When engineers need to design or improve fluid flow systems, it is often impossible or not cost-effective to see inside these systems and measure data. Through the use of numerical methods and powerful computers, computational fluid dynamics (CFD) algorithms allow engineers to model fluid and how it interacts within a specified system. With CFD models, engineers can quickly and cost-effectively design or improve the performance of a system without physically building and testing it.
Above: Vortices caused by thermowell inclusion in a pipe.
What Is It?
Computational fluid dynamics (CFD) is a branch of fluid mechanics that uses computers to simulate fluid flows such as the interactions between gases, fluids and structures. The idea behind CFD is that engineers can predict the real life physics of fluid flows in a system without physically building and testing that system, which is usually very costly both in time and money. Through the use of numerical methods and algorithms, engineers can create mathematical models to predict fluid flows outside or inside a system. Many different flows such as laminar, turbulent, compressible, incompressible, single-phase, multi-phase, subsonic, and supersonic can all be analyzed.
Take, for example, a small rod inserted into the side of a pipe with gasses moving past it. In a refinery vessel, this is a thermowell in the vapor overhead line. The refinery wants to increase production, which would begin a domino effect of changes. This increase in production will increase the velocity of the gas. If the velocity of the gas increases, the forces on the rod will increase. If the forces on the rod increase too much, the rod could begin vibrating and break. Because of these possibilities, the refinery needs to know how know much force will be applied to rod by the increased gas velocity to avoid breaking the rod. Building a test rig to test this scenario would take months to complete and could cost a small fortune. CFD can provide the answer with just a computer and a couple of weeks.
A CFD model has three major parts: 1) the fluid inlet, 2) the modeled domain (containing the point of measurement), and 3) the fluid outlet. Determining the correct locations for the inlet and outlet with respect to the measurement point are two of the most important steps in building a good CFD model.
The dimensions of the run of pipe.
A good inlet is one where both 1) the fluid velocities can be reasonably approximated and 2) there is enough distance before the point of measurement (the rod) for flow patterns to develop. In this model, the gas may be assumed to have a constant velocity at the top of the refinery vessel. This location provides enough distance for flow patterns to develop before the point of measurement. Therefore, the top of the refinery vessel is our model inlet.
A good outlet is one where there is enough distance for flow patterns to stabilize after the point of measurement. In this model, 20-feet of pipe downstream of the rod is considered a reasonable distance. The model domain is therefore from the top of the vessel to a point 20-feet after the rod.
Once the model domain is selected, an exact digital replica of domain is drawn using Computer Aided Design (CAD) software. Next, the domain is broken down into tiny cubes called a 3D mesh. This 3D mesh allows the computer to solve very small flows and then combine the flows together. The computer solves the flows for millions of tiny cells at one moment in time. It then moves forward in time a very small step (maybe 0.00001 seconds) and solves the flows again. It takes a string of 1 million of these modeled time steps to get 10 seconds of predicted flow. CFD therefore requires a lot of computational power over a long time to model even 1 minute of flow.
Why Do It?
Performing CFD analyses help engineers to see inside of systems where physically measuring data would be cost-prohibitive or simply impossible. These simulations help engineers understand the physical processes that occur in their fluid flow systems. By performing these tests, engineers can prevent problems from happening, explain why problems occurred and even predict how prototypes may act in the real world. Time and resources that would have been used to physically create the system to obtain data are saved, and engineers are able to produce the information they need to influence business decisions that lead to better returns on investment.
Above: Looking down along the thermowell as flow passes across it from left to right. Velocity vectors are shown as arrows pointing in the direction of the flow with arrow color indicating the speed of the flow.
How We Do It
We begin performing CFD studies by getting as much information from the client as we can. We either go directly to the worksite and collect ourselves or have the client send the necessary data to us. Next we establish and approve assumptions about the system to be modeled. We create a study basis that contains the assumptions, known data, and purpose of the study, which is then reviewed and agreed upon by the client. We then create a model of the system, run or simulate the model and validate the results. Due to the complexity of the calculations being performed, running the model and collecting the required data from it may take several days or weeks. Once done, the results are compiled into a draft report that is discussed with the client. After everything has been settled, the report is finalized and delivered to the client.
To learn more about CFD studies at Opgrade LLC, contact our sales staff today!