Use of life cycle assessment (LCA) to advance optimisation of radiological protection and safety

Life cycle assessment (LCA) is a modelling technique used to determine the cradle-to-grave environmental and human health impacts from the production of a good or the provision of a service. Radiological protection may benefit from employing tools like LCA to obtain a broader perspective and enable comparison with analyses of non-radiological systems. Despite structural similarities to other well-established decision-aiding techniques (DATs), the impact assessment within LCA (i.e. LCIA) is not commonly used in the optimisation of radiological protection process. This paper provides a brief review of LCA, including LCIA, along with more traditional DATs (such as multi-attribute utility analysis) used in the optimisation process for comparison. Basic concrete shielding was considered as a simple, illustrative example; concrete attenuates emissions from a radiation source but is also associated with a financial cost as well as costs with respect to energy, material, and water use. LCA offers quantification of these and other key resources (termed ‘impact categories’). Ultimately, we offer that, depending on the circumstance, LCA can be a useful tool in radiological protection decision-making, complementing existing techniques.


Introduction
Radiological protection is an important aspect of occupational health and safety for those who work with radiation or radioactivity, and it seeks to protect patients, members of the public, and the environment from detrimental effects of unnecessary exposure to ionising radiation (ICRP 2007).The radiological protection community aims to follow a graded approach in the optimisation of protection and safety (Clement et al 2021); the implementation of tools like life cycle assessment (LCA) in radiological protection decision-making may yield a broader perspective than traditionally taken and enable comparison with analyses of non-radiological systems.LCA is used to evaluate the potential environmental and human health impacts of products or services throughout their life cycle from raw materials extraction (cradle) to end-of-life (grave) (ISO 2006a(ISO , 2006b)).To the best of the authors' knowledge, LCA has not been extensively used in the context of optimisation of radiological protection.This work aims to demonstrate the applicability of LCA as a decision-aiding technique (DAT) for optimisation of radiological protection using the specific example of concrete as a shield for a hypothetical radioactive source, acknowledging that LCA may also play a useful role in other areas of decision-making, e.g. in deciding if a practice or action is justified.

Optimisation of radiological protection and DATs
The principle of the optimisation of radiological protection states 'the likelihood of incurring exposure, the number of people exposed, and the magnitude of their individual doses should all be kept as low as reasonably achievable (ALARA), taking into account economic and societal factors' (ICRP 2007).International Commission on Radiological Protection (ICRP) Publication 146 explicitly incorporates environmental factors in the optimisation of radiological protection as well (ICRP 2020).
Optimisation of radiological protection is a continuous, cyclical process (i.e.iterative).In brief, the process for the optimisation of radiological protection is as follows: evaluate the situation, identify protective options, select the best (i.e.optimal for a given scenario) option, implement the protective option, and regularly review the situation (ICRP 2006; figure 1).
The process of optimisation of radiological protection has progressively become more reflective of stakeholder involvement, personal safety, and safety culture (ICRP 2006).As such, the optimisation of radiological protection process has become more judgment-based (via emphasis of operational procedures, good practices, and qualitative approaches) rather than strictly quantitative (ICRP 2006).While determination of 'reasonable' lacks universal benchmarks, an example, qualitative framework describing factors to consider is the '3Rs' approach, which broadly includes relationships, rationale, and resources (Wieder et al 2022;figure 1).'Relationships' refers to the fostering of positive stakeholder relationships, building mutual trust, and meaningfully engaging in the co-expertise process.'Rationale' refers to ensuring and communicating sound technical, contextual, and ethical reasoning in the determination or pursuit of a protection goal or goals in a given circumstance.'Resources' refers to consideration of availability, and responsible use, of natural resources, technology, time, finances, human capital, etc in development and implementation of a decision.
DATs can be used to quantitatively support decision-making in radiological protection.Example DATs include cost-benefit analysis, cost-effectiveness analysis, and multi-attribute utility analysis (MAUA;ICRP 2006).The use of a particular DAT depends on the data available and the complexity of the given situation.Arguably, the most straightforward DAT is cost-benefit analysis, which balances factors affecting cost and collective dose (ICRP 2006, IAEA 2019; figure 1).In this approach, the cost of detriment and cost of protection curves are summed to yield the total cost; the minimum of this curve is the optimum solution.The monetary value of a person-sievert (cost/personSv) has historically been used to compare the costs of protection options (dollar value of the protective option to reduce the dose by a given amount) and the costs of detriment (dose avoided due to implementing protective option of a certain cost) in the cost-benefit analysis (ICRP 2006).Cost-effectiveness analysis (figure 1), is a technique used to remove non-cost-effective options by comparing the cost-effectiveness ratio (∆X/∆S in units of monetary cost/personSv; see figure 1 where X is the cost for protection, monetary cost, and S is the individual dose, Sv) to a selected monetary value of the person sievert reference criteria (ICRP 2006).MAUA (elaborated on further in section 2.3) is used if there are relevant attributes (i.e.metrics) to consider outside of cost and dose or detriment (see section 3 ).Such attributes may include, for example, spread of exposure in space and time, perceived level of risk, collective dose, and individual dose (ICRP 2006).Each attribute of interest is qualitatively or quantitatively weighted for importance, and then evaluated for each alternative (i.e.protective option) under consideration.The overall utility for each alternative is calculated as a weighted sum of individual attribute utilities, and the alternative with the highest overall utility is presumably the preferred option (Jansen 2011).DATs provide the opportunity to conveniently compare protective options to determine the best one for a given situation taking into account any necessary external factors, particularly the opinion or relevant stakeholders.The ICRP (2006) recommends that while there are quantitative metrics to consider, the qualitative metrics should not be ignored in the decision-making process.

