Method and system for reducing marine fuel consumption

JP2022519354A5Active Publication Date: 2024-11-21SHELL INTERNATIONALE RESEARCH MAATSCHAPPIJ BV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
JP2021544806
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Priority Date
2019-02-07
Filing Date
2020-02-03
Publication Date
2024-11-21
Estimated Expiration
2040-02-03

AI Technical Summary

Technical Problem

Current methods for optimizing vessel trim and draft do not account for real-time operational changes, leading to sub-optimal fuel consumption and increased greenhouse gas emissions, and are often costly and inaccurate due to reliance on static analysis and limited sea condition testing.

Method used

A method and system that collects and analyzes historical vessel data to determine optimal trim and draft conditions using artificial neural networks and regression models, filtering out noise and weather effects, and providing real-time recommendations for draft and trim adjustments to reduce fuel consumption.

Benefits of technology

Achieves fuel savings of 3-7% per year by optimizing draft and trim based on real-time operational parameters, reducing fuel consumption and greenhouse gas emissions while being cost-effective and deployable on existing vessels.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

A method for reducing a vessel's fuel consumption by optimizing the vessel's draft, speed, and trim using historical vessel data. Historical global online data is collected for multiple vessel operating parameters associated with its previous voyages. After initial filtering and cleaning of the collected data, a process is described in which the data is analyzed to determine the optimal draft, speed, and trim for a given speed of the vessel. The determined optimal draft, speed, and trim values ​​are then presented to the vessel's master or an automated draft and trim optimization system to adjust the current draft and trim. Thus, the present application discloses a method for analyzing historical vessel data to provide advice regarding optimal draft, trim, and speed. Methods for predicting achievable fuel savings and recording achieved fuel savings are also disclosed.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present disclosure relates to a method for improving fuel consumption of a ship or class of ships through optimization of operational parameters. The present disclosure also relates to a process for collecting and analyzing ship data through various mathematical models to further determine optimal operational parameters. [Background technology]

[0002] Generally, a vessel will start a voyage with a given trim and draft, which will then change during the voyage as a result of internal transfers, such as fuel or lubricating oil transfers, and due to the consumption and production of fresh water and other fluids. For liquefied natural gas (LNG), liquefied petroleum gas (LPG), and other liquefied gas carriers, the draft and trim will also change as the cargo is reduced due to the consumption of boil-off gas (BOG). This draft and trim may not be the optimal condition for the vessel's current speed. This non-optimal draft and trim condition leads to increased resistance, requiring more power from the main engine or engines to overcome the increased resistance to maintain a constant speed. This increases overall fuel consumption and increases the emissions of CO2 and other greenhouse gases. This can be prevented by better optimizing the vessel's trim and draft. Increased operating costs are also seen due to increased fuel consumption.

[0003] Currently available techniques for determining optimal vessel speed or power source configuration and utilization are typically based on static analyses performed when the engine is factory tested prior to delivery or on board the vessel during its first sea trials. These calculations are often performed as manual calculations using approximate data.

[0004] After factory testing and initial sea trials, the ship's optimal operating profile will change due to wear, aging, maintenance, operation, etc. of the ship and onboard equipment. Each ship looks slightly different and the optimal operating parameters will change over time. Existing technologies do not consider these various changes in the ship's operating profile, which directly relate to optimal speed and power source configuration and utilization. These existing technologies are based on previous static analysis and lack real-time analysis of optimal ship performance based on the ship's real-time operating parameters (i.e., current state).

[0005] Despite the availability of performance optimization systems that are partially automated or involve operational staff, our assessment shows that ships currently typically operate in suboptimal conditions for a significant portion of their total voyage time, e.g., around 30%.

[0006] Current industry practice typically only considers trim optimization as a single parameter to reduce main engine power and fuel consumption, and so far initiatives have met with limited success. Trim values ​​are typically obtained by CFD computational modeling or measured during sea trials, both of which cover a limited number of sea conditions. Information obtained in such a way is less accurate and does not maximize the fuel reduction potential (low number of sea conditions tested). Traditional solutions also have relatively high development costs and are slow to deploy.

[0007] Currently, only trim parameters are optimized to reduce fuel consumption. US7243009 discloses a system that uses a statistical model of the input data to determine only the optimal trim. This method provides some performance improvement, but even greater improvements can be achieved. Furthermore, this method is limited in accuracy because it relies on the current vessel trim, power, and speed.

