The International Maritime Organization’s
targets for decarbonization
The shipping sector plays a key role in the global economy,
transporting people and goods worldwide. Carrying around 80%
of the world’s trade volume and 70% of its value, marine vessels
are estimated to account for 2.9% of worldwide carbon dioxide
emissions [1].
The International Maritime Organization (IMO) has adopted a
strategy to progressively reduce the marine industry’s greenhouse
gas (GHG) emissions, in line with the Paris Agreement on
climate change in which in 2015 adhering countries agreed to a
commitment to limit the greenhouse effect.
The IMO strategy to progressively reduce the GHGs from shipping,
adopted by the Marine Environment Protection Committee (MEPC)
in 2018 [2], to progressively reduce the GHG from shipping
includes the objectives of:
- reducing CO2 emissions per transport work, as an average
across international shipping by at least 40% by 2030,
compared to the levels of 2008
- reducing total annual GHG emissions by at least 50% by
2050, compared to 2008.
Methanol’s role as a decarbonization fuel
Methanol (CH3OH) is a chemical used in thousands of products.
While it can be produced from different sources, it is traditionally
made from fossil feedstocks via syngas. Renewable methanol,
instead, is produced either from biomass (bio-methanol) or
from captured CO2 and H2 produced from water by electrolysis
via renewable electricity (e-methanol). Compared to other fuels,
methanol can reduce CO2 emissions by 65% to 95%, depending
on the feedstock [3]. The use of renewable methanol as a fuel
is therefore strategic for those sectors, including shipping, which
are transitioning to decarbonization. In addition, its combustion
is sulfur oxide-free and generates low nitrogen oxides emissions
compared to other conventional fossil fuels [3].
Methanol in marine diesel engines
The main marine engine manufacturers have developed
technologies to burn methanol in diesel engines [4].
With its expertise in marine fuel handling and conditioning,
Alfa Laval has contributed to methanol technology development
from the very beginning, working on the first methanol fuel
supply system prototype. Today, Alfa Laval
has a strong experience in designing and
supplying methanol fuel supply systems, with
12 systems currently installed and operating
onboard methanol carriers, with a total of
100,000 hours of operation, plus several other
systems in the final stages of development.
The use of methanol as a fuel for the first
container vessel is expected by 2023. This
is a step towards sustainable zero-emission
vessels, in line with the IMO’s decarbonization strategy. With
its commitment to sustainability, Alfa Laval is fully involved in
the development of technology to support methanol and other
alternative fuels [5].
Simulation’s role in product development at Alfa Laval
In addition to traditional and consolidated engineering practices,
Alfa Laval is adopting advanced engineering tools to support the
product development process, with the aims of ensuring costeffective
design and increasing the efficiency of methanol fuel
supply systems.
One of the first modeling and simulation activities was dedicated
to the methanol fuel supply system because of its strategic role
in Alfa Laval’s portfolio, which is expected to increase in the near
future. The objective is to develop virtual models of methanol fuel
supply systems and validate them through field data retrieved
from operating systems. Once validated, the models can be used
as the basis for analyzing the behavior of the process under nonstandard
conditions, and for providing remote customer support.
Methodology
In this study, Alfa Laval worked with EnginSoft to develop a fluid
dynamic model of an existing methanol fuel supply system. This
model will generate numerous benefits, such as simplifying the
understanding of real system behavior and providing a tool for
making better engineering decisions.
Software
The Flownex simulation environment was used to model the system.
Flownex is a one-dimensional computational fluid dynamic (CFD)
modeling software. The 1D CFD modeling approach is suitable for
system simulations and enables the physical behavior of the entire
module to be modeled, studied and analyzed, taking into account
different operating conditions.
Process description
The methanol fuel supply system modeled in this study is a system
designed to feed a two-stroke, dual-fuel marine diesel engine with
methanol. It consists of a two-stage pressure module with an
intermediate mixing tank (see Fig. 1), designed to pump fuel from
the storage tank to the engine at the operating conditions required
by the engine, under varying loads. The system also includes heat
exchangers, filters, and valves to meet engine requirements for
temperature and degree of filtration in a fully automatic mode.
The methanol fuel supply system is also equipped with an
ethylene-glycol/water solution circuit to provide heating/cooling
media for the heat exchangers in methanol operation, preventing
contamination of any of the ship’s utilities in the event of an
internal leakage in the heat exchangers.
Model basis
The main parameters studied were pressures, temperatures,
flow rates, and valve opening in both the low-pressure and highpressure
recirculation loops.
Methanol pumps
Both the supply pump (LP) and the circulation pump (HP) were
modeled using the “Fan or Pump” component available in the
Flownex library. The flow-prevalence curves, the net positive
suction head required (NPSHr) curves, and the efficiencies were
taken from the datasheets of the pumps installed in the system.
This permitted the pumps to be simulated in terms of performance,
power consumption, heat transferred to the fluid, and cavitation risk.
Filters
The filters were modeled as pressure drop-generating components,
since this is the main effect related to the fluid dynamics of the
model. In fact, fuel purity, which is an important parameter in the
actual supply system, was not considered as a parameter in the
present study.
Therefore, the filters were modeled using the “User specified
pressure drop” component. The flow-to-pressure drop curves
were obtained from the datasheets of the filters actually installed
in the methanol fuel supply system.
Heat exchangers
The modeled system includes two heat exchangers in methanol
operation, one in the low-pressure section (LP HE), and one in the
high-pressure section (HP HE). The model was implemented and
validated against the system’s data.
Initially, the “heat exchanger primary”component was used. Four
thermal balances at different liquid-phase methanol (MeOH) mass
flows and inlet temperatures were used as reference cases to
calculate the required heat transfer rate as input to the component.
The pressure drop values were used to interpolate the factors Ck,
α, and β used in the following equation:
Subsequently, to obtain more accurate results, the “Shell Tube
Heat Exchanger” component available in the Flownex library was
used to model a more accurate geometry1. All data needed as
input to the model (geometry, fouling factor, materials, etc.) was
obtained either from the available datasheet or from Alfa Laval’s
in-house experience.
Mixing tank
The mixing tank was initially modeled by splitting the model of
the liquid portion and the model of the gas portion, and using
the “Open Container” and “Air Volume” components available in
Flownex Library, respectively, and associating them via a script
(see Fig. 2). By doing so, any changes in level during dynamic
load variations, or at start-up and shutdown, were automatically
reflected in a corresponding change in the available vapor
space above the liquid level, and thus in pressure. Later, the
“Accumulator” component was used, which simplified the system
while obtaining the same results1.
[1 The validation results presented in the “Results” section refer to the model without
this improvement.]
Other components
All pipes and bends were modeled using the “Insulated pipe” and
“Bend” components according to the actual 3D geometry of the
methanol fuel supply system. The on/off valves were modeled
using the “Basic Valve” component, with the actual valve flow
coefficient (Cv) of the valves used in the system. The same
approach was taken for the control valves, modeled using the
“Ansi control valve” with the actual valve Cv and characteristics.
The flow meter, similarly to the filters, was modeled with a “user
specified pressure drop”.
Glycol-water circuit
The glycol-water circuit consists of a pump that circulates an
ethylene glycol-water (GW) solution to a plate heat exchanger to
achieve the desired GW temperature set point. The GW is circulated
to the two heat exchangers in methanol operation and then back
to an expansion tank. The plate heat exchanger maintains the GW
temperature by exchanging heat with low temperature water (LT
water). To obtain accurate results and make use of the available
data, the GW system was also modeled.
Process control parameters
In order to deliver methanol at the operating conditions required
by the engine at varying engine loads, the methanol fuel supply
system is operated under pressure control at two points in the
process (low-pressure and high-pressure sections), and under
temperature control at the outlet battery limit. The control logic
that acts on the methanol fuel supply system is based on software
developed in-house by Alfa Laval, which enables fully automated
system operation.
At this stage of the model development, the pressure and
temperature controls were modeled using the proportional,
integral, derivative (PID) control available in Flownex. The PID
parameters were tuned to reproduce the trends in some actual
pressure and temperature datasets.
Validation of the model
To validate the developed model, several datasets from methanol
fuel supply systems in operation on vessels were analyzed in
depth and used as references. The parameters used as inputs to
the model, as well as the resulting outputs, are listed in Table 1.
The boundary conditions applied to the model are the methanol
temperatures and pressures at the module inlet and the
corresponding flow rate at the module outlet. For LT water, the
flow rate, pressure, and temperature conditions provided during
the module design were used as boundary conditions. The GW
flow rate is directly dependent on the flow-head curve of the GW
pump, and on the GW pressure. Since the latter had already been
considered in Table 1, and the characteristic curve was taken from
the pump’s datasheet, the GW flowrate was not further validated.
The data available for the methanol flow rate refers to a flowmeter
placed at the inlet of the mixing tank, i.e. between sections LP
and HP. Therefore, this dataset can be used as-is as a boundary
condition for the model only in steady-state conditions, assuming
the same flowrate at the system outlet. By contrast, during transient
phenomena characterized by flow variations over time, this set
of measured flowrates cannot be used directly as a boundary
condition.
Transient state simulation/test
Outlet battery limit’s on/off valve
The validated model was initially used to simulate the transient
phenomena that occur:
- at a sudden opening of the on/off valve at the outlet battery
limit when the fuel supply system is in full recycle mode
(startup condition and filling line to the engine);
- at a sudden closing of the on/off valve at the outlet battery
limit when the fuel supply system is in operation (in case of
engine switch to diesel oil).
The PID parameters of the model were tuned to reproduce the
experimental pressure peaks generated during the opening and
closing of the outlet battery limit valve.
Cyclic flow variations
The model was used to simulate a cyclic mass flow trend
observed in a set of experimental data measured during a sea
trial to evaluate the system’s functionality assuming extreme sea
conditions. Based on the key engine requirements, the methanol
fuel supply system must be designed and controlled to withstand
load variations without exceeding ±0.5 bar at the module outlet.
Fig. 3 represents a dataset taken over a 1,000-second time period.
This dataset accurately represents very severe conditions under
which the system should maintain its pressure variations within
the required limits. These values were rounded off to the following
sinusoidal function2 for input into the simulation software:
where ±1,250 kg/h is the amplitude of change in flow with an
average period of 25 seconds and a mean value of 1,250 kg/h, as
inferred from the trend shown in Fig. 3.
[2 The recent software update allows raw data to be entered directly into the simulation.]
For the transient simulations, the coefficients of the PID controllers
controlling the pressure were specified according to the control
logic being used in the actual system. The pressure variation at
the LP/HP interstage (before the mixing tank) and at the module
outlet were considered as parameters for validation.
Results and discussion
Validation results
Table 2 shows the values of the input parameters to the model
as well as the results obtained from the simulation, and the
corresponding plant values. The validation refers to a steady-state
dataset. A maximum deviation of 5% of the results was considered
acceptable for the methanol lines, while a larger tolerance was
allowed for the auxiliary lines.
The deviations in the LP and HP loop parameters are well below
the considered threshold. Therefore, the model can be considered
to be in line with the required accuracy.
In the GW loop, there is a larger discrepancy between the
simulated and experimental pressure values. This is due to the
presence of a manual throttling valve in the loop, an element for
which no data was available at the exact opening point from the
operational plant.
Transient simulation results
Simulation of the module outlet valve opening
Fig. 4 shows the trend in methanol flow rate and the supply
pressure upon the sudden opening of the on/off valve located at
the outlet battery limit of the methanol fuel supply system. The
two charts on the right in Fig. 4 reveal how consumption increases
due to the empty pipe attached to the module outlet and the
consequent pressure drop. Due to the sampling rate of the flow
meter and the maximum detectable flow, the top right graph has a
stepped shape. The charts on the left represent this behavior quite
well, although the simulated pressure drop is less significant.
The magnitude of the simulated peak flow cannot be compared,
but the trend is reliable. Overall, it can be said that the peak
consumption and pressure drop following the opening of the outlet
battery limit valve is well represented by the model. However,
adjustments to the flowmeter sampling rate and maximum
detectable flow will be further evaluated to provide a more reliable
dataset.
The good correlation between the resulting trends and the actual
behavior of the module also reveals that the model represents the
real system well, while the control logic can be improved. This
will be possible due to the recently added functionality that allows
the actual control logic to be implemented in the software, instead
of converting it into pre-built PID controllers.
Cyclic consumption simulation
As mentioned in the section on cyclic flow variations, the
approximation of the input mass flow data only allows for the study
of the resulting pressure variation and its magnitude. Fig. 5 shows
the experimental pressure trend at the two control points. At the
module outlet, the pressure variation is ±0.1, while before the
mixing tank the pressure variation is ±0.05 bar. Fig. 5a shows
some pressure spikes that cannot be reproduced as input using
the function approximated in the section on cyclic flow variations.
Fig. 6 represents the simulated trends over a shorter time frame.
The outlet pressure (Fig. 6a) has a variation of ±0.1 barg, while
the interstage pressure (Fig. 6b) has a pressure variation of ±0.08
barg, slightly higher than actual pressure.
The simulation of the severe cyclic flow variation reflects the
experimental pressure variation and complies with the engine
requirements. The acceptable accuracy obtained allows this
model to be used to simulate other conditions and understand
if the resulting pressure variations are within acceptable ranges.
Conclusions
Alfa Laval is adopting new methods of approaching fuel supply
system design and development based on data analysis and
system modeling.
In this paper, the model-based approach was applied to a fuel
supply system processing methanol, which is a key step towards
sustainable shipping.
The results of the modeling activities presented show a reliable
degree of prediction of the actual system’s behavior, fit for
purpose, at a degree of approximation that was judged to be
acceptable. Further optimization to the same system that can be
performed starting from this study includes:
- Implementing Alfa Laval’s automation software in the model,
instead of using standard simple PID logic;
- Simulating other transient states by directly inputting
experimental data without having to define a function to
describe the data trend;
- Simulation based on the experimental data related to the
configuration of the methanol fuel supply system with the
flowmeter located at outlet battery limit;
- Simulation of the system behavior under different boundary
conditions.
The availability of additional data from the deployed systems
would be useful to further validate the 1D CFD model in various
scenarios and to support future modeling activities.
Newsletter EnginSoft Year 18 n°4
By Nicoletta Spazzadeschi, Danish Taufiq, Davide Rossin | Alfa Laval
Erik Mazzoleni, Marco Gatti | EnginSoft
