Introduction

The flexibility of life cycle inventory (LCI) background data selection is increasing with the increasing availability of data, but this comes along with the challenge of using the background data with primary life cycle inventory data. To relieve the burden on the practitioner to create the linkages and reduce bias, this study aimed at applying automation to create foreground LCI from primary data and link it to background data to construct product system models (PSM).

Methods

Three experienced LCA software developers were commissioned to independently develop software prototypes to address the problem of how to generate an operable PSM from a complex product specification. The participants were given a confidential product specification in the form of a Bill of Materials (BOM) and were asked to develop and test prototype software under a limited time period that converted the BOM into a foreground model and linked it with one or more a background datasets, along with a list of other functional requirements. The resulting prototypes were compared and tested with additional product specifications.

Results

Each developer took a distinct approach to the problem. One approach used semantic similarity relations to identify best-fit background datasets. Another approach focused on producing a flexible description of the model structure that removed redundancy and permitted aggregation. Another approach provided an interactive web application for matching product components to standardized product classification systems to facilitate characterization and linking.

Discussion

Four distinct steps were identified in the broader problem of automating PSM construction: creating a foreground model from product data, determining the quantitative properties of foreground model flows, linking flows to background datasets, and expressing the linked model in a format that could be used by existing LCA software. The three prototypes are complementary in that they address different steps and demonstrate alternative approaches. Manual work was still required in each case, especially in the descriptions of the product flows that must be provided by background datasets.

Conclusion

The study demonstrates the utility of a distributed, comparative software development, as applied to the problem of LCA software. The results demonstrate that the problem of automated PSM construction is tractable. The prototypes created advance the state of the art for LCA software.

中文翻译：

---

自动化产品系统模型组装的原型

介绍

生命周期清单 (LCI) 背景数据选择的灵活性随着数据可用性的增加而增加，但这伴随着将背景数据与主要生命周期清单数据一起使用的挑战。为了减轻从业者创建链接的负担并减少偏见，本研究旨在应用自动化从原始数据创建前景 LCI，并将其链接到背景数据以构建产品系统模型 (PSM)。

方法

三位经验丰富的 LCA 软件开发人员受委托独立开发软件原型，以解决如何从复杂的产品规范中生成可操作的 PSM 的问题。参与者获得了一份材料清单 (BOM) 形式的机密产品规范，并被要求在有限的时间内开发和测试原型软件，将 BOM 转换为前景模型并将其与一个或多个背景联系起来数据集，以及其他功能要求的列表。将所得原型与其他产品规格进行比较和测试。

结果

每个开发人员都对问题采取了不同的方法。一种方法使用语义相似性关系来识别最适合的背景数据集。另一种方法侧重于生成模型结构的灵活描述，去除冗余和允许的聚合。另一种方法提供了一种交互式网络应用程序，用于将产品组件与标准化产品分类系统进行匹配，以促进表征和链接。

讨论

在自动化 PSM 构建的更广泛问题中确定了四个不同的步骤：从产品数据创建前景模型，确定前景模型流的定量属性，将流链接到背景数据集，并以一种格式表达链接模型，现有的 LCA 软件。这三个原型是互补的，因为它们处理不同的步骤并展示替代方法。在每种情况下仍然需要手动工作，特别是在必须由后台数据集提供的产品流描述中。

结论

该研究证明了分布式比较软件开发在解决 LCA 软件问题时的效用。结果表明，自动化 PSM 构建的问题是可以解决的。创建的原型推进了 LCA 软件的最新技术。