Connecting water footprint and water risk assessment: case packaging board

Fresh water is an essential resource for people, for the environment and for the industries. Water scarcity and the availability of fresh water is a global concern (Rockström et al. 2009), making numerous industries vulnerable to water disruption throughout their operations, posing a threat to company's production levels, profit margins and even ‘licence to operate’ in water stressed regions (Ceres 2010). This has led to investors and other stakeholders seek for more information and demanding companies to measure and communicate their water use and the related risks.

Measuring and reporting water-related risks and opportunities is complex, and the methodologies are still developing (Ceres 2010). However, a variety of approaches and institutions promoting sustainable water management exist (Kounina et al. 2013), guiding the development of water saving products and services. Water footprint (WF) is a tool that aims at describing the impact of water use on humans and ecosystems due to changes in water quality and quantity, making it possible to reveal mitigation points and critical phases along the supply chain of a product, process or a service. The development of the method has been rapid and multifaceted, and as a result, the ISO Standard 14046 has been launched in 2014 to harmonise the methodology. A WF assessment conducted according to this standard is based on a life cycle assessment, is a sum of the WF of different life cycle stages, identifies potential environmental impacts related to water, and includes geographic and temporal dimensions (ISO 14046 2014). Even though several methods have been developed proposing different freshwater use inventory schemes and impact assessment models to account for freshwater use in life cycle assessment, no single method is available which comprehensively describes all potential impacts derived from fresh water use (Kounina et al. 2013). As distinct from the product carbon footprint, which describes the global warming potential of a product and where the impact is global, the WF is a local indicator. The WF is intended to be used as a multi-use environmental communication tool at both product and organisation level.

WF and risk assessment methods can offer complementary elements for water sustainability assessment. Several types of risk analysis methods can be applied in environmental risk assessment, i.e. quantitative, semi-quantitative or qualitative methods. The selection depends mainly on the goal of the analysis, the available data, and resources:

• - Quantitative risk assessment is mainly used when the goal of the analysis is to reduce the probability of the already known risk. Statistical data of probabilities and expected values of consequences are used. The results are expressed by PLL values (the expected number of fatalities during 1 year), FAR values (the expected number of fatalities per 100 million exposed hours), IR values (individual risk – the probability that a specific individual being present at a certain position is killed during 1 year) and F–N curves (the frequencies of accidents with at least N fatalities) (Aven 2008).

• - Semi-quantitative methods are used to prioritise the risks according to probability, consequences or risk indexes. In these cases, the likelihood and consequences of risks are based on expert opinions of numerical values.

Qualitative risk assessments are used when the purpose is to identify new risks. In this case, the risks are prioritised according to the descriptions of likelihood and consequences. Qualitative methods are also successful when the aim is to develop the safety culture of the enterprise as there is possibility to use group methods to identify new risks.

At sectors, where water is an important processing medium, such as the pulp and paper industry, product quality is linked to the quality of process water, which in turn can be related to water consumption in the process. While reducing water can have net positive effect due to economic savings achieved by lower costs of raw water and lower volumes of waste water to be treated, it is difficult to reduce water use to maximum without sacrificing product quality (Wessman et al. 2012) or causing risks for production or environment (Kiuru 2011). It is thus important to know the quality and operational consequences of the water reduction measures aiming at minimising WF. Combination of risk assessment with the environmental assessment can assist in optimising the level of measures leading to the most sustainable end result.

This study aims at providing a framework for companies to evaluate their water impacts and improve water sustainability with a risk conscious approach. The overall goal is to connect two environmental tools, WF and water risk assessment, to create value for the water user by revealing hotspots of water impact within the product supply chain and providing additional information of the risks in planning the improvement measures and the threat the risks pose to the business. Product system of a packaging board will be used as an example to provide an overview of the implications of water within a supply chain and to evaluate the impact of different influent and effluent qualities on product WF. In this study, the water withdrawals do not take place in high water scarcity areas and the lack of water is not an issue. Therefore, the purpose is to point out how the WF is affected by the changing quality of input and output waters and what are the risks for the company related to it.

