Rotorcraft Icing Computational Tool Development

The formation of ice over lifting surfaces can affect aerodynamic performance. In the case of helicopters, this loss in lift and the increase in sectional drag forces will have a dramatic effect on vehicle performance. The simulation of rotorcraft flow fields is a challenging multidisciplinary problem that lags in development over its counterpart in the fixed wing world by more than a decade. Successful aerodynamic simulation of a rotor/fuselage system requires the modeling of unsteady three-dimensional flows that include transonic shocks, dynamic stall with boundary layer separation, vortical wakes, blade/wake and wake/wake interactions, rigid body motion, blade deformations and the loss of performance caused by ice accretion.

(Photo Courtesy U.S. Army)

Stand-alone ice accretion prediction tools, as well as ice accretion fully integrated with aerodynamics, currently exist for 2D airfoils and 3D aircraft configurations. Ice accretion predictions are typically two-dimensional in nature and based on the classical Messinger model. The analysis consists of four critical steps: flow-field calculation, water droplet impingement calculation, heat transfer prediction, and ice accumulation normal to the surface.

LEWICE and LEWICE3D

LEWICE is NASA's flagship code for 2D ice accretion prediction, and it is the core of the 3D ice accretion tools as well. LEWICE development was initiated in the early 1980s, with the first general release (Version 1.0) in 1991. Four major updates to the code followed, in 1993 (Version 1.3), 1995 (Version 1.6), 1999 (Version 2.0) and 2002 (Version 2.2). Recent updates were released in 2005 (Version 3.0) and 2006 (Version 3.2), and a mixed-phase modeling capability was added in 2008.

The code uses a potential panel method to determine the flow field about a clean surface, then calculates water droplet trajectories from some upstream location until they impact the surface or until the body is bypassed. Collection efficiency is then determined from the water droplet impact location pattern between the impingement limits. A quasi-steady analysis of the control volume mass and energy balance is performed next using a time-stepping routine. Density correlations are used to convert ice growth mass into volume. LEWICE also features multiple drop size distributions, multiple airfoil elements, thermal models for anti-icing/deicing systems, and an interface with structured grid codes, allowing the use of viscous Navier-Stokes flow solutions.

The thermal models in LEWICE combine the features of previous codes, LEWICE/Thermal and ANTICE, to simulate de-icing and anti-icing with electrothermal or hot air systems. Features are included to allow determination of optimized heater sequencing (for electrothermal analysis) and multiple boundary conditions (for bleed air analysis).

LEWICE has been thoroughly validated for a wide range of conditions, with a database of over 3,000 ice shapes on 9 different geometries. Validation has been documented in numerous papers as well as NASA reports. The validation database lies mostly within the Appendix C continuous maximum or intermittent maximum envelopes, but there are some exceedance and supercooled large droplet conditions for comparison as well. This validation, along with significant research into recommended test methods and advanced component models, has led to a degree of acceptance for use in reducing the cost of development and certification programs.

However, this acceptance does not exist for rotary-wing applications.

LEWICE does not simulate a fully rotational system, but does allow the user to input a number of simple parameters-distance from the hub to the 2D section of interest, rotation speed, and orientation of the plane of rotation (vertical for propellers, horizontal for rotors). LEWICE performs three additional calculations in rotating body cases. The rotation speed is used to calculate an increase in the aerodynamic heating term in the energy balance, the rotational force is included in the ice shedding determination, and the rotational force is used to find the resultant force of the shed ice particle, which is then used to track the particle after it is shed. The rotating body information is not used by the potential flow solver in LEWICE, nor is the rotating body information used by the trajectory equation.

Figure 1. Methodology for Developing a 3D Ice Shape.

LEWICE3D is a suite of codes, developed by NASA and used widely by industry, to determine the amount and location of ice accretion on an aircraft. It is used to calculate water loading on aircraft surfaces so that the size and location of ice protection systems can be determined, to optimize the placement of icing sensors, and to determine ice shapes used in failed ice protection system tests. It is also used to determine corrections for cloud measurement instruments, such as droplet size probes or liquid water content probes on NASA research aircraft.

LEWICE3D uses a Monte Carlo-based collection efficiency calculation using droplet impact counts. Trajectories are calculated using an Adams-type predictor- corrector method. Tangent trajectories and collection efficiencies for simple 2D or 3D regions can also be calculated using a modified version of the 2D LEWICE method. Streamlines are calculated using a 4th-order Runge-Kutta integration scheme.