Life cycle (impact) assessment
LCA is a modelling technique aimed at determining the potential impacts on humans and the environment from the production of a good or the provision of a service (Rebitzer et al 2004, Huijbregts et al 2016, Paulillo et al 2020).The method uses a 'cradle-to-grave' approach, meaning that it considers the impact from all steps to create a good or provide a service.Steps include: raw material extraction (cradle), energy acquisition, processing materials, manufacturing products, consumption/use, reuse, storage, recycling, and disposal (grave) (Pennington et al 2004, Rebitzer et al 2004;figure 2).
The LCA framework includes four stages: (1) goal and scope definition, (2) life cycle inventory (LCI), (e.g. using an LCI database such as EcoInvent; Wernet et al 2016), (3) life cycle impact assessment (LCIA, e.g. using an LCIA database such as ReCiPe; Huijbregts et al 2016), and (4) interpretation.An LCI database reports and quantifies the elementary flows to and from the environment for a given process (e.g. for a concrete production process, cement, aggregate, water, and energy are removed from the environment or technosphere whereas CO 2 , 222 Rn, etc. are emitted).The technosphere encompasses all flows within the economy, so in the aforementioned case, the energy used (e.g.electricity) comes from other processes in the economy/technosphere (e.g. a power plant).An LCIA database converts the emissions or input of individual energy and material flows to or from the environment and quantifies the midpoint or endpoint impacts (discussed below) associated with that flow.An analyst may choose among dozens of LCIA databases depending on the impact categories of interest, processes required for the system, or robustness of the database as documented in the literature.The open-access software OpenLCA (https://openlca.org)enables utilisation of LCI and LCIA databases for cradle-to-grave assessment of a product or service of interest.
During the LCI (stage 2 above), the entire process of creating a product or providing a service is considered and modelled.The final output from the LCI analysis defines the elementary flows (i.e.materials and energy) to and from the environment for the steps within the system boundaries.These flows, collectively stored in an inventory database (e.g.EcoInvent) then feed into the impact assessment (stage 3 above).An LCIA is conducted to calculate potential impacts to the environment and human health (Rebitzer et al 2004).Impacts can be assessed at two different stages: midpoint and endpoint.Midpoint impact categories focus on a single environmental or human health impact, such as global warming potential Table 1.Life cycle impact assessment steps, their purpose, and a relevant example.

LCIA step a
Purpose Example

Categorisation
To choose appropriate impact categories.
Global warming potential, marine ecotoxicity, and particulate matter are categories one could pick.

Classification
To connect elements in the inventory with associated impacts.
Emissions such as CCl2F2, CO2, and CH4 contribute to global warming potential.Note that CCl2F2 also contributes to the stratospheric ozone depletion impact category.

Characterisation
To determine the strength of the connection to the midpoint impact category.
GWP is expressed in kg CO2 eq. in ReCiPe 2016.For example, in a 20 year time horizon, methane has a characterisation factor of 84 kg CO2 eq./kg CH4 emitted whereas as CCl2F2 has a characterisation factor of 10 800 kg CO2 eq./kg CCl2F2 emitted.

