1 Introduction

In the transition of our society towards lower energy consumption and higher sustainability, science has several roles and responsibilities. It provides the knowledge of our climate that is the basis for the Intergovernmental Panel on Climate Change (IPCC) reports, as well as the associated treaties, policies and action plans. Science also provides the input to industrial innovation to enable a circular economy and more sustainable products. At the same time, science is a domain that should become sustainable itself, just as any other domain. Scientists are also users of energy, sometimes heavy users, especially in the big science areas where large scale research infrastructures are used. Scientists have the responsibility not only to provide objective input to the climate debate, but also to operate their infrastructures in a sustainable way. An essential first step is the analysis of the carbon footprint of research infrastructures.

Aujoux et al. have presented the analysis of the carbon footprint of the astro-physical GRAND project [1]. An area where large scale research infrastructures are also used is astronomy. To get a better understanding of our early universe, astronomers need to observe objects and processes that are extremely faint and often billions of light years away. The only way to do this is by building telescopes that are extremely powerful. In the case of radio astronomy, this requires many antennas, often dishes or dipoles. Telescopes like the LOw Frequency ARray (LOFAR) and the Square Kilometre Array (SKA) are therefore extremely data-intensive [2]. Previous research has focused on the development of energy efficient technologies [3,4,5,6], sustainability of astronomical research infrastructures [7,8,9], institutes [10] or communities [11, 12]. This paper provides a detailed Life Cycle Analysis of the LOFAR from the perspective of its energy consumption and carbon footprint. LOFAR is a radio telescope with more than 100,000 antennas spread over 10 countries in Europe [13] and was designed by ASTRON, the Netherlands Institute for Radio Astronomy, part of the Dutch Research Council. We realize that environmental footprint is broader than the carbon footprint alone and should also include the usage of raw materials, water and land. To our knowledge, this is the first in-depth environmental impact analysis of a large-scale astronomy research infrastructure. The outcomes serve three main goals: 1) provide accountability and transparency to our scientific community and societal stakeholders, 2) serve as a baseline for actions to reduce the energy consumption and the carbon footprint, and 3) serve as a blueprint for the footprint analysis of other research infrastructures. This publication extends the work of Aujoux et al. [14].

The LOFAR system, its processes and its boundaries are defined in Section 2. The methodology to analyze the system is explained in Section 3. The outcomes are an approximation since they rely on benchmark data for conversion factors. Despite being approximative, results provide insight into the breakdown of the footprint and serve as an input for policies to reduce the footprint. Sections 4, 5, 6, 7 and 8 analyze the phases of the Life Cycle from initial development of LOFAR to its future end of life. Within each phase the analysis is structured according to the three scopes of the Green House Gas protocol [15]. Section 9 summarizes the carbon footprint and identifies the major contributions that can be influenced. The energy consumption is summarized in Section 10, including electricity, gas and fuel. Although scientific processing is outside the system boundaries, we discuss a number of representative science projects in Section 11 to get a better understanding of the ratio of energy consumption in pre- and post-processing.

2 System definition

LOFAR is a distributed radio telescope with 52 stations in several European countries (See Fig. 1). In 2022 the countries participating in LOFAR were The Netherlands (38 stations), Germany (6 stations), the United Kingdom (1 station), Sweden (1 station), France (1 station), Poland (3 stations), Ireland (1 station) and Latvia (1 station). Additional stations are planned in Bulgaria and Italy. Stations have two types of antennas, the Low Band Antenna (LBA) and the High Band Antenna (HBA). Figure 2 shows a schematic view of a LOFAR station. All stations have 96 LBA antennas. The HBA antennas are configured as tiles with 4x4 antennas. Dutch (NL) stations have 48 HBA tiles and International stations 96 HBA tiles. Coaxial cables are used to transmit analog signals to station electronics that are located in a shipping container (International stations) or cabinets (NL stations).

Fig. 1
figure 1

The sites of the International LOFAR Telescope

Full size image
Fig. 2
figure 2

Schematic view of a LOFAR station in The Netherlands with 96 LBA and 48 HBA tiles connected to a cabinet with station electronics

Full size image

After an initial stage of digital signal processing at the stations, the data is transported to the High Performance Computing (HPC) clusters COBALT2 and CEP4 at the Center for Information Technology (CIT) of the University of Groningen for further processing to standard data products. Three sites (SURF in The Netherlands, Forschungs Zentrum Jülich (FZJ) in Germany, and the Poznan Supercomputing and Networking Center in Poland (PSCN) provide data processing capabilities for post processing and the storage for the Long Term Archive (LTA). Scientists get access to the data products in the LTA for further processing towards science data products and publications.

The processes development, maintenance, operations, support and governance are staffed with employees of ASTRON and LOFAR partners. The manufacturing of components, transportation of components and decommissioning processes are outsourced to external parties. Excluded from this analysis is carbon footprint from manufacturers’ processes (the production of raw materials, shipping of raw materials to manufacturers, packaging of products, waste treatment) the development of the area around LOFAR to a natural reserve, scientific activities, and the waste treatment at the end of life.

The system used in this analysis includes the processing towards standard data products in the LTA, but excludes further post-processing by scientists. In Section 11, some typical cases for post-processing have been included. An overview of the LOFAR system from the perspective of the energy and sustainability analysis is provided in Fig. 4. It distinguishes between the 3 emission scopes defined in [15]. Scope 1 is the direct emission of green house gases controlled by the reporting organization. Scope 2 is the indirect emission from the purchasing of electricity for the own use of the reporting organization. All other emissions contribute to Scope 3, which is generally the biggest contributor of the three scopes to the overall carbon footprint. The Life Cycle and the processes for LOFAR are provided in Fig. 3. LOFAR is operated by ASTRON on behalf of the International LOFAR Telescope (ILT) consortium (Fig. 4). Each country is responsible for the site maintenance and the operation of the LOFAR station if used in local mode.

