Back to BlogWritten by Adam Kahraman
The boundary layer is a key element of a propeller's aerodynamic performance, but it can be a difficult concept to understand.
As a current of air moves past a propeller blade at high speed, the air molecules directly in contact with its surface have zero relative velocity. Within a thin region near the blade surface, the air velocity gradually increases until it approaches the velocity of the surrounding airflow. This region is known as the boundary layer.
This thin region affects aircraft skin friction, flow separation, lift, thrust and power consumption. To improve aerodynamic performance, aircraft designers sometimes add surface features such as dimples or vortex generators on wings. However, a feature that improves performance on one design may reduce it on another.
So what determines whether aerodynamic surface features help or hinder aircraft performance? And how can this inform better UAV design?
Researchers at Khulna University of Engineering & Technology performed an experiment with modified airfoils to explore this very question. They took the NACA-4415 airfoil and added hexagonal dimples to the surface to see if they could delay flow separation. And that’s exactly what they found - the dimples resulted in delayed flow separation and improved aerodynamic performance at higher angles of attack (source).
Figure 1: Constructed model airfoil with hexagonal dimples from Mustak et al. (source)
Inspired by this work, we wanted to explore a similar surface modification, but on another lift generating device - the propeller. Could a modified surface improve propeller performance?
In our experiment, two 3D printed propellers with the same base geometry were tested. One had a smooth surface while the other had raised features along the leading edge on the suction (top) side of the blade. A 3D render of these propellers with an illustration of the flow is shown in figure 2 below. The propellers were designed with a relatively high blade pitch to investigate performance under conditions where separation might occur.
Figure 2: 3D illustration of how the flow is expected to behave between the smooth propeller and the modified propeller (source: Tyto Robotics)
The tested hypothesis was that the raised features would energize the boundary layer, delay separation and help the modified propeller operate more efficiently. Our surprising results are described in section 7.
But first, to understand why such a design may prove beneficial, we need to look at how the boundary layer forms.
Table of Contents
- Introduction
- What is the Boundary Layer?
- Boundary Layer and Lift Generation
- How to Delay Boundary Layer Separation?
- Why Aren’t Aircraft Wings Covered in Dimples?
- How to Improve Boundary Layer Attachment on Propellers
- Test and Experiment Results
- Conclusion
What is the Boundary Layer?
The boundary layer is the thin region of air next to a surface where airflow velocity gradually increases from zero relative to the surface to the free-stream velocity. On a propeller blade, the air directly touching the surface has zero relative velocity due to the no-slip condition. Moving away from the blade, the air velocity gradually increases until it approaches that of the surrounding airflow.
The Reynolds number is a dimensionless quantity that compares the effects of inertial and viscous forces, and it is one factor that influences the behaviour of the boundary layer. Lower Reynolds numbers generally support laminar flow, while higher Reynolds numbers make the transition to turbulent flow more likely. Surface roughness, flow disturbances, geometry, and other factors can also affect when this transition occurs. The figure below shows the difference between laminar and turbulent boundary layers near the surface of an object.
Figure 3: Boundary layer formation with laminar and turbulent flows (source)
Boundary Layer and Lift Generation
An airfoil generates lift by creating a pressure difference around its surfaces and deflecting the airflow as it passes over the airfoil. The boundary layer is part of this process because it affects how closely the airflow follows the shape of the airfoil. As long as the boundary layer remains attached, the airflow can continue to follow the surface and support the pressure distribution that produces lift.
If the boundary layer can no longer stay attached, the airflow separates from the surface. This changes the pressure distribution around the airfoil which can limit lift and can lead to stall.
The position of the separation of the boundary layer from the airfoil is influenced by a number of factors including the angle of attack, the shape of the airfoil, Reynolds number and surface roughness. The angle of attack is the angle between the chord line of an airfoil and the direction of the oncoming airflow. A higher angle of attack can create a stronger adverse pressure gradient, making it more difficult for the airflow to remain attached to the surface. This can cause the boundary layer to separate and, as the separation becomes more extensive, eventually lead to stall. Figure 4 below shows an example of an airfoil in a wind tunnel at a low angle of attack compared to the same airfoil at a higher angle of attack, where it experiences stall.
Figure 4: The airfoil on the left has a low angle of attack, allowing the airflow to remain attached to the surface for longer. The airfoil on the right has a higher angle of attack, which causes the airflow to separate and the airfoil to stall (source)
How to Delay Boundary Layer Separation?
