Life Cycle Assessment of Dairy Products: A Case Study of a Dairy Factory in Brazil

Abstract

The production of dairy products generates several environmental impacts, and life cycle assessment (LCA) is a useful methodology to quantify and understand those impacts. In Brazil, some traditional dairy products have not yet been evaluated using the LCA methodology. Based on this gap, we conducted a cradle-to-gate LCA of six dairy products from a plant in Minas Gerais, Brazil. We also performed two sensitivity analyses. The first analysis was on how the environmental profiles of the products changed depending on how the multifunctional processes were allocated. The second analysis evaluated how these changes in environmental profiles occurred depending on the way that the impacts were allocated to products and by-products (whey and buttermilk) produced within the dairy factory. Among the dairy products studied, the impacts of mozzarella cheese and butter substantially surpassed those of other products; cheese spread and dulce de leche had a similar impact; and yoghurt and milk had the lowest values for the impact categories that were assessed. The inclusion of by-products in the analysis proved to be an effective way to reduce the environmental impacts attributed to the dairy products, especially for cheese and cheese spread, the impact values of which decreased by 56% and 46%, respectively. Additionally, the use of different strategies to deal with the multifunctional processes significantly affected the impact results of the dairy products. The subdivision of processes combined with causal allocation was the best alternative as opposed to the allocation by milk solids. These results could offer a better understanding of the environmental profiles of dairy products from Brazil, especially the traditional products, such as dulce de leche and cheese spread. Other contributions of this study include the proposal of alternatives that could improve the environmental profiles of products (such as the processing of by-products and the questioning of the use of allocation according to milk solids, which have been commonly used in other life cycle assessment studies) and the proposal of a better method for assessing the environmental impacts of dairy products.

1. Introduction

Dairy farming is an important branch of Brazilian agribusiness due to the value it adds to the domestic economy [1]. In 2020, milk production generated a gross value of around BRL 40.4 billion (or USD 7.78 billion considering the most recent value for the last month of that year [2]), which represented approximately 0.543% of the GDP [3,4]. The importance of this sector for the country is also reflected in its international production performance. In 2020, Brazil stood out as the sixth largest milk producer in the world, with 25.5 billion liters of milk being generated [5,6]. In addition to the economic benefits of this industry, milk and its derivatives contribute essential ingredients to the human diet and provide the population with a high level of nutrients, energy, and health from childhood onward [7]. Thus, dairy products play a key role in the country’s food security [8,9].

However, dairy production generates a series of environmental liabilities and exacerbates global warming, soil acidification, the eutrophication of water resources, and the demand for energy from fossil sources [10]. According to Bacenetti et al. (2016), the potential for increased global warming, acidification, eutrophication, and energy demand are some of the main environmental impacts of milk production due to the production of feed for the livestock [11]. Santos et al. (2017) stated that the most impactful processes of agro-industrial production are energy consumption and the standardization of milk, with the former contributing the most to ozone depletion and the latter significantly influencing climate change and demand for energy from fossil sources [12]. To deal with these issues, some studies have pointed to the importance of developing a more sustainable dairy production chain [13,14]. For this, the development of the sector is necessary and must include production methods that favor social progress, poverty reduction, and animal health and generate fewer negative impacts on the environment [8].

One of the industrial ecology methodologies that are suitable for the holistic analysis of environmental impacts throughout the various stages of a production process is life cycle assessment (LCA). This tool considers the environmental impacts of the life cycle stages of the product, which can include the extraction and procurement of raw materials, energy production, product production, end use, treatment, and disposal [15,16]. The systematic observation that LCA offers makes it possible to assess the environmental loads of the internal processes during each stage of the life cycle of dairy products. Using LCA, it is possible to identify possible improvements in terms of environmental performance thanks to a better understanding of the production chains [17,18].

The development of the LCA technique over the last few decades has been closely linked to the agro-industrial and food sectors. In addition, with the growing concern for the environmental sustainability of food production and the complexity of mitigating the environmental impacts, there has been an increase in applied research and scientific production around this topic, which has also been evident in Brazil [19,20]. The use of LCA for activities associated with agribusiness has been presented as the fourth largest application of this technique by the Brazilian Institute of Information in Science and Technology, representing 9% of all studies within this area [21].

Applying the LCA technique to the production chain of milk and its derivatives has now become common within the scientific community. For example, the environmental impacts of popular dairy products, such as milk, cheese, and butter, have become widely known. Additionally, the main impacts associated with each production stage and possible ways to reduce those impacts have also been discussed [10,22]. However, despite the similarities among the qualitative results, which have often been repeated on different continents, the quantitative study of environmental performance can differ spatially as production chain scenarios vary regionally [15]. Thus, the life cycle assessment of dairy products from different countries is justified as it could help to understand better the regional variability of products and their environmental profiles.

In Brazil, LCA studies have been conducted on the production processes of various dairy products, such as cheese and butter [12,23]. However, one area that has not yet been investigated is the characterization of the environmental profiles of traditional Brazilian products, such as dulce de leche and cheese spread. To the best of our knowledge, these products have not yet been evaluated using LCA, which is a point that requires clarification if the environmental impacts of dairy products are to be better understood. Moreover, compared to the rest of the world, Brazil has not produced a significant number of life cycle assessments that consider several dairy products from the same plant. Finally, within the context of the evaluation of dairy products using LCA, there is also a need to discuss strategies to manage production flows given the challenges of the multifunctional processes and the impacts of the valorization of by-products as co-products.

Thus, a life cycle assessment was carried out in this study to present a comparison between dairy products, considering six products popular with Brazilians. In addition, sensitivity analyses were performed to understand the effect that the transformation of by-products into co-products represents on reducing the environmental profile and whether the method of dealing with multifunctional flows affects the environmental profile of the products. The intention was to help recognize the level of environmental impacts relating to the production of various milk derivatives and to propose measures to reduce emissions and create an increasingly cleaner and more sustainable production.

2. Materials and Methods

The methodology developed was based on the recommendations of NBR ISO 14.040/2009 and 14.044/2009 standards and following the guidelines of Bulletin 479/2015 of the International Dairy Federation (IDF) [17,24,25]. Thus, the study was subdivided into (i) mapping the production process, (ii) defining the goal and scope of the life cycle assessment, (iii) preparing the life cycle inventory of products, and (iv) conducting the environmental impact assessment and (v) carrying out the sensitivity analysis.

