Evaluation of Input Data Quality in Standardized LCA for System Improvements in Continuous Manufacturing Systems

Other conference contribution, 2024

Background & Purpose:The rise in demands for environmental
information concerning products has now seen the use of LCA
within the EPD framework rise among product producers,
particularly in the construction sector. Although some construction
products are manufactured in repetitive or batch processes due to
their specifications, others are produced in continuous processes due
to their relationship to a customizable criterion, for example, size or
mass.
Aggregates are manufactured using continuous processes, however
many of the system inputs occur at discrete intervals which can be
challenging for allocation. Further for the data collection, even inputs
that are used continuously in the process can be collected at discrete
intervals (e.g. electricity bills once a month). Therefore, not only are
there temporal variations in the inputs to the system but also
temporal variations in the data collection which all lead to
assumptions being made while modelling the LCI for a product and
can affect the data quality. This contribution explores these variations
and their potential impact on data quality for identifying system
improvements at the manufacturing level.
Methods:The evaluation is based on observations from three research
projects to identify the need for preparing and processing input and
output data.
Results: The results show multiple sources for input and output data
coming at various time intervals throughout a one year period. The
granularity varies for key input and output data.
Conclusions: The results indicate automating data collection is
needed to reduce the data collection burden on operators if LCA
results are to be used for system improvements at the manufacturing
level. However, there are questions around what level of granularity
and processing of the data is needed to be useful in identifying
improvements at a finer scale than one year. Further, it raises
challenges with quality questions that need to be addressed so that
the data can be trustworthy for operators considering the various
sources for the data that are observed.