1 Introduction

Although maritime transport is the most sustainable transport mode, emissions from the maritime transport sector account for a significant portion of total emissions, affecting air quality and contributing to climate change. Thus, in recent years, public concerns regarding the environmental impacts of maritime transport have increased.

International shipping was estimated to have emitted 870 million tons of CO2 in 2007 (no more than 2.7% of the global total of that year) and 949 million tons of CO2 and 972 million tons of CO2e greenhouse gases (GHG), combining CO2, CH4 and N2O, in 2012.

A multi-year average estimate for all shipping, using bottom-up totals for 2007–2012, was 1016 million tons of CO2, which accounted for approximately 3.1% of annual global CO2, 20.9 million tons of NOx (as NO2) and 11.3 million tons of SOx (as SO2) (IMO 2014).

In the context of port-city areas, emissions released by vessels operating in port negatively affects local communities, albeit with a small percentage compared to the total amount released by shipping (Dalsoren et al. 2009). Nevertheless, it inevitably constitutes a source of pollution concentration in the air and has a significant environmental impact on the coastal communities, as 70% of the ship emissions occur within 400 km of land (Eyring et al. 2005).

Moreover, the urban character of some ports and their populated surroundings are the main focus of the negative effects of exhaust pollutants (NOX, SOX, VOC, CO and PM) due to the associated local impacts on human health. Of particular importance to the human health in urbanised ports is that around 95% of the ship-generated total PM is of an aerodynamic diameter of less than 2.5 μm (PM2.5) (Whall et al. 2007). Thus, the need to control air pollution at ports is widely acknowledged as an active policy issue by various authoritative port associations (IAPH 2007; ESPO 2003) as a reaction of main regulations (IMO, EC, EPA etc.), which are indicated in Table 1.

Table 1 Main regulations for the prevention of air pollution from ships

As a consequence, relatively recently ports in North America (Los Angeles-Long Beach, Seattle, Vancouver, New York etc.) and Europe (Venice, Barcelona, Gothenburg, Antwerp etc.) have started to introduce specific measures and policies to directly address GHG emissions (through the reduced use of conventional fuel) and, indirectly, to control local air pollutants since a significant share of emissions are derived from the time the vessels remain in port (Gibbs et al. 2014). Most of the measures are related to the introduction of LNG bunkering infrastructure, cold ironing, the provision of shore-side electricity at berth or by defining incentives for fuel switching or green ships (Merk 2014).

A fundamental requirement for emission control, assessing the impacts of growing shipping activity and planning mitigation strategies is developing accurate emission inventories for ports (ICF 2006). Furthermore, as stated in Tzannatos (2010a), port emission inventories would aid policy makers in developing effective regulatory requirements or port environmental management systems. In such a context, in the port of Naples, two experimental campaigns were carried out in 2012 to investigate the air quality (sulphur dioxide, nitrogen dioxide and benzene, ethylbenzene, toluene and xylene) and to compare the observed concentration values with limits established by European legislation (Prati et al. 2015).

With regard to emissions in urban ports, the growth of cruise activities should be underlined since cruise shipping is a relatively large emitter, due to large hoteling load and extended turnaround times, which sometimes exceed 48 h (home ports). As an example, the cruise activity in the five busiest Greek ports contributed 6.2 and 3.1%, respectively, to the relevant national NOX and SO2 inventory (Maragkogianni and Papaefthimiou 2015).

In 2014, the cruise industry met a demand of more than 21 million global passengers through the supply of 296 cruise vessels and a total of 500,854 berths, mainly concentrated in America (Caribbean and North America) and Europe (Mediterranean and North Europe). Looking at long-term projections, the cruise industry is expected to exceed 25 million cruise passengers in 2018 and 30 million in 2030 (Pallis 2015); therefore, main cruise ports have recognised the need to reduce emissions from the cruise industry, mainly in cruise terminals (e.g. Venice and Barcelona) that are close to city centres and where the exposure of the population will be high.