The ice growth methodology in LEWICE3D uses a single time step strip approach and requires a steady or time-averaged flow solution, supplied by the user. The strip approach is based on the classical Messinger energy balance procedure with an integral boundary layer technique used to generate heat transfer coefficients, and is a modified version of the method used in the 2D LEWICE code applied along streamlines. LEWICE3D supports multi-block structured grids, adaptive Cartesian grids and unstructured grids, as well as panel-based binary-tree grids.

LEWICE3D includes extensions which allow generation of a full 3D ice accretion for surfaces and generation of a new iced surface, calculation of off-body concentration factors, and determination of shadow zones. The program has been parallelized using OpenMP and MPI (message passing interface) to complete jobs faster on parallel machines. The parallel version has been ported to SGI and Linux machines.

However, the current NASA icing codes, LEWICE and LEWICE3D, cannot be applied directly to the ice accretion of rotorcraft flows for several reasons. These codes are acceptable for the majority of fixed-wing applications, such as general aviation, business jets or commercial transports, but there are still shortcomings for some vehicle types, notably rotorcraft.

Past Research in Rotorcraft Icing Codes

The importance of rotorcraft airfoil oscillation during ice accretion was recognized in the early 1980s and NASA sponsored an icing wind tunnel test of a 2D six-inch chord oscillating airfoil. This Sikorsky-run test confirmed that the variation in angle of attack altered the ice shape and produced changes in the drag coefficient. About this same period of time, Sikorsky and the United Technologies Research Center designed, fabricated, and tested in dry air 2D and 3D models with a chord of 17 inches. Meanwhile, as part of an earlier project, Sikorsky designed and fabricated an airfoil test rig to span the NASA Icing Research Tunnel (IRT) test section. This apparatus had a chord of 15 inches and mounted to the upper and lower IRT turntable. Maximum angle of attack for the airfoil was 10 degrees at the maximum operating speed of the IRT of 250 knots (with the model installed and with ice on the model leading edge), with an angle of attack of at least 20 degrees at 150 knots in the IRT.

To further research on rotorcraft icing, a Government-industry consortium, composed of NASA, Texas A&M University, Bell Helicopter Textron, Boeing Helicopters, McDonnell Douglas Helicopters, and Sikorsky Aircraft, was created to better understand the impact of rotor blades ice accumulation on aircraft performance, increase in vibration and ice shedding. The program was to also validate the industry existing performance models and assessed the benefits of rotor blade scaled model testing.

A two-model approach was selected as the most effective means to accomplish the program goals. A lightly instrumented OH-58 tail rotor that had been modified to operate as a main rotor was chosen as the initial test article.

The initial experimental program was conducted in 1988 in the NASA Lewis Research Center Icing Research Tunnel in which the OH-58 tail rotor assembly was operated in a horizontal plane to simulate the action of a typical main rotor. Ice was accreted on the blades in a variety of rotor and tunnel operating conditions and documentation of the resulting shapes was performed. Rotor torque and vibration were recorded and presented as functions of time for several representative test runs, and the effects of various parametric variations on the blade ice shapes were shown. This OH-58 test was the first of its kind in the United States.

Figure 2. Approach for Simulating Oscillation of Rotor.

Based on the results of these two tests, it was clear that the CFD methods developed during future studies must include airfoil oscillation and that data must be acquired to validate the method(s). Although investigations into rotor blade ice protection systems continued 1997-99, and commercial use of the IRT continued, NASA essentially put rotorcraft icing research on hold in 1994 to focus on fixed wing following a highly visible fixed-wing accident.

Progress in Tool Development

When opportunity finally arose again at NASA in 2004-2005, the state-of-the-art of subsonic rotary wing icing was quasi-steady and quasi-3D. Grid generation for complex icing shapes was 2D and interactive or crudely automated. There could be a variously loose coupling between the aerodynamics and the ice accretion-but not yet a true, multi-phase solution. The icing analysis procedure for prediction of performance degradation must address both aerodynamic performance and ice accretion. Aerodynamic performance degradation involves the calculation of the aerodynamic coefficients of the iced geometry. The section lift, drag and moment characteristics with ice must be accurately known in order to predict the performance degradation from an icing encounter. Lifting line theory is one commonly used method, but momentum source methods coupled with blade element theory are also widely used, as are panel methods. Often, proprietary methods or Navier-Stokes solvers are used to calculate aerodynamics.

