Duke Energy uses a GE Frame 5 gas turbine to drive a compressor used in a natural gas pipeline. The intake system required retrofitting with a silencer to meet current regulations and the need also existed to reduce particle contamination. The pressure drop in the initial design was found to be 8.5 inches of water, much higher than the design specification which was 5.3 inches.
Mueller Environmental Designs Inc. which provides air filtration, noise control, evaporative cooling, mist extraction, and emissions control products to the oil and gas transmission industry, worked with consultants from ANSYS, Inc. to simulate flow in the inlet duct and identified variations in flow velocity across the cross-section of the inlet and a large recirculation zone. They made a series of design changes including modifying the inlet scroll configuration, adding radii to the ductwork, and adding guide vanes to the outlet region of the duct. Then they evaluated the effects of these changes by re-running the simulation. The changes reduced the pressure drop to just 4.2 inches water.
Updating The Duke Intake System
The four Duke Energy compressor stations with a total of eight compressors were installed in the 1960s and originally equipped with intake systems using single-stage fiberglass filters. Recently, Duke Energy decided to update these intake systems. The primary objective was to reduce inlet pressure differential. Other objectives included improved inlet air cleanliness and reduced inlet air noise. The filters in the old intake systems had to be cleaned frequently to protect the turbine blades from being eroded, corroded or deformed due to dirt.
The design specification mandated limits on the pressure drop of the intake system because additional fuel must be burned by the engine to overcome the pressure drop. At the same time, the intake system had to make several sharp turns to get into the compressor building. Space considerations placed severe limitations on the overall allowed length of the intake system, creating a substantial design challenge.
Mueller engineers designed a new intake system that includes a silencer and state-of-the-art filters. But they were concerned about the amount of pressure drop generated by the new design. The geometry of the intake system was far too complex to calculate the pressure drop using engineering calculations.
The standard approach for addressing this sort of problem begins after the equipment is installed. Then, if performance does not meet specifications, physical measurements of airflow are taken at various locations along the ductwork. Based on these measurements, engineers attempt to determine what is causing the problem. Based on this diagnosis, some type of flow control device is installed within the ductwork. Then the measurements are repeated to see if the device actually solved the problem.
The problem with this approach is that it is very expensive to send engineers and technicians out to a customer’s site to perform measurements and modifications. In addition, customers are often anxious to get their equipment up and running and do not welcome the delay caused by this kind of iterative process. Chances are that the first design iteration might not be successful and in that case additional construction expense and downtime will be required to modify and retest the design. Physical testing usually provides little diagnostic information on why a particular approach is not working, so many iterations may be required in order to get the design right.
Using Simulation Instead Of Trial And Error
Mueller engineers decided instead to use computer simulation to evaluate their initial design. A computational fluid dynamics (CFD) simulation provides fluid velocity, temperature, and chemical concentration values throughout the solution domain for systems with complex geometries. As part of the analysis, a designer may change the geometry of the system or the operating conditions and view the effect on fluid flow patterns. Mueller contracted with consultants from ANSYS, Inc., to perform fluid flow simulation.
Figure 1 shows the geometry that was included in the analysis, from the inlet, through the filter and silencer to the slant duct, then through the horizontal duct, 90-degree bend, an additional silencer and on to the duct outlet. The filter was modeled as a porous medium, with the pressure drop specified as a local function of velocity. This methodology allows the simulation of the essential effects of the filter on the flow without resolving its extremely fine-scale geometric details. The pressure drop for the entire system was found to be 8.5 inches of water, which confirmed concerns that it might be above the allowable level. Additionally, the k-epsilon turbulence model used in the analysis showed very high levels of turbulence downstream of the second silencer, which was unacceptable for the inlet to the turbine.
Figure 1: Computational fluid dynamics (CFD) model of the original design of filter and silencer duct system.
Diagnostic Information Leads To Quick Solution
The CFD analysis results provided a substantial amount of diagnostic information that guided Mueller engineers and ANSYS consultants in their efforts to resolve these concerns. First of all, the analysis results of the original design (Figure 2) showed substantial variations in velocity, particularly in the region near the outlet. These variations meant that only a small portion of the cross-section of the duct was actually being utilized by the air moving into the turbine. The variations were clearly caused by a large recirculation zone near the outlet.
The recirculation zone in turn was associated with high levels of turbulence within the duct, ranging up to 23.6 square meters per second. The CFD analysis results also helped understand what was causing the turbulence. In several areas of the duct, the air had to make sharp turns, and when this occurred centrifugal force concentrated the air on the outside edge of the duct, generating turbulence and increasing pressure drop.
[inline:Figure 2 – right.jpg][inline:Figure 2 – left.jpg]
Figure 2: Outlet region of duct without second silencer or vanes, showing (left) contours of velocity magnitude with a large recirculation zone and (right) high levels of turbulence kinetic energy.
The use of simulation made it possible for Mueller engineers and ANSYS analysts to work together to propose possible design changes that had the potential to reduce the turbulence.
If a design change appeared promising, the analysts modified the model and then re-ran the simulation to determine the impact on the performance of the intake system. The first change was modifying the inlet scroll configuration to add four turning vanes. Re-running the analysis showed that the vanes significantly reduced turbulence but that there were some additional turbulence points within the duct systems.
Mueller engineers suggested changing the sharp angles in the original design to radii in order to smooth the flow in the system. The ANSYS consultants modified the design and re-ran the analysis, which showed further improvements but still indicated some turbulence was being generated in the filter housing. Mueller engineers proposed additional modifications to the geometry, and – when implemented – the analysis results showed that turbulent kinetic energy had been reduced to 4.88 square meters per second. (Figure 3).
[inline:Figure 3 – right.jpg][inline:Figure 3 – left.jpg]
Figure 3: Outlet region of duct with guide vanes showing (left) contours of velocity magnitude with improved flow distribution and (right) greatly reduced levels of turbulence kinetic energy.
The final design showed that the variations in velocity across the cross-section of the intake system were substantially reduced, which in turn resulted in the pressure drop being reduced to just 4.2 inches of water.
These improvements were made entirely through the use of software prototypes without having to build hardware or shut down the compressor. The intake system was built and installed only after the design had been validated in the simulation domain. Since it was installed, the new intake system has worked almost exactly as predicted, meeting all the customer’s requirements and providing a lower pressure drop than specified which will mean substantial fuel savings.