Poka-yoke Methods in Make-to-Order Production

Abstract

Poka-yoke applications are tools aimed at creating defect-free production environment by removing the possibility of mistake or recognizing the mistake at as early a stage as possible. Thus, the poka-yoke applications ensure that customers do not receive defected products. In made-toorder production it is possible to fulfill the customer needs by providing customizable products which leads to the manufacturing company having to be able to function in a complex manufacturing environment. There is little existing research of poka-yoke methods in make-to-order production. The goal of this master’s thesis is to define what kind of mistakes operators make in make-toorder production and how those mistakes could be prevented using poka-yoke methods. The research is completed as a quantitative action research in cooperation with a case company that manufactures weather and industrial measurement devices. The product catalogue of the case company is diverse and production volumes are small. The primary data used in the research is a collection of customer complaints from the year 2019. A secondary set of data is gathered from work instructions, order forms and standard work sheets. Based on the data it is identified which types of defects cause customer complaints and in which production phase they occur. After the production phase is identified poka-yoke methods are applied to eliminate the mistakes that lead to the defects form the production phase. The research shows that a large portion of customer complaints result from a part missing from the order or the wrong part being in the order. Most of the missing or wrong parts are accessories that are collected at the end of the production line. Based on the root cause analysis, the mistakes are a result of the production phase not guiding the operator enough, instead the production phase requires the operator to make multiple production related choices. To eliminate the mistakes in the packing phase four poka-yoke applications are evaluated. Based on the evaluation a method with opening lids to indicate what to pack is deemed best, but the method did not fit into the existing manufacturing environment. A light-guided system guides the operators to pick the right products based on the product configuration and was evaluated to be the second-best option. A scale system that checks whether the products that have been packed weigh the correct amount but does not guide the operator in making the choices during the picking process. A barcode reading method could help the operator by checking every product as it is being packed, but scanning every product while packing is too slow. The light-guided system and scale system were piloted in two different production cells. The results were promising, because no mistakes that would lead to defects were made during the pilot. However, the pilot was rather short, and more research is required to prove that the pokayoke methods applied lead to mistake-free production. What can be said on the results is that poka-yoke methods can be used for decreasing mistakes and mistake-proofing packing mistakes in the case company. Based on the results, the challenge with poka-yoke applications in MTO production is that there are so many different errors that multiple poka-yoke applications are needed to prevent them. This being the case, it could be possible that the defining factor for mistakes is the ‘decision-making density’, which would mean that the tasks where decision making by the operator is required the most are the most potential tasks for mistakes. More research in different types of production environments are needed to generalize the findings.