CONCURRENT ENGINEERING: Research and

CONCURRENT ENGINEERING: Research and Applications Team Formation in Concurrent Engineering Using Group Technology (GT) Concepts Esra Aleisa,1,* Nallan C Suresh2 and Li Lin3 1

Industrial and Management Systems Engineering Department, College of Engineering and Petroleum, Kuwait University, Khaldia Bldg. # 8KH, 3rd Floor, P.O. Box 5969, Safat 13060, Kuwait 2 Operations Management and Strategy, School of Management, University at Buffalo, 326F Jacobs Management Center, Buffalo, NY 14260-4000, USA 3 Department of Industrial and Systems Engineering, 438 Lawrence D. Bell Hall, University at Buffalo, The State University of New York, Buffalo, NY 14260-2050, USA Abstract: Concurrent engineering (CE) has emerged as an essential design principle that facilitates rapid and efficient product development that is necessary to survive in today’s fiercely competitive environment. However, the lack of tools that address design team formulation and proper design task assignment in CE has made its application an immense challenge. In this article we explicitly address this issue by pointing out the relevance of its structure to cell-formation problem in cellular manufacturing (CM) and group technology. Particularly, interdisciplinary CE teams will be formulated as multifunctional machine-cells that process similar parts, in our case, design tasks requiring involvement of similar individuals. The fruitful marriage of the two disciplines, concurrent engineering and cellular manufacturing, opens the path towards utilizing the ample existing efficient algorithms of cellular manufacturing to concurrent engineering rather than re-inventing the wheel in developing methods in this respect. The paper in-hand also incorporate sequences (or routing in cellular manufacturing) to create work teams that can conduct similar tasks commonly known as work packages in a smoother manner that avoids unnecessary back and forth alterations and overlapping. In addition, to enable solving the task assignment problem in reasonable time for relatively larger problems, the developed framework integrates a fuzzy ART algorithm proven efficient in cellular manufacturing literature. The article also provides an illustrative example on executing the methodology on a design example and highlights its implications on task team assignments. Key Words: concurrent engineering, team formation, group technology (GT), cellular manufacturing, sequence-based clustering, fuzzy artificial neural networks.

1. Introduction Concurrent engineering (CE) is a design principle that incorporates various aspects of a product life-cycle at an early stage of the design [1–3]. CE benefits have been realized across many practices that led to shorter development times, improved product quality, and lower development costs, which are essential to survive in today’s fiercely competitive market [1,4]. Krezner [5] indicates that organizations no longer have the luxury to perform work in series. His research also indicates that this style of operation requires individuals from R&D, engineering, and production to be all actively involved in early project phases. This indicates that CE is by large a managerial challenge [6,7]. A successful CE application requires efficient and flexible formation

*Author to whom correspondence should be addressed. E-mail: [email protected]; [email protected] Figures 1, 4, 5 and 6 appear in color online: http://cer.sagepub.com

and organization of design teams [3,7–10] and utilization of the most advanced technology [11–13]. This aspect is of a particular importance. Specifically, that recent literature indicates that concurrent projects in projectized institutions has become a modus operandi (method of operation) [14,15]. Temporary teams are created constantly on transitory short-term bases to accomplish specific projects or work packages. The ideas of this research could be adopted to create work packages (group of similar tasks) to be assigned to projects. The methods of scheduling and software are relatively mature and available for individuals and institutions. Schedules define the sequence and time frame within which a project should be accomplished. However, the formation of work teams according to work schedules still lacks a scientific procedure and is basically formed using soft science. Our research addresses this particular need by integrating the ample methods available in cellular manufacturing, rather than reinventing the wheel in this respect. When the schedules are fed to our developed methodology as sequences, better work packages could be created and, therefore, better task

Volume 19 Number 3 September 2011 1063-293X/11/03 0213–12 $10.00/0 DOI: 10.1177/1063293X11418014 ß The Author(s), 2011. Reprints and permissions: http://www.sagepub.co.uk/journalsPermissions.nav

213

214

E. ALEISA ET AL.

assignment to teams could be established, a key to excel in today’s team based work environments. Trends in team formation for CE have been inclined towards supporting concurrent product development across geographically dispersed teams [16] and encompassing human resource management, supply chain, and marketing information into its formulation [17]. Although there have been widely used tools to facilitate CE design for assembly [18], design for serviceability [19] and design for compatibility [6], there has not been tools specialized in interdisciplinary design teams’ assignment and coordination that incorporates sequences for CE [12]. In this regard, we intend to provide methods for efficient CE team formation, by utilizing the ample advances and tools from the field of group technology (GT). We achieve that by modeling design tasks as manufacturing parts and team-members as machines in cellular manufacturing (CM) settings. Such that, similar design tasks are grouped into clusters and modeled as part-families that need to be processed using a specific machine-cell, where a machine-cell corresponds to a particular design team in CE context. As in CM, each manufacturing cell consists of a number of functionally dissimilar machines [20,21]. This dissimilarity of machines within a specific cell corresponds to the interdisciplinary nature of team formation classically undergone CE [9,22]. The former research provides deliberate review and procurements of the extensive cellformation literature, led to the conclusion that sequence based cell-formation clustering is most suitable for the formation of task families, commonly known as work packages and multi-functional design teams. Sequencebased cell-formation in CM contains sophisticated methods that can tractably establish efficient teamtasks assignments without inducing feedback coupling, which will result in more realistic team formations. In this article, we apply a fuzzy artificial neural network algorithm that has been developed in Suresh et al. [23] and was originally intended for sequence-based partfamilies and machine-cell-formation. This algorithm was chosen mainly due to its ability to eliminate cycles in concurrent engineering, which in turn eliminates redundancy and unnecessary back and forth task coupling. Also, the fuzzy ART was tested in CM literature and was shown to handle industrial size problems (typical to cellular manufacturing) in tractable time while eliminating exceptional elements. Many cellular manufacturing algorithms could have been adopted, but these do focus on other operational aspects relevant to cellular manufacturing but not necessarily concurrent engineering. For instance, incorporation of setup times, flexible assembly production lines and robotics. In addition, it is used here due to its added convenience of enabling analyst to adjust the similarity sequence threshold as appropriate and compare the results of team formations accordingly.

