Introduction

Over the past few decades, scientific research has made remarkable progress in various fields of knowledge, exploring innovative pathways in all human activities and needs. With the advent of online platforms and journals, scientists can now access a vast store of insights with ease. However, despite the convenience of online resources, most scientists still consider in-person meetings as an essential and stimulating tool, primarily due to the networking, social interaction, and traveling opportunities they offer (Bonnett 2006; Bossdorf et al. 2010; Alberts 2013; Porpiglia et al. 2020). One of the significant challenges of today’s research community is to reduce the environmental footprint of anthropogenic activities and pave the way toward sustainable development, which is clearly impacted by human behavior from a variety of perspectives (Pisello et al. 2015). Unfortunately, scientific research produces extended impacts on the environment (Holden et al. 2017; Abbott 2020; Kier-Byfield 2020). The environmental burden of events and meetings has been an object of interest to several academics concerned about emissions and behavior patterns. Some scholars suggest limiting the number of in-person gatherings to those strictly necessary, rethinking locations and transportation, among other actions to reduce their environmental footprint. The literature provides several editorials with reflections and opinions about this topic (Green 2008; Ponette-González and Byrnes 2011; Boreman 2013). Several frameworks and standards guide the sustainability of meetings and events, as reported by Cai et al. (2014). Among them are the ISO 20121:2012 (event sustainability management systems) (ISO Central Secretariat 2012) and the Green Globe Certification (Green Globe Ltd 2015). The same authors highlighted the importance of the Green Meeting Standards established by the Convention Industry Council (CIC), the Accepted Practices Exchange (APEX), and the American Society for Testing and Materials (ASTM). These standards provide specifications for evaluating and selecting matters related to meetings, events, conferences, and shows.

Numerous studies have been conducted to assess the environmental impact of events, categorized by type, size, and data sources. These studies analyze the effects of various aspects such as travel, accommodations, food services, and the site activities before, during, and after the event (promotional items and marketing). Based on these categories, Edwards et al. (2016) conducted a study on two sports events hosted by the University of Arizona, attended by 60,000 people. The study found that primary data collection is crucial, and traveling was responsible for most of the carbon dioxide emissions, accounting for around 80% of the 2400 metric tons of CO2-eq emitted in 2012 and 1900 metric tons of CO2-eq in 2013. Accommodations were responsible for approximately 18% of the total emissions. The Homecoming events were responsible for the equivalent of 39.62 kgCO2-eq per person in 2012 and 31.50 kgCO2-eq in 2013 (Edwards et al. 2016).

Similarly, Desiere (2016) tracked the carbon footprint of a congress in Ljubljana (in 2014) attended by 646 participants from over 40 countries. The study focused on transportation emissions, concluding that each participant emitted about 500 kgCO2-eq. The authors suggested that selecting a more central location and encouraging the use of trains instead of airplanes could significantly reduce emissions.

Stroud and Feeley (2015) emphasized the crucial role of selecting an appropriate location in reducing the overall emissions of a conference. The authors analyzed the travel distances for four international conferences with 200–400 participants and found an average of about 9564 km. The results showed that each attendee contributed on average to 2.5–3.0 tons of CO2-eq. However, by simply changing the conference venue from Germany to the UK (London), the authors found a reduction of about 200 kgCO2-eq per person. In another study, Hall (2007) designed four different scenarios for an Annual International Conference and calculated the emissions according to the Tyndall Centre for Climate Change Research formula. The estimated emissions from air displacement were 7.725 × 105 kgCO2-eq for 357 participants or 2163 kgCO2-eq per person.

Spinellis and Louridas (2013) assessed the carbon footprint of conference papers by collecting all available data from Scopus and crossing it with additional information from ISI Web of Science. They then classified the papers based on conference location and traveler origin. The researchers found about 1.17 million conference papers in 2008, which were associated with 9.39 × 108 kgCO2-eq emissions, corresponding to 0.23% of emissions from international flights in that year. The findings highlight the significant environmental impact of the scientific community on our society.

Neugebauer et al. (2020) investigated the paradox of sustainability in conferences through life cycle assessment (LCA), including preparation, execution, traveling, and finishing activities related to a 3-day international conference. The authors also quantified the emissions related to hospitality, such as accommodation and catering services. The estimated carbon footprint was 4.55 × 105 kgCO2-eq, i.e., 570 kgCO2-eq per participant.

In a similar study conducted by Hischier and Hilty (2002), the effect of different measures to reduce the environmental impact of a 3-day international conference was investigated. Although the study did not consider the impact of accommodations and meeting site structure due to lack of information, it revealed that reducing promotional items and printed material had a positive effect despite more than 95% of the environmental impact derived from traveling. In addition, the centralized meeting scenario reduced from 80,000 kgCO2-eq to about 45,000 kgCO2-eq considering simultaneous decentralized meetings.