Normalisation
To compare different flows and processes via a suitable reference value.Values will be unitless and range from 0 to 1 after being normalised.
Compare across categories by using some internal or external reference value for each impact category (consider relative biological effectiveness or relative toxicity).

Grouping
To group midpoint impact categories into higher-level endpoint categories such as human or ecosystem health.
Global warming potential and particulate matter could be grouped for damage to human health.Marine ecotoxicity would be in its own grouping for damage to the ecosystem.

Valuation
To establish the weight of each midpoint impact to its associated endpoint impact.
Value-judgements such as age-weighting, time horizon choice, and distance to target.This step is currently under discussion in the community.
(GWP) or ozone depletion, whereas endpoint categories aggregate midpoint impacts into higher-order impact categories, such as ecosystem health or human health.For example, the ReCiPe 2016 database includes midpoint impact categories such as human toxicity, water use, acidification, mineral resource scarcity, fossil resources, stratosphere ozone depletion, ionising radiation, fresh water and marine ecotoxicity, and particulate matter formation (Rebitzer et al 2004, Huijbregts et al 2016 ).The three endpoint impact categories included in ReCiPe 2016 are damage to human health, damage to ecosystem quality, and damage to resource availability (Huijbregts et al 2016 ).The specifics of midpoint and endpoint categories are defined within the framework of the LCIA database and may vary for different databases.
Although LCIA is one stage within the overall LCA process, it is itself comprised of several steps (summarised in table 1) the first of which is the determination of the midpoint impact categories of interest for a given analysis (referred to as 'categorisation').Then, each flow to or from the environment or technosphere is linked with their associated impact (referred to as 'classification').Note that one flow can be linked to multiple impacts (e.g.Freon (CCl 2 F 2 ) emissions are linked to both GWP and stratospheric ozone depletion).Overall impacts are then quantified by multiplying each flow by their respective characterisation factor for each impact category it is linked with (referred to as 'characterisation').A characterisation factor is a value that demonstrates the relative importance of an emission.Characterisation factors are defined within each LCIA database.The methodology used for each characterisation factor is based on the unique impact category as defined within a specific impact database.For example, the GWP characterisation factor, as described in the ReCiPe 2016 database (Huijbregts et al 2016), is calculated as a ratio of the absolute GWP (AGWP) of 1 kg of a given greenhouse gas (GHG) with respect to the AGWP of 1 kg of CO 2 , where the GWP is a measure of the additional radiative forcing integrated over time.A description of all the characterisation factors used to quantify the impact categories in this study can be found in the ReCiPe 2016 database documentation (Huijbregts et al 2016).
Only the first three steps of the LCIA are required (i.e.categorisation, classification, and characterisation), providing the practitioner with a set of impact scores expressed across a variety of midpoint impact categories.Normalisation expresses relevant flows and processes relative to a specified reference value thus providing a unitless value for each impact category that are all within the same order of magnitude for ease of comparison.To be an ISO compliant study, the unnormalised and unweighted scores must be reported (ISO 14044 2006b).However, normalisation at the midpoint impact categories enables a practitioner to directly compare impacts of different products or services.Additionally, endpoint impact categories are computed based on the normalised midpoint impact categories (i.e.grouping).Weighting factors can also be applied to better reflect different stakeholder priorities (i.e.valuation).It is ultimately at the discretion of the practitioner whether to use the midpoint or endpoint impact categories for the final analysis, although the decision should be made in consultation with stakeholders.Some stakeholders may prefer midpoint impact categories, for example, to reduce inherent subjectivity and uncertainty in the analysis that would otherwise be present in the assessment of endpoint impact categories (Pennington et al 2004).More specifically, the modelling of endpoint impact categories includes value judgements such as (a) what reference value to normalise the data against, (b) which impact categories are of main concern to the stakeholders, and (c) how to weight those impact categories.A multi-criteria decision-making analysis or multi-criteria decision analysis (MCDA) method can be used within LCA to consider the input from various stakeholders (Rajagopalan et al 2021).For example, stakeholder preference can be incorporated into the MCDA method when determining end of life strategies, assessment of those strategies, and refinement with final evaluation of the strategies used (Alamerew and Brissaud 2019).Multi-criteria methods are discussed in greater detail in section 2.3.Regardless of whether midpoint or endpoint impacts are modelled, the assumptions (e.g.cultural biases, discussed below) for the models should be consistent for the impacts evaluated (Pennington et al 2004).
The ReCiPe 2016 impact database includes three cultural biases: individualist, hierarchist, and egalitarian.The cultural biases in LCIA are intended to reflect and incorporate the different values and beliefs (i.e.philosophies) of different, representative culture types.The individualist culture prioritises individual safety and well-being, and it values liberty and self-reliance (Douglas 2003).In the context of LCIA, the individualist approach assumes that an abundance of resources exists (Steg and Sievers 2000) and as such, future generations will be able to adequately meet their needs in any scenario (Frischknecht et al 2000).In the ReCiPe 2016 database, the individualist cultural bias employs the shortest time horizon at 20 years (Huijbregts et al 2016).The hierarchist culture values hierarchy, order, and stability, whereas the egalitarian culture values equality and justice (Douglas 2003).In the context of LCIA, both the hierarchist and egalitarian cultures are concerned for current and future generations, and within the respective biases special attention is given to impacts on children and the elderly (Frischknecht et al 2000).The hierarchical bias in LCIA incorporates the belief that resources are scarce and that expert bodies should determine acceptable risks (Steg and Sievers 2000).The time horizon considered within the hierarchist bias is 100 years (Huijbregts et al 2016).The egalitarian approach aims to reduce exposures to future populations, is more conservative than the other two cultural biases (Frischknecht et al 2000), and views resources as depleting (Steg and Sievers 2000).The egalitarian bias considers an infinite time horizon, typically modelled using 100 000 years (Huijbregts et al 2016).The cultural bias selected will affect how the characterisation factors are determined due the different underlying philosophies.For example, depending on the bias implemented, carcinogenic substances are included if there is a definite link to cancer (individualist), if there is sufficient proof of a link to cancer as determined by international scientific consensus (hierarchist), or if there is any possible link to cancer (egalitarian; Hofstetter 1998, Goedkoop andSpriensma 2001).In any of these biases, age-weighting could be applied to emphasise the importance of certain age-groups to the scenario, although age-weighting is typically not included in the default assessments.
Results of the LCIA can then be interpreted to inform a decision.For example, consider concrete as the product of interest for an LCA.The production of concrete requires input flows of materials such as water, aggregate, energy, and cement.The impact assessment may consider a midpoint impact category such as GWP for the concrete production, where the impact is primarily driven by GHGs emitted from manufacturing and transporting the concrete as well as any materials to make the concrete.For illustrative purposes, this example of concrete for shielding is referenced throughout the work to demonstrate the LCA process.

