Temporalis, a generic method and tool for dynamic Life Cycle Assessment
Graphical abstract
Introduction
Life Cycle Assessment (LCA) is a well-established method to estimate the potential environmental impacts of services and products throughout their entire life cycle. One of the shortcomings of LCA practice is the lack of consideration of the temporal and spatial variation of flows and emissions (Huijbregts, 1998). Already in the early days of LCA Finnveden and Nielsen (1999) stressed the importance of considering the long term emissions from landfills. The lack of temporal considerations is still considered an unresolved problem and an important limitation for the accuracy and representativeness of LCA (McManus and Taylor, 2015; Reap et al., 2008). Methodologies to include time and space in LCA have been proposed (Mutel and Hellweg, 2009; Beloin-Saint-Pierre et al., 2014, Beloin-Saint-Pierre et al., 2017; Tiruta-Barna et al., 2016; Yang and Heijungs, 2016), but it is still difficult to easily perform dynamic and spatialized LCA for practitioners. This is also due to the lack of accessible and transparent (possibly open source) software. Since “LCA is primarily a steady-state tool” (Udo de Haes, 2006) the conventional approach sums all the emissions for a given pollutant into a single value in the Life Cycle Inventory (LCI), regardless of its time of occurrence. Subsequently, the impacts of the aggregated environmental interventions are characterized during the Life Cycle Impact Assessment (LCIA), irrespective of their timing.
Time can be considered at the level of: (i) the Functional Unit (FU), by giving it a temporal dimension (e.g. one year of energy use); (ii) the LCI, by explicitly considering the temporal relationship between flows; (iii) the LCIA, by using dynamic characterization factors (dCF) or characterization functions (CFun) in place of characterization factors (CF) and (iv) the weighting of impacts, for example by discounting them (Collet et al., 2014; Hellweg et al., 2003). Regardless the level of complexity considered, to take time into account in LCA, the LCI must be dynamic, which means that emissions and resource consumptions are explicitly distributed over time. In their seminal book on the computational aspects of LCA, Heijungs and Suh (2002) already discussed a theoretical extension of the matrix-based method to include both spatial and temporal differentiation of the inventory. But already at that time the authors warned the reader that, despite the solid theoretical base, the method's operationalization posed problems. This is due to the huge amount of temporal data required and its high computational demand. In the first studies talking about dynamic LCA (dLCA) (Pehnt, 2006; Kendall et al., 2009; Zhai and Williams, 2010) time was not explicitly considered. In these works the temporal changes in the processes studied were implicitly considered and eventually both emissions and impacts were still aggregated following the traditional LCA approach.
To be dynamic a LCI must be able to locate and differentiate activities and flows in time. This ability to consider and compute temporal characteristics in LCIs, to the best of our knowledge, has been presented in three methodological proposals. In Collinge et al. (2013) the traditional approach based on matrix inversion (Heijungs and Suh, 2002) is used and improved with the inclusion of temporal information. Although it is possible with this method to consider time for each dataset in the LCI, it shows the important operational limitations already recognized from Heijungs and Suh (2002). Beloin-Saint-Pierre et al. (2014) developed the enhanced structure path assessment (ESPA), which extends on structural path analysis, a widely known technique in input-output analysis. It makes use of power series expansion to solve the dynamic inventory, and the matrix inversion is replaced with a product of convolution of the discrete distribution functions. The ESPA has recently been further integrated with the possibility to consider time also at the level of LCIA by applying time-dependent characterization factors (Beloin-Saint-Pierre et al., 2017). The major drawback of this approach is that it is still insufficiently documented and, to date, it has not been made operational and thus not available for the LCA community. A final approach consists in a direct traversal of the supply chain graph, as done by Tiruta-Barna et al. (2016). They recently introduced a very promising method for dynamic LCI that has been developed as a prototype web application. It is based on a process flow network structure and makes use of a graph search algorithm to build the temporal model. Despite the promises of this methodology, it is still a proof of concept that needs to face the implementation challenges of a desktop application. For example, the need for a reduced utilization of memory and computational resources in comparison to a server application. Moreover, it is not coupled to a LCIA framework and it is not clear if the method can deal with datasets without temporal information, raising doubts over its integration potential with existing LCA databases. Regarding the treatment of the LCI as a graph, it is worth mentioning that this approach poses a key methodological challenge due to the cyclic nature of the supply chain graphs. Loops can be encountered, and a cutoff function must be applied to halt potentially infinite loops in supply chain traversal.