The article is organized as follows: first, we review GT cell-formation methods in general and sequence-based cell-formation in particular. Next, we reformulate the team assignment problem in CE as a CM problem and adapt the fuzzy artificial neural network sequence-based cell-formation to fit the objectives of team formulation and assignment. Finally, we apply the developed approach on an illustrative CE task-team formation and assignment problem and discuss the results.

2. Group Technology and Cellular Manufacturing Originally proposed by Mitrofanov [24] and Burbidge [25], Group technology (GT) is a manufacturing concept that is principled on grouping parts that are similar in their design, geometry or processing requirements [26]. Cellular manufacturing (CM) is the application of GT concepts on the facility shop-floor. CM is intended to facilitate and accommodate the norm of demands of greater variety and smaller order volumes [27]. A basic step in CM is the recognition of machine components in a process referred to as cell-formation. A manufacturing cell is an independent group of functionally dissimilar machines that manufacture a family of similar products [21,28]. Methods to form machine-cells and product families have been extensively addressed in the literature and usually employ two basic steps, the metric of similarity and an algorithm for converting this measure to relationship among cluster [29]. Along these lines a wide latitude of methods were developed for cellformation which then has been classified to either product flow analysis (PFA), array based methods (ABM), similarity coefficient methods (SCM) and many others. An interested reader can look up references [26,28,30,31] for an exhaustive list of classified cell-formation methods. Recent advances incorporate practical consideration such as dynamic cell configuration, alternative routing, sequence of operation, capacity, workload, and setup times [32]. In addition, research in this area have applied advanced tools to classify parts using simulated annealing hybrid algorithms [33] and fuzzy ART neural networks for machine-part clustering based on sequence data [34–37]. The latter is discussed in further detail in the next section. 2.1 Sequence-Based Cell-Formation Incorporating process sequencing requirements when identifying part-families and machine-cells has earned considerable attention in the literature of CM. This is due to its contribution in forming more realistic machine-cells that have smoother flow patterns with less backtracking in the material handling system and reduction in work-in-process (WIP) levels [23]. Here, we

Team Formation in Concurrent Engineering Using Group Technology (GT) Concepts

215

Table 1. Literature review summary of operation sequence-based clustering for the cell-formation problem. Contributors

Contributions

Selvam & Balasubramanian [38] Vakharia & Wemmerlo ¨ v [39] Choobineh [44]

First to develop a heuristic procedure for cell-formation based on approximate similarity in processing sequences. This was not practical for large size problems. Developed a two stage procedure for cell-formation based on similar sequences of some length L between two parts. But cumbersome for real world large problems. Proposed similarity-based coefficient using Lavenshtein distance measure and dynamic programming. Cumbersome for typical industrial applications. Developed a similarity coefficient based on identifying routing sequences that contains sequences of other parts. Proposed a procedure to minimize backtracking. Similar to Vakharia & Wemmerlo ¨ v [46] but considered probabilistic operation reallocation. Developed a cell-formation approach for class of problems that require exceptional elements to flow in one direction using 0-1 integer programming model. Developed a new generalized machine-cell-formation technique considering production sequences, lot sizes, processing times, of machine grouping, and the plant layout via the self-organizing feature maps (SOFM). Presents a heuristic for the machine-part grouping problem which incorporates relevant production requirements such as routing sequence, production volume, unit handling size, unit processing time and cell size. Used non-hierarchical clustering methods for sequence-based cell-formation. Developed bond-efficiency measure for sequence based clustering. Identified part-families based on similarity of operation sequences using pattern recognition based on artificial neural networks and using Fuzzy ART neural networks. Performed hierarchical clustering of operation sequences by identifying common substring of operations in the operation sequences. Developed a similarity coefficient that incorporates alternative process routing, operation sequence, operation time, and production volume factors. Developed a methodology for forming machine-cells and part-families based on sequence in a unidirectional material handling facility using a divide and conquer algorithm. Conducts thorough experimental procedures to compare the performance of Fuzzy ART neural networks with traditional hierarchical clustering methods for large-scale industrial-type data sets. Extended the Fuzzy ART neural networks approach to process integer values rather than binary representation to be used in within the clustering analysis for machine cells in group technology. Developed an algorithm CLASS that identifies sequence of machine and layout design of machine cells within the cellular layout context.