In such a context, the goal of this paper is to develop accurate emission inventories (CO2, SOX, NOX and PM) and emission indicators for cruise ports by estimating, firstly, the fuel consumed by each vessel on the basis of its activities in port. By integrating the evaluation over time (i.e. 1 year) and over the fleet that calls at a specific port, a yearly inventory can be achieved. On the other hand, the development of emission indicators will facilitate reliably estimating the emission inventories of cruise ports at the port level. Indeed, this information is essential to properly assess the impacts of strategies for regulating and controlling air emissions from vessels at ports.

The paper is organised as follows: Section 2 reviews relevant literature on the issue. Section 3 introduces the methodological approach and the formula used to estimate inventory emissions. Section 4 introduces the data used for the particular case study and the main results. Section 5 presents the most relevant emission indicators. Finally, Section 6 highlights the main conclusions.

2 Literature review

According to published research, which incorporates extensive reviews of ship emission estimation methodologies (Miola et al. 2010; Tichavska and Tovar 2015), two different approaches can be used to estimate atmospheric emissions arising from maritime transport: top-down and bottom-up approaches.

The top-down approach calculates emissions without considering the characteristics of the individual vessels, which are instead spatially assigned later. The bottom-up approach evaluates the individual pollution emitted by a single vessel in a specific location and then, by integrating the evaluation over time and over the fleet, obtains the total emissions. In addition, as it is stated in Miola et al. (2010), a combination of bottom-up and top-down approaches in the evaluation of total emissions is possible if geographical factors are considered. Thus, two factors must be considered in order to evaluate atmospheric emissions: the quantity of emissions produced and where they are emitted.

2.1 Emission inventories at global, regional and port-level

With regard to the state-of-art, a wide variety of studies relates to emission inventories at global or regional levels but only a few do so at the port-level (local approach). The most relevant studies at global or regional level are Endresen et al. (2003, 2004, 2007), Corbett and Kohler (2003), Eyring et al. (2005), Corbett et al. (2007), Wang et al. (2007) and IMO (2009), whose estimations where based on fuel sales statistics. These studies reported average CO2 emissions, as well as upper and lower levels and the important uncertainties between them were quantified (Miola et al. 2010). In addition, the study of Moreno-Gutiérrez et al. (2015) should be highlighted since it compares several different methods of estimating emissions and fuel consumption and makes a comparative analysis between the main papers and reports published in areas of the EU and the USA.

On the other hand, methodologies to evaluate emissions due to port activity, which sometimes are included in city inventories, have increasingly become an important research topic over the last two decades and the number of scientific studies addressing this concept has broadly increased. The representative approach for emission estimation in port studies was the bottom-up approach, based on port calls and estimated vessels operating at a port (Tichavska and Tovar 2015). Furthermore, normally, activity-based and/or fuel-based estimations were made since they are more accurate than top-down methodologies that require detailed data such as routing, engine workload, ship speed, location and duration (Song 2014).

For instance, the study conducted by Saxe and Larsen (2004) analysed the urban dispersion of air pollutants (nitrogen oxides) originating from ships in three Danish Ports using an operational air quality model. De Meyer et al. (2008) gave a better insight to emission inventories on a national scale (Belgian seaports) by using a bottom-up activity-based model. Tzannatos (2010a, b) addressed the issue of air pollution generated by passenger shipping alone at the port of Piraeus. He developed an in-port ship activity-based methodology that was applied for manoeuvring and berthing operations in order to estimate the main vessel exhaust pollutants (NOX, SO2 and PM2.5) over a 12-month period in 2008–2009.

Then, Berechman and Tseng (2012) estimated the emission costs of ships and trucks in the Port of Kaohsiung (Taiwan) by calculating the time spent at berth, the mean load on the auxiliary engines, the load factor and the emission factors of auxiliary engines for each pollutant. Villalba and Gemechu (2011) used the same methodology to calculate GHG emissions (CO2 equivalents) in the Port of Barcelona. In particular, they accounted for the emissions due to electricity and fuel consumption in the port area. McArthur and Osland (2013) also estimated the emissions from ships hoteling in the Port of Bergen and placed monetary value on these emissions, whereas Song (2014) estimated both the in-port ship emissions inventory and the emission-associated social costs in Yangshan port of Shanghai for the entire fleet. In that case, a methodology, supported by ship-by-ship and real-time data from the modern automatic identification system (AIS), was developed to obtain accurate results.

