Microsoft Word - Proposal_ETASR_MLCH-ed.doc ETASR - Engineering, Technology & Applied Science Research Vol. 2, �o. 2, 2012, 182-189 182 www.etasr.com Chew et al: A Decision-Analytic Feasibility Study of Upgrading Machinery at a Tools Workshop A Decision-Analytic Feasibility Study of Upgrading Machinery at a Tools Workshop M. L. Chew Hernández Dept. of Grad. Studies and Research Tech. of High Studies of Coacalco México City, México mchew@tesco.edu.mx E. K. Velázquez Hernández Dept. of Industrial Engineering Tech. of High Studies of Chalco Chalco, México ing.evelazquez@yahoo.com.mx S. León Dominguez Dept. of Industrial Engineering Tech. of High Studies of Chalco Chalco, México chavalol@msn.com Abstract— This paper presents the evaluation, from a Decision Analysis point of view, of the feasibility of upgrading machinery at an existing metal-forming workshop. The Integral Decision Analysis (IDA) methodology is applied to clarify the decision and develop a decision model. One of the key advantages of the IDA is its careful selection of the problem frame, allowing a correct problem definition. While following most of the original IDA methodology, an addition to this methodology is proposed in this work, that of using the strategic Means-Ends Objective +etwork as a backbone for the development of the decision model. The constructed decision model uses influence diagrams to include factual operator and vendor expertise, simulation to evaluate the alternatives and a utility function to take into account the risk attitude of the decision maker. Three alternatives are considered: Base (no modification), C+C (installing an automatic lathe) and CF (installation of an automatic milling machine). The results are presented as a graph showing zones in which a particular alternative should be selected. The results show the potential of IDA to tackle technical decisions that are otherwise approached without the due care. Keywords-decision analysis; equipment replacement; integral decision analysis; maximum expected utility I. INTRODUCTION This work presents a study of the feasibility of the inclusion of automated equipment into an existing metal manufacturing workshop located near México City. The feasibility analysis is approached from a Decision Analysis (DA) perspective. DA is a discipline which aims to bring clarity, insight and definition to messy decision situations [1-3], and has been viewed as a mixture of Systems Analysis and Decision Theory [1]. It´s usage for decision making guarantees the satisfaction of a set of desiderata (axioms) of rational choice [4]. In the context of DA, several methodologies for problem analysis have been proposed, as is the case of the PrOACT [5], the Integral Decision Analysis (IDA) [2] and Value Focused Thinking (VFT) [6]; similar methodologies are discussed in [7]. These methodologies aim to convert an initially blurry situation (in which the stakeholders don´t know exactly which consequences they care about or what can be done), into a structured decision model, in which alternatives and objectives have been clearly defined and measured [7]. The IDA consists of the following steps: 1) Problem Framing; 2) Analysis of Objectives; 3) Creation of Alternatives; 4) Identifying Uncertainties; 5) Decision Modeling; 6) Alternative evaluation; 7) Alternative selection and 8) Implementation. A distinct feature of the IDA is its careful determination of the decision frame, involving the creation of several frames of different sizes and emphases, using a graphic tool called diagram of decision frames. There is a vast body of work related to the problem of finding an optimal replacement policy of industrial machinery [8-17]. A more complicated problem arises when incorporating the effects of technological change [18-20], inflation and taxes [21], a limited budget [22], imperfect repairs [23-24] or warranties from the equipment supplier [25-27]. Other researchers have approached the problem through fuzzy models [28-29] or treated the replacement of several equipments [31- 32]. The consideration of several objectives can be found in [33-39] while the introduction of risk attitude is shown in [40]. The decision treated here is whether or not to include new equipment at a workshop, and it can be considered equivalent to the problem of determining a policy of equipment replacement. However, the above mentioned research starts with a problem that is already structured, that is, objectives and alternative courses of action are taken as a given. Related to this, a four-step method for selecting a model for a replacement problem is shown by Fraser and Posey [41] while Hart and Cook [42] propose a systematic approach to the decision process with stages of objective identification, indicators of achievement, alternatives and problems of