Tam [45] Vakharia&Wemmerlo ¨ v [46] Kang &Wemmerlo ¨ v [47] Dahel [40] Jang & Rhee [31]

Hwang & Sun [41]

Nair &Narendran [42] Suresh et al. [23]

Irani et al. [48] Yin & Yasuda [43] Lee & Chiang [51]

Park and Suresh [34]

Ozdemir et al. [35]

Mahdavi and Mahadevan [37]

review the literature of sequence-based clustering and discuss its relevance to the task-team assignment problem in CE. A summary of the literature review is provided in Table 1. Most research in sequence-based clustering proposes a mathematical formulation for incorporating more shop-floor operational characteristics such as set-up times, volumes, lot sizes, processing times, routings, etc. [31,38–43]. Another trend in this area is to propose a novel similarity measure based on sequence [23,44–48]. However, most sequence based clustering approaches are combinatorial and is tedious when applied for large-scale problems, which encouraged the resolution to AI-based techniques such as, neural networks and fuzzy theory [23,34–37,49,50]. In this research, we shall pursue fuzzy neural network for sequence-based clustering developed by Suresh et al. [23] to formulate task-team assignment

problem in CE. This is due to its convenient matrix representation of processes’ (task) sequences, which could be readily adapted for team formation. In addition, this approach enables the use to manipulate threshold values and its capacity to handle large industrial-type data sets with relatively low execution time, and superior solution quality. Experimental analyses that show the advantages of this approach are provided in references [34,36].

3. Adapting Sequence–Based Clustering of CM to Be Used for Task-Team Assignments in CE In this section, we reformulate the classical cellformation part – machine incidence matrix to represent task-team-member interaction in an attempt to be able to utilize cell-formation techniques of CM to team formation in CE.

216

E. ALEISA ET AL.

3.1 The Task-Team-Member Incidence Matrix The classical part-machine incidence matrix is defined as:  aij ¼

1 0

if part i requires machine type j otherwise

ð1Þ

for all i ¼ 1,2, . . . n and j ¼ 1,2, . . . m we shall replace part i with task i and machine type j with specialized team-member j. Accordingly, Equation (1) will be modified to represent task-team-member incidence matrix Amn ¼ ½aij :  aij ¼

1 if task i requires team member type j 0 otherwise

ð2Þ

The above formulation allows general GT concepts and most popular cell-formation methods such as rank ordering clustering algorithm (ROCA) [52], Direct clustering algorithm (DCA) [53], Bound energy algorithms (BEA), similarity coefficient based approaches such as single linkage clustering algorithm (SLCA) [54], and others to be utilized for task-team assignment in CE [55]. However, in CE it is necessary to maintain task sequence requirements. Task sequence corresponds to part routings; therefore, cell-formation algorithms that emphasize part routing are necessary to create reasonable task-team assignments in CE.

3.2 Representation of Task Sequence Requirements Routing sequence refers to the sequence of machine types required by a part [46]. To incorporate sequences, Suresh et al. [23] have modified the traditional incidence matrix definition given in Equation (1) as follows: 8 > < k is the routing index by which part i a ij ¼ uses machine type j > : 0 if part i does not require machine type j ð3Þ Adapting the above equation according to the A matrix defined in Equation (2), results in the following sequence-based task-team-member incidence matrix  mn ¼ ½a ij : A

a ij ¼

8 > > > > > < have to precede the ¼ contribution of team member q > > > in operation sequence of task i > > > : 0 otherwise ð5Þ

Equation (5) forms a suitable platform to construct task-team-member assignments using CM sequencebased clustering such as Suresh et al.’s [23] fuzzy ART ANNs approach for sequenced based cell-formation or any other reasonable substitute thereof.

4. Applying Fuzzy Neural Network for Sequence–Dependent Clustering to Task-Team Assignments of CE Fuzzy ART neural network is based on unsupervised learning [56]. Therefore, the network exhibits no a priori knowledge of the number of task families (class) that exists nor that attributes of each cluster of similar tasks. Furthermore, no training is provided to acquire correct responses to the network [57,58]. In addition, Suresh et al. [23] algorithm is operated as a leader algorithm. Hence, as each task sequence is fed, the network clusters each of these sequences into a distinct class. An exemplar vector is created for each class. Within some specified limit, referred to as the vigilance threshold, if a new input is similar to an existing exemplar, that input is classified under the same category of that exemplar, and the matching exemplar is updated accordingly. Otherwise, the exemplar will be categorized as a new one. When all the inputs are fed to

Team Formation in Concurrent Engineering Using Group Technology (GT) Concepts

Output layer

Input layer

Precedence matrix

this application, the vigilance parameter represents the threshold upon which exemplars are considered similar. That is the higher the value of  the larger the number of task families to be identified. In the algorithm, j ¼ 1,2, . . . U correspond to the index of task families, where U is the total number of clusters of similar tasks identified. Within the algorithm the array of task sequencing is fed to the algorithm and is converted to a precedence matrix, PMi . Each PMi is fed to the lower-level neurons (input neurons). Accordingly, the output value Tj is computed using:      Tj ¼ PM ^ wj  =  þ wj  ;

Design tasks

Team members

Routing

Routing sequence

Figure 1. The layers for the fuzzy ART network for sequencebased clustering.

the network, several exemplars are created each of which represent a cluster of task that require similar teammembers (task families). The layers of the fuzzy ART network are shown in Figure 1. As illustrated, it consists of two layers of neurons. The lower layer, the input or comparison layer, consists of neurons that are fully connected to the upper (output) layer through weight vectors. The lower layer interacts with the routing input corresponding to the team-members ð1  j  mÞ that are required to work on some design tasks. As shown in Figure 2, the lower layer consists of a matrix of neurons, each of which interfaces with a particular PMi matrix ð1  i  nÞ, where families of similar design tasks construct the upper layer neurons in the artificial neural network. The upper layer, the recognition layer, is dynamically constructed whenever a new team (class) is identified. Associated with each of these neurons is an exemplar vector that designates a common sequence of design task processing.

