming problem encountered in the cylinder head and increased the process capability from 0.49 to 1.28. Further it had a significant financial impact (saving over US$110, 000 per annum) on the bottom-line of the company. Jagadeesh (1999) traced the growth and spread of TQM in India from its initiation to current status. Further, he described about the success stories of Indian companies such as Escorts Limited and J.K. Synthetics, which adopted TQM and have taken necessary initiatives for being more customer focused. On the other hand, Narain et al. (2004) presented the findings of case studies carried out in two large Indian manufacturing organisations (involved in making shoes and railway coaches) to highlight the status of adoption of FMS. They explained that the move to implement FMS was triggered by the external stimuli of customers and competitors for the first firm (shoe manufacturer), while it was the sheer varieties of products, which forced the second firm (railway coach manufacturer) to adopt FMS. But in all these cases, some intriguing questions remain unanswered:

How would the managers or executives of these organisations have made a decision of implementing such advanced manufacturing management philosophies or technically sophisticated systems in their organisations as part of their manufacturing strategy?
Further, these decisions are strategic in nature and the investment involved in implementing these decisions will be very high and tend to be irreversible. Though, cost can be one of the important deciding factors, these strategic decisions cannot be made based on just one parameter alone. Rather, a host of other factors, which are both qualitative and quantitative in nature should be considered and analysed. Hence, it is necessary to understand “what factors would the managers have considered in making such decisions when so many alternatives are available?”
Since, multiple factors or elements are to be considered what decision-making methodology would the managers have used to supplement their decisions?

In this paper, an attempt has been made to answer the above-raised questions by utilising the following approach/methodology:

A case study of a small- and medium-sized enterprise (SME) involved in valve production is presented in which the managers were contemplating on implementing either CIMS or LMS.
A multi-criteria decision making (MCDM) model, namely, the preference ranking organisation method for enrichment evaluations (PROMETHEE) is used to supplement the decision-making efforts.
In addition to this, a detailed literature review was carried out to analyse about the alternatives chosen by the manufacturer and also to understand the algorithm of PROMETHEE, its advantages and limitations apart from its applications in diverse fields.

The paper is arranged as follows: Section 2 provides an overview about the case organisation and discusses about the problems faced by the company while Section 3 describes the literature review about the alternative manufacturing systems considered by the case organisation. Section 4 gives an introduction to MCDM models and in particular about PROMETHEE, its application and its suitability for the problem under study. Section 5 describes the algorithm of PROMETHEE and explains in detail how it was utilised for the problem under study. Section 6 deals with the results and discussion, while Section 7 ends with conclusions.

2 An overview about the case organisation

The organisation considered for case study is a medium-sized valve manufacturer located in the north-western part of India. It manufactures different types of valves (relief valves, flow control valves, etc.) and its associated components. These valves are predominantly used in pressure vessels. The case organisation is one of the first tier suppliers to the pressure vessel manufacturers. Table I presents a summary about the case organisation.

The organisation is currently facing lot of problems in terms of not being able to meet its competitive priorities. Following are some of the problems that are faced by the organisation:

High variety and low volume. The design of valves is highly varying because it is customised for the type of pressure vessels built. This resulted in a variety of valves under each type and naturally, the number of associated components and spares is also very high. On the other hand, the volume for each type of valves is low, which naturally increased the number of stock keeping units for the organisation. In addition to this problem, most of the valves and its associated parts differ in terms of dimensions, design (shape) and materials used, which makes the organisation to carry out a lot of setup and material handling activities.
Quality concerns. Valve is considered to be one of the critical components in the pressure vessel assembly as it is concerned with the safety of the product as well as that of user. Hence, the valve and its associated components have to be precisely machined and there is no room for even a slight deviation from the specifications. In the past, the company had faced few quality problems and some of their lots were returned back even for a slight deviation from the specifications resulting in significant losses to the tune of around Rs 12 lakhs.
Delivery. Since, the requirement of power is growing in India, the market for pressure vessels is also increasing. Naturally, the demand for the valves and its associated components are increasing and it is expected to rise further in the future too. Hence, on one hand, the company was expecting more orders from the customers, but on the other hand, the orders were not appreciably increasing as expected by the company. On analysing the problem, they found that the delivery performance of the company was not well appreciated by their clients. Even though they made efforts to supply a fairly good quality product, they had problems in meeting the deadlines and target. They found that their on-time delivery record was just 75 per cent.
High cost. Adding fuel to their existing internal problems, the number of competitors in the valve market has started to increase resulting in increased cost pressure for the organisation. Further, in the last two years, the material, labour and energy costs were also spiralling upwards, but the clients of the case organisation were emphasising on continuous price reduction every year as per their long-term contract.

