Determining the absolute sustainability of products with 1 case studies on laundry and food production 2 3

10 In this work, a new metric called ‘Service-weighted Product Level Absolute Sustainability’ is 11 proposed as a numerical indicator to determine if a product is sustainable. The service offered 12 by a product was found to be crucial to normalize its environmental impact and permit 13 comparisons between products. Service-weighted Product Level Absolute Sustainability is 14 demonstrated here with examples of water use for laundry and food production. The 15 maximum justifiable environmental impact of these products has been calculated based on 16 their performance, i.e., the quantity of clothes washed or nutritional content. Now the 17 environmental impact of products can be rationalized as either sustainable or unsustainable, 18 informing sustainable choices by manufacturers as well as consumers. 19 20


Introduction 25
The deterioration of the environment undermines efforts to sustain essential services and 26 habitable living conditions. Accordingly, environmental sustainability is now embedded into 27 many aspects of governance, business, and society. Tools for monitoring sustainability 28 include the Environmental Performance Index, 1 and the Sustainable Society Index. 2 National 29 or global scale multi-criteria indicators such as these may introduce emission targets to 30 normalize an impact category, but do not typically provide a well-defined absolute ecological 31 limit to those environmental impacts. Therefore, while it is possible to identify an 32 environmentally preferable practice, whether it is sustainable or not is unclear. 33 The proposal of planetary boundaries has introduced absolute limits on human activities, 34 including water use, land use, and pollution. 3,4 A planetary boundary defines the tipping point 35 of an Earth system process, beyond which the ecosystem becomes unstable with potentially 36 disastrous consequences. The best-known planetary boundary is the safe limit to atmospheric 37 CO2 concentrations with respect to climate change. Other examples relevant to this work are 38 provided in Table 1

51
Planetary boundary Global scale. 3,4 Safe operating space. 5 Agricultural allocation. 7 Freshwater use (km 3  The scale and ambition of the planetary boundary concept is suited to inform 53 international policies, 6 but they can also be divided into allocations to suggest a maximum 54 environmental impact for different activities. This 'downscaling' exercise has been 55 performed for agriculture by Springmann et al. 7 Note that the sustainable limit to fertilizer 56 use was actually increased compared to the full planetary boundaries (Table 1), suggesting a 57 larger environmental impact can be tolerated than previously thought. 58 Downscaling the planetary boundaries and combining with life cycle assessment (LCA) 59 is the basis of absolute environmental sustainability assessments. 8,9 For example, the annual 60 environmental impact of a municipal water company has been interpreted relative to a 61 calculated maximum permissible impact. 10 An allocation of each planetary boundary was 62 determined based on the population being supplied with water and the household expenditure 63 on this utility. The resulting 'share of safe operating space' reports if the allocated share of a 64 planetary boundary for a specific purpose has been exceeded. It was found in this example 4 that some impacts were sustainable (e.g. relating to stratospheric ozone depletion) but many 66 were not (e.g. climate change indicators). Algunaibet et al. investigated the environmental 67 impact of the USA power industry in a similar way but concentrated their efforts on 68 understanding three future scenarios. 11 Bjørn et al. identified the environmental impact of 69 laundry detergent manufacturing and use by introducing geographically resolved allocations 70 of the planetary boundaries. 12 The absolute sustainability of each process in the life cycle was 71 then calculated using an economic allocation, and by doing so revealed that producing the 72 raw materials from vegetable oils was responsible for the majority of the economic-weighted 73 land use and biogeochemical flow impacts (i.e. fertilizer use). The carbon emissions of the 74 New Zealand horticultural sector have also been evaluated with an absolute sustainability 75 assessment. 13 The allocation of the global carbon budget to this sector was based on its 76 historical share of emissions (globally) and then one of 4 methodologies was applied to 77 attribute it exclusively to New Zealand. Of which, only the economic allocation suggested the 78 foods (apples, kiwifruit, wine) were sustainably produced. 79 In environmental sustainability assessments, the societal need satisfied by services and 80 products is typically defined by their monetary value. The primary aim of this work is to 81 show that the environmental sustainability of products can be interpreted in a way that is 82 relatable to how we use them, thus the function and performance (i.e. service) of a product 83 can be represented as a variable in absolute sustainability assessments. Combining 84 environmental impacts with the societal benefit obtained from the function of a product 85 reveals how the choices made in the design of products and the implementation of services 86 defines their sustainability. Specifically, the ratio between the service provided by a product 87 and demand for that service, compared to the ratio between its environmental impact and the 88 maximum permissible impact, can be used to indicate if a product is sustainable. The 89 resulting metric is called Service-weighted Product Level Absolute Sustainability (Fig. 1)  impact. This generic example is for non-agricultural products. See Note S1-S7 for more information.

