Metering Marine Fuel

The ability to identify non-ideal operating conditions for an instrument is particularly important when metering a high value product such as fuel. In-built intelligence and the monitoring of additional data channels can not only detect non-ideal conditions but potentially provide corrected measurements and on-line uncertainty assessments where models can be developed. The $150 billion ship fuel (bunkering) industry still uses largely manual methods for determining the transfer of fuel from the bunkering barge to the receiving ship. These techniques are prone to error and potential fraud, with Maersk, the world's largest ship operating company, estimating it is defrauded of 1.5% of its $7 billion annual spend of fuel. Conventional flow metering does not work because of entrapped air in the viscous fuel. However, a new generation of meter is able to detect and correct for this condition. The detection of air can in turn improve the operation of the barge leading to better asset utilisation. Detailed trials and studies have been carried out demonstrating the benefits of the new approach.

bunker pride in action

 

 

In 2006, approximately 350 million tonnes of marine fuel were supplied worldwide. At current prices, this equates to approximately $150 billion of transactions annually. Increasing economic and environmental importance is attached to the measurements taken during the bunkering process (i.e. the physical transfer of ship fuel [1]), in terms of both quantity (usually measured in volume but sold by mass), and of fuel quality (such as viscosity, density and sulphur content). For example, the International Marine Organisation (IMO) is introducing a series of measures to impose significant cuts in the permitted levels of sulphur in marine fuel [2], in order to reduce harmful emissions. For the efficient enforcement of such measures, reliable methods are needed for monitoring both the sulphur content and the quantity of fuel at the point of supply.

 

There are two distinct classes of marine fuels: distillate grades, often described as gas oils, and the residual fuel grades. “Residual” refers to the crude oil content remaining after the lighter and more valuable components have been removed by the oil refining process. Currently about 90% of marine fuel is residual, however many ships use at least some gas oil to drive auxiliary power systems. Conventionally, residual fuels have been specified simply by viscosity; current standards (e.g. ISO 8217:2005 [3]) provide more thorough specifications for the fuels grades (including density, flash point, pour point and limits on components such as sulphur), but the grades are still labelled according to their nominal viscosities, so that for example the fuel grades RMF-180 and RMG-380 specify maximum kinematic viscosities of 180cSt and 380cSt respectively, measured at the standard temperature of 50ºC. At ambient temperature in high latitude ports, some fuel grades are effectively solid. As most bunker fuel is supplied to vessels via bunker barges, it is thus necessary to ensure that the fuel supply pipework on the barge is kept entirely clear between transfers (usually achieved by “blowing down” with high pressure air at the end of each transfer), to prevent solid fuel blockages. Even at an elevated pumping temperature, the viscosity of the fuel is high, and so air entrainment is a significant challenge to any metering system that might be used for fuel bunkering.

 

Another regular feature of bunkering operations is “tank stripping”, whereby the last dregs of fuel (together with increasing amounts of air) are pumped out of one storage tank on the bunker barge before the supply is switched to another. A typical modern bunker vessel of perhaps 4,000 tonnes capacity may have 6-10 such fuel storage tanks operating in port/starboard pairs: a high proportion of bunker transfers will thus include a period of tank stripping.

 

Currently, the most widely used means of measuring the fuel transfer from bunker barge to receiving ship is by manual tank dipping. In this procedure, the level of fuel in each bunker barge tanks is recorded, along with its temperature, prior to the transfer of fuel. This exercise is repeated at the end of the transfer. Barge-specific calibration tables are used to map the recorded dip levels into corresponding volumes, and the difference in volume between the beginning and end of the bunkering operation gives the transferred quantity. The tank calibration tables are normally derived from calculated data and the tank is not calibrated by measurement and filling. Further calculations, based on the density of the fuel (as certified by the supplier) yield the delivered mass. One difficulty of current practice is that regular recalibration of the measurement system, as practiced in most custody transfer applications, is rarely possible. A further difficulty is the measurement error that may be introduced by entrained air.

 

Dissastifaction with the current state of the industry has been most notably declared by Maersk, the largest shipping company in the world [4]. On an annual basis Maersk buys 13 million metric tonnes of marine fuel and undertakes 12,000 bunkering operations, but claims an average discrepancy of 1.5% between the readings taken by Maersk vessels and those provided by suppliers or barge operators. Until recently, all but the most specialised flow instrumentation was incapable of providing accurate measurement of two-phase flow (e.g. oil/air) mixtures. Indeed Coriolis metering was known to be especially vulnerable.

Developments at the UTC (described in Coriolis Research) have pioneered Coriolis metering technology capable of maintaining flowtube oscillation. Thus although the uncertainty of the measurements are higher than for single-phase fluids, it is becoming accepted that Coriolis meters are capable of generating usable measurements with two-phase flow.