Similarly, Ng et al. (2012) used AIS data to determine typical main engine load factors through vessel speed and operation mode characterisation for emission inventories of ocean-going vessels in the port of Hong Kong. Finally, a study by Tichavska and Tovar (2015) presented vessel emissions in the port of Las Palmas by developing a full bottom-up model and data transmitted by the AIS in 2011.

2.2 Cruise ship emission inventories at port-level

With regard to cruise ship emissions at ports, Maragkogianni and Papaefthimiou (2015) presented a “bottom-up” estimation based on the detailed individual activities of cruise ships in the Greek ports of Piraeus, Mykonos, Santorini, Katakolo and Corfu. For each studied port and for all approaching cruise vessels, they registered ship movements during manoeuvring and berth operations, engine types and sizes, load factors, the type of fuel consumed and the time spent in each mode. For each ship call, the air pollutants (NOX, SO2 and PM2.5) produced during the ship’s activity in the port was estimated. They stated that emissions during hoteling accounted for 88.5% of the total emissions and highlighted the seasonality effect as summer emissions and associated impacts were significantly amplified.

In addition, Dragovic et al. (2015) estimated ship exhaust emission inventories and their externalities in the Adriatic ports of Dubrovnik (Croatia) and Kotor (Montenegro) for the period 2012–2014. The methodology for emission estimation relied on the distinction of various activity phases (manoeuvring and berth/anchorage) performed by each cruise ship call (bottom-up) as a function of energy consumption during each activity multiplied by an emission factor. The results showed that the application of ship activity-based methodology improves the understanding of ship emissions in ports and contributes toward the implementation of effective port policies to control air quality.

The present paper proposes a methodology based on the full bottom-up approach and begins by evaluating the fuel consumed by each vessel on the basis of its individual port activities (manoeuvring, berthing and hoteling) and differentiating between the main vessel propulsion, auxiliary propulsion (thrusters), boilers and electrical generators. Unlike previous studies, this paper also provides accurate cruise ship emission indicators (rates per hour, per passenger, per GT or a combination of all three), which can be used by other researches and stakeholders to reliably and quickly estimate emission inventories in other cruise ports at the port level.

3 Evaluating emissions from cruise ships

According to the literature review, the first step in the evaluation of emissions is the estimation of the fuel consumed by each vessel (or fleet) on the basis of its activities. Specific fuel oil consumption (measured in g/kWh) is therefore an important input to the appraisal. Once the fuel consumption is calculated, it is possible to use emission factors to estimate the emission of different pollutants.

This paper considers, in general terms, the full bottom-up approach but takes into account separately the fuel consumption and emissions of the following propulsion systems of cruise vessels during port operations:

  • Cruise vessel engines: modern ships use diesel, diesel-electric engines or gas turbines as a source of power for propulsion (main propulsion)

  • Transversal propulsion (thrusters) for berthing and unberthing operations (auxiliary propulsion)

  • Boilers for steam production used to heat up heavy fuel oil (HFO) fuel and modify its viscosity and for heating up water

  • Auxiliary engine generators for providing electrical energy used during hoteling

Then, for every vessel call, the fuel consumption (based on the power consumed) and corresponding emissions will be estimated for the following: (a) incoming manoeuvring from the Landfall Buoy to the cruise terminal dock, (b) berthing approach, (c) stay at the cruise terminal dock (port time), (d) unberthing operations and (e) outgoing manoeuvring from the cruise terminal dock to the Landfall Buoy.

3.1 Methodological approach

3.1.1 Propulsion power consumption for incoming/outgoing manoeuvring

The Admiralty coefficient method is proposed for estimating the propulsion power for manoeuvring, which is based on the basic assumption that the all resistance is frictional and that the power varies as the cube of the speed. This method, which determines the required propulsion power according to the given ship speed and the displacement, has been used by several authors, such as Tupper (2013), Watson (1998), Taylor (1996) and Schneekluth and Bertram (1998) because of the advantages of the practicality of this methodology.