Fig. 3
figure 3

Processes in the LOFAR Life Cycle

Full size image
Fig. 4
figure 4

LOFAR system overview

Full size image

3 Methodology

The methodology in this paper does not follow a strict standard. A Life Cycle Analysis (LCA) is used and within the phases of the lifecycle, the contributions to the emission are structured according to the scopes of the Green House Gas protocol [15]. The following aspects are important:

  • The conversion factors used are benchmark data from public sources. No specific data is available on the carbon footprint of individual suppliers in the LOFAR supply chains. Some conversion factors are industry averages and these may substantially differ from the actual values for LOFAR. The conversion factors applied for the Initial Development Phase (2001 - 2010) are different from the Operational Phase (2010 - today). This is explained in more detail in Section 4.

  • Reference year. For the Operational Phase, 2020 is used as reference year except for commuting, air travel and initial development. Commuting post-corona is lower than pre-corona due to the adoption of hybrid working. Commuting during the Covid pandemic was very low. The commuting data from 2022 provides a representative data set for the entire Operational Phase. Air travel is used for governance processes and maintenance and was in 2022 at the same level as pre-corona. For the Initial Development Phase 2005 is used as a reference year. For air travel, 2022 is used as a reference year. It should also be noted that the realization of embodied footprint in, for instance, buildings and electronic products differs from the reference year used for the conversion factors.

  • Lifetime of assets. For the lifetime of components and buildings, a panel of ASTRON experts was used to assess the initial lifetime at design time, the expected lifetime, failure and replacement rates, the decommissioning information, and the differences between lifetime of individual components within an asset. This resulted in an average lifetime per component.

  • The embodied carbon of a component is attributed to an operational year by dividing the embodied footprint of that component by its lifetime.

  • International stations. The values for station electricity consumption, transport of LOFAR components to the site, and site preparation are well understood. For numbers relating to the footprint of the different organisations owning LOFAR stations across Europe, (buildings, commuting of personnel), data from ASTRON is extrapolated to other LOFAR partners. This is explained in Section 7.

  • Recycling is excluded from the carbon footprint calculations since the recovery of embodied carbon has not materialized yet.

  • Electricity. The two major accounting mechanisms applied worldwide are the market-based approach and the location-based approach 12.2. Both mechanisms have their pros and cons and debates are ongoing [16]. In this report, we use the location-based approach. The discussion section is used to analyze the difference with the market-based approach. Grid emission data is used from [17]. For the Netherlands, the location-based emission is 0.37434 kgCO2e/kWh. For the international stations, the average location-based emission is 0.36657 kgCO2e/kWh.

  • Life time. LOFAR operations started in 2010 and the science potential is huge, even in the era where SKA LOW will be operational. No decommissioning decisions have been planned or are being considered. The lifetime used is 25 years and is fictitious.

  • Trends. Calculating with a lifetime of 25 years, the Operational Phase of LOFAR is now half way. In the next decade, several conversion factors are expected to decrease. A first example is the location-based emission factors that will decrease due to the large scale investments in European renewable energy supplies. A second example is the strong increase of electric cars that will cause the conversion factors for commuting to decrease. A third example are trends in computing where the available processing per unit of energy is decreasing over time. The reference year in this analysis is halfway within the lifetime of LOFAR which is assumed to give a good approximation to the total lifetime footprint.

  • Data Availability. The data and the calculations used in this publication are available at [18].

The generic model for the life cycle of an arbitrary asset is also applicable to LOFAR. It starts with the Development Phase \(p=1\), followed by the Manufacturing Phase \(p=2\), the Deployment Phase \(p=3\). After the Operational (or Use) Phase \(p=4\), the lifetime ends in the End-of-Life (or Decommissioning) Phase \(p=5\). The life cycle is shown schematically in Fig. 3. The assembly, verification, technical and scientific commissioning of LOFAR are not described separately in this paper. They are activities in the phases described. In reality, the life cycle phases overlap, e.g., the Manufacturing Phase continues during the Operational Phase for the production of spares and replacement of electronics and clusters. For the goal of this analysis, this can be ignored. Emissions E are measured in units of kgCO2e or tCO2e. In all phases, the total emission is the sum of emissions in Scopes \(s=1\) to 3 as defined in [15]: Scope 1 emissions are direct emissions from owned or controlled sources. Scope 2 emissions are indirect emissions from the generation of purchased energy. Scope 3 emissions are all indirect emissions (not included in scope 2) that occur in the value chain of the reporting company, including both upstream and downstream emissions.

The total emission during the life time of LOFAR is the sum of the emissions in phases \(p=1\) to 5:

$$\begin{aligned} E_{tot} = \sum \limits _{p=1}^{5} E_{p} \end{aligned}$$
(1)

Emissions are classified as capital emissions C, measured in units kgCO2e or tCO2e, when it is an asset with a certain lifetime (e.g. buildings, electronics), or operational emissions O, measured in units kgCO2e/yr or tCO2e/yr, (e.g. gas, electricity).

Within each phase p and scope s, a specific capital emission \(C_{p,s,i}\) is calculated by multiplying the value of \(C_{p,s,i}\) in the applicable units (e.g. kg, m3, kWh) with the conversion factor \(F_i\) in the corresponding units of (e.g., kgCO2e/kg, kgCO2e/m3, kgCO2e/kWh).