Have you ever wondered why a golf ball is covered in dimples? While it is true that laminar flow around an object produces less skin-friction drag, it also has less momentum near the surface and can separate early from a curved body like a golf ball. This early separation can create a large wake and increased pressure drag. For a golf ball, pressure drag occurs when the low pressure wake behind the ball creates a force that pulls it backward, thus slowing it down.
This is where the dimples come in. The dimples on a golf ball create small disturbances that cause the boundary layer to transition to turbulent flow. Turbulent mixing brings higher momentum air closer to the surface, helping the boundary layer remain attached for longer. This delays separation, reduces the size of the wake, and ultimately reduces the ball’s total drag. Figure 5 below illustrates the difference in the wake produced by a golf ball with and without dimples.
Figure 5: Aerodynamics of a golf ball with and without dimples (source)
There is, however, a tradeoff with this design. The turbulent boundary layer creates more skin-friction drag, but the smaller wake greatly reduces pressure drag. The result is lower overall drag, allowing the ball to travel farther than it would with a smooth surface.
So why aren’t wings covered in dimples if they can actually increase overall aerodynamic efficiency?
Why Aren’t Aircraft Wings Covered in Dimples?
Unlike golf balls, aircraft wings are streamlined and specifically designed to generate lift while keeping the airflow attached over much of their operating range. Therefore, adding more disturbances on the airfoil could actually do more harm than good. However, this isn’t always the case.
Vortex generators are small vanes attached to an aircraft’s wings that deliberately disturb the boundary layer by generating small vortices within the flow. Figure 6 below shows a wing of a small aircraft with vortex generators. These vortices mix higher momentum air into the slower moving air near the surface, helping the boundary layer resist separation. Because of the turbulence caused by the vortex generators, they also introduce additional drag.
Vortex generators can be worthwhile when improvements in lift, stall behaviour, or control effectiveness outweigh their drag penalty.
Figure 6: Micro vortex generators on a small airplane wing (source)
Some flight scenarios where vortex generators may improve performance include:
- Low speeds
- High angles of attack
- Takeoff
- Approach
- High-speed manoeuvring
How to Improve Boundary Layer Attachment on Propellers
Deliberately adding small disturbances to an otherwise smooth aerodynamic surface can sometimes improve performance by helping the airflow remain attached. This idea has been explored before by researchers at Khulna University of Engineering & Technology, who added hexagonal dimples to an airfoil to delay flow separation at high angle of attack (source).
Their results showed that deliberately disturbing the flow could improve aerodynamic performance. With the regular surface, the authors reported that flow separation occurred at an angle of attack of 12 degrees. With the hexagonal dimples, separation was delayed until 16 degrees. Figure 7 below shows the surface pressure distribution of the airfoils tested in the research paper. The dotted vertical line indicates the coefficient of pressure at the same location on each airfoil. The graph shows that the hexagonal dimpled airfoil has a more negative pressure coefficient on the suction side of the airfoil compared to the regular airfoil.
Figure 7: Surface pressure distribution of regular airfoil vs hexagonal dimpled airfoil at 12 and 16 degrees angle of attack (source)
Inspired by the findings of this paper, we wanted to explore a similar concept using drone propellers - could vortex generators contribute to improved performance?
To test this, we 3D-printed two 4” propellers with the same blade geometry. The only difference was that one propeller had tiny vortex generators along the leading edge, while the other had no surface modifications. Both were tested under the same conditions so their performance could be compared directly. Figure 8 below shows an image of the 3D printed propellers.
Our hypothesis was that the modified propeller would perform better than the unmodified propeller, as measured by overall efficiency. The reasoning followed that if small disturbances could lengthen and sustain the boundary layer and reduce the wake, then this would correspond to improved performance.
Figure 8: Our 3D-printed propellers with a smooth surface and with vortex generators (source: Tyto Robotics)
Test and Experiment Results
The propellers were mounted on the Series 1585 thrust stand in front of the 2x2 WindShaper to test their performance in three conditions: no wind, laminar flow and turbulent flow. Data was recorded using the Flight Stand Software.
The software’s automatic control feature was used to create a throttle step test from 1,000 μs to 1,750 μs, with seven throttle steps in total. The highest step corresponded to a maximum speed of approximately 12,000 RPM. This provided a practical operating range for testing a 4” propeller.