A WF assesses the potential, water-specific environmental impacts associated with a product, process or organisation. The recently accepted International Standard (ISO 14046 2014) provides principles, requirements and guidelines for water footprinting. The standard is based on life cycle assessment (ISO 14044). This study is carried out according to the ISO 14046 (2014), as a stand-alone assessment meaning that only impacts related to water are assessed. No consideration is given to other environmental impacts, such as global warming. WF inventory analysis includes compilation and quantification of water withdrawals and releases, types of water resources used, water quality data, and forms and locations of water use. The results are presented as water availability footprint (WAF), which gives an assessment of the contribution of the studied product's impacts related to pressure on water availability.

Because any change in water quality and in water quantity may have an impact on the availability or possible uses of water, it is important to consider both aspects in the impact assessment. The new standard does not however provide specific characterisation methods for assessing the impacts. Nor is there currently a consensus on the method for consistently associating inventory data with specific potential environmental impacts. For the purposes of this study, the stress method introduced by Boulay et al. (2011b) is chosen, because it measures the potential impacts on water stress caused by water consumption and degradation. Boulay scarcity indices distinguish different freshwater types and functionalities. WAF is assessed by differentiating source and quality of water by weighing the stress of each water type.

To determine quality of withdrawn water entering different unit processes within the supply chain, median surface water quality data is used, published by Boulay et al. (2011a). Water qualities for released effluents are determined by utilising available primary data and Quantis water database. Online tool is applied to determine water quality categories (Ciraig 2012) of effluent waters.

Before carrying out WF assessment, a pre-screening is recommended according to Figure 1.

Qualitative risk assessment method is applied to identify water risks by using the results of the WF analysis. Risk assessment is carried out on the identified hotspots of the supply chain.

The risk assessment consists of the risk assessment phase of risk management procedure (SFS-EN31010), i.e. risk identification, risk analysis and risk evaluation. As in many earlier risk analysis (Khan et al. 1997; Pollard & Guy 2001; Modarres 2006) the semi-quantitative index methods have been used in this study.

The risk assessment covers local risks which may threaten either the production processes or environment. Before the actual risk assessment can take place, possible risk objects must be connected with the local environment. This can be carried out, e.g. by binding up the process flow chart with the plant lay-out and by identifying geographical locations of the supply chain.

The first phase of risk assessment, water risk identification, can be carried out either in internal or external workshops by brainstorming to find out the main risks (Figure 2). It is important that meteorological and cultural experts who are familiar with local environmental circumstances are included in this process.

The risk analysis phase includes definition of the likelihoods and consequences of identified risks, and is recommended to be performed in brainstorming sessions, especially if there is no scientific data of probabilities or consequences. The likelihood of a risk can be assessed in a very pragmatic way, and it is tied both to the real return time of the risk and to the efficiency of risk treatment (Table 1).

Table 2 categorises the consequences of risks to the manufacturing process, quality of the product, environmental issues and image. The environmental and image consequences are modified from the environmental risk analysis method which is focussed on environmental consequences (Wessberg et al. 2007). This classification takes also into account the consequences of process water release to the effluent treatment plant. This is an important issue because a failure-free operation of a biological treatment plant normally requires a steady input effluent quality.

The final risk evaluation is done in a matrix of likelihood and consequences (Figure 3). It is notable that the analysed organisations should define the classes for the matrix in regard to their risk resilience by themselves. Depending on a case, a very unlikely phenomenon with extensive consequences can be classified into the risk category III, or even the category II.

 Figure 4 expresses how the water risk is approached from the WF to the risk treatment. WF assessment is carried out to reveal impacts of water use and the hotspots, i.e. points in the value chain where water use is most critical. Risk analysis will focus on those processes by analysing the local environmental circumstances and the main risks, and evaluating the likelihood and consequences to define the risk category. The results will help recognise environmental impacts and risk hotspots, and can be integrated in water management and strategic planning, and disclosed through reporting.

The case study provides an overview of the water-related impacts of producing packaging board in a water abundant area (Finland). The impact of different influent and effluent qualities on product WF and the related risks that may therefore threaten either the environment or production processes will be assessed.