In the development phase \(p=1\), the annual emission is the sum of capital emissions \(C_{1, s,i}/L_i\) and operational emissions \(O_{1, s,j}\) multiplied by a fraction \(f_1\) that corrects for the fact that not all of ASTRON’s staff was working on LOFAR development, see also Section 4. This annual emission is multiplied by the duration of the development phase, which is 10 years. The development phase includes prototyping activities.

$$\begin{aligned} E_{1} = 10f_1 \sum \limits _{s=1}^{3} \left( \sum \limits _{i=1}^{n} \frac{C_{1,s,i}{F_i}}{L_i} + \sum \limits _{j=1}^{m} O_{1,s,j}{F_j}\right) \end{aligned}$$
(2)

In the manufacturing phase, all the initial components were produced. They have a lifetime \(L_i\) smaller than 25 years, which is the fictitious lifetime of LOFAR. The ratio \(25/L_i\) is used to include the spares and replacements during the operational phase. In the manufacturing and the deployment phase, the total emission of components with a life time of \(L_i\) is

$$\begin{aligned} E_{(p=2,3)} = 25 \sum \limits _{s=1}^{3} \sum \limits _{i=1}^{n} \frac{C_{p,s,i}{F_i}}{L_i} \end{aligned}$$
(3)

For the operational phase \(p=4\), the data that is used to calculate emissions is only available from ASTRON and not from its partner organizations. The fraction \(f_4\) corrects for the fact that not all of ASTRON’s staff is working on LOFAR, see also Section 7. The emissions from the LOFAR partners are included by the factor \(g_i\). The emission from operations in the fictitious lifetime of 25 years is thus

$$\begin{aligned} E_{4} = 25 \sum \limits _{s=1}^{3} \left( f_4 \sum \limits _{i=1}^{n} g_{i} \frac{C_{4,s,i}{F_i}}{L_i} + \sum \limits _{j=1}^{m} O_{4,s,j}{F_j}\right) \end{aligned}$$
(4)

The decommissioning of LOFAR stations at the end of life is a relatively small activity which will not take more than a few weeks per station. In the decommissioning phase, the total emission is:

$$\begin{aligned} E_{5} = \sum \limits _{s=1}^{3} \sum \limits _{i=1}^{n} O_{5,s,i}{F_i} \end{aligned}$$
(5)

4 The initial development phase (2001 - 2010)

The development of LOFAR to its first deployment by ASTRON lasted from 2001 - 2010. The reports from 2004 - 2010 show a total number of person years invested in LOFAR of 470 full time equivalent (FTE). This is extrapolated to the full initial Development Phase resulting in 470 * 10 / 7 = 671 person years of initial development. The average number of people working at ASTRON during these years is 150, so that the fraction of personnel working for LOFAR \(f_1 = 0.4476\). This number will be used to attribute organizational footprint to LOFAR. The total Development Phase lasted 10 years and the lifetime of the assets from the Development Phase (designs, prototyping, testing, site selection etc) is 25 years. Test stations were in use between 2003 and 2009. The reference year used for heat and electricity consumption for this phase is 2005. The contributions from refrigerants over 2001 and 2010 are low and will be ignored.

4.1 Scope 1

Scope 1 consists of the consumption of natural gas for the heating of buildings and the use of fuel for traveling to the test sites.

4.1.1 Heating of buildings

The fraction of natural gas to be attributed to LOFAR to heat the building at the ASTRON Headquarters (hereafter referred to as ’Dwingeloo’) from 2001 - 2010. In the reference year 2005, the total natural gas consumption was 48,230 m3 gas. The emission factor for natural gas is 1.8 kgCO2e/m3 natural gas [19]. The attribution of carbon emission to LOFAR is thus (48,230)*1.8*0.4476 = 38,860 kgCO2e per development year. The total Development Phase lasted 10 years and the lifetime of the developed assets is 25 years. The attribution to LOFAR operations per year is therefore 38,860*10/25 = 15,544 kgCO2e per operational year.

4.1.2 Traveling to test sites

Fuel was used for site selection and traveling to test sites. Data is not available and the total distance is estimated. The total distance traveled by car for site selection, site contracts and measuring Radio Frequency Interference is estimated to be 5 days of traveling to 14 stations with a total distance of 5 * 1385 km = 6,925 km. Traveling to/from the test sites was done 80 times per year, with a distance of 100 km during 6 years. This results in 48,000 car km. The conversion factor is 0.132 kgCO2e/km [19]. The total emission is thus (48,000 + 6,925)*0,132=7,250 kgCO2e and 7,250/25 = 290 kgCO2e per operational year.

4.2 Scope 2

Scope 2 consists of the consumption of the fraction of electricity to be attributed to LOFAR for building operations in Dwingeloo from 2001 - 2010. In the reference year 2005, the total consumption was 130,849 kWh electricity. The location based emission factor for The Netherlands in 2005 is 0.494 kgCO2e/kWh [20]. The attribution of CO2 emission to LOFAR is thus (130,849)*0.494*0.4476= 28,933 kgCO2e per development year. The total Development Phase lasted 10 years and the lifetime of the developed assets is 25 years. The attribution to LOFAR operations per year is therefore 28,933*10/25 = 11,573 kgCO2e per operational year.

4.3 Scope 3

4.3.1 Embodied carbon of ASTRON’s buildings

The embodied carbon of ASTRON’s buildings is estimated using data from a benchmark study [21]. The average embodied carbon in buildings is 400 - 800 kg CO2e/m2 Here we take as a first estimate the mean of this range, 600 kgCO2e/m2 A more accurate value requires a detailed analysis of building materials and volumes. The floor area in Dwingeloo in 2005 was 5,953 m2. The lifetime of the buildings and installations used is 30 years. The fraction of total carbon footprint related to the buildings in the Development Phase is thus 5,953*600*10*0.4476/30= 532,935 kgCO2e. The attribution per operational year is 532,198/25 = 21,317 kgCO2e per operational year.