The WindShaper was kept at a constant setting throughout the test, with the wind speed averaging 8 m/s (28.8 km/h) for both the laminar and turbulent flow conditions. Thrust, torque, rotation speed, and overall thrust efficiency were recorded throughout the test. Data was exported to .CSV for analysis.
Boundary layer transition and separation were not measured directly, so some of the explanations below remain hypotheses.
Thrust
First, the smooth propeller produced more thrust throughout most of the RPM range tested. At approximately 11,700 RPM, the modified propeller produced 5.8% to 7.4% less thrust than the smooth propeller across the three conditions.
Figure 9 shows the thrust produced by the smooth and modified propellers under each wind condition. Under no wind conditions, the two propellers produced similar thrust at lower rotation speeds, but the difference increased as RPM increased, with the smooth propeller producing more thrust at the higher speeds. A similar trend was observed under laminar flow, although the difference between the two propellers remained smaller overall. Under turbulent flow, the difference was more pronounced across most of the tested RPM range, with the smooth propeller consistently producing more thrust than the modified propeller.
The similarity between the no wind and laminar flow results may be related to the relatively steady conditions experienced by both propellers. In both cases, the aerodynamic loading on the blades would remain relatively consistent, whereas turbulent flow introduces fluctuations in the incoming airspeed and direction.
Figure 9: Thrust produced by the smooth propeller and modified propeller under each wind condition (source: Tyto Robotics)
Torque
Figure 10 shows the torque required by the smooth and modified propellers under each wind condition. Under no wind conditions, the torque required by the two propellers was nearly identical across most of the tested RPM range.
Under laminar flow, the modified propeller required increasingly more torque as rotation speed increased, producing the largest difference between the two propellers. At 11,700 RPM, the modified propeller required 3.7% more torque with no wind, 11.2% more torque under laminar flow, and 7.9% more torque under turbulent flow compared to the smooth propeller.
The larger torque required by the modified propeller may be due to the increased drag created by the vortex generators, requiring more torque to maintain the same rotation speed.
Figure 10: Torque required by the smooth and modified propellers under each wind condition (source: Tyto Robotics)
Efficiency
Figure 11 shows the overall efficiency of both propellers, defined as the grams-force of thrust generated per Watt of power supplied to the system, under each wind condition. Measurements below 2,000 RPM are not shown so that differences at higher speeds are easier to see. These measurements remain in the dataset.
Under no wind conditions, the efficiency of the two propellers remained relatively close throughout most of the operating range. At around 11,700 RPM, the modified propeller was 9.2% less efficient than the smooth propeller. The larger differences seen at the lowest rotation speeds should be interpreted cautiously because both thrust and power are close to zero in this region, making the calculated efficiency highly sensitive to small measurement variations.
A similar trend was observed under laminar flow. Once the propellers reached the normal operating range, the efficiency curves remained relatively close, although the modified propeller became less efficient at higher rotation speeds. At around 11,700 RPM, the modified propeller was 11.1% less efficient than the smooth propeller.
Under turbulent flow, the difference between the two propellers was more pronounced across the tested range. The modified propeller showed lower efficiency, particularly at lower rotation speeds, although the difference decreased as RPM increased. At around 11,700 RPM, the modified propeller was 14.9% less efficient than the smooth propeller, representing the largest efficiency difference between the two propellers at this rotation speed.
Figure 11: Overall efficiency of the smooth and modified propellers under each wind condition with measurements below 2,000 RPM not shown in the graph (source: Tyto Robotics)
The efficiency values change sharply at the lowest RPM points because both thrust and power are close to zero. The smooth and modified propeller curves begin to converge at around 4,500 RPM under turbulent flow conditions. From approximately 6,500 to 12,000 RPM, the efficiency remains relatively stable for both propellers in all flow conditions, with the unmodified propeller having a slightly higher efficiency throughout the range. Above 10,000 RPM, the efficiency of both propellers begins to decrease slightly.
The efficiency results help explain why the torque trends do not directly match the overall performance trends. Higher torque generally means the motor has to work harder and use more power to maintain the same RPM, but efficiency also depends on how much useful thrust is produced from that power. Under laminar flow, the modified propeller required more torque, but its thrust remained relatively close to the smooth propeller, so the overall efficiency stayed similar. Under turbulent flow, the torque difference was smaller, but the modified propeller lost more thrust, meaning it produced less useful output for the power being used. This is why the largest efficiency difference appeared under turbulent flow rather than under laminar flow.