Multi-criteria decision making (MCDM)
This work suggests that using LCA in radiological protection as a supplement to more traditionally utilised DATs, such as MAUA could be beneficial.An overview of MAUA methodology is provided for context and comparison.MAUA is a DAT currently used in the radiological protection community, along with other fields, and is broadly applicable to decision-making situations where not all attributes are quantifiable in terms of money (ICRP 1990).In this method, the total attractiveness (i.e.utility) for each alternative option is calculated considering all relevant utilities for each attribute (Roth 1994, Roth and Bobko 1997, Davis et al 2000, Kelly and Thorne 2001).For example, for a radiological release, some attributes that will be considered are: individual dose, spread of exposure in both space and time, perceived risk, and collective dose (ICRP 2006) whereas in a field like industrial or organisational psychology, MAUA could be used to evaluate tradeoffs in training programs or selection methods (Roth 1994).The MAUA method is considered beneficial as it is both repeatable and modifiable (Davis et al 2000).(Boudreau 1988, ICRP 1990, Roth and Bobko 1997, Davis et al 2000, Kelly and Thorne 2001, Scholz and Tietje 2002, Carretero-Gómez and Cabrera 2012).

MAUA MAUT
State the objective.Analyse the situation.Define the system.Consider alternatives and associated consequences.Identify all relevant attributes (utilities).
Describe alternatives and attributes.Weight the attributes.
Describe weights that will be assigned.Assign weights to each attribute.Quantify each attribute.
Describe quantification.Convert it to a utility between 0 (best case) and 1 (worst case).
Determine if there needs to be a hierarchy.Ensure there is a value for each attribute in every alternative option.Determine the utility function for each attribute.Calculate the total utility of each option.→ sum individual weighted utilities.
Calculate the total utility for each option.→ consider necessary cutoffs.Evaluate results and make a decision.
Evaluate and disseminate the results.
The steps of the MAUA method are well documented (e.g.Boudreau 1988, ICRP 1990, Roth and Bobko 1997, Davis et al 2000, Kelly and Thorne 2001, Carretero-Gómez and Cabrera 2012).The general steps are outlined in table 2 and juxtaposed with the steps utilised in multi-attribute utility theory (MAUT) as described by Scholz and Tietje (2002).Both MAUA and MAUT have been abbreviated as MAU (e.g.Bose et al 1997, Davis et al 2000).Like MAUA, MAUT is used to evaluate tradeoffs for alternatives (i.e.options) having multiple attributes, particularly when no explicit comparison method is available (von Winterfeldt and Fischer 1975, Keeney 1977, Dyer et al 1992, Scholz and Tietje 2002, Jansen 2011).Following the methodological steps outlined in table 2, one can see the clear similarities between MAUA and MAUT methods, even though some rows have more detail than others.Thus, these two methods are functionally the same, just applied in different fields.
Due to the possible uncertainties associated with value judgements, measures should be taken to include views of all stakeholders, consider all factors relevant to the decision-making process, consider all alternative options, be unambiguous, and follow consistent logic (Kelly and Thorne 2001).In MAUA and MAUT, the preferred option of all alternatives considered is the one that produces the highest total utility, which represents the lowest adverse effect or the best outcome (ICRP 1990, Dyer et al 1992, Kelly and Thorne 2001, Jansen 2011).However, if multiple alternatives have total utilities that are identical, there is no preference between these alternatives.
MCDA, which is used in LCA, is a broad field that contains a variety of approaches.Decision methods that employ 'multi-criteria' attributes are often collectively referred to as methods in MCDM (Velasquez and Hester 2013, Morgan and Cohon 2017, Taherdoost and Madanchian 2023).Velasquez and Hester (2013) provide an excellent review of multi-criteria methods.

Framework for LCA as a DAT
This paper seeks to describe how LCA could be utilised as a DAT within the process of optimisation particularly when balancing local impacts with global impacts, that is, provide an option for provision of quantitative metrics for considering 'societal, environmental, and economic aspects' (ICRP 2020) of a scenario.It could also be used as a supplement to DATs more commonly used in radiological protection, e.g.MAUA.The midpoint impact results from LCA can be incorporated as additional quantitative metrics into a MAUA calculation to promote the consideration of potential environmental and human health impacts associated with the creation of the protective option (e.g.concrete).In essence, if LCIA midpoints are considered in MAUA, at the end of the assessment, they are converted to a higher-order aggregated metric analogous to the endpoint impact categories as a result of the way the overall utility is calculated in MAUA.Due to these similarities in how endpoint impact categories are calculated in LCA (i.e.weighting, considering stakeholder input, etc), it would be repetitive to include the endpoint impact categories in MAUA.However, an LCA reporting endpoint damages could be considered as a stand-alone, alternative, DAT that could be utilised individually or in conjunction with other established DATs already used in the optimisation of radiological protection such as cost-benefit analysis.The inclusion of LCA into the optimisation of radiological protection enables consideration of different time horizons and potential global environmental and human health impacts associated with creating or using a given protective option.
Depending on the chosen boundary conditions, an LCA can evaluate impacts over the entire lifecycle from extraction of raw materials (cradle) to end-of-life (grave) for a protection option.Something not currently considered in the optimisation of radiological protection is the associated risks/impacts that occur from the production of that protection option, for example, in the production of concrete.
In the ReCiPe 2016 LCIA database, the damage to human health endpoint impact category is expressed in terms of disability adjusted life years (DALY) for radiological and non-radiological related exposures (Huijbregts et al 2016).In the optimisation of radiological protection, the concept of detriment (Sv −1 ) is commonly used rather than DALY, but the two methodologies represent the same type of idea: to quantify the burden of disease from exposure to low dose or low dose-rates (Murray 1994, Shimada and Kai 2015, Cléro et al 2019, ICRP 2022).The concept of detriment was established by the ICRP in Publication 26 (1977) and is only used in radiological protection.Stochastic health effects, expressed as detriment, are predicted by calculating the nominal risk and then weighting this value in terms of years of life lost and quality of life by considering the severity of the effect (Cléro et al 2019, ICRP 2022).DALY was created for the World Health Organization in the mid 1990s (e.g.Murray 1994, Murray andLopez 1996) and is used in fields such as policy, economics, and epidemiology (Chen et al 2015).Further, DALY quantifies health effects by determining the years lived disabled and the years of life lost from an unexpected accident or disease by combining mortality and morbidity (Chen et al 2015, Shimada andKai 2015).
Since LCA considers the global impacts for the production of a product, can vary the time horizon for both the mid-and endpoint impact categories, and the endpoint impact results are expressed in terms comparable to those used in radiological protection (DALY and detriment), the addition of LCA as a DAT is warranted whether it be considered alone (using endpoint impacts) or within MAUA (using midpoint impacts).A simple LCA of concrete shielding follows as proof of principle.