Analysing the production process, the manufacturing of valves and other components are currently manufactured with the help of semi-automatic, general purpose machines and few fully automatic turning and machining centres. The number of people on roll at present is around 80. Though the valve manufacturer is poised for growth, the management is worried about the above-mentioned problems. The managers in the top level would like to make changes and transform their existing or traditional manufacturing systems (TMS) and are in the process of laying out strategies and policies to become a world-class valve manufacturer in India within the next five years. They were contemplating on the following alternatives to resolve the above-mentioned problems:

a highly sophisticated and technically intensive CIMS; and
a highly practical and management-oriented LMS.

Though it is a medium-sized enterprise, the managers have identified CIMS as one of the alternatives based on the existing technology they possess and from the perspective of economy of scale, assuming an increase in demand in the future. The organisation is currently using the following computerised systems:

Computer aided design (CAD). They use software packages like AutoCAD for the purpose of designing the tools, fixtures and other material handling systems apart from generating the drawings and documents of their products.
Computer aided manufacturing (CAM). They also use computer numerical control (CNC) machines as they possess a couple of turning and machining centres apart from semi-automatic machines. In addition to this, they have also incorporated local automation for some machines as part of their productivity improvement activities carried out earlier.
Computer aided production planning and control. They perform the production planning and scheduling activities such as material requirements planning (MRP) and order processing using standalone planning software developed indigenously, which utilises the spreadsheet applications like Microsoft Excel and Access.

On the other hand, the top management was also open to implement management philosophies such as LMS. This is because, as a first tier supplier to pressure vessel manufacturers, they have obtained the ISO 9000 certification, which have shown them good results in the past as they could standardise various processes apart from reducing defects. Hence, they were contemplating on implementing such manufacturing management practices and philosophies. But the issue here is “how to choose between LMS and CIMS?”. For a better understanding about the capability, tools, techniques and the overall applicability of CIMS and LMS, a detailed literature survey was undertaken.

3 Literature review

3.1 Computer integrated manufacturing systems

According to Groover (2001), CIMS denotes the pervasive use of computer systems to design the products, plan the production, control the operations, and perform various business-related functions needed in a manufacturing firm. Milling (1997) discussed briefly about the components of computer integrated manufacturing (CIM) as shown in Figure 1.

Boaden and Dale (1990) have listed out some of the benefits of CIMS, which includes: cost reduction, reduced lead time, improved flexibility of response, etc. Hassard and Forrester (1997) commented that the introduction of CIM can be a catalyst to human resource and organisational change within manufacturing companies. They reviewed the strategic benefits, which are expected to accrue from full CIM implementation and have also suggested ways in which organisational and social problems may be overcome in the implementation of CIM. Attaran (1997) presented different case studies, in which US firms like Motorola, Allen Bradley, Texas Instruments and Tandem Computers have successfully applied CIM and capitalized on the advantages of such advanced technology. He explained about the barriers to factory automation and the steps for successful implementation apart from examining technologies that enhance CIM implementation. Caputo et al. (1998) developed a methodology for introducing CIM technology in small companies and have assessed the capabilities of the proposed approach using a case study of a small Italian company (Italpneumatica Sud) producing pneumatic components. Based on the literature review, they identified the various situations or factors which favour small companies in developing and implementing CIM applications. They have also claimed that the introduction of CIM technologies may be one of the most promising strategies to acquire and maintain a competitive edge, particularly for small companies.