The results did not support the original hypothesis, as the modified propeller did not outperform the smooth propeller under any of the tested conditions. At around 11,700 RPM, the modified propeller produced 5.8% to 7.4% less thrust while requiring 3.7% to 11.2% more torque, depending on the wind condition. Its overall efficiency was also 9.2% lower with no wind, 11.1% lower in laminar flow, and 14.9% lower in turbulent flow.
Why did the modified propeller perform worse?
The results show that the effect of the vortex generators depended strongly on the incoming wind condition. The modified propeller generally produced less thrust, while the increase in required torque was most noticeable under laminar flow. Despite these differences, the two propellers showed nearly identical overall efficiency under no wind and laminar conditions throughout the meaningful operating range, while a clearer reduction in efficiency was observed for the modified propeller under turbulent flow. This may be explained by several factors.
Unsuitable position of vortex generators
The improved performance observed in the paper was attributed to the hexagonal dimples delaying boundary layer separation and reducing wake formation at higher angles of attack. In our case, the modified propeller may not have achieved the same effect because of the position of the vortex generators. When a blunt feature is introduced too close to the leading edge, it can disrupt the flow and create wake turbulence, causing early separation and additional drag rather than improved performance.
The vortex generators were too large
When looking at airplane wings, vortex generators are relatively small compared to the overall size of the wing. They are typically only a fraction of an inch tall and a few inches long. For example, the Boeing 767 uses seven vortex generators on each wing, with a height of 19.05 mm, which is about 0.316% of its 6.03 m aerodynamic chord (source). The vortex generators on the modified 4” propeller were 1 mm high, which is about 5% of the blade chord. Although small, these features were relatively large compared with the blade and may have disrupted the airflow, increased drag, and reduced efficiency.
Different airspeed along the length of the propeller
A fixed airfoil in front of a wind wall experiences a nearly constant flow of air along the length of the airfoil. Since a propeller rotates around its hub, the tip travels at a higher airspeed compared to the propeller’s blade closer to the hub. Another factor is that the local blade angle changes from root to tip of the propeller. The size and position of the vortex generators must account for this variation. In our modified propeller design, however, the vortex generators were placed uniformly along the leading edge, with no variation in their height, length, or distance from the leading edge.
Centrifugal force and coriolis effects
The centrifugal force can drive air outward along the rotating blade toward its tip. The Coriolis effect is the sideways deflection of moving air when viewed from the rotating blade. It occurs because the blade rotates while the air moves relative to it. In some cases, this can improve boundary layer attachment and help prevent stall even without vortex generators. However, when vortex generators are added, they can interfere with the attachment already caused by the Coriolis effect and disrupt the flow instead of improving it.
How to Improve the Next Test
In a second iteration of this test, the design of the modified propeller could be improved by determining where the boundary layer separation occurs. Once the separation point is known, the following changes could be made:
- Place the vortex generator just upstream of the separation point.
- Limit the vortex generators to blade regions where separation occurs, to avoid added drag where flow already stays attached.
Other design changes could be tested independently:
- Testing at different angles of attack
- Different vane angles with counter-rotating pairs
- Testing with larger propellers
- Using recessed dimples instead of protruding vanes
The height of the vortex generators should be determined by the relative size of the propeller’s length and chord. Because they are 3D printed, multiple samples should also be tested to account for 3D printing defects and variations.
Conclusion
The boundary layer is an important phenomenon that explains how air interacts with surfaces like propellers and airfoils. Changing how that thin layer interacts with the surface can improve performance, but each surface feature also creates its own drag. This becomes a delicate balancing act of changing the surface just enough to improve performance without hindering the airflow enough to create additional drag.
In this study we aimed to determine whether vortex generators could improve propeller performance. The modified propeller generally produced less thrust and required more torque at comparable RPM. These results are consistent with an aerodynamic penalty from the added features, but the test did not directly measure their effect on boundary layer separation.
While this test demonstrates the challenges of modifying boundary layer behaviour on a propeller, it also highlights the potential for carefully designed surface features to improve aerodynamic performance.
Leave a comment