Simplified case study
We propose an illustrative scenario as a proof of principle for utilisation of LCA in the optimisation of radiological protection.Consider a room housing a gamma-emitting point source.The source is shielded by concrete, which exponentially attenuates source emissions.
Although dose is reduced as shield thickness increases, there are additional cradle-to-grave impacts associated with that increased concrete production, e.g.GWP, terrestrial ecotoxicity, mineral resource scarcity, and water consumption as defined in the ReCiPe 2016 impact database.Of note, the ionising radiation potential (IRP) midpoint category refers to the radiation that is emitted through the creation of the shielding option (e.g. the concrete) as opposed to any unshielded radiation emitted from the gamma-emitting source.To avoid confusion between the IRP impact category and the ionising radiation shielded by the concrete, this impact category was not used for this simple proof of principle but could be used in future analyses to evaluate the balance between local radiological protection and global radiological emissions.
Four concrete processes with varying compositions from EcoInvent v3.7 (Wernet et al 2016) were chosen for the LCIA analysis.All four concrete processes in the EcoInvent LCI database were titled 'concrete production, 20 MPa, ready-mix' assigned for 'rest of world' (RoW) with different compositions (Supplementary data A).The four predefined concrete compositions analysed were: Portland cement (Portland), 21%-35% of limestone (Limestone), 21%-35% of alternative constituents (Alternative), and 36%-55% pozzolana and fly ash (PFA).All three cultural biases were evaluated for the Portland concrete composition to illustrate how bias may influence midpoint impacts.The hierarchist cultural bias was further applied for all four concrete compositions for all impact categories considered.A concrete thickness up to 200 cm was considered for this illustrative example.The linear rate of change of the LCA impacts vs. concrete thickness was visualised using John's Macintosh Project (JMP Pro 16, SAS Institute, Inc., Cary, NC).After determining the marginal impact (i.e.rate of change in the impact versus thickness; slope) for each impact category, the data was normalised by calculating the slope as a percentage of the largest slope for a given impact category.