In a more recent study, Toniolo et al. (2017) investigated and confirmed the validity of the LCA method in measuring the potential environmental impacts of an event. The study accounted for all materials, energy, and water used before, during, and after the event, as well as waste products. However, traveling was not considered. By doing so, the authors managed to appreciate and compare minor impacts usually dominated by transportation, e.g., accommodation, and food services. In addition, the calculations demonstrated the importance of emissions related to energy use and materials employed for organizing these meetings.

Numerous studies have shown that traveling has a significant impact on the environmental performance of conferences, events, and meetings. Indeed, flights produce large amounts of greenhouse gases (GHG)—primarily carbon dioxide from burning fuel. Therefore, the location of an event is a crucial factor to consider when planning. In recent years, researchers have explored the possibility of resorting to online meetings as a partial substitute for in-person gatherings to reduce the overall environmental footprint of research. For instance, Periyasamy et al. (2022) estimated that physical meetings can impact 55 times more than virtual ones.

The topic gained even more attention after the COVID-19 pandemic forced gatherings to be replaced by virtual ones (Bottanelli et al. 2020; Porpiglia et al. 2020; Periyasamy et al. 2022). However, the actual feasibility of this new form of research is still under debate. While there is an expectation of a reduction in environmental burdens, there are also concerns about the loss of human connection in the virtual environment (Porpiglia et al. 2020; Arul Vallarasi and Regi 2022). Tables 10 and 11 in Appendix provide a summary of research papers that delve into the impact of events and meetings, specifically focusing on the emissions attributed to science-related activities. Literature can be divided into three distinct lines of study. First, several papers evaluate the ecological aspect of these practices. A second group of papers discusses the social impacts of modifying the status quo. Finally, researchers have written open letters on the subject.

The present study aims to assess the environmental impact of an ongoing European project, the SWS-Heating, which aims to create a new seasonal thermal energy storage system using an innovative sorbent storage material embedded in a compact multimodular sorption unit. The project involves 16 organizations, including universities and companies in seven European countries, with five board meetings concluded by March 2020. This work explores the intersection of social and environmental impacts, covering ecological and social perspectives of international project meetings. The authors aim to improve the environmental performance of these meetings without any negative consequences on their social role. Most literature assesses only one of these sustainability aspects at a time. The findings of this study will provide valuable insights into the overall sustainability of international research meetings and support coordinators in the strategic planning of future projects to optimize their productivity and ecological footprint.

Research methodology

In this work, we conducted a comprehensive sustainability assessment of the first five board meetings of the SWS-Heating European project (Grant Agreement No. 764025). Our assessment focused on two key areas: (i) environmental footprint from a life cycle perspective and (ii) user acceptance through language processing analysis. To do so, two dedicated online forms were designed and distributed to meeting organizers and participants between February and May 2020 (see Fig. 1). The utilization of online surveys has raised concerns about the potential impact of unequal Internet connectivity and technology (Fabiani et al. 2021). Despite these concerns, we determined that this approach was suitable given the specific group of participants and the onset of the pandemic. The obtained responses were thoroughly analyzed and are presented in Fig. 2.

Fig. 1
figure 1

The overall flow of the research analysis

Full size image
Fig. 2
figure 2

Participants’ presence in each meeting (Athens 1, Perugia, Messina, Lleida, and Athens 2) and collected replies divided by country

Full size image

Elaboration of the questionnaires

Participants’ questionnaires

The participants of the SWS-Heating group meetings were asked a series of qualitative and quantitative questions, which can be found in Supplementary Material 1. The survey was divided into two parts. The first part consisted of open-ended questions that focused on personal perception of virtual meetings, projections for face-to-face meetings, and opinion about the SWS-Heating project meetings. The second part of the survey covered personal information, including eating habits, accommodation preferences, travel standards, and specific questions about participation in work meetings and social activities. Each group meeting was assessed separately, using a dichotomous question, followed by detailed multiple-choice and short open-ended questions.

Hosts’ questionnaires

The organizers of each group meeting were asked to complete a second survey, which was divided into four sections, and primarily consisted of open-ended questions (refer to Supplementary Material 2 for details). The first section of the survey focused on promotional items provided during the meeting, including details on the type of item, its source, delivery method, and packaging. The second section inquired about food services, including catering and restaurant selection, using open-ended and multiple-choice questions. The third section of the survey asked about building facilities, equipment, and waste management, using both open-ended and dichotomous queries. Finally, the organizers were asked to report any transportation arrangements for secondary activities, such as dinners or visits.

