Model engineers

Computational fluid dynamics (cfd) has been something of a controversial subject over the last few years. "Use or abuse of modelling?" asked a recent feature in this magazine (BSJ, August 2002). On the other hand, use of cfd continues to grow, and software is increasingly sophisticated – it's a popular piece of equipment used in the leading consultancies. The reason for this dichotomy is that while cfd is widely used, sometimes, according to some engineers, the results aren't all that good.

However, there is a growing band of experts who specialise in working with cfd, and similar tools such as dynamic thermal modelling. They say that it's not the software which is inaccurate, but the users who don't necessarily know what they're doing. Working on cfd is, they say, more complicated than you think.

One such expert is Paul Kingston, head of virtual engineering at Hoare Lea. He says: "The impression I get in building services is that people try to be a building services engineer/modeller. If you're trained to be a modeller, that's what you do. If you're trained to be a building services engineer, that's what you do. There needs to be an understanding between the two. Most cfd packages are on a par in terms of their capabilities, but the accuracy is driven by the competency of the people using them. It's a case of those who know what they are doing with it are the ones who get the right results, those who are chancing their arm at it get what they get."

At Hoare Lea, the virtual engineering department is made up of eight individuals all dedicated to that specific field.

This isn't to say that no one else is extended the opportunity to try out modelling to a certain degree, as Kingston points out: "All graduates within Hoare Lea will do an amount of modelling up to a certain level. It then comes to a stage, because of time constraints and skills, where we'll then stop them and take over, or sometimes take over half way through a project and take it that bit further. My guys aren't building services engineers, they come from different backgrounds so there are cases where we'll get it done quicker and work on a project right through to the end."

Arup also has its own department of simulation modelling experts. Dr Darren Woolf believes that experience is key to accurate interpretation of results. "It is important that your deductions from the results of cfd modelling are reasonable, and moving in the right direction. You have to use your judgement at all times. We do have groups which carry out their own analyses, but it is important to have some sort of peer review in place for those who aren't using cfd or similar software 100% of the time."

It is this issue of accuracy that determines when and where the virtual engineering department will take sole charge of the modelling on a project, but Kingston believes it is important for engineers to do some modelling where possible: "We encourage limited modelling by the engineers to give them a better appreciation of how the buildings work, but it's not feasible on certain large projects or sensitive projects where its critical to get it right."

Dr Garry Palmer, group director of Buro Happold's Computational Simulation & Analysis division (CoSA) says that integration is key. "It's true that where cfd is concerned, a little knowledge is a dangerous thing. Commercial packages make it look easy; people think it's easy to get accurate results. On the other hand though, there are good specialists out there, but they are not always integrated with the team. Analysis that way is not much good either. We have to work as an integral part of the design team."

It would seem that the only way for the accuracy of results from cfd to continue improving and for the confidence in computational methods to grow in the industry, is for experience to be gained and built up over a longer period of time.

Kingston feels his group is in a position now where they can be extremely assured of the results they are producing. "Over the past five years we've built up a large portfolio of experiments and computational models to give us that level of confidence and accuracy in what we're doing. The people working here range from mechanical engineer/physicist to actual computational fluid dynamicist."

Both Arup and Burro Happold have a team of experts with backgrounds in engineering and often with PhDs in cfd-related areas. Again though, where accuracy of results is concerned, nothing counts like experience, both personal and across the organisation.

"Because we're a multi-disciplinary practice, we have access to buildings post-occupancy. This means we can evaluate our work. We also use analysts with appropriate backgrounds to whatever project they're dealing with," says Palmer.

At the moment, there is a relatively small number of computer simulation specialists, although most engineers have some experience of using these tools as a small part of their responsibilities.

But this is a trend that may be set to change. In light of the new Part L regulations and the growing issue of building labelling, performance of the building is now paramount for the vast majority of new projects. Being able to model a new building at an early stage can go some way to helping clients achieve the requirements laid down by legislation.

I think it’s like cad or the internet. At certain times there is a real need for people who can master those tools.

Tim Dwyer, principal lecturer, South Bank University.

Kingston sees this as the key to an increase in the use of cfd and other engineering software: "I think people are demanding more modelling to be done on buildings prior to construction. The last thing you want to do is build a building nowhere near what you want to achieve, so if you're modelling you can bring it within the realms of what you're aiming for. Because of legislation people are asking for modelling because they want to be confident that the building design is going to meet the requirements. That's the driving force."

Woolf and Palmer agree. Getting the modelling team involved early can save costs, and create better buildings.

The other driver behind increased used of cfd and similar software is that it's getting much easier to run complex modelling projects on a desk top pc, rather than across a whole server.

Kingston says: "We have modelled buildings where we've got three buildings on the side of a hill. We modelled inside and out and we can see where all the airflows go. Those kind of calculations cost a lot of money, but they are what I would consider cutting edge; they are transient. They have the effect of a dynamic model, but go into really fine detail and we have to run those on big super-computers across the internet. It's a case of moving on but purely because of the computational powers available."

Woolf adds: "Computing speeds and capacity have increased, even in the last year. I could now double our capacity for less money than a few years ago. When you get these kinds of improvements, you are opening more doors to explore more techniques, such as greater accuracy in turbulence modelling, for example."

So with the demand for modelling and other virtual engineering techniques looking set to increase in the near future, one might expect that the number of 'virtual engineers' would also begin to rise.

Although the term 'virtual engineer' is one that Kingston isn't altogether comfortable with, he does agree that this kind of specialism is required: "I suppose that's what they are, more virtual prototyping engineers. I'm all for a limited level of modelling being carried out by engineers, but to get things done more effectively, you need a dedicated engineer with all the skills required to do it. That's not just computational skills, that's report writing through to understanding the building services engineer and vice versa."

Kingston also believes that in order for more specialist people to come through, more dedicated courses are required: "The universities are including modelling and other skills in their courses but I think there needs to be a more dedicated course to actual building services. You can do aeronautical based modelling courses and specialised car industry courses and the same is going to have to happen in building services. It's one of those things that will have to develop, but as it takes off you'll hopefully see more specialist people coming through. How well it develops will be seen over the next five to ten years."

For the universities, however, teaching use of cfd and similar software packages is something of a challenge. Tim Dwyer, principal lecturer at South Bank University, says: "I think it's one of those things like cad or using the internet. At certain times in their development there is a real need for specialists – someone who can master these tools."

But Dwyer believes that even though cfd is now more widely available to engineers, teaching it at an advanced level as part of a building services course isn't practical. "It would be very difficult to include as part of the existing course, and still find time to fit everything in. Also, it's a very transient thing. The software would develop so much that we would find it difficult to keep students up to date."

Students at South Bank are encouraged to view cfd as a useful tool, and made to feel at home with the concept of using it – they have to demonstrate its use as part of their final project. But ultimately, says Dwyer, the role of the building services engineering course is to turn out graduates who are useful engineers, not cfd software specialists – which is a fair point.

It seems that we are looking at a time when use of software modelling before, during and even after buildings are designed and constructed will increase. Technology is rapidly catching up with the requirements of engineers.

The important issue seems to be experience. The software can be used by anyone, but accurate results can only be produced by someone with a combination of skills including an understanding of engineering and how buildings really work.

  It is important that undergraduate courses give students a flavour of this new technology, but post-graduate specialisation seems to be the most intelligent route, for the individual and for consultancies overall.