Virtual Engineering and the future of Aerospace

RachaelO

On Ask The Experts which is a cool new site for technology enthusiasts, there is an article in its ATE Defence Space and Aerospace which looks at the what can be derived and achieved when virtual engineering is applied to the aerospace industry http://tinyurl.com/39q2eqh its suggestions are futuristic.

mugaliens

The Boeing 777 was the first aircraft designed entirely on a computer. Throughout the design, they also designed maintenance people in digits, as well, to test for proper human access to all serviceable aircraft systems.

jhae2.718

Actually, some of the things mentioned are already being done. I believe Lockheed Martin, for example, created of the JSF production facility in CAD and ran through the production phases, as well as using virtual environments to test maintenance and carrier operations.

Boeing's also heavily computerized the design of the 787.

In some ways, I think we're seeing a shift towards more computer simulations as opposed to physical testing, as computer models become more accurate. There's quite a bit of research being done in improving simulation performance and reducing the time it takes to run them.

simplex

Virtual Engineering does not exist or it has existed for thousands of years!

Tell me, when Gustav Eiffel designed his well known tower in Paris, what did he do?
Answer:
Virtual engineering.
Why, because he designed everything on paper and after that, he built the tower according to the plans.
Nobody had ever buit something similar to Eiffel Tower, before and the viability of the entire construction was based only on mechanical calculations (using the strength of materials theory).

Cyrus

Would you guys stop using and abusing this meaningless word 'virtual engineering'.

JaredJames

http://en.wikipedia.org/wiki/Virtual_engineering

"Virtual engineering is defined as integrating geometric models and related engineering tools such as analysis, and simulation, optimization, and decision making tools, etc., within a computer-generated environment that facilitates multidisciplinary collaborative product development."

jhae2.718

"Virutal engineering" is a rather pointless term; "virtual engineering" is doing engineering work with computers. There's not really a need for it.

Sounds rather like a "buzzword".

(For the record, I never actually used that phrase...)

JaredJames

I think it's just used to mark the difference between using facilities such as wind tunnels for aerodynamic testing to using only computer simulations. So far as engineering goes, it makes no difference. It's more about marking a difference in how things are done.

I remember watching a documentary on designing a new navy submarine. The engineers were showing the computer models used and how they helped them visualise how every section went together and interacted, something that was difficult to do without the 3d modelling. They also explained how this method allowed a reduction in design and development costs.

Cyrus

Before I respond, I'll ask you to clarify what you meant in bold.

I don't think CAD needs to be glamorized to the point of being called 'virtual engineering' - it was called drafting for years before a computer was used to do the same thing.

JaredJames

As in the work completed by the computer is no different to what a human would do. For example, a human would use a wind tunnel to test a new aircraft design, analysing the aerodynamics. Now you just design your model in a computer and it can perform the fluid dynamics simulations on it without the need for a wind tunnel and to a far greater degree of accuracy.
I will agree there, CAD is certainly nothing particularly special, but these days the CAD software has developed from being something that you could use for rapid prototyping to give you a base model for testing and improvement, to giving you a full model, with simulations and tests run on it (as above) which can go straight to final production.

Years ago, you would have to manually draw and calculate everything by hand. Now, you can insert the components / systems into a software suite and produce a 3D model, showing all components and how they interact / interfere with each other, highlighting any issues. In the case of the submarine programme, they showed how they could do a walk through of the ship so they could get an idea of how it all fits together.

I think the whole 'virtual' side of things mainly refers to the ability to run simulations and generate these 3D models which can be far more useful to the engineer than the hand drafted designs.

The ultimate 'virtual' engineering suite would be like that on Iron Man, where you get the 3D holographic model in front of you and you can interact with it. But that's way in future, so far I've only seen a touch screen design which allows you to interact using your hands on what is essentially a giant mobile phone screen.

