Search

Supercomputer power has been doubling approximately every 14 months for several decades, increasing the capabilities of scientific modelling at a similar rate. However, to utilize these machines effectively for applications such as computational fluid dynamics, improvements to strong scalability are required. Here, the particular focus is on semi-implicit, viscous-flow CFD, where the largest bottleneck to strong scalability is the parallel solution of the linear pressure-correction equation — an elliptic Poisson equation. State-of-the-art linear solvers, such as Krylov subspace or multigrid methods, provide excellent numerical performance for elliptic equations, but do not scale efficiently due to frequent synchronization between processes. Complete desynchronization is possible for basic, Jacobi-like solvers using the theory of ‘chaotic relaxations’. These non-deterministic, chaotic solvers scale superbly, as demonstrated herein, but lack the numerical performance to converge elliptic equations — even with the relatively lax convergence requirements of the example CFD application. However, these chaotic principles can also be applied to multigrid solvers. In this paper, a ‘chaotic-cycle’ algebraic multigrid method is described and implemented as an open-source library. It is tested on a model Poisson equation, and also within the context of CFD. Two CFD test cases are used: the canonical lid-driven cavity flow and the flow simulation of a ship (KVLCC2). The chaotic-cycle multigrid shows good scalability and numerical performance compared to classical V-, W- and F-cycles. On 2048 cores the chaotic-cycle multigrid solver performs up to faster than Flexible-GMRES and faster than classical V-cycle multigrid. Further improvements to chaotic-cycle multigrid can be made, relating to coarse-grid communications and desynchronized residual computations. It is expected that the chaotic-cycle multigrid could be applied to other scientific fields, wherever a scalable elliptic-e

Since 2004, supercomputer growth hasbeen constrained by energy efficiency rather than raw hardware speeds. Tomaintain exponential growth of overall computing power, a massive growth inparallelization is under way. To keep up with these changes, computationalfluid dynamics (CFD) must improve its strong scalability – its ability tohandle lower cells-per-core ratios and achieve finer-grain parallelization. Amaritime-focused, unstructured, finite-volume code (ReFRESCO) is used toinvestigate the scalability problems for incompressible, viscous CFD using two classicaltest-cases. Existing research suggests that the linear equation-system solveris the main bottleneck to incompressible codes, due to the stiff Poisson pressure equation. Here, these results are expandedby analysing the reasons for this poor scalability. In particular, a number ofalternative linear solvers and preconditioners are tested to determine if thescalability problem can be circumvented, including GMRES, Pipelined-GMRES,Flexible-GMRES and BCGS. Conventional block-wise preconditioners are tested,along with multi-grid preconditioners and smoothers in various configurations.Memory-bandwidth constraints and global communication patterns are found to bethe main bottleneck, and no state-of-the-art solution techniques which solve thestrong-scalability problem satisfactorily could be found. There is significantincentive for more research and development in this area.

Some species of fish are able to alter their mode of swimming to interact with naturally produced vortices; the use of these gaits reduces the energy expended by the fish. To analyse the feasibility of autonomous underwater vehicles (AUV) replicating these gaits, a series of experiments are performed with unpowered rigid and flexible bodies positioned in the Kármán wake of a rigid cylinder. Simple motion capture techniques are used to capture the bodies’ lateral and upstream motion in the flow. The results demonstrate that manufactured bodies are capable of passively mimicking fish behaviours, to a limited extent. More importantly, it was concluded that while significant upstream movement was possible for a manufactured object, it was achievable irrespective of the stiffness of the material. For AUVs operating in unsteady flow regimes an ability to utilise energy saving gaits may improve the range or operational time.

This paper investigates the influence of the propeller race on upstream and downstream self propelled AUVs. Initially simulations of a self-propelled hull are performed at the Reynolds Number 3.2x10^6 with software commercial RANS code ANSYS CFX 12.1, utilising a body force model to replicate the impact of the propeller utilising momentum source terms. This is then extended to consider a fleet of two self-propelled vehicles operating at a range of longitudinal offset and transverse separations. The results highlight that operation in close proximity to another self-propelled vessel has a significant impact of both the flow around the hull and drag experienced by the vehicle. A propeller race deduction is proposed to account for the increase in vehicle drag due to the propulsors of other vehicles. The propeller race deduction is dependent upon both longitudinal and transverse separation. From a vehicle or mission design perspective, it is important to correctly understand the true propulsive energy budget of the vehicle and its impact on both range and endurance. This study highlights the importance of considering both thrust deduction and any propeller race deductions when calculating the propulsive power consumption of an individual or fleet of vehicles.