98
Service can be defined as the benefit received from the intended purpose of a product. 99 Increased performance or an extended product lifespan improves the service that is obtained 100 from that product. Annual demand corresponds to the collective receipt of a service by a 101 given population, and so it is defined by consumer behaviours. Various future market 102 scenarios can be analyzed with Service-weighted Product Level Absolute Sustainability to 103 predict necessary improvements in technology or determine a sustainable level of 104 consumption for a given population. 105 6 Metrics describing products have previously incorporated an efficiency scale to justify 106 resource use, 14 but do not directly measure sustainability. The European Commission's 107 Product Environmental Footprint (PEF) methodology will introduce a standardized LCA 108 approach designed to permit fair comparisons between products within the same category. 15 109 However, PEF is not an absolute sustainability assessment and comparisons between 110 dissimilar products with different functions are not valid. This is because a LCA reports 111 environmental impacts relative to a functional unit (e.g. the grams of CO2 emitted by a 112 vehicle per kilometre). Service-weighted Product Level Absolute Sustainability normalizes 113 product performance by demand for that service, and so eliminates specific functional units 114 for different products. This achieves valid comparisons between unrelated products. 115 As is true of 'share of safe operating space' calculations, a proportion of the planetary 116 boundaries (specifically the safe operating space) must be allocated to the demand category 117 relevant to the product in question. Appropriate methods are debated, 16,17 but the basis of 118 relative economic value is typically applied. In this work, a significant proportion of relevant 119 planetary boundaries has been reserved for agriculture, as was previously determined by 120 Springmann et al., 7 and only then is the remainder allocated to non-agricultural sectors 121 according to their economic value (Fig. 1). 122 The case studies in this work have been chosen because equivalent regional assessments 123 have been previously published, 5,12 and therefore the results can be compared. The absolute 124 sustainability of freshwater use for laundry and producing tomatoes are evaluated here on the 125 basis of a single wash cycle and 1 kilogram of tomatoes respectively. The scope is defined so 126 that the result is the same for the whole operational lifespan of the washing machine or if a 127 single tomato from that harvest is considered. 128 129 130 7

Results 131
Water use by washing machines 132 The first demonstration of Service-weighted Product Level Absolute Sustainability 133 describes the freshwater use of washing machines. This case study was chosen to permit a 134 comparison with the first 'share of safe operating space' assessment, 5  Accordingly, the resulting 'share of safe operating space' is 554%, considerably exceeding 141 the sustainable threshold of 100%. 142 We now compare the regional assessment to the Service-weighted Product Level 143 Absolute Sustainability of a single washing machine, accounting for its water efficiency (Fig.  144 2). The service provided by a washing machine can be considered as a single wash cycle 145 instead of the cumulative number of wash cycles over its lifespan because a washing machine 146 consumes water as a linear function of its use. The water use of washing machines was 147 sourced from manufacturer specifications. 18,19,20 It was assumed all water is bluewater 148 (surface water and groundwater) to match the planetary boundary definition. The washing 149 machine water use quoted in other assessments falls between that of the products used in this 150 work (33 L and 72 L per wash). 5,21 The amount of water required to manufacture a washing 151 machine has been excluded as it was previously shown to be minimal, 5 but for consistency 152 the GVA contribution to the allocation of the freshwater planetary boundary is for clothes 153 washing services only and excludes the GVA generated from manufacturing washing 154 machines (see Table S1 To calculate demand for wash cycles, it was assumed a 6 kg load household washing 163 machine is used 260.1 times a year, as obtained from a previous life cycle assessment. 22 The 164 annual demand for UK wash cycles was calculated by multiplying the number of households 165 by the clothes washing frequency stated above (see Table S3). This was in preference to 166 using an estimate of the number of operational household washing machines so that 167 launderette users contribute to the total demand for laundry. 168 Service-weighted Product Level Absolute Sustainability emphasizes the importance of 169 service, and commensurately the value of unpaid household services in the UK have been 170 9 valued and a GVA assigned for the year 2016. 23 Laundry accounts for 2.9% of this expanded 171 GVA measure (see Table S1). The quantities of water required by agriculture are much 172 higher than would be allocated according to GVA, and so not to impair food production an 173 allocation of freshwater use can ringfenced for agricultural purposes. 7 The contribution of 174 laundry to UK (expanded) GVA after excluding food production is 3.0% (of the non-175 agricultural economy), meaning 0.53 km 3 of freshwater is available as the sustainable limit to 176 satisfy annual UK laundry demand by this measure (Fig. 2A). 177 The Service-weighted Product Level Absolute Sustainability of laundry, adjusted to UK 178 demand according to population, 22 is calculated as 44%, rising to 96% for more water 179 intensive washing machines (Fig. 2B). After considering the increase in UK population since 180 2016, the latter washing machine represents the limit of a sustainable product with Service-181 weighted Product Level Absolute Sustainability recalculated as 100% (retaining the same 182 economic allocation, see Table S15). A washing machine that consumes more than 72 L of 183 freshwater per wash is therefore unsustainable with respect to water use in the UK market. 184 The discrepancy with the regional analysis by Ryberg et al. is mostly caused by the choice of 185 economic allocation (compared in Fig. S1). 5 The present analysis is more proportionate with 186 the overall evaluation of Steffen et al., 4 who calculate current freshwater use globally is about 187 two-thirds of the sustainable limit. 188 189