4.1 The sequence-Based Clustering Fuzzy ART Algorithm Here, we explain the fuzzy ART sequence based clustering algorithm that uses the PMi to produce clusters of task families that require similar processing from team-members and accordingly generates suitable teams to assign to these tasks in CE settings (see Figure 2). The algorithm is shown in Figure 3. It starts by initializing: the weight vectors that connect upper and lower level neurons to the value of one; and the specification of the fuzzy ART parameters. These are the choice parameter ð > 0Þ, the learning parameter ð ¼ ½0,1Þ, and the vigilance parameter ð ¼ ½0,1Þ. In

217

ð6Þ

where ^ designates the auzzy AND operator: ðx ^ yÞ ¼ minðxi ,yi Þ, and wj denotes the weight of node j. For every upper-level neuron j ¼ 1,2, . . . ,U, the neuron that exhibits the largest Tj and satisfies the pre-specified similarity parameter, , is appended to the current family of tasks. The best-matching exemplar is updated accordingly by modifying the associated weight vector. This process is repeated until all task processing sequences are fed to the algorithm. Experimental verification for this algorithm and the effect of the parameters , , and  are provided in Suresh et al. [23]. An application for the algorithm shown in Figure 3 in the concurrent engineering context will be demonstrated in the next section.

5. Illustrative Example Consider a problem of a concurrent manufacturing facility design that consists of 46 tasks given in Table 2. The design tasks ði ¼ 1, 2, . . . 46Þ given in Table 2 need to be addressed by 30 individuals ðj ¼ 1,2, . . . 30Þ from eight departments ðD1 , D2 , . . . D8 Þ. The eight functions correspond to finance and administrative services, marketing, product design, process planning and scheduling, storage and warehouses management division, human resources and safety division, construction and project administration, and the architectural and infrastructure department. The distribution of individuals in each of the aforementioned departments is: D1 ¼ f1,2,3,4g; D2 ¼ f5,6g; D3 ¼ f7,8,9,10g; D4 ¼ f11,12,13,14g; D5 ¼ f15,16g; D6 ¼ f17,18,19g; D7 ¼ f20,21, 22,23,24,26g; and D8 ¼ f27,28,29,30g. The sequence in which the design tasks need to be addressed by the individuals from the above departments is provided in Table 3. Prior to running the neural network, the choice parameter (1  i  n), the learning parameter ð ¼ ½0,1Þ, and the vigilance parameter ð ¼ ½0,1Þ need to be identified. As indicated earlier, the vigilance parameter represents the threshold upon which exemplars are considered similar.

218

E. ALEISA ET AL. Identify design task and project individuals

CE principles

Identify process sequencing based on CE principles

Create sequence based task-team– member incidence matrix A 1 1 2 3 4 3 2 4 3 1 2

2 5 4

4

31

5

For each task (1 ≤ i ≤ n) create PMi matrix 1 1 1 ⎛1 ⎛1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ⎝1 1 1 1 1 1 1 ,⎝

⎛

⎜

⎜

⎝

⎜

⎛

⎜

⎝

Specification of Fuzzy ART learning parameters

2 5 1 1

... n

Upper layer (task families) Fuzzy ART algorithm for sequencebased grouping

Lower layer (team-members)

Team formation and assignment to task families Figure 2. Team and task formation using sequence-based cell-formation fuzzy artificial network method.

5.1 Application of the Fuzzy ART Neural Network Algorithm Here, we feed the routing sequence provided in Table 3 to the fuzzy Art neural network. The code for the neural network is written in Java programming language. Prior to running the neural network, the choice parameter (1  i  n), the learning parameter

ð ¼ ½0,1Þ, and the vigilance parameter ð ¼ ½0,1Þ need to be identified. As indicated earlier, the vigilance parameter represents the threshold upon which exemplars are considered similar. That is the higher the value of  the larger the number of task families to be identified. The effect of the choice and learning parameters will be further tested in the experimental design section later in this article.

Team Formation in Concurrent Engineering Using Group Technology (GT) Concepts

219

Initialization: Set values of a > 0, b, r ∈[0,1], wjkl = 1(∀j, k, l )

Read input: convert routing to precedence matrix PMi (1 ≤ i ≤ n)

For every upper or lower level neuron j compute: Tj = [⏐PM ∧ wj⏐] / [a +⏐wj⏐] , ∀j

Select the upper-level neuron whose with an exemplar that best matches the input exemplar Tq = max {Tj }

Tj = [⏐PM ∧ wq⏐] / [⏐PM⏐] ≥ r

Resonance test: degree of similarity among best matching exemplar

Select new upper -level neuron and exemplar Set Tq = –1

Learning law→ update best-matching exemplar wqnew = b [PM ∧ wqold ] +[1=b]wqold

Figure 3. Fuzzy ART algorithm for sequence-based clustering (adapted from Suresh et al. [23]).

