PACKAGING SPECTRUM: Comparative Life Cycle Assessment (LCA) of selected cosmetics packaging – uncertainty analysis using the Monte Carlo computational algorithm 
19 Oct 2018 13:31

ABSTRACT: Life Cycle Assessment (LCA) is one of the primary methods of assessing sustainability of products, services, processes, building and even whole industries and economies. The aim of the study is to conduct methodological research scenarios of the Life Cycle Assessment for selected packaging applications. Three packaging types – jars for cosmetics – were chosen for this study. They share the same function, and have identical product capacities, but are manufactured from different components and with the use of different packaging materials, which makes them ideal as LCA results examples. This selection will allow the study to focus specifically a factor that is very often quoted as the main disadvantage of LCA – quality of input data. Therefore, the main objective of this study is to show and discuss approaches available to limit this disadvantage. This parameter can be assessed using an uncertainty analysis based on the Monte Carlo statistical methodology. This method is used to mathematically model processes that are too complex to predict using a singular analytical approach.

STRESZCZENIE: Ocena cyklu życia (LCA) to jedna z głównych metod oceny zrównoważonego rozwoju wyrobów, usług, procesów, budynków, a nawet całych gałęzi przemysłu i gospodarek. Celem zadania badawczego jest przeprowadzenie metodologicznych scenariuszy badawczych Oceny Cyklu Życia dla wytypowanych opakowań. Do badania wytypowano trzy opakowania – słoiczki przeznaczone do produktów kosmetycznych. Wszystkie opakowania cechują się tą samą funkcją, i mają identyczną pojemność, ale wyprodukowane są z innych materiałów opakowaniowych i składają się z różnych komponentów. Taki wybór opakowań pozwoli dokonać interesującej analizy porównawczej, skupiającej się na aspekcie uznawanym w literaturze za jedną z największych wad metodologii oceny cyklu życia – niejednolitej jakości danych wejściowych. Jakość danych wsadu może być badana analizą modelowania kalkulacyjnego Monte Carlo. Metoda ta pozwala modelować matematyczne procesy, które są zbyt złożone, aby można było przewidzieć ich wyniki metodą analityczną..

1. Introduction

1.1. Packaging, environment and sustainability assessment

Widespread use of packaging, mostly produced from non-renewable resources, causes a noticeable increase in environmental burdens – the consumption of natural resources, emissions during production, as well as the need for management of increased waste. Increasing public awareness, more stringent legal regulations and the development of knowledge about the environmental impact of products, makes the protection of the natural environment and sustainable development more and more important. More attention is paid to the type of raw materials and their impact on the environment, energy consumption, mode of transport, storage and disposal of post-consumer waste.

The activities of environmental organizations, increased awareness of residents, increasing legal requirements, and above all, the development of knowledge about the impact of many products on the state of the environment led to the development of various methods to assess this impact in the context of environmental threats. The Life Cycle Assessment (LCA) is an example of a method effectively implemented in industrial practice, aimed at limiting the negative impact on the environment. It complies with the international standard ISO 14040 Environmental management – Life cycle assessment – Principles and structure. 

The LCA method seems to be a natural development of both the environmental management system and the waste management strategy. The life cycle assessment analyses the environmental hazards associated with the product throughout its life, including: extraction and processing of raw materials, production (production process), distribution, transport, use, and waste management [1].

The life cycle is defined as subsequent, interrelated processes – from the collection of raw materials to the production of materials, through the production and distribution phase, up to the stage of waste generation and the processes of their recovery and / or disposal [2]. 

The LCA method should form part of the concept of extended producer responsibility for the product and be used by industry as a basic factor supporting the decision-making process related to the selection of packaging for specific product groups.