[0008] US20140336853A1 discloses a method for determining an optimal speed for a vessel using historical data and a method for calculating fuel savings achieved from this optimal speed, the method optimizing speed solely by comparing current vessel data with historical data stored on a computing device.

[0009] SG190462 discloses a method for optimizing fuel efficiency of a vessel comprising various steps of retrieving operational data of the vessel, comparing said data with optimum data to obtain a variance, and thereby automatically adjusting ballast tank pairs of the vessel based on the variance to obtain an optimum trim of the vessel to achieve optimal fuel efficiency of the vessel.

[0010] US2016 / 121979 discloses a system that allows identification of a cruising condition such that a ship can provide higher cruising performance than when cruising under the predetermined cruising condition. A ship management device obtains input data indicative of the predetermined cruising condition and extracts state data corresponding to the input data. The state data includes data describing a combination of a cruising condition and a fuel consumption amount and is determined in advance. The ship management device uses the state data to identify a fuel consumption amount at the voyage condition indicated in the input data and determines whether the fuel consumption amount can be improved. The ship management device then generates and outputs monitoring data corresponding to the result of the determination.

[0011] WO2010 / 031399 discloses a ship system including a number of sensors measuring a number of data sets including settings controllable by the ship operator, each of which defines a state of the ship at a particular sea condition, the system generating a statistical regression model of the ship's fuel efficiency, and the optimal settings providing the highest fuel efficiency for the current state of the ship are determined by optimizing the statistical regression model of fuel efficiency with respect to the settings controllable by the ship operator. In this system, the current state of the ship is determined by optimizing the statistical model with respect to control variables. The control variables can be controlled by the ship operator. Incoming data is compared with the control variables and the ship operator is requested to adjust the control variables. Noise filtering is performed to derive the unknown parameters Pf, Pg and corresponding noise parameters.

[0012] Other methods commonly used in the industry for trim optimization use computational fluid dynamics (CFD) or measured sea trials. This allows only a limited number of speed and trim conditions to be tested, and not the full range of normal operating conditions. Furthermore, the vessel is only tested in "as new" condition, and trim optimization does not take into account performance degradation over time. These CFD methods often do not include metocean or weather data, or the effects of hull fouling or coating performance, thereby reducing the accuracy of the results. The described methods of trim optimization also result in high costs due to expensive CFD software, computing power, or the need to perform sea trials or model tests of the vessel.

[0013] Optimizing ship performance reduces total ship fuel consumption and costs, maximizing ship profitability. Fuel consumption is based on actual engine and generator performance only, as an example. It is based on the ship's operational parameters. Fuel consumption is related to both the amount of fuel required to propel the ship along the way and the fuel required to power the necessary equipment on board the ship during the ship's voyage. The optimal ship speed must balance the benefits of slowing the ship to conserve propulsion fuel with the associated costs of the impact of additional electrical load (i.e., the power required to operate the necessary equipment) on fuel consumption due to the excess time required to sail at a slower speed. This optimal speed can also take into account opportunities to optimize the ship's profits through more revenue by running more voyages if there is additional unmet demand. Additionally, ship performance may be improved based on various power source configurations and utilization rates. Summary of the Invention

[0014] The present disclosure aims to provide a more accurate method and system for optimizing navigation parameters.

[0015] According to a first aspect, the present disclosure provides a method for determining optimal trim and draft for a vessel in ballasted and loaded conditions using analysis of historical vessel data for the vessel being optimized, comprising: - collecting operational data from the vessel for one or more of its previous voyages, the operational data including one or more operational parameters or data tags; - filtering out errors and noise created by selected sources of operational data; - filtering out the effects of bad weather, hull fouling and / or other conditions that have been found to reduce the accuracy of the process; - processing the operational data by placing the operational data into classes or bins of increasing size for speed, draft, and trim to determine an average power output for each historical speed, draft, and trim condition; - creating a database of optimal draft and trim conditions based on operational data and providing the database to an operator, such as a ship's captain, or as input to an automated draft and trim optimization system; - calculating projected fuel consumption for each speed, draft, and trim condition based on operational data to estimate achievable fuel savings; comparing the predicted fuel consumption to the achieved fuel consumption of the vessel to determine the achieved savings using information about current fuel prices.

[0016] The step of filtering out error and noise may include removing data that falls outside of what is considered reasonably practicable.

[0017] Filtering out the effects of bad weather, hull fouling, and / or other conditions that have been found to reduce the accuracy of the process may include removing data that falls outside of set points based on previous experience.

