Moving mesh or MRF?

In this article, we describe a recent work in which we compared the advantages and disadvantages of the steady and transient approaches to the analysis of a four-bladed aircraft propeller. In the study, we examined a concept model of a ground based turbo-prop engine operating within an enclosure.

Fig.1a: CAD generated concept design of engine, stand and enclosure	Fig. 1b: CAD geometry built in STAR-Design– the separately created propeller mesh was inserted into the cylindrical region.

As engineers routinely applying CFD to a wide range of turbomachinery and aerospace applications, we often face technical judgments as to the applicability of certain numerical approaches or physical models used in a simulation. One such judgment for rotating machinery relates to the choice of either applying a time accurate transient moving mesh approach or a simplified steady-state multiple rotating frames (MRF) approach. This modeling decision can be critical especially when simulating the flow through rotating systems which contain a low blade count.

Fig. 2: Surface mesh used to begin automated pro-STAR trim mesh, (rotating region inside cylinder)	Fig. 3: Final trim mesh; propeller mesh was trimmed separately from surrounding trim mesh and then assembled for flow analysis

We used STAR-CD to consider several facets of the propeller design, principally:
i) maximum torque load on the propeller blades
ii) time varying cyclical loading of the blades
iii) mass flow through the system
iv) engine outlet temperature
v) flow over the tip of the propeller blades

Our overriding question concerned the trade off between the expense of the numerical calculation technique and the accuracy of the solution it predicted. In order for a computational method to qualify as a valid and useful simulation technique, calculations are required to be both accurate and practical. We needed to understand whether the Implicit MRF approach could meet the technical challenge and whether the transient moving mesh approach could meet the schedule requirements of the project.

Fig. 4: Pressure boundaries assigned to inflow and outflow regions	Fig. 5: Comparative view of velocity magnitude for steady and transient analyses (isometric view)

The CAD geometry of the concept design examined in the study was built in STAR-Design (Figure 1). All the components were created and meshed separately, using trimmed cell technology (Figure 2) before being assembled into a single model (Figure 3). The final assembly consisted of 1.75 million computational cells. Fixed pressure boundaries were prescribed at inflow and outflow regions, and
rotating wall boundaries to the surface of the propeller (Figure 4). The flow was considered compressible, consisting of large temperature gradients in the system due to the hot exhaust of the gas turbine engine. Identical flow properties, solver settings and geometric configurations were simulated for both the MRF and moving mesh approaches.

Fig. 6: Individual propeller blades were monitored for torque loads	Fig. 7. Transient analysis shows cyclic torque loading of blades as they rotate through 360 degrees (3 of the 4 blades are monitored here; peak-to-peak represents a single propeller revolution)

Our analysis revealed that both the steady MRF and the transient moving mesh approaches proved meritorious. The steady analysis was computationally stable and converged monotonically in a timely fashion. The steady simulation captured the basic flow structure across the propeller tips, as well as the temperature mixing of the engine exhaust. The MRF predicted a fixed torque loading on the blades; however, due to the steady nature of MRF, the analysis is not able to predict the cyclic torque loading that the blade experiences naturally during rotation. The steady analysis provided quick, general results in which the gross flow structure was predicted (Figure 5).|

The transient moving mesh analysis provided more than the gross flow structure; the analysis additionally provided critical engineering data concerning the effects of the blade rotation in time. Specifically, we noticed a high torque loading experienced by all blades as they passed a particular point in the 360° revolution (Figure 6 & 7). The time accurate results of transient blade loading provided torque spike magnitudes; the results allowed us to determine if additional engineering of the engine mounting system was warranted to mitigate the high cyclic loading. The transient analysis also captured the temperature mixing as did the steady analysis, and predicted a system mass flow rate 4% higher than that of the steady case. The transient simulation required approximately 4-6 times more computational runtime to establish a “cyclically steady” solution, yet the analysis provided more insight for understanding the flow physics of the system.

From our examination, we conclude that the transient moving mesh analysis more appropriately captures high resolution, high accuracy flow behavior and cyclic fatigue characteristics. Although MRF is less expensive and acceptable for understanding the basic flow structure, the steady state MRF approach is not able to provide potentially critical time accurate information.

Jacobs Sverdrup provides a range of advanced technology engineering services to government and industry. One of our core customer bases is the aerospace and defense industry, for which we deliver a full range of design and build services for aero-propulsion and space systems facilities.

For further information, contact: connorch@sverdrup.com