Environmental impact assessment

The environmental impact assessment utilized data gathered from the surveys to analyze the impact of various factors, such as transportation, accommodations, building energy, printed materials, consumables, and meals provided during the event (including coffee breaks, lunch, and dinners). The information is presented in Figs. 3 and 4. In cases where survey data were insufficient, secondary data were obtained through a comprehensive literature review. The study was conducted using the LCA framework standardized by ISOs 14040 and 14044 (International Organization for Standardization 2006a, b). “Goal and scope” outlines the aim and approach of the study, while “Functional unit and boundaries” defines the functional unit and the system boundaries. The life cycle inventory for six activity groups is presented in “Life cycle inventory”. Finally, the results are presented in “Results” using a single-point method for transparency and simplification. The results include a comprehensive impact assessment for organizing the meeting as well as individual guests.

Fig. 3
figure 3

Study boundaries for the hosts’ comprehensive emissions

Full size image
Fig. 4
figure 4

Study boundaries for the participants, and individual emissions

Full size image

Goal and scope

The purpose of the environmental analysis is to assess the potential environmental impacts resulting from five meetings of the ongoing SWS-Heating European project. This evaluation considers all aspects of the meetings, including planning, meeting site requirements, meals, materials, accommodations, and transportation from a cradle-to-grave perspective. Furthermore, this study aims to raise awareness within the scientific community about their environmental footprint and resource depletion habits.

Functional unit and boundaries

The functional unit of this study is the sum of all inputs and outputs associated with the conduction of five 2-day-long board meetings in four different European cities: Athens (Greece), which hosted two meetings, as well as Perugia (Italy), Messina (Italy), and Lleida (Spain). By presenting the results for each meeting separately, this research aims to evaluate the differences and provide valuable insights for future events. It is important to note that this study did not consider the additional technical meetings taken in Messina in September 2018, Athens in March 2019, and Malmo in January 2020 where the technology was being implemented.

The boundaries of our study excluded the construction and infrastructure of the facilities. Unfortunately, we were unable to model waste and water collection systems due to inaccuracies in the replies and a lack of information regarding the water network from the sites. Additionally, we did not make specific distinctions regarding the food’s origin and preparation methods due to the system complexity and the lack of information.

To evaluate the environmental burdens related to the SWS-Heating group meetings, the Simapro version 8.4.0.0 software and the ecoinvent database were used (Wernet et al. 2016). In addition, the Agribalyse 3.0 (Agence de la Transition Ecologique 2020) data were included for missing food items. The impact assessment in terms of comprehensive and individual emissions using the single-point Global Warming Potential (GWP) method for the horizon of 100 years developed by the Intergovernmental Panel on Climate Change (IPCC 2013 V1.00) was used. This method relates greenhouse gas emissions in the air to climate change, which impacts the ecosystem, human health, and resources (PRé-Consultants 2020). The final impact is expressed in kgCO2-eq as in most studies as CO2 emissions’ estimation is essential for benchmarking.

Life cycle inventory

This section presents and discusses the life cycle inventory separated by activity. The ecoinvent market specification refers to the processes from a specific location (different production countries) and includes average transportation and eventual losses. Country-specific data were preferred for this study. However, when these are unavailable, the selection follows the sequence: European, Swiss, or Global datasets..

Transportation

Tables 1 and 2 present a comprehensive transportation inventory. The data were collected in a systematic manner, as illustrated in Fig. 5, and categorized according to the mode of transportation, including aircraft, buses, cars, trains, and bicycles. The distances traveled by air were calculated using the Great Circle Mapper website (meco media & communication GmbH 2008). Table 3 displays the distances between the airports and the meeting center, as reported by survey respondents. The means of transportation used to travel to and from the airport were also determined based on survey responses, with options including bus, car, or train. The distances traveled by the participants were calculated using Google Maps, considering the shortest route possible based on the itinerary provided by the respondent. For travel to accommodations, meetings, and event dinners, we considered the position of a central hotel, events location (specified by the host), eventual visits according to the agenda, and the means of transportation reported by the participant.

Table 1 Transportation inventory
Full size table
Table 2 Accommodation inventory divided per hosting country
Full size table
Table 3 Car distance from the city center to the main airport hub assessed by Google maps
Full size table
Fig. 5
figure 5

Mapping scheme for transportation. The section was mandatory, and the interview could skip the steps that did not apply

Full size image

Unfortunately, most of the participants did not specify the type of car used. Therefore, we assumed the process market dataset for medium-sized passenger cars, petrol, category EURO 4, based on references from the literature (De Camillis et al. 2010; Neugebauer et al. 2020). Distances between cities assumed the regional passenger train dataset in Switzerland, while distances inside cities were calculated based on the passenger train urban dataset. Moreover, since no detailed information was available, the distance covered by bicycle was estimated in two-daily 5 km trips, for 2 days.

