Aircraft health and usage monitoring system for in-flight strain measurement of a wing structure

FBG sensors were selected for the in-flight strain measurement of aircraft structures due to their superior performance characteristics in practical applications. FBG sensors hold numerous advantages over the conventionally used strain gauges, including their small size, light weight, embeddability, easy multiplexing, corrosion resistance, and non-EMI. Thanks to such advantageous characteristics of FBG sensors, they are suited for the multiplexed measurement of aircraft structures when operated under extreme environments, over long time periods, and with various on-board electrical equipment.

FBG sensors consist of regular interval gratings on the fiber core, which acts as an optical mirror reflecting a certain wavelength of light. When broadband light enters the optical fiber, a particular wavelength that satisfies certain Bragg conditions undergoes multiple Fresnel reflections due to the variation in the refractive index of the Bragg gratings. The specific wavelength that is reflected is described by

$$\begin{eqnarray}&&{\lambda }_{B}=2{n}_{e}\Lambda ,\end{eqnarray} \tag{ 1 }$$

where λ_B is the Bragg wavelength, n_e is the effective refractive index of the grating, and Λ is the grating period produced on the optical fiber. When an external load is applied to the Bragg grating, its effective refractive index and/or grating period change, leading to a shift in the Bragg wavelength of the reflected light. This shift in terms of strain and temperature variation is

$$\begin{eqnarray}&&{\rm{\Delta }}{\lambda }_{B}={\lambda }_{B}[(1-{p}_{e})\varepsilon +(\alpha +\xi ){\rm{\Delta }}T],\end{eqnarray} \tag{ 2 }$$

where p_e is the photoelastic constant, α is the thermal expansion coefficient, and ξ is the thermo-optic coefficient of the optical fiber. FBG sensors can detect strain and temperature variations in the Bragg grating by measuring the center wavelength shift of the reflected light from each FBG sensor.

In this study, a low-speed FBG interrogator (SFI-700, Fiberpro Inc., Republic of Korea) was utilized by taking into consideration the long-term monitoring of aircraft structures. The interrogator is composed of vertical-cavity surface-emitting lasers (VCSEL) in an array as the broadband light source. The cost of VCSEL is relatively low compared to other broadband light sources, making it possible for the interrogator to be relatively low cost. As the interrogator is a commercialized product for general purposes, it was modified to fit for aircraft. To directly receive electric power from the aircraft power supply system, a dc–dc converter was applied. Furthermore, vibration isolating materials were added in advance to the inner side of the device to prevent possible malfunction caused by high levels of vibration from the propeller during flights. The interrogator has a total of two channels, and each channel can measure up to 12 points with a sampling rate of up to 10 Hz. Its measurement range is 1510 ∼ 1550 nm and repeatability is less than 10 pm. The interrogator has a size of 420 (length) × 210 (width) × 85 (height) mm³ and weight of 4.5 kg, which allows for the device to be mounted into small aircraft. The application of the affordable interrogator allows for the simplification of the overall system and results in a more cost-effective HUMS for aircraft.

Figure 1 shows a conceptual diagram of the FBG-based HUMS. The HUMS is divided into two major parts, an airborne system and ground system. The airborne system consists of a sensing module and a data acquisition/processing/storage (DAPS) module. In the airborne system, multiple FBG sensors are installed on aircraft structures and connected to an on-board FBG interrogator to detect their wavelength shift during flight. The detected wavelength data are transferred to the DAPS module. Additionally, various flight parameters such as altitude, roll, pitch, and yaw angles, which are recorded from the altitude and heading reference system, are digitalized and transferred simultaneously to the DAPS module, in which all of the transferred data are stored on a compact flash memory card. After each flight, all of the stored data are transferred to a ground system via the memory card. In the ground system, the validity and relevance of the raw data are checked before they are uploaded to the main database server using a portable ground support equipment (GSE) device. The raw data are also converted into strain data through an internal algorithm of the portable GSE device. Finally, all the data are uploaded to a ground database server. The uploaded data are configured so that authorized users can directly connect to the server and access the flight test data online. This system configuration is convenient for various users.

Zoom In Zoom Out Reset image size

The JABIRU UL-D aircraft (JABIRU Aircraft Pty. Ltd, Australia; see figure 2) was selected as the testbed aircraft for the HUMS implementation and subsequent flight tests. The aircraft was equipped with two seats, and the maximum take-off weight was limited to 450 kg. According to its specifications, the JABIRU UL-D aircraft is categorized as an ultralight aircraft by the UK Civil Aviation Authority. It has a rectangular wing with a strut-braced wing configuration. The fuselage, wings, and struts were hinged. The aircraft was composed of a pair of main wings, horizontal stabilizers and one vertical stabilizer.