implementation. These methodologies, however, do not treat problem framing explicitly and don´t take advantage of any of the well- established tools of the DA discipline. By contrast, in a real life situation, once the idea of replacing equipment comes to mind, the engineer should proceed to carefully define a decision frame for the situation, so relevant objectives and alternatives are uncovered. These steps are omitted in the previous works and are presented here, as they are part of the IDA methodology. Also, in this work, ETASR - Engineering, Technology & Applied Science Research Vol. 2, �o. 2, 2012, 182-189 183 www.etasr.com Chew et al: A Decision-Analytic Feasibility Study of Upgrading Machinery at a Tools Workshop relevant uncertain knowledge from the plant engineers and vendors are expressed as subjective probabilities and incorporated to the model. In this respect, except for Arueti and Okrent [39], none of the previous authors explicitly use subjective probabilities in the decision. Finally, while this work follows the original IDA methodology for the most part, the IDA methodology is here expanded by adding the usage of the strategic Means- Objectives Network as a map for decision-model building. To the best of our knowledge, there are no reports of the application of the IDA, or other DA methodology with a similarly careful procedure for problem framing, to an industrial equipment replacement problem. II. PROBLEM STATEMENT The workshop under study is located near the town of Chalco, México. It produces several types of iron and steel tools: manual and bench drills, vises, clamps, etc. Its customers are mainly local carpenters and the nearby wood furniture industry. The main concern of the manager is a perceived low efficiency in the processing of bench vises, which happens to be the top seller product of the company. One proposal for improving this situation is to substitute old equipment with modern one, thus allowing operation with fewer workers and an increased productivity, as the modern machinery is more automated. Several issues need to be settled so the problem can be modeled correctly 1. The metric over which the modifications should be evaluated: It can be productivity, production costs or profits. The adequate metric depends on the manager’s objectives. 2. The modifications that are to be considered when evaluating each proposal (i.e. are changes in inventory or layout to be considered in the decision?) 3. Are there any uncertainties that should be considered in the model? If so, the stakeholder´s dislike of uncertainty should be introduced in the model. In the following we apply the IDA steps to the problem, showing how it helps to clarify the decision. All shown tables and figures are the authors’ own production. III. DEVELOPMENT AND RESULTS A. Problem Framing In order to define a decision frame (what to decide and with which objectives), several frames should be explored. This can be conveniently done using a frames diagram. The construction of this diagram starts with the Base frame, which represents the current understanding of the decision situation, and then several other frames are defined by changing the amplitude and emphasis of the Base frame. Figure 1 shows a decision frames diagram, whose parts are explained below. Base Frame: The trigger of the decision is the idea of automating the manufacture process, thus this frame is stated as: Deciding the automation of the manufacture process. It comprises the decision of whether or not to automate, and the type and size of the new machines. Its objective is to maximize the plant productivity. +arrow Frames E1 and E2: The decision frame E1 is Deciding the degree and extent of the automation and E2 is Deciding the equipment provider. The objective of E1 is to maximize productivity and that of E2 is to minimize the time and costs involved in fixing possible equipment failures. Wide Frame A1: Deciding about improving the manufacture process contains the base frame plus other alternatives, like modifications of inventory, staff, policies of inspection and outsourcing. The objective of this frame is to maximize product quality and to minimize costs. Frames B1, B2, B3, B4 and B5: These frames are contained into A1, so their amplitude is similar to that of the Base Frame but their emphases are different: B1 emphasizes layout, B2 inspection, B3 staff, B4 inventory and B5 outsourcing. The objectives of B1-B5 are means of the objective of A1, related to the scope of each frame. Fig. 1 Diagram of Decision Frames Wide Frames A2 and A3: These frames shift the focus to different aspects of plant operation. A3 is Decide about raw material procurement and A2 is Decide marketing and A4. Deciding about mid and long term company operation a) Raw material providers b) Improvements of