Plant Layout Computerised Design: Logistic and Relayout Program (LRP) PDF

Preview

Summary

Int J Adv Manuf Technol (2003) 21:917–922 Ownership and Copyright  2003 Springer-Verlag London Limited Plant Layout Computerised Design: Logistic and Relayout Program (LRP) E. Ferrari1, A. Pareschi1, A. Persona2 and A. Regattieri1 1 Department of Mechanical and Industrial Plants, Bologna University...

Description

Int J Adv Manuf Technol (2003) 21:917–922 Ownership and Copyright  2003 Springer-Verlag London Limited

Plant Layout Computerised Design: Logistic and Relayout Program (LRP) E. Ferrari1, A. Pareschi1, A. Persona2 and A. Regattieri1 1

Department of Mechanical and Industrial Plants, Bologna University, Italy; and 2Department of Technique and Management of Industrial System, Padova University, Italy

Markets are affected by strong competition in terms of continuous innovation of products and processes, high customer satisfaction and low cost of production. In order to achieve these strategic results it is recommended, or necessary, to rectify lack of organisation. For instance, the increasing costs due to material handling, force factories to check the facility layout and, when necessary, to improve it. The evaluation of a prospective improvement requires a large effort and a wellestablished skill. The study and the optimisation of plant layout is a strategic activity. This paper represents the last step of a research program on the automatic design of plant layout. The aim is to support the design activity of plant layout by means of an integrated approach, taking into account many criteria, both quantitative and qualitative. In particular, this paper presents a global approach, based on material flow and activity relationships. It is carried out using new software, called LRP (layout and re-layout program), introduced by the authors. A real application in a factory, specialising in the production of electronic devices, is presented. Keywords: Electronic devices; Facility layout; Flow of materials; LRP software; Real application; Relationships

1. Introduction The design process of plant layout is influenced by many factors which can have a great impact on the factory, e.g. plant-engineering, technology, management and also economicfinancial. The best solution must guarantee a harmonic synthesis and a well-balanced combination of all of these. During the last few years the systematic layout planning [1] (SLP) approach is being widely applied. SLP consists of:

Correspondence and offprint requests to: Professor A. R. Regattieri, DIEM – Plant Section, Faculty of Engineering, Bologna University, Via le Risorgimento 2, Bologna 40136, Italy. Received 10 April 2001 Accepted 9 April 2002

Data collection (i.e. product types, quantities, process, and plants). Block layout planning (with freedom inside blocks). Intra-block design. Plant realisation. It is very important to underline the relevance of data collection, as it represents the fundamental starting point of the procedure. Insufficient information can cause an explosion of the layout problem, and can give misleading conclusions. The focal point of the SLP procedure is block and intra-block design. Different authors have proposed some models with corresponding algorithms [1–4]. Fundamentally, the models proposed are focused on the flow of materials and the minimisation of transportation. Several solutions are based on the relationships among activities, and their relative closeness is the target function to maximise [1]. This simple approach is often inadequate for real situations involving more complex relationships among facilities, for which a deeper analysis is required. Some multicriteria approaches have been proposed [1,5,6] in order to take into account simultaneously both the relationships and the flow of materials in an integrated way. The use of a personal computer is required by the great amount of data to process in the layout study. Many software programs and techniques [7–9], based on single or multiple criteria, have been proposed. The critical point in most of them is the strictness of constraints derived from the analytical approach, which are normally too rigorous for a direct application in a real case. A real situation requires a “systemic approach” that takes into consideration a large number of factors. Because of this, and on the basis of the experiences of different layout studies in various industrial sectors, the logistic and re-layout program (LRP) [10], a new modular platform, is proposed. The LRP platform was developed in 1999 and it is in permanent improvement; this paper presents the current status of LRP. After a short overview of the literature (Section 2) on layout design, the most relevant features of LRP logic are described in Section 3. A representation of the scheme proposed is

918

E. Ferrari et al.

presented in Section 4. Section 5 focuses on a real application of LRP in a multi-plant factory operating in the field of electronic devices. Finally, conclusions are drawn in Section 6.