[0018] In one embodiment, the operational parameters include one or more of vessel draft, trim, fuel consumption, date and time of sample collected, speed on land, speed in water, main engine power, main engine rpm, true wind speed, relative wind angle, engine fuel mass flow, fuel consumption, water depth, shaft rpm, and time since last hull wash.

[0019] In another embodiment, the method includes the use of artificial neural networks and regression tree models to further improve the accuracy of the model.

[0020] In yet another embodiment, the method includes displaying real-time results to the master or an autodraft and trim control system using computer software or code and / or a graphical display.

[0021] In one embodiment, the graphical display provides the captain with the optimum draft and trim of the vessel to alter the current state of the vessel.

[0022] Optionally, each individual step is combined into a workflow that is installed on a computer or chip to perform the analysis automatically.

[0023] The method may include the step of continually measuring the current vessel speed to further select the optimum draft and trim.

[0024] In one embodiment, a method includes using an artificial neural network to analyze and provide optimization of draft, trim, and speed for an entire fleet of vessels, the fleet including a plurality of substantially similar or identical vessels.

[0025] The method may include using data regarding fuel quality and primary engine mechanical failures to refine projected fuel consumption for economy estimation.

[0026] According to another aspect, the present disclosure provides a system including tools for implementing the method as described above.

[0027] To overcome the limitations of current methods of trim optimization, a method is disclosed that analyzes vessel data to determine the optimum trim and optimum draft for any given vessel speed in both loaded and ballast conditions. The method can also determine the optimum draft and trim conditions for the vessel in its current condition, rather than the current method of applying an optimized trim found in an "as new" condition throughout the vessel's life. By using "real world" data from vessels in service, the effects of weather, hull fouling, and coating performance can also be considered when using this method. CFD and other methods can only predict real world weather based on models and may not accurately represent real world events. The disclosed method is also substantially less costly than currently deployed methods. The described method requires a source of data that can be obtained from any number of data collection providers. As a further cost reduction, data can also be collected by information reported manually through "midday" or daily reports. The process can be run on a common laptop or computer using freely and widely available low-cost software with minimal time required to run the process. Currently, existing methods using historical data are not able to provide advice to similar or sister vessels in the fleet. Using the disclosed method, this draft, trim and speed optimization can be applied to an entire fleet of similar vessels.

[0028] By employing the disclosed method, fuel consumption and associated greenhouse gas emissions can be reduced. Reducing fuel consumption can save operational costs. By optimizing draft, trim, and speed, the vessel can maintain a higher speed with the same fuel consumption or maintain the same speed with lower fuel consumption. It has been found that a typical vessel sails in non-optimal draft and trim conditions for a low percentage of the total voyage time. By using the present invention, the percentage of voyage time that the vessel spends in optimal draft, trim and speed conditions can be increased. When draft and trim change during a voyage due to consumption and movement of fuel or other liquids, then the draft and trim can move away from the optimal condition and the draft and trim can be adjusted to correct back to the optimal condition.

[0029] The disclosed system provides a more accurate, lower cost, higher gain solution for optimized voyage parameters. The system also allows for rapid deployment and can be implemented on existing vessels. [Brief description of the drawings]

[0030] The figures in the drawings depict one or more implementations according to the present teachings, by way of example only, and not by way of limitation. In the figures, like reference numbers refer to the same or similar elements, where: [Figure 1] FIG. 1 illustrates an exemplary draft and trim optimization workflow diagram, demonstrating placing data into bins or classes, according to one embodiment of the disclosed method. [Diagram 2] 1 shows an example plot of fuel consumption and power to estimate potential savings. As disclosed, it is possible to predict the fuel savings achieved by optimizing draft and trim using optimal polynomial equations relating fuel consumption and vessel power. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0031] Certain terms used herein are defined as follows: "LNG" refers to liquefied natural gas, which is typically cooled to a temperature at which the gas can be in a liquid phase at a pressure of at least about 1 bar; for liquefied methane, this temperature is about minus 162°C. "Draft" may refer to the depth of water required to float a vessel, the depth from the waterline to the bottom of the vessel's hull, or the depth of water a vessel can pull. In the present disclosure, "draft" may refer specifically to the depth from below the waterline to the bottom of the vessel's hull, and may also refer to the depth of water a vessel can pull. "Trim" refers to the position of a vessel relative to the horizontal. In other words, "trim" refers to the difference between the draft at the forward end of a vessel and the draft at the aft end of the vessel.