On the other hand, Gunasekaran et al. (2000) developed a generalized practical framework for the design and implementation of CIM in SMEs. They have validated the same using a case study of an industry in the SME sector adopting CIM. In another study, Gunasekaran and Thevarajah (1999) analyzed empirically, the implications of CIM in British SMEs. They analyzed the economic impact (which includes the combined influence of profitability, operating risk and present net value), strategic impact (which includes the characteristics of a company in terms of customer satisfaction, reduction in lead time, improved quality of products and improved market share), the social impact (concerning the changes in the nature and level of labour loading, union relations, labour productivity, training requirements, and motivation), the operational impact (which includes aspects such as delivery schedule performance, productivity, inventory, maintainability, flexibility and quality control). Marri et al. (1998) referred the definition of Lefebvre et al. and discussed about the components of CIMS as follows: “CIM is concerned with providing computer assistance, control and high-level integrated automation at all levels in manufacturing (and other) industries, by linking islands of automation into a distributed processing system”. These isolated automated production islands include NC machines, distributed numerical control (DNC), CNC, CAD, CAM, MRP, manufacturing resource planning (MRP II), computer-aided process planning (CAPP), automated storage, computer controlled material handling equipment, and robotics. Similarly, in another study, Gunasekaran et al. (2001) analysed the implications of organisation and human behaviour due to implementation of CIM in SMEs and explained that it requires cross-functional co-operation, and involvement of employees in product and process development. Apart from this, they highlighted that a successful CIM initiative in SMEs must have top management involvement and commitment and a CIM compatible organisational infrastructure which includes requisite skills, appropriate training and education, and adequate incentives and rewards.

3.2 Lean manufacturing systems

The researchers Womack et al. (1990) from Massachusetts Institute of Technology, USA have coined the word “lean production (LP)” after their landmark study titled “International Motor Vehicle Program (IMVP)” which investigated productivity and management practices in the motor industry involving 52 vehicle assembly plants in 14 different countries around the world. Smeds (1994) described the constituents of LM as follows:

The LM principles include the integration of production activities into self-contained units along the production flow. These units produce flexibly, with short throughput times and high quality similar parts or whole products. Flexible manufacturing technology and a group of multi-skilled operators with a high degree of autonomy and self-regulation characterize these production cells. They are mainly controlled by cell output in a simple pull mode: just in time for the need of the next “customer”. Thus, the lean production process also cuts across traditional organizational functions and the changes spread into the connected processes. In consort, these “lean” improvements in all business processes can create the radical innovation, a new organizational configuration “lean enterprise”.

The proponents argued that LMS has a universal applicability. On the other hand, some researchers like Cooney (2002) questioned the universality of lean and emphasised that it cannot replace the TMS like craft and mass production systems. This is indeed true; however, LM can really transform them. There are adequate evidences available in the literature in the form of case studies, which support the claim that the LM practices have a universal appeal. For example, Parry and Turner (2006) described three case studies about Rolls Royce, Airbus and Weston Aerospace in UK that has been practicing LM. As described earlier, Seth and Gupta (2005) used the VSM, a LM technique to achieve productivity improvement at the supplier end (motorcycle frames manufacturer) for an auto industry in India, while Dhandapani et al. (2004) presented a case study of a steel company that applied some aspects of lean thinking and have explained that per annum production costs can be reduced by 8 per cent of turnover, while capital equivalent to 3.5 per cent of turnover can be released through the removal of inventory. Lee and Allwood (2003) investigated how LM can be applied to temperate dependent processes with interruptions. They have used simulation to model an extrusion process and used dynamic programming to optimize the profit for the same. Sohal and Egglestone (1994) reported about the adoption of LP by many Australian organisations. Thus, these cases substantiate that LM has been implemented in project shops (aerospace industries), in a discrete mass production industry (auto-component supplier) and also in a batch production environment (steel mills). On the other hand, Karlsson and Ahlstrom (1997) developed a framework to represent their view of the theoretical concept of the lean enterprise and studied how it can be applied in a SME. Various authors have described about the benefits of LMS. For instance, Detty and Yingling (2000) developed a simulation model to quantify the benefits of the lean system (relative to the existing system) and they found that:

Average time parts spent in system reduced by 55 per cent.
Model changeover time reduced in the assembly cells from 11 to 3 minutes.
Average inventory throughout the system was reduced as shown below:
- per cent lower warehouse inventories;
- per cent lower exchange inventories;
- per cent reduction in assembly cell inventories;
- per cent reduction in pre-assembly and kitting inventories; and
- per cent reduction in finished goods inventory.
Floor space requirement was reduced by:
- per cent in warehouse area due to reduced maximum inventory levels; and
- per cent in exchange area from lower maximum inventory requirements, etc.