Recent high profile ship collision and grounding events show that such incidents occur with a higher frequency than they should; hence the survivability of ships subject to damage needs to be better understood. The long-term aim of this project is to develop a numerical, hydroelastic model that can predict the survivability of a damaged ship. This paper presents the results of the initial development of a two-dimensional hydroelastic model to numerically model the influence of abnormal loading due to floodwater ingress on the motions and global loads of a ship in regular waves using a quasi-static approach. Predictions are validated using experimental data. Results presented show promising agreement between experimental measurements and numerical predictions. The ability to carry out a systematic study of the influence of damage location, severity and ship speed on the effect of damage on a vessel is demonstrated. Future work will investigate the inclusion of both transient flooding and three-dimensional effects.

The range of an AUV is dictated by its finite energy source and minimising the energy consumption is required to maximise its endurance. One option to extend the endurance is by obtaining the optimum hydrodynamic hull shape with balancing the trade-off between computational cost and fluid dynamic fidelity. An AUV hull form has been optimised to obtain low resistance hull. Hydrodynamic optimisation of hull form has been carried out by employing five parametric geometry models with a streamlined constraint. Three Genetic Algorithm optimisation procedures are applied by three simple drag predictions which are based on the potential flow method. The results highlight the effectiveness of considering the proposed hull shape optimisation procedure for the early stage of AUV hull design.

A sprint kayak experiences an unsteady flow regime due to the local influence of the paddle. However, kayak designs are usually optimised for steady-state, naked hull resistance. To determine whether unsteady paddle effects need to be included in kayak design, the hydrodynamic interactions between a kayak paddle and a hull are assessed using computational fluid dynamics. A body force model of a drag-based paddle stroke is developed using a blade element approach and validated against experimental data. This allows the paddle-induced local velocities to be simulated without the need to fully resolve the detailed flow around a moving paddle geometry. The increase in computational cost, compared to the naked hull simulation, is 8%. A case study investigating the impact of different paddle techniques on the hydrodynamic forces acting on a self-propelled kayak is conducted. A 0.23% difference in self-propelled resistance was observed, while an estimated 0.5% additional increase can be attributed to paddle-induced draught increases. An estimate of small changes in resistance on race times indicates that reductions of even a fraction of a percent are worth pursuing, indicating that the developed methodology may provide a useful design tool in the future.

In this work model predictive control is used to provide transit and hover capabilities for an autonomous underwater vehicle where the description of the system dynamics used include terms measured experimentally. The resulting controller manoeuvres the vehicle in the presence of constraints on the actuators and results obtained from the deployment of the vehicle in an inland lake for the study of the Zebra mussel, an invasive species, are also given.

In the application of an autonomous underwater vehicle a critical requirement is to keep the level of the actuation signals within operational limits to avoid, for example, actuator nonlinearities and reduce peak power consumption. The most common approach to this problem for AUVs that have been deployed is, if required, to trade-off performance in order to keep the actuation signals and power required within the operational limits. This paper addresses depth control of an AUV using model predictive control with constraints on the both the amplitude and rate of change of the entries in the control vector. The model predictive control algorithm is designed by solving a quadratic programming problem in real-time when implemented on an AUV prototype. Experimental test results for depth control are also given and demonstrate that physically relevant constraints on the thrust and actuation power, critical factors for the use of these vehicles, can be achieved. Moreover, there is agreement between the control action used and the underlying physics of a body moving in water.

The purpose of this paper is to provide guidance for operators on suitable spacing for multiple vehicles’ missions. This paper investigates the combined drag of a pair of propelled prolate spheroids and compared to the towed models for the length Reynolds Number of 3.2×106. The model has a length-diameter ratio of 6:1. A series of configuration of a pair of spheroids is simulated at various longitudinal offsets and transverse separations. Three-dimensional simulations are performed using a commercial Reynolds Averaged Navier Stokes (RANS) Computational Fluid Dynamics code ANSYS CFX 12.1 with the SST turbulence closure model. In each case, the fluid domain has a mesh size of approximately nine million cells including inflated prism layers to capture the boundary layer. Mesh convergence is tested and then validated with wind tunnel test results. The drag of each spheroid is compared against the benchmark drag of a single hull. The three-dimensional cylinder is modelled to simulate the thrust distribution of propeller. The drag of the propelled model is compared against the single bare hull model. The results show that the transverse separations and longitudinal offsets determine the interaction drag between both hulls. The increasing of separation results in lower interference drag. The decreasing of offset results in higher drag reduction. By implementing the body force propeller, the combined drag and drag of the follower is interfered by the accelerated flow. Based on the numerical information, operators can determine the optimal configurations in transvers separation and longitudinal offset based on energy considerations.

In this work a depth and pitch controller for an autonomous underwater vehicle (AUV) is developed. This controller uses the model predictive control method to manoeuvre the vehicle whilst operating within the defined constraints of the AUV actuators. Experimental results are given for the AUV performing a step change in depth whilst maintaining zero pitch.