Accommodation

Table 2 presents the inventory section that focuses on the number of overnight stays reported during the meeting. Interestingly, despite the meeting’s duration being 2 days, most participants stayed for 3 days. One person even reported an extended visit to Lleida. The attendees lodged in centrally located three- or four-star hotels, but unfortunately, most of them did not provide the exact names of their hotels, making it difficult to determine their locations. Consequently, the hotels’ energy consumption (electricity and heating) and water standard consumption were derived from the XENIOS project (Dascalaki and Balaras 2004). XENIOS is the only reference that reports data from audited buildings in Italy, Spain, and Greece, making it highly compatible. To calculate the energy consumption, a room of 4.5 m × 7 m (31.5 m2) was assumed in each case, based on an executive room according to Neufert (2019).

Office organization

This section of the inventory considers all impacts that arise from office activities related to meetings attended by the participants. These activities include printing (energy, ink, and paper), computer work, and Internet use, as outlined in Table 4. The printing services considered recycled paper printed on both sides, as reported by the participants. For computer work and Internet use, we estimated 2 h of work dedicated to the meeting preparation over a period of 5 days, based on the indications provided by the University of Perugia team.

Table 4 Office organization inventory
Full size table
Meeting site organization

The preparation of the meeting site (as outlined in Table 5) involves all the necessary activities required to ensure that the site is ready for the event and the following 2 days. To determine the resources required for promoting the event and marking the building, we relied on data provided by the hosts and specific compositions found in the literature (Toniolo et al. 2017). Additionally, the University of Perugia’s organizing team provided data on the use of hygiene materials which were taken as reference.

Table 5 Site meeting organization inventory
Full size table

The meetings were held in various locations, including the School of Mechanical Engineering (75 m2) and Stavros Niarchos foundation Cultural Center (64 m2) in Athens, Palazzo Murena (115 m2) and Centro Interuniversitario di Ricerca sull’Inquinamento da Agenti Fisici (52 m2) in Perugia, ‘Auditorium of Istituto di Tecnologie Avanzate per l’Energia’ (75 m2) in Messina, and ‘Aula Magna Saló Víctor Siurana’ (170 m2) in Lleida. Unfortunately, the hosts were unable to provide information on the energy consumption of these buildings. Therefore, we estimated the non-residential energy consumption to be 280 kWh/m2 based on data from the Buildings Performance Institute Europe (Economidou et al. 2011). It is important to note that several factors can influence the final energy consumption, including construction technology, internal temperature setpoint, and user behavior. Additionally, the organizations reported the number of computers used (desktop or laptop), and we included the transportation of 50 kg of paper by light commercial vehicle (considering the database input Europe without Switzerland) corresponding to the material printed outside the institutions (graphic services). A total of 3.3 km was considered for Perugia, 2.4 km for Messina, and 1.5 km for Lleida. Furthermore, we took into account the Internet access and computer operation per attendee, as well as standby mode during coffee breaks. During lunch and site visit periods, all personal equipment was considered to be in off mode. Finally, we assumed 45 h of computer work for each organization based on the University of Perugia team’s experience.

Promotional items

This section pertains to the promotional items that were reported by the hosts. These items include posters, lists of attendees, agendas, blocks of notes, and brochures. We also considered the packaging and transportation as reported by the hosts. However, since we had no detailed information about the items like badges, pens, and cotton bags, the amounts were estimated based on the University of Perugia case and simplified in the model as in Table 6.

Table 6 Promotional items inventory
Full size table
Meals

The model considered the specific catering profile, menu description provided by the hosts, and observations made during the event. Therefore, the inputs were simplified, as reported in Table 7. In addition to ecoinvent, emissions for various items, such as tea, coffee, pork, cheese, wine, sausage, ham, egg, beer, fish, and yeast, were collected from the French database Agribalyse 3.0 (Agence de la Transition Ecologique 2020). It is worth noting that none of the 62 participants had any special dietary requests.

Table 7 Meals, packaging, and catering products
Full size table

To estimate energy and water data inputs, the study by Neugebauer et al. (2020) was used, which considered two coffee breaks and two lunches served to 800 people. The overall energy consumption of the restaurants was calculated based on the CIBSE Guide F for a typical practice in a restaurant with a bar (Filippín and Larsen 2009). The menu considered was designed based on the Italian balance diet requirements (SINU—Società Italiana di Nutrizione Umana 2014) representing Mediterranean countries. However, it is important to note that the menu does not replicate the real complexity of all the ingredients in this diet. Additionally, it is not possible to reproduce the precise portions and preparation methods, storage techniques, and waste patterns as they were.

Researchers’ stance evaluation

The purpose of this study was to evaluate the environmental and social impacts of a global, interdisciplinary, and multicultural project. To gather qualitative feedback on virtual meetings, participants were asked to share their thoughts and impressions through open-ended questions (as explained in “Participants’ questionnaires”). While natural language can provide valuable insights, it can be difficult to analyze quantitatively. Therefore, we utilized a two-stage language processing approach. First, we manually analyzed the responses to identify the overall sentiment of the participants’ dissertations. Then, we applied a multi-stage natural language processing (NLP) procedure to verify the main results using established statistical methods.