4.3.2 Commuting

In Section 7.3.2 the commuting is calculated in detail for 2022. We assume that the average commuting distance per day in 2005 was not different, but that the fraction of days working remotely was higher. Data from ASTRON’s Human Resources system show that pre-covid, the average fraction working from home was  20% whereas that is 50% in 2022. For the Development Phase we will use the value 10,350 km per FTE per year from Section 7.3.2 multiplied by 1.5 to correct for the difference in the remote working situation between 2005 and 2022. The average commuting distance per year in 2005 per full time equivalent employee is thus 1.5 * 10,350 = 15,525 km/year. Multiplying by the number of FTE working on LOFAR provides the total commuting distance per development year = 15,525 * 150 (employees)* 0.4476 (fraction working on LOFAR) = 1,040,175 km/year. Using the conversion factors from [19] and ‘fuel unknown’, and an average distance of 15 km/liter of fuel [22], this results in 0.193 * 1,042,393 = 201,182 kgCO2e per development year. The attribution to LOFAR operations is then 201,182*10/25 = 80,473 kgCO2e per operational year.

4.3.3 Testing and prototyping

Test stations were used between 2003 and 2009, consisting of ½LBA station plus two ¼LBA stations near the town of Exloo and a HBA station on the site of the Westerbork Synthesis Radio Telescope. This hardware is not reused during operations and will be accounted for as 1 NL LOFAR station. The carbon footprint of 1 NL station is 113,018 kgCO2e (see Section 5.5) and the attribution to an operations is 113,018/25=4,521 kgCO2e/year.

A broad range of equipment and test facilities was used in the development phase. Examples are spectrum analyzers, Electro Magnetic Compatibility (EMC) chambers, equipment for climate testing, cables, and production facilities for hardware prototyping. No data are available to calculate carbon footprint of this equipment. Most of this equipment has a lifetime more than 20 years and is also used for other purposes. The contribution to the total carbon footprint will be ignored.

4.4 The total carbon footprint of LOFAR initial development

The total footprint of the initial Development Phase \(E_1\) is thus 388,596 (Heating of ASTRON’s buildings, Scope 1) + 7,250 (Fuel) + 289,338 (Electricity for ASTRON’s buildings, Scope 2) + 532,935 (Embodied carbon of ASTRON’s buildings, Scope 3) + 2,011,818 (Commuting, Scope 3) + 113,018 (test station)= 3,342,956 kgCO2e. With a lifetime of 25 years, the contribution to the annual footprint of the Initial Development Phase = 133,718 kgCO2e per operational year.

5 The Manufacturing Phase

The manufacturing of parts was carried out by suppliers and contributes to Scope 3. No supply chain specific information is available on conversion factors. The conversion factors applied, are industry averages. These conversion factors are a lower limit, since they do not include the energy consumption and carbon footprint related to for instance the production of raw materials, packaging and shipping etc. The main components are the Low Band Antenna (LBA), the High Band Antenna (HBA), the coaxial cables, the station cabinets (Dutch stations) and shipping containers (International stations) for station electronics. The latter includes the switches and routers for the connections to the central processing facility.

5.1 Low Band Antenna

The carbon footprint is calculated using the Bill of Material (BOM) for the Latvian station which is representative for all stations in combination with [23]. The carbon footprint of printed circuit boards (PCB) is calculated using a conversion factor of 216 kgCO2e/kg for surface mounted PCBs [24]. The LBA antenna structure consists of weed blocks, ground plane, a polyvinylchloride (PVC) pole, the LBA antenna wires, the PCB, rubber bands and pegs. Each antenna is connected with 2 coaxial cables to the station cabinets or shipping container with standard cable lengths of 115 or 85 meters. The total carbon footprint per antenna is 75 kgCO2e (structure, lifetime = 20 years) + 24 kgCO2e (cabling, lifetime = 25 years) + 6.48 kgCO2e (PCB, lifetime = 20 year)= 105.48 kgCO2e/ antenna. The total carbon footprint associated with the manufacturing of LBAs is 623,826 kgCO2e.The annual contribution to the carbon footprint is 24,953 kgCO2e per year.

5.2 High Band Antenna

The carbon footprint is calculated using the Bill of Material (BOM) for the Latvia station in combination with [23]. The carbon footprint of printed circuit boards (PCB) is calculated using a conversion factor of 216 kgCO2e/kg electronics [24]. An HBA-tile consists of a bottom bag, a ground plane, styrofoam (Extruded Polystyrene, EPS), aluminum antennas, analog electronic boards (HBA front end, P-summator, C-summator), coaxial cables of standard length 4.7 meters, top bag and tile fixation. Each tile is connected with 2 coaxial cables to the station cabinet or shipping container with standard lengths of 115 or 85 meters. The embodied carbon footprint of manufacturing an HBA-tile is 636.25 (EPS, lifetime is 20 years) + 44.44 (coaxial cables, lifetime is 25 years) + 125,00 (top/bottom bags, groundplane, lifetime is 20 years) + 1,103,98 (PCB, lifetime is 20 years) = 1,782.20 kgCO2e. All HBA stations have a dummy tile in the center with EPS, top and bottom bags. A Dutch station consists of 48 HBA-tiles, an International station of 96 HBA-tiles. The total carbon footprint of manufacturing HBA-tiles is 7,063,254 kgCO2e. The attribution to an operational year is 282,530 kgCO2e/year.

5.3 Station electronics

The carbon footprint of printed circuit boards (PCB) is calculated using a conversion factor of 216 kgCO2e/kg electronics [24]. The station electronics in a shipping container/cabinet consist of the receiver units (RCU), processing boards (RSP), transient buffer boards (TBB), power supply (SPU), clock board (TDS), test board (JTB), and local control unit (LCU). The total weight of the PCBs is 124 kg (International station) and 67 kg (NL station). The total carbon footprint is therefore (67*38 + 124*14)*216 = 924.912 kgCO2e. The lifetime of the components is 12.5 years and the stations electronics of LOFAR are currently replaced in the LOFAR2.0 program. For the full life time of 25 years, the total footprint is thus 25 * 924,912 /12.5 = 1,848,385 kgCO2e. The annual contribution to the carbon footprint is 73,935 kgCO2e per year.

