CFD Analysis of a 15-Stage Axial Compressor
ANSYS CFX software has been used to perform a full 3-D flow analysis of 15-stage axial compressor in one computation. Because these simulations included tip gaps, mass bleeds, hub leakage flows, and ranged from single passage to full 360 degree analysis, this was an extremely challenging project. Within the CFD simulations, various effects were analyzed: mesh style and refinement, boundary conditions, steady or transient, tip clearance and numerical issues including turbulence model choice and advection model choice. The total number of nodes used for the entire simultaneous analysis of the inlet guide vane (IGV) and 15 stages (31 components, single passage) was approximately 6.2 million nodes. For the 360 degree simulation, a mesh with a total of 32 million nodes was used to compute the entire middle section of the compressor (5 stages).
Overview of the mesh for the 15 stage, single passage analysis
Mesh (32 million nodes) for the 360 degree analysis of the middle portion of the compressor (5 stages)
Convergence history for a typical simulation with a 2nd order advection scheme on the fine mesh, from the initial guess to convergence. Tight steady-state convergence is obtained in less than 120 iterations.
Mass flow rates are usually required for high power output of the gas turbine. The high mass flow rate is critically dependent on the first compressor stage where choking can occur. The mass flow rate is ultimately limited by the mechanical stresses, operating range and the stage efficiency. The figure below illustrates the normalized relative Mach number evolution near the tip region for the first stage rotor. The local peak Mach number is on the suction side of the rotor. Under these particular operating conditions, choke occurs at about 82% of the chord at the suction side and 15% of the chord at the pressure side.
Relative Mach number
The chart below shows the dimensionless static pressure evolution versus axial distance. The axial position 0 corresponds to the inlet of the guide vane while position 1 is the outlet of the 15th stator. “Pref” corresponds to the static pressure at the last casing measurement point. The CFD data points are obtained at all nodes along the periodic/shroud line of intersection. The experimental data was obtained from a series of casing pressure taps, located ahead and behind of each stator blade. The agreement between the static pressure measurements and numerical results is extremely good, even for the coarse mesh. Small differences are seen between the coarse mesh and the fine mesh. An expected disturbance in the predicted static pressure is seen at each mass bleed location on the casing (mb1 to mb5). The largest mass bleed occurs near the trailing edge of stator 13 (axial position of approximately 0.92), corresponding to the largest casing static pressure disturbance. The agreement between the static pressure measurements and numerical results is satisfactory. CFD predictions are obtained from simulations on the coarse mesh, and on the fine mesh with and without leakage flows. The agreement to data is similar for all three simulations. The result from the simulation including leakage flow is slightly closer to the data. The main effect of the stator leakage flows is to disturb the boundary layer upstream of the stator passage, as well as cause a slight shift in the meridional mass distribution as the stator passage accommodates the increased local passage mass flow.
Comparison between predicted and experimental static pressure on the shroud
The chart below illustrates the dimensionless total pressure developments, along the span, for three different axial positions located in front of stator 5, stator 7, and stator 14. In these figures, Pref corresponds to the total pressure at the exit of the compressor respectively. The comparison between the predicted results and measurements shows a good agreement in both radial gradients and absolute values. The leakage flow introduces additional losses mainly due to a reverse flow near the hub region. Thus, modeling the leakage flows better predicts the flow near the end wall regions.
Local profile developments of the total pressure
References
[1] T. Belamri, P. Galpin, A. Braune and C. Cornelius, “CFD Analysis of 15 stage axial compressor Part I: Methods”, GT2005-68261, ASME Turbo Expo Conference, Reno June 6-9, 2005.
[2] T. Belamri, P. Galpin, A. Braune and C. Cornelius, “CFD Analysis of 15 stage axial compressor Part II: Results”, GT2005-68262, ASME Turbo Expo Conference, Reno June 6-9, 2005.