In this context, the estimation of the fuel consumption for manoeuvring is calculated as follows:

$$ {C}_P=\sum_{\mathrm{ij}}\left({P}_{B_{\mathrm{ij}}}{t}_{\mathrm{ij}}\right){c}_e $$
(1)

where C P denotes the amount of fuel consumed by the main propulsion of the vessel moving (tones), i represents those sections in which the travel distance between the dock and the Landfall Buoy is divided and velocity data is registered, j is the vessel’s activity stage (incoming/outgoing manoeuvring), t ij is the time (h) the vessel spends moving within the port, c e is the specific fuel oil consumption (g/kWh) and \( {P}_{B_{\mathrm{ij}}} \) is the propulsion power required (kWh) during manoeuvring, which is calculated according to Eq. (2):

$$ {P}_{B_{\mathrm{ij}}}=\frac{{\varDelta_{\mathrm{ij}}}^{2/3}{V}_{\mathrm{ij}}^3}{c_a} $$
(2)

where ∆ ij is the real vessel displacement, V ij is the vessel speed (nm) and c a is the Admiralty coefficient, which is related to the vessel’s resistance, that is:

$$ {c}_a=\frac{\varDelta^{2/3}{V}^3}{P} $$
(3)

in which ∆ is the vessel’s displacement related to the propulsion power at maximum speed, V is the maximum vessel speed and P the effective energy power (kW). For diesel-electric engines, P e is equivalent to the electric power engine, and for diesel engines, the effective energy power is equal to the maximum propulsion power.

3.1.2 Hoteling consumption

Following the methodological approach, the fuel consumption for hoteling (C H ) during port time at the cruise terminal and during manoeuvring is estimated as follows:

$$ {C}_H={c}_e\left({P}_H{t}_d+{P}_{H^{*}}{t}_{\mathrm{ij}}\right) $$
(5)

where t d is the dwell time at the terminal dock, P H is the hoteling power (kW) and \( {P}_{H^{*}} \) is the hoteling power developed when the vessel is moving.

3.1.3 Thrusters consumption for berthing/unberthing operations

The fuel consumption required for a cruise vessel to manoeuvre around can be estimated as follows:

$$ {C}_T=\sum_{\mathrm{jk}}\left({n}_k{P}_k{c}_e\right)\left({t}_{l_{\mathrm{kj}}}{r}_{l_{\mathrm{kj}}}+{t}_{e_{\mathrm{kj}}}{r}_{e_{\mathrm{kj}}}\right) $$
(6)

where c T is the fuel oil consumption of the thrusters (kg/h), k is the type of thruster propeller (stern and bow), n k is the number of propellers, \( {t}_{l_{\mathrm{kj}}} \) is the time that each type of propeller is working on load, \( {t}_{e_{kj}} \) is the time that each type of propeller is working empty, \( {r}_{l_{\mathrm{kj}}} \) is the ratio (%) corresponding to the load factor and \( {r}_{e_{\mathrm{kj}}} \) is the ratio (%) corresponding to the empty factor.

3.1.4 Boiler consumption

Finally, the fuel consumption provided to the boilers will be estimated as follows:

$$ {C}_B=\left(\sum_{\mathrm{ij}}{t}_{\mathrm{ij}}+{t}_d\right){c}_B $$
(7)

where c B is the fuel oil consumption of the boiler (kg/h). In this paper, this parameter is obtained through a survey completed by ship owners. In particular, it is usually registered in the “Engine Room Log Book”.

3.1.5 Total fuel consumption

Once the individual fuel consumption is estimated, the next step is to quantify vessel emissions per air pollutant by multiplying fuel consumption and emission factors (g/Kwh), that is:

$$ {E}_z=\left({C}_P+{C}_H+{C}_T+{C}_B\right){\mathrm{EF}}_z $$
(8)

where z is the type of air pollutant.

Combustion emission factors (EF) vary by the following: engine type (main and auxiliary engines, auxiliary boilers); engine rating (SSD, MSD, HSD); whether engines are pre-IMO Tier 1, or meet IMO Tier I or II requirements; the type of service in which they operate (propulsion or auxiliary); type of fuel (HFO, MDO, MGO and LNG) etc.

Therefore, a differentiation is made between those emissions that only depend on the fuel consumption and those that depend on the previous engine properties. Table 2 shows the main details and data sources.