Packaging tests incorporating the LCA method rely on the recording of environmental burdens in the particular stages of their life cycle (system boundaries). According to this, it is possible to depict the impact of the assessed packaging on individual categories of environmental mechanisms based on life science (like for example quality of soil, use of minerals, water, air, animals and plants, landscape and climate). Knowledge on this subject allows making choices more beneficial for the environment, and thus enables rational management of resources in accordance with the principle of the sustainable development [3]. If the environmental impacts of specific packaging are known, strategies can be defined to reduce them, for example through material changes, technological development, better process management, etc. [4]. An extremely important stage in the Life Cycle Assessment is the selection of the Environmental Impact Assessment Method (LCIA). The chosen method determines which substances used and emitted during the entire life cycle of the packaging are included in the final result and to what extent the selected substances are responsible for individual environmental mechanisms, translated into environmental impacts and damages. The choice of method can significantly affect the final result and the interpretation of the results, which is why it is crucial that the method correctly reflects the environmental mechanisms relevant to the packaging. This was demonstrated in the previous article of the series titled „Comparative Life Cycle Assessment (LCA) of selected cosmetics packaging – comparison of different impact assessment methods” that appeared in June 2018 in this very journal. 

Another critical element of the objective life cycle assess-ment is the quality of input data, which will be the main subject of this article. This parameter can be assessed using an uncertainty analysis based on the Monte Carlo statistical methodology. This method is used to mathematically model processes that are too complex to predict using a singular analytical approach.

1.2. Purpose of the study

The aim of the paper is to conduct methodological research scenarios of the Life Cycle Assessment for selected packaging applications. Three packaging types – jars for cosmetics – were chosen for this study. They share the same function, and have identical product capacities, but are manufactured from different components and with the use of different packaging materials, which makes them ideal as LCA results examples. This selection will allow the study to focus on the factor that is very often quoted as the main disadvantage of LCA – namely, the quality of inventory data. Therefore, the main aim of this paper is to show and discuss approaches available to limit those disadvantages.

This will be done by performing a comparison of uncertainty of input data using the Monte Carlo computational algorithm. Inventory data quality is of utmost importance in LCA as all results are based solely on what is inputted in the beginning of the study. In addition to that, even if data quality is assured, in industrial practice data is never stable – inputs and outputs needed to produce something often significantly vary from one measurement to another. Hopefully, variation in the data can be described by a distribution, expressed as a range or standard deviation. Statistical methods, such as Monte Carlo techniques can be used to handle these types of uncertainties, and calculate the data uncertainty in the LCA results [5]. In to apply the Monte Carlo technique, the input data needs to have variation assumptions included in its core. Fortunately many databases have a very vigorous data uncertainty section build into them, and those databases will be used for the scenarios of this study.

2. Methodology

2.1. Life Cycle Assessment standards and software

Life Cycle Assessment (LCA) is a standardised method. ISO 14040:2006, Environmental management – Life cycle assessment – Principles and framework, provides a clear overview of the practice, applications and limitations of LCA to a broad range of potential users and stakeholders, including those with a limited knowledge of life cycle assessment. While ISO 14044:2006, Environmental management – Life cycle assess-ment – Requirements and guidelines, is designed for the preparation of, conduct of, and critical review of life cycle inventory analysis. It also provides guidance on the impact assessment phase of LCA and on the interpretation of LCA results, as well as the nature and quality of the data collected.

This study uses both standards, and the LCA examples presented here follow the internal LCA guidelines of ISO 14044:2006. SimaPro 8 software is LCA assessment tool in line with ISO 14040: 2006. SimaPro 8 software allows to create full LCA’s, LCA reports, export results, calculate uncertainty and most importantly includes numerous databases with input and output data of thousands of feedstock products, processes, transport and energy mixes. [5]

2.2. Impact assessment methods

For the interpretation of lists of emitted chemical substances used in the uncertainty analysis this study utilises the ReCiPe method. It was chosen on the merit of giving opportunity to assess individual categories of environmental impacts and enabling to recalculate these inflows into categories of environmental damages.

ReCiPe method. 