Results and discussion
Depending on the impact category, the cultural bias implemented may yield meaningful differences in the overall impacts.Overall, the water consumption, mineral resource scarcity, and GWP across the three cultural biases are fairly comparable.Water consumption is identical for all three biases, the mineral resource scarcity is slightly higher in the individualist as opposed to the hierarchist and egalitarian scenarios, and the GWP decreases slightly as the time horizon in question increases (figures 3(a)-(c)).This slight decrease as the time horizon increases is likely due to a constant amount of kg CO 2 eq.being emitted and the exposure being extended to a larger number of people.The terrestrial ecotoxicity impact is over double in the hierarchist and egalitarian bias as compared to the individualist bias (figure 3(d)).The underlying details that drive the change in impact with respect to the chosen cultural bias is beyond the scope of this paper.For this proof of principle, the cultural bias chosen was essentially inconsequential for the GWP, mineral resource scarcity, and water consumption impact categories but very impactful for the terrestrial ecotoxicity impact category.
Comparative analysis of dose reduction and key LCA impact categories can elucidate the potential impacts from the production of concrete radiation shielding.Photons are exponentially attenuated in concrete.Consequently, the dose rate drops off quickly as shielding thickness increases.Despite the dose rate of the gamma-emitting source exponentially decreasing with increasing thickness, the environmental and human health impacts associated with increasing the concrete thickness continue to expand, linearly, as modelled using LCA.Impacts of mineral resource scarcity, GWP, terrestrial ecotoxicity, and water consumption associated with the production of concrete with Limestone, Alternative, PFA, and Portland compositions show the expected linear relationship with respect to the changing thickness of the concrete (figure 4).
The rate of change (i.e.slope or marginal impact) of the aforementioned impact categories with respect to concrete thickness differs based on concrete composition and impact category (figure 4).The impact of water consumption to produce the concrete is least dependent on concrete composition (as seen by the close grouping of impact curves in figure 4(c)), where Limestone yielded the highest marginal impact of 0.112 and Alternative yielded the lowest marginal impact of 0.091 (Supplementary data Table B1).While the marginal impacts are nearly the same for each concrete composition, the absolute water consumption increases with the concrete thickness.For example, the water consumption at the lowest thickness considered (i.e.20 cm) ranged from 1.81 m 3 for Alternative to 2.24 m 3 for Limestone, a difference of only 0.43 m 3 .This difference increases to 2.13 m 3 for a concrete thickness of 100 cm.Thus, when evaluating or interpreting rate of change (for comparative purposes) it is important to keep in mind that differences in impact become more pronounced as thickness increases.Further analysis of the marginal impact normalised to the most impactful concrete composition shows that the impact of water consumption associated with the Portland composition is 8% less than the impact of water consumption associated with the Limestone composition.The normalised impact of water consumption with respect to concrete composition is up to nearly 20% lower for the Alternative concrete composition.As previously discussed, normalisation of LCIA results eases direct comparisons across impact categories.For demonstrative purposes, figure 5 is included here to show the normalised marginal impacts alongside the impact curves of figure 4.
For the GWP, the marginal impacts are similar for the Limestone and Portland concrete compositions, with marginal impacts of 32.23 and 29.74, respectively, while the Alternative and PFA concrete compositions yield similar marginal impacts of 22.31 and 19.66, respectively (figure 4(b); Supplementary data Table B1).The normalised GWP marginal impact associated with the Portland composition is within 8% that of the maximum (Limestone) whereas the PFA and Alternative compositions are 30% to almost 40% less than the normalised global warming marginal impact from Limestone (figure 5).
Like the water consumption and GWP categories, the Limestone composition was the most impactful concrete composition for the terrestrial ecotoxicity impact category.However, impact of the Limestone composition on the terrestrial ecotoxicity impact category was 16%-25% greater than the other concrete compositions evaluated (figures 4(d) and 5).The Portland, Alternative, and PFA compositions were within 8.5% of one-another.