For the NLP analysis, we parsed and tokenized all responses to obtain a unique set of alphanumeric characters separated by blanks and punctuation marks. This allowed us to identify the primary terms (or tokens) in the text and create a list of unique terms (or types) representing the reference vocabulary of the collection. We then carried out additional pre-processing procedures to reduce language variability and eliminate possible noise in the data set. This included converting all characters to lowercase, deleting all numbers, and correcting all misspelled terms. We also applied lemmatization to reduce language variability at a morphological level, transforming each inflected term into its canonical form. Finally, we pruned the vocabulary by deleting message stop-words and removing the most common words in the English language to avoid non-informative terms. After completing the initial analysis, we further processed the textual content as structured data within the R environment. Our goal was twofold: (i) to gain insight into the general sentiment expressed by participants regarding online group meetings, and (ii) to evaluate the frequency and distribution of words across the corpus, along with their most common connections.

To achieve this, we utilized a dedicated language processing analysis within the ‘tidy tool’ ecosystem in R. We selected a word and emotion connection, also known as the ‘NRC lexicon,’ which was constructed and later validated using crowd-sourcing and categories of words and their associations with eight basic emotions corresponding to four polar opposites (anger, fear, anticipation, trust, surprise, sadness, joy, and disgust) (Plutchik 2001) and two sentiments (negative and positive) (Mohammad and Turney 2010). To measure frequency distribution, we calculated the correlation factor among the different words in the previously processed corpus and the PageRank centrality of each node. The PageRank centrality is a crucial parameter that takes into account the number of links received by the node, the link propensity of the linkers, and the centrality of the linkers (Page and Brin 1998). In other words, a node is considered essential if it is linked to other important nodes and links parsimonious nodes, or if it is highly linked.

Results

Environmental impact assessment

The organization of the five group meetings resulted in a total of 1377 kgCO2-eq emissions. This breaks down to approximately 120 kgCO2-eq for promotional products and 1257 kgCO2-eq for the meeting site activities. In more detail, 84% of the emissions produced by the meeting site and organization are due to the energy consumed by the buildings. When divided by the total amount of participants (143), the emission per capita is about 9.6 kgCO2-eq.

Regarding the estimation of the average individual emissions produced by each meeting participant, the main results are summarized in Fig. 6, which is divided into impacts due to transportation, accommodation, organization, and meals. The emissions obtained from these categories add up to 28,088 kgCO2-eq. Specifically, 1259 kgCO2-eq is due to accommodation, 24,299 kgCO2-eq is due to transportation, 61 kgCO2-eq is due to office organization, and 2468 kgCO2-eq is due to meals. Therefore, we can estimate that, on average, each participant emitted about 455 kgCO2-eq.

Fig. 6
figure 6

Emissions in kgCO2-eq for 62 entries are divided by meeting activity

Full size image

Transportation is the primary contributor to emissions, accounting for a staggering 86.5%. Meals and accommodations follow making up 8.8% and 4.5%, respectively. Within transportation, air travel is responsible for a whopping 73% of CO2 emissions, while passenger cars account for 26% (refer to Fig. 7). When we consider the combined impact of both hosts and visitors, the total emissions per participant amount to 463 kgCO2-eq. This translates to a comprehensive environmental impact of approximately 66,161 kgCO2-eq across all five meetings.

Fig. 7
figure 7

Emissions in kgCO2-eq are divided by the mean of transportation

Full size image

The results per meeting were separated and are reported in Table 8. Once again, the impact of transportation on the meeting’s performance is clearly visible. However, a fascinating insight regarding the different locations can be observed in terms of meals and meeting sites. It is interesting to note that the higher emissions during both Athens meetings can be partially attributed to the impact of the Greek energy mix, as demonstrated in Table 9. In fact, the impact is almost 50% of the total for the meals. Furthermore, when it comes to meeting site emissions (as depicted in Fig. 8), it was discovered that using a larger conference room had a significant impact on Lleida’s environmental performance. This impact corresponds to 24.3% of the emissions for this specific activity.

Table 8 Total emissions calculated per meeting and per activity
Full size table
Table 9 Total consumption per country, emissions according to the energy mix, and total emissions
Full size table
Fig. 8
figure 8

Sum of emissions in CO2-eq per ‘meeting site’ related activities. (The numbers from 1 to 5 indicate the specific meeting: 1 = Athens; 2 = Perugia; 3 = Messina; 4 = Lleida; 5 = Athens, second meeting)

Full size image

Life cycle interpretation

In agreement with the literature presented in “Introduction”, traveling contributes significantly to the environmental burdens produced by the project meetings. In fact, our calculations have estimated that a total of 24,299 kgCO2-eq are emitted, with 17,806 kgCO2-eq coming from air transport alone, which equates to 287 kgCO2-eq per person, i.e., approximately 62% of the overall emission. These findings confirm the importance of considering the impact of air travel, even for relatively short distances within Europe. The obtained shares are in line with the reported literature (Desiere 2016; Neugebauer et al. 2020), which highlight the crucial role of meeting location in the environmental sustainability of a project, rather than the number of participants. Therefore, it is clear that the impact of air transport must be a key indicator to be estimated. However, recent research has raised concerns about the accuracy of the SimaPro characterization factors under the IPCC method. Specifically, it has been argued that these factors do not consider the increased effects caused by the emission of GHG at high altitudes (Jungbluth and Meili 2019), primarily due to the considerable uncertainty regarding this calculation.