manufactures process c) Marketing and distribution d) New products and installations A3. Deciding raw material provider A1. Deciding about improving manufacture process a) Automation Decisions b) Layout c) Inspection d) Staff decisions e) Inventory f) Outsourcing A2. Decide about marketing and distribution a) Distribution channels b) Sales policies c) Promotions, warranties B1. Deciding workshop layout Modify the machines layout and facility location. B2. Deciding product inspection policies a) Sampling policies b) Type and frequency of inspections B3. Staff Decisions a) Number of workers/ supervisors b) Training of the workers/ supervisors Base. Deciding the automation of the manufacture process a) Degree and extent of the automation b) Provider of technology B4. Inventory Decisions Modify the number and type of parts at the warehouses E2. Deciding the provider of the automation technology National or Foreign provider, distance to nearest offices etc. E1. Decide the degree and extent of the automation a)Degree: Total or partial reduction of the number of workers b) Extension: Total or partial automation of the process B5. Outsourcing Decisions Parts of the process to be outsourced ETASR - Engineering, Technology & Applied Science Research Vol. 2, �o. 2, 2012, 182-189 184 www.etasr.com Chew et al: A Decision-Analytic Feasibility Study of Upgrading Machinery at a Tools Workshop distribution policies. The objectives of A3 are to minimize purchase costs, maximize availability and quality of raw material while those of A2 are to maximize sales and minimize marketing and distribution costs. Wide Frame A4: A4 is the widest frame to consider, including frames A1, A2 and A3 plus additional decisions, like the installation of more workshops and new product development. A4 objective is to maximize profits. Once the Decision Frames Diagram is complete, the generated frames are analyzed. A key element of our problem is that we haven´t decided whether or not to automate the process, and the automation will proceed only if it has a reasonable chance of generating economic benefits. Thus we discard the narrow frames E1 and E2, as these assume that it has been decided to automate the process. The objective of the base frame is to increase productivity. While automating the process may increase productivity, the costs of the new equipment may outbalance the economic benefits of that increase. This would be an unacceptable scenario for the stakeholders, so the objective of the base frame is inadequate, as it doesn´t refer to costs. Frame A1 objective (maximize quality and minimize costs) is more appropriate than the objective of the Base Frame. However, not all the decisions of A1 are to be considered in the present problem: we are not allowed to change inventory, inspection or outsourcing. By pruning the decisions of A1, we produce the frame A1* which has the same objective of A1 but only decisions in the context of our problem (Figure 2). A1* is not yet adequate, as the decisions should be assessed by their economic implications and A1* has the objective of maximizing quality and minimizing costs. A4 has the adequate objective (maximize profits) but the decisions included in it are too wide. We thus define A4*, eliminating the decisions of A4, not included in A1*. The decision frame to be used A4*, shown in Figure 3. Fig. 2 Reduced Diagram of Decision Frames B3. Staff Decisions a) Number of workers/ supervisors b) Training of the workers/ supervisors B1. Deciding workshop layout Modify the machine, warehouse and general facility location. Base. Deciding the automation of the manufacture process a) Degree and extent of the automation b) Provider of automation technology A4* Deciding about increasing plant profitability a) Decision about automation b) Decision about layout c) Decisions about staff Fig. 3 Final Decision Frame B. Objective Analysis The first stage of the clarification of objectives is their identification [6]. To do so we begin with a “wish list” that provides the following 19 objectives 1. Maximize profits 2. Minimize total costs 3. Maximize incomes 4. Maximize sales 5. Maximize product quality 6. Minimize the number of jobs that need re-work 7. Maximize productivity 8. Minimize process times 9. Minimize transport time 10. Minimize delivery times 11. Minimize raw material waste 12. Optimize facility lay-out 13. Minimize required man-hours 14. Minimize inventory costs 15. Minimize inspection cost 16. Maximize market share 17. Maximize competitiveness of company 18. Maximize skill of work force 19. Minimize delays in product delivery One of the most important steps in a DA approach to problem solving, is to understand the relationships among the identified objectives. The objectives that are important by themselves are called Fundamental