For the final impact category considered in this proof of principle, i.e. mineral resource scarcity, the Portland composition, as opposed to the Limestone composition, is the most impactful with a marginal impact of 0.266 (figure 4(a)).The Alternative and Limestone compositions are 33%-36% less impactful than the Portland composition.Interestingly, the PFA composition was the least impactful with a marginal impact of 0.106; a difference in normalised marginal impact of over 60%.
To help conceptualise some of these impacts, it is useful to consider everyday examples like carbon dioxide emissions from air travel and water volumes in a bathtub.Consider the impact for the Alternative composition of a 30 cm thick concrete shield.A transcontinental flight between San Francisco, California (SFO) and New York (JFK) would emit 296.1 kg of CO 2 per passenger (ICAO n.d.).About 2.3 people could travel across the continental United States and produce the same amount of CO 2 equivalents that are emitted in the production of a 3 m × 3 m × 0.3 m concrete shield.The amount of water consumed in the production of that same volume of concrete would fill about 17 bathtubs at maximum capacity (159 l; Kohler 2019).For comparison, the use of the Limestone composition, for a shield of the same dimensions, would result in approximately 21 bathtubs of water consumed (an increase of about 24%) and CO 2 emissions equivalent to nearly 3.3 people flying across the country (an increase of 44.5% more GHGs emitted).
Once the midpoint impact assessment is complete, the practitioner can use LCA as a DAT by (a) incorporating the un-normalised, un-weighted midpoint results into MAUA, or (b) continuing the LCA through to the endpoint impacts.For the example assessment presented here, the LCIA results were normalised to the maximum marginal impact for each midpoint category to show the relative impacts for four concrete compositions (figure 5).Continuation of the LCIA to the endpoint impacts requires implementation of a weighting scheme, which should be developed in collaboration with the stakeholders.The weighting scheme can have a drastic influence on the overall mid-and endpoint impacts; thus, inclusion of key stakeholders yields more applicable outcomes.For example, consider two different weighting schemes (that could be based on stakeholder preference): (1) equal weighting and (2) 50% weighting on water consumption, 10% on mineral resource scarcity, and 20% on both GWP and terrestrial ecotoxicity.In the case of equal weighting, Portland has the largest overall midpoint impact (92.0%) followed by Limestone (90.9%),Alternative (73.7%), and PFA (65.5%).With regards to the second weighting scheme, the Alternative (76.5%) and the PFA (74.8%) remain the least impactful, but Limestone (96.4%) becomes more impactful than Portland (91.1%).This variation can occur with any method where weighting is involved (e.g.LCA, MAUA, etc).Thus, for this proof of principle, depending on the weighting scheme used, either Portland or Limestone are the most impactful, while PFA is the least impactful in both cases.Of course, other weighting schemes are possible, which will influence the relative impact of the concrete compositions.Recall, the unweighted impacts show that the Limestone has the largest impact for water consumption, terrestrial ecotoxicity, and GWP, while Portland has the largest impact for mineral resource scarcity.The difference between Portland concrete and the most impactful concrete (Limestone) for water consumption, terrestrial ecotoxicity, and GWP is a total of 32.5%.For mineral resource scarcity, the Portland concrete is at least 33.1% more impactful than the next most impactful concrete (Alternative) and 36.4% more impactful than Limestone concrete.Meanwhile, had mineral resource scarcity not been chosen as a midpoint impact category, Limestone would have clearly yielded the largest midpoint impacts, particularly for an equal weighting scheme.These results emphasise the importance of both the applied weighting scheme and the impact categories selected for the analysis.
Additional DATs, cultural biases, weighting, or impact categories could aid in the decision for the best holistic option to use for shielding the (hypothetical) radiation source.While the dose rate is reduced exponentially with increasing thickness, the overall impacts from producing the concrete continually increases (figure 4).The goal is to optimise the shielding ability without unduly increasing associated impacts.Although of course adequate shielding is essential, superfluous concrete shielding will be associated with an unnecessary financial cost as well as costs with respect to energy, material, and water use.LCA in this respect is useful for decision-makers and other stakeholders to have additional quantitative information concerning impacts beyond the radiological in order to make the most informed decision possible.