A new scenario was developed to assess the impact of the database used for the simulation model. The first modification involved changing the type of bus from a regular bus to a trolleybus (database input ‘Transport, regular bus {GLO}| market for, Alloc Def, U’ to ‘Transport, trolleybus {GLO}| market for, Alloc Def, U’), resulting in a 6% reduction in emissions.

The second modification involved substituting a more generic input for a medium-sized petrol car (database input ‘Transport, passenger car, medium size, petrol, EURO 4 {GLO}| market for, Alloc Def, U’ to ‘Transport, passenger car {RER}| market for, Alloc Def, U’). The newly selected input considers a mix of different sizes, EURO classification, and fuels, such as gas, diesel, petrol, and electricity, and produces 12% emissions’ reduction.

Finally, the alternative scenario considered more localized data for passenger aircraft, considering an exclusive European set, rather than a general intracontinental and intercontinental dataset, resulting in a 23% increase of emissions mainly due to the fuel geography. This highlights the importance of database selection and improving details when approaching transportation. The contributions related to accommodation, materials, communication, and food were less significant to the overall environmental footprint of the SWS-Heating project meetings. However, they still represent a non-negligible source of pollution. Among these activities, the energy (electricity and heating) consumption associated with meals accounted for 63% of their total emissions.

Regarding the food provided during the meetings, Fig. 9 shows the CO2-eq emissions associated with one kilogram of each item in the menu reported by the hosts according to the ecoinvent (Wernet et al. 2016) and Agribalyse 3.0 (Agence de la Transition Ecologique 2020). The emissions can vary based on production methods, product origin, and other factors, which can add uncertainty to the results. To provide some context, a study by Clune (2019) examined the environmental impact of various food items in terms of embodied CO2 by constructing a Food Global Warming database. The author estimated that field-grown fruits and vegetables have emissions of about 0.40 kgCO2-eq/kg, while grains, cereals, and pulses have emissions of 0.51 kgCO2-eq/kg. Poultry has an average of 3.71 kgCO2-eq/kg, non-ruminants have about 5.72 kgCO2-eq/kg, and ruminants have an average of 26.61 kgCO2-eq/kg. It is important to note that the environmental burden attributed to food can be characterized by a large amount of uncertainty due to several assumptions.

Fig. 9
figure 9

Comparison of kgCO2-eq emissions per kg of food

Full size image

Accommodation is a crucial factor in our model, ranking third in terms of relevance. Our survey results revealed that 84% of participants stayed in four-star hotels, while 13% opted for three-star hotels, and only one person stayed in a five-star hotel. It is worth noting that all hosts were assumed to stay at their homes, so their impact was excluded from these percentages. Numerous literature studies have highlighted the importance of electricity and heating consumption in hotel stays. For instance, Santamouris et al. (1996) conducted an audit of 158 hotels in Greece to study energy consumption reduction solutions. They found that the average annual consumption was 273 kWh/m2, second only to the hospital sector. Similarly, Beccali et al. (2009) studied the hotel sector in Sicily (2002) and classified the samples by clusters according to the hotels’ categories. They discovered that the CO2 emissions by overnight stay were 4.69, 7.05, and 15.77 kg for one-, two- or three-, and four- or five-star hotels, respectively. The resulting average was 9.17 kgCO2-eq, which is very similar to the 10.58 kgCO2-eq considered in our calculations, which did not differentiate between hotel categories. Despite the building performance, many services that are related to the hotel category, occupant behavior, size, and others can increase the potential impacts of this activity (De Camillis et al. 2010). Concerning meeting organization and promotional products, the highest shares of the impacts are associated with electricity and heating consumption and printing. However, they only contribute narrowly and marginally to the comprehensive impact of the event.

Researchers’ stance interpretation

Manual language processing

The initial inquiry into participants’ feelings about virtual meetings yielded a generally favorable response, particularly for straightforward topics and smaller working groups, as it can expedite the decision-making process. However, five respondents expressed caution, emphasizing that face-to-face meetings are irreplaceable and more suitable for complex discussions.

The interviewed participants were not entirely convinced about replacing face-to-face meetings with virtual ones. Most of them believe that virtual meetings are appropriate only in specific situations. The more confident participants (three of them) emphasized the need for technological improvements to make virtual meetings more effective. The group unanimously agreed that face-to-face meetings were valuable to the project’s development.