According to Treville and Antonakis (2006), these benefits are obtained through the demanding factory physics of LP, which will be achieved over time through a combination of synergistic and mutually reinforcing practices. They have grouped those practices into several complementary sub-systems including (but not limited to) just-in-time (JIT) manufacturing, TQM, TPM, Kaizen (continuous improvement), design for manufacturing and assembly, supplier management and human resources management (HRM) practices under the “respect-for-workers” umbrella, which serves as the glue to hold the overall system together. Feld (2001) identified that LM consists of five primary elements: manufacturing flow, organisation, process control, metrics/performance measures and logistics. Under these elements, he categorized various tools, techniques and practices of LM as sub-elements. Shah and Ward (2003) listed out 21 elements based on the literature review and categorized them into four practice bundles associated with JIT, TPM, TQM and HRM. Bonavia and Marin (2006) conducted an empirical study to determine the degree of use of some of the most representative LP practices in the Spanish ceramic tile industry apart from understanding their relationship with plant size and their effect on the operational performance of the companies in the sector. They concluded that in the ceramic sector, there is one set of practices that has been scarcely implemented (group technology, kanban, reduction of set-up time, development of multi-function employees and visual factory) while another set (standardisation of operations, TPM and quality controls) was used quite frequently.

3.3 Choosing the manufacturing system

A detailed literature review on CIMS and LMS revealed that there is a lack of research in the area of how to choose between these two manufacturing systems. It is very normal to make a decision of choosing LMS or CIMS by analysing the cost aspects. But these decisions are part of a manufacturing strategy of an organisation. Hence, due care should be taken to analyse such problems from various perspectives. Hence, in this case, apart from the cost, the problem will be analysed from the following perspectives:

What is its impact on the organisation as a whole?
What is its impact on the stakeholders of the organisation? In other words, how these alternative manufacturing systems will affect the top management, employees, suppliers, customers and shareholders?
What are the perceived benefits for each of the alternative manufacturing systems?

To make a decision based on these perspectives, it is necessary to identify the factors or criteria or attributes (which we would like to call as “elements”) that are relevant to each of these perspectives. The elements were identified from the literature survey and discussions held with the experts. The experts from the case organisation include the president – operations, and the manager – production engineering department, who participated in the study in addition to the authors. Table II shows the list of elements for the selection of suitable manufacturing systems.

Since, there are many elements and sub-elements to be analyzed during decision-making, the use of MCDM models was suggested. The next section will describe in detail about a MCDM model, namely, the PROMETHEE.

4 The PROMETHEE and its applications

There are numerous MCDM models available in the literature, which are used under different situations. Some examples are elimination and choice translating reality (ELECTRE), technique for order preference by similarity to ideal solution, joint probability decision-making, equivalent cost analysis, multi-attribute utility theory, analytic hierarchy process (AHP), etc. Amongst these models, the most commonly used model is AHP. The AHP methodology as explained in Saaty's (1980) book has three main steps: structuring the hierarchy, performing paired comparisons between elements/decision alternatives and synthesizing results. AHP would be appropriate whenever a goal is clearly stated and a set of relevant criteria and alternatives are available. When there are numerous criteria involved, AHP is one of the very few MCDM approaches capable of handling so many criteria, even if some of the criteria are qualitative. Hence, for the current problem too, AHP can be applied, but it was not utilised because AHP is unable to handle decision problems that are subjected to constraints (Pandey and Kengpol, 1995). In addition to this, some of the authors such as Macharis and Springael (2004) and Albadvi et al. (2007) have compared PROMETHEE with AHP and found that:

the PROMETHEE I does not aggregate good scores on some criteria and bad scores on other criteria, as in AHP;
it has less pair-wise comparisons when compared to AHP; and
it does not have the artificial limitation of the use of the nine-point scale for evaluation as in AHP.

Similarly, L'Eglise et al. (2001) explained that they chose the PROMETHEE for its ease of application, its efficiency and its interactivity as it has a transparent influence of each criterion and weight on the solution. According to them, another main advantage of this evaluation methodology is that it is based on the importance of a performance difference between two solutions, which is best describing whether a solution should be preferred to another one. In addition to this, user-friendly software (Decision Lab, 2000) is available for performing the calculations even though the level of complexity of this algorithm is low. Considering all these factors, PROMETHEE was chosen as the decision aid for the problem under study.