Contemporary food production 190
The second case study addresses food production. As alluded to, agriculture is a major 191 water user, both in scale and importance. The service provided by food is not straightforward 192 to define, and its different nutritional benefits must be taken into account. Energy in the form 193 of calories, protein, and portions of fruit and vegetables have been considered here as the 194 10 basis of the service provided by food. A worked example for water used to grow tomatoes is 195 given in Fig. 3. The planetary boundary reservations for food production, 7 were split into contributions 209 towards the provision of different macronutrients. To do so, the energy (kcal), protein 210 (grams) and equivalent portions of fruit and vegetables (one portion is 80 g) in 1 kg of farmed 211 foodstuffs was sourced from the USDA 'FoodData Central' database. 24 Food production data 212 (by mass) was sourced from FAOSTAT, 25 to establish the daily demand for food (inclusive 213 of waste) per capita (see Fig. 3A and Table S4). The nutritional content of every foodstuff 214 was then divided by the daily demand (per capita) for each respective macronutrient (see 215   Table S4 and Section 2.2) to calculate nutritional units (NU, per kg). The global gross 216 production value of a foodstuff, 25 was multiplied by its NU to assign a monetary value to the 217 provision of each macronutrient. The summation of all foodstuffs attributed 35% of each 218 planetary boundary reserved for agriculture to energy (calories). The provision of protein was 219 assigned 45% and fruit and vegetables 17% (Fig. 3B). The remaining 3% is the sum of the 220 production value generated from non-foods. A summary is given as Table S8 and provided in 221 full in the supplemental data file. This resulting weighting of planetary boundary agricultural 222 allocations is shown for freshwater use in Fig. 3C and for other planetary boundaries in Table  223 S6. 224 Land use and water use impacts were sourced from the work of Poore and Nemecek 225 because mean, median, and percentile data was made available and land use was also 226 reported inclusive of grazing pasture. 26  The environmental impact incurred during food production must also be distributed 232 proportionally according to the relative provision of energy, protein, and portions of fruit and 233 vegetables. Taking the example of water use to produce tomatoes, nutritional content (Fig.  234 3D) was converted into NU and weighted with the same economic allocation used for the 235 planetary boundaries (Fig. 3E). This was then used to assign a share of the environmental 236 impact to each macronutrient (Fig. 3F) Sustainability of 144% with respect to freshwater use (Fig. 3G). By this measure, the 245 maximum sustainable quantity of freshwater that can be used for the production of one 246 kilogram of tomatoes is 257 litres. Water use to produce tomatoes, potatoes, and pork are 247 tabulated in Table S5. 248 Water use in food production varies considerably, and when Service-weighted Product 249 Level Absolute Sustainability is applied to specific products (e.g. tomatoes produced in 250 different regions with different farming practices) it can differentiate between sustainable and 251 unsustainable sources of the same foodstuff. For instance, the median freshwater use to 252 produce tomatoes is sustainable. Figure 3H also shows the freshwater use Service-weighted 253 Product Level Absolute Sustainability of potatoes, pork, and tofu (from soybeans), including 254 the range between the 10th and 90th percentile. 26 A significant amount of tomatoes and pork 255 are produced unsustainably, but the majority of potato and tofu production requires 256 sustainable quantities of irrigation water. The sustainability of water use and land use for a 257 13 further 27 foods are analyzed in Fig. S3-4, revealing unsustainably high water use for most 258 meat products and rice production in particular. 259 A regional assessment evaluating the water use to produce tomatoes is available in the 260 literature and provides a means of comparison with the service and demand interpretation of 261 environmental sustainability developed in this work. 12 Bjørn et al. used temporally as well as 262 spatially resolved water demand and the value of tomato farming to the regional economy as 263 methods to assign a sustainable volume of water use to this industry. 12,21 In some regions, 264 they found freshwater use for producing tomatoes was more than 5000% of the indicated 265 sustainable maximum. 12  advances that enable a reduction to water use, land use, nitrogen and phosphorus fertilizer 280 application were previously determined by Springmann et al. 7 In addition, different food 281 production scenarios for the year 2050 were also considered. Firstly, it was assumed there 282 14 will be no change to food demand per capita, and so global demand increases proportionally 283 with population. A second future food production scenario was designed to reflect lower 284 consumption of animal products and an average nutritional intake equivalent to minimum 285 daily dietary requirements (i.e. an average of 2000 kcal, 50 g protein, and 5 portions of fruit 286 and vegetables per capita) but also factoring an additional 17.75% food waste factor across 287 all macronutrients (see Table S4). This food surplus was chosen to provide leeway in 288 providing sufficient nutrition and to match energy (kcal) availability to that suggested by 289 The economic allocation in the future food production scenarios was unchanged (from 302 that shown in Fig. 3B) when the daily demand per capita was maintained. The 2050 reduced 303 diet scenario uses the alternative daily nutritional demand in Table S4 to produce the NU, and 304 accordingly the division of the planetary boundary agricultural allocation between 305 macronutrients was adjusted (Table S7). The gross agricultural production value in 2050 was 306 estimated in line with the dietary changes in Table S4 and scaled proportionally with the 307 estimated population change to 2050 (see supplemental data file). To do so it was assumed 308 the relative monetary value of foodstuffs is the same in 2050. 309 For tomatoes and pork to be produced (on average) with sustainable amounts of water in 310 2050, both improved technology and diets are required (Fig. 4A). Land use (Fig. 4B, an 311 expanded chart is available as Fig. S6) and nitrogen fertilizer application (Fig. 4C) for 312 producing pork remains unsustainable regardless of what interventions are enacted. 313 Phosphorus recycling could make tomato and pork production sustainable (Fig. 4D), while 314 the quantities of fertilizer needed to produce potatoes will remain sustainable (on average) to 315 2050 without changing diets or needing technological advances in farming. Current day 316 fertilizer use has already transgressed planetary boundaries, 4