ReCiPe is the successor of the methods Eco-indicator 99 and CML-IA. The purpose at the start of the development was to integrate the ”problem oriented approach” of CML-IA and the ”damage oriented approach” of Eco-indicator 99. The ”problem oriented approach” defines the impact categories at a midpoint level. The uncertainty of the results at this point is relatively low. The drawback of this solution is that it leads to many different impact categories which makes the drawing of conclusions with the obtained results complex. The damage oriented approach of Eco-indicator 99 results in only three impact categories, which makes the interpretation of the results easier. However, the uncertainty in the results is higher. ReCiPe implements both stra-tegies and has both midpoint (problem oriented) and endpoint (damage oriented) impact categories. The midpoint characterization factors are multiplied by damage factors, to obtain the endpoint characterization values. ReCiPe comprises two sets of impact categories with associated sets of characterization factors. At the endpoint level, most of these midpoint impact categories are multiplied by damage factors and aggregated into three endpoint categories: The three endpoint categories are normalized, weighted, and aggregated into a single score. 

Figure 1 portrays relations between the 18 midpoint categories, and the 3 endpoint categories [6,7].

 

2.3. Uncertainty analysis in LCA

As stated in the introduction to this study, inputs and outputs needed to produce or process a product often significantly vary from one measurement to another. Hopefully, variation in the data can be described by a distribution, expressed as a range or standard deviation [5]. For those uncertainties SimaPro software can use Monte Carlo statistical technique. A very basic example of the way Monte Carlo method works is described in the figure 2 – the description is taken from the software documentation:

 

3. Goal and scope 

3.1. Samples and functional unit

All packaging types for the study were provided by the Novo-Pak company from Otwock (Mazovia region). They comprise of three cosmetics jars of identical capacity (50 ml):

1. Polypropylene (PP) jar

2. Glass/Polypropylene (PP) jar

3. Polylactic acid (PLA) jar

Tested packaging types are presented on figure 3.

The functional unit for this study is 50 ml of cosmetics packed (cream). This allows to make direct comparisons between all three packaging types.

Packaging types differ by their material composition as well as their construction. 

Polypropylene (PP) jar comprises of 5 different elements (all produced from Polypropylene), glass jar comprises of glass base and polypropylene cover in three parts, while PLA jar – the lightest and simples packaging solution – consists of just the base and one element cover. Specific masses of construction materials (in grams) are shown on table 1. 

3.2. System boundaries

The LCA study uses standard Cradle to Grave scenario. This scenario considers the whole life cycle of the packaging. It includes all the processes of the Cradle the Gate scenario (above), and goes on further until the packaging becomes post-consumer waste. It is important to note that all three packaging life cycle assumptions after production are identical (i. e. identical transport routes, identical product packing systems and processes, identical distribution channels, and identical usage trends), and therefore are not indicated in the final results. This was a deliberate choice of the authors, because it allows the results to reflect on the actual differences of the packaging systems. The main difference in the Cradle to Grave boundary is the waste management stage. To illustrate the impact of the waste management options, Polish packaging centric waste management scenario was chosen from the database. It includes recycling, incineration and landfilling of particular packaging elements, and takes into account the environmental credits from recycling (i. e. impact offset, showing the positive environmental benefit of recycling). Recycling, incineration and landfilling percentages of particular packaging materials that are used in this LCA are presented on table 2. 

Those percentages are derived from Polish data of EcoInvent 3 database. It is worth noting the fact that composting is still not a viable end-of-life option in Poland, according to this database.

Cradle-to-Grave system boundary diagrams for all three products are presented on figures 4-6.

3.3. Data collection

There were 2 main sources of the data used for this study:

1. Primary data – the physical samples were obtained from the Novo-Pak company from Otwock (Mazovia region). The company also shared their processing data and explained their manufacturing procedure, which is reflected by the changes to the data from database – the second source of data.

2. Secondary data – data from EcoInvent 3 database concerning feedstock materials, granulate, packaging glass making, additional processing and end-of-life waste management, recycling, incineration and landfilling. Data from literature concerning the energy use of various manufacturing equipment and processes of glass making. Where possible data from database was changed to reflect primary data collected from Novo-Pak.

4. Results

Results are presented in the following tables:

1. Uncertainty analysis (Monte Carlo Method) – of three packaging types using single score probability distribution of the ReCiPe method.