Taking inspiration from nature the NEMOdeep vehicle described in this paper is being developed as a laboratory demonstrator to showcase potential technologies to achieve small deep diving autonomous underwater vehicles. The design of the vehicle is a hybrid of conventional AUV components with engineered analogues of marine animal organs. The internal structure is comprised of a spine, ribs and sternum, which support a hydrodynamic fairing or ‘skin’. The ‘brain’ and actuators are developed using pressure tolerant electronics, negating the need for pressure vessels. The pressure tolerant electronics are surrounded by light mineral oil which transmits the hydrostatic pressure uniformly to the components while also providing buoyancy. The internal fluid is more compressible than seawater and the batteries and internal structure are less compressible than seawater: a ‘Buoyancy/Compressibility’ organ is being developed to passively maintain neutral buoyancy over the 0 to 6000m depth range.

The range of physiological adaptations possessed by marine animals allowing them to successfully operate in the marine environment is a plentiful source of inspiration for the designers of Autonomous Underwater Vehicles. This chapter compares the total energetic cost of straight line swimming for both marine animals and AUVs, using cost of transport (COT) as a comparative metric. COT is a normalised measure of the energetic cost of transporting the animal’s or vehicle’s mass over a unit distance. It includes non propulsion power requirements as well as considering the energy lost by actuators and mechanical couplings and energy lost in the wake. Comparisons presented in this chapter show that marine animals typically have higher optimum COT than engineered systems of equivalent size. However parallels may be drawn, for example, to increase range both marine animals and AUVs appear to favour reducing non-propulsion power costs and travelling slowly to ensure operating at the minimum COT.

The maneuvering performance of an over-actuated AUV (2m long, 254mm diameter) has been tested operating on the free surface (97% displacement submerged). Straight line speed tests and turning circle tests have been performed. Both rear control surfaces and through-body tunnel thrusters were used to maneuver the vehicle. The turning circle diameter, turning rate and forward velocity data is presented with rudder, thruster and hybrid tests. The rudder actuated turning diameters (TD) reduced as forward speed increased. The thruster actuated TD are smaller than the rudder actuated and the TD for thruster actuated increases with forward speed. For the smallest TD, a hybrid system using both the rudders and thrusters were used. An inverse open loop control algorithm is presented that delegates actuator set-points for a given propeller set-point and TD

The missions being proposed for autonomous underwater vehicles (AUVs), by both marine scientists and industry, are becoming increasingly complex and challenging. In order to meet these demands the next generation of AUVs will need to be faster, to operate for longer durations, and to be more manoeuvrable than existing vehicles. It is therefore vital that the hydrodynamic forces and moments acting on a self-propelled manoeuvring AUV can be predicted accurately at the initial design stage. In order to achieve this, the use of a computational-fluid-dynamics-based analysis is suggested. The approaches developed are predominantly steady state and suitable for running on a workstation personal computer using a commercial software licence. It is estimated that the proposed simulations would take a competent user less than 1 month for a new concept design. The total cost of these simulations is significantly lower than the cost of building a model and having it commercially tested to capture the same level of detail for the resistance, propulsion, and manoeuvring performance. Based on the validation studies presented, it is estimated that on a 2×106 element structured mesh a competent user should be able to predict hydrodynamic forces to within at least 10 per cent and moments to within 20 per cent of in-service performance.

It has long been understood that swimming marine animals have evolved capabilities in terms of speed, manoeuvrability, and efficiency which are desirable for underwater vehicles. Despite this, solutions inspired by nature, or bio-inspiration, are very rarely applied to solve engineering challenges. In particular, it is understood that fish have the ability to alter their mode of swimming to interact with naturally produced vortices as a method of conserving energy and in certain instances extracting energy from a flow. This paper considers whether a bio-inspired flexible autonomous underwater vehicle (AUV) could exploit unsteady flow features to reduce its cost of transport. An analytical model is developed which allows an AUV designer to predict which flow frequencies excite the natural vibration modes of a flexible cylinder. It is demonstrated that by placing a flexible cylinder in an unsteady flow, such as downstream of a bluff body, a similar mechanism to that used by fish may be exploited to move the cylinder upstream with no power input except that extracted from the flow.

The use of an unsteady computational fluid dynamic analysis of the manoeuvring performance of a self-propelled ship requires a large computational resource that restricts its use as part of a ship design process. A method is presented that significantly reduces computational cost by coupling a blade element momentum theory (BEMT) propeller model with the solution of the Reynolds averaged Navier Stokes (RANS) equations. The approach allows the determination of manoeuvring coefficients for a self-propelled ship travelling straight ahead, at a drift angle and for differing rudder angles. The swept volume of the propeller is divided into discrete annuli for which the axial and tangential momentum changes of the fluid passing through the propeller are balanced with the blade element performance of each propeller section. Such an approach allows the interaction effects between hull, propeller and rudder to be captured. Results are presented for the fully appended model scale self-propelled KRISO very large crude carrier 2 (KVLCC2) hull form undergoing static rudder and static drift tests at a Reynolds number of 4.6×106 acting at the ship self-propulsion point. All computations were carried out on a typical workstation using a hybrid finite volume mesh size of 2.1×10^6 elements. The computational uncertainty is typically 2–3% for side force and yaw moment.