Objectives, and are organized into a hierarchy shown in Figure 4. The objectives of the wish list that are not fundamental should either be A3. Deciding raw material providers A4. Deciding about mid and long term company operation a) Raw material providers b) Improvements of manufactures process c) Marketing and distribution d) New products and installations A1* Deciding about improvements of the manufacture process a) Decision about automation b) Decision about layout d) Decisions about A2. Decide about marketing and distribution a) Distribution channels b) Sales policies c) Promotions, warranties B3. Staff Decisions a) Number of workers/ supervisors b) Training of the workers/ supervisors Base. Deciding the automation of the manufacture process a) Degree and extent of the automation b) Provider B1. Deciding workshop layout Modify the machine, warehouse and facility location. ETASR - Engineering, Technology & Applied Science Research Vol. 2, �o. 2, 2012, 182-189 185 www.etasr.com Chew et al: A Decision-Analytic Feasibility Study of Upgrading Machinery at a Tools Workshop equivalent to a fundamental objective, or be a mean to accomplish one. In this latter is true they are called Means Objectives and are structured in the Mean-End Objectives Network of Figure 5. Fig. 4 Hierarchy of fundamental Objectives Min. Number of reworks Max Profits Raw Material Process Distribution Equipment Min Costs Min. Penalties Max. Sales Max Income Inventory Min. Waste Min. Man- Hours Min. Inventory costs Min. Delays in deliveries Max. Quality Min. Process and Transport Times Min. Inspection costs Max. Productivity Max. Market share Min. Number of workers Fig, 5 Mean-Ends Objective Network Automate process Min. Number of reworks Max Profits Raw Material Process Distribution Equipment Min Costs Min. Penalties Max. Sales Max Income Inventory Min. Waste Min. Man- Hours Min. Inventory costs Min. Delays in deliveries Max. Quality Min. Process and Transport Times Min. Inspection costs Max. Productivity Max. Market share Min. Number of workers Fig 6. Alternatives and Mean-Ends Objectives Network C. Alternatives The Means-Ends Objective Network is useful for generating alternatives; however, in this case the class of alternatives to be considered has already been identified by the framing process. The alternative “Automating the process” implies either the introduction of automated machines that substitute the workers in a part of the process or changing the existing machines for new automated ones that require less supervision or fewer workers. This alternative implies purchase, installation, start up and maintenance costs, and, potentially, costs of worker training. To indentify all effects of this alternative on the objectives, we locate the alternative to the right of the Mean-Ends Objective Network and draw arrows to the objectives affected (Figure 6) Once the general class of alternatives has been set, the following three concrete alternatives are defined for further consideration 1. C+C: Introduce an automatic Lathe. 2. CF: Introduce an automatic Milling Machine. 3. Base: Keep using the current machinery D. Analysis of Uncertain Events The main uncertainties to be considered can be identified from the Means-Ends Objective Network of Figure 6. The decision should be valued by its economic implications, so we´ll need to model the arrow paths that go from “Automate Process” to “Max. Profits”. The arrow between “Max. Productivity” and “Max. Sales” and that between “Min. Process and Transport times” and “Max. Productivity” Inventory Raw Material Process Distribution Equipment Min Costs Gas Electricity Staff Min. Penalties Max. Sales Max Income Max Profits ETASR - Engineering, Technology & Applied Science Research Vol. 2, �o. 2, 2012, 182-189 186 www.etasr.com Chew et al: A Decision-Analytic Feasibility Study of Upgrading Machinery at a Tools Workshop represent uncertain relations. While the latter may be dealt with by a simulation model, for the former we will likely have to rely on subjective probabilities from the vendor staff. E. Decision Modeling As the decision should be justified economically, it should be evaluated by its effect on profits Profits=Income−Cost (1) Income equals the number of tools (vises) sold 'V (vises/day) times the selling price of each vise PV ($/vise). Income='V×PV (2) If 'M is the size of the vise market (the maximum number that can be absorbed by the market on a daily basis) and 'PROD the daily production then 'V=min('M , 'PROD). The expertise of the company vendors can be used to construct the following contingency table of 'V. TABLE I. EXAMPLE OF CONTINGENCY TABLE FOR 'V �M Probability of nM,k �V nM.1 p(nM,1) nM,1 nM.2 p(nM,2) nM,2 : : : nM.i p(nM,i) nM,i nM.i+1 p(nM,i+1) 'PROD : : : nM.n p(nM,n) 'PROD Being nM,i such that nM,i < 'PROD