Data was collected through interviews with the production and quality coordinators during a visit to the manufacturing facility and subsequent communication via questionnaires. The data correspond to an average of the dairy’s monthly production for the first half of the year 2021. Once collected, they were organized in electronic spreadsheets that were reviewed by a factory employee and served as the basis for carrying out the calculations and creating the inventory. The choice of data requested in the questionnaire was also obtained from the list of technical data suggested by IDF Bulletin 479/2015 and from literature consultation on the use of the methodology with dairy products [25].

2.1. Mapping the Production Process

2.1.1. Location of the Dairy Facility

The dairy plant is in the city of Viçosa, in the state of Minas Gerais, the state that concentrates the largest domestic milk production (Figure 1). Currently, the factory’s line of dairy products consists mainly of whole milk, skimmed milk, butter, mozzarella cheese, yoghurt, cheese spread, and dulce de leche. Cheese spread and dulce de leche are both traditional products of Brazilian cuisine. The first is a creamy cheese produced from skimmed milk and cream, while the second is a dessert made by boiling milk and sugar.

2.1.2. Characterization of the Dairy Products

Characterizing the dairy products is an important step due to the nutritional variability of their composition and the relevance of the nutritional content for comparison with similar foods. For this reason, a list of the nutritional elements of the dairy products under study is shown (

Table 1. Monthly production and nutritional characteristics of the evaluated dairy products.

Product		%
	Monthly Production, in kg	Milk Solids	Protein	Fat
Dulce de leche	158,400	81.5	10.0	6.5
Yoghurt	62,481	6.9–27.9	1.8–4.7	0.0–3.6
Milk	123,823	8.2–11.3	3.4	0.0–3.1
Butter	2188	84.0	0.0	84.0
Mozzarella cheese	8000	46.3	20.0	23.0
Cheese spread	5200	33.3	10.0	23.3

). Some dairy products display variations in certain nutritional characteristics due to the production of more than one type of the same product, such as yoghurt, with different flavors. It was also important to obtain the milk solids content of each dairy product due to its relevance for impact allocations used in the sensitivity analysis [25,26]. This indicator represents the number of solids in the product coming exclusively from milk. To obtain the product milk solids content, it was calculated the ratio of the number of carbohydrates, proteins, and fats present in the product by a standard serving size of the same product (e.g., 10 g of butter). These informations are present in the nutritional table of each dairy.

2.2. Definition of the Goal and Scope of the Life Cycle Assessment

The life cycle assessment carried out was of the attributional type. Its goal was to evaluate and compare the environmental impacts related to the production of the six dairy products of the dairy plant. To achieve this goal, the environmental performance of each product was compared, and the possible critical points within the production of each one were analyzed. The target audience for this information was the factory itself, to enrich its understanding of the impact of emissions and evaluate solutions that might improve the environmental performance of each product.

In the scope of the evaluation, the six dairy products were considered in relation to the line of products produced, and the functional unit used was 1 kg of end product. A cradle-to-gate system boundary was adopted in which the impact of the production of inputs, their transportation to the factory, and the activities from product production until dispatch were considered [15]. Given this, the system was subdivided into three systems: the production and transportation of inputs to the industry, the factory itself, and the treatment of effluents from production (Figure 2).

The first subsystem, or subsystem I, consists of the production and transportation logistics of the raw materials necessary to produce each dairy product and is further subdivided into the following categories: raw milk, ingredients (sugar), cleaning product, refrigerant gas, energy, and packaging. Subsystem II comprises the dairy plant, which receives the raw material flows from subsystem I and uses them in the processes of milk pasteurization and dairy product production. The first process at the factory consists of receiving the raw milk, pasteurizing it, and processing it into whole milk, skimmed milk, and cream. Once the first process is finished, its output flow enters the dairy production process as needed. Finally, subsystem III receives the effluents from all seven processes of subsystem II.

Since the factory produces different products with common inputs and outputs, it was important during data collection to determine these flows in a way that best represented this situation. Faced with multifunctional processes, the literature suggests a hierarchy of solutions that prioritizes a breakdown of the processes by means of subdivision. For cases where this is not possible, there is an opportunity for allocation [15,24]. Considering this hierarchization, to obtain information about the use of inputs and production of by-products, the subdivision of multifunctional processes of the system was prioritized for the majority. For those cases in which it was not possible, a causal allocation was used. This was made possible by recording the approximate quantity of inputs and outputs referring to the seven processes involving dairy production (subsystem II), i.e., from the milk standardization process to the production processes of the six other products.

The milk standardization process is a step common to all dairy products since it is responsible for processing raw milk received from farms and producing cream and milk (whole and skimmed). This is achieved through pasteurization, standardization, and homogenization performed on raw milk to yield a pathogen-free standardized product in terms of nutrients and better quality. Since the standardization process offers multiple outputs of inputs for production, it was decided to use the allocation of the impacts of each one of them to the product to which it was destined. The allocation of impacts suggested by the dairy literature is based on the milk solids present in the product [25]. With this, the impacts of the outputs of the standardization process were subdivided according to the content of milk solids present in the cream, whole, and skimmed milk, giving distinct values of impact for these products when entering new processes.

In addition to standardization, other processes produce multiple outputs, such as the production of cheese, cheese spread, and butter production, which produce by-products such as whey and buttermilk. For these processes, it was decided to attribute the impact of the flows only to the main product (e.g., butter) since these by-products are not wanted and, in the current scenario, are donated as pig feed to farms in the region.

The limitations of this LCA study consisted in disregarding the life cycle inventories of some production inputs (

Table 2. Input production processes that are not considered in the construction of the model and the products affected by their absence.

Process Production	Affected Products
Infrastructure	All
Equipment	All
Ingredients
Fruit pulp	Yoghurt
Sodium chloride	Cheese, butter, and cheese spread
Potassium chloride	Cheese
Lactic acid yeast	Yoghurt and cheese
Coagulant	Cheese
Inorganic colorant	Yoghurt
Packaging
Aluminum seal	Yoghurt, milk, butter, and cheese spread
Solid waste	All

), considering foreign inventories, and the use of similar processes. The limitation of disregarding certain inputs was due to their having little influence on the result of the impact of the product. On the other hand, the use of global inventories was due to the scarcity of process inventories for the Brazilian scenario, making it impossible for some studies to be carried out only with national information. Despite this, it was possible to use inventories of some inputs appropriate to the regional reality of the study. Data from the Ecoinvent 3 database of the SimaPro program version 9.1.1 developed by Pre-Sustainability Amersfoort City, Netherland, were used to obtain the inventories. In addition to the limitations cited, some database inventories were added to the model for flows that had a similar material or function. This occurred, for example, with some packaging materials and for the transportation of inputs to the factory.