Available temporal information can be absolute (e.g. May 25, 1978) and relative (e.g. two weeks ago) in time. While for most impact assessment methods it is necessary to know the absolute calendar date of the emissions (Beloin-Saint-Pierre et al., 2014), both relative and absolute distributions can be encountered in the inventory. This is essentially dependent on how the data are collected during the LCI construction and there are no specific indications to use one or the other. The work of Collinge et al. (2013) is based on absolute temporal data while Beloin-Saint-Pierre et al. (2014) and Tiruta-Barna et al. (2016) use relative temporal information. Ideally both types of temporal information can be handled by a dynamic LCA framework.
The timing of emission is also relevant in impact assessment (IA). In conventional LCIA methods, emissions are integrated over the life cycle, hence they are treated as a pulse rather than a temporally distributed flux. But the moment when the emissions occur can affect the impact. An example is those impact categories influenced by the background concentrations of the pollutants, like aquatic eutrophication (Udo de Haes et al., 2002) and acidification (Potting et al., 1998). Noise impact on human health (Cucurachi et al., 2012), photochemical smog production (Shah and Ries, 2009) and water scarcity (Kounina et al., 2013) are other examples of time-dependent environmental responses. Timing of emissions is also relevant when their impact assessment is performed on a finite time horizon (TH). The typical example of a time horizon-dependent CF is the Global Warming Potential (GWP). This metric, in fact, is very sensitive to the time horizon considered, and the impacts are directly related to its length (IPCC, 2013). In the non-dynamic approach it is implicitly assumed that all the life cycle emissions occur at year 0 and remain in the environment for the entire TH. Levasseur et al. (2010) applied time-dependent CFs to temporally differentiated LCI, overcoming the inconsistencies due to the application of a static approach in the IA.
Numerous authors have demonstrated how neglecting time consideration in LCIA can lead to mis-estimation of impacts (Almeida et al., 2015; Kendall, 2012; Lebailly et al., 2014; Levasseur et al., 2012; Levasseur et al., 2010; Levasseur et al., 2013; Pinsonnault et al., 2014). The limits of the non-dynamic approach are further amplified when biogenic carbon and long life cycles are studied (Jørgensen and Hauschild, 2013). To address the issue of emissions timing in LCA Kendall (2012) also proposed the use of the Time Adjusted Warming Potential (TAWP), a static, time-corrected GWP metric that weights the global warming impact on the basis of the timing of the emissions.
While the systematic introduction of temporal dynamics would increase the representativeness of the LCA results, the process needs to be confronted with the increase in complexity of the LCA modelling and the lack of temporal parameters in LCI databases. In addition, the collection of temporally differentiated data can be a long and costly task, and it should be undertaken only for those datasets that are more sensitive to time. Pinsonnault et al. (2014) demonstrated that temporally differentiated information, on first approximation, are not needed for every process, and their use can be restricted to the ones more sensitive to time. Collet et al. (2014) also introduced a method to identify the specific flows requiring such a temporal differentiation. The method uses a stepwise approach to assess the sensibility of the results to the temporal variability of environmental and product flows. Despite the limitations due to the upfront choice of the LCIA method, this method can represent an important instrument to help in understanding where temporal explicit data are needed and further efforts are necessary during data collection. The possibility to deal also with datasets without temporal parameters is a necessary feature of a dLCA framework.
In short, despite the substantial work done on developing dynamic LCA in the past ten years, no methods have been defined and implemented to provide (i) efficient resolution of temporally differentiated life cycle inventories (LCI); (ii) handling of both absolute and relative temporal distributions, as well as exchanges with databases that have no temporal information; (iii) dynamic characterization of emissions, including both distribution over time and characterization as a function of time; (iv) correct temporal accounting of biogenic carbon (i.e. no carbon neutrality assumption); (v) implementation in accessible and open source computer code. In this paper, we present a novel numerical computational approach to dynamic LCA which meets all our criteria. We implemented our approach in the open source Temporalis software library, built on top of the Brightway2 LCA framework (Mutel, 2017). In this paper we will present the methodology, validate it with a virtual example, and introduce its software implementation. We will use Temporalis to calculate the cradle-to-grave climate impact of 1 m3 of glued laminated timber (glulam) and show how, by explicitly considering the temporal information, the LCA results diverge from the conventional steady state approach.
Section snippets
Methods
We first introduce the computational framework and the way temporal information is stored. We then explain the functioning of the implemented best-first graph traversal used to solve the dynamic inventory problem, validating it with a virtual example. Finally, the dynamic impact assessment and software implementation are explained. To ensure the necessary transparency and reproducibility of the study requested by several scholars (Frischknecht, 2004; Pauliuk et al., 2015), all the analysis have
Results
Fig. 4 shows the cumulative climate impact for the case study over a time horizon TH of 20, 100 and 500 years using both static and dynamic LCA.