NASA's objective was a robust, validated coupling of a rotor performance code with an ice accretion code. The CFD analysis of a clean rotorcraft configuration is well within the range of current technology. Fully time accurate simulations of ice accretion on rotorcraft blades are not currently feasible, however. The complexity of the problem demands high-fidelity tools based on first principles, and a tightly-coupled, physics-based approach is not currently available.

Rotorcraft aeromechanical studies involve coupling the rotor aerodynamics with the structural dynamics of the system. The airloads computed by the CFD solver is used to drive a forced response simulation with the CSD solver. The computed structural deflections are used in the CFD analysis, leading to a change in the airloads. The two solvers are thus inherently coupled. The CFD-CSD coupling may be performed primarily in two ways - loose and tight. In tight coupling, the data is exchanged every time step of the simulation. In loose coupling, the data is exchanged between the two solvers at periodic intervals, typically once per revolution. Since loose coupling is driven by the inherent periodicity in the solution, it is used for analysis of rotors in steady flight conditions.

Coupling Ice Accretion Models with Aeromechanics

Model Rotor in the Icing Research Tunnel

One successful approach is an integrated tool set capable of modeling ice accretion and the overall effects of rotor performance. This loosely coupled suite of tools (LEWICE, GT-Hybrid, and DYMORE) has been applied to a representative rotor for detailed study. The entire process (clean rotor grid generation, clean rotor analysis, ice accretion simulations, and iced rotor analysis) is automated and modular. NASA entered into a two-year cooperative agreement with the Georgia Institute of Technology, to develop improved coupling techniques for icing computational fluid dynamics. Georgia Tech was partnered with the Sikorsky Aircraft Corporation.

The project utilizes a 3-D Navier-Stokes analysis and a multi-body dynamics tool, coupled with the GT-Hybrid unstructured Cartesian grid-based flow solver to represent the ice shapes. Several different Navier-Stokes flow solvers have been used in this framework including OVERFLOW, TURNS, and GT-Hybrid. GT-Hybrid, a three-dimensional unsteady viscous compressible flow solver that uses a free wake solver to model the effects of the rotor wake. The flow is modeled from firstprinciples using the Navier-Stokes methodology. The three-dimensional unsteady Navier-Stokes equations are solved in the transformed body-fitted coordinate system using a time-accurate, finite volume scheme. A thirdorder spatially accurate Roe scheme is used for computing the inviscid fluxes and second order central differencing scheme for viscous terms. The Navier-Stokes equations are integrated in time by means of an approximate implicit time marching scheme. A Spalart-Allmaras turbulence model is used to compute the eddy viscosity. The flow is assumed to be turbulent everywhere, and hence no transition model is currently used.

A single blade is resolved in the Navier-Stokes domain. The influence of the other blades and of the trailing vorticity in the far field wake is accounted for by modeling them as a collection of piece-wise linear bound and trailing vortex elements. The near wake is captured inherently in the Navier-Stokes analysis. The use of such a hybrid Navier-Stokes/vortex modeling method allows for an accurate and economical modeling of viscous features near the blades, and an accurate “non-diffusive” modeling of the trailing wake in the far field.

Coupled CFD/CSD Analysis for Rotorcraft in Forward Flight

Another icing analysis process that has been developed involves the loose coupling of OVERFLOW-RCAS for rotor performance prediction with LEWICE3D for thermal analysis and ice accretion. This method uses three-dimensional analysis for rotor performance and degradation and two-dimensional analysis for ice accretion. The automated process allows for rapid analysis in a parametric study or for the analysis of an airfoil subject to the many conditions existing on a rotor. For validation, predictions of performance and ice shapes were compared with experimental data for rotors in hover and in forward flight.

NASA entered into a two-year contract with the Boeing Company (Ridley Park, PA), to develop these improved coupling techniques for icing computational fluid dynamics. The project resulted in a process by which OVERFLOW, RCAS and LEWICE can be loosely coupled to assess ice accumulation and rotor performance degradation for helicopters in forward flight, as shown in Figure 1. The system has been tested and evaluated using existing wind tunnel or flight data, and effort is still ongoing. The result is a computational approach for performing high fidelity simulations with ice accretion of rotorcraft blades. The approach is appropriate to address ice accumulation on rotors in flight regimes from hover to high-speed forward flight.

The high fidelity icing analysis approach developed for rotor systems follows three basic steps:

Comparison between scanned data (top) and photograph (bottom) of ice shape on a rotor blade.
  • Establish rotor trim, clean rotor performance and the initial flow field environment using CFD or coupled CFD-CSD as appropriate;
  • Extract representative 2D airfoil conditions for blade sections at radial and azimuthal locations and predict ice buildup on the rotor accounting for the diverse operating environment of the rotor;
  • Reestablish rotor trim and performance for the iced blades.