The CML-IA baseline V3.06/World 2000 methodology was used to assess the impacts of production systems, adopting the following impact categories: climate change (kgCO₂eq), ozone layer depletion (kgCFC11eq), acidification (kgSO₂eq), eutrophication (kgPO₄eq), photochemical oxidant formation potential (kgC₂H₄eq), abiotic depletion—fossil sources (MJ). This impact assessment methodology and its impact categories are common for LCA studies applied to the dairy sector [27] and thus are most convenient for comparing studies.

2.3. Preparation of the Life Cycle Inventory of Product Systems

The data collected on the quantity of each flow in the plant were organized in electronic spreadsheets by assigning the flows for each production process; for those that served more than one product, a percentage corresponding to each product was assigned. As highlighted above, the logic for the distribution of these percentages in the face of multifunctional processes was carried out using two strategies: process subdivision and causal allocation. Through the subdivision of processes, approximate values of certain multifunctional flows that make up each product were obtained. This strategy was reserved for the following categories: distribution of milk and cream, production of by-products, amount of sugar, and packaging materials. The flows related to thermal energy, electricity, ammonia, water, and cleaning products followed a strategy of causal allocation, in which the factory team suggested percentages of each of these flows based on the relevance that this flow represented for each product. The resulting relationship is shown in

Table 3. Percentage distribution of flows for each process in the dairy industry.

Category	Flow	Milk Standardization	Milk Production	Cheese Production	Cheese Spread Production	Butter Production	Yoghurt Production	Dulce de Leche Production
Wastewater	Wastewater	25.00	10.00	10.00	10.00	10.00	10.00	25.00
Packaging	Tin	0.00	0.00	0.00	0.00	0.00	0.00	100.00
	Cardboard	0.00	0.00	0.00	3.76	3.08	33.00	60.16
	LDPE	0.00	19.30	2.77	0.00	0.00	8.04	69.90
	PET	0.00	11.81	0.00	0.00	0.00	88.19	0.00
	PP	0.00	0.00	0.00	53.39	46.61	0.00	0.00
Energy	Electricity	0.00	10.00	5.00	3.00	3.00	15.00	64.00
	Thermal	6.00	0.00	6.00	6.00	6.00	6.00	70.00
Ingredients	Sugar	0.00	0.00	0.00	0.00	0.00	10.00	90.00
	Cream	0.00	0.00	0.00	40.00	60.00	0.00	0.00
	Raw milk	100.00	0.00	0.00	0.00	0.00	0.00	0.00
	Skimmed milk	0.00	50.60	0.00	49.40	0.00	0.00	0.00
	Whole milk	0.00	19.05	12.70	2.48	0.00	9.90	55.87
Cleaning	Nitric acid	25.00	10.00	10.00	10.00	10.00	10.00	25.00
	Tap water	25.00	10.00	10.00	10.00	10.00	10.00	25.00
	NaOH	25.00	10.00	10.00	10.00	10.00	10.00	25.00
Refrigerants	Ammonia	16.00	16.00	16.00	16.00	16.00	16.00	4.00
Byproducts	Buttermilk	0.00	0.00	0.00	0.00	100.00	0.00	0.00
	Whey	0.00	0.00	75.00	25.00	0.00	0.00	0.00

.

Once a strategy was outlined to determine the direction of the flows for each process, it was also considered convenient to analyze the values of each of the flows qualitatively. This was carried out because certain inputs were measured directly, as in the case of the flows of the packaging category, in which each product’s package and packaging were weighed. In contrast, other inputs were not measured directly or were calculated via mass balance in the system. For this classification, we used the methodology suggested by Santos et al. (2017) and proposed four categories: measured, verified, estimated, and calculated [12]. With this classification, it was possible to add qualitative indicators to the inventory that convey a degree of confidence in the data of each flow (

Table 4. Inventory for dairy production considering 1 kg of end product.

Flow	Unit	Milk Standardization	Milk Production	Cheese Production	Butter Production	Cheese Spread Production	Yoghurt Production	Dulce de Leche Production	Qualitative Analysis
Input
Packaging
Cardboard	kg × kg−1	-	-	-	0.05288	0.02717	0.01985	0.01427	Measured
LDPE	kg × kg−1	-	0.00485	0.01075	-	-	0.00400	0.01372	Measured
PET	kg × kg−1	-	0.00136	-	-	-	0.02008	-	Measured
PP	kg × kg−1	-	-	-	0.07164	0.03452	-	-	Measured
Tin	kg × kg−1	-	-	-	-	-	-	0.07845	Measured
Energy
Electricity	kWh × kg−1	-	0.07208	0.55781	1.22374	0.51490	0.21426	0.36061	Verified
Thermal energy	MJ × kg−1	0.34809	-	28.29000	43.52308	103.43882	3.62221	16.66919	Calculated *
Firewood	m3 × kg−1	0.00004	-	0.00300	0.00462	0.01097	0.00038	0.00177	Verified
Ingredients
Sugar	kg × kg−1	-	-	-	-	-	0.09603	0.34091	Verified
Cream	kg × kg−1	-	-	-	1.64537	0.46154	-	-	Verified
Raw milk	kg × kg−1	1.00000	-	-	-	-	-	-	Calculated
Skimmed milk	kg × kg−1	-	0.13314	-	-	3.09559	-	-	Verified
Whole milk	kg × kg−1	-	0.86686	10.31862	-	-	1.03052	2.29303	Verified
Cleaning
Tap water	kg × kg−1	0.00065	0.00136	0.02100	0.07678	0.03231	0.00269	0.00265	Verified
Nitric acid	kg × kg−1	0.00035	0.00073	0.01125	0.04113	0.01731	0.00144	0.00142	Verified
Sodium hydroxide	kg × kg−1	0.00023	0.00048	0.00750	0.02742	0.01154	0.00096	0.00095	Verified
Refrigerant
Ammonia	kg × kg−1	0.000004	0.00002	0.00036	0.0013299	0.00056	0.00005	0.000005	Verified
Output
Wastewater	m3 × kg−1	0.00065	0.00136	0.02100	0.07678	0.03231	0.00269	0.00265	Verified
Buttermilk	kg × kg−1	-	-	-	0.74224	-	-	-	Verified
Whey	kg × kg−1	-	-	9.21780	-	4.72708	-	-	Verified

* The thermal energy values used were calculated from the amount of firewood.