Conclusions
LCA, as emphasised with the proof-of-principle scenario presented here, is a viable addition to the DAT toolbox by providing an additional, broad metric for consideration in radiological protection decision-making.Understanding how each impact changes depending on the flows into and out of the process(es) of interest for a given protective option provides more information than typically considered in radiological protection.For example, LCA considers the time horizon, cultural bias (individualist, hierarchist, egalitarian), and the cradle-to-grave impacts that are associated with the concrete production to give a broader understanding of the impact from activities, such as creating concrete shielding.This proof of principle is not intended to suggest a whole-scale change in the radiological protection optimisation process to an LCA approach.Rather, we emphasise the usefulness of LCA (and LCIA) to broaden the landscape of impacts considered in optimisation of radiological protection.

Figure 1 .
Figure 1.The four components of the optimisation of radiological protection cycle are listed in the cycle.A few things that are considered when determining the best option for protection are shown as well including the 3 Rs of Reasonableness and a few decision-aiding technique examples such as cost-benefit analysis and cost-effectiveness analysis.Adapted with permission from Wieder et al (2022).Adapted with permission from ICRP, (2006).The Optimisation of Radiological Protection -Broadening the Process.ICRP Publication 101b.Ann.ICRP 36 (3).

Figure 2 .
Figure 2. The LCA framework is shown on the left (©ISO.This material is reproduced from ISO 14040 (2006a) with permission of the American National Standards Institute (ANSI) on behalf of the International Organization for Standardization.All rights reserved.)and the 'cradle-to-grave' steps generally considered for an inventory analysis are emphasised on the right.

Figure 4 .
Figure 4.The hierarchist LCIA results are shown compared to thickness of the concrete shield where (a) is mineral resource scarcity in kg Cu eq., (b) is the global warming potential in kg CO2 eq., (c) is the water consumption in m 3 , and (d) is the terrestrial ecotoxicity in 1,4-DCB eq.The cement composition in the concrete is differentiated by both the colour and shape: blue asterisks denote Alternative, orange circles denote Limestone, green squares denote PFA, and purple triangles denote Portland cement compositions.

Table 2 .
Steps to conduct MAUA or MAUT