Modeling and experimental testing of a centrifugally powered pneumatic de-icing system for rotor blades

Open Access
Bailey, Matthew John
Graduate Program:
Aerospace Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
Committee Members:
  • Jose Palacios, Thesis Advisor
  • Deicing
  • Deicer
  • Icing
  • Pneumatic
  • Rotorcraft
  • Helicopter
  • Centrifugal
The goal of this research was to test, evaluate, and model a novel pneumatic approach to protect helicopter rotor blades from ice accretion. The system relied on centrifugally generated pressures to deform a 0.02 in. (5.08 x 10^-4 m) thick titanium leading edge cap. This leading edge cap was protected from erosion by a titanium aluminum nitride (TiAlN) coating. Six pneumatic diaphragms were installed beneath the titanium leading edge. The diaphragms were normally deflated under vacuum against the surface of the blade and were inflated when the ice accretion thickness reached a critical value. The deformation of the leading edge introduced transverse shear stresses at the interface of the ice layer that exceeded the ice adhesion strength of ice to the TiAlN coating (126 psi, 8.68 x 10^5 Pa) promoting instantaneous ice debonding. The applied input pressures to the system (+/- 3.7 psi, +/- 2.55 x 10^4 Pa) were representative of the pressures generated centrifugally by a medium size helicopter rotor system. With these pressures, the maximum deformation of the leading edge was quantified to be 0.2 in. (5.08 x 10^-3 m). The aerodynamic performance degradation effects related to the leading edge deformation were quantified during low speed (Re=1,000,000) wind tunnel testing. Results were compared to the previously recorded aerodynamic performance degradation data due to ice accretion. It was measured that the penalties related to the deployment of the pneumatic diaphragms were 35% lower than the aerodynamic drag penalty due to ice accretion. The lower aerodynamic penalty of deploying the proposed de-icing concept with respect to that of ice accretion case indicated that the system would not introduce any aerodynamic penalty while removing accreted ice. The system was tested under representative rotor icing conditions at the Penn State Adverse Environment Rotor Testing Stand (AERTS). The system was subjected to centrifugal loads that ranged from 110 to 514 times the force of gravity. The de-icing system successfully promoted instantaneous shedding of ice thicknesses ranging from 0.06 in. to 0.1 in. (1.52 x 10^-3 m to 2.54 x 10^-3 m) and for varying icing conditions within the Federal Aviation Regulations (FAR) Part 25/29 Appendix C Icing Envelope. A finite element model (FEM) was developed to predict the de-icing capabilities of the pneumatic system. The cohesive zone method was selected to model the ice/leading edge interface. This modeling approach was mesh-dependent and the cohesive interface parameters needed to be determined for the final converged mesh density and testing conditions. Two bench top tests were conducted and modeled to empirically quantify the failure criteria used in the FEM. All experimental tests were conducted under the same freezer icing conditions. Aluminum with the same surface roughness was used across all three tests. The experimental shear adhesion strength criterion for this aluminum was quantified to be 39.94 psi (2.75 x 10^5 Pa). The experimental peel adhesion strength failure criterion was gathered in a mixed-mode test. The force require to delaminate an ice patch at the root of a plate in bending was quantified. This loading condition which induces ice delamination was used in tandem with the FEM to determine the necessary cohesive zone properties. The pneumatic de-icing system installed on the wind tunnel experimental model was also used to validate the FEM predictions during freezer ice testing. This wind tunnel specimen was modeled in Abaqus and predictions were made to determine the delamination behavior of ice patches placed at different chordwise locations of the leading edge cap. Five locations were modeled and predictions indicated that 4 of the 5 locations would fully delaminate. Only the aft-most position would fail to completely delaminate. These predictions were validated with experimental results. Finally, the FEM was used in conjunction with the Penn State AERTS Rotor Icing, Shedding, and Performance code to investigate the relationship between input pneumatic pressures vs. impact ice delamination using a pneumatic de-icing system. This code predicts natural ice shedding and was modified to determine ice delamination using a pneumatic de-icing system. The model was initially used to determine the critical ice adhesion area required to achieve ice shedding. A percentage of the ice shape becomes delaminated when the pneumatic de-icer is activated. The remaining area must then carry all of the shear loads to resist shedding. Once a critical ice adhesion area was reached, ice shedding was predicted. Based on a theoretical study for a generic rotor setup - 55.5 in. (1.41 m) radius rotor, 280 RPM, 16 in. (0.406 m) chord NACA 0012 - having a leading edge surface roughness of 26 +/- 0.9 Ra μin., the critical ice adhesion area was determined to be 61% of the original ice/leading edge interfacial area. The modeled icing environment had a liquid water concentration of 2 g/m^3, a median volume water droplet diameter of 25 μin, and a temperature of -14 ̊C. These modeled testing conditions were chosen to match the low centrifugal forces at the root of a full-scale rotor and the ice/leading edge interfacial properties gathered in the experimental results obtained in this research effort. Two pneumatic de- icing designs were created and the ice adhesion surface length was determined depending on input pneumatic pressure. The first design represented the NACA 0012 design used during wind tunnel testing. The second design removed two of the pneumatic diaphragms and reinforced the area behind the leading edge with solid aluminum for increased ballistic protection. The critical pneumatic pressure to induce ice shedding was determined to be 1.0 psi (6.89 x 10^3 Pa) for the design tested in this research and 1.1 psi (7.58 x 10^3 Pa) for the improved design suggested for future research. These pressures were well within the available pressure supplied from centrifugal pumping (3.7 psi, 2.58 x 10^-4 Pa).