Rotor blades experience pitch oscillations as they rotate around the shaft. Pitch oscillations introduce time varying conditions that influence the distribution of ice along the leading edge. The process to predict ice on an oscillating airfoil is built on the premise that the shape is not a strong function of the frequency of the oscillation and is predominantly influenced by the mean and amplitude of the pitch variation. With this assumption, the time history of an oscillating airfoil can be represented by a very slow moving blade. Furthermore, if we assume the shape can be approximated by only considering the mean angle of attack and the extreme excursions from the mean, the blade motion can be represented as the series of quasi-static events, as shown in Figure 2.

Once the ice has been established on the blade, the CFD-CSD analysis process is repeated for the iced rotor. The 3D rotor grid is modified to account for the ice shape on the blade. The input to the CSD analysis is also modified. Accreted ice adds to the blade section mass and chordwise inertia. It is assumed that accreted ice has no effect on stiffness properties. The mass of ice is determined from post-processing the icing analysis, and the placement of the ice is assumed to be at the section leading edge. Updates to the section mass, center of gravity offset, and chordwise inertia are computed and used in RCAS. Rotor performance degradation is obtained by comparing the forward flight performance characteristics of the iced rotor to the baseline rotor.

High Resolution CFD Analysis of Rotorcraft Icing

Additionally, various Eulerian approaches (with both one-way and twoway coupling) for simulating ice accretion have also been examined, but significant work is still required in order to fully demonstrate this alternate method. NASA entered into a two-year cooperative agreement with the Pennsylvania State University to develop improved coupling techniques for icing computational fluid dynamics. Penn State partnered with Bell Helicopter Textron, Inc.

The project applied a zonal approach to the unstructured FUN3D flow solver, extended the existing NASA LEWICE ice accretion formulation to the rotorcraft environment, and coupled this module with the outer CFD flow solution. Initial validation will be conducted by comparison with existing test data. The work will result in an advanced software tool for performing high fidelity CFD simulations with ice accretion of rotorcraft blades. Work is currently ongoing on the thermal modeling for generalized rotorcraft flows, although some issues include the computation of recovery temperature with variable stagnation conditions, transitional flow effects, and working out details to handle them with access to a limited amount of data.

Conclusions

An integrated tool set capable of modeling ice accretion and the overall effects of rotor performance was developed and demonstrated. Key computational parameters were explored, and preliminary results for cases of practical interest were encouraging. Modifications to LEWICE were demonstrated which allow for the retention of previous time-step ice shapes.

Preliminary development of a three-dimensional Eulerian analysis for modeling droplet impingement was also undertaken, to improve more efficient calculation of collection efficiency. The development of a tightly-coupled truly multi-physics approach is still a goal to work towards, but several promising efforts have been undertaken recently. Additional effort is still needed to improve methods for predicting rotor blade shedding and de-icing/anti-icing system performance. An icing analysis process involving the loose coupling of OVERFLOW-RCAS for rotor performance prediction with LEWICE3D for thermal analysis and ice accretion was developed and demonstrated. The method uses 3D analysis for rotor performance and 2D analysis for ice accretion. For validation, predictions of performance and ice shapes were compared with experimental data for rotors in hover and in forward flight.

Studies have also been conducted to examine the effects of grid spacing, grid density, turbulence model, flow-field update frequency and number of spanwise cuts, to name a few. Likewise, simulations of ice accretion prediction and associated rotor performance degradation have been conducted for multiple 2D airfoils and for various 3D rotors in hover and in forward flight. Ice accretion and detailed aerodynamic measurements for 2D clean and oscillating airfoils undergoing both steady and transient behavior was obtained in the IRT. Ice accretion, rotor performance and de-icing/anti-icing system behavior was obtained for a 3D rotating tail rotor in the IRT. For the first time the coupling of an icing code with a computational fluid dynamics code and a rotorcraft structural dynamics code has been demonstrated. The codes and research conducted here are already being transitioned and used by industry.

This article is based on SAE Technical paper 2015-01-2088 by Richard E. Kreeger, NASA John Glenn Research Center; Lakshmi Sankar, Georgia Institute of Technology; Robert Narducci, Boeing Co.; and Robert Kunz, Penn State University, doi:10.4271/2015-01-2088.