).

From the information collected at the dairy factory, it was found that of the total raw milk processed in the standardization stage, 94.12% was used to produce whole milk (at 3.00% fat and 3.30% protein), 4.97% was used to produce skimmed milk (at 0.50% fat and 3.30% protein), and 0.91% was used to produce cream (at 50.0% fat). With this relationship obtained from the monthly total of these inputs, the total amount of raw milk was corrected in terms of fat and protein (fat protein corrected milk—FPCM). This correction was made to represent in the model the impact of the nutritional quality of milk. Thus, the FPCM formulation considered the average fat content of raw milk received by the dairy (3.4%), average protein content (3.3%), and the average specific mass of raw milk received of 1032 kg m⁻³. It was then possible to calculate the impact of raw milk on the factory.

The amount of heat needed for each production was calculated considering the average amount of firewood used by the factory. Thus, the value for monthly thermal energy in MJ was obtained from the product of the total volume of firewood by the lower calorific value offered by the Ecoinvent process used, of 9430 MJ m⁻³. To calculate and convert units of other flows, we used the specific mass values presented by the dairy for the following inputs: pasteurized milk (1031.7 kg m⁻³), skimmed milk (1034.4 kg m⁻³), butter (911.6 kg m⁻³), and yoghurt (1054.0 kg m⁻³). The specific mass of whey used in the conversion from liters to kilograms was 1024.2 kg m⁻³ [28], and the specific mass of the ammonia used for refrigeration was 682.0 kg m⁻³ [29]. The plant gave the value for buttermilk as 1015.0 kg m⁻³. Finally, the processes that best represented the operating reality of the factory were considered when preparing the inventory using the Ecoinvent 3.7.1 database (

Table 5. Flows that were considered in the model and their respective process in the database.

Category	Flow	Process in the Database (Ecoinvent 3.7.1)
Input
Packaging	Cardboard	market for carton board box production, with gravure printing|carton board box production, with gravure printing|APOS, U–GLO
	LDPE	market for packaging film, low density polyethylene|packaging film, low density polyethylene|APOS, U
	PET	market for polyethylene terephthalate, granulate, bottle grade|polyethylene terephthalate, granulate, bottle grade|APOS, U–GLO
	PP	market for polypropylene, granulate|polypropylene, granulate|APOS, U–GLO
	Tin	Tin {GLO}|market for|APOS, U
Energy	Electricity	market for electricity, low voltage|electricity, low voltage|APOS, U–BR-South-eastern grid
	Thermal energy	heat production, mixed logs, at furnace 100 kW|heat, central or small-scale, other than natural gas|APOS, U–RoW
Refrigeration	Ammonia	market for ammonia, anhydrous, liquid|ammonia, anhydrous, liquid|APOS, U–RoW
Ingredients	Sugar	market for sugar, from sugarcane|sugar, from sugarcane|APOS, U–GLO
	Cream	This process was created by the authors
	Raw milk	market for cow milk|cow milk|APOS, U–GLO
	Skimmed milk	This process was created by the authors
	Whole milk	This process was created by the authors
Cleaning	Tap water	tap water production, direct filtration treatment|tap water|APOS, U–BR
	Nitric acid	market for nitric acid, without water, in 50% solution state|nitric acid, without water, in 50% solution state|APOS, U–RoW
	Sodium hydroxide	market for sodium hydroxide, without water, in 50% solution state|sodium hydroxide, without water, in 50% solution state|APOS, U–GLO
Output
Wastewater	Wastewater	treatment of wastewater from potato starch production, capacity 1.1 × 1010 L/year|wastewater from potato starch production|APOS, U–GLO
Byproducts	Buttermilk	This process was created by the author
	Whey	This process was created by the authors

).

2.4. Performing the Life Cycle Impact Assessment

This step consisted of the life cycle impact assessment of the different product systems modeled in SimaPro software, interpreting the respective results, and comparing them with the literature. Its importance lies in presenting not only the results of the environmental performances of the different product systems but also in providing results for the subsequent sensitivity analysis and informing about possible errors in the results.

From these results, it was possible to present their performance graphically and in tables for the different impact categories proposed and compare the products’ performance for each category under investigation. The processed data were used to produce the life cycle impact assessment model for this study.

2.5. Sensitivity Analysis

Sensitivity analysis is an important and necessary contribution to life cycle assessment in the agroindustry since this stage presents various critical points producing these products [10,12]. Through this analysis, it is possible to relate the change in the environmental performance of dairy products according to the change in the elements of the life cycle impact assessment, such as the presence of by-products and how the way of allocation or the way of dealing with the flows of the system reflect impacts on the products.

2.5.1. Influence of By-Products on the Assessment of Product Impacts

The production of by-products is one element of the case study since whey is produced in the manufacture of cheese and cheese spread, and buttermilk in the manufacture of butter. Since these by-products have no commercial value for the company and are also not wanted, it was decided for the model that the impact of their production would be linked to the impact of the main product of that process. This results in allocating all impacts from the production of these potential by-products to the main products, which could be reversed if these by-products were processed and considered as co-products in the dairy. Given this context, a sensitivity analysis was conducted to verify the degree of impact reduction generated by the proper allocation of impacts for whey and buttermilk, considering a scenario of using them as co-products by the factory.

The sensitivity analysis considered the volume of whey and buttermilk produced in each system in which these by-products were produced. In addition, we considered an impact allocation per milk solids present in each of them (

Table 6. The ratio of milk solids content and total produced per product and by-product for the milk solid content allocation calculation.