Cyrus

Sorry, but I'm afraid that's not true. CFD always requires validation in a wind tunnel - it is not at a point to replace a wind tunnel. I'm not sure where you got this notion from. CFD generally does a bad job, sometimes is O.K., rarely its good.

Edit: Okay, it's not generally 'bad', but it's not 'great' - just 'O.K'. CFD is generally good for predicting the lift curve. The drag curve is so-so, prior to stall. Pitching moments are generally 'fair'. When the flow starts to enter into stall CFD stinks. The problem with CFD is that it has a hard time accurately reproducing the boundary layer and shear stresses. As for replacing a wind tunnel - don't hold your breath .

The inter-relationship of components would fall under systems engineering. The production aspects, manufacturing engineering. CAD is just a tool that can be used, I still scoff at it being called 'virtual engineering'.

JaredJames

Just did a very quick bit of reading and I agree, it's not good for low airspeed situations in particular so yes, a wind tunnel is required. It did throw up that space ship one was designed completely on a computer and didn't under go wind tunnel tests (apparently).

Well according to the wiki definition it is called that because it is all done within a computer environment. I agree with what you say though.

Cyrus

Let me put it this way, space ship one had aerodynamic instability problems in pitch rate. That being said, I don't particularly care much for the validation of their CFD. They also killed two people when a rocket motor blew up in their faces - total disregard for safety. So, I would take their work with a grain of salt.

minger

Cyrus, I'm like standing right here. I'm going to have to do some defending here....

No it doesn't. Researchers were nailing boundary layers and shear stresses using algebraic models 40 years ago based on the Boussinesq approximation. Today's turbulence models are far more general. Like most CFD, it is up to the user to know when to use the appropriate model; no model works for everything.

Aside from that, one "could" do a DNS simulation.

As for validation on complex analyses, there are numerous studies and journal articles showing really damn close validation for i.e. multi-stage components including compressors and turbines, all solved simultaneously using various GGI approaches.

My former adviser once told me that, "A good engineer can get great CFD results, a bad engineer will get horrible CFD results." The quote couldn't be more true. The fact of the matter is that the state of the art in this so-called "virtual engineering" is good enough to design without testing. The problem is the engineers; we are fallible. We necessitate testing, just to be sure.

Could CFD replace wind tunnels? Maybe, but what would replace the ubiquitous engineer with a smoke wand standing in front of the brand new 2011 [Insert Make] [Insert Model] car's commercial?

p.s. I agree. This phrase "virtual engineer" is BS. Even the smallest design shop owns an ANSYS or Patran/Nastran license. The fact is that virtual engineering as we're calling it here, is just engineering nowadays.

Cyrus

I detect a WT/CFD battle brewing!!

Let me pose this question to you then. I'm designing an aircraft for the first time - which turbulence model do I choose? I ask this because there is a difference between predictive analysis (which is what you want CFD to do) and postdictive analysis - which is adjusting the turbulence models until it matches experimental data.

Brute force! (computationally expensive, isnt it?)

I don't know - I've seen some presentations of people in industry using CFD and comparing it with WT data. The results were as I described. I've seen some really good CFD, and similarly I've seen it do not so great.

You'll take my smoke wand out of my cold dead hands!!

minger

External flows are fairly easy to predict. I would be very confident using an SST k-w method which uses k-e model near the wall and k-w away.

I just started a new position with a major player. The methods and techniques are VASTLY different than the little guy I was previously with. The standards and best practices are well-documented, and they continually do rig-testing and CFD validation tests. That's just a difference between them, the big guys have the capacity and resources to do in-depth validation with their rig testing.

Now, like you, I've seen some pretty bad CFD from industry folks, as well as academia. Would I recommend no wind tunnel testing? Absolutely not. However, if I mess up a design, I can have a new model running in minutes. How long to get another scale model?

I do agree though, eventually one would like "real" data to confirm the design.

Cyrus

How do you know these are the right ones to use a-priori, and, where does the switch over between k-e and k-w occur?