4.1 Introduction to PROMETHEE

The ELECTRE and the PROMETHEE are the two most popular families of the outranking methods introduced by Roy (1973). PROMETHEE is a MCDM method developed by Brans and Vinke (1985). It is still evolving and developing as evident from the works of Brans et al. (1986), Diakoulaki and Koumoutsos (1991), Brans and Mareschal (1994), Goumas and Lygerou (2000), etc. In this method, the intensity of the preference for alternative “a” over alternative “b” with regard to each criterion “j” is measured in terms of a preference function P _j(a, b), which is evaluated based on the generalised criterion for each “j”. Brans et al. (1986) proposed the following six possible types of generalised criterion as shown in Figure 2.

In order to define these criterions and evaluate the preference functions, one or two of the following thresholds have to be fixed:

Indifference threshold (q). It is the lowest value of d _j(a, b) below which the decision maker considers, there is indifference between “a” and “b”.
Strict preference threshold (p). It is the lowest value of d _j(a, b) below which the decision maker considers, there is a strict preference of “a” and “b”.
Standard deviation (σ). It is a well-known parameter directly connected with standard deviation of a normal distribution.

A weighted average of the preference functions is calculated to obtain a rank ordering of the alternatives. “PROMETHEE I” provides a partial pre-ordering of the alternatives through a pair-wise dominance comparison of positive and negative outranking flows, while, “PROMETHEE II” provides a complete pre-ordering through a comparison of net outranking flows. More details regarding the methodology can be found under the algorithm section.

A review of literature on PROMETHEE revealed that it has received wide attention and has been applied in diverse areas. Raju and Pillai (1999) utilized PROMETHEE II to select the best reservoir configuration for the case study of Chaliyar river basin, Kerala, India. They compared the PROMETHEE II with four other MCDM methods, namely, ELECTRE-2, AHP, compromise programming (CP) and EXPROM-2 and commented that though these methods follow different approaches, the analysis has shown that the same preference strategy is reached by all the methods. Further, they concluded that CP was best suited for their case problem. Cavallaro (2005) utilised PROMETHEE II for selecting renewable energy installations from a number of alternatives that are operating in the area of Messina in Sicily. They explained the algorithm using a case study approach in which Wind was suited as the best among the alternatives for various scenarios like economic-, environment-oriented, etc. Similarly, a lot of applications of PROMETHEE can be found in the literature. But no such application was found in the field of LMS. Table III shows the summary of PROMETHEE applications in various fields.

5 Algorithm of PROMETHEE for the selection of LMS

The algorithm of PROMETHEE II for the problem under study is discussed below in a step-by-step manner:

Define the problem and determine the objectives. The problem for the case organisation is “how to select a suitable manufacturing system from the available alternatives as part of their manufacturing strategy?”
Identify the alternatives (a _j) available. The alternatives considered in this case are:
- Existing/traditional manufacturing systems – TMS.
- Computer integrated manufacturing systems – CIMS.
- Lean manufacturing systems – LMS.
Determine the attributes/criteria/performance indicators g _j (where j=1, 2, 3 … J) that govern the problem. As discussed earlier, the selection of alternative manufacturing systems will be based on various perspectives, namely, the drivers, impact on the stakeholders of organisation and the perceived benefits. The elements corresponding to these perspectives have been identified earlier and are shown in Table II. A brief explanation about each of these elements was provided to the participants from the case organisation for the sake of clarity and understanding.
Classify the attributes/criteria/performance indicators into direct (performance grows while measure increases) and indirect categories (performance grows while measure decreases). In other words, identify those elements, which have to be maximised and minimised. For instance, if we consider sales, it should always be maximised with the implementation of new systems. Hence, it will falls under the direct category. On the other hand, if we consider any of the cost factors (say implementation cost), it should always be minimised, in which case it fall under indirect category. Along with the support of authors, the participants were able to classify the identified elements into maximum (direct case) and minimum categories (indirect case).
Choose the preference function for each attribute/criterion/performance indicators. A guideline to choose the preference function was provided by Routroy and Kodali (2007), which are as follows:
- Type I (usual criterion). It is a basic type without any threshold and very seldom used.
- Type II (U-shape criterion). It uses a single indifference threshold, which is generally used with qualitative criteria.
- Type III (V-shape criterion). It uses a single preference threshold and often it is used with quantitative criteria.
- Type IV (level criterion). It is similar to U-shape but with an additional preference threshold and it is mostly used with qualitative criteria.
- Type V (V-shape criterion with indifference threshold criterion). It is similar to V-shape but with an additional indifference threshold and often used with quantitative criteria.
- Type VI (Gaussian criterion). It is seldom used.
1. The concept and features of each preference function were thoroughly explained to the participants by the authors. Utilising the above guidelines, the preference function was assigned to each element.
Form the threshold matrix using the strong preference threshold value (p _j) and indifference threshold value (q _j) for each attribute/criterion/performance indicator if required depending upon the preference function. Again the authors explained the concept of threshold to the participants and they were asked to provide the threshold values.
Absolute weight value w _j on a suitable scale (say 1-10) is assigned for each attribute/criterion/performance indicator reflecting the normative judgment of the decision maker. In this case, the participants were asked to rate the importance of each attribute/criterion/performance indicator with respect to the problem. For instance, the participants were asked verbally – “how do you rate the importance of ‘implementation cost’ with respect to the objective?” In a similar manner, the importance rating for all elements was collected from the participants. Steps 4-7 are shown in Table IV. Table IV shows the classification and assignment of preference function for each element.
Obtain the relative weight value (W _j) for each attribute/criterion/performance indicator (g _j) from absolute weight value w _j using the following equation: Equation 1
Form the performance matrix, by filling up the co-efficient g _ij related to the attribute/criterion/performance indicator g _j ( j=1, 2, 3, … J ) for the alternative a _i (i=1,2,3, … I ). For most of the quantitative elements, data were obtained from the case organisation. For instance, the training cost (TRC) under the existing system was Rs 3 lakhs and the participants estimated that the TRC for alternative manufacturing systems will be around Rs 10 and 8 lakhs for CIMS and LMS, respectively.
Quantify the qualitative attributes using the scale of 1-10, where 1 refers to very low, 3 means low, 5 means medium, 7 means high, and 9 means very high. Similarly, for the qualitative attributes, the participants were asked to compare the alternatives with respect to each of the attributes. They were asked to rate each element/attribute with respect to the alternatives. After a thorough discussion and deliberations, they provided the rating values. For example, the participants provided the rating of 4 for TMS, 7 for CIMS and 6 for LMS for the criterion “union issues”. This is due to the reason that CIMS will result in complete automation of the existing manufacturing system resulting in lesser amount of work, thereby lesser labour, which may result in the reduction of manpower. Similarly, implementing LMS will result in the reduction of non-value added (NVA) activities, which may also result in the requirement of lesser labour. Hence, considering these issues, the participants felt that there will be more union intervention and problems, while implementing CIMS and LMS and in particular, they felt that CIMS will have more impact than LMS on such union issues. In a similar manner, the data were obtained and the performance matrix was constructed. Table V shows the performance matrix. Steps 8-10 are shown in Table V.
Calculate the preference index for each alternative over all criteria. The preference index is defined as: Equation 2 where, W _j refers to the weight assigned to the criterion j and P _j(a ₁, a ₂) is represented as P _j[d _j(a ₁, a ₂)]. Where, P _j(a ₁, a ₂) refers to the value of the preference function according to the difference between the evaluations of the alternatives a ₁ and a ₂ on the criterion j, where (d _j(a ₁,a ₂)=g _j(a ₁)−g _j(a ₂)). π(a ₁, a ₂) represents the intensity of preference of the decision maker of alternative “a₁” over action “a₂” when considering simultaneously all the criteria. It is a figure between “0” and “1” and:
- π(a ₁, a ₂)=0 denotes a weak preference of “a₁” over “a₂” for all the criteria; and
- π(a ₁, a ₂)=1 denotes a strong preference of “a₁” over “a₂” for all the criteria.
Compute positive (where alternative is dominating) and negative (where alternative is dominated) outranking flows for each alternative as shown in equations (3) and (4): Equation 3 Equation 4
Compute the net flow as shown in equation (5): Equation 5
1. The higher the leaving flow and the lower the entering flow, the better the action. Table VI shows the positive, negative and net flow for each alternative with varying weights. Steps 11-13 are shown in Table VI.
Compute the pre-orders of the alternatives using the following conditions: Equation 6
Compile the partial pre-order (PROMETHEE I) of the alternatives by considering the intersection of the above pre-orders: Equation 7
1. By using the PROMETHEE I method, some actions still remain incomparable, because only confirmed outrankings are given by the partial pre-order. Table VII shows the partial pre-orders for the alternatives.
To derive the complete pre-order of the alternatives, use the following conditions of the net flow (PROMETHEE II) and rank the alternatives by their net flow: Equation 8

Table VIII shows the complete pre-orders for the alternatives along with their rank.