Process	Product	Milk Solids	Production (kg)	Allocation
Butter Production	Butter	84%	2400	0.89839
	Buttermilk	13%	1600	0.10161
Cheese Production	Cheese	46%	8000	0.44200
	Whey	6%	73,742	0.55800
Cheese spread Production	Cheese spread	33%	5200	0.53778
	Whey	6%	24,582	0.46222

). Information regarding the quantity of milk solids present in the by-products was obtained in the literature [26,28], except for the specific mass of the buttermilk offered by the dairy plant as being 1.015 kg m⁻³.

2.5.2. Influence of the Option of Subdivision of Multifunctional Processes versus Just Allocation

Subdivision of multifunctional processes is an option suggested for life cycle assessment as the most suitable compared to allocation [15]. This is because it highlights the actual flows that occur within systems with multiple product outputs, indicating the approximate share of input entering each product. Other strategies used are system expansion and causal allocation, which are alternatives when process subdivision is not possible. Despite these strategies, many studies perform LCA using allocation related to physical or economic characteristics because they do not have enough information regarding how much input goes into each production, or they only have general information about how much input goes into how much product is produced [10,26].

Given this issue, a sensitivity analysis was performed to verify the degree of variation in the results obtained with the model, comparing the results made with subdivision of multifunctional processes and causal allocation against the results of the allocation made exclusively by milk solids content, which is the one suggested in allocation situations [25] With this analysis, it is possible to identify if a more generic investigation of the inputs and outputs of the dairy does not significantly alter the results compared to a more detailed and closer to the reality of the flows.

In performing this analysis, we considered modeling using the subdivision of multifunctional processes and causal allocation as the default scenario, with the alternative scenario considering only allocation by solids content. The standard scenario considered information offered by the production team on how much input was offered for each product. In the alternative scenario, on the other hand, the solids content allocation formula was considered, and the input data were the product quantity and its respective milk solids content, obtained through its nutritional table (Table 1). It is important to emphasize that the processes that are common to all the products are allocated considering all the products during the calculation of the allocation percentages, for example, electricity use. On the other hand, more specific processes are considered in the allocation calculation only the products that have the input in the recipe, such as adding sugar, where only the yoghurt and the dulce de leche are affected. This last measure was taken to avoid allocating ingredient impacts to products that do not contain them in their recipe.

3. Results

3.1. Comparing the Impacts of Dairy products

After modeling, we observed the environmental performance of the products for different impact categories being analyzed (

Table 7. Results of the life cycle environmental impact assessment for the different dairy products.

Product (1 kg of Each)	Climate Change	Acidification Potential	Eutrophication Potential	Abiotic Depletion	Ozone Layer Depletion	Photochemical Oxidation
	(kgCO2eq)	(kgSO2eq)	(kgPO4eq)	(MJ)	(kgCFC-11eq)	(kgC2H4eq)
Dulce de leche	4.830	0.036	0.023	25.790	2.25 × 10−7	3.39 × 10−3
Yoghurt	1.820	0.012	0.007	8.568	7.50 × 10−8	9.81 × 10−4
Milk	1.545	0.010	0.006	6.055	5.42 × 10−8	3.19 × 10−4
Butter	13.888	0.093	0.051	60.061	5.32 × 10−7	8.61 × 10−3
Cheese	16.306	0.109	0.061	60.947	5.77 × 10−7	4.88 × 10−3
Cheese spread	5.959	0.040	0.022	25.997	2.28 × 10−7	3.65 × 10−3

) and the relative performance among them (Figure 3). For all the categories, milk had the lowest impact among the six dairy products evaluated. In contrast, mozzarella cheese stood out as the product with the greatest impact, standing out in five of six environmental impact categories. Thus, the descending order of impact by product was cheese, butter, cheese spread, dulce de leche, yoghurt, and milk.

The results of this study show that the environmental impact values of the dairy products were similar to or less than the ranges presented by the Djekic et al. (2014) LCA study of several dairy products [10]. The only exception is cheese, which showed superior results. The impacts of milk and yoghurt were within the range of values found for climate change. However, for acidification potential and eutrophication, milk was below the values found, and yoghurt was within the range of values found by the authors. On the other hand, butter presented considerably lower values than those of the authors for the three categories analyzed, which were 20.69~21.30 kgCO₂eq (climate change), 0.2636~0.2658 kgSO₂eq (acidification potential), and 0.145~0.1248 kgPO₄eq (eutrophication potential) [10]. Furthermore, all products showed lower ozone depletion potential than the products presented by Djekic et al. (2014), which were in order of 10⁻⁴ to 10⁻⁵ kgRC11eq, while those evaluated in this study were in the range of 10⁻⁷ and 10⁻⁸ kgCFC11eq [10].

Table 8. Comparison of the results with other similar LCA studies published in the literature.

Product	Reference	Climate Change (kgCO2eq)	Acidification Potential (kgSO2eq)	Eutrophication Potential (kgPO4eq)	Country
Yoghurt	Present study	1.820	0.012	0.007	Brazil
	[30] *	1.776	0.029	0.010	Portugal
	[31] *	4.210	0.070	0.242	Turkey
Milk	Present study	1.545	0.010	0.006	Brazil
	[32] **	1.500	0.010	0.007	Italy
	[7]	1.740	0.031	0.009	Portugal
	[33]	0.946	0.008	-	Iran
Butter	Present study	13.888	0.093	0.051	Brazil
	[34]	9.600	0.076	0.060	United Kingdom
	[34]	9.000	0.092	0.044	Germany
	[34]	7.200	0.050	0.044	France
Cheese	Present study	16.306	0.109	0.061	Brazil
	[35]	7.490	0.180	0.065	Portugal
	[36]	9.530	0.095	0.055	Spain

* cradle-to-grave; ** cradle-to-distribution center.

compares the present study’s results with those of other studies relevant to the dairy products’ life cycle impact assessment segment, which also used the CML method.

3.1.1. Mozzarella Cheese

The values obtained for cheese were higher than for the other dairy products, except for butter, which showed similar values. The impact category with a greater effect was climate change, with ozone layer depletion coming as the lowest and more similar to the other products.

Cheese is one of the dairy products with the largest number of LCA studies, alongside milk and yoghurt [27]. Considering studies with the same impact assessment methodology (CML), when comparing the cheese of the present work to those carried out with Galician and mature Portuguese cheeses, the values obtained here were higher than those of Spain and Portugal for the climate change category and closer for the other categories [35,36]. It is believed that the reason for the higher impact of the Brazilian product compared to the Portuguese product is due to the larger amount of milk present in the Brazilian product and the fact that the Portuguese cheese allocates part of its impact to whey. Another reason for the greater prominence of the product of this study is due to the high impact of milk production present in the database since the study of Galician cheese presents smaller usage of milk than that of our study.