5.4 Station containers and cabinets

The 20-ft shipping containers used for International LOFAR stations are made of steel and have a standard weight of 2,200 kg [25]. The shipping containers are equipped with an EMC room with a weight of 2,000 kg. The carbon footprint of the shipping container is calculated using the conversion factor of 2.3 kgCO2e/kg [10]. The total embodied carbon footprint for an International station is 9,660 kgCO2e. Station cabinets for NL stations are made of steel and have a weight of 1,000 kg. Using the conversion factor of 2.3 kgCO2e/kg [23], the embodied carbon footprint of a station cabinet is 2,300 kg CO2e. The lifetime is 25 years. The total footprint from containers and cabinets is thus 9,660 * 14 + 2,300 * 38 = 222,640 kgCO2e. The annual contribution to the carbon footprint is 8,906 kgCO2e per year.

5.5 Computing clusters

The compute clusters for central processing have a lifetime of 5 years. No industry data are available on the embodied carbon footprint of the HPC clusters. In [26], the analysis for a DELL R470 server shows that the embodied footprint generated in the Manufacturing Phase is roughly equal to the operational footprint in the Operational Phase. Using the location based emission of 0.374 kgCO2e/kWh for The Netherlands [17], the total footprint of 1 computing cluster with an annual electricity consumption of 500,812 kWh/year is thus 5*500,812*0.37434=937,370 kgCO2e. Five generations of clusters are required for the lifetime of LOFAR so that the total carbon footprint of central computing is 5*937,370=4,686,850. The embodied footprint to be attributed to an operational year equals 4,686,850/25=187,474 kgCO2e/year.

5.6 The total carbon footprint of the Manufacturing Phase

The total footprint of the Manufacturing Phase \(E_2\) is thus 623,826 (LBA) + 7,063,254 (HBA) + 1,848,385 (station electronics) + 222,640 (containers and cabinets) + 4,686,849 (computing clusters) = 14,444,955 kgCO2e. Ignoring the carbon footprint from repairs and replacements, the initial carbon footprint of producing a NL station is 131,018 kgCO2e and an International Station is 218,148 kgCO2e. Including repairs and replacements, the contribution of the Manufacturing Phase to the annual operational footprint = 577,798 kgCO2e/year.

6 The Deployment Phase

The Deployment Phase consists of the transport of materials to the sites and the site preparation which contribute to Scope 3.

6.1 Transport to the sites

The transport of parts to an NL station requires 13 trucks and to an International station 26 trucks.The average single journey distance to an NL station is 250 km and to an international station 1,300 km. The emission factor for the truck transport used for LOFAR is 1.158 kgCO2e/km (data from supplier). The total emission related to truck transport of parts to stations is thus (38*2*250*13+14*2*1,300*26)*1.158 = 1,381,957 kgCO2e. These transports take place only once during the lifetime of LOFAR, so that the emission per operational year from truck transport is 55,278 kgCO2e.

6.2 Site preparation

LOFAR stations are built on a site that requires flattening and digging of trenches. This work is usually outsourced to local subcontractors and is roughly the same for Dutch and International stations. The main contribution to the carbon footprint is the usage of fuel by excavator (8 hours/station, 18 liters diesel/hour), tractor with dumper/scraper (32 hours/station 10 liters diesel/hour), mini crane (160 hours/station, 3 liters diesel/hour) [27]. This includes digging trenches for the network connections. The total consumption of diesel is 18*8+10*32+160*3= 944 liter diesel/ station. The station lifetime is 30 years. With a conversion factor of 3.473 kgCO2e/liter diesel [23], the total carbon footprint of site preparation is 944 * 3.473 = 3,279 kgCO2e/station and 52*3,279=170,483 kgCO2e for all the stations. For the sites a lifetime of 25 years is also applied so that the attribution to an operational year is 6,819 kgCO2e.

6.3 Buildings

All the signals from the 24 stations in the central part of LOFAR (the core), are transported via network connections to a small building, the concentrator node. This building has an area of 50 m\(^2\). The embodied carbon of buildings is estimated using data from a benchmark study [21]. The average embodied carbon in buildings is 400 - 800 kg CO2e/m2. Here we take as a first estimation the mean of this range, 600 kgCO2e/m2. The total carbon footprint of the concentrator node is therefore 600*50=30,000 kgCO2e. The concentrator node has a lifetime of 25 years. The annual contribution to an operational year is thus 30,000/25 = 1,200 kgCO2e/year. For the assembly in The Netherlands, floor space was rented from an industrial partner (10,000 m\(^2\) for 1 year) and additional space in barns near the LOFAR stations (10,000 m\(^2\) in total during 3 years).

For these buildings, we use the a lower value of 400 kgCO2e/m\(^2\) due to the type of building materials applied. The lifetime of these buildings is 30 years. The total carbon footprint for the assembly buildings is thus (10,000 + 3*10,000)*400/30=533,333 kgCO2e. Per operational year, 533,333/25=21,333 kgCO2e is attributed.

6.4 The total carbon footprint of the Deployment Phase

The total footprint of the Deployment Phase \(E_3\) is 1,381,957 (Transport) + 170,483 (Site preparation) + 30,000 (concentrator node) +533,333 (assembly buildings) = 2,115,773 kgCO2e. Per year of operations the attributed carbon footprint is 84,631 kgCO2e.

7 The Operational Phase (2010 - today)

After 2010, the Deployment Phase of LOFAR was finished and LOFAR was put in operations. From that moment the LOFAR activities comprise operations, maintenance, continued development, governance and science. The total ASTRON staff is 200 employees. On average, 30 employees are working on maintenance and operations of LOFAR (both in Dwingeloo and Westerbork). The staff working on the continued development and governance of LOFAR is estimated to be 50 employees. This equals 50% of the staff in the engineering and astronomy departments, therefore, 50% of the support staff is attributed to LOFAR as well. The fraction of ASTRON staff working directly and indirectly for LOFAR is therefore \(f_4=0.50\) (see Equation 4). This fraction is used to attribute organizational emissions to LOFAR in the Operational Phase. The scientific activities that do not contribute to the processes mentioned above are considered to be outside the system boundaries and are therefore excluded from the calculations.

