CFD Simulation of Control Valve
Guest post by :Centre for Computational Technologies, Pune
dynamics (CFD) is the use of computers and numerical techniques
to solve problems involving fluid flow. CFD is used in predicting fluid flow,
heat and mass transfer, chemical reactions, and related phenomena by solving
numerically the set of governing equations. Governing equations are conservation
of mass, momentum and energy. The results of CFD analyses are relevant in conceptual
studies of new designs, detailed product development, troubleshooting, and redesigning.
How Does CFD Work
Most of the commercial CFD solvers are based on the finite
volume method in which domain is discretized into smaller control volumes. On these
control volumes, conservation equations are solved. Partial differential
equations are discretized into a system of algebraic equations and these
algebraic equations are then solved numerically over each control volume. Basic
transport equation for transport of mass, momentum and energy is shown.
Processes involved in CFD
Basically there are three main stages in a CFD, they are Pre-processing, Solving and Post-processing. Pre-processing involves problem formulation, generation of geometry and computational mesh, applying the physics to the mesh under consideration. Solving is nothing but the computation algebraic equation of governing equations to yield the result. Post-processing involves analyzing the result both qualitatively and quantitatively.
CFD helps in
problem understanding. Experiments provide only reliable integral value for
a limited range of problems and operating conditions. Suppose if you want to predict the Cv of a
control valve, by experimentation we would be able to calculate the pressure
drop across the valve. From experiment is very difficult and in some cases it
not possible to get the pressure distribution over the surface involved. Apart
from this, pressure drop measured may the overall effect; we cannot isolate the
contribution from individual participants. But CFD Provide local information of
all the variables like pressure, velocity and temperature. Apart from this CFD
will help calculate the gradients thereby providing additional information
about the shear rate or heat transfer coefficient. Specific optimization of geometry
variations is possible for more efficient designs. We can predict the
performance before the valves are being manufactured. This will save
considerable time and cost to the individual manufactures. It is possible to
identify critical system conditions and to eliminate them from the experiments
in advance. In short we say that CFD provides detailed information, at faster
rate, eliminate or reduce the expensive design cycle and testing with flow
assurance and at much lower cost. The cost of doing CFD has decreased
dramatically in recent years, and will continue to do so, as computers become
more and more powerful and cheaper.
CFD in Control Valves
CFD is being applied by major valve manufacturers in broad
applications like flow performance evaluation, cavitations prediction, noise
prediction, torque prediction and to study the different geometrical effect for
efficient performance. Below figure outlines some of the applications for
control valves industry.
From CFD simulations,
data can be extracted for pressure distribution throughout the domain, velocity
magnitude at critical points to check if it is attainment beyond specified
limit, turbulence quantities and secondary flow. This Information is useful to
determine discharge, flow separation and recirculation zones and to calculate
the forces and torque acting on the valve body.
In the right hand a
complete CFD for ball valve is shown starting from geometry generation. The
geometry can be generated using commercial software like ProE or SolidWorks or
in ICEM CFD. In CAD model we have solid parts like pipe, ball and other seating
and sealants for the geometry shown. But for CFD analysis we require fluid
volume. This geometry was imported into ICEM CFD for fluid volume extraction. Fluid
volume was extracted keeping in mind the straight length required before and
after the valve. This will help in specifying the boundary conditions with more
Once the fluid volume was
extracted we need to create the mesh. One can have different options for mesh
generation. Depending upon the accuracy needed and the computing power we have,
we need to decide the meshing strategy. Picture
shows a hexahedral mesh generated in ICEM CFD. It is generally preferred to
have flow aligned mesh for better accuracy and smaller number of mesh count.
But for some complicated geometry, it would be too time consuming to generate
the hexahedral mesh. In that case one can go for tetrahedral meshing with prism
layers for better capturing the flow physics near the walls of the domain.
We have option of solving only for the fluid domain or if the application demands to include the solid parts of valve as well. Sometimes application may be high temperature and we may need to calculate the heat loss from the valve to the surroundings or it may be calculate the heat loss from the pipe and the temperature experienced by the different components inside valve. In these cases we need to solve for energy equation by including the solid parts also in CFD simulations. Appropriate solver setting is required to be applied over the fluid and solid domain in Pre-processor. Once this is done, iterations are carried out to yield the solution. Field variables are plotted and quantitatively evaluated at desired location in the post-processor. One may wish determine the exact pressure acting over sealant, is pressure in any location is falling below the cavitations limitations and many other parameters. If anything is required to judged, we need to consider the mesh independent study and accuracy of boundary conditions applied along with solver settings. Once this checked, one can go for the design changes if required and redo the CFD simulations.
This was very simple case study that we did at Centre for Computational Technologies (CCTECH), Pune. A horizontal axis butterfly valve was installed in fluid circuit to control flow of water through a metering valve. This setup was to calibrate the metering valve. Since the valve is of large diameter it had motor fitted on to it for actuating the butterfly valve rotation. What observed was that actuator is not able to open butterfly valve beyond 70o turn. The reason being the torque required to open the valve is higher than the actuator can supply. But we were sure that the actuator installed was in accordance with the manufactures guidelines. There was a conflict between the valve manufacturer and end user. Below picture shows the geometrical arrangement of the valve and the pipelines. To reduce the computational power, symmetrical assumptions was done. Flow enters the pipe from right top, flows through the valve and finally goes to the sump located beneath. Metering valve is not shown in the picture.
CFD simulation was carried out to determine what will be the maximum torque on the butterfly valve disc. From the graph it is clear that maximum torque is in between 60 to 70o turning of valve disk. Also it was observed that the total torque on the valve is more than the actuator can handle. In this CFD simulation only torque due to flow i.e. hydrodynamic is considered.
A very important conclusion can be drawn from the velocity contours near the valve as depicted in the pictures below. It can be clearly seen that velocity is maximum at two locations, near the valve and on the lower part of the bend section. For smaller angle of opening valve disc do not enter into the zone of influence of bend section. Hence torque is mainly due to pressure distribution around the valve disc. But as the disc rotates further, it enters into the zone of bend section. But in the bend section, pressure distribution is not uniform. This non-uniform pressure distribution increase the torque required to open the valve disc further. This can be further verified by simulation of valve for 90o opening. For 90o opening, we should get approximately zero torque. But simulation indicates it has significant torque because of uneven pressure distribution in the bend section.
Finally we did simulate a case
in which valve was moved by one diameter away from the bend. The results show a
very good agreement with theoretical. The torque on the valve disc for 90o valve position is almost zero. This clearly indicates the error was in installation
of the butterfly valve. Below picture shows the velocity distribution for 90o case with valve one diameter away from the bend.
a good insight into complex flow physics inside control valve. We can extract valve
characteristics without time-consuming and expensive experimentation. CFD enable designers a possibility
to understand on how geometrical and installation changes might impact valve
performance. This will help innovate and redesign. CFD helps in reduce
life cycle cost of a valve from conceptual design to market. Apart from this, computers
are becoming powerful and cheap, which further suggests making us of CFD to the
fullest to reduce cost and time.
Guest post by :Centre for Computational Technologies, Pune