6 Results and discussion

Highly user-friendly software, i.e. PROMETHEE II has been developed in VC++ to aid the user for comparison of the elements with respect to the alternatives and for performing the analysis utilising the user inputs. Though a commercial software package called Decision Lab 2000 is available, the same was not utilised due to the budgetary constraints (The price of the software is approximately CA$1,500). The indigenously developed software is capable of generating graphical outputs and also supports the data to be retrieved in the form of a spreadsheet similar to that of Decision Lab. The only disadvantage of the indigenously developed software is that, it does not have the graphical functionality of geometrical analysis for interactive assistance (GAIA). For instance, Figures 3-5 show the graphs depicting the positive, negative and net outflow of alternatives for the criterion – “TRC” respectively, which were generated from indigenously developed software.

From these graphs, it can be inferred that under the criterion – “TRC” TMS is found to be better as per positive outflow (Figure 3) as the TRC was least. Similarly, referring to Figure 4, since CIMS and LMS incur more TRC, naturally, the negative outflow for these alternatives were very high. Considering the net outflow (Figure 5), it can be concluded that TMS performed better than the other alternatives. On the other hand, if you consider the overall positive, negative and net outflow for the alternatives by considering all the elements in tandem as shown in Figure 6, it can be found that LMS has outscored both CIMS and TMS, as it has ranked well in other criteria.

To check for the sensitivity of the decision made, the PROMETHEE analysis was repeated with same data values, but the weight values for individual elements were kept equal. In other words, the managers of the case organisation want to give equal importance to all the elements and hence, equal weight values (say 1) was assigned for each criterion. Table IX shows the positive, negative and net flow for each alternative with equal weights. In this case also, it was found that LMS has outranked the other two alternatives. Thus, it can be said that the decision is highly reliable.

6.1 Research limitations

The above-described problem can also be extended by incorporating the constraints (such as financial, technical, social, etc.) of the organisation along with the factors/elements considered and it can be modelled by using an extended version of PROMETHEE called the PROMETHEE V. Since, a single case study approach has been utilised, the findings cannot be generalised for any other industry.

6.2 Limitations of PROMETHEE

Though a lot of advantages are reported in the literature, various authors have also identified the shortcomings of the PROMETHEE. For instance, De Keyser and Peeters (1996) remarked that more and more practitioners are applying PROMETHEE with the help of PROMCALC, the software to handle their multiple criteria problems without being aware of the consequences of the model assumptions made. Hence, they have provided a short overview of some drawbacks of PROMETHEE methods. Parreiras et al. (2006) compared three decision-making methods – simple multi attribute rating technique using swings (SMARTS), PROMETHEE, and a fuzzy decision algorithm and concluded that both SMARTS and PROMETHEE cannot always be applied to non-convex problems. Similarly, many researchers have identified the following shortcomings:

Ranking, not ratings. Waeyenbergh et al. (2004) mentioned that because of the outranking principles, no independent ratings, but rankings, are only produced by PROMETHEE. Moreover, they noted that the inclusion of new alternatives and criteria requires the repetition of pair-wise comparisons for re-establishing a ranking order. But, repetition of comparisons has to be performed even in other MCDM models too, if alternatives and new criteria are added.
Black box approach. Wolfslehner (2007) commented that though PROMETHEE is very powerful in the communication of results and can be easily used, it lacks structural features. He observed that these outranking models are rather uncommunicative and tends to behave as a black-box to the non-professional user.
Uncertainty. Hyde et al. (2005) discussed about various sources of uncertainty in the application of PROMETHEE methods especially during the definition of criteria weights and the assignment of criteria performance values. They explained that generalised criterion functions were incorporated in PROMETHEE to take the uncertainty in the criteria performance values into account; however, users find it extremely difficult to select the generalised criterion functions and their associated thresholds for each criterion, resulting in an additional source of uncertainty. Hence, to overcome this, they proposed a reliability-based approach which enables the decision maker to examine the robustness of the solution obtained from PROMETHEE.
Computational limitations. Marinoni (2006) recognized that outranking methods such as PROMETHEE, ELECTRE are subject to computational limitations with respect to the number of decision alternatives. They explained that when these models work with large raster datasets, they reach their computational limits quickly. Hence, to overcome this limitation, he presented an iterative approach which enables a modeller to easily and transparently apply outranking methods with practically no limit in the sizes of the raster datasets.
In accurate preference functions. Lahdelma and Salminen (2007) explained that preference information in real-life MCDM problems is always more or less inaccurate, imprecise or uncertain. Sometimes preference information can be missing. To overcome this problem, they discussed various methods for representing different kinds of incomplete preference information through probability distributions for preference parameters and have shown how to treat this information in MCDM methods through simulation techniques.
Others. Moffett and Sarkar (2006) have listed out the following disadvantages of PROMETHEE:
- PROMETHEE I requires the calculation of a value function for each criterion. As such, the method assumes substantially more of a decision maker than do most other outranking methods.
- Since, PROMETHEE I rely upon value functions, it proves to be a less objective method than the alternatives models available.
- PROMETHEE I requires the attribution of quantitative weights to the criteria; however, it does not provide a clear method by which to assign these weights.
- Since, PROMETHEE II involves assigning a single quantitative value to each alternative, it requires the assumption that the criteria are commensurable.