2.

Literature

Many studies have been made on the facilities layout problem. Fundamental formulation is due to Koopman and Beckmann [11] with their quadratic assignment problem (QAP). Based on QAP, different formulations are developed. Bazaraa [10] introduced the quadratic set covering problem (QSP), Lawler [12] the linear integer formulation, and Kaufman the mixed integer programming problem. Beyond QAP and their corollaries, non-linear formulations are introduced, e.g. Tam and Li model attempts to minimise an objective function defined as f = wij d2ij where wij represents the flow of materials and dij the distance between facilities. The solution of these models often requires very complex algorithms and a large amount of computational resources. Analytical methods can be classified as optimal, e.g. the wellknown branch and bound and cutting plane techniques, and heuristics. The exacting requirements of optimal methods, in terms of high memory and computer time, limit their application; the largest problem solved optimally is a layout with 15 facilities [1]. For this reason heuristics have become more widely adopted. They can be classified into construction, improvement, and hybrid types. Construction algorithms produce the solution ab initio without requiring any starting layout, as in the well-known CORELAP and ALDEP [13]. Improvement algorithms, such as CRAFT [14] and COFAD [15], start with an initial layout and try to improve it with facility exchanges. Hybrid approaches provide a first construction phase and a final improvement arrangement. Artificial intelligence (AI) has become a powerful tool; expert and hybrid systems, employing both algorithms and knowledge-based systems, have been developed, and studies have been made by Heragu and Kusiak [9] and Fisher and Nof [4]. The importance of a multilevel approach has already been underlined. However, in spite of the large amount of work on layout design, very few papers consider the flow of material and the relative closeness among activities, at the same time. Further, the multi-criteria models proposed [16–18] have non-trivial constraints and are difficult to adapt to real cases, thus having few applications. These problems are quite common for all of the presented approaches; in real problems, all factors can vary over a wide range and, generally, the analytical models introduce unfeasible restrictions. A new software package called Logistic and Re-layout Program (LRP), centred on continuous dual analysis (flow of materials and activities relationships), has therefore been developed. Its most important features, i.e. interactivity, modu-

larity and integration, permit great flexibility and, as a consequence, can fine tune solutions in real cases.

3. Integrated Approach by Logistic and Re-layout Program (LRP) Before introducing LRP logic, the most relevant features are described. 3.1 Dual Approach

It is universally recognised that the flow of materials is a very important factor in plant layout, but an accurate study can be performed only if the relationships among activities are considered. A complete approach must merge all the factors [18]. LRP presents a dual track: the flow of materials and the relationship among activities are treated by the same algorithms, in a parallel or in a combined manner. The system works on both factors at the same time and permits great flexibility during the development of the layout. 3.2 Integration

For a program dedicated to layout design, the compatibility and integration with CAD software are very important. LRP supports this facility and allows an easy data interchange with many advantages in accuracy and saving of elaboration time. 3.3 Modularity

LRP has a modular structure. It has two macro-sections, which can work individually, i.e. data input and improvement of solutions. The section for data input is realised in Microsoft Access for maximum compatibility with widespread information systems. It presents a user-friendly interface that guides the designer, or the customer on data introduction. The optimising section, based on Visio CAD support and the VBL language, elaborates the data stored in the previous section for the interactive generation of the plant layout. Each of the two sections has an internal structure made of basic modules, as shown in the LRP scheme. 3.4 Interactivity

A completely automatic approach can be too rigid: the extended set of factors involved in layout design cannot be handled independently of human control. With LRP the user operates interactively during the generation of the layout and keeps control at any moment of what LRP suggests. 3.5 Multilevel Analysis

The LRP platform has different levels of detail both on the flow of materials and the activity relationships. In particular,