which is reflected by the high 317
Service-weighted Product Level Absolute Sustainability of most foods in this respect (see 318 Sustainability has been shown to provide an absolute measure of product sustainability that 325 had previously remained elusive. This calculation can be applied to any product that provides 326 a quantifiable service. Comparisons between products are permitted because of the 327 introduction of societal need (i.e. demand) to normalize environmental impacts, thus also 328 introducing a natural link between social and environmental sustainability. 329 The laundry case study indicated that contemporary washing machine water use can be 330 considered sustainable (in the UK market), but less efficient products are close to the 331 acceptable limit. The sustainable volume of water that may be used to provide a laundry 332 service was determined with an allocation of planetary boundaries that was generous toward 333 unpaid household services. Compared to a strictly economic allocation, this approach permits 334 a greater environmental impact within the defined sustainable limits. Conversely, processes 335 that are not consumer-facing will need to have lower impacts for Earth-systems to operate 336 within planetary boundaries. There is yet to be unanimous agreement on a fair allocation 337 system, and this is recognized as the greatest source of variance between assessments (further 338 analysis in Fig. S7), 5 but an emphasis on what people do, rather than how much they pay for 339 it, is commensurate with an equitable society. 340 The sustainable amount of water use, land use, and fertilizer use in agriculture was also 341 justified using the nutritional content of the food produced. It was found that the mean 342 average environmental impact of food production is unsustainable in several instances, 343 particularly for animal products. However, when considering the range of environmental 344 impacts incurred by different farming practices in different locations, there are many 345 examples of sustainable agricultural practices. Service-weighted Product Level Absolute 346 Sustainability was used to imply a sustainable limit to the environmental impacts associated 347 with several foodstuffs, and in doing so introduced targets for future practices. 348 In defining food demand categories (energy, protein, portions of fruit and vegetables) it 349 is assumed the consumption of fruit and vegetables per capita is diverse enough to deliver 350 sufficient micronutrients, and protein intake provides sufficient quantities of essential amino 351 acids. Service-weighted Product Level Absolute Sustainability could be calculated to 352 consider individual vitamins and minerals with a more complex economic allocation. An 353 additional allocation accounting for fiber was considered, but ultimately discounted because 354 whole plant-based foods contain large quantities of fiber, meaning a very high allocation of 355 planetary boundaries was attributed to the provision of fiber and very little to the other 356 macronutrients. Therefore, it has also been assumed that diets can be adopted to provide 357 sufficient fiber (30 g per capita per day) by virtue of consuming whole foods (grains, 358 vegetables, etc.). 359 Service-weighted Product Level Absolute Sustainability can identify the excessive use of 360 fertilizers relative to nutritional benefit, 29 and provide product-level objectives for 361 agriculture. 30 This exercise reiterates well understood consequences of farmed meat and the 362 need for sustainable diets, 31 but also identifies areas and practices that support sustainable 363 food production. Where environmental impacts are identified as unsustainably high, Service-364 weighted Product Level Absolute Sustainability calculations indicate the required reduction 365 to (for example) freshwater use, or perhaps whether different crops could be grown 366 sustainably in their place. The responsibility of consumers is also recognised in the Service-367 weighted Product Level Absolute Sustainability framework. To take one example, the land 368 use associated with producing potatoes has a Service-weighted Product Level Absolute 369 Sustainability of 104% based on the demand created by a future flexitarian diet (Fig. 4B). 370 18 Slightly reducing food waste from 17.75% to 13.25% of our basic nutritional requirement 371 would make land use associated with potato production sustainable in this scenario. 372 Some general limitations to the methodology have also been inferred through these case 373 studies. The emphasis on the service provided by finished products means Service-weighted 374 Product Level Absolute Sustainability does not evaluate the individual components in a 375 product or the stages of a manufacturing processes to identify sustainability hotspots. 376 However, the benefit of improved product performance can be evaluated, thus sacrificing the 377 producer-orientated assessment of other absolute sustainability methodologies and replacing 378 it with an end-user focus. Regardless, manufacturers can still use the Service-weighted 379 Product Level Absolute Sustainability concept to introduce overall performance and 380 sustainability targets for their products, and then determine which raw materials or 381 manufacturing processes need to be reviewed for an acceptable environmental impact. 382 There are knowledge gaps that prevent Service-weighted Product Level Absolute 383

