聚氯乙烯廠作業勞工職務暴露矩陣模式之建立

Development of a Job Exposure Matrix Model for Polyvinyl Chloride Workers in Taiwan

本研究中由五家聚氯乙懦工廠提供逐年之相關資料，檢視資料，可將工程改善分 成四個時期，且其工人可歸納成三大類，24 種職務。重合反應後之氯乙懦逸散量依照 質量平衡之原則或製程改善之效率兩種模式來計算。結合以擴散公式建立氯乙懦逸散 模式以及近年來之實際環測資料，我們建立推佔工廠內勞工氯乙懦暴露之多重迴歸經 驗公式。除了逸散量、距離因子之外，其他工作相關之變項如工作地點、時間、風向、 風速也予以加權或放入最後之模式中，建立了這五家工廠從 1965 至 1995 年 24 種職 務勞工之八小時時量平均職務暴露矩陣。結果顯示在 1986 年各廠脫除設備設立之前，兩種模式所推佔之職務暴露矩陣接 近，在 1986 年之後製程改善效率之模式較接近近年實際環測之數據。我們結論認為 結合既有之生產或製程資料可以來建構職務暴露矩陣。然而因為變異性仍大，將來用 於聚氯乙懦世代追蹤研究時需注意其不確定性之程度。本研究所建立之職務暴露矩陣是一系統化之資料庫，它可以彈性放入各種不同之 影響因子，並可藉由日後資料的修正、驗證而使得誤差減少。其他假設也可以後續修 正。在其他石化工業製程之逸散推佔上，本研究之方法可能具有參考價值。

After collecting and reviewing five polyvinyl chloride (PVC) factories in Taiwan annual records, we classified the implementation of engineering control into four stages, and the workers into three categories, 24 job titles. The emission amount of vinyl chloride monomer (VCM) from the polymerization process was calculated by mass 、balance or process efficiency model. After adopting multiple regression analysis VCM dispersion equations and current measurements, an empirical model to predict the field exposure of workers within factory boundary was developed. Along with the variables of emission amount, distance in the prediction model, other parameters related to each task such as place, duration, wind direction and velocity were weighted and integrated into the final model. 8-hr time weighted average job exposure matrix (JEM) of PVC workers in these factories were then reconstructed from 1965 to 1995. The results showed that around 1986, before the setup of stripping process, JEM estimation in two models posed no distinct difference, and a mass balance model was applied. The process efficiency model showed more consistency with recent ambient measurements results for the time period after 1986. Though adopting several assumptions, we remain convinced that reconstructing JEM in PVC workers is feasible by simple integration of production or process data. However, because of the variations, it is prudential to note the uncertainty while applying this technique for exposure assessment in future studies among VCM cohorts. The JEM is stored as a computerized database, flexible for incorporating other affecting factors, and it contains bias and permits refinement if more valid measurement or verification data become available. The developed JEM methods can also be applied to estimate the exposure of other petrochemical processes.