The result of using parameters  ¼ 0:1,  ¼ 0:1, and  ¼ 0:1 results the creation of six work groups or teams. The tasks within each work group need to be accomplished concurrently by a work team. In other words, each work group corresponds to a work package that can be addressed separately i.e. by a separate team, department, company or a contractor. In another run, using  ¼ 0:1,  ¼ 0:1, and  ¼ 0:4 results in seven work groups/task families, while using  ¼ 0:1,  ¼ 0:1, and  ¼ 0:7 results in nine work groups. The values ,, and  of these parameters need to be adjusted until a reasonable plan for work groups or task families is created based on reasonable work load balance that is distributed among work teams. For this particular example, with both and  set to a value of 1.0 the rate of increase in the number of workgroups is depicted in Figure 4.

vigilance parameters on the number of task families and the maximum number of tasks per family as these criteria mostly affects the workload balancing effort. The analyses are carried out using a multilevel factorial design that consists of three factors namely: ,, and . Each of these factors consists of five levels: 0.1, 0.25, 0.5, 0.75, and 0.9. The design is replicated once. Two responses are associated, response 1: The number of task families (n) and response 2: the maximum number of tasks per family (m). Thus the design is 53 full factorial design with 125 runs. Using a confidence level of 5%, the results show that only  and  are significant with respect to both responses, while the  is insignificant. The regression model for the first response is: n ¼ 1:69 þ 0:09 þ 14:4 þ 21:6,

ð7Þ

where S ¼ 2.10612, R2 ¼ 93.3%, R2(adj) ¼ 93.2%. The regression model for the second response is: 5.2 Results and Analysis m ¼ 18:1  0:297  6:10  11:7, In this section, we conduct further analyses to investigate the effect of the learning, choice, and

where, S ¼ 2.21707, R2 ¼ 76.5%, R2(adj) ¼ 75.9%.

ð8Þ

220

E. ALEISA ET AL.

Table 2. Tasks of designing a manufacturing facility. i

Task

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Definition of facility objectives, corporate strategy, customer, investment policy Specification of general product lines Planning of production and operation strategy Identification of product volumes and varieties and capacity growth planning Human Resource planning and labor force composition Component list of products CAD drawings of manufactured parts (dimensions, tolerances) Material composition for manufactured parts Bill of materials Design of manufacturing operations Type of workstations Number of workstations and preliminary assembly line balancing Determination of on-line/off-line inspection methods Specification of shop floor control devices (Hand written report, manual data entry terminals, bar codes, cameras, voice data entry systems, etc.) Production schedule design and policies (min WIP, idle time, setup time, average flow time or meet due dates, etc.) Layout type (product, processes, GT, CM, etc) Conduct layout analysis (interrelationship, flow patterns, flow, layout method) Estimate WIP/Scrap (workstation) Layout simulation Determine non-production areas and their relationships Determination of required configuration non-production departments Transported material condition/orientation/speed/unit load size (workstation 1, workstation 2) Material handling equipment for production area, warehouse and storage (unitizing equipment, conveyors type, sortation equipment, storage retrieval and stacking, transportation equipments and trucks) Degree of automation Product arrangement principles for warehouse and storage (popularity, similarity, size characteristics, etc.) Future space requirements for production area, warehouse and storage Amount of safety stock required for warehouse and storage Finished product material condition for warehouse Warehouse and storage operations Select storage equipment for warehouse and storage Aisle arrangement/allowance, cubic space required for production area, warehouse and storage Shipping/receiving operations for warehouse and storage Design of docking areas for warehouse and storage Establishment of management policy for warehouse and storage Identify security precautions for plant Plan general accessibility means Design atmospheric system (regulating temperature, humidity, noise, etc.) Design structural system (steel skeleton, column spacing, height clearance, etc.) Design enclosure system (walls, roofs, doors, windows, etc.) Design electrical/lighting system Design sanitation system Prepare plot plans for facility total site Prepare blue print for facility total site

15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

Since only  and  proved to be significant with respect to both responses, here we show the surface plots for these factors with respect to response 1. These are shown in Figure 5. Figure 5 shows that higher values of  and  creates a larger number of parts families, while smaller values of  and  results in bigger families. An expert investigation to the number and size of part families to promote a balanced workload among work teams for the illustrative example in hand indicates that 6–11 task families or teams is adequate. To obtain this level, we need to pick  and  values that will produce this range. The contour plot in Figure 6 could assist in obtaining theses values.

6. Conclusions Concurrent engineering in projectized institutions has become a modus operandi (method of operation). Institutions are running hundreds of projects by temporary teams that are created on transitory shortterms bases to accomplish specific tasks. The developed framework in this article addresses the idea of assigning group of tasks, commonly referred to as work packages to individuals and forming teams accordingly. The assignment is aimed to provide seamless accomplishment of tasks by incorporating sequences or work schedules. We achieved that by formulating the

Team Formation in Concurrent Engineering Using Group Technology (GT) Concepts Table 3. Task sequencing for design of manufacturing facility example. Task

Sequence

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43

51,2,3,4,5,6> 52,3,4,5,6,7,8,11,12> 51,2,3,4,17,18,18> 57,8,9,10> 55,6,7,8,9,10> 5,7,8,11,12,13,14> 59,10,11,12,13,14> 53,4,11,12,13,14> 511,12,13,14> 511,12,13,14> 57,8,11,12,13,14,16> 59,13,14,15,16> 54,5,15,16> 515,16,15,2,5,1> 515,17,16,18> 57,15,17,15,18,3,2,1> 511,15,12,16> 515,17,20,27> 518,28,21> 530,23> 518,28,21> 530,23>