The functional unit (FU) is one tonne of 295 g/m² folding box board. The product system is assessed with cradle-to-gate approach, i.e. the system boundaries include raw material suppliers, raw material transportations and manufacturing of product. Integrated board mill produces chemical pulp, board and part of the needed energy, and purchases additional energy and pulp (chemi-thermomechanical pulp (CTMP)), chips and softwood, referred as fibre raw materials; fillers and production chemicals referred as non-fibre raw materials; and fuels (peat, forest residuals and heavy fuel oil), that are transported to the mill site mainly by truck. Of the fibrous raw materials, the end product contains about 65% of CTMP and 35% of chemical pulp.

The assessment covers 99% of the mass of substances used in production of the board, or 17 out of 31 input flows. The inventory data for WAF calculation is a combination of primary data from the industry, and data obtained from Quantis (2013), which is based upon LCA database Ecoinvent (2010). Each unit process in the product system is studied to find out the key characteristics affecting the functionality of water: quantity and quality of water inputs and outputs, type of resources (surface water, groundwater) and geographical location.

Assessment of five scenarios of varying influent and effluent water qualities are carried out (Table 3). Different water qualities express different types of potential risks for production or environment. Case A is ‘business as usual’, i.e. the board mill withdraws ambient quality surface water (class 2a of the Boulay et al. (2011a) classification of water categories) and discharges treated effluent, that meets water class 3 criteria. Case B, namely ‘malfunction’, represents a case where the board mill's effluent treatment is malfunctioning and the released water is categorised as unusable (class 5). In case C, the input water is of highest quality water (class 1) instead of median quality surface water (2a) is withdrawn for the functions of the board mill, while the effluent is treated as usual. Cases D and E represent situations where the mill utilises water that is of lower quality than the ambient surface water. In case D, the mill circulates the process water by withdrawing the treated effluent (class 3). In case E, the input water is of lower quality (class 4) than the released effluent water, i.e. the water is purified by the mill.

For the purposes of increasing understanding of how geographical location impacts the result, water stress assessments in cases A, B and C are calculated with (1) Finnish country level stress factors and (2) generic European stress factors. In cases D and E, only Finnish factors are used. Impacts of potential improvements needed for pre-purification techniques at the mill site in cases D or E are not reflected in calculations.

To reveal the role and potential impact of a single actor on the total impact of the supply chain, only board mill's input and output water qualities are varied, while supply chain input and output water qualities are kept constant. Supply chain unit processes are assumed to be located near board manufacturing.

The consumption of soil moisture, also called as green water, is not considered in this study. The recently accepted WF standard (ISO 14046 2014) does not include the terms green, grey and blue water, which are terms related to the WF methodology that has been developed by WF network (Hoekstra & Chapagain 2008), in parallel with the LCA approach. In ISO 14046 (2014), the main term is water use, defined as use of water by human activity. In a note level, the standard gives an option to use water consumption which can occur because of evaporation, transpiration, integration into a product, or release into a different drainage basin or the sea. Extracting wood from semi-natural forests in northern countries like Finland or Sweden will likely not contribute to water scarcity, since water is an abundant resource in these areas (Launiainen et al. 2013). However, in this study, the use of fibre raw materials and fuel wood residuals contribute to product impact by the volume of water integrated to wood that becomes removed from forest.

As this study represents an ‘average’ board mill located in either Finland or in European area in general, also the likelihood and consequences of the risk were evaluated by using common information on the risk, rather than site-specific information.

In these cases risk mapping tools, such as the Water Risk Atlas provided by the World Resources Institute, offer valuable worldwide information on water availability and quality and even on reputational factors (Gassert et al. 2013). In addition, water stress indices, published, e.g. by Boulay et al. (2011a) or Pfister et al. (2009), offer a first insight into screening where water risks and opportunities may emerge and whether risk assessment should be carried out. This information was used also in this study instead of more site-specific information to evaluate the consequences and the likelihood of risks. As a matter of fact, in the risk assessment the used data was derived mainly from Water Risk Atlas. However, for the Finnish cases also the data available from Finnish studies and databases were used. To clarify the nomenclature, the term ‘hazard’ used in this study refers to the risk factors used in Water Risk Atlas.