2. Uncertainty analysis – comparisons (Monte Carlo Method) – comparison between pairs of assessed packaging types (1 and 2; 1 and 3, 2 and 3), utilising ReCiPe method, including the following graphs:

a. Single Score difference (A – B) normal probability distribution

b. Weighting – end point – probability difference bars

4.1. Uncertainty assessment

4.2. Uncertainty assessment – comparisons between pairs, ReCiPe method

5. Results

5.1. Uncertainty analysis 

Figures 7-9 portray uncertainty results for each packaging types using Monte Carlo method. Figure is in a form of a histogram of single score uncertainties, with probabilities of single score occurring on Y axis, with dotted line vertical line showing median, and three straight lines red vertical lines showing mean and 95% confidence intervals. For each packaging type separate LCA calculation was run 1000 times, with different values assigned to inventory data, according to its uncertainty classification. As can be seen, all three results resemble normal distribution bell curve, which is an indicator of reliable uncertainty data of inventory processes of EcoInvent 3 database. High-low differences of 95% confidence interval results are relatively small in all three packaging types and amount to roughly +/- 15%, which indicates that the data used for the analysis is of rather high quality, which is reinforced by the fact that mean and median values are really close together.

Figures 10-15 show comparisons between pairs of assessed packaging types. Unfortunately uncertainty analysis methodology of Monte Carlo method, does not allow for comparisons of more than 2 results, therefore for the purposes of this study, three different uncertainty comparisons were made (packaging 1 with packaging 2, packaging 1 with packaging 3 and packaging 2 with packaging 3). Methodology of comparisons is similar to single packaging comparison. LCA calculation is made 1000 times for both assessed pairs and impact assessment scores are subtracted to show the difference. When single score result on the x axis on figures 10, 12 and 14 is positive, it signifies that packaging A statistical environmental impact (accounting for uncertainty) is equal or larger with given probability on axis Y, and vice versa. For easier reference negative scores are coloured red and positive scores are coloured green. When both 

colours appear within 95% confidence intervals, uncertainty analysis demonstrates that based on the uncertainty data of the inventory, environmental impact of compared products may overlap in certain scenarios. 

As can be observed in the instance of three tested packaging, this is not the case, and uncertainty analysis illustrates, that no matter what uncertainty values will be inserted, the mean results of comparisons of three packaging types in ReCiPe method will almost never change. This is elaborated further with subsequent figures (figure 11, 13 and 15) showing probabilities of, damage assessment (end-point). It is important to note that in this portrayal of results some damage assessment categories vary. For example, figure 13 (PP jar and PLA jar comparison) shows that the while resources damage category will always be higher for PP, ecosystems damage category will be always lower for PP jar and in 83% of cases human health damage category will be lower for PP.

Interpreting comparative results of LCA with uncertainty data is very interesting and dynamic and often produces areas for further investigations of specific scenarios and sensitivity analysis. It is why it is recommended to be used in the interpretation stage of the LCA by the ISO 14044 standard ‘Environmental management -- Life cycle assessment -- Requirements and guidelines”.

BIBLIOGRAPHY

[1] Żakowska H., Ganczewski G., Environmental trends in packaging. LCA and „carbon footprint” for selected types of consumer bags [w:] „CURRENT TRENDS IN COMMODITY SCIENCE. Environmenal and Market Research, Zeszyty Naukowe” nr 216/2011, red. Foltynowicz Z., Witczak J., Wydawnictwo Uniwersytetu Ekonomicznego w Poznaniu, Poznań 2011, s. 79–88.

[2] Żakowska H., Systemy recyklingu odpadów opakowaniowych w aspekcie wymagań ochrony środowiska, Wydawnictwo Akademii Ekonomicznej w Poznaniu, Poznań 2008.

[3] Żakowska H., Wytyczne do wykonywania analizy cyklu życia (LCA) opakowań i ograniczenia tej metody, „Opakowanie” nr 11/2004, s. 20–23.

[4] Żakowska H., Ganczewski G., Nowakowski K., Kilanowski M., Przeprowadzenie ekologicznej oceny cyklu życia (LCA) toreb wielokrotnego użytku, Raport końcowy na zlecenie Ministerstwa Środowiska, COBRO, 2010.

[5] Goedkoop M., Oele M., Leijting J., Ponsioen T., Meijer E., Introduction to LCA with SimaPro, © 2002-2016 PRé Consultants. Some rights reserved. January 2016.