[0032] Whenever a ship sails at a particular speed, whether carrying cargo or under ballast, there exists an optimum trim and draft condition for that ship. For a ship in service, a large amount of data related to its performance is generated. This data can be collected by the crew, for example through "noon" reports, or through automated radio frequency data acquisition systems connected to the ship's machinery. While in operation, the ship's master may select a draft and trim setting at a particular ship speed that requires less engine power than a previous draft and trim setting at the same ship speed.

[0033] The disclosed method allows data to be analyzed to determine which historical trim and draft conditions performed better than others for a given speed. By providing this information to the vessel's captain, draft and trim settings can be selected that result in lower main engine power at a particular speed, which would reduce overall fuel consumption. By implementing the disclosed process, savings of 3-7% per vessel per year are achieved.

[0034] The disclosed method uses collected historical vessel data that includes one or more of the following parameters: - date and time when the sample was collected, - ground speed, -Speed ​​in water, -Main engine power, -Main engine rpm, -True wind speed, - Relative wind angle, -draft, -trim, -Engine fuel mass flow rate; - Water depth, -shaft rpm, and -Time since last hull cleaning.

[0035] This data may be collected using either radio frequency data loggers (there are numerous providers), "midday" data reported by the vessel's crew, or other reliable data sources. The period of data used should be relatively short to avoid the impact on power requirements due to increased hull fouling. Data beyond one year would begin to include the effects on the hull of increased frictional resistance caused by marine growth, thereby reducing the accuracy of the method.

[0036] Data collected from high frequency data loggers is often referred to as "noisy" data, with errors introduced by the data logger. This may include missing data, zero values, or impossibly high or extreme data being captured in the log. Before using this data for analysis, processing or filling of the data by removing data outside of ranges deemed reasonably practicable may be performed to remove or reduce the effect of these errors. For example, zero values, empty data, and values ​​other than physically possible values ​​may be filtered out and removed from the data set. Upper and / or lower thresholds may be set, for example based on the vessel's expertise or specifications, to remove data values ​​other than physically possible values. Filtering thus removes noise introduced by the sensing equipment.

[0037] For example, the disclosed method may filter out any value where the speed is greater than the design speed of the vessel. Thus, if the speed measured by the sensor is more than a predetermined percentage above a predetermined design speed of the respective vessel, the measured speed will be ignored. The design speed may be provided, for example, by the manufacturer of the respective vessel, or may be otherwise pre-determined. For example, if the measured speed is greater than 100.1% of the design speed, it is ignored. For example, if the measured speed is greater than 101% of the design speed, it is ignored. For example, if the measured speed is greater than 102% of the design speed, it is ignored. For example, if the measured speed is greater than 103% of the design speed, it is ignored. For example, if the measured speed is greater than 105% of the design speed, it is ignored.

[0038] For example, a data point that captures a vessel's speed at a speed that exceeds the vessel's maximum achievable speed (e.g., 40 knots) is clearly erroneous and may be deleted.

[0039] It may also be possible to remove or filter the effects of adverse weather conditions to perform draft and trim optimization. Extreme weather conditions require greater main engine power to overcome increased wave and wind resistance, which would prevent the effects of draft and trim changes from being analyzed. For example, some data recorded when sea states are greater than 4 may be ignored if the Beaufort wind scale is greater than 5, but the current effect is greater than 3%. The current effect is a measure of how much the current has affected the vessel's speed. It may be calculated as a percentage, for example, as shown in Equation 1:

[0040]

number

[0041] Herein, speeds on land and speeds in water are expressed in knots, where 1 knot is equal to approximately 0.514 m / s. Note that although the adverse weather data is removed for analysis, the draft and trim optimization advice can still be used to adjust draft and trim in adverse weather conditions. However, in adverse weather conditions, it may be difficult to measure the savings achieved.

[0042] Next, an averaging method is performed, whereby the historical data is placed into groups or "bins" of increasing size. Data for the previously listed data tags is collected over time, including the vessel's speed, draft, trim, and power during previous voyages. The historical data is placed into bins of increasing size by rounding the vessel's speed, draft, and trim to the nearest increment, and the remaining data for that sample is also placed in a bin. The vessel speed historical data is averaged in increments of 0.5 knots, draft is averaged in increments of 0.25 meters, and trim is averaged in increments of 0.5 meters. Each bin then has multiple data points with the same draft, trim, and speed, while the other data tags are left in the form captured by the data logger. For all the data captured in each bin, an average of the power measurements for that bin can be calculated. There will then be one draft, trim, speed, and average power value associated with each bin.