Water inventory results of the product system are shown in Figure 5 by lifecycle stages, and represent the business as usual case. Total amount of withdrawn water per ton of product is 63 m³. Water withdrawals are dominated by the board mill (34% of the overall water inputs) just as are the outputs (35%, respectively). Board manufacturing includes water withdrawals of rather large volumetric amounts, but also intensive water recirculation, where substantial amount of energy is required to move the aqueous fibre suspensions (i.e. pumping and water transfers). Amount of consumed water (input minus output) per tonne of board is 9.4 m³ H₂O. At the board mill, largest share of water is consumed (evaporated) at the drying section of the board machine. Within the whole product chain, water consumption is dominated by the non-fibre raw materials production, i.e. chemicals and pigments used in board production (50% of the overall water consumption of the product system). The evaporative losses of water (1.7 m³/FU) in hydropower generation were also considered as consumed water and are included in the inventory results.

In the water inventory, the input waters were also classified by source: water is mainly withdrawn from surface waters (93%) and smaller amount from ground waters (7%). Of the withdrawn water, 44% are used for cooling. Of the consumed water within the whole system, 59% is evaporated from the cooling waters.

Water stress results obtained are shown in Figure 6. The results depend greatly on the input and output water qualities. Similar results have been obtained by Boulay et al. (2011a, 2014). The results of business as usual scenario (A) indicate that board mill is responsible for 30% of the stress impact (WAF) of the studied product system, while production of non-fibre raw materials causes 52% of the impact. Here, especially the impact of sodium chlorate production plays a significant role.

In scenario B, decreased quality of output water in the case of effluent treatment plant malfunction approximately doubles the stress impact of the studied product system. The board mill becomes the hotspot within the product chain: water degradation and the associated water consumption explain 60% of the stress impact. Increase of BOD level from 12 to 21 mg/L is enough to cause this effect and the effluent to be categorised as unusable (category 5). Because water quality category is changed from 2a to 5, more human user groups are impacted than in case A (Table 3).

The potential impact (WAF) of the board mill is increased when the operations are located in Central Europe and average quality European surface water is withdrawn (cases A and B). One might expect a lower water impact if lower quality water is consumed, due to a smaller difference between qualities of withdrawn and effluent waters. In this case, the median ambient poor quality water (class 3) in Europe is a more scarce resource than the good quality (2a) ambient water in Finland. This leads to a greater deprivation for competing water users in median European areas compared to Finland.

Case C assumes that board mill withdraws highest quality surface water and releases treated effluent as usual. The high impact of this scenario compared to others is due to the high difference in the availability of the withdrawn and released water. Owing to the high scarcity (α_i = 0.985) of category 1 (excellent) surface water compared to category 2a surface water (α_i = 0.00334), the footprint result would be increased nearly 300-foldedly in Finland. In both Finland and in the median European area, the board mill would responsible for 98–99% of the stress impact of the studied product system in scenario C.

By withdrawing water of lower quality, the board mill can reduce its WF. In case (D) where the mill withdraws water equal to the quality of water it releases, e.g. by recirculating treated effluent back to the process, the stress impact of the site is decreased by 70% compared to business as usual, causing a 20% reduction in WF of the whole product chain. Further, a negative impact (credit) is presented in case E, where a very poor quality (class 4) input water is assumed to be withdrawn, and released later as a higher quality water.

Loss of water functionality is created in cases A, B and C, leading to water deprivation for users who need water at a higher quality level than the released one (Table 3).

The potential changes needed in process water pre-purification techniques due to a lowered quality of input waters (cases D and E) were excluded from the study based on cut-off criteria (less than 1% of total water consumption volume). Another important note and a cause of potential uncertainty in results is that the region-adjusted data does not necessarily take into account the adaptation of industries into local circumstances, i.e. water efficiency in areas of high water scarcity can in some cases be assumed to be higher than in areas where water is very abundant.

The results of risk category assessment on the bases of likelihood and consequences of hazards are shown in Table 4.

The Finnish regional water scarcity data indicated slight seasonal water scarcity to occur during summers in some specific river areas (Ahopelto 2013). A peek into the Water Risk Atlas indicated that Finland suffers a serious draught especially in the middle of the country, being inconsistent with the new European water exploitation index, indicating that there is no annual water scarcity in Finland.

