Can carbon emissions estimates play a significant role in the race for decarbonisation?

In this article, we briefly discuss the challenges of estimating shipping emissions on a bigger scale. We summarise our research findings for estimating emissions from all cargo ships (i.e container ships, drybulk carriers, tanker ships, roro, etc) that have entered Felixstowe port from January 2019 till December 2021.

Then, results are compared with a recent publication from the Port of Felixstowe authority that reports expected annual CO₂ emissions generated from ships to follow a steady increase between 60 and 70 thousand tonnes from 2019 to 2030 (see Figure 1).

Figure 1: Port of Felixstowe CO₂ emissions 2019-2030. Source: Port of Felixstowe Authority

Air pollution and greenhouse gases from port activities are considered a significant public health hazard as ports are usually located very close to city centres. According to IMO 2020 GHG report, annual CO₂ emissions from shipping are estimated to account for 2.9% of global emissions, and 70% of these ship emissions are generated within 400 km of the coast.

On another side, accessing actual ship’s data is a crucial aspect to accuracy. Calculating emissions for a single ship or a fleet of ships is a simple task if based on actual fuel consumption reported by vessels (i.e. noon reports). However, calculating shipping emissions based on fuel consumption estimates is a challenging task and requires advanced modelling and a data science approach (i.e. big data, machine learning).

As such, this article aims to discuss the following questions:

What are the different approaches to estimating fuel consumption?
What are the challenges in estimating carbon emissions on a large scale?

Maritime researchers have implemented two approaches to estimate emissions, top-down and bottom-up. The former is used to estimate annual emissions on a macro level and is the least accurate compared to the latter. Therefore, 91% of research that focuses on estimating shipping emissions uses a bottom-up approach (see Figure 2).

Figure 2: Research methods for estimating shipping emissions.

The bottom-up method calculates the emissions of individual ships in different operating conditions separately. The results are summed up to obtain the total emissions during the port time. Therefore, it is an estimation method that requires a high data resolution. Only a few major international ports have publicly available data to meet this demand, and most researchers need to work with local ports to obtain relevant information.

In summary, the top-down method requires simple data but is limited by the resolution of the data. So, the results cannot explain the details of vessel emissions based on port activity. This limitation is addressed in the bottom-up method but is challenging as it involves many technical parameters and a detailed record of ships’ movements.

Additionally, research published by IMO divides the bottom-up approach into two categories, energy-based and fuel-based. The main difference between both methods is the emission factor. The emission factor is an empirical constant that depends on the engine type, emissions standards applicable in the operation and fuel type region, expressed in g/kWh (IMO, 2014a). Empirical research (also in this article) shows that the fuel-based method provides more accuracy. Therefore, 66% of published research focuses on using this method.

Both methods are based on the following formula:

E: emissions in g or ton
MCR: maximum continuous rated power in kWh
LF: load factor (unitless)
time: activity time in hours
EF: emission factor, expressed in terms of g/kWh, depends on engine type, IMO NOx standards and fuel used.
i,j,k: represent the operating mode (cruising, manoeuvring, and at berth), engine type (main engine, auxiliary engine, or boiler), and the pollutant species.

This formula can be used for estimating shipping emissions on a big scale. However, empirical research has shown that this approach overestimates emissions, and a more dynamic approach needs to be adopted to improve estimation accuracy and, consequently, long-term forecasts. Only 64% of published research considers the vessel’s speed in estimating the Load Factor, and only 32% considers specific fuel consumption (SFC) in their calculations, with only 7% taking this further and considering the use of the SFC curve.

Sea Edge implements a bottom-up approach based on an adjusted fuel-based method for the SFC curve and a dynamic load factor that is influenced by the change in vessel’s speed and weather. Figure 3 depicts our overall framework to estimate emissions.

Figure 3: Sea Edge's research framework for estimating emissions.

The empirical challenges

Collecting data on a big scale is challenging and a complex task. This requires the use of complex applications, simultaneous access to multiple databases and machine learning applications. For example, a series of algorithms are required to deal with missing shipping technical data (i.e. engine RMP, engine power, LOA, cargo capacity, design/service speed, etc.). Ship’s activity (AIS) data will have missing and incorrect values that require cleaning and a quality control procedure (i.e. the great circle distance between two points can be calculated using the haversine formula to check the accuracy of reported AIS speed).

Case study

The Port of Felixstowe is the largest and busiest container port in the UK, and one of the largest in Europe, handling over 4 million TEUs per year and home to around 2,000 vessels, including today’s largest container ships.

This research focuses on estimating carbon emissions for all the ships within the boundaries of Felixstowe port from 2019 to 2021, excluding miscellaneous. First, understanding the profile of ships calling at the port and relative cargo capacity is required. Figure 4 shows percentages of the different types of ships that called at the port during the study period, indicating that over 98% of ships are Ro-Ros and container ships.

Figure 4: Profile of ships calling at Felixstowe port.

Figure 5 shows the number of calls at the Port of Felixstowe. From January 2019 to August 2021 monthly calls remained relatively stable between 140 and 180, with an evident decline after that.

Figure 5: Number of calls at Felixstowe port.

Figure 6 shows the impact of different ship sizes (cargo capacity) on time spent in port, which is a critical emission factor. There is, obviously, a positive correlation between cargo capacity (DWT) and time in port.

Figure 6: The relationship between cargo capacity and time at port.

Second, we estimate emissions for the sample period for all vessels using the framework depicted in Figure 3, illustrating, and reporting the results in Figure 7 and Table 1, respectively. In Figure 7, annual carbon emissions are reported based on both methods described earlier, the energy and fuel-based methods. By comparing our results with Felixstowe’s official report, we can see that the energy-based method overestimates carbon emissions compared to the fuel-based.

Figure 7: Port of Felixstowe CO₂ emissions 2019-2021.

Table 1: Annual carbon emissions (in tonnes) estimated vs reported.

Table 2 reports the total and intensity of carbon emissions according to ship status within the port. Carbon emissions intensity is much higher when ships are steaming at average speed (sea status) than when ships are manoeuvring or at berth. For this case study, ships that are within the port and manoeuvring or at berth, their carbon emission intensity is less than half compared to when they are at sea, which we are using as a benchmark (BM). However, most carbon emissions produced by ships at the port are generated at berth due to the time spent on cargo operations.

Table 2: Total and intensity of carbon emissions by ship status.

Table 3 reports more details regarding carbon emissions totals and intensity, and average port time based on ship’s type.

Table 3: Total and intensity of carbon by ship type.