Guidelines for using Transient Blade Row features are described in the following sections:
CFX-Pre Turbomachinery Wizard:
When setting up your transient blade row problem, it is always recommended that you use the Turbomachinery wizard in CFX-Pre.
Flow initialization:
A steady-state solution can be used as the initial condition for all transient flow simulations. You may use the Stage model (mixing-plane) or Frozen Rotor method, when modeling multiple blade rows.
Time-integration:
For all transient blade row simulations that have Transient Method > Option set to
Time Integration, a uniform time step must be used.You should not change the time step (number of time steps per period or timestep multiplier) in the middle of a simulation. However, you can stop a simulation, change the time step, then restart (continue) the simulation. For details, see Stopping and then Restarting Simulations with an Increased Number of Time Steps Per Period.
Inflow and Outflow Boundary Conditions:
For proper comparison between the full-domain solution and any of the transient blade row solutions, the inflow and outflow boundary conditions must be equivalent. For example: an average pressure specification on an outlet boundary should not apply on the full extent of the outlet in the full domain model but rather it should be applied to individual passage outlets to replicate the transient blade row solution boundary condition specification. Therefore it is advised to divide the full-domain inlet and outlet boundaries into multiple boundaries accordingly.
For an inlet disturbance problem: When working with profiles as boundary conditions, be sure to double-check the periodicity of the profiles on both reference test cases and the transient blade row test cases. You can use monitors at the inlet to do this check.
Solution Monitoring:
You can use solution monitor points to monitor the results of the transient simulation.
When comparing the full-domain reference solution with the transient blade row solution, it is important to locate the monitors in the passages of the full-domain geometry with relative locations as in the transient blade row passage configuration.
Pitch ratio:
There is a limit on the range of pitch-ratio allowed when using the Time Transformation method. This range is strongly dependent on the Mach number associated with the rotor rotational speed (or signal speed in the case of an inlet disturbance problem). The pitch-ratio limit is defined in Time Transformation.
Solution Monitoring
The Time Transformation method does not support monitor points that use CEL expressions. Instead, you can use solution variables (pressure, temperature, and so on) to monitor the solution.
The Fourier Transformation method works for compressible as well as incompressible flows and for all ranges of disturbance pitch ratios from small to very large. For example, the Fourier Transformation method can handle a once-per-revolution disturbance on a blade passage, which is a situation that cannot be handled with the Time Transformation method.
However, in general for compressible flow and with small pitch ratios (typically between 0.75 and 1.35) it is recommended that you use the Time Transformation method instead because the calculation will be more efficient than when using the Fourier Transformation method.
When not using the Turbomachinery wizard in CFX-Pre, an additional step is needed for creating turbo topology; you must create two instances of the blade passage.
There are two options by which you can create a copy of the original passage:
Perform a rotation on the copy of the original geometry (mesh) by its own pitch. The rotation angle must be specified with many decimal digits for minimum round-off error.
Perform a turbo rotation and update the number of passages to model and the number of passages in 360° in each domain.
When running a case in serial, it is highly recommended that you run CFX-Solver using double precision. This recommendation also applies for steady-state cases that are used to initialize Fourier Transformation cases.
When running a case in parallel:
You may run CFX-Solver using single precision, but it is highly recommended that the partitioning is done using double precision. This recommendation also applies for steady-state cases that are used to initialize Fourier Transformation cases.
Each pair of sampling/phase shift corrected interfaces must belong to a single domain. This is a condition that should be guaranteed if CFX-Pre Turbo Wizard is used to set up the case.
When setting up a Fourier Transformation case, any domain interface that is part of the Fourier Transformation Transient Blade Row setup should not span more than one domain.
It is very important that the boundary condition is periodic in time for the passages being modeled. In the case of blade flutter, this can be achieved by selecting a number of timesteps that is a multiple of
.In cases where you observe solution instabilities, you should apply frequency filtering. Instabilities are typically seen in elongated domains where the periodic signal becomes very weak at the furthest point away from the source of the disturbance. Frequency filtering removes unwanted frequencies from the solution thereby avoiding the onset of instability. The setting for enabling frequency filtering is described in Frequency Filtering in the CFX-Pre User's Guide.
Instabilities could also exist related to the mass flow Fourier reconstruction for cases with high speed flow where a shock hits the periodic interface. The recommended workaround is to turn off the mass flow Fourier reconstruction with one of the following expert parameters:
For inlet disturbance and flutter cases:
ps reconstr mflow from sfc = fFor cases involving a transient rotor stator Fourier interface:
ps reconstr mflow from sfc trs = f
Animations:
Because the resulting animations of transient blade row cases are usually periodic in time, you must take care not to repeat the start/end point of the period when these are identical. The animation panel has an advanced option to avoid this (the option is named Don't Encode Last MPEG frame), and you should select this option when working with a transient blade row animation.
Note that, for a continued run of a transient blade row case, the
Accumulated Time Stepvariable does not reflect the solver run history.Note that hybrid values of injection quantities might not be available for FT and HA cases.