伊呂原 隆

前報〔1〕では配置対象物の形状の多様性及びその数という観点から大規模な配置問題に対応するためのレイアウト技法(CARD-SA)を提案した.本報では, このCARD-SAを発展させて石油精製・石油化学プラントの実問題を解くことを試みる.本問題のデータは実績プラントから各機器のStock情報, 機器間のFlow(配管)情報を読み出して作成し, メンテナンスエリアや機器間で保つべき保安距離などの現実的な制約条件を考慮したものである.機器の数は100を越え, 機器サイズのばらつきが大きいゆえ, 従来の技法では実用上満足出来る解が得られない.本報では前報〔1〕における建屋を敷地に職場を機器に対応させ, 目的関数は総配管長の最小化と考える.また, 有効性の検証では特に, 対象とする機器数が多くまたサイズのばらつきが大きいことより, クーリングスケジュールと解の収束性の関係等に注目する.

In this paper, a layout technique for unequal area and shape departments with a main aisle is proposed. In this technique, layout is represented by the flexible aisle structure (FAS),which is an improvement on the flexible bay structure (FBS), by considering the main aisle inside the building. FBS and FAS treat unequal area and unfixed rectangular shape departments. By using FAS, assignments of departments and aisle structure are optimized simultaneously during the search. In the objective function, which is based on the material handling cost between departments, we use the distance between departments along aisles, in order to take into account the existence of the main aisle. Also, a new measurement for evaluating department shapes, which have unequal area, unfixed rectangles, is proposed. Experimental results indicate the effectiveness of controlling department shapes by this new measurement and of using distance along aisles in the objective function.

Recently, due to consumer demands have been varied, it is essential that a facility layout is able to adapt to the demands dynamically. In this paper, we present a technique to solve a dynamic layout problem. We deal with the problem by simulataneously considering two kinds of flexibility. One is involved with the rearrangement and the changing size of departments. The other is for the fixed area and shape of the building. The Bottom-Left (BL) algorithm is integrated in the technique in order to treat rectangular departments. The two steps Geretic Algorithm (GA) to improve the solutions, which involved an operation called GENITOR algorithm, is proposed. Numerical experiments are carried out to demonstrate the effectiveness of the proposed technique.

配置問題に対する数理的アプローチの多くは配置対象物間の距離とコストペナルティ(Flow)の積の総和を最小化する組み合わせ最適化問題として定式化され, 配置対象物数が10程度の問題に対しては分枝限定法により最適解が求められている.また, その数が10を越え計算時間の制約から最適解法が事実上適用できない問題に対してはSimulated Annealing(SA)に配置変更アルゴリズム(CARD)を組み入れた近似解法(CARD-SA)が提案されている.しかしこのCARD-SAでも, プラントレイアウトとして頻繁に発生する配置対象物数が100程度にもおよび, かつ形状の多様性が高い大規模な問題に対しては適用が困難である.そこで本研究では配置変更アルゴリズム及びSAを改良することにより現実の配置問題に対して実用上満足のいく解を得るための技法を示す.

This paper presents a technique to solve the facility layout problem with rectangular departments of unequal areas. This problem can be formulated as a combinatorial optimization problem to minimize the total cost owing to material flow between departments. It is, however, computationally timeconsuming to find optimal solutions for large problems, because they are highly combinatorial in nature. Furthermore, these problems exhibit many local minim. These properties make them suitable candidates for the application of simulated annealing, which is a stochastic optimization procedure. A suboptimal method is then developed, which is a new improvement-type technique based on simulated annealing. In this technique, a new algorithm for updating placement of departments is proposed. This algorithm enables the application of simulated annealing to solve the problems considered here. This technique also offers increased flexibility to the optimization process, making the solution much less dependent upon the initial layout. Numerical results are given to demonstrate the effectiveness of the presented technique.

A Space Filling Curve (SFC) has been proposed, which allows for exchange among departments and offers increased flexibility to the optimization process, as a facility layout technique which can treat any departments of unequal area. However little attention has been paid to the department shape in spite of the fact that SFC determines the department shape. The Hilbert Curve (HC) is used as the SFC, but it gives poor estimation of department shape and is not applicable for use in some rectangular buildings. A new SFC is thus developed, which offers better estimation of department shape than SFC based on HC, and applicability for any rectangular building. Numerical results are given to demonstrate the effectiveness of the presented SFC.