One consequence of the new Coriolis metering technology is that the previous vulnerability to two-phase flow has been transformed into a very high sensitivity to two-phase flow, whereby the damping on the flowtube (as represented by the parameter known as the drive gain) can be used to detect the presence of even small quantities of gas, a feature particularly useful to the bunkering industry.

This is illustrated in figure 1, which shows data from an actual bunker transaction recorded during a trial in Singapore, as discussed later. As the first results using Coriolis metering for bunkering are shown in conferences and publications, the question most frequently raised is “how can you prove that there really is air in the bunker fuel?” The data in figure 1 is used to illustrate how additional channels of data, such as the drive gain, can be used to determine the presence of air.

 bunker drive gain and air

 Figure 1. Detection of entrained air in a commercial bunker transaction

 

The top graph in figure 1 shows the mass flow rate of fuel during the bunker transfer from the barge to the receiving ship, over the course of seven hours, as observed by the Coriolis mass flow meter. The second graph down shows the observed density of the fluid, again as measured by the Coriolis mass flow meter. Where there is no possibility of any air or gas being entrained in the fuel, then the density reading can be used as a further indication of the quality of the fuel – there are fairly strict limits placed on the permitted densities of bunker fuels, so that 380 cSt grade should be expected to have a density close to 980 kg/m3. It can be seen that prior to the start of the batch the density reading is low – the meter is partially but not entirely drained from the previous bunker - but once the batch begins the density reading rises to approximately 1000kg/m3. Later in the batch, at about 2:30am and then at 04:00am, the density drops significantly, indicating the presence of air in the fuel. The third graph shows the same density data, but on a restricted y-axis scale. Note that the difference between 980 kg/m3 and 970 kg/m3 is approximately one percent. The characteristics of the density reading become more readily apparent on this scale. Between the start of the batch and approximately 01:20am, the density reading is very smooth and relatively steady, although a slight decline is evident. At around 01:20am however, there is a clear change in behaviour: the density reading becomes much noisier and its mean value drops by about one percent. This is one indication of the presence of air in the fuel. The lowest graph in figure 1 shows the drive gain, an indication of the energy required to maintain flowtube osciallation and hence of the damping on the flowtube, a measure which is entirely independent of the mass flow and density readings. This can vary by up to two orders of magnitude, and so is plotted on a logarithmic scale. It is clear that from the start of the batch until about 01:20am, the drive gain, like the density reading, is also steady, but the subsequent 1% drop in density is associated with a doubling or more of the drive gain. This provides an independent confirmation that the drop in density is associated with entrained air in the bunker fuel, rather than say, a change in the composition of the fuel. 

 Laboratory Trials

Trials have been carried out at the National Engineering Laboratory in East Kilbride near Glasgow (NEL). A 200mm diameter Coriolis flowtube was selected based on typical bunkering flowrates of up to 140kg/s (504 tonnes/hour). The two-phase flow modelling and measurement techniques used were similar to those described by Henry et al. [5], which discusses trials on a smaller meter at the same facility for an application with Venezuelan heavy (i.e. high viscosity) crude oil. Tests on single phase oils and steady-state two-phase mixtures yielded satisfactory results, as discussed by Gregory et al [6].

 

These were followed by a set of simulated bunker loading tests. Each experiment began with an empty flowmeter. Single phase oil was introduced, which was later interrupted by short bursts of gas (simulating tank stripping), before finishing with an empty flowmeter. Figure 2 presents the time profile of a bunker simulation experiment. The top graph shows the mass flow rate from the test meter (solid line), and from a reference meter (dot-dashed line), which was kept full throughout the experiment. After the onset of flow, steady, single-phase flow is established at about 120s. However, between 240-300s, bursts of air are injected into the flow stream to simulate tank stripping or other process disturbance. After a further period of single phase flow, the flow is reduced, a valve between the reference and test meter is closed to keep the reference meter full of liquid, and a blast of high pressure air (at about 520s) is used to clear the lines and return the test meter to its empty starting condition. Note that this final slug of liquid seen by the test meter but not by the reference meter, is balanced at the start of the batch (at about 30s) by the delay in the onset of flow as seen by the test meter. Overall, therefore, the test and reference meters should report the same totalised flow.

 bunker NEL batch

Figure 2. Profile of Bunker Transaction Simulation at National Engineering Laboratory.

Table 1 shows the results of five such bunker simulation trials. The average error is less than 0.5%, and the repeatability is also about 0.5%. Overall, the trial was deemed to be a success, and a field trial in Singapore was arranged.