Considering the impact of climate change exclusively, the cheese of the present study presented values higher than 7.203 kgCO₂eq, obtained in Ireland [37], and 5.300 kgCO₂eq, observed in Canada [38]. On the other hand, the cheese of the present study presented a value close to that obtained in another Brazilian study, in which the result of 14.447 kgCO₂eq was reported [12]. It is believed that the main reason justifying this difference is the allocation chosen in some of these works, in which part of the impact of the cheese was transferred to a co-product, such as whey protein isolate production.

3.1.2. Butter

Butter showed relative values from 84% to 99% to the impact values for cheese. Butter also showed the highest impact in the photochemical oxidation category. Despite being high, its impact values were in accordance with those presented in the literature. The characterization of butter varies greatly by author. This is due to several factors, such as the type of allocation, the physical and chemical characteristics of the products, the impact assessment methodology, regional characteristics, and the system boundaries of the model.

For climate change, the value of butter was above the values presented in another study conducted in Brazil, which ranged from 1.258 to 5.816 kgCO₂eq [23], but close to the average presented in Ireland, with 9.680 kgCO₂eq [37], and European and North American countries, which ranged from 8.080 to 16.930 kgCO₂eq [39]. Considering papers using the CML impact assessment methodology, the values obtained for climate change and acidification potential were slightly above those presented for the UK, Germany, and France, ranging from 7200 to 9600 kgCO₂eq and 0.050 to 0.090 kgSO₂eq, respectively. In contrast, for the eutrophication potential, the butter value was between the presented values of 0.040 to 0.060 kgPO₄eq [34].

3.1.3. Yoghurt and Milk

Yoghurt and milk have lower impacts compared to the other products. Both showed similar values as in other studies that evaluated them [10,40]. For climate change, our study indicated similar values to that of Djekic et al. (2019), of 1.511 kgCO₂eq for milk and 1.672 kgCO₂eq for yoghurt [40]. When we compared the values for other categories in the present study to the Djekic study, we also found that the values were similar [40].

Yoghurt showed a pattern of impacts similar to milk, except for ozone depletion. Its values were similar to those evaluated by other studies that used the same impact assessment methodology as used in the present study. When analyzing an LCA study of yoghurt in Portugal, a lower value than the one found in this study was observed for the climate change indicator. In contrast, the values for Portuguese yoghurt were higher for acidification potential and eutrophication [30]. When compared to a study of yoghurt production in Turkey, values in the present study were near or lower for all categories in common, with the biggest difference being the climate change category [31].

Comparing the impacts of milk production in the factory with a study for UHT milk that used the CML methodology [7], values obtained in that study were close to but greater than those of the present study for all categories in common. For example, the impacts of climate change, acidification potential, and eutrophication were 1.740 kgCO₂eq, 0.03128 kgSO₂eq, and 0.00943 kgPO₄eq, respectively. Naturally, this comparison should suggest higher values for the UHT milk since this product presents higher processing than the pasteurized milk evaluated in this work. Considering pasteurized milk, another study, which used the CML methodology, offers as impact values of 0.946 kgCO₂eq, 0.084 kgSO₂eq and 1.48 × 10⁻⁷ kgCFC11eq [33]. Compared to that Iranian study, the results obtained here were substantially superior. Nevertheless, another study in the same country suggested similar results than the present study, with a value of 1.57 kgCO₂eq [41]. Finally, our results were also in agreement with one pioneering work in the investigation of industrial milk production, which presented values of 1.05 kgCO₂eq for climate change and 0.00531 kgPO₄eq for eutrophication potential [42].

3.1.4. Impact Associated with the Functional Unit Compared to Monthly Production

Dulce de leche showed values similar to the impacts of cheese spread, being comparatively lower or equal in all categories. Although dulce de leche is the product with the fourth greatest environmental impact, this finding is based on the sum of all its impacts normalized in reference to the functional unit adopted in this study (1 kg of final product). When considering the impact related to the total kilograms of product produced per month—multiplying the impact related to the functional unit by the total product produced monthly—dulce de leche stands out as the first among the other dairy products (Figure 4). On the other hand, under this analysis, cheese spread presents a lower performance than almost all other products, which is explained by the low quantity produced compared to the others. In an inverse manner to the relation presented for the functional unit, butter, which stood out as the second most polluting, presents itself as the least polluting dairy in this analysis that considers the impact generated by the product for the factory in one month. In addition, inversely, milk and yoghurt, which presented lower impacts relative to the others, stand out as the second and fourth most impactful, respectively.

3.2. Analysis of the Contribution Due to the Different Processes

3.2.1. The Impacts of Raw Milk on the Production

The amount of raw milk inputs employed in the production of each product was the decisive factor in determining the impact of the dairy products (Figure 5). The environmental impacts concerning the milk production step ranged from 20.7 to 99.0% of the total dairy products, affecting less of the photochemical oxidation potential of the dulce de leche and more of the eutrophication potential of the processed milk. It was observed that raw milk mainly influenced the categories of climate change, acidification potential, and eutrophication potential of dairy products. The impact of milk for these categories was in the range of 72.6% to 97.5% for climate change; 65.0% to 98.5% for acidification potential; and 57.6% to 99.0% for eutrophication potential. In addition, the product that had its impact categories most affected by the presence of raw milk were processed milk and cheese, and the one that showed the lowest relationship was dulce de leche. Thus, the results indicated that, in general, raw milk is a determinant input in the impact of dairy products, overlapping all other inputs for most of the categories studied and, to a lesser extent, for the category of photochemical oxidation.

The dairy products received different inputs derived from raw milk. Of these, the one with the highest impact was whole milk at 3.0% fat, followed by cream and skimmed milk. Whole milk had an impact allocation 50.0 times greater than skimmed milk, while cream had an impact allocation 2.3 times greater. This was because we used an allocation by milk solids content in the methodology, which attributed the impacts of this input according to their quantities and percentage of milk solids content. Thus, the products that received skimmed milk, such as cheese spread and part of the processed milk, presented a lower impact ratio per liter of milk—due to the reduced content of solids and the low amount produced. Conversely, the rest of the products that received whole milk or cream had a higher impact ratio per milk-derived input received.