The data required for the Operational Phase is not available for most of the International LOFAR partners. We assume here that per international station, 3 employees should be attributed to LOFAR, which is 42 employees in total, or \(g_{i}=1.42\) (see Equation 4). The exception is electricity for station operations for which \(g_{i}=2\) due to the double number of HBA tiles in an International station.

7.1 Scope 1

7.1.1 Heating of ASTRON buildings

The consumption of natural gas in 2021 = 56,448 m3 gas per year for heating of ASTRON in Dwingeloo and Westerbork. The conversion factor is 1.8 kgCO2e/m3 natural gas [19]. The total annual emission of carbon is thus 56,448*1.8*0.50*1.4286 = 72,141 kgCO2e/year.

7.1.2 Refrigerants leakage

The cooling machine in Dwingeloo was replaced in 2022 because the old cooling machine was leaking refrigerant. The refrigerant used until 2022 was R407c with a conversion factor of 1,624 kgCO2e/kg [19]. The new cooling machine uses R1234ze with a conversion factor of 1.0 kgCO2e/kg. The total leakage of R407c from 2010 to 2022 was fully dominated by the leakage in 2022 and was 32.42 kg. Converted to carbon emission, the total emission from 2010 - 2022 was 32.42 * 1,624 * 0.5 * 1.42 = 37,382 kgCO2e. For 2023 - 2035 we use the average emission from [28] for heat pumps which is 6% per year. The total mass is 69 kg refrigerant so that the expected leakage from 2023 - 2035 is 68 * 12 * 0.06 * 0.5 * 1.0 * 1.42 = 35 kgCO2e. The total attribution of refrigerant leakage to LOFAR operations is thus 37,382+35 = 37,416 or 1,497 per operational year.

7.1.3 Fuel

ASTRON uses company cars for maintenance of the national and international stations. The total fuel consumption is 36,000 km, using diesel of which 90% is attributed to LOFAR. The conversion factor is 0.132 kgCO2e/km [19]. The emission attributed to LOFAR and related to fuel is thus 36,000*0.90*0.132*1.42 = 6,073 kgCO2e/year.

7.2 Scope 2

7.2.1 Electricity for buildings

The electricity consumption in Dwingeloo = 1,600 MWh/year and in Westerbork =564 MWh/year. The fraction we adopt to attribute to LOFAR is 0.50. Including the factor \(g=1.4286\) to account for the fraction of electricity used by buildings of international partners, the total electricity consumption is 1,536.4 MWh/year. For all locations we use location based emissions [17] so that the total carbon emission related to Electricity for buildings is 575,151 kgCO2e/year.

7.2.2 Station operations

The average electricity consumption of an NL stations is 60 kWh/day and for International LOFAR stations is 120 kWh/day. The total electricity usage of the stations is (60*38 + 120*14)*365= 1,445.4 MWh/year. Using the grid emission factors of [17], the carbon emission is 536,297 kgCO2/year.

Table 1 Electricity consumption and carbon emission related to the central processing of LOFAR
Full size table

7.2.3 Electricity for Central Processing and Storage

The electricity consumption at CIT for central processing is 500,812 kWh/year (Table 1). The grid emission factor in The Netherlands is 0.371 [17]. The carbon footprint of central computing is thus 500,812*0.374=187,474 kgCO2e/year.

Table 2 Carbon emission related to tape storage of the LOFAR LTA
Full size table

The total volume of data in the archive is 29.376 (SURF) + 18.035 (FZJ) + 9.642 (PSCN) = 57.1 PByte on tape and 1.178 (SURF) + 1.242 (FZJ) + 0.391 (PSCN) = 2.811 PByte on disk (Tables 2 and 3). No details are available on the electricity consumption of the 3 sites (SURF, FZ Jülich, Poznan). According to [29] the energy consumption for storage is 11,900 kWh/year/PByte for disk and 1,530 kWh/year/PByte for tape. These industry averages are applied for the storage of the LOFAR Long Term Archive. The energy consumption of LOFAR data storage is therefore 57.1*1,530 kWh/year + 2.811*11,900 kWh/year = 120,742 kWh/year.

The total carbon emission per year from processing and storage is therefore 187,474 (processing) + 37,376 (tape storage) + 13,783 (disk storage) = 238,632 kgCO2e/year.

7.3 Scope 3

7.3.1 Embodied carbon of buildings

The embodied carbon of ASTRON’s buildings is estimated using data from a benchmark study [21]. The average embodied carbon in buildings is 400 - 800 kg CO2e/m2. Here we take as a first estimation the mean of this range, 600 kgCO2e/m2. A more accurate value requires a detailed analysis of building materials and volumes. The total floor area is 9,741 m2 in 2021. The embodied CO2 in the buildings is 9,741 * 600 kg/m2 = 5,844,600 kgCO2e. Buildings are used 30 years and the factors \(f_4=0.5\) to calculate the fraction of ASTRON’s emission to attribute to LOFAR. The factor \(g=1.4286\) is applied to attribute the contribution of the embodied emission from buildings of international partners. The total embodied carbon attributed to LOFAR per year = 5,844,600*0.5*1.42/30 = 138,322 kgCO2e/year.

7.3.2 Commuting

The contribution of commuting of staff is calculated using an anonymized overview of the personnel system and taking into account the type of traffic (foot, bicycle, public transport, car, carpooling) and the number of days employees work at the office and the fraction \(f_4=0.5\). This results in a total distance of 1,035,000 km/year fossil fuel based commuting distance to be attributed to LOFAR. Using the conversion factors from [19] and ‘fuel unknown’, an average consumption of 15 km/liter of fuel [22], and using \(g=1.4286\) this results in 0.193 * 1,035,000 * 1.42 = 283,652 kgCO2e/year.