Plant Layout Computerised Design

a flow control point (FCP) is introduced for a punctual analysis of the flow of materials. The FCP represents an interchange point for the materials, e.g. at machinery in shops or even at a storage point for materials at a single machine. The introduction of FCPs enables the user to make a detailed analysis of the materials moved among blocks, shops or areas, and inside a single block, at different moments in the layout under construction. The first step in a correct procedure, ideally concentrates the flow of materials in the centre of gravity of the blocks, thus obtaining a quick definition of a draft layout in a “shop logic”, with a little computational effort. This first solution can be followed by a screening phase in depth, contrary to many traditional approaches, which are based only on shop logic, and ignore the intra-block flow of materials and introduce a distortion of the final solution (Fig.1). 3.6 Entry List

A strategic target for the LRP platform is a very fast decisional support system for layout construction. By experience, the final layout corresponds strongly to the sequence by which shops, or activities are introduced: that is, the entry list. This matter is developed in [10,19]. LRP presents three different methods for the selection of the entry list [10]: Weighted ratio method. Total maximum value. Punctual maximum value.

of experience and support, can help to achieve a fast and good solution. Automated algorithms can find an optimal position based on the minimum of a target function (see target function), but a critical contribution by the user is always needed. The LRP platform supports the user’s interaction by a process called gradient: starting from the optimal position, LRP shows the trajectory, described by the barycentre of blocks, to be positioned in order to determine decreasing values of a target function, e.g. the cost of material transport (Fig. 2). 3.8

Target Functions

The evaluation of the efficiency of the plant layout is carried on by means of target functions. The most used target functions are the annual cost of the material handling policy and the total evaluation of relationships, expressed by coefficients representing the proximity among blocks or activities [20,21]. Each function calculates a numerical value, as the design of the plant layout is still in progress, in order to verify its efficiency, or to compare different layout alternatives in the final choice. In synthesis:

冘 冘 cij * zij * dij FO2 = total evaluation of activities proximity = 冘 冘 R

FO1 = material handling policy =

i

j

h

These different methods are suitable for real applications depending on the different application characteristics [10]. 3.7 Position Research

Another fundamental problem is related to the positioning procedure. In this phase a contribution by the user, in terms

919

cij zij dij Rhk

hk

k

cost of handling per unit of length between i and j ($/m) number of travels per year between i and j (trips/year) distance between i and j (m) value of the relation between h and k inside the influence area

Both these functions are very important because they influence simultaneously the plant layout, as shown during real applications of the procedure. The LRP platform estimates, at the same time, a set of target functions based on the two previous formulations, respectively: FO1, FO2 and based on a hybrid version defined as P1*FO1+P2*FO2 where P1 and P2 are user-defined weights. LRP updates these functions in real-time.

Aggregate logic (shop)

3.9 M5

M1-M2-M3-M4

Distance of Influence

It is logical to think that there is a correlation and a mutual influence among activities, even if they are not directly contigu-

Production line 2

gradient DRILLING

MILLING

Shop 1

Flow control point logic M2

M1

relationship representation

M3 M5 Production line 2

CUTTING

M4

Shop 1

LATHENING

Fig. 1. Shop logic and FCP logic.

STORAGE

Fig. 2. Application of gradient tool.

920

E. Ferrari et al.

FACILITY :

According to a graphical representation, the density values of these two aspects are transformed in arrows changing in width and shape (Fig. 4).

Cutting

Actual Distance of Influence (m) New Distance of Influence (m)

Distance of Influence

Apply

OK

4. LRP Logical-Scheme Cancel

CUTTING Form for re-setting of Distance of Influence

Fig. 3. Distance of influence.

ous. In LRP, the target for the proximity of activities is calculated by the sum of every contribution due to those activities, which is in an influence area, whose extension is defined by the user around each single block or facility (Fig. 3). 3.10

The LRP platform presents two fundamental frameworks: the input setting background and the drafting environment. The input setting background permits an organic and systematic collection of data about areas, activities, machines, and corresponding data on intended relationships and flows. Different options, for example, on distance type or on closeness scale are permitted. The final framework is the working environment where software supported by the user develops and optimises the layout. It is CAD compatible both in input and in output, it permits very powerful graphical functions and a real-time analysis for different perspectives (flows, closeness of activities and hybrid). A logic scheme is represented in Fig. 5.