In nearly 20 years of LCA studies about dairy products, authors have shown that the raw milk production phase on farms confers the greatest impact on dairy products [10,31,32,43]. Santos et al. (2017) evaluated the impact of cheese in another region of Brazil and confirmed that milk was the input that impacted production the most. Furthermore, in that study, it was pointed out that the share of milk in the composition of impacts was in an approximate range of 70.0 to 98.0% for the categories evaluated [12]. For this reason, it is important to emphasize how raw milk production affects the life cycle of dairy products, thus being a necessary step within the strategies for reducing the impacts of these products since it also provides their main ingredient.

Although LCA research in dairy industries in Brazil is still incipient compared to countries in the global north—which traditionally investigate this topic—the few studies conducted suggest lower values than those usually found in the literature. Since the present study used data from the Ecoinvent inventory, the results here may be overestimated compared to the Brazilian reality. The raw milk inventory presented by Ecoinvent has impact results that reflect the reality of South Africa and Canada [44]. On the other hand, studies conducted in the southern region of Brazil [45,46,47] and the state of Minas Gerais [48,49] indicated lower values than the inventory present in the Ecoinvent database (

Table 9. Comparison of the impact of raw milk from work with other Brazilian LCA studies published in the literature.

Reference	Climate Change	Acidification Potential	Eutrophication Potential
	(kgCO2eq)	(kgSO2eq)	(kgPO4eq)
Present study (based on Ecoinvent)	1.429	9.64 × 10−3	5.60 × 10−3
[45]
Confined feedlot system	0.776	1.27 × 10−2	5.44 × 10−3
Semi-confined feedlot system	1.065	6.62 × 10−3	1.36 × 10−2
Pasture-based system	1.013	7.73 × 10−3	4.07 × 10−3
[49]
With anaerobic digester	0.881	4.20 × 10−3	1.60 × 10−3
Without anaerobic digester	1.202	2.40 × 10−3	1.80 × 10−3

). Thus, the impact of the dairy products evaluated in this study could be lower than those presented, a point that can be investigated in future studies.

In addition to raw milk, other inputs affected the environmental profile of dairy products, most notably energy sources, packaging materials, and sugar. Energy sources accounted for 0.31 to 70.6% of the products’ impact composition, acting mainly on photochemical oxidation. Packaging inputs influenced with just under 0.20 to 38.3%, with prominence in the category of abiotic depletion from fossil sources. Finally, sugar had impacts ranging from 4.39 to 42.6%, emphasizing the photochemical oxidation category. It is possible to observe that, although small for some categories, these fluxes have a significant influence. Even so, to better understand the contribution of these and the other input flows, it is necessary to analyze them without considering the inputs generated by raw milk, to facilitate the analysis of their relative impacts.

3.2.2. The Impacts of Other Inputs without Considering Raw Milk

Disregarding the raw milk flow, it is also possible to evaluate the environmental impacts generated by the other inputs in dairy production. In this way, we observed that those that contributed the most to the categories studied were energy consumption and materials for packaging the products. More precisely, the decreasing order of influence on the impacts by the other inputs were energy, packaging, sugar, cleaning, effluent treatment, and refrigeration (Figure 6).

Within the energy inputs, thermal energy was the flow that most influenced the impacts of most products. Apart from industrialized milk—which did not consider the use of thermal energy in its production line directly, the contribution of thermal energy varied from 8.90% to 96.4%, with the lowest value for the eutrophication category for the dulce de leche product and the highest for the photochemical oxidation category for the cheese spread. The thermal energy influenced mainly the photochemical oxidation and acidification categories; on the other hand, it had the least influence on abiotic depletion and ozone depletion. The products that were most affected by thermal energy were butter and cheese spread, and those that were least influenced were dulce de leche and yoghurt.

Electricity was a flow that was present in all products. Although it contributed less than thermal energy, electricity was the second flow that most affected the products, disregarding for this analysis milk. Its share of energy contribution in the categories varied from 18.9 to 50.3%, the lowest participation being in eutrophication and the highest in ozone layer depletion. The products most affected by electricity were milk, followed by cheese. The smallest contribution was for dulce de leche. The categories that electricity influenced the most were ozone depletion and global warming; the least affected was eutrophication.

After milk, energy is considered one of the largest contributing inputs to most impact categories in dairy production, especially in the agro-industrial part of the process [50]. One study that evaluated dairy production from a cradle-to-grave perspective identified that thermal energy followed by electrical energy significantly affected the environmental profile of products at the stage from milk production to shipping products to the dairy facility, mainly affecting the categories of climate change and acidification, which was also observed in the present study [51]. One of the studies that evaluated dairy products in Brazil also identified that, when disregarding milk and its processing, thermal energy and electricity use were the main critical points in the production of the product [12].

Within the input stream, in our study, packaging materials represented a notable influence on the impact of products, presenting even greater relevance than energy for some products, such as dulce de leche. The most affected categories were abiotic depletion from fossil sources and climate change; on the other hand, the least affected was photochemical oxidation. The products that were most influenced by this flow were dulce de leche and milk, and the least affected was cheese. Among the materials used in packaging, those that contributed the most were the cans used for dulce de leche. Another material that came close to the values of cans was low-density polyethylene (LDPE), which, however, is used by four products, including the 5 kg dulce de leche. Finally, the material that presented the lowest impact among the others used was cardboard.

Some studies on life cycle assessment in dairy production have also pointed to packaging materials as a critical issue affecting most of the categories, with the category climate change as one of the most affected [50,52]. Similar to our study, in their study on LCA in yoghurt production, the authors González-García et al. (2013) identified that one of the most affected categories is abiotic depletion due to the origin of some packaging being from fossil sources [30]. Regarding Brazil, although LCA research on packaging materials and new food packaging materials is still incipient, there are indications that these fields will be fully developed in the coming years and may go together in this process [53].

