- Email: [email protected]
Structural health monitoring: composite skins are getting a nervous system
Reinforced Plastics Volume 59, Number 3 May/June 2015
www.reinforcedplastics.com
FEATURE
Structural health monitoring: composite skins are getting a nervous system Django Mathijsen Django Mathijsen discusses structural health monitoring (SHM) and how it is entering real world applications. Imagine not needing testing or measurement equipment to assess how much wear and damage a composite structure has sustained. Just plug your laptop into the component and its built-in sensors tell you if it needs replacing. That’s what structural health monitoring is promising for the future. But it takes a lot of research to make it work. Structural health monitoring (SHM) is integrating sensors into a composite component so you can real time measure if it is still functional and safe. The concept of SHM was first applied in civil engineering: attaching sensors to steel and concrete structures like bridges and buildings, especially in earthquake sensitive areas. The field is shifting to composites, and applications like aerospace and wind turbines. But the technology cannot easily be transferred. In composites the damages are of a different order. And the structures tend to be more complex, as well as the damage: it can be internal and is influenced by fiber orientations. Still, the technology is close to leaving the laboratory and entering real world applications.
Fast and effective damage analysis ‘‘You can measure anything that will give you information about the performance of a structure,’’ says Richard Loendersloot of the University of Twente about SHM. ‘‘Temperature in bearings for example. But we are especially looking at dynamic behavior, measured for example with strain gauges.’’ The university is participating in the project Wibrate (http:// wibrate.eu/), integrating SHM in a helicopter rotor and in milling machines for the automotive industry. They are cooperating with NLR (the Dutch National Aerospace Laboratory) to develop SHM for fixed wing aircraft. And they are looking at implementing SHM in composite aircraft panels with Airbus in the project Saristu (http://www.saristu.eu/). ‘‘The aim is to get to a higher technology readiness level,’’ Loendersloot says. ‘‘So we are initially limiting the level of SHM. For example I am investigating a monitoring system for
the composite parts around a door surround in an aircraft (together with the partners of the Saristu project). There is a lot of traffic around the doors when an aircraft is stationary. So the impact risk is high. If an impact happens, you need fast and effective damage analysis.’’
Detecting invisible damage Composites are being used more in airplanes, like the Boeing 787 and Airbus A350. ‘‘Composites have many advantages, like low weight and high stiffness and strength, but a disadvantage is that you cannot see internal damage on the outside,’’ says Ted Ooijevaar who earned his PhD at the University of Twente for an SHM system to monitor composite skin-stiffener structures, a common construction in the aircraft industry. Aircraft maintenance crews use the ‘‘Barely Visible Impact Damage’’ criterion: if you can’t see any damage under typical lighting conditions from a distance of five feet, the structure is still good to fly. Fine for metal structures, but not for composites. ‘‘Internal damage, invisible from the outside, can have significant consequences,’’ says Ooijevaar. ‘‘Delamination for example between a stiffener and the skin. So one reason for developing SHM is to be able to guarantee the integrity of the structure. A second reason is: if you exactly know when a part will fail, you can go from maintenance at fixed periods to maintenance as needed. My research was financed by the European project Cleansky (http://www.cleansky.eu/), which focuses on the development of breakthrough technologies to improve the environmental performance of airplanes. That’s the third reason: if I can measure the behavior of the structure, I may be able to lengthen its working life.’’
Embedded sensors Maintenance crews need specialist knowhow and equipment (like ultrasonic scanners) for assessing internal damage. SHM means
0034-3617/ß 2015 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.repl.2015.02.013
139
FEATURE
Reinforced Plastics Volume 59, Number 3 May/June 2015
FEATURE
them for quality control during and after production,’’ Ooijevaar adds. Don’t the embedded sensors weaken the part? ‘‘Not if you do it right,’’ Loendersloot answers. ‘‘In composites it’s a matter of not disturbing the intended path of the fibers. If the sensor pushes fibers aside, especially if they are under an angle, you have a problem.’’ Critics have other qualms about integrated sensors. They increase the cost of a component. And if a sensor breaks, should you replace the whole component? ‘‘No manufacturer wants an aircraft with sensors everywhere,’’ says Ooijevaar. ‘‘So I think SHM will mainly be used for hot spot monitoring.’’ From the design you know where the part is likely to fail critically first: that’s where you want sensors. Chiefly piezoelectric elements are used for the strain gauges. They require wires coming out of the component: creating a robust connection is a challenge. ‘‘A wire can break, especially if it is long,’’ Loendersloot says. ‘‘You might prefer to process the data in a small processor near the sensors, and maybe send the results wireless to a central unit. We are working with such a wireless node in the Wibrate project. The processor can also accept commands from the central unit. Wire sensors on the other hand give you more freedom with algorithms. So SHM requires an integrated design process.’’ They are also looking at using SHM in wind turbine blades to prevent blade breakage. There, wireless data transfer from the rotating to the static part has big advantages. Going wireless offers its own challenges, like the power supply: do you install a battery or can you harvest energy? Another way of reducing the number of wires is to use an optical fiber Bragg grating: a fiber that can contain multiple sensors. It has a local periodical variation in the refractive index of the fiber core, acting as a wavelength-specific mirror. The reflected frequency changes as the fiber is strained. These fibers offer other challenges. integrating sensors into the structure so you only have to plug in a laptop, which will run a measurement cycle and tell you if the structure is healthy. Integrating the sensors instead of sticking them on increases their chemical and mechanical resistance. ‘‘And you can use
Skin-stiffener structures, a common construction in the aircraft industry. 140
A skin-stiffener damage index