Table 2 Emission factors (EF) in terms of grams of fuel consumed per air pollutant. Sources: IMO (2009), IMO (2014), ENTEC (2002) and EIAPP certificates from ship owners

In summary, Fig. 1 shows the methodological framework considered in this paper, in which steps 1 and 2 are related to the input data model and steps 3 to 6 are methodological aspects that are described in Section 3.1.

Fig. 1
figure 1

Methodological scheme to estimate air pollutant emissions

4 Inventory of cruise vessels

In this section, emission inventory values (i.e. CO2, SOX, NOX and PM) for cruise vessels in the Port of Barcelona are presented.

4.1 Data samples

The data sample for this particular study comprises 30 cruise vessels that were monitored during 2015. According to the statistics of the Port of Barcelona, those 30 vessels accounted for more than 520 calls which represents about 70% of total cruise vessel calls in 2015 (the total number of cruise calls was 749). This statement denotes the selection of the data set is suitable because of their relevant significance on the cruise traffic. It also should be mentioned that these vessels are also representative of other European and Caribbean ports that specialise in the cruise shipping industry.

In addition, the sample includes data from 125 vessel calls (the number of calls per vessel is indicated in Figs. 2, 3, and 4) during 2015. For every vessel call, manoeuvring and berthing time and cruise speed real-time data are obtained from the modern AIS. Secondly, for each vessel, engine details (typology, ratings, electrical power, specific fuel consumption), vessel characteristics (GT, LOA, draught, beam, passenger capacity) and thruster and boiler properties (power and specific fuel consumption) come from IHS Sea-web database. Thirdly, the load factor and working time of the thrusters, type of fuel used (HFO, MGO/MDO) and hoteling electric power (kW) used during berthing activity are obtained through surveys and interviews of cruise shipping companies (steps 1 and 2 from Fig. 1).

Fig. 2
figure 2

Emissions annual inventory per type of propulsion for cruise vessels

Fig. 3
figure 3

CO2 annual inventory for cruise vessels

Fig. 4
figure 4

Pollutant emissions (NOX, SOX, PM) annual inventory for cruise vessels

4.2 Results

The total GHG (CO2) and air pollutant emissions (NOX, SOX and PM) for 30 cruise vessels during 2015 at the Port of Barcelona (about 520 vessel calls and 6277 hoteling hours) are estimated in this section. The emissions distribution per type of power used is depicted in Fig. 2, whereas emissions per air pollutant are represented in Figs. 3 and 4. In both figures, the vertical axis shows the identification code for each chosen vessel, the number of calls per vessel and its GT.

Hoteling emissions (electrical generators) were found to be dominant (79%), followed by those emitted by boilers (12%) and thrusters (6%) during manoeuvres. The remaining percentage (3%) corresponds to the main propulsion used to move the vessel from/to the Landfall Buoy/Cruise terminal dock.

It should be stated that the above rates are in accordance with the study of Maragkogianni and Papaefthimiou (2015) for Greek ports, which concluded that emissions during hoteling corresponded to 88.5% of total and those produced during ship manoeuvring activities about 11.5% of total. However, it was said that emissions during ship operations were overestimated.

In absolute terms, the total emissions derived from the 30 cruise vessels amounted to 41.750 tons of CO2, 955 tons of NOX, 900 tons of SOX and 94 tons of PM. On average, per vessel call, the estimation of emissions was 80 tons of CO2, 1.85 tons of NOX, 1.75 tons of SOX and 0.20 tons of PM.

5 Cruise vessel emission indicators

Based on the estimation of emissions represented above, the next step is to estimate indicators with the aim of extrapolating the estimations for other cruise vessels based on vessel dimensions (GT and capacity) and port time (manoeuvring and berthing time).

In order to choose appropriate indicators, a regression analysis is performed between total pollutant emissions/hoteling emissions and independent variables (port time, passenger capacity and vessel GT). In case the regression model (linear regression) is deemed satisfactory, in the sense that a relationship exists among variables, then an indicator combining those independent variables will be chosen. That is, the estimated regression equation or indicator can be used to predict the emission values based on the vessel dimensions (GT) and/or port time.

