An automated optimisation of a fully parametric vessel for real world conditions.

Abstract

New environmental regulations and fluctuating fuel prices have motivated studies to reduce carbon footprints and fuel consumption of a vessel. It is therefore naval architects' responsibility to design an eco-friendly and energy efficient ships. More radical designs are stretching the boundaries of conventional ship geometry and increasing the need for generating a more sophisticated ship analysis method.
Hull form optimisation is known as the most effective technique as a means of improving energy efficiency of vessels. Through careful and precise adjustments in the flow exposed hull region significant improvement can be achieved. An automated optimisation can produce 3 to 5% gains in energy efficiency of an advanced design. In fully-parametric modelling the entire ship geometry is described by a set of parameters. A hierarchical model is then set to define the importance of each parameter. This parametric model can be regarded as a tool which takes parameters as inputs and gives a shape as an output. In terms of an automated optimisation a fully parametric model is key as it allows both large changes in the early design stage and minor adjustments when refining the optimum design.
The main aim of this project is to investigate a fully automated shape optimisation through a combined Computational Fluid Dynamic (CFD) and Finite Element (FE) analyses to improve ship efficiency. This project will contribute to help the industry to achieve the forthcoming IMO regulations and safeguard the environment for future generations. The fully automated hull optimisation technique to be developed will be easily applied to any vessel type to rapidly optimise the hull shape.
Applicants should have a bachelor degree in naval architecture with at least a good 2.1 degree and also should have (or expect to obtain by the start date) a Master's degree in a naval architecture or a related subject. Some experience of CAD optimisation programming (e.g. CAESES, formerly known as FRIENDSHIP-Framework) would be an advantage but is not essential. The project requires a mixture of skills, including numerical and experimental fluid dynamics, ship structures, computer programing and basic statistics.

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/N509760/1 01/10/2016 30/09/2021
1811031 Studentship EP/N509760/1 01/10/2016 31/08/2020 Anthony Romanowski
 
Description The research has developed a novel approach to modelling and simulating irregular seas within Computational Fluid Dynamics. This is the first know method to be proposed within this research area. The method allows advanced simulations to be conducted, enabling a much greater and realistic creation of simulations of the meantime environment. This has wide ranging advantages to the industry, including a greater accuracy in calculating the forces acting on the marine structures. This enables an advanced evaluation of the structures design at an early stage, thus significantly improving the structures overall life planning and in turn reducing the overall environmental impacts. There are further advantages to such simulations, these include accurate real world motion analysis, fatigue analysis and operability studies, all of which can significantly help to reduce the overall structures negative environmental impact.

From this point the analysis moved to focusing on vessel maneuvering characteristics in varying conditions. This section focused primarily in developing a fully parametric maneuvering geometry setup to enable an automated CFD setup to be produced. This setup took advantage of the key lessons learnt from the initial irregular sea analysis and built of the methodology developed in it. To ensure a stable and accurate CFD simulation was created, an extensive verification and validation study was conducted based of the model tests conducted for SIMMAN 2014. These ensured the simulation setup were accurate and stable, this in turn enabled an extension of the work conducted by the model tests. This initially included adding an extra degree of freedom to the model to compare the vessel responses and forces withe the model tests run. In addition to this, the CFD enabled an even greater step forward, and allowed an in depth analysis into the effects of varying regular wave conditions on the vessels maneuvering characteristics. This also tested the complete parametric geometry setup by now including waves, and thus utilizing the methods developed in first section.

This will now be taken further by developing a fully parametric hull based on the NPL series to conduct the optimization with. This is currently under development and being refined at the moment.

This approach is the first key step in completing the whole phd research completed through this award. Further to this, a journal paper has been published presenting the first approach.
Exploitation Route The findings may be used to reduce the overall life time environmental impact by accurately modelling a marine structures power requirements, motions and operabilty at the early stages of design. This enables the design to actively improve these factors through optimization, and thus maintain the desired design requirements while reduce any negative environmental impact.

The findings can also be used to retro fit and improve existing vessels to reach ever changing and demanding environmental requirements defined by governments and maritime bodies.
Sectors Aerospace, Defence and Marine,Environment
 
Title Applied simulations with various vessels 
Description To further investigate the accuracy and reliability of the methodology, multiple simulations were created to model the resistance and motions of vessels moving within an irregular sea state. This stage was split into two, however both stages followed the same procedure and thus are combined in this one description. Two vessels were modelled, a small catamaran and large containership (KCS). Both simulations were modelled in fully scale, as the method proposed used fully scale domains, as well as having zero scaling factor. The simulations and models focused on the integration of overset regions with solid bodies capable of motion due to the waves being simulated. The catamaran simulation focused on the smaller wave heights and higher speeds. This significantly increased the complexity and size of each simulation, as well as increasing the overall run time due to the introduced body. 6 Initial test simulations were created, each simulation equated to around 20Gbs in size, along with around 1Gbs in related data, primarily images and graphical data. These simulations defined the final simulations, which consisted of three main variants, each of which had 6 auto saved simulations. Theses had a final simulations size of 21Gbs, and 1.1Gbs of related data. This in turn equated to 27 simulations and 590Gbs of data at the final output time. The KCS simulations were smaller due methodologies effect of smaller simulations for larger waves. A smaller number of simulations were run for this body, a total of only 9 simulations were run, each equation to 18.5Gbs, and 1.2Gbs of related data. Thus a total of 177.3Gbs of data was simulated and saved for this investigation. This finally comes to a total of around 780Gbs of data for all the simulations and related data, including MATLAB investigations and EXCEL sheets. It is important to note as well at this point that the data found and compiled was not included in the paper, currently with editor at time of writing. However, these been vital to the next stages of the project and award. 
Type Of Material Computer model/algorithm 
Year Produced 2018 
Provided To Others? No  
Impact The data and models allowed a clear evaluation on the next stages, along with the key changes required to improve and define the key points in next stage of the award in investigating maneuvering of ships in waves. 
 
Title Automated Volume Generation 
Description To accompany the CFD simulations, a further program is used to develop a fully parametric setup for generating the key volumes for ensuring stable simulations. This is a vital point in the entire process as it will key to the research and completed work. The file itself does not exceed 0.25Gbs, but the generated data equates to 0.5Gbs. This data is produced for every simulation using hull and body geometry. Therefore this file has produced data for 166 simulations and intern equates to 83Gbs of data. In addition the main file has been modified and varied 15 times throughout the process, producing 3Gbs of data. 
Type Of Material Computer model/algorithm 
Year Produced 2018 
Provided To Others? No  
Impact This simulation is the overall key to final project and the fully automated process. It is therefore constantly being updated and modified to accommodate new simulations. In addition, the fully parametric ship will be modeled within this file, thus creating the fully automated setup for this award and investigation. 
 
Title Irregular sea initial simulation set 
Description The data set is composed multiple simulations focusing on the first iterations of simulating irregular seas within CFD. These simulations applied the known methods of modelling waves within CFD, and thus acted as the baseline simulations for future work to be compared with. The data set investigated multiple sea states to develop a clear foundation to work off. All simulations failed to match the significant wave height defined, thus highlighting and justifying the need to develop the new methodology. Each simulation was around 8Gb, with around of 0.25Gb related data. Three sea states were tested, with 2-3 simulation variants produced for each. The final simulation variants had a further 3 copies produced when running and auto saving. A total of 19 simulations were saved and complied for this initial investigation, equating to around 175Gb of simulation and data being saved. 
Type Of Material Computer model/algorithm 
Year Produced 2017 
Provided To Others? No  
Impact This data set defined the initial plan and further investigations into irregular sea simulations and thus defined the first key section of research conducted under this award. 
 
Title Maneuvering Gen 1 
Description This data set include the initial generation 1 of the maneuvering simulations attempting to integrate overset region and a user defined PMM path. The data set includes around 25 simulations, each evolving and varying in attempt to successfully couple and write a user defined PMM path and overset region. The simulations varied in size, ranging from 2.5Gbs to 9Gbs in size. The total accumulation size of the data set came to around 150Gbs. These simulations failed to achieve the desired result and thus were not taken forward using this process in the later sections. This however was key in defining the next stages and the next group of simulations to be modelled and the methodology to be used in linking the PMM and Overset regions. 
Type Of Material Computer model/algorithm 
Year Produced 2018 
Provided To Others? No  
Impact Although these simulations failed to achieve the goal aimed for them, they provided the key information in defining the generation 2 simulation methodology, and thus the successful simulations in generation 2. 
 
Title Maneuvering Gen 2 (A) 
Description This data set is comprised of the methodology developed to overcome the issues found in generation 1 as well as the focus on the KCS hull form. The method uses the PMM setting within Star CCM+ but only directed at the overset to overcome the combination issues. The set is built up from three initial simulations, each have 5 variations focusing on different factors effecting the simulation, the most accurate and stable simulation was taken forward and had a further 8 simulations developed from it. This included and investigation into the interpolation methods used for overset in Star CCM+. 6 out of the 8 simulations were ran and auto saved 3 times extra for each. The best simulation is being tested at various speeds and displacements, these equate to a further 20 simulations, thus bringing the total up to 61 simulations. These simulations vary in size from 2-7Gbs, with a total accumulated data size of 320Gbs. 
Type Of Material Computer model/algorithm 
Year Produced 2018 
Provided To Others? No  
Impact This data set is key to validating the methodology compared with towing tank tests, and thus ensuring accurate results will be found when moving away from towing tank tests. This is the most vital stage in ensuring the methodology is valid for investigating the maneuvering in waves. 
 
Title Maneuvering Gen 2 (B) 
Description This data set is comprised of the methodology developed to overcome the issues found in generation 1 as well as the focus on the DMTB hull form. The method uses the PMM setting within Star CCM+ but only directed at the overset to overcome the combination issues. The set is built up from the optimum simulation found in Gen (A). The simulation is then modified and varied a total of 10 times to create a stable simulation. Once the simulations is found that is stable and accurate, another interpolation investigation is conducted creating another 6 simulations, each saved a further 3 times when running. These are currently running and therefore no more simulations and models have been made. This comes to a total of 29 simulations, equating to a total of 165Gbs of data. 
Type Of Material Computer model/algorithm 
Year Produced 2019 
Provided To Others? No  
Impact This data set is key to validating the methodology compared with towing tank tests, and thus ensuring accurate results will be found when moving away from towing tank tests. This is the most vital stage in ensuring the methodology is valid for investigating the maneuvering in waves. 
 
Title Maneuvering Gen 3 
Description This data set is comprised of the methodology developed to overcome the issues found in generation 1 as well as the focus on the DMTB hull form. The method uses the PMM setting within Star CCM+ but only directed at the overset to overcome the combination issues. The set is built up from the optimum simulation found in Gen 2. The simulation is then modified and varied a total of 3 times to create a stable simulation. Once the simulations is found that is stable and accurate, another interpolation investigation is conducted creating another 20 simulations, each saved a further 3 times when running. This comes to a total of 63 simulations, equating to a total of 360Gbs of data. 
Type Of Material Database/Collection of data 
Year Produced 2019 
Provided To Others? No  
Impact A stable, highly accurate and fast running parametric setup, that allowed for a comprehensive verification and validation study to be conducted 
 
Title Maneuvering Test 1 
Description The data set is composed of multiple simulations focusing on the first test iterations for simulating PMM simulations within CFD. These simulations investigated the integrated setup within the CFD code Star CCM+. These simply tested the method and the limitations of the system. These simulations acted as the foundation for the primary large maneuvering simulations for later. Around 15 simulations were conducted in total, each around 4.5Gbs in size, equating to a final data set size of 70Gbs. 
Type Of Material Computer model/algorithm 
Year Produced 2017 
Provided To Others? No  
Impact This data set defined foundation of the maneuvering simulations, which was further evolved and investigated in initial ship and towing tank simulations. 
 
Title Maneuvering Test 2 
Description This data set is comprised of the methodology developed to overcome the issues found in generation 1 as well as the focus on the DMTB hull form. The method uses the PMM setting within Star CCM+ but only directed at the overset to overcome the combination issues. The set is built up from the optimum simulation found in Gen 3. The simulation is then modified and varied to add an extra degree of freedom. This is tested a total of 6 times to ensure stability in varying conditions. Once the simulations are found to be stable and accurate, an investigation is conducted to compare the CFd with model testes, this created around 45 simulations, each saved a further 3 times when running. This comes to a total of 63 simulations, equating to a total of 750Gbs of data. Not all data has been kept, with a final data total closer to 400Gbs. 
Type Of Material Database/Collection of data 
Year Produced 2019 
Provided To Others? No  
Impact The ability to compare how varying levels of degrees of freedom have on the overall maneuvering characteristics, and in turn how they impact the maneuvering coefficients. 
 
Title Maneuvering Test 3 
Description This data set in the most comprehensive set out of all the maneuvering data sets. The method uses the PMM setting within Star CCM+ but only directed at the overset to overcome the combination issues. The set is built up from the optimum simulation found in Gen 3. The simulation is then modified and varied to add an extra degree of freedom. From here the simulations added waves and varied the DOF when running in waves to build a complete and comprehensive picture of how waves effect maneuvering ships. Validation simulations were run with a total of 15 simulations, ensure stability in varying conditions. Once the simulations are found to be stable and accurate, an investigation is conducted to compare the CFD with model testes, this created around 57 simulations, each saved a further 3 times when running. This comes to a total of 63 simulations, equating to a total of 1Tb of data. Not all data has been kept, with a final data total closer to 650Gbs. 
Type Of Material Database/Collection of data 
Year Produced 2019 
Provided To Others? No  
Impact The key results shown from this data allow a comparison and investigation into how waves effect maneuvering characteristics of ships. In addition to validating the methodology produced in the previous sections. 
 
Title Methodology Generation 1 Simulations and Data 
Description The data set is composed multiple simulations focusing on the first iterations of the stage one proposed methodology for simulating irregular seas within CFD. These simulations applied the known methods of modelling waves within CFD with a focus on greater refinement through the use of statistical break down of irregular wave data. The data set investigated multiple sea states to develop a comparison to the control work. The methodology showed improvement, but still failed to meet the desired accuracy. Each simulation was around 12Gb, with around of 0.15Gb related data. Three sea states were tested, with 3-4 simulation variants produced for each. The final simulation variants had a further 5 copies produced when running and auto saving. A total of 27 simulations were saved and complied for this initial investigation, equating to around 335Gb of simulations and data being saved. 
Type Of Material Computer model/algorithm 
Year Produced 2017 
Provided To Others? No  
Impact This data set defined generation one of the methodology, which was further evolved and investigated in generation 2 by using the most accurate simulations in this data set. 
 
Title Methodology Generation 2 Simulations and Data 
Description The data set is composed multiple simulations focusing on the second iterations of the stage one proposed methodology for simulating irregular seas within CFD. These simulations applied the methods proposed and evaluated in generation 1, however a greater focus was applied to the statistical evaluation and breaking down irregular sea states. The data set investigated a single sea state to develop a comparison to the control work and generation 1. The methodology showed significant improvement over generation 1, but was much larger in simulation size and running time. In addition to the CfD simulations, the statistical investigation was conducted through the use of MATLAB and an in house code. This produced further data along with the CFD simulation outputs. Each simulation was around 19Gb, with around of 0.15Gb related data. A single sea state was tested, with 8 simulation variants produced for each. The final simulation variants had a further 3 copies produced when running and auto saving. A total of 12 simulations were saved and complied for this initial investigation, equating to around 250Gb of simulations and data being saved, including the MATLAB statistical data. 
Type Of Material Computer model/algorithm 
Year Produced 2017 
Provided To Others? No  
Impact This data set defined generation 2 of the methodology, which was further evolved and investigated in generation 3 by using the most accurate simulations in this data set, along with the corresponding MATLAB simulations. 
 
Title Methodology Generation 3 Simulations and Data 
Description The data set is composed multiple simulations focusing on the final iterations of the previous stages of proposed methodology for simulating irregular seas within CFD. These simulations applied the methods proposed and evaluated in all the previous generations, however a greater focus was applied to the stability and speed up times for each simulation. The data set investigated a three sea states to develop a comparison to the control work and generation 1 and 2, as well as produce data to be presented in the submitted paper, currently with editor at time of writing. The methodology showed significant improvement over generation 1 and 2, due to increased cell number from 1 while reducing from 2, thus improving the overall reliability of each simulation. Each simulation was around 14Gb, with around of 0.15Gb related data. Three sea states was tested, with 6 simulation variants produced for each. The final simulation variants had a further 6 copies produced when running and auto saving. A total of 30 simulations were saved and complied for this initial investigation, equating to around 425Gb of simulations and data being saved. 
Type Of Material Computer model/algorithm 
Year Produced 2018 
Provided To Others? No  
Impact This data set defined generation 3 of the methodology, which was further evaluated and prepared for presentation. Generation 3 was the final step in completing the fully automated irregular sea set up, and the initial stage in the overall project. 
 
Title Parametric hull generation 
Description To accompany the CFD simulations, a further program is used to develop a fully parametric setup for generating the key volumes for ensuring stable simulations. This is a vital point in the entire process as it will key to the research and completed work. The initial file itself did not exceed 0.25Gbs, but the generated data equates to 0.5Gbs. From here the file has been extended and add to multiple times to produce the fully parametric hull. This addition increased the overall size by 50%, leading to a final file around 0.75Gbs in size. Around 17 total file iterations have been created. This file was also used to test the optimization algorithm, and was run multiple times producing extensive results. This led to a total data set size of around 20Gbs. 
Type Of Material Computer model/algorithm 
Year Produced 2019 
Provided To Others? No  
Impact The development of a fully parametric hull, that has parametric CFD volumes directly linked and thus produces the key foundation for an automated analysis in any desired condition.