# Computational Fluid Dynamics (CFD)

## Fluid flow and thermal simulation as a cost effective alternative or supplement to series of sophisticated experiments

When designing and optimizing new or existing processes and systems often interactions between flow and structure play an important role. To better understand these processes in order to derive important lessons for the design and optimization of components and systems involved, termoflow.com offers Computational Fluid Dynamics service (CFD,). This makes it possible to play through various test cases, geometry modifications and flow parameters on inexpensive computers, which, consequently, leads to a decrease in design time and to a significant cost reduction. Ideally, such expensive prototype series and costly laboratory tests may be omitted. CFD has become the third, indispensable link between theory and experiment in research and development.

CFD simulations can be alternative to experimental and mathematical methods if not the only viable analytical path in certain cases when conventional measurements are difficult or impossible to accomplish due to extreme temperatures, high pressure, aggressive media, or other hazards to the test personnel or study area.

## How is it possible to simulate fluid flows?

### CFD in the right hands - a powerful development tool

The movement of fluids is difficult to imagine and to understand, as even the smallest forces cause large progressive form changes. Turbulence as a factor is a highly unsteady, seemingly chaotic behavior, which manifests itself in rapid fluctuation movements of fluid particles. In general, this movement can be explained with the Newtonian equation of motion. However, this microscopic modelling approach is very expensive in terms of computational time and memory ressources and runs downright past the practical use. For industrial applications the fluid is considered to be a continuum instead, i.e. looking at the microscopic average (macroscopic) properties of the fluid; the fluid motion is then described by the derived in the 19th century Navier-Stokes equations of motion. These governing equations of fluid dynamics are based on three fundamental physical principals: conservation of mass, momentum and energy. Because of their difficult mathematical nature, it is not yet able to provide a solution in an analytical way. Therefore, it is dependent on the use of computer-aided methods in which the equations of motion are converted at discrete points in the entire flow region by mathematical operations in numerically solvable set of equations, which can be solved iteratively using a computer. ## Typical sequence of a CFD simulation

During the course of a flow simulation, a real existing flow process is displayed by a digital model and made accessible to numerical calculation using the computer. The process is described below in more detail:

### Geometry processing

At the beginning of each flow simulation, a geometry file of the fluid domain, the space filled up with fluid, is created using CAD software. As far as heat transfer processes are concerned, the surrounding walls and the flow space adjacent components additionally have to be considered as solid domains. However, irrespective of the type of the simulation, it is at the the engineers discretion to determine the level of detail of the geometry and filter only the flow-related components that have a potential impact on the flow field. It should be in mind that this simplification represents a first approximation to the original. This simplification is unwanted sometimes and small geometry features have to be captured. Just imagine the flow through an engine compartment. The cooling air is making its way through a maze of pipes, cables, and other equipment. Consequently, the involvement of all these factors is crucial for the correct determination of the volume flow dependent pressure drop in the engine compartment (system impedance) and thus for the selection of a suitable fan. At termoflow.com we intervene in geometry as little as possible to achieve a higher degree of detail and simualtion accuracy.

### Discretization

The numerical determination of the fluid flow takes place at discrete points in the computational domain. For this purpose, a 3D mesh is generated, which divides the computational domain in a lot of non-overlapping mesh cells or in control volumes (CV) from a physical point of view. Each CV is associated with a mesh node on which all governing analysis parameters such as the flow speed, temperature and concentration are stored. The spatial distribution of mesh nodes has a decisive influence on the accuracy of the numerical solution, especially at the boundaries of the computational domain. Experience has shown that the generation of high quality meshes is very time consuming and requires both a big amount of project time as well as experience of the user. The innovative approach of termoflow.com for mesh generation uses an iterative method, which adjusts the mesh resolution locally on the flow conditions and geometry, at the same time ensuring a reasonable resolution of the near wall flow and temperature boundary layers, which are very important for the prediction of heat transfer. This physically reasonable approach allows shorter project runs and produces robust and reliable results. Also it enables us to react quickly and flexibly to short-term variant and construction changes.

### Preparation of models (Pre-Processing)

Before the simulation run, the physical complexity of the flow model has to be defined in a further processing stage. Depending on the expected flow conditions, a decision is made for either transient or stationary simulation. Liquids can be simulated in a good approximation as incompressible media. For gas flows compressibility effects have to be accounted for Mach numbers greater than about 0.3. As the Mach number is sufficiently large, the inertial forces are dominand and therefore friction can be neglected. In this case, the much more simple Euler equations are preferably utilized. Above a critical Reynolds number, the flow loses its laminar, structured character and goes into the turbulent state. The effects of turbulence are then included via implementation of statistical turbulence modelling by means of additional transport equations. For flows with heat transfer a further transport equation for the total energy is needed. Additional effects such as radiation or combustion can be simulated by eather including additional physical models or source terms. By defining constant or variable material properties, the numerical simulation model is finally completed.

The resulting differential equation system cannot be clearly solved without further parameters because there is an infinite number of solutions. Thus appropriate physical conditions have to be applied at all boundaries of the flow domain to obtain a well defined flow scenario. These could be the flow variable itself or their gradient. The quality of the simulation depends on this essential step since even small variations of the boundary conditions produce large changes in the results and thus lead to false conclusions. The result of bad calculations is usually costly rework to optimize the products and processes. In some worst case scenarios a costly redesign may be the only solution. The simulation process itself is then puted into question or often rejected from these considerations in advance. We offer our customers our experience and extensive expertise in the field, so they can assess and are able to benefit from the numerical flow simulation to its full capacity.

### Presentation of results (Post-Processing)

The result of a CFD simulation is a deep data jungle; from this which we extract all information you need and systematically analyze it. For selected 1D/2D/3D investigation areas (space curves, section planes, surfaces, volumes) we provide the following data representation:

• Flow pattern (Streamlines)

• Particle paths

• Vector plots (forces, speeds, etc.)

• Contour plots for scalar quantities (pressure, density, temperature, concentration, etc.)

• Time-dependent animations   