Overcrowding Analysis

5. Real Application Another facility supported by LRP is the hunting of “critical” overcrowded points or zones, e.g. zones of high density in flow of materials or in closeness among conflicting activities.

The last release of the LRP platform has been applied for a relayout case, in a factory involved in the production of

LATHENING

PACKAGING

DRILLING

Target Functions – L.R.P. 2.0.0

STORAGE DELIVERY

$ Shop $ FCP Rel Hybrid

Fig. 4. Overcrowding analysis example.

Plant Layout Computerised Design

921

INPUT section Existing file CAD

Area loading New definition

Dimensions

Activities / Shop data Flow Control Point

Type of distance

Material flows data

Flow from to chart Travel cost from to chart

User-defined scale

Relationships data Relations from to chart

WORK section (CAD)

Entry list choice

Block position design Target functions Intra-block optimisation

Different Layout options

Overcrowding analysis

Fig. 5. The LRP scheme.

electronic devices, e.g. inverters, battery chargers, and welding machines (300 skilled personnel and 700.000 units sold every year). The main target of this complete relayout is to merge four existing production areas, not close each other, and to replace them by only three zones. The flow of materials is large in this case history; a first analysis is carried out centred on this topic, and the solution is fine-tuned according to the relationships as second step. As the first step, a mapping of the starting situation was carried out by a reconstruction in the LRP environment, both of the flow of materials and the activities relationships.

In the initial four facilities configuration of the company, the number of kilometres travelled (without intra-facilities travels) was about 8000. The intra-facilities path length was about 1200 kilometres per year. A new solution based on three sites, with a different configuration of machines and shops, is developed. In the LRP final relayout (on three sites), the total distance travelled per year of the handled material amounts to about 6500 km (intra-facilities travel included), that is a saving of about 1500 km/year, which is equal to 18%. On closeness of activities, the new situation produces an about 32% gain on the corresponding target function.

922

E. Ferrari et al.

References

Fig. 6. A product sample.

Because of CAD compatibility and the great interactivity, it was possible to apply a set of constraints, using an analytical approach, producing optimum results in real-time.

6.

Conclusions

Plant layout optimisation is relevant both for new plants and for existing companies. In general, it is a “capital intensive” question. During recent years, many approaches have been proposed by different authors. The resort to artificial intelligence appears the most promising, but real cases usually present constraints, in terms of flexibility and variability, that conflict with the analytical model proposed. The last release of the logistic and relayout program (LRP) approach is presented, in order to give information on the procedure for the dual contemporaneous approach (flow of materials and relationship), and CAD integration involved (for a better layout design) [11,12]. Modularity, interactivity, twolevel analysis (shop and single flow control point), and a variable entry list system are other significant features. The LRP factors for the final decision-making are a set of different target functions, overcrowding analysis and graphic analysis of relationships. A real application of LRP software in a company making electronic devices is briefly presented. Using LRP, a new solution is designed, with a saving of material transport costs of about 18% (conservatively) and a gain in activity closeness of about 32%. The LRP platform has a modular architecture, and allows a continuous improvement with the design or in the testing phase; the integration of LRP with modules of warehouse design and of manufacturing cells design appears very interesting.

1. M. Bonfoli, A. Di Giulio, L. Ferrari and P. Garulli, “Definizione del layout di impianti complessi mediante sistema esperto”, Impiantistica Italiana, 12(8), 1994. 2. M. Bonfioli, A. Di Giulio, P. Di Camillo and R. Salandra, “Definizione del layout di impianti complessi mediant...