Experimental procedures 396
Resource availability 397 All the source data used in this article is available from the cited references. The 398 reinterpretation of this data is documented in the article and the Supplemental Information. 399 Washing machine case study data was sourced from DEFRA, 22 In The Wash, 18 ONS, 23 400 Samsung, 19 Springmann et al., 7 and Whirlpool. 20 Food production case study data was 401 sourced from FAO, 25,33 Gerten et al., 28 Poore and Nemecek, 26  The allocation of the freshwater planetary boundary (PB water , km 3 /year) to UK laundry 406 demand was determined with Equation 1 according to the population affected (P) and GVA 407 (also see Fig. 2A). The absolute sustainability of laundry freshwater use was then able to be 408 calculated with Equation 2 for a given washing machine model (results in Fig. 2B). Data is 409 provided in Fig. S1 and Table S1-3.  Table S5). Equation 4 represents the gross production value 414 (V, $) attributable to energy provision of global tomato production. For foods without 415 nutritional data, the average for that class of food was used (categorized into grains, roots, 416 sugar crops, oil crops, pulses, nuts, fungi, animal products, vegetables, and fruit). The 417 economic value of agricultural crops not intended as food (e.g. cotton, tobacco) and herbs and 418 spices were not converted into NU. is required to obtain the sum of the global production value of food 422 attributable exclusively to energy provision in the form of calories (Fig. 3B). This value was 423 used to assign a proportion of a planetary boundary to the provision of food calories 424 according to Equation 6 ( Fig. 3C, also see Table S6 and S7).

427
The sub-division of environmental impact (Impacttomatoes) into individual contributions 428 for each macronutrient is given in Equation 7 for the example of water use for tomato 429 production (Fig. 3F).