Tank Farm Wind Load Analysis using Computational Fluid Dynamics
Wind load is one of the main concerns in construction codes. Besides gravitational loads, earthquakes, and other factors, wind gusts-induced loads can cause failure in structures. This is mostly because of the lateral forces, induced turning moments, and pressure/suction on roof and walls, all of which deviates from the regular gravitational loads’ effects on buildings and other load-carrying structures.
The wind is more or less influential in a given structural design case, depending on the geographical location, as for example coastal areas and plains are more affected by wind storms and fast wind gusts. Also, the tallness of the structure, its shape and surrounding elements affect the wind flow and therefore the pressure distribution on the surfaces. Construction codes take into account all of these factors and deliver methods to calculate loads, always in approximate and conservative ways. But if the engineering department needs to know with precision the details of wind load over a structure, because of unusual conditions or irregular structures, other methods must be used, such as wind tunnel testing or Computational Fluid Dynamics. In this article, I will make use of the later to predict the pressure distribution over a set of oil tanks and make comparisons with a single one.
Study Case: Tank Farm Analysis
We have a 2600 square meters tank farm located in an oil terminal at a bay side, with ten 15 meters tall tanks together in a space confined by a damn. This is a typical configuration for these facilities. The proximity to the sea and related lack of obstacles to gusts make wind loads a special concern. Also, care must be taken, as a failure in any of the tanks could lead to oil spills and contamination, so negative environmental impact is a big risk. A maximum knowledge of the real loads and produced stress state of the structural systems is desired, for safety and reliability assessments.
We will use the wind gust speed of 36 m/s from the local construction code and apply it in a CFD wind tunnel model of the tank farm, then analyze the vortex formation, turbulence, and pressures generated on the tanks.
Modelling and Simulation
A virtual wind tunnel was modeled using a hexahedral-dominant mesh, with the tank farm in the middle. The mesh was computed in the SimScale platform, resulting in 1.5 million cells:
The following modelling techniques were used:
- K-Omega SST turbulent model
- Steady state simulation
- Incompressible fluid
- No-slip wall conditions on floor and bodies
- Inlet velocity condition with v=36 m/s
- Outlet atmospheric pressure condition on other outer box faces
- Local refinements and boundary layers on walls
- Bounding box big enough to avoid influence of boundary conditions on the internal flow at the region of interest
Results of the CFD Analysis
The turbulent 1.5M cells simulation took 86 minutes to run on the SimScale platform. Below are some plots showing the pressure distribution and flow streams:
We can see that at regions of strangulation of the flow, the maximum speed tops at 57.5 m/s, a 60% rise over the gust speed. Let’s also examine in detail the pressure distribution:
We find a maximum positive pressure of 872 Pa, and a maximum vacuum of 2272 Pa. The main result to note here is the asymmetry in the pressure distribution on the walls of the tanks. This would induce moments and asymmetric stress distributions which are different from analytical wind load calculation methods. To highlight this fact, a second simulation featuring a single tank, with equal dimensions as one of the farm tanks, was carried out, then compared to the corresponding tank in the farm:
We can see that the pressure levels are lower on the farm tank, as expected, but distribution is irregular and asymmetric, so the resulting stress condition should be quite different. It is also interesting to note that besides the symmetry of the lone tank, at this flow speed a turbulent vortex is being developed. The result is that some asymmetry is to be expected in this pressure distribution also. We can, for example, examine the total normal forces exerted on the surfaces of the tank to compare the action of the pressure:
There are some interesting things we can examine here. First is the force in the X direction being not zero, which is the result of the asymmetries in the flows. Also, note that the total X force is higher in the farm tank case than in the solo tank case by 49%. The higher forces occur in the Z direction, and if we see the plot we find that its effect is in lifting the tank roof, by suction:
These simulation results clearly show that designing and building tank farms, especially in regions where environmental and weather conditions are not ideal or steady, requires the use of a Computational Fluid Dynamics software. Very interesting insights in the flow around and inside the tank farm were gained, which can be used for engineering design decisions such as placement of equipment around the tanks.
The realism of the results, with the particularities found — such as asymmetries in loads — could be very useful for safety and reliability assessments. The next natural step in this direction is to translate the pressure loads over each of the tanks to Finite Elements structural models to compute stress levels and safety factors. This can also be carried out with high modelling detail using the SimScale platform.
You can take a look at this tank farm simulation project and even copy it and use it as a template for your own design. Please note that this is possible with a simple trial of the Professional account on SimScale, which is free. This is how I also started for my consulting business and once I tested the platform, I switched to a paid account so I can use it for my customer projects. Given that the platform is cloud-based, the investment in hardware is no longer required and the subscription cost is quite fair. You can just give it a try here: Plans & Pricing.