運用實驗設計技術(DOE)於改善建築物粉刷工程剝離之研究-以高雄市A個案工程為例

建物粉刷工作從業人員日益減少，且客戶對其品質要求越趨嚴苛，故粉刷工程之品質良品率更顯重要，本論文針對粉刷施工，探討有關粉刷工作各相關因子間的關係，希望能提高施作品質之良率，降低額外成本支出。讓相關單位之粉刷施作品質能有效獲得提升，使購屋者能得到施工品質的保障。本研究運用「實驗設計」(DOE, Design of Experiments)方法，6標準差(6 Sigma)之D-M-A-I-C的步驟及方法，繪製特性要因圖找出粉刷施作過程中影響粉刷良率的因子並建立實驗因子篩選表，因子水準對照表，並利用點線圖確定實驗因子於直交表之關聯性。使用尋智專業顧問有限公司之DOE軟體，進行第一階段實驗因子分析，經過實驗後，將所得到的實驗數據製成平均值比較表，並進行變異數分析（Analysis of Variance, ANOVA），貢獻度柏拉圖分析找出影響品質的非常顯著因子為素地表面清潔度及素地表面含水量。再以此兩因子規劃進行第二階段實驗，依據主因子之各水準平均值回應圖找到平穩的區間，將得到最佳操作條件的組合。由A個案工程進行再現性實驗粉刷施作，做為確認實驗完成後的成效。再藉由他案實驗操作下，粉刷工作良品率有效從96%提升至98%。

關鍵字

實驗設計(DOE) ；粉刷工程；良品率； Six Sigma ； DMAIC

並列摘要

The painting industry has been experiencing a continuous decline in labor; meanwhile, customers are becoming more demanding with respect to quality. Consequently, there is great importance attached to the question of how to produce both yield and quality in painting work. This thesis focuses on painting work and explores the relationships between the related factors in painting processes. Hopes are that the results will lead to higher yield rates and quality while allowing for reductions in unnecessary costs so that related organizations can see effective improvements in painting work quality and house buyers will be able to be sure of the quality of the work. This study follows a Design of Experiments (DOE) approach according to the 6-Sigma D-M-A-I-C steps and methodologies. A cause and effect diagram was created to find the factors that impact yields during the painting processes and an experimental factor screening table was constructed. A factor level comparison table and a linear graph were used to determine correlations between experimental factors in the orthogonal array. DOE software from Eureka Consultant Corp. was used to conduct the first stage in the experimental factor analysis. After the experiments were conducted, the data was used to create a comparison of means table and conduct the Analysis of Variance (ANOVA) test. A Pareto Chart and analysis of the contribution levels show that the factors which have a very significant effect on quality are the cleanliness and water content on the plain surfaces. These two factors were used as the basis for a second stage experiment. According to the average level value of the main factors, a response table indicated a stable region with the best operating conditions. During the work done at the Case A site, reproducible painting experiments were conducted, serving to confirm the effectiveness of the experiment. While conducting experiments in other cases, the painting work yield rates effectively rose from 96% to 98%.