Many engineering systems rely on interpretation of data collected at different spatial scales. For example, process engineering data may be available at scales such as specific equipment, plants and supply chains. Similarly, data for life cycle assessment (LCA) is often available at equipment, value chain and economy scales. Data rectification is a popular approach for reducing errors in process data based on using domain specific information at each scale (intra-scale models) and across scales (inter-scale models) [1]. Our work is motivated by the availability of multiscale data in life cycle assessment, but the proposed approach is general and can be applied to rectification of data from other sources. This presentation will focus on new methods for guiding the procurement of additional data at single or multiple scales to improve the quality of rectified results. This approach is relevant for making decisions about buying new data or installing additional sensors.

The reliability of results obtained from LCA depends on the quality of available life cycle inventory (LCI) data, which mandates the need for comprehensive and accurate databases. Among the other challenging steps in LCA, procurement of reliable dataset seems to be the most intriguing task. The present status of various LCI databases shows a need for rigorous techniques to enhance quality due to the presence of errors such as missing streams, random and gross errors in measurement, errors due to the combination of data from various sources and errors at multiple spatial and temporal scale. There has been much research in the field of data rectification for application in chemical process industries and in the past few years these methods have been implemented for rectification of LCI to improve their quality [1,2]. Despite the long history of data rectification methods and the recognition of their relevance to improving LCI data, applying these methods to LCI data introduces many new challenges. These include the need for information that is usually not included in LCI databases such as the underlying life cycle process network or process chemistry and inconsistencies arising due to aggregation of LCI databases at different scales for use in hybrid LCA. With the availability of several LCI databases another challenge is to find the most reliable dataset for a particular LCA study. Although recently there has been some work to address the need of multiscale rectification [2], it still has limitations in terms of excluding the specific process chemistry and extent of reactions as proper constraints. Including such information in data rectification is expected to have significant effect on the quality and reliability of results.

This work particularly focuses on developing a multiscale data rectification technique by using the process chemistry as constraints to provide a more reliable dataset for hybrid LCA studies. Reliable external data usually improve the LCI, but inclusion of the external data might not always be worth the cost and effort in terms of improvement of quality of LCI. The approach taken here uses "value of information" metrics to assess the marginal increase in quality of LCI by procurement of external data to guide decisions about purchasing the external dataset. Using the developed technique this work also illustrates the need of using multiple databases as complimentary to each other and then apply data rectification techniques to come up with a more reliable dataset. The results obtained by using the dataset developed by combination of data from two different datasets can give more reliable results from LCA studies. With the current scenario of environmental concerns it would prove worthy to put these extra efforts.

The need of multiscale data rectification and not relying on a single database to make decisions is illustrated by a case study based on data from commercial and public domain databases of chemical processes in a life cycle. which demonstrates the strength of approach taken here to improve the quality of LCI data by external data procurement. The case study also shows the augmentation of quality of LCI data available in a public database by use of fine scale data available from other database. These approaches are expected to add to the value of LCA studies in present day world which needs sustainability.

References

1.Hau, J. L., Yi, H., & Bakshi, B. R. (2007). Enhancing life-cycle inventories via reconciliation with the laws of thermodynamics. {JOURNAL OF INDUSTRIAL ECOLOGY}, {11}({4}), {5-25}.

2.Yi, H., & Bakshi, B. R. (2007). Rectification of multiscale data with application to life cycle inventories. {AIChE JOURNAL}, {53}({4}), {876-890}.