Analitik hiyerarşi prosesi yardımıyla kalite fonksiyon açınımında önceliklendirme

Abstract

Günlük yaşamımızda gerek özel gereksede işimizle ilgili bir çok kararlar vermek zorunda kalırız. Bu kararlar genelde bir çok kısıt altında, birden fazla faktörün etkileşimde bulunduğu durumlar altında alınır. Bu çalışma, karar verme destek sistemleri dahilinde mevcut çok ölçütlü deterministik yöntemleri incelemeyle başlamıştır. Çok ölçütlü karar verme destek sistemleri içinde insan beyninin doğal çalışma sistemine benzer bir yaklaşım sunan analitik hiyerarşi prosesi yöntemi üzerinde durulmuştur. Analitik Hiyerarşi Prosesi, bir amacı veya hedefi etkileyen etmenleri biribirine benzer özellikler allında gruplandırır ve amaca uygun hiyerarşiler kurarak her düzeyin amacı ne kadar ve nasıl etkilediğine dair önceliklendirmeler yapan bir karar verme destek sistemidir. Pazarlama fonksiyonlarını dikkate aldığımızda, tüm kurulan stratejilerin, planların ürünün müşteri tarafından daha doğru algılanması, bir marka bilinci yaratılması ve ürün bağlılığın yaratılması üzerine olduğunu görürüz. Bunun dışında kalan aktiviteler destekleyici satış, dağıtım gibi aktivitelerdir. Tüm bunlar dikkate alındığında ürünün tasarım ve fonksiyonduk açısından müşteri beklenti ve ihtiyaçlarını tam olarak karşılaması hatta müşterileri özellikleri bakımından müşteriyi heyecanlandırması gereklidir. Bu nedenle Kalite Fonksiyon Açınımı tekniği müşteri beklentilerini ürün üzerine yansıtmada çok kullanışlı bir metodtur. Bu çalışmada, Kalite Fonksiyon Açınımı metodunda kullanılacak müşteri isteklerinin önceliklendirilmesinde Analitik Hiyerarşi Prosesi yönteminin uygunluğu gösterilmektedir. Sonuçta müşteri beklentileri AHP ile önceliklendirilmiş ve bulunan sonuçlar ile Kalite Fonksiyon Açınımı yöntemi uygulanmıştır.