As the risk of shallow water levels and drought for the European conditions, sourced from the Water Risk Atlas, is on a general level and gives only an overall picture of the situation, it is essential that the risk should be studied more deeply by using regional data.

The freshwater withdrawal is most intensive in production processes of non-fibre raw materials and in mill production, which indicates that the risks connected to water level and drought should be taken into account at these production sites, especially during the summer months, and especially if competing users, such as hydropower generation exist. The drought, shallow surface or ground water levels can reduce the production temporarily. Risk of drought in general does not differ significantly in different parts of Europe. Instead, flood severity is higher in Central Europe than in Finland (Gassert et al. 2013).

The water contamination risks should be taken into account at the sites where non-fibre raw materials, energy and pulp as well as the paper mill itself are located. In these cases, the quality of the receiving water body and the availability of water are the most important factors to analyse. The availability of water is an important issue for two reasons; first there should be enough good quality water for mill processes to ensure the high quality of product; and secondly there should be enough receiving water body to ensure the sufficient dilution of waste water. The upstream storages can ensure the efficiency of diluting water in dry seasons. According to the Water Risk Atlas there is more upstream storage in Finland than in Europe in general.

If the manufacturing sites are located in areas with lowered water availability, also the image risks should be taken into account. The Water Risk Atlas approaches the issue with media coverage, determining the percentage of water-related articles among all media articles in the specific area (Gassert et al. 2013). High values indicate high public awareness of water issues. Security and safety risks as well as regulation and contract risks should be analysed especially if sites are located in foreign cultures.

In cases D and E, where the mill withdraws water with poorer quality, most of the water stress comes from the production of chemicals and fillers. Poor quality of board mill input water must however be taken into account as a risk source to poor machine operation or end product quality (Kiuru 2011), and needs to be treated to a higher quality level prior to use.

All the analysed main risks are evaluated to define the likelihood and consequences of the identified hazards (Table 4). The results show that the board mill in an average location in Central Europe cause a slightly higher water-based risk for manufacturing process than in Finland, mainly due to potential droughts and shallow water levels (cases A and B). In addition, the upstream storages which are more general in Finland than in other parts of Europe can ensure the efficiency of diluting water in dry seasons. Hence, the likelihood of polluting the receiving water body is minor in Finland. As mentioned before, the use of poor quality process water decreases WAF but causes the risk of poor product quality. In these cases, the risk is equal in all parts of Europe as a technical process risk.

The utilised method helps to understand the various water flows related to the studied product system and allows assessing hotspots within the product value chain. This study showed that magnitude of water stress impact assessment results depend greatly on the input and output water qualities. Changes of water quality at one part of the supply chain can have a major effect on total WAF result, i.e. results are sensitive to changes occurring at only one single unit of the supply chain, in this case the board mill. It is recommended to screen the most essential hotspots within the value chain including both water quality and quantity. Water-efficient process can still cause impact in water quality.

The presented risk analysis gives a hands-on way to compare different locations and their water-related risks when deciding far-reaching investments or choosing suppliers. Specific regional data and indicators are recommended to be used in risk assessment. The importance of use of site-specific data can be recognised in the results of this study. Risks related to measures aiming at decreasing the impacts were not included in this study but should be taken in focus in further studies.

The Boulay midpoint stress method together with the water balance data for various unit processes, available through Quantis Water database (Quantis 2013) was considered as a straightforward way to proceed with the WAF assessment. To reveal the overall water impacts, water quality indicators such as freshwater eutrophication, aquatic acidification or aquatic toxicity with water availability footprint results should be used. Stress and quality indicators in WF assessment complement each other as the stress method categorises water based on human user functionalities while the ecosystem quality indicator captures direct impacts from pollution. In some cases availability of water to human uses may not be affected, but quality of water may however be substantially decreased with a significant impact on the ecosystem.

WF and environmental risk assessment should be connected to derive complementary data on product water sustainability. The suggested framework provides companies a way to manage and foresee water use related impacts and risks. It can be used as a basis for a broader water disclosure, providing a deeper understanding of water risks for the companies themselves, the investors and other stakeholders. The commonly global value chains make water stress indicator relevant also at regions with abundant water resources, like Finland.