# 3 Approaches To Fluid Mechanics

Posted October 28, 2015 at 07:01 by adarsh tiwari

Tags cfd, fluid dynamics, fluid mechanicscfd, openfoam, starccm

In my previous blog, I had talked about the interpretation of CFD concepts and my discussion was limited to the nature of solution obtained from CFD simulations. In this blog let’s talk about the nature of solutions and its interpretations in real-world applications.

As we look into the classical approach of fluid mechanics, we will see that there were two approaches for the determination of fluid phenomenon. Let’s look into these three approaches in detail.

The 1st approach is mostly theoretical. In order to validate the 1st approach, 2nd 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 fails 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 3rd Approach.

Let’s dig a little deeper into this.

As we discussed the two approaches, we can consider CFD as the 3rd approach in fluid dynamics. 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 ‘3rd 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.

MYCADCFD 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.

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.

As we look into the classical approach of fluid mechanics, we will see that there were two approaches for the determination of fluid phenomenon. Let’s look into these three approaches in detail.

**The 1st and 2nd Approach**The 1st approach is mostly theoretical. In order to validate the 1st approach, 2nd 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 fails 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 3rd Approach.

Let’s dig a little deeper into this.

**3rd Approach**As we discussed the two approaches, we can consider CFD as the 3rd approach in fluid dynamics. 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 ‘3rd 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.

MYCADCFD 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.

Total Comments 0