Task

Sequence

2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42

51,2,3,4,5,6,7,8> 51,2,3,4,5,6,7,8,11,12> 55,6,7,8,9,10> 54,5,6,7,8,9,10> 56,9,10,11,12,13,14> 54,5,11,12,13,14> 53,7,8,11,12,17,18,19> 53,4,6,7,8,9,11,12,13,14> 511,12,13,14> 517,18,18,20,21,27,28> 59,13,14> 54,7,8,11,12,16> 515,16,1,2,5> 57,15> 515,17,16,18,2> 511,15,12,16> 52,3,15,17,16,18> 515,17,27,20> 519,29,22> 519,29,22> 528,26>

Table 4. Task families or work groups. Work group or Task family

Tasks

Task families or work groups for  ¼ 0:1,  ¼ 0:1, and  ¼ 0:1 1 2 3 4 5 6

123458 6 7 9 10 11 13 12 14 15 16 17 18 19 20 21 22 23 24 25 26 27 29 30

Task families or work groups for  ¼ 0:1,  ¼ 0:1, and  ¼ 0:4 1 2 3 4 5 6 7

Number of workgroups

35 Number of workg roups

1 2 3 4 5 6 7 8 9 10 11 16 24 12 13 14 15 17 18 19 21 22 23 32 33 20 37 41 25 26 27 29 30 31 34 35 36 28 38 40 39 43

30 25 20 15 10 5 0 0

0.2

0.4

0.6 r

0.8

Figure 4. Number of workgroups with respect to parametersand  set to a value of 1.0.

1

1.2

221

222

E. ALEISA ET AL.

Acknowledgements The authors would like to thank the office of vice president for research and the research administration at the Kuwait University for funding this project under Grant Number E102/08.

Response 1

30

20

10 0.00

0.25 0.50 Beta

0.75 0.50 0.25 Rho 0.00

0.75

Figure 5. Surface plot for response 1 (number of task teams) vs q and b.

Contour plot of number of task families versus r and b 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.9 Response 0.9 0.8

0.8

0.7

0.7

0.6

0.6

b 0.5

0.5

0.4

0.4

0.3

0.3

0.2

0.2

0.1 0.1

0.2

0.3

0.4

0.5 r

0.6

0.7

0.8

1 < 10 10 - 15 15 - 20 20 - 25 25 - 30 > 30

0.1 0.9

Figure 6. Contour plot for the number of task families or teams with respect to q and b.

concurrent engineering task-team assignments using cell-formation models in cellular manufacturing and group technology through revealing the relevance between the two problems’ structures. Particularly, the nature of interdisciplinary designs teams in concurrent engineering resembles the idea of mulwwti-functional machine-cells in cellular manufacturing. This formulation avails most cell-formation algorithms to be utilized for this purpose. When schedules are fed to our developed framework as sequences, better work packages could be created and, therefore, better task assignment to teams could be established, a key to excel in today’s team based work environments. The algorithm uses Fuzzy ART neural networks which is suitable for large scale CE applications. Future research is attributed to investigate whether other performance measures besides reducing task overlapping and improving tractability need to be incorporated. Ideas such as percent dedication of team members on other routine tasks or other projects and team work load balancing are also new research pathways in this regard.

References 1. Pennell, J.P. and Winner, R.I. (1989). Concurrent Engineering: Practices and Prospects, In: IEEE Global Telecommunications Conference and Exhibition Publ by IEEE, IEEE Service Center, Piscataway, NJ, USA: Available from IEEE Service Cent (cat(89CH2682-3), Vol. 1, pp. 647–655. 2. Nevins, J. L. and Whitney, D.E. (1989). Concurrent Design of Products and Processes, A Strategy for the Next Generation in Manufacturing, New York: McGraw-Hill. 3. Kusiak, A. and Park, K. (1990). Concurrent Engineering Decomposition and Scheduling of Design Activities, International Journal of Production Research, 28(10):1883–900. 4. Linton, L., Hall, D., Hutchison, K., Hoffman, D., Evanczuk, S. and Sullivan, P. (1992). First Principles of Concurrent Engineering: A Competitive Strategy for Product Development, Washinton D.C.: CALS/ Concurrent Engineering Electronic System Task Group. 5. Kerzner, H. (2009). Project Management A System Approach to Planning, Scheduling, and Controlling, 3rd edn, New York: John Wiley & Sons, Inc. 6. Ishii, K. (1993). Modeling of Concurrent Engineering Design, In: Kusiak, A.K. A. (ed.) Concurrent Engineering: Automation, Tools, and Techniques, New York: John Wiley & Sons, Inc. 7. Yassine, A. and Braha, D. (2003). Four Complex Problems in Concurrent Engineering and the Design Structure Matrix Method, Concurrent Engineering Research & Applications, 11(3): 165–176. 8. Smith, R.P. and Eppinger, S.D. (1997). Identifying Controlling Features of Engineering Design Iteration. Management Science, 43(3):276–293. 9. Chen, S.-J.G. and Lin, L. (2004). Modeling team Member Characteristics for the Formation of a Multifunctional Team in Concurrent Engineering, IEEE Transactions on Engineering Management, 51(2 May): 111–124. 10. Ke, L., Archetti, C. and Feng, Z. (2008). Ants Can Solve the Team Orienteering Problem, Computers & Industrial Engineering, 54(3):648–665. 11. Singh, N. (1996). Systems Approach to ComputerIntegrated Design and Manufacturing, New York: Wiley. 12. Jin, Y., Levitt, R.E., Kunz, J.C. and Christiansen, T.R. (1995). Virtual Design Team: A Computer Simulation Framework for Studying Organizational Aspects of Concurrent Design, Simulation, 64(3 Mar): 160174. 13. Gardoni, M. (2005). Concurrent Engineering in Research Projects to Support Information Content Management in a Collective Way, Concurrent Engineering-Research and Applications, 13(2):135–144. 14. Grey, C.F. and Larson, E.W. (2008). Project Management the Managerial Process, 4th edn, Boston: McGraw-Hill Irwin.