The impact of the other flows was mainly from sugar and, to a lesser extent, from cleaning products and effluents. In addition, the stream that affected the products the least was ammonia refrigeration. Sugar, present in dulce de leche and yoghurt, had a great influence in some impact categories, overlapping the sum of the other input streams in the categories of acidification, eutrophication, and photochemical oxidation. Compared to dulce de leche, yoghurt had a greater impact on using this raw material. Regarding the use of cleaning products, these inputs mainly affected the categories of ozone depletion and climate change. Of the cleaning materials, sodium hydroxide had a close but greater influence than nitric acid and had a prominent role in the ozone depletion category, ranging from 3.70% to 18.6% of the total impact of the products. Similar to the use of ammonia for cooling, the use of water for cleaning represented a low influence for all categories, with values below 1.00%. Finally, the outflow represented by production effluent mainly affected the eutrophication potential category, with 0.30% for dulce de leche and 27.0% for milk.

3.3. Sensitivity Analysis

3.3.1. Influence of By-Products on the Assessment of Product Impacts

The production of cheese, cheese spread, and butter in factories generates by-products. The first two dairy products produce whey, and the last one produces buttermilk. In this work, only the impacts of producing the main products were considered since the by-products do not provide a function to the product system and are destined for pig farming on a farm. Thus, the production impact of the volume of whey and buttermilk during the processing of the main products was all attributed to the main products. With this sensitivity analysis, we wanted to identify how the impact of the main products would be reduced in an alternative scenario in which the by-products were used as co-products and to identify if the impacts of their production would be appropriately attributed to them. After conducting sensitivity analysis, we obtained the impact values of the product and the co-product for the three productions (

Table 10. Results of the life cycle environmental impact assessment of some dairy products and their by-products.

Production	Climate Change	Acidification Potential	Eutrophication Potential	Abiotic Depletion	Ozone Layer Depletion	Photochemical Oxidation
	(kgCO2eq)	(kgSO2eq)	(kgPO4eq)	(MJ)	(kgCFC11eq)	(kgC2H4eq)
Cheese production
Whey (co-product)	9.099	0.061	0.034	34.008	3.22 × 10−7	2.72 × 10−3
Cheese (product)	7.207	0.048	0.027	26.939	2.55 × 10−7	2.16 × 10−3
Cheese spread production
Whey (co-product)	2.754	0.018	0.010	12.016	1.05 × 10−7	1.69 × 10−3
Cheese spread (product)	3.205	0.021	0.012	13.980	1.23 × 10−7	1.96 × 10−3
Butter production
Buttermilk (co-product)	1.411	0.009	0.005	6.103	5.41 × 10−8	8.75 × 10−4
Butter (product)	12.477	0.084	0.046	53.958	4.78 × 10−7	7.74 × 10−3

). Figure 7 presents a percentage ratio of the results of environmental impacts for whey and buttermilk compared to their respective main products.

The alternative scenario showed how significant the environmental impacts of co-products are and that their consideration can substantially change the way we look at the environmental profile of producing the main product. As soon as whey was treated as a co-product, it was possible to identify that its impact affected the production of the main product in an amount close to or greater than the main product’s value. On the other hand, it was observed that buttermilk did not influence the impacts on the butter production process in a high proportion similar to whey, representing only 10% of the influence on butter production. What explains the variation in these impacts is the allocation of impacts used, which was determined by the total product produced and its respective milk solids content. Since there is a high production of whey in the manufacture of cheese and cheese spread, the allocation factor for the two productions was high for this co-product. Likewise, the low allocation factor justifies the impact of buttermilk being lower than butter since it is produced less and has low solids content. Thus, the alternative scenario represents an opportunity for a considerable reduction in the impacts of dairy products, especially cheese and cheese spread.

3.3.2. Influence of Multifunctional Process Subdivision and Causal Allocation against Allocation by Milk Solids Content Alone

Conducting a life cycle assessment in a dairy plant implies dealing with processes with multiple outputs—i.e., with process multifunctionality. There are different strategies to deal with this situation, and it is common in studies to opt for the one that fits the availability of data. In cases where there is greater detail of data, the subdivision of multifunctional processes should be used; on the other hand, when there is no possibility of subdividing the flows, allocation is indicated, with allocation by the milk solids content being one of the most used [25]. To evaluate the ability of the choice of strategies to influence the environmental profile of the products, we conducted this sensitivity analysis, in which we compared (i) the standard scenario of subdivision of multifunctional processes and causal allocation with (ii) the alternative scenario of allocation by milk solids content.

With the sensitivity analysis, it could be seen that allocating by milk solids differs from the comparative table of impacts of products presented in the standard scenario, offering more impacts for products of higher production and higher solids content (Figure 8). The alternative scenario generated an allocation of impacts mainly for dulce de leche, for its high quantity produced monthly and its high milk solids content. On the other hand, this product did not present a substantial overestimation in its environmental profile, since its production is large, which amortizes the impact related to its functional unit. In general, there was a reduction in the products’ impacts. Cheese and butter presented an expressive variation, which is observed by the decrease of almost half of the value presented in the standard scenario. On the other hand, milk, and yoghurt, which have high production and lower solid contents, presented lower impacts but close to the standard scenario. These results indicate that the allocation by milk solids content may present values incompatible with a scenario closer to reality (standard scenario).

4. Conclusions

In this study, the environmental impact associated with the production of six dairy products was compared and calculated using the life cycle assessment methodology. In addition, through two sensitivity analyses, we investigated how the valorization of by-products as co-products and how different methods for dealing with multifunctional processes can affect the environmental profile of the products. The study showed that products’ descending order of environmental impacts was cheese, butter, cheese spread, dulce de leche, yoghurt, and milk. As in other studies, it was found that raw milk production is the most critical part of producing dairy products, followed by energy and packaging. Valorizing by-products as co-products afforded a reduction in the impacts present in the main products; this reduction was significant in the case of whey, given its large production. Regarding the strategy to deal with multifunctionality, it was observed that allocating by the level of milk solids only, despite facilitating the calculations, can return unrealistic values that are better represented by the subdivision of processes together with causal allocation. Thus, the main contributions of this work are: (i) to provide a better understanding of the environmental impacts of dairy products in Brazil; (ii) to present the environmental profile of two traditional products (dulce de leche and cheese spread) not yet explored in other studies in the life cycle assessment literature; (iii) to present the degree of influence of processing the by-product into co-product for reducing the impacts of other products; (iv) to question the quality of results offered by using only milk solids allocation in life cycle assessment studies of dairy products.