A cursory review of these shortcomings revealed that researchers are attempting to overcome these shortcomings by finding new approaches (Hyde et al., 2005; Marinoni, 2006; Lahdelma and Salminen, 2007) or by upgrading the basic version of PROMETHEE (for instance, the current version is PROMETHEE V). In this paper, too, an attempt has been made to overcome some of the shortcomings identified by researchers such as Hyde et al. (2005), Wolfslehner (2007) and Moffett and Sarkar (2006). The algorithm of PROMETHEE I and II have been explained in a step-by-step manner, which will eliminate the black box nature of PROMETHEE to a certain extent. Similarly, the guidelines developed by Routroy and Kodali (2007) were presented to assign the generalised criterion to the factors/elements, which can reduce the uncertainties identified by Hyde et al. (2005) and Moffett and Sarkar (2006). Thus, it can be concluded that though PROMETHEE suffers from some shortcomings, the advantages it possess makes it a valuable tool for modelling MCDM problems.

7 Conclusions

This paper started with the following questions: how would the managers or executives of the organisations have made a decision of implementing either an advanced manufacturing management philosophy or a technically sophisticated system in their organisations as part of their manufacturing strategy? What factors would they have considered in making such decisions when so many alternatives are available? What decision-making methodology the managers would have used to make such decisions? It was found from the literature review that such questions have not been addressed properly neither in the field of CIMS nor in LMS. To answer these questions, a case study of a SME involved in valve manufacturing was presented in which the managers and executives of the organisation were contemplating on implementing either CIMS or LMS. To make such a strategic decision, analysis was carried out from different perspectives like cost/financial, organisational, impact of employees, suppliers, shareholders and top management apart from the expected benefits. The factors or elements, which were identified from the literature review, were classified based on these perspectives. Since, it represented a multi-criteria problem, a MCDM model, namely, the PROMETHEE II was utilised. A detailed step-by-step methodology was presented and from the extensive analysis, it can be concluded that the LMS is the best under the circumstances for the developed case situation. Though PROMETHEE methodology has certain shortcomings, researchers are attempting to identify a solution by proposing a new model or approach or by developing an updated version of the methodology to resolve those shortcomings. Hence, it is believed that this paper will provide an opportunity for the managers to understand the concept of PROMETHEE and apply the same in real-life situation to supplement their decision-making efforts.

Equation 1

Equation 2

Equation 3

Equation 4

Equation 5

Equation 6

Equation 7

Equation 8

Figure 1The components of CIM

Figure 2Six possible types of generalised criterion

Figure 3Positive outflow of the alternatives for the criterion –TRC

Figure 4Negative outflow of the alternatives for the criterion –TRC

Figure 5Net outflow of the alternatives for the criterion –TRC

Figure 6Overall positive, negative and net outflow for the alternatives

Table IA summary about the case organisation

Table IIList of elements for the selection of suitable manufacturing systems

Table IIISummary of PROMETHEE applications in various fields

Table IVClassification and assignment of preference function for each element

Table VPerformance matrix

Table VIPositive and negative and net flow for each alternative with varying weights

Table VIIPartial pre-orders for the alternatives

Table VIIIComplete pre-orders for the alternatives along with its rank

Table IXPositive, negative and net flow for each alternative with equal weight

References

Abu-Taleb, M.F., Mareschal, B. (1995), "Water resources planning in the Middle East: application of the PROMETHEE V multicriteria method", European Journal of Operational Research, Vol. 81 pp.500-11.