Zoom In Zoom Out Reset image size

The main wing was considered as the target structure for the HUMS implementation. This is a major part of the aircraft and supports most of the lift force during flight. In this study, FBG sensors were installed on the upper and lower flange of the main spar to measure the strain response mainly caused by the bending moment on the main wing. Sensor installation was conducted during the wing manufacturing process at the JABIRU Pty. manufacturing plant. The main wing of the testbed aircraft was composed of an upper and lower skin and a single main spar made of glass fiber epoxy composite. Additionally, a Styrofoam core was filled between each component. Strain and temperature FBG sensors were prepared in an array because their responses are affected simultaneously by strain and temperature variation. Figure 3 shows the details of the sensor installations for the strain and temperature sensors. All of the sensors were installed on the inner side of the main spar channel section. One strain sensing array consisted of 10 FBG sensors, with five strain sensors attached to the inner surface of the upper flange and the remaining five to the inner surface of the lower flange on each wing using an epoxy adhesive. One temperature sensing array consisted of two FBG sensors, and the temperature sensors were installed in a strain-isolated state along the neutral axis of the spar web to minimize the effect of the bending moment on them. One strain sensing array and one temperature sensing array were directly connected with each other to simultaneously measure the strain and temperature through a single optical fiber line. A total of 12 FBG sensors on a single optical fiber line per wing were connected to each channel of the on-board FBG interrogator for flight tests. In the case of sensor damage that might occur during the wing manufacturing, assembly, transport, and long-term operation after sensor installation, two strain sensing arrays and one temperature sensing array were additionally installed on each wing for redundancy. Thus, a total of five FBG sensor arrays, which consisted of 30 strain sensors and four temperature sensors, were installed on each wing. The strain sensing arrays were laid side-by-side with the same strain sensing locations on the spar flanges. The distance between each strain sensing array in the chord direction was about 5 mm. The temperature sensors were installed in different locations along the spanwise direction. As the optical fiber lines are very thin and brittle, there exists a large risk of damage. To protect the optical fiber lines from possible external damage, preventive methods were considered during sensor installation in the following manner. There is a weak point between the buffer and the non-buffer coating areas, where stress concentrations exist due to abrupt changes in the cross-section diameter. These stress concentrations can lead to damage in the pertinent optical fiber line segments during sensor installation or wing manufacturing and assembly. Therefore, possible damage to these segments was prevented in advance by using epoxy adhesive to completely fix these areas to the neutral axis of the spar web near the wing root. Additionally, all segments of the FBG sensor arrays were fixed to the upper and lower flanges using epoxy adhesive to prevent any breakage of the optical fiber lines. All of the FBG sensor arrays were checked using the low-speed FBG interrogator and showed normal output signals after sensor installation.

Zoom In Zoom Out Reset image size

A ground test was conducted to confirm the integrity and availability of the installed sensors before installing the main wings to the testbed aircraft fuselage. Figure 4 shows the experimental setup for the ground test to examine the status of the output signals after the final wing manufacturing process. The output signals from all of the FBG sensor arrays were examined using the low-speed FBG interrogator. From the ground test, all of the sensor arrays installed in both wings functioned, but several problems arose in a few sensors on the strain sensing arrays after the wing manufacturing process. These problems included multiple center wavelength shifts being detected from a single FBG sensor, flickering of the detected wavelength signal, poor repeatability, and the inability to measure the center wavelength shift for certain FBG sensors. Figure 5 shows the wavelength spectrum measured from one of the problematic strain sensing arrays. Some of the FBG sensors showed optical spectrum splitting and/or bandwidth widening of the reflected Bragg wavelength peak. These phenomena were caused by the physical interaction between the FBG sensors, the adhesive, and the inner foam core, whose installation process is shown in figure 6. The inside of the wing structure was filled with the foam core to prevent buckling during flight. In this process, there was direct contact between the FBG sensors and inner foam core as well as non-uniform adhesive bonding during hardening due to the work being performed manually. The problematic sensors could not be corrected physically after completion of the wing manufacturing because all of the FBG sensors were already embedded into the wing structure. Therefore, to obtain reliable output data from the FBG sensors, such problems have to be taken into consideration before the sensor implementation. For example, packaging the FBG sensors in protective films or fabricating very thin extruded/intruded grooves at the sensing areas can prevent the direct contact between FBG sensors and the other components, which would be helpful for implementation on real wing structures consisting of an inner foam core. Among the three strain sensing arrays on each wing, only one of 10 FBG sensors showed normal output signals with a high level of the signal repeatability below 10 pm during the ground test. The other two strain sensing arrays on each wing included a few problematic FBG sensors. As a result, the normal strain sensing array on each wing, which did not include any problematic FBG sensors, was selected for the in-flight strain measurement. After the ground test, the main wings were assembled to the testbed aircraft fuselage. Additionally, the FBG interrogator and the DAPS module were loaded into the cockpit and connected to the selected FBG sensor arrays.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

