(464b) Enhancing Life Cycle Inventories Via Reconciliation across Multiple Scales

Due to the data intensive nature of Life Cycle Assessment (LCA) it is common to combine data available from a variety of sources. These include: Process LCA databases such as those contained in SimaPro [1] and being compiled by NREL [2], aggregated data about economic sectors from economic input-output models and US EPA's toxic release inventory [3,4], and detailed engineering data. The increasingly popular approach of hybrid LCA [5] enables utilization of such data representing multiple spatial scales and levels of aggregation. An unfortunate but common characteristic of LCI data is that they often violate the laws of thermodynamics including conservation of mass and energy, casting doubt on the validity of the LCA results [6]. Such violation of conservation laws is often exacerbated in hybrid LCA due to very different underlying assumptions in the disparate sources of data.

This presentation will describe a novel approach for enhancing hybrid LCI data by reconciling it with the first law of thermodynamics. This approach is related to our previous work [7], but extends it to combine data at different spatial scales (such as engineering data, process LCA, and economic input-output LCA) in a statistically rigorous manner and improves it by ensuring satisfaction of conservation laws at each scale and across different scales. This approach provides insight into the quality of the available data, helps identify missing data, and reduces the effect of errors. The proposed approach is not automatic but relies on interaction with the users, so that they can provide information about the relevant processes such as the underlying chemistry and other prior knowledge. The resulting LCI is scientifically sound and usually more complete than the original data.

Techniques for improving the quality of data by reconciliation with conservation equations have been used in process engineering for many years. When implementing data reconciliation, gross errors or outliers must be identified first followed by implementing data reconciliation using serial elimination, serial compensation or collective compensation of errors [8]. These methods require redundancy in the available data and process information. Using data reconciliation for improving hybrid LCI data is much more challenging than reconciliation of typical chemical process data because LCI usually lacks all the required information. These challenges are addressed in this work via an iterative reconciliation approach that encourages the user to provide more information about the processes in the selected life cycle. The hybrid approach also requires reconciliation at each scale and across multiple scales.

Collective compensation by mixed integer nonlinear programming is used to estimate the magnitudes of gross errors and reconciled values simultaneously. Because system boundary for LCA includes many interconnected processes, the mixed integer nonlinear programming for the collective compensation requires extensive computation time. Therefore, the rectification problem of life cycle inventory data is decomposed to result in a multiscale approach using the augmented Lagrangian relaxation method. For a case study, LCA is applied to the fixed bed maleic anhydride process using the rectified life cycle inventory data and the inconsistent life cycle inventory data. A refinery plant, a natural gas processing plant and a power plant are included as process life cycles of the maleic anhydride process, and hybrid LCA is utilize to deal with incompleteness of process life cycles. The results show that emission impacts and resource consumption by the rectified inventory data gives more reliable solution than that obtained from the inconsistent inventory data. This work indicates that reconciliation of hybrid life cycle inventory data using multiscale information can significantly improve the scientific rigor in LCA and should be a prerequisite procedure for all hybrid LCAs.

References:

1. PRe Consultants, http://www.simapro.com

2. Life-Cycle Inventory Database Project, http://www.nrel.gov/lci

3. Hendrickson, C., Horvath, A., Joshi, S., Lave, L., Economic Input-Output Models for Life-Cycle Assessment, Environmental Science and Technology, vol. 13, 198A-191A (1998).

4. Toxics Release Inventory Program, http://www.epa.gov/tri

5. Suh, S., Lenzen, M., Treloar, G. J., Hondo, H., Horvath, A., Huppes, G., Jolliet, O., Klann, U., Krewitt, W., Moriguchi, Y., Munksgaard, J., Norris, G., System Boundary Selection in Life-Cycle Inventories Using Hybrid Approaches, Environmental Science and Technology, vol. 38, 657-664 (2004).

6. Ayres, R. U., Life Cycle Analysis: A Critique, Resources, Conservation and Recycling, vol. 14, 199-223 (1995).

7. Hau, J. L., Ph. D. Dissertation, The Ohio State University, Columbus, Ohio, USA (2005).

8. Bagajewicz, M., A Brief Review of Recent Developments in Data Reconciliation and Gross Error Detection/Estimation, Latin American Applied Research. Vol. 30, pp. 335-342, (2000).