Figures 5 and 6 represent the satisfactory regression models for total emissions and hoteling emissions per cruise vessel call, respectively. As emissions differ with the type of pollutant (depending on fuel consumption and/or engine ratings), CO2 and NOX emissions are analysed separately. It should be mentioned that SOX and PM emissions analyses are equivalent to CO2, as both of them also depend on fuel consumption (see Table 2).

Fig. 5
figure 5

Regression model figures with regards to total emissions

Fig. 6
figure 6

Regression model figures with regards to hoteling emissions

From the regression analysis, it can be stated that the independent variables capacity (passengers) and vessel GT cannot be individually used to predict the total emissions and hoteling emissions, as the correlation coefficient is weak, indicating that there is no relationship between the two variables. However, by combining them with the port-time variable, the regression model results indicate an excellent relationship.

Therefore, it can be concluded that the best independent variable to predict total inventory emissions or hoteling emissions emitted by cruise vessel at ports is the port time—GT. Alternative variables to estimate cruise vessel emissions are dwell time—passengers and port time.

Finally, Table 3 lists average emission values for every selected indicator and the 25th and 75th percentile values in order to show the range variability.

Table 3 Emission indicators (average values and 25/75% percentile values [in square brackets]) regarding port time, gross tonnage (GT) and number of passengers per vessel

It should be said that hoteling emission values from Table 3 included both time at dock (85% of total hoteling) and manoeuvring time (15% of total hoteling) within the port area.

6 Conclusions

The need to control air pollution at ports is widely acknowledged as an active policy issue by numerous ports and international port associations. In such a context, a fundamental requirement for emission control and planning mitigation strategies to reduce the environmental shipping impacts is the development of accurate emission inventories for ports.

Under this framework, this paper addresses the estimation of air emissions released by cruise vessels in urban ports. This is of great importance due to a significant share of emissions produced during the time cruise vessels stay in ports. In addition, this paper provides useful cruise ship emission indicators, which could facilitate reliably estimating the in-port ship emission inventories of cruise ports without requiring large amounts of data and high levels of detail.

The proposed methodology is based on the “full bottom-up” approach and begins by evaluating the fuel consumed by each vessel on the basis of its individual port activities (manoeuvring, berthing and hoteling at the terminal dock). The methodological scheme also separately considers different types of vessel propulsion: main propulsion (diesel or diesel-electric engines), auxiliary propulsion (thrusters), boilers and generators providing electrical energy for hoteling. Once the fuel consumed is determined, the next step is estimating air emissions from cruise vessels by employing the corresponding emission factors per air pollutant.

The methodology was implemented to a particular case in which 30 cruise vessels and 125 calls were monitored in the Port of Barcelona during 2015. The emission estimations led to the following considerations:

  • Hoteling emissions (electrical generators) were found to be dominant (79%), followed by those emitted by boilers (12%) and thrusters (6%) during manoeuvring. The main vessel propulsion accounts for the remaining percentage (3%).

  • Hoteling emissions produced during berthing time represent about 85% of the total hoteling emissions, whereas the remaining 15% are produced during manoeuvring activities.

  • According to the sample data, the average estimation of emissions per vessel call was 80 tons of CO2, 1.85 tons of NOX, 1.75 tons of SOX and 0.20 tons of PM.

With regard to emission indicators, it was found through a regression model that the best independent variable to predict total inventory and hoteling emissions was the combined variable port time–GT. Nonetheless, the variables port time–passenger and port time are also quite robust. In relation to the indicator emission per port time and GT, the following values could be used to estimate total emissions at ports: 69.80 g CO2/h-GT, 1.68 g NOX/h-GT, 1.50 g SOX/h-GT and 0.16 g PM/h-GT.

With respect to the reliability of the emission indicators, it should be mentioned that information regarding vessel activities, hoteling power, engine ratings, fuel use, emission factors related to NOX and load factors are based on empirical and real information (work field) received from shipping crew companies, which means that estimations are quite consistent.

In summary, this paper contributes to the development of ship cruise emission indicators, which can be extended to other cruise ports to reliably and quickly estimate emission inventories and to calculate emission inventories, which could help to understand cruise emissions when proposing environmental and policy measures.