Team Formation in Concurrent Engineering Using Group Technology (GT) Concepts 15. Milosevic, D.Z. (2003). Project Management ToolBox: Tools and Techniques for the Practicing Project Manager, 1st edn, Hoboken, New Jersey: John Wiley & Sons, Inc. 16. Li, W.D. and Qiu, Z.M. (2006). State-of-the-Art Technologies and Methodologies for Collaborative Product Development Systems, International Journal of Production Research, 44(13): 2525–2559. 17. Tseng, K.C. and Abdalla, H. (2006). A Novel Approach to Collaborative Product Design and Development Environment, Proceedings of the Institution of Mechanical Engineers Part B-Journal of Engineering Manufacture, 220(12): 1997–2020. 18. Boothroyd, G. and Dewhurst, P. (1983). Design for Assembly: A Designer’s Handbook, Wakerfield, RI: Boothroyd Dewhurst Inc. 19. Makino, A., Barkan, P., Reynolds, L. and Pflaff, E. (eds). (1989). Design for Serviceability Expert System, In: SME Winter Annual Meeting: Concurrent Product and Process Design, December; San Fransisco, CA. 20. Waghodekar, P.H. and Sahu, S. (1984). MachineComponent Cell Formation in Group Technology: Mace, International Journal of Production Research, 22(6): 937–948. 21. Ham, I., Hitomi, K. and Yoshida, T. (1985). Group Technology: Application to Production Management, Hingham, MA: Kluwer-Nijhoff. 22. Chen, L. and Li, S. (2001). Concurrent Parametric Design Using a Multifunctional Team Approach, Proceedings of the ASME Design Engineering Technical Conference, 2: 287–296. 23. Suresh, N.C., Slomp, J. and Kaparthi, S. (1999). SequenceDependent Clustering of Parts and Machines: a Fuzzy ART Neural Network Approach, International Journal of Production Research, 37(12): 2793–2816. 24. Mitrofanov, S. (1966). The Scientific Principles of Group Technology, National Lending Library Translation. 25. Burbidge, J.L. (1975). The Introduction of Group Technology, New York: Wiley. 26. Irani, S.A. (1999). Handbook of Cellular Manufacturing Systems, New York, NY: Wiley. 27. Salum, L. (2000). The Cellular Manufacturing Layout Problem, International Journal of Production Research, 38(5): 1053–1069. 28. Waghodekar, P.H. and Sahu, S. (1984). MachineComponent Cell-Formation in Group Technology Mace, International Journal of Production Research, 22(6): 937–948. 29. Mosier, C.T. (1989). Experiment Investigating the Application of Clustering Procedures and Similarity Coefficients to the GT Machine Cell Formation Problem, International Journal of Production Research, 27(10): 1811–1835. 30. Singh, N. and Rajamani, D. (1996). Cellular Manufacturing Systems: Design, Planning and Control, New York: Chapman & Hall. 31. Jang, I.H. and Rhee, J.T. (1997). Generalized Machine Cell Formation Considering Material Flow and Plant Layout Using Modified Self-Organizing Feature Maps, Computers & Industrial Engineering, 33(3-4): 457–460. 32. Defersha, F.M. and Chen, M. (2006). Machine Cell Formation Using a Mathematical Model and a GeneticAlgorithm-Based Heuristic, International Journal of Production Research, 44(12): 2421–2444.

223

33. Tiwari, M.K. and Roy, D. (2003). Solving a Part Classification Problem Using Simulated Annealing-Like Hybrid, Robotics and Computer-Integrated Manufacturing, 19(5): 415–424. 34. Park, S. and Suresh, N.C. (2003). Performance of Fuzzy ART Neural Network and Hierarchical Clustering for Part-Machine Grouping Based on Operation Sequences, International Journal of Production Research, 41(14): 3185–3216. 35. Ozdemir, R.G., Gencyilmaz, G. and Aktin, T. (2007). The Modified Fuzzy Art and a Two-Stage Clustering Approach to Cell Design, Information Sciences, 177(23): 5219–5236. 36. Won, Y. and Currie, K.R. (2007). Fuzzy ART/RRR-RSS: a Two-Phase Neural Network Algorithm for Part-Machine Grouping in Cellular Manufacturing, International Journal of Production Research, 45(9): 2073–2104. 37. Mahdavi, I. and Mahadevan, B. (2008). CLASS: An Algorithm for Cellular Manufacturing System and Layout Design Using Sequence Data, Robotics and ComputerIntegrated Manufacturing, 24(3): 488–497. 38. Selvam, R.P. and Balasubramanian, K.N. (1985). Algorithmic Grouping of Operation Sequences, Engineering Costs and Production Economics, 9(1–3): 125–134. 39. Vakharia, A.J. and Wemmerlov, U. (eds). (1986). Designing a Cellular Manufacturing System: a Cell Layout Approach Using Operation Sequences, In: Proceesings of the Decision Science Institute, Hawaii. 40. Dahel, N.E. (1995). Design of Cellular Manufacturing Systems in Tandem Configuration, International Journal of Production Research, 33(8): 2079–2095. 41. Hwang, H. and Sun, J.U. (1996). A Genetic-AlgorithmBased Heuristic for the GT Cell Formation Problem, Computers & Industrial Engineering, 30(4):941–955. 42. Nair, G.J. and Narendran, T.T. (1998). CASE: A Clustering Algorithm for Cell Formation with Sequence Data, International Journal of Production Research, 36(1): 157–179. 43. Yin, Y. and Yasuda, K. (2002). Manufacturing Cell’s Design in Consideration of Various Production Factors, International Journal of Production Research, 40(4 Mar 10): 885–906. 44. Choobineh, F. (1988). A Framework for the Design of Cellular Manufacturing Systems, International Journal of Production Research, 26(7): 1161–1172. 45. Tam, K.Y. (1990). An Operation Sequence Based Similarity Coefficient for Part Families Formations, Journal of Manufacturing Systems, 9(1): 55–68. 46. Vakharia, A.J. and Wemmerlov, U. (1990). Designing a Cellular Manufacturing System - a Materials Flow Approach Based on Operation Sequences, Iie Transactions, 22(1): 84–97. 47. Kang, S.L. and Wemmerlov, U. (1993). A Work LoadOriented Heuristic Methodology for Manufacturing CellFormation Allowing Reallocation of Operations, European Journal of Operational Research, 69(3): 292–311. 48. Irani, S.A., Zhang, H., Zhou, J., Huang, H., Udai, T.K. and Subramanian, S. (2000). Production Flow Analysis and Simplification Toolkit (PFAST), International Journal of Production Research, 38(8): 1855–1874. 49. Kaparthi, S. and Suresh, N.C. (1991). Neural Network System for Shape-Based Classification and Coding of