Table 3 Carbon emission related to disk storage of the LOFAR LTA
Full size table

7.3.3 Air travel

The organization of ILT requires on average 9 European flights and 1 intercontinental flight per year. To calculate the CO2 emission, the emission associated with a return flight Amsterdam-Rome is used (450 kgCO2e). For the intercontinental travel, Amsterdam - Calgary (2,250 kgCO2e) is used as reference [30]. For maintenance of international stations, flights Amsterdam - Dublin (260 kgCO2e) and Amsterdam - Riga (460 kgCO2e) are used. The total emission related to international flights is 7,020 kg CO2e. Staff from international partners use flights e.g. for the ILT LOFAR Board meetings. The factor g=1.42 is used here as well, resulting in a total emission related to air travel of 9,968 kgCO2e/year.

7.3.4 Site maintenance

During the Operational Phase, mowing of the site is the main activity contributing to the carbon footprint. NL stations require 5 days of mowing per year. The fuel consumption is 330 liters of HVO Biodiesel per station per year. The emission factor of HVO Biodiesel is 0.314 kg CO2e/liter [19]. The emission during the Operational Phase for an NL station is 104 kg CO2e/year. This number is applied to the International LOFAR stations as well, resulting in a carbon emission of 52*104 = 5,388 kgCO2e/year.

7.4 The total carbon footprint of the Operational Phase

The total footprint of the Operational Phase \(E_4\) is 79,710 (Scope1) + 1,350,080 (Scope 2) + 437,331 (Scope 3) = 1,867,120 kgCO2e per year.

8 End of Life Phase

LOFAR is a world-class research infrastructure for the coming decades. At some point in time, LOFAR will be taken out of operations after which all the sites need decommissioning and all the components on the site removed and where possible recycled. Consistent with Section 4, the lifetime for LOFAR used here is 25 years.

8.1 Site decommissioning

The future decommissioning of LOFAR involves returning the sites to their original states. This implies the removal and recycling of all material. The recycling of materials is covered in Section 8.2. The energy consumption of the decommissioning is much lower than the site preparation, since no requirements on flattening etc are needed. It is assumed that 1 day of excavator + 1 day of tractor/scraper = 18*8 + 10*8=224 liters diesel/station kgCO2e/station. With a conversion factor of 3.473 kgCO2e/liter diesel [19], the total carbon footprint of site decommissioning is 224 * 3.473 = 778 kgCO2e/station and (38+14)*778 = 40,456 kgCO2e in total. The attribution of the total carbon footprint of decommissioning LOFAR to operations is 40,456/25 = 1,618 kgCO2e per year.

8.2 Transport of waste

After the decommissioning of the sites, all materials have to be transported to a recycling organization. It is assumed that this is done locally. The distance from any LOFAR station to a local recycling company is< 200 km. The transport is dominated by the EPS, so that the same number of trucks will be needed as in the deployment phase. The total distance is therefore (38*13+14*26)*2*200=343,200 km and with a conversion factor of 1,158 kgCO2e/km, the emission of the transport is 397,426 kgCO2e in total, and 15,897 kgCO2e/operational year.

8.3 Avoided future greenhouse gas emissions due to recycling

A full analysis of avoided future emission of green house gases due to recycling requires an analysis of the recycling rates, and the carbon footprint of the waste treatment and the recycling processes. This data is currently not available and this area is considerd future research.

8.4 The total carbon footprint of the End Of Life Phase

The total footprint of the End of Life Phase is 40,456 (site decommissioning) + 397,426 (transport of waste) = 437,882 kgCO2e.

9 The total carbon footprint of LOFAR

The carbon footprint per phase of the Life Cycle is shown in Table 4. All numbers are rounded to tCO2e. The total life cycle carbon footprint of LOFAR is 67,020 tCO2e and 2,681 tCO2e per operational year.

The Operational Phase contributes 70% of the total carbon footprint. Of the total operational footprint, 43% is directly related to the operations of the facility. The other 57% is related to buildings and commuting of staff and attributed to LOFAR.

Scope 1 (direct emissions from LOFAR) contributes 3.5%, scope 2 (electricity) contributes 48.9% and scope 3 (other indirect emissions) contributes 47.6%. Electricity for storage contributes 2% and is included in Scope 3.

The initial investment in LOFAR was 200 Meuro and the annual operations budget is 10 Meuro so that the total life cycle budget is 200 + 25*10 = 450 Meuro. The carbon footprint per euro of budget is therefore 65,594 tCO2e/450 Meuro = 145 tCO2e/Meuro. The construction footprint is 19,904 tCO2e and 99,5 tCO2e/Meuro of the initial investment. The operational footprint is 187 tCO2e/Meuro operations budget.

Table 4 Carbon emission per phase of the life cycle of LOFAR
Full size table

10 The energy consumption of LOFAR

The three main contributions to the energy consumption of LOFAR consist of Electricity, ’Fossil’ (Natural) Gas and Fuel. Table 5 shows the quantity per phase in the life cycle and the total consumption over the life time of LOFAR. The energy consumption by subcontractors for the production of assets has not been included. The consumption of natural gas is fully related to the heating of buildings. For the buildings in Dwingeloo, Aquifer Thermal Energy Storage (ATES) is already in use and the dependency on natural gas will be reduced over time. The use of fuel is dominated by the commuting. Trends and legislation related to the automotive sector in Europe are expected to reduce the usage of fuel over time. The electricity consumption is an important component to investigate. LOFAR2.0 is the major upgrade program for the station electronics and the programme uses operational requirements to restrict the power consumption after the upgrade. More opportunities for the reduction of power are investigated.