 

Table 1. Results from bunker simulation lab trials

 

Run

Index

Reference

Total (kg)

Meter

Total (kg)

 

Meter error

(kg)

(%)

1

13679.86

13651.15

-28.71

-0.21

2

14300.85

14273.60

-27.25

-0.19

3

12444.67

12419.29

-25.38

-0.20

4

13092.49

13017.37

-75.12

-0.57

5

13356.90

13255.77

-101.13

-0.76

 

Mean Error

-51.52

-0.39

Spread
(max - min)

75.75

0.57


Trials in Singapore

The next stage of the testing has been the construction of a half-size container skid (Figure 3), further calibration trials on water at the Mogas test lab in Singapore, and installation onto a bunker barge. Figure 4 illustrates the location and function of the skid on board the vessel. The barge has four main pairs of tanks, each served by two outlet valves – a 200mm main suction valve, and a100mm tank stripping valve. The latter is positioned lower in the tank body to access the last few inches of liquid. Fuel from the tanks is sent via the cargo pump room and the flowmeter skid to the hose boom and then onto the customer vessel. The monitoring equipment for the skid is located in the cargo control room, with a direct communications link to the research team.

 bunker skid on pride

Figure 3. Metering skid installed on bunkering barge during trials in Singapore

 

bunker pride layout

Figure 4. Metering skid location on bunkering barge during trials in Singapore

 

The trial monitored commercial bunker transactions on the barge, performing a detailed analysis of the meter performance. Statistics on the trial period are shown below:

 

•         Number of monitored bunkers with full data available                78

•         Number of days covered                                                     71

•         Total fuel monitored by Coriolis meter                                    75,608.420 tonnes

•         Value of fuel (based on $300 per tonne)                                 $30.2M

•         Estimated annual turnover (based on $300 per tonne)               $120M

 

A key finding was the extent to which two-phase flow was indeed present in the bunkering transfers, which exceeded expectations and initially drew considerable scepticism from colleagues from the bunker industry. However, as further trials (especially by Maersk) have revealed similar findings, the industry is beginning to accept the need to detect and reduce the presence of air in the fuel. Of course the Coriolis meters are able to provide accurate measurement despite the air entrainment, but the accuracy would improve still further if it were removed entirely.

The mass-weighted average gas void fraction (GVF) was found to be 3.11% i.e. on average, the fluid stream contained about 3% air. However, this varied from bunker to bunker, with the highest recorded average being 16.38%. Much of this behaviour can be explained in terms of tank stripping and other known procedures, but some of the data is not fully explained. For example, as shown in figures 5 and 6, two consecutive binkers, produced only hours apart and using the same fuel cargo, showed very low and very high levels of gas entrainment respectively.

bunker low air

 Figure 5. Commercial bunker transaction with low air content.

 

bunker high air

Figure 6. Commercial bunker transaction with high air content.

Discussions with the barge operators suggest that possible additional sources of air might include pump seals and the use of bypass pipes to control the flow rate.

 

There are many benefits to being able to detect and estimate the presence of two-phase flow, especially if the data is shared between the receiving vessel and the bunker barge. Changes in operating procedures, armed with the additional data provided, can lead to reduced disputes, more efficient use of bunker barge and port facilities, and more rapid turnaround of cargo vessels.

In the light of the rapid uptake of metering on both bunker barges and ships, the Singapore Marine Port Authority has established a committee to develop a draft standard for the use of Coriolis metering for bunker transfers.

 

References

[1]   Fisher, C., Lux, J. ‘Bunkers: An Analysis of the Practical, Technical and Legal Issues’, 3rd edition 2004, Petrospot, ISBN 0-9548097-0-X.

[2]  International Maritime Organisation, IMO environment Committee makes progress’, Briefing 10, 26 March 2010.

[3]  International Standards Organisation, ‘Petroleum products -- Fuels (class F) -- Specifications of marine fuels’, ISO 8217:2005.

[4]  Peterson, C.M., ‘Measure for Measure’, Bunkerspot, Feb/March 2009, pp37-38.

[5]  Henry, MP, Tombs, M, Duta, MD, Zhou, F, Mercado, R, Kenyery, F, Chen, J, Morles, M, Garcia, C. “Two-phase flow metering of viscous oil using a Coriolis mass flow meter: a case study”, Flow Measurement and Instrumentation 17 (2006), pp399-413, doi:10.1016/j.flowmeasinst.2006.07.008

[6]  Gregory, D, West, M, Paton, R, Casimiro, R, Boo, S, Low, YK, Henry, MP, Tombs, MS, Duta, MD, Zhou, FB, Zamora, ME, Mercado, R, Machacek, M, Clarke, DW. “Two-Phase Flow Metering using a large Coriolis Mass Flow Meter applied to Ship Fuel Bunkering”, Measurement and Control, Sept 2008.