224

50.

51.

52.

53. 54. 55.

56.

57.

58.

E. ALEISA ET AL.

Rotational Parts, International Journal of Production Research, 29(9): 1771–1784. Zhang, H. and Huang, S. (1995). Appliations of Neural Networks in Manufacturing: a State-of-the-Art Survey, International Journal of Production Research, 33: 705–728. Lee, S.D. and Chiang, C.P. (2002). Cell Formations in the Uni-Directional Loop Material Handling Environment, European Journal of Operational Research, 137(2): 401–420. King, J.R. (1980). Machine-Component Group Formation in Group Technology, Omega-International Journal of Management Science, 8(2): 193–199. Chan, H.M. and Milner, D.A. (1982). Direct Clustering Algorithm for Group Formation in Cellular Manufacture. McAuley, J. (1972). Machine Grouping for Efficient Production, Production Engineer, 51: 53–57. Pharn, D.T. and Afify, A.A. (2007). Clustering Techniques and their Applications in Engineering, Proceedings of the Institution of Mechanical Engineers Part C-Journal of Mechanical Engineering Science, 221(11): 1445–1459. Carpenter, G.A., Grossberg, S. and Rosen, D.B. (1991). Fuzzy ART. Fast Stable Learning and Categorization of Analog Patterns By an Adaptive Resonance System, Neural Networks, 7(6): 759–771. Suresh, N.C. and Kaparthi, S. (1994). Performance of Fuzzy ART Neural Network for Group Technology Cell Formation, International Journal of Production Research, 32(7 Jul): 1693–1713. Burke, L.I. and Kamal, S. (eds). (1992). Fuzzy ART for Cellular Manufacturing, Intelligent Engineering Systems Through Artificial Neural Networks, Fairfield, NJ: ASME.

Esra Aleisa Dr. Esra Aleisa is an assistant professor of Industrial and Management Systems Engineering (IMSE) at Kuwait University. She has received her Masters and PhD in Industrial Engineering and production systems from the department of Industrial Engineering at the State University of New York at Buffalo in 2001 and 2005 respectively. Her research interests includes, Planning and design of large scale facilities, multilevel planning and design of complex engineering design, group technology (GT), and design structured matrices (DSM), simulation and improvement of manufacturing and service systems, especially of that related to environmental areas such as wastewater treatment and reuse. She is a member of Omega Rho, the international operations research honor society, IEEE, INFORMS, IIE, ASEE.

Nallan C Suresh Dr. Nallan C. Suresh is UB Distinguished Professor and Chair of the Department of Operations Management & Strategy at UB School of Management. Dr. Suresh specializes in the areas of manufacturing, supply chain & logistics management. He is an internationally recognized scholar and his papers have been published in almost all the leading journals of these fields. Over the years, his research has been in four broad streams, including: 1) Economic evaluation of flexible manufacturing systems (FMS) and robotics; 2) Design and performance evaluation of cellular manufacturing systems; 3) Design and Performance of virtual manufacturing cells, and 4) Design and performance of global manufacturing and sourcing networks, and avoidance of disruptions in global supply chains. Dr. Suresh is particularly recognized internationally for his contributions to cellular manufacturing, including the resolution of a long-standing paradox through new analytical models, and an influential research volume in this field. Li Lin Dr. Li Lin is Professor of Industrial and Systems Engineering at University State University of New York at Buffalo, where he has been a faculty member since he obtained his Ph.D. in 1989 from Arizona State University. Dr. Lin’s research areas include manufacturing and healthcare systems simulation and concurrent engineering and design, including environmentally conscious design and manufacturing. Dr. Lin research has been supported by the National Science Foundation (NSF), the Environmental Protection Agency (EPA) and the Agency for Healthcare Research and Quality (AHRQ). He has also worked extensively with many manufacturing companies and healthcare organizations in helping them improve operational efficiency, cost and quality of products and services.