For the strut-braced wing structure, it was known that the largest bending moment occurs near the strut-wing joint section, and the smallest bending moment occurs near the wing root and wing tip because there is no reaction force against the bending moment at those regions due to the hinged joint boundary condition [20]. Figure 7 shows the strain responses of 10 FBG sensors on the left wing, upper and lower flanges as well as the strain distribution of the wing structure along the spanwise direction at around 2000 s. In the case of the upper flange, there were large compressive loads on the strut-braced section 'b' and FBG8 showed maximum strain level during the flight. In the case of the lower flange, there were low strain levels on the strut-braced section 'b', close to zero, and FBG4 showed maximum strain level during the flight. Such strain distribution characteristics were caused by its wing configuration and boundary conditions. The simple beam section 'a' experienced only the bending moment due to the distributed lift force; however, the strut-braced section 'b' experienced the complex loading conditions of not only the bending moment but also the longitudinal compressive load due to the lift force on the section 'a', as shown in figure 8. Therefore, the upper flanges of section 'b' bear much more load than the lower flanges for the strut-braced wing structure due to the longitudinal compressive load. The strain measurement results revealed that the latter had a significant impact on the strain distribution of the strut-braced wing configuration, which was quantitatively verified. The critical section of the strut-braced wing structure was confirmed to be where the high levels of flight load occurred.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Pilots control the aircraft attitude by adjusting the flight control surfaces of the aileron, rudder, and elevator, which influence the loading conditions of the main wings by altering the static equilibrium state of the aircraft. Therefore, the effect of the adjusted control surfaces on the loading conditions of the main spars was investigated through the in-flight strain measurement results.

Figure 9 shows the strain response of FBG9 from both wings and corresponding roll angle during the turning maneuvers. The steep turning maneuvers of the aircraft caused the strain variations, as shown in the figure. The strain variations in both wings showed opposite trends during turning. Such contrasting trends in the strain are caused by the different lift conditions of each wing because the ailerons are adjusted in opposite directions to turn the aircraft in a desired direction. For example, for a left turn, the aileron of the left wing is adjusted to the up position, decreasing the lift force due to a shorter camber length and a smaller angle of attack (AOA). On the other hand, the aileron of the right wing is adjusted to the down position, increasing the lift force due to a longer camber length and a larger AOA, as shown in figure 9.

Zoom In Zoom Out Reset image size

Figure 10 shows the strain responses of FBG9 and corresponding pitch angle during the climbs and descents. The strain variation showed high correlations with the pitch angle of the aircraft. The elevators of both horizontal stabilizers were adjusted upward or downward to make the aircraft pitch-up or pitch-down, respectively. For example, to make the pitch-up motion of the aircraft, the elevators are adjusted upward, which decreases the lift force generated on the horizontal stabilizers due to a shorter camber length and a smaller AOA, as shown in the figure. The decrease in the lift force on the horizontal stabilizers alters the moment equilibrium state of the aircraft about its center of gravity, thus causing the pitch-up motion. This increases the lift force generated on both main wings due to a larger AOA resulting in an increased strain response. Conversely, the pitch-down motion decreases the lift force on the main wings and decreases the strain response.

Zoom In Zoom Out Reset image size

The in-flight strain data showed that steep maneuvers of the aircraft caused drastic changes in the loading conditions of the main wings, and resulted in significant strain variations. From the comparison results between the in-flight strain response and corresponding flight parameters, the in-flight strain data obtained from the HUMS qualitatively corresponded with the loading conditions of the main wings according to the aircraft maneuvers.

Aircrafts use flaps during take-off and landing for more efficient flight. Using flaps significantly increases the lift coefficient by changing the camber length and the AOA of the wing airfoil so that sufficient lift force for flight is obtained at a relatively low aircraft speed. This use of flaps changes the lift characteristics of the wing airfoil and influences the loading conditions of the main spars. The testbed aircraft had simple-hinged fixed slotted flaps on the rear portion of the main wings and controlled the flaps over two stages from 15.5° to 40° for take-off and landing. Therefore, the effect of the flaps on the loading conditions of the main spars was investigated through the in-flight strain measurement results.