First, we compared the results obtained using static (sLCI) and dynamic LCI (dLCI) for a static GWP over 20 (Fig. 4a), 100 (Fig. 4b) and 500 years TH (Fig. 4c). It can be seen that the closer t0 to the end of TH, the greater is the discrepancy between the two results. This is due to the fact that when using a static LCI all the environmental
Discussion
The methodology reported in this paper goes a step further compared to what has been already done in the field of dynamic LCA. It allows for the accounting of time at all the levels outlined in the introduction and is fully flexible for what the temporal information is concerned. This flexibility makes it possible to easily and efficiently use the methodology and the Temporalis software with already existing databases that traditionally lack temporal information. In our case study, for example,
Acknowledgement
This work was conducted as part of the collaborative project FORest management strategies to enhance the MITigation potential of European forests (FORMIT), funded by the European Union's Seventh Framework Programme under grant agreement no. 311970. The work of C. Mutel was supported by the Swiss Competence Center for Energy Research-Supply of Electricity (SCCER-SoE).
References (56)
- et al.
Recent developments in Life Cycle Assessment
J. Environ. Manag.
(2009) - et al.
Introducing a localised spatio-temporal LCI method with wheat production as exploratory case study
J. Clean. Prod.
(2017) - et al.
SCALE: Software for CALculating Emergy based on life cycle inventories
Ecol. Model.
(2013) - et al.
The changing nature of life cycle assessment
Biomass Bioenergy
(2015) Dynamic life cycle assessment (LCA) of renewable energy technologies
Renew. Energy
(2006)- et al.
Meta-analysis of greenhouse gas displacement factors of wood product substitution
Environ. Sci. Pol.
(2010) - et al.
Framework and computational tool for the consideration of time dependency in Life Cycle Inventory: proof of concept
J. Clean. Prod.
(2016) - et al.
Greenhouse gas emission timing in life cycle assessment and the global warming potential of perennial energy crops
Carbon Manage.
(2015) - et al.
The ESPA (Enhanced Structural Path Analysis) method: a solution to an implementation challenge for dynamic life cycle assessment studies
Int. J. Life Cycle Assess.
(2014) - et al.
Implementing a dynamic life cycle assessment methodology with a case study on domestic hot water production
J. Ind. Ecol.
(2017)
Implementation of IPCC impact assessment method 2007 and 2013 to ecoinvent database 3.2
15804: Sustainability of construction works-Environmental product declarations-Core rules for the product category of construction products.
CO2 emissions from biomass combustion for bioenergy: atmospheric decay and contribution to global warming
GCB Bioenergy
Application of probability distributions to the modeling of biogenic CO2 fluxes in life cycle assessment
GCB Bioenergy
How to take time into account in the inventory step: a selective introduction based on sensitivity analysis
Int. J. Life Cycle Assess.
Dynamic life cycle assessment: framework and application to an institutional building
Int. J. Life Cycle Assess.
AGRIBALYSE, the French LCI Database for agricultural products: high quality data for producers and environmental labelling
Oilseeds Fats Crops Lipids
Towards a general framework for including noise impacts in LCA
Int. J. Life Cycle Assess.
Long-term emissions from landfills should not be disregarded
Int. J. Life Cycle Assess.
Transparency in LCA - a heretical request?
Int. J. Life Cycle Assess.
The Computational Structure of Life Cycle Assessment
Approaches for inclusion of forest carbon cycle in life cycle assessment – a review
GCB Bioenergy
Discounting and the environment should current impacts be weighted differently than impacts harming future generations?
Int. J. Life Cycle Assess.
Application of uncertainty and variability in LCA
Int. J. Life Cycle Assess.
The Physical Science Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
Need for relevant timescales when crediting temporary carbon storage
Int. J. Life Cycle Assess.
Time-adjusted global warming potentials for LCA and carbon footprints
Int. J. Life Cycle Assess.
Accounting for time-dependent effects in biofuel life cycle greenhouse gas emissions calculations
Environ. Sci. Technol.
Cited by (45)
Synergy between industry and agriculture: Techno-economic and life cycle assessments of waste recovery for crop growth in glasshouses
2023, Journal of Cleaner ProductionSynergies and gaps between circularity assessment and Life Cycle Assessment (LCA)
2023, Science of the Total EnvironmentCoupling big data and life cycle assessment: A review, recommendations, and prospects
2023, Ecological IndicatorsCalifornia's harvested wood products: A time-dependent assessment of life cycle greenhouse gas emissions
2023, Science of the Total Environment