Table 5 Energy consumption per phase of the life cycle of LOFAR
Full size table

11 Science and post-processing

LOFAR realizes annually 6,100 hours on sky. Per observation hour 594 kWh, 8.0 m3 natural gas and 22.2 liters of fuel are required and 430 kgCO2e is emitted on average. The range of LOFAR observations is extremely large due to its extreme flexibility and some observations are executed commensally with other observations. The table below shows 6 example science projects together with the post-processing required. The energy for post-processing strongly depends on the type of observation. For pulsar timing, the post processing is 0.02%, for a standard imaging observation it is 3% [31, 32] for a wide field long baseline imaging observation [33], it is 40% of the total electricity. The power consumption from a typical system = 12.5 W/core-hour (Table 6).

Table 6 Electricity consumption and carbon footprint of typical LOFAR observations including post processing
Full size table

12 Discussion

12.1 Accuracy of the results

A sensitivity analysis was done on the assumptions in relation to the overall outcome.

  • For the conversion factors related to manufacturing, an error value of 25% is applied to account for differences between the industry averages and the specific LOFAR supply chains.

  • Using the value range in [21], the error for the conversion of embodied carbon in buildings is set to 25%.

  • The error in the embodied carbon for clusters is set to 50% due to the fact that we extrapolate data from CPU-based clusters with both GPU’s and CPU’s.

  • The error on the factor of 0.44 applied to account for the footprint of international LOFAR partners is set to 25%

Using these values, the overall error on the carbon footprint is estimated to be 25%. The additional error due to applying a reference year in the middle of the LOFAR lifetime reference year to the entire lifetime is difficult to assess.

12.2 Location-based versus Market-based reporting for Electricity

In the Section 2 the difference between location-based reporting and market-based reporting was mentioned. The data reported here are all location-based. For the Dutch stations and the central processing, green electricity has been purchased with European Guarantees of Origin. This information is not for all the international partners available. Applying the market-based approach to the Dutch stations, buildings and central processing only would lead to a decrease of 40% of the total footprint.

12.3 Policy aspects

The outcome of the analysis is for any research facility extremely useful as input to various policy areas:

  • Incorporating sustainability requirements in the design of research instruments and infrastructures will lower the footprint. These requirements can relate to energy efficiency requirements, lifetime requirements of components to reduce the replacement rate and end of life requirements (cradle to cradle).

  • Procurement processes can take sustainability criteria in the evaluation of contractors in tender procedures. Reporting by contractors on supply chain specific carbon footprint can help to improve the accuracy of the carbon footprint analysis. For electricity contracts, the Guarantees of Origin are most valuable if they ensure a direct coupling between the certificates and the increase in total renewable energy supply.

  • Operational policies related for instance to the upgrading of components. A long lifetime of components enables maximum usage of the embodied carbon whereas the upgrading of equipment often results in a lower operational footprint. An example where this applies is computing where additional operational risk occurs when a component is operated beyond the period where vendor support is offered.

  • Organizations operating research facilities may consider to use carbon offset plans to compensate for the carbon footprint, but care should be taken since carbon offset plans vary strongly in cost and impact, see also [34].

  • Facilities can communicate the carbon footprint of an observation to the principal investigators. This may help to increase the awareness on the environmental impact of science. It may also help PI’s to design their research with minimal impact.

  • Science that reuses archive data has a much lower footprint. Research facilities can set up policies to stimulate the reuse of archive data.

12.4 Comparison with other publications

Knödlseder et al. have recently investigated the carbon footprint of astronomical research infrastructures using an economic input-output (EIO) analysis [9]. Knödlseder et al. use as average value 240 tCO2e/Meuro for the carbon emission associated with the construction of ground-based astronomical observatories. In Section 9 it was shown that this value for LOFAR is 99,5 tCO2e/Meuro investment. The lower value than used in [9] can be explained by the fact that some processes were excluded from the analysis (see Section 2.

Knödlseder et al. use as average value 250 tCO2e/Meuro for the carbon emission associated with the operations of ground-based astronomical observatories. The operations budget for LOFAR consists of a budget managed by the central ILT organization and budgets from partners. The operations budget of ILT is 5 Meuro/year which covers the operations, maintenance and support staff and the cost for central computing and storage. The partners each cover in addition the local cost of electricity and site maintenance and staff. Budgets for continued development such as LOFAR2.0 [35] are managed separately. The total operations budget is estimated to be 10 Meuro. A more accurate number would require an in depth analysis of the financial accounts of all the partners which is not feasible. Using the total operational carbon footprint per year (1,866 tCO2e/year) results in 187 tCO2e/Meuro operations (location-based approach) which is quite close to the number of 250 tCO2e/Meuro operations.

12.5 Comparison with societal numbers

The total annual carbon footprint of LOFAR of 2,624 tCO2e/year equals the emission of 386 EU citizens (location-based approach), who emit on average 6,800 kgCO2e/year/capita [36]. The annual electricity consumption of LOFAR 3,603 MWh equals the consumption of 2,258 EU citizens who consume 1,596 kWh/year/capita [37].

13 Conclusion

Providing transparency on the environmental impact of research infrastructures is important since they are accountable to public funding agencies. This paper reports a baseline for the carbon footprint and the energy consumption of a European research infrastructure LOFAR. The carbon footprint of LOFAR per year is 2,624 tCO2e using a Life Cycle Analysis. This includes the attribution of the carbon footprint from development, manufacturing, deployment and the end of life. The accuracy of this analysis is determined by the usage of industry averages for conversion factors. The annual energy consumption is dominated by the total electricity consumption of 3,603 MWh/year.

This analysis can serve as input to the strategy and policy of astronomical facilities to quantify and reduce their impact. The methodology can serve as a blue print for the analysis The accuracy of such an analysis can be improved if supply chain specific data can be used for conversion factors instead of benchmark data from industry.