Figure 11(a) shows the strain response of FBG9 located on the right wing spar upper flange and corresponding pitch angle during the take-off procedure. The testbed aircraft entered the runway with the first stage deflected flaps for take-off at a lower speed with a higher lift coefficient. As the aircraft speed increased on the runway, the strain response progressively increased due to the generation of the lift force on the main wings at interval (a) in the figure. When the aircraft reached its take-off speed, the aircraft was taking off and the strain response stepwise increased due to the increasing AOA at interval (b). After the aircraft reached a certain altitude with adequate speed for stable flight, the first stage deflected flaps were raised to the up position for more efficient flight by reducing the drag force on the wing airfoil. Raising the flaps to the up position resulted in a stepwise increase in the strain response, as shown in the figure at interval (c).

Zoom In Zoom Out Reset image size

Figure 11(b) shows the strain response of FBG9 located in the right wing spar upper flange and corresponding pitch angle during the landing procedure. The testbed aircraft used the flaps in two stages to obtain lower stall speed and a lower approach speed of the aircraft as well as a higher lift coefficient for a safe landing. The aircraft approached the runway with the flaps in the up position. During descent, there existed a stepwise decrease in the strain response at intervals (a) and (b) in the figure. These sudden decreases in the strain response were caused by adjustment of the flaps from the up position to the first and second stages, respectively. This decreasing trend in the strain response was found to be more significant as the flap deflection angle increased. After the flaps were adjusted to the second stages, the aircraft began the final approach to the runway. When the aircraft touched the ground, there was a stepwise decrease in the strain response due to the decreasing AOA at interval (c). After the aircraft landed, the strain response progressively decreased due to the loss of the lift force on the main wings at interval (d).

The in-flight strain measurement results revealed that the use of flaps caused stepwise decreases in the strain response of the spar flanges. This was caused by the change in the aerodynamic characteristics of the wing airfoil. Wenzinger and Delano [21] investigated the pressure distribution of the wing airfoil through wind tunnel tests by employing slotted flaps. From these tests, it was found that there was a significant increase in the pressure on the flap surface near the trailing edge when the flap was deflected at certain angles and the center of pressure (CP) location gradually moved to the rear. Additionally, it is expected that the direction of the resultant lift force applied to the CP will move to the rear as the flap angle increases. Thus, the use of such high lift devices does not necessarily lead to an increase in the bending moment directly applied to the main spars causing the decrease in the strain response. From the comparison results between the in-flight strain response and corresponding pitch angle, the in-flight strain data from the HUMS showed reasonable responses for the corresponding maneuvers of the testbed aircraft during the take-off and landing procedures.

Stall recovery tests were conducted to simulate an abnormal state of the aircraft. Stall is defined as a sudden reduction in the lift force on the main wings that results in the loss of aircraft control. The effect of aircraft stall on the loading conditions of the main wings was investigated. A power-on stall test was conducted to simulate a stall that can occur during the take-off procedure caused by a large AOA which exceeds the critical value. The testbed aircraft was equipped with a stall warning system that consisted of a stall warning funnel, tubes, and a horn. The stall tests were conducted until the stall warning horn was activated, and then recovery maneuvers were taken immediately afterwards.

Figure 12 shows the strain response of FBG9 located in the right wing spar upper flange, corresponding altitude, and pitch angle during the power-on stall test. During climbing, the engine power was set to be 2500 rpm to gradually reduce the aircraft speed. When the aircraft speed became less than 1 kt, the pilot simultaneously pulled back the yoke and increased engine power to simulate the departure stalls. During the early stages of the stall at interval (a) in the figure, there was a temporary rise in strain due to the increasing AOA. Subsequently, the aircraft nose was continuously raised, and once the nose reached over a critical AOA, the back pressure and airflow separation on the upper airfoil became more severe, which caused the aircraft to enter a stall and gradually lose lift force. This significantly decreased the bending moments on the main wings and consequently led to a rapid reduction in the strain response and pitch-down motion of the aircraft, as shown in the figure at interval (b). After the occurrence of the stalls, the recovery maneuvers were made immediately. As a result, the lift forces on the main wings were fully recovered, and the reduced strain levels immediately increased to regular levels, as shown in the figure at interval (c).

Zoom In Zoom Out Reset image size

From the results, the aircraft stalls affected the loading condition of the wing structure causing the significant decrease in the strain response due to the loss of the lift force on the main wings. The in-flight strain monitoring of the spar flanges was able to sensitively respond to the stalling of the main wings and showed reasonable responses for the corresponding maneuvers of the testbed aircraft during the stall recovery test.