Walking on air

Working in 3D is increasingly becoming a standard practice for engineers and designers, with more complex computational fluid dynamics (CFD) systems capable of simulating and analysing ever-larger amounts of data.

But the sheer quantity of the data is creating two problems — how to manipulate it so that users can extract the relevant information; and how to display it meaningfully.

Cambridge Flow Solutions (CFS), a spin-out from Cambridge University’s CFD Laboratory, is working on these problems, and has linked up with 3D specialist Virtalis to make the link to 3D visualisation.

CFS was formed in 1999, but its technologies have been in development since 1985, taking shape from software to simulate the flow of gases through and over turbine blades. Over this time, it has seen CFD develop from a highly-specialised process, accessible only to institutions with supercomputer-level processing power, to a tool that virtually any engineer can access.

‘You can build a computer to handle CFD from standard PC hardware these days,’ said Ed Lewis, CFS’s business manager. However, he added that the technique has almost become a victim of its own success.

Most CFD is now run on parallel computing systems, to reduce the time needed for the vast amount of calculations and data handling. But even with such systems, not all the steps can be parallel, and bottlenecks are inevitable.

In particular, said CFS software development engineer Simon Harvey, the post-processing stage, which translates the results of the fluid dynamics equations into a graphical form, is a serious bottleneck — and this means that the systems are very slow and cumbersome to use.

‘The problem is, that people are now capable of going into much more detail with their CFD than they could before,’ said Lewis.

‘The commodity systems, with parallel computing, allow them to do things like seeing the effect of changing the positions of rivets and bolts, but this creates a huge amount of data, and processing it so that it can be displayed is a huge task. The information is there, but the more detail you go into, the harder it is to actually access it.’

The GigaCell VR system means complex graphical displays, such as isobaric surfaces on turbines and shockwaves forming on aircraft, can be produced in a fraction of the time of a conventional system.

CFS’s response to this is the GigaCell VR system, a parallel postprocessor which keeps all the data on the system which calculated it and only processes the relevant data for viewing purposes. This allows it to produce complex graphical displays, such as isobaric surfaces on turbines and shockwaves forming on aircraft, in a fraction of the time of a conventional system.

‘Everything stays on the server and doesn’t migrate to the viewing workstation,’ said Harvey. ‘The server “listens” to the cluster for a request from the client, and only sends through what it’s asked for. You don’t have to ship gigabits of data, which makes it that much quicker.’

It’s the combination of GigaCell VR with Virtalis’s 3D visualisation systems that the companies hope will change the way people use CFD.

Using stereoscopic projection on to a big screen and viewing through polarised 3D goggles, the system allows users to look at their simulations in a much more intuitive way than they would on a flat workstation screen.

‘The system will be able to run CFD calculations over a few hundred million cells, and we are developing a parallel post-processor that will be infinitely scalable. When in use, it will appear to be effortless, but in fact up to a terabyte of data will be interrogated in real-time and displayed in 3D,’ explained Andrew Connell, technical director at Virtalis.

‘The difference from a flat display is quite startling. You start to see connections between the design and the way gases flow around it as quite obvious, whereas before, you would never have seen it.’

Connell believes that the 3D system allows CFD users to use the distance and depth cues which are a vital component of stereoscopic vision, and are ‘hardwired’ into the human brain’s perception centres. ‘These are very important in the understanding of flow,’ he said.

‘Using a 3D visualisation system allows you to use a computer model in exactly the same way as you’d use a solid model in an air-tunnel test, for example. You can walk around it, apply the digital equivalent of smoke trails, and see exactly how the airflow moves and changes. Because of the speed of the postprocessing, you can do it in real-time.’

The applications for the system spread across a variety of sectors, said Lewis. Familiar hi-tech pioneers, such as Formula 1 teams and the aerospace sector, are obvious targets, but some of the actual applications are likely to be markedly different.

‘This could be incredibly useful for the design of safe airliner cabin interiors, because you can use it to model how smoke would flow around the cabin in case of fire,’ he said.

By the same token, the system has potential for security applications. ‘People are very interested in modelling airflow around cities,’ said Lewis.

‘In the case of something like a nerve gas attack, you need to know how weather patterns will interact with the buildings, rivers and so on to move the agent around. That’s a problem requiring a huge dataset, to model the city and so on, and massive calculations. This sort of system would be able to handle it, and the 3D visualisation is a massive help in interpreting the results.’

Similar calculations would help in the design of safety systems for oil rigs and process plants, especially for explosion tests. The system could also be used in the design of gas-phase reactors, where the geometry of reaction vessels, configuration of baffles, inlets and outlet tubes, the location of sensors, and the generation of heat from reactions can all have complex and unexpected effects on the flow of gases.

3D visualisation is a new tool for this application, and Lewis believes that it is so useful that plant designers will be quick to see its potential.