Finally, participants expressed concerns about the consequences of substituting face-to-face meetings with virtual ones in a European project like SWS-Heating. Most people considered the possibility of misunderstandings, delays, and networking losses as potential side effects. Only two participants considered virtual meetings feasible once people get used to them, but with reservations.

Automated language processing

Figure 10 shows the results of the NRC classifications of the analyzed text. These findings support the impression derived from manual processing and are consistent with the results of Kim et al. (2022) from a virtual medical conference. Indeed, the NRC results indicate that over 82% of the investigated words are positive. The most common sentiment toward online meetings is trust, accounting for 37%, followed by anticipation at 19%, and joy at 18%. These emotions have a high degree of similarity in positive sentiment, according to the wheel model from Plutchik (2001). To gain a better understanding of the general opinion about online meetings in European projects, a closer look at the open answers of the meeting participants is necessary. The stronger correlations among the words were filtered and visualized in a network. The dimension of the node denotes its Page Rank centrality, and the color of the word gives the number of its repetitions, while the thickness of the line gives the intensity of the correlation (see Fig. 11). For example, Fig. 11a considers a minimum number of repetitions of 3 and a minimum correlation of 0.45. In this case, four different groups of words can be detected.

Fig. 10
figure 10

Classifications of the participants’ words in the reference NRC lexicon categories

Full size image
Fig. 11
figure 11

Word networks based on the correlation of word counts considering only connections with a a correlation greater than 0.45 and a word frequency of at least 3 repetitions and b a correlation greater than 0.35 and a word frequency of at least 4 repetitions

Full size image

The bottom left group in Fig. 11a is densely packed with many connections among most of the nodes, characterized by high correlations. Indeed, most of these nodes share a similar centrality value and are equally important and connected. This group indicates a strict correlation among words like “short,” “conversation,” “partners,” “faster,” “communicate,” “positive,” “focused,” “solving,” and “directly.” This suggests that most meeting participants consider online meetings as a faster and more direct way to interact with other partners. Additionally, the presence of terms like “task,” “specific,” “tool,” and “project” indicates that online meetings are also seen as a valuable tool for efficiently and rapidly solving specific problems. Moving to the bottom right group, we see fewer terms, but the words “person,” “help,” “difficult,” “aspect,” and “issue” have higher centrality values. This suggests the opinion that the most challenging issues should be discussed in person. The two smaller groups at the top of Fig. 11a, on the other hand, show a significant correlation between the pair “opinion–involved” and the pair “misunderstanding–communication”. However, due to the poor number of significant connections with other words, it is difficult to evaluate these strong correlations. Figure 11b shows the word network obtained by considering a minimum number of repetitions of 4 and a minimum correlation of 0.35. In this case, only the two main groups described before can be noticed.

Discussion

Limitations of the study

The LCA is a simplification based on reality. This work attempts to reconstruct and report past events that occurred in different geographies and timeframes, which can add complexity to the model. As a result, the assessment is limited to the details provided by the organizers and participants of the event. To limit uncertainties in the model, this study made assumptions and estimations based on the literature, consolidated database, and expert experience. When information was not provided, it was estimated using secondary data from existing literature. However, it is important to note that the lack of information regarding actual building energy and water consumption, total waste produced by organizations, and energy flows related to all activities should be considered. Overall, events would benefit from the environmental calculations before their execution, starting from the planning stages and continuing throughout the event. Post-meeting evaluations show high levels of uncertainty, even if, for instance, some high-impact activities such as traveling are traceable by receipts.

On the other hand, NLP is a powerful tool that enables computers to comprehend text in the same way humans do. However, it is not without its limitations and challenges. One of the most important obstacles is contextual understanding, where the same word or phrase can have different meanings depending on the context. Another limitation is detecting irony and sarcasm, which often use words and phrases that, strictly by definition, may be positive or negative but connote the opposite. Ambiguity, whether lexical, semantic, or syntactic, is also a significant concern in NLP, referring to sentences and phrases that can have multiple interpretations. To address these issues, future research could limit some of these issues using supervised machine learning and model pre-training. However, this would require many domain-specific corpora and expensive computational resources.

Finally, using online surveys for retrieving the information for the LCA and the NLP is intriguing but susceptible to under-coverage issues due to uneven Internet connectivity and technology distribution. Nevertheless, in this study’s case, which involved researchers participating in project meetings, it was deemed acceptable. Future research could replicate this study outside of a single European project, combining in-person and online questionnaires and employing different and more advanced sampling and data collection techniques.

Integration of different sustainability perspectives

The investigation presented here delves into the overall sustainability of international research and provides intriguing insights that could benefit from an organic interpretation. By integrating considerations about environmental sustainability and participants’ opinion mining, we can gain a more comprehensive understanding of the topic at hand. In today’s world, researchers have the luxury of using online platforms to access a vast amount of information and share ideas without having to meet face-to-face in a specific location. However, in-person meetings still play a crucial role in research projects.

Despite this, manual interpretation and the text-mining investigation carried out in this work suggest that most researchers have a positive outlook of online meetings. They are considered an effective way to discuss specific issues that need to be addressed rapidly within the project’s tasks. However, in-person meetings are still perceived as the most appropriate solution for addressing complex issues that require face-to-face interaction. This finding is also supported by Straus and McGrath (1994). Sniezek and Crede (2002) conducted a study comparing decision-making processes in in-person groups to technology-assisted meetings. They found that there were no significant differences related to the quality of the decisions. Most distinctions were related to the participants’ perception of confidence in the outcomes of the meetings.

From an environmental point of view, as demonstrated in “Environmental impact assessment” and according to similar literature studies (Desiere 2016; Neugebauer et al. 2020), traveling has the highest environmental impact. Even though all SWS-Heating project meetings were organized within Europe and among European participants, most attendees traveled by airplane, resulting in 287 kgCO2-eq emissions per capita. However, other activities related to accommodation, materials, communication, and food also contributed to the overall environmental footprint of the meetings, albeit to a lesser extent than transportation.

To provide a more detailed breakdown, per capita emissions were calculated for each meeting. The first Athens meeting resulted in 600 kgCO2-eq, Perugia in 362 kgCO2-eq, Messina in 511 kgCO2-eq, Lleida in 299 kgCO2-eq, and the last meeting at Athens 474 kgCO2-eq. The 30% difference between the average emissions of Athens and Lleida can be attributed mainly to transportation and a smaller scale for meals. Most participants who responded to the first Athens meeting were from abroad, while most of the responses obtained for the meetings in Spain and Italy were from locals, resulting in a reduced environmental impact from air travel.

These combined findings represent good insights for optimizing the overall sustainability of scientific research. As such, we propose a novel management approach for European projects that leverages remote meetings in specific circumstances. This approach can significantly reduce the environmental impact of a project, facilitate the swift and effective resolution of minor issues, and ultimately improve the lifestyles of researchers. Moreover, project leaders worldwide could promote scheduled or online meetings as a partial substitute for in-person meetings, which should only be organized a few times per year to discuss the most controversial or crucial research topics. Additionally, meeting locations could be planned based on a preliminary impact that considers the most relevant impact activities, such as transportation. This approach can also coordinate other activities like site visits in the surroundings or courses to optimize time and displacement. Limiting emissions and offsetting unavoidable ones can help mitigate these impacts. However, it is essential to evaluate how this approach would affect the dynamic among participants from a social perspective.

Conclusions

In recent years, there has been a growing concern about the impact of human activities on the environment. As a result, environmental research has been working tirelessly to reduce the environmental footprint of all kinds of human activities. Researchers from all over the world have been collaborating and sharing their knowledge in multifaceted international research projects to achieve this goal. However, it has been observed that most of these activities also generate severe impacts, partly because of the many conferences and group meetings that are held. Therefore, this work aims to investigate the overall sustainability of an ongoing H2020 European project by examining the environmental burden produced by the group meetings and evaluating researchers’ opinions about online events. To achieve this aim, we designed and carried out a dedicated online survey campaign during the project. The survey was completed by almost 70 respondents, including hosts and meeting participants. The results were first checked and classified, and then, the environmental burden due to transportation, accommodation, office organization, site meeting organization, promotional items, and meals was evaluated from a life cycle perspective, based on the data from the survey. Later, data were analyzed to assess researchers’ stance on online meetings. The environmental investigation revealed the significant impact of transportation, particularly air travel, on the overall greenhouse gas emissions generated by the five SWS-Heating project meetings. While accommodation, promotional items, communication, and food had a smaller impact, they still contributed to emissions that cannot be ignored. The data collected during the investigation highlighted the potential for digitalization to reduce the environmental impact of paperwork and promotional items. However, the investigation also revealed the challenges of mapping processes that are not directly controlled by the hosts, such as food production and waste management.

The results of our study have highlighted that in-person meetings are the most effective tool for discussing complex issues and debating crucial topics related to a project. However, online meetings are also valuable for addressing specific issues within the project’s confined tasks in a rapid and effective manner. To reduce the environmental impact of face-to-face meetings, strategies such as selecting an optimized site and offsetting measured emissions can be implemented.

Overall, online meetings have the potential to be a promising management tool for researchers and policymakers involved in international projects. They can improve problem-solving for specific issues and tasks within the project work frame. Additionally, integrating and coordinating online meetings with in-person meetings can reduce the environmental impact of international projects, especially when multiple countries are involved. As an alternative, hybrid meetings with participants on virtual platforms and in-person can be a feasible strategy for reducing emissions and costs, while also aiding in time management for participants. This approach has become even more relevant in the post-pandemic world, where remote work has become the norm.