In my previous blog Simulation Science And Computational Fluid Dynamics, I had talked about the interpretation of CFD concepts and my discussion was limited to the nature of solution obtained from CFD simulations. As we look into the classical approach of fluid mechanics, we will see that there were three approaches for the determination of fluid phenomenon.

In this blog, I will talk about the nature of solutions and its interpretations in real-world applications. Let’s look at the three approaches in detail.

**The 1**^{st} and 2^{nd} Approach To Fluid Mechanics

^{st}and 2

^{nd}Approach To Fluid Mechanics

The 1^{st} approach in fluid mechanics is mostly theoretical. In order to validate the 1^{st} approach, 2^{nd} approach (i.e. experiment) is used. We can’t completely rely on theory because of assumed values and simplifications. Similarly, it is not very economical to conduct experiments and to articulate the real life conditions.

For example, if we want to test an aircraft at the speed of say M=5, i.e., Mach 5. In this case, most of our theoretical calculations fail because of unusual curved geometries and complex structures. The experiments at ground testing facilities require wind-tunnels of such capacities. We are lacking the wind tunnels that can simultaneously maintain the higher Mach number and higher flow field temperatures. Also, there is a big question about the accuracy of the results. Hence, it is always better to start with a better estimate of the values. This ‘better estimate’ is provided by Computational Fluid Dynamics (CFD), the 3^{rd} Approach.

Let’s dig a little deeper into this.

**3**^{rd} Approach To Fluid Mechanics

^{rd}Approach To Fluid Mechanics

As we discussed the two approaches, we can consider CFD as the 3^{rd} approach in fluid mechanics. I am not saying that CFD will completely replace the theory or it will completely replace the experiments. The only thing I am saying here is that the ‘3^{rd} approach’ will assist both the approaches in reducing the overall effort and cost. Hence, we can more confidently say that CFD is just an engineering tool. A good computer programmer can write a CFD code and solve some problems, but the results he/she will get may not make any physical sense.

The point I am trying to make here is, learning some software is not CFD. For using a complete potential of CFD, you need to learn the governing principles and ideas driving CFD. Once you are familiar with the theory and algorithms in CFD, you can use any of the software with a little practice. It doesn’t matter whether you are using Ansys Fluent, CFX, StarCCM or OpenFOAM, of you are familiar with the concepts any software can be used effectively. focus on OpenFOAM

Workshop Technologies focus on OpenFOAM because it is free in all respect. It is easy to program, fast and reliable. All the software, if used intellectually will give similar, but not exactly same results. This is because of the inherent mathematical approximation in algorithms.

**Conclusion**

CFD is a developing science. The current state of CFD is such that we can easily solve the laminar flow problems, but due to extensive computational cost and time required, we need turbulent models for calculating the turbulent flows. Direct Numerical Simulation (DNS) is limited to Reynolds value of few tens of thousands. If nothing new came up recently, then we are correct in mentioning that the achieved DNS till date is for Re=30,000. Still, DNS is not practically useful for most of the engineering problems due to computational limitations. Grid independence and governing equations are key players in the estimation of the correctness of results. Next blog will be dedicated to these topics.

*Have any questions about CFD? I’d be more than happy to discuss them in the comments below.*

Rohan Hingmire

said:Excellant information. how much error would be in actual performance vs cfd genarraly would we get?

View Commentworkshopwale

said:Hi Rohan,

Errors typically depends upon the following major factors

1) Computational Simplifications (Assumptions)

2) Mesh size and mesh density

3) Boundary conditions

4) Division of mesh for parallel solving

In a single line, it can be said that, with proper BC and modeling, the CFD results can vary up to +/- 5% of actual results.

Hope this helps 🙂 🙂

View CommentAndreea

said:Hi, I would like to talk about a collaboration with you. How could I give you more details?

View Commentworkshopwale

said:Hi Andreea,

We are always ready for healthy discussion and fruitful partnerships. You can directly write to our directors at: adarsh@workshopwale.com ; khushali@workshopwale.com

You can also call us at:

View Comment1) L-line : +91-020-2727-0144

2) Mob: +91-9986-999-219

Jay Patel

said:what types of boundary conditions used in simple modeling

View CommentADCS

said:I like this web blog very much, Its a real nice

View Commentberth to read and find information.