There are a number of rather different approaches available for the decision support systems. A few classification approaches have been proposed by researchers in the past. One of them classified the systems as single objective deterministic methods, multi-objective deterministic methods, probabilistic methods and fuzzy set methods. Another one classified as single criterion methods that are divided into deterministic and non deterministic methods and multiple-criteria methods that are again divided into deterministic and non-deterministic methods. Single criterion deterministic methods include net present value method, internal rate of return method, benefit/cost ratio method, payback period, mathematical programming and minimal annual revenue requirement. Single criterion non- deterministic methods include sensitivity analysis, decision trees, optimistic/pessimistic, Monte-Carlo simulation. Multiple criterion deterministic methods include scoring models, analytic hierarchy process, goal programming, decision support systems, productivity model, dynamic programming. Multiple criteria non-deterministic methods include fuzzy linguistics, expert system, utility models, game theoretic model. Scoring models feature the ability to accommodate the consideration of intangible, or economically non-quantifiable elements involved in an investment decision in an analytical fashion. A linear additive scoring model aids decision makers in evaluating the desirability of a firm's long term and short term decisions. Goal programming features the ability to analyze multiple, conflicting goals. It has been used to model investment decisions in a flexible manufacturing system from a multiple-objective context. Decision support system refers to a computerized approach to establishing an information system for managerial analysis. The applications of traditional economic evaluation methods, simulation, mathematical programming or accounting techniques for the evaluation of advanced manufacturing systems suffer from limitations when used alone, but better analysis is possible when all the techniques are combined. With a combined decision support system several alternatives can be studied in less time. Productivity mode is used to access the impact of the proposed equipment investment on profitability. If a project has a total productivity level greater than or equal to a predetermined productivity level, it is accepted. XI Dynamic programming is an appropriate tool for usage in the justification problem since the decision in the level of investment depends not only the current realizations of costs and revenues but also on previous decisions regarding investment. Fuzzy linguistic models permit the translation of verbal expressions into numerical ones, thereby dealing quantitatively with imprecision in the expressions of the importance of each strategic goal. Expert systems are computer programs that have the capability of solving complex problems by certain rules and logical reasoning mechanism. These rules represent the problem solving approaches of experts. The main feature of expert systems is the ability to handle problems with inexact data. Several applications of expert systems have been developed by researchers. Utility models evaluate the decision maker's preferences expressed as utility functions of multiple attributes in order to determine the choice that satisfies him the most. Game-theoretic models feature the ability to handle the element of strategic interdependence between firms, which is an important factor because the firms that invest in new products exhibit significant strategic interdependence between competitors. The Analytic Hierarchy Process is particularly useful for evaluating complex multiattribute alternatives involving subjective or intangible criteria. The Analytic Hierarchy Process has an axiomatic base that establishes its mathematical viability. The essential steps in the application of the Analytic Hierarchy Process involve decomposing a general decision problem in a hierarchical fashion into sub problems that easily can be comprehended and evaluated, determining the priorities of the elements at each level of the decision hierarchy, and synthesizing the priorities to determine the overall priorities of the decision alternatives. A hierarchy is an abstraction of the structure of a system to study the functional interactions of its components and their impacts on the entire system. This abstraction can take several related forms, all of which essentially descend from an apex, down to sub-objectives, down further to forces which affect these sub-objectives, down to the people who influence these forces, down to the objectives of the people and to their policies, still down to the strategies and finally the outcomes which result from these strategies. Two questions arise in the hierarchical structuring of systems: 1. How do we structure the functions of a system hierarchically? 2. How do we measure the impacts of any element in the hierarchy? Hierarchies are basic to human understanding. After constructing of a hierarchy, one must assess the relative importance of the elements at each level. In the Analytic Hierarchy Process, the decision maker is asked to compare the elements at a given level on a pairwise basis to estimate their relative importance/preference in relation to the elements at the immediately proceeding level. Pairwise comparison is the heart of the method. Each pairwise comparison represents an estimate of the ratio of the xn priorities or weights of the compared elements. The preferences are translated into pairwise weights of 1, 3, 5, 7, or 9, respectively, with 2, 4, 6, and 8 as intermediate values. Applying eigenvector method to these data, estimates of the weights/priorities are calculated for each pairwise comparison matrix for each level of the hierarchy. To synthesize the results over all levels, the priorities of the each level are weighted by the priority of the higher level criterion/attribute with respect to which the comparison was made. The weighted priorities of the decision alternatives are added component wise to obtain an overall weight or priority at each alternative over the entire hierarchy. The resulting priorities represent the intensity of the decision maker's judgmental perception of the preferences of the alternatives, taking into account the relative importance of the criteria represented in the hierarchy, and considering the importance of and tradeoffs among the attributes. An important consideration in using the Analytic Hierarchy Process is the notion of consistency. A respondent who reports that A is twice as important as B and that B is three times as important as C is providing consistent judgement if he also reports that A is six times as important as C. If the respondent reports any other value for the A-C comparison, the judgement is said to be inconsistent. The eigenvector method permits a quantitative assessment of consistency. The consistency ratio should be less than 0.10 for acceptable results. When judgments are inconsistent, the decision maker should be given the opportunity to revise the pairwise comparisons. When a group result is sought in Analytic Hierarchy Process, consensus on the judgments is obtained. In order to aggregate the judgments of a group of individuals, the collective judgement itself must satisfy the reciprocal property. It has been demonstrated that the geometric mean of the set individual pairwise comparisons. The geometric mean of all responses for each pairwise comparison is calculated, and the resulting comparison matrix consisting of geometric means is analyzed using the eigenvector method to estimate the priorities. Quality Function Deployment is a means of assuring the customer requirements (needs, wants, demands) are accurately translated into relevant technical requirements throughout each stage of the product development process. Customer requirements can be traced from the start of product planning down to the most detailed instructions at the operating level. The voice of the customer is the point of departure for QFD and drives the process. Listening, understanding, interpreting, and translating what the customer says forms the heart of QFD. Quality Function Development addresses the need to start the design process with clear objectives for a product that if met will not only satisfy the customers' wants, but also actually excite or delight them. Quality Function Development also emphasizes the need to know as much as possible about a product before it is introduced to the manufacturing process. Quality Function Development emphasizes multifunctional teams and helps to integrate all corporate functions to be responsive to the customers' requirements so that product planning, product design, process planning, and production planning provide a coherent response to customer needs. It provides discipline to the decision making process, so that all requirements are taken into account, decisions are documented and all information is brought to bear. The identification of customer requirements provides the foundation for the process. These requirements are related xni to the technical aspects of the product through a planning matrix, termed the "House of Quality" that provides various mechanisms for evaluating how well the technical aspects satisfy the customer requirements. QFD's fundamental objectives are to: 1. Identify the customer, 2. Identify what the customer wants, 3. Identify how to fulfill the customer wants. A broad description of constructing House of Quality is given at the thesis. Quality Function Development process involves four phases. Phase 1: This phase is called product planning phase and is usually led by the marketing department. Document customer requirements, competitive opportunities, product measurements and technical ability of the organization to meet each customer requirement. Getting good data from the customer in this phase is critical to success of the entire Quality Function Development process. Design requirements can be gotten from this phase. Phase 2: This phase is called product design phase and is usually led by the engineering department. Product concepts are created during this phase and pan specifications are documented. Critical part/mechanism characteristic can be gotten from this phase. Phase 3: This phase is called process planning phase and is usually led by the manufacturing department. The manufacturing process is flowcharted and process parameters are documented. Key processes can be determined at this phase. Phase 4: This phase is called process control planning phase and is usually led by the quality assurance department.. Performance indicators are created to monitor the production process, maintenance schedules, and skills training for operators. Control methods can be set at this phase. Many Quality Function Deployment approaches require the QFD team to evaluate the importance of the customer requirements and assign predetermined weights to each attribute. However, these approaches do not seem to capture the importance of the requirement from the customer's perspective. Moreover these approaches seem to have little ability to capture aggregate importance from a group of diverse customers. Identification of the requirement and prioritizing them is a nontrivial problem. In fact, it is found that inconsistency in quantifying expert judgement remains a common weakness at Quality Function Deployment methodologies. An innovative approach to this problem that is involving the use of the Analytic Hierarchy Process, has been suggested. Analytic Hierarchy Process is used to develop and prioritize customer requirements for an application of Quality Function Deployment to the manufacture of an exterior structural wall panel. xiv The results demonstrate that the Analytic Hierarchy Process does provide an effective framework for determining the priorities of the customer requirements in Quality Function Deployment. After the priorities are found by using Analytic Hierarchy Process, it is applied Quality Function Development process to these results and House of Quality had been constructed.