Computational Fluid Dynamics (CFD)
Simulating flow of a fluid such as air or water around an object with a computer program
CFD can provide both qualitative and quantitative results. As well as hard data about the forces, velocities and pressures, we can use a number of visualisations to indicate their distributions or variation over time. Displaying stream lines and pressure distributions gives an insight into the flow’s behaviour and indicates where geometry changes might improve the forces or flow characteristics – such as stall angle or flow separation.
The process involves setting up a domain – or volume to analyse – and setting the flow conditions at the domain boundaries. Usually this is an inlet flow speed, and outlet and some constraints at the sides, top and bottom of the domain. Then we locate the object to be analyses in the domain, oriented to the flow according to what we want to analyse.
A boundary condition is applied to the surfaces of the model that are of interest – the keel, bulb or hull and so on – and then the model is meshed. Meshing generates a large number of nodes and element faces that are used to represent the fluid and the energy, direction and speed, within it. We generally model hulls with a mesh size between 50mm and 100mm, and our CFD models range from around 2 million elements to up to 8 million, depending on the mesh size and density.
Some of our recent projects include:
Aeronautical and Automotive
The investigation into the helicopter airframe was to provide two key pieces of data – the total drag of the airframe and pressures on various panels, all at a range of yaw angles and speeds. The pressure – and its distribution – over the glazing/door panels can be used to specify glazing thickness and the supporting structure. With accurate loads defined the structural aspects can be designed more efficiently, saving weight – always a benefit in an aircraft. An automotive company, making a replica E-type Jaguar, wanted to explore the addition of a rear diffuser to help improve the down-force and hence road holding of this classic 1960’s sports car. With over double the original’s horsepower installed, speeds of over 200kmh were being attained, so some updating of the aerodynamics was in order.
Amphibious craft launch simulation
One analysis that sits squarely in between quantitative and qualitative is the investigation in to launching the ARC600 Amphibious Rescue Craft. Exploring a range of gradients and entry speeds to determine when the bow is inundated allows the envelope of safe operations to be entered in to the owner’s manual. While this study was done retrospectively, the ability to define the safe limits for slope and speed for another amphibious craft will inform the design at the early stages and help define sheerline/bow freeboards early in the process. We started with a 2D study to ensure the range of motions was as expected, then moved to a 3D study. The last image is a screen grab from a video of the craft launching as modelled.
Headseas – behaviour in waves
The ability to model a vessel’s behaviour in headseas is important for both predicting the added resistance in waves but also the design of the forward sections and, in the case of a multihull, the wing-deck clearance and extents forward. The increase in drag due to pitching in a seaway can be a large proportion of the calm water resistance, requiring additional horsepower to be installed. The amount of bow flare, as well as chine height and width, can have a bearing on the pitch response and thus both comfort and resistance. The CFD analysis can indicate panel pressures for engineers to incorporate in their structural calculations, and the output of pitch and heave over time allows accelerations to be calculated to assess passenger/crew comfort levels. While we cannot model irregular waves, we can define a sinusoidal wave form, specifying the wave amplitude, wave length and period of encounter, so an approximation to WMO sea states can be modelled. Animations of a 19m ocean rowing skiff in a chop can be seen here and a 56ft power cat here.
Free Surface – the wave pattern
One of the more interesting outputs for hull design is the free surface – or a representation of the wash or wake pattern. Along with the drag forces (which determine the amount of power required) the free surface helps us with the less numerical aspects such as locating chines at the most effective positions according to the way the bow wave is formed. We can look at the flow with a pressure map, vectors or streamlines to understand ‘where it is going’ and refine the design of prop tunnels, or the energy in the wash as it meets the beach or bank to minimise erosion and so on. The hull can be analysed in a fixed orientation – such as a yaw angle that might be used for a towing tank simulation for a yacht sailing upwind – or it could be free trimming so the vessel will change it attitude as the forces evolve, giving us information for a powerboat about running trim and sinkage or stern squat at the transition from displacement to planing modes
Keels, bulbs and rudders.
The first two images here show the keel and bulb analysis from an IRC 42ft design, followed by a new rudder design for a 30m sloop. While a racing yacht’s keel and bulb might seem to be the obvious candidates for CFD analysis – and they are and area where large gains can be made – we have also carried out some less straight forward investigations. Our 56ft cruising yacht had a twin rudder and twin engine configuration, and we wanted to be certain that the prop shafts and p-brackets would not interfere with the flow leading on to the rudders, especially upwind at higher yaw angles. We also carried out a study into the design of skegs for a 40m dumb barge that yawed excessively under tow. Originally fitted with only one skeg on centre, we found the bluff stern was preventing the skeg from seeing any lateral flow and so couldn’t correct the yaw. By placing two skegs outboard and investigating the effects of splay angles we were able to remove the yaw issue completely.
Aerofoils – wings and things
Slotted and multi-element wingsails are becoming more common place, and analysing them is something we can undertake. The images show a slotted leading edge design intended to increase lift/drag ratio and the stall angle , and a conventional hinged two-element wingsail against which it was compared. Other studies have included the design of a cowling and straightening vanes for a frost fan. They are intended for drawing down warmer air from above an inversion layer, and blowing it over several hundred metres along a vineyard to prevent frost damage to the crop. The cowling was improved to remove the ‘dead air’ at the tips of the impellor blades, and the vanes had an element of twist introduced to account for local variations in flow direction across the prop disc.
Aero studies of superstructures
The Elite 10m was an exercise in determining the total aero drag at a number of speeds as the speed requirements were only marginally meet when calculated using the standard formulations for aero drag. This investigation gave us the confidence to state the contract speeds were attainable. The second image shows the 2D investigation into a visor for the Global 80 cruiser. The client wished to settle the argument as to whether a vertical, forward raked or aft raked visor would afford the best protection from a head wind when lying at anchor. (Vertical was the winner). The owner of the Origami 560 wanted to be sure that windscreen would deflect the oncoming airflow at 30knots over his 1.9m head height and would not be unduly affected by the bimini hard top.