[0043] To further increase the reliability of the advice given to the ship's master, bins containing a limited number of data points are ignored in the process. Typically, states where the ship has made fewer than 50 voyages are rejected, but various thresholds can be used. Data from sister ships can also be used to further improve accuracy. By using data from sister ships with the same length, width, and other geometric parameters, a larger data set can be made available for the analysis. Similarly, sister ships may have sailed in different draft and trim states relative to the original ship, or sailed in different parts of the world relative to the original ship, increasing the number of states that can be made available for the analysis. Sister ships should not only be geometrically similar, but also mechanically similar in terms of main engine design and arrangement so that fuel economy comparisons can be performed.

[0044] A database was then developed showing the average main engine power at various vessel speeds, drafts, and trim conditions, removing the effects of extreme weather conditions and hull fouling. Table 1 below shows an example table showing the change in main engine power when draft and trim are changed at a given speed, and illustrates how the data is processed to be placed into classes or bins as described. Although only one speed is shown in this example, the process can be repeated for a range of vessel speeds. Speed ​​is expressed in knots (symbol: [kn] or [kt]), a unit of speed at sea. A knot is defined as one nautical mile per hour, where a nautical mile is 1,852 meters. A knot equates to approximately 1.852 kilometers per hour or 0.514 m / s (so 13 knots is approximately 6.69 m / s). As shown, even if the speed remains constant, the main engine power requirements can change dramatically when draft and trim are changed. In the example in Table 1, the lowest engine power requirement (7.93 MW) is more than 35% lower than the highest engine power requirement (12.27 MW).

[0045] [Table 1]

[0046] The described averaging method provides a simple and easy to deploy solution for predicting draft and trim based on previous operational data. However, the averaging method can result in some inaccuracies, such as a 10% average error being seen. The averaging method is also limited by its inability to predict outside the data set. When a vessel sails to conditions not previously captured in the existing data, the optimal draft and trim conditions cannot be directly compared. Various mathematical prediction models have been trialled to determine the optimal draft and trim outside the existing data set and to remedy these limitations. A random forest regression model can be used to reduce the inaccuracies seen with the averaging method. A reduction in the average error from 10% to 5% is seen. However, the performance of the random forest regression model is limited when predicting outside the existing data set. An artificial neural network has been tested and implemented as a further optimization of the method. The artificial neural network has been confirmed to reduce the error from 10% to 0.5% compared to the averaging method, with a maximum error of 23% when predicting outside the data set. For the artificial neural network to perform as optimally as possible, the weather filtering step mentioned above must be ignored.

[0047] An artificial neural network (ANN) is a mathematical structure composed of nodes and connections, called edges, designed to replicate the functions of the human brain. Each network is composed of an input layer and an output layer of nodes interconnected by edges. A network may also be composed of hidden layers, which are between the input and output layers. Each node is a mathematical transfer function. Data is passed to the input layer, where the node sends a value to the output layer. This predicted value is compared to the input value to determine the level of error between the predicted value and the input value. To reduce this error, the network is trained multiple times. The network can then make predictions based on the original data set presented to it.

[0048] Random Forest Regression (RFR) is a learning model that can perform both mathematical regression and classification tasks using a technique called bootstrap aggregation or bagging. A single bootstrap aggregation occurs within a decision tree, and then a random forest regression is a combination of multiple decision trees. The goal of the decision tree is to split the training data set into two optimal child subsets. A process called tree growing aims to split the training data into branches and leaves based on this optimal splitting process, until no further splits are possible. Each split in the decision tree effectively asks a yes / no question about the data set, which leads to a predicted value based on the input data set.

[0049] Using RFR or ANN techniques to predict the output for each draft and trim condition improves accuracy compared to a simple averaging of the output values ​​found in the data set.

[0050] With the optimum draft and trim determined using either the disclosed averaging method or a mathematical model, a database of these optimum values ​​for each vessel speed is created. This data is then packaged into a usable display and installed on the ship's bridge. Depending on the infrastructure available on the vessel, this could be in the form of a paper readout, a display, a software package, or another method of presenting the data.

[0051] With the system deployed on the bridge, the vessel's current speed, draft, and trim can be measured. These can then be compared to the speed, draft, and trim stored in the system, and the optimum draft and trim for that particular speed can be displayed to the captain, or as an input to an automatic draft and trim control system. The captain, or an automatic controlled ballast control system, can then adjust the vessel to the optimum conditions to begin to achieve fuel savings. The captain, or an automatic ballast control system can then maintain the vessel close to the optimum conditions for as much of the voyage time as possible.

[0052] The term "comparison" herein above relates to implementation, not to the generation of results. Unlike prior art systems, the disclosed method and system does not use any comparison to generate results or analysis. Instead, once optimal draft and trim values ​​are generated by the disclosed process, an operator, e.g., the captain of a vessel, can check the vessel's current speed with the same speed in the results generated using the disclosed method. Thus, the operator identifies the optimal draft and trim. In effect, the captain compares the current speed with the same speed in a table (e.g., see Table 1 for an example) to look up the values. However, the disclosed method lacks a comparison step in generating results.

[0053] The methods of the present disclosure may be automated into a workflow, for example as detailed in Figure 1. The workflow 10 may include automated data collection, processing, filtering, averaging, and output to a table, web-based interface, or display for use by the vessel's captain to select optimal trim or draft conditions. The workflow is made up of several stages.

[0054] Stage 1 may include a first step 12 that involves collecting data 14 from data sources (which may include field device instrumentation and / or dedicated sensors). A second step 16 involves processing the vessel data 14 that is optimized to remove errors such as negative and extreme values ​​as described above.

[0055] Stage 2 may include either a weather filtering step 18 and an averaging step 20 as described above, or a step 22 using a mathematical model such as an artificial neural network to provide a prediction 24 of optimal draft and trim.

[0056] Stage 3 may include a step 26 of displaying the optimum draft and trim 24 determined in stage 2 for the current vessel speed to the captain and / or may be provided as an input to an automatic ballast control system. Optionally, stage 3 may include a step 28 of calculating estimated (potential) fuel savings based on the prediction of the optimum draft and trim and reporting the calculated estimated savings.

[0057] Stage 4 involves adjusting the draft and trim, either manually or by the automatic ballast control system, to the conditions displayed in stage 3, i.e., the optimum draft and trim predictions 24 .

[0058] Stage 5 may include a step 32 of calculating the actual fuel consumption saved by comparing the average fuel consumption before the optimization date with the average fuel consumption after the optimization date.

[0059] In another step, fuel savings are predicted. To predict potential savings achievable by the vessel, main engine power 50 (x-axis; e.g., expressed in [MW]) is plotted against fuel consumption 52 (y-axis; e.g., expressed in mass flow rate [kg / s]). A best-fit polynomial curve 54 can be applied to the plot, as shown in FIG. 2. The extreme outliers of the power curve or polynomial curve are filtered to improve the best fit of the power curve. The curve should show that as engine power increases, fuel consumption also increases. Using the equation of the best-fit curve 54, the relationship between fuel consumption 52 and power 50 can be approximated to calculate the fuel consumption for each average power at the optimized draft and trim settings.

[0060] The fuel consumption vs. power curve 54 can be continually updated by an automated system using real-time data from the vessel to provide an up-to-date estimate of potential savings, taking into account changes over time in the relationship between fuel consumption and power. Using the calculated fuel consumption for each non-optimum draft and trim setting, the fuel consumption difference or "delta" can be determined by calculating the difference in fuel consumption between the optimal draft and trim conditions. Through the sum of these delta fuel savings values, the total potential fuel savings achievable by the vessel is determined. This can be multiplied by the current fuel price and the number of expected sailing days per year. Factors can be incorporated to account for the percentage of time the captain chooses to implement the disclosed method, and the possibility that adverse weather conditions will prevent the use of the system. From experience, it is found that the disclosed method can be deployed by the captain for, for example, about 30-50% or more of the total sailing time.

[0061] With the disclosed method employed on board a vessel, the ship's performance can be compared before and after introducing draft and trim optimization using the same or different data sources from the vessel. The fuel consumption recorded before and after can be compared to determine the savings achieved. Trials have shown that by employing the disclosed method, fuel consumption savings of approximately 3-7% or more per year per vessel are achievable. As a further embodiment, the comparison of achieved savings can be automated as an additional part of the workflow, as detailed in Figure 1. Measured fuel savings can also be compared to predicted fuel consumption savings to validate the accuracy of the method.

[0062] The disclosed method details a process for determining optimal draft and trim to reduce fuel consumption in ballasted and loaded conditions. The process uses sources of data from one or more vessels being optimized to determine the optimal trim and draft of the vessel at any given speed. The data may be in the form of "noon" reported data, electronic radio frequency data, or another data source. Noon reported data is vessel operational data reported once a day, typically by the chief engineer, providing a once a day update on the vessel's condition.

[0063] Several filtering and data optimization processes are detailed to ensure good data standards for completing the process. Typically, data coming in from high frequency data recorders tends to be "noisy" and can result in erroneous values ​​being recorded. Filtering to remove extreme values ​​must be performed before applying the method detailed in this patent. Adverse weather that requires increased engine power to overcome increased wind and wave resistance is filtered out. Adverse weather has been determined to be sea states greater than 4, Beaufort wind scale greater than 5, and current impacts greater than 3%. Various restrictions can be applied to adverse weather without compromising the process.

[0064] Next, an averaging method is disclosed to see which trim and draft conditions of the historical data gave lower fuel consumption. During the period covered by the data, the vessel may have sailed with a draft and trim at a particular speed that required less main engine power than another draft and trim at the same speed. This is compiled into a database or matrix showing which draft and trim conditions at each speed contribute to a reduction in main engine power. The information in the database or table may then be presented to the master or serve as input to an automatic draft and trim control system. This information may be presented on paper or using an electronic, web-based, or similar type of display. Using the information regarding the optimum draft and trim for the vessel's current speed, the draft and trim may be changed to this optimum condition.

[0065] Then, as a further step in the process, fuel consumption can be predicted based on historical data to estimate the possible savings achieved by the implementation of the method. To estimate the potential fuel savings, the fuel mass flow rate of the historical data needs to be plotted against the power of the vessel, and an equation for a polynomial curve relating these two parameters can be found. This can then be used to estimate the fuel consumption for the average power calculated at each draft and trim state stored in the database. The estimated fuel consumption at each optimized draft and trim state, as well as knowledge of the number of sailing days per year and current fuel prices, can be used to predict potential savings. Data from vessels equipped with devices to measure fuel quality can also be used to improve the accuracy of the fuel savings estimate, since changes in fuel quality can affect fuel consumption. Similarly, reported information about mechanical problems or failures of the vessel's main engines can also be included, since an engine with a mechanical failure can consume too much fuel and affect the fuel consumption estimate. Further factors need to be included to consider when draft and trim optimization cannot be performed due to bad weather. It is found that draft and trim optimization cannot be performed because bad weather or other navigation priorities typically account for 30-40% of the total voyage time.

[0066] Using the same data collection methods used to calculate optimal draft and trim, actual fuel savings can be further measured. Fuel savings can be measured by comparing fuel consumption data collected before and after implementation of draft and trim optimization. Fuel savings of 3-7% per vessel per year have been measured with the implementation of this process.

[0067] The various detailed steps are incorporated into a workflow or process that may be automated to various levels. This may be performed in a semi-manual fashion using a spreadsheet application such as Microsoft Excel. To reduce the time required to perform this optimization, a programming language such as Python or C++ may be used to automate the described workflow. As a further embodiment, the output of the process may be automated such that the draft and trim are automatically changed by an automatic control system without the master's input.

[0068] One limitation of the averaging method described is that it can only give optimal draft and trim for states in which the vessel has sailed before. If the vessel then sails to a state in which it has not sailed before, the averaging system cannot advise on the optimal draft and trim. Another limitation of the averaging method is that there is an imprecision built into the analysis. An average error of 10% is seen using the averaging method. To improve this, some mathematical models can be built into the process to make predictions of optimal draft and trim based on previous performance for new, unseen states. Random forest regression and artificial neural network models have been deployed to improve predictions outside of existing data sets.

[0069] The artificial neural network was the most successful in predicting optimal draft and trim conditions outside of the existing dataset, with a maximum percentage error of 23%. The artificial neural network was also able to significantly reduce the imprecision of the analysis, reducing the average error from 10% to 0.5%.

[0070] The disclosed system and method allows reducing a ship's main engine power and fuel consumption by applying an advanced analytical process to identify optimal voyage parameters. Advice is then provided to the ship's master on which parameters (and how much) to change to optimize the voyage and reduce fuel consumption. Advice here may be for example on adding ballast water to change draft and trim or adjust speed. The end product may be a user-friendly web-based tool with a unique analytically driven output.

[0071] The disclosed method and system does not compare the measured operational data to a predetermined default or optimal data set. The disclosed method processes and analyzes historical data to calculate the optimal. The disclosed method is not a comparison method to an already optimized data set residing in the memory or other storage of the data processor, or some measured or otherwise identified data is compared to a predetermined sailing condition. Conventional methods and systems typically include a step of comparing the measured data to a pre-determined optimal base case. The processing of data in the disclosed method and system does not include a comparison, but actually a further step of processing by processing the operational data into classes of increasing size of speed, draft, and trim to determine the average power for each historical speed, draft, and trim condition. For example, each row of Table 1 corresponds to a "class" as previously described.

[0072] The system and method allows for a holistic assessment of a vessel's hydrodynamic performance, enabling the identification of sweet spots of performance not previously observed with conventional methodologies. The disclosed system and method utilizes a series of advanced algorithms to process and analyze real-time, high-frequency data recorded on the vessel, further highlighting the optimum at which the vessel can operate at a particular time and within a particular context (wind, current, etc.).

[0073] With respect to conventional systems, the disclosed system and method provides an increased number of optimized parameters, increased choice for their selection, and increased accuracy. The advanced algorithms applied in the disclosed system and method overcome the limitations of the prior art systems and can successfully predict the optimized values ​​of more parameters, providing the captain with more choice (number of interventions).

[0074] The results obtained by the data analysis have been fully tested and validated on the applicant's vessels, with a margin of error of less than about 5%. The disclosed system and method have achieved an average increase in main engine fuel efficiency of more than about 3% on typical cargo ships, including but not limited to crude oil tankers (MR tankers, VLCCs, etc.), LNG carriers, LPG carriers, bulk carriers, container ships, etc.

[0075] The present disclosure is not limited to the embodiments as described above and in the appended claims, many modifications are possible herein and the features of each embodiment can be combined.

Claims

1. 1. A method for determining optimum trim and draft for a vessel in ballasted and loaded conditions using analysis of historical vessel data for the vessel being optimized, comprising: - collecting operational data from said vessel for one or more of its previous voyages, said operational data including one or more operational parameters or data tags; - filtering out errors and noise in said operational data; - filtering out the effects of bad weather, hull fouling and / or other conditions that have been found to reduce the accuracy of the process; - processing the filtered out operational data to generate processed operational data by placing the filtered out operational data into different bins for speed, draft and trim and determining an average power in each bin; Generate bins with different speed, draft, and trim magnitudes for each increment, placing the filtered out operational data including speed, draft, and trim into the bin in which speed, draft, and trim are closest; Calculate the average power in each bin, Steps and - creating a database of optimal draft and trim based on said processed operational data and providing said database to the pilot or as input to an automatic draft and trim optimization system; - calculating a predicted fuel consumption at each speed, draft and trim based on said processed operational data to estimate achievable fuel savings; - comparing said predicted fuel consumption with the achieved fuel consumption of said vessel to determine the savings achieved using information about current fuel prices; A method comprising:

2. The method of claim 1, wherein the step of filtering out errors and noise in the operational data includes removing data whose values ​​are outside an upper threshold and / or a lower threshold to remove zero values ​​and / or one or more values ​​other than physically possible values.

3. A method as described in claim 1 or 2, wherein the step of filtering out errors and noise in the operational data filters out data outside a set range from the operational data.

4. The range is selected based on weather conditions when the current impact is less than 3%, the sea state is less than 4, and the Beaufort wind scale is less than 5, and the impact of the false accusation is [0010] The method of claim 3, wherein the calculated

5. 5. The method of claim 1, wherein the operational parameters include one or more of: draft, trim, date and time of sample taken, speed on land, speed in water, main engine power, main engine rpm, true wind speed, relative wind angle, engine fuel mass flow, fuel consumption, water depth, shaft rpm, and time since last hull wash.

6. A method as described in any one of claims 1 to 5, comprising a step of displaying real-time results to the pilot or to an automatic draft and trim control system using a graphical display.

7. The method of claim 6 , wherein the graphical display provides the operator with the optimum draft and trim of the vessel for modifying the current state of the vessel.

8. A method according to any one of claims 1 to 7, comprising using data relating to fuel quality and primary engine mechanical failures to refine the predicted fuel consumption for savings estimation.

9. 9. The method of claim 8, including using an artificial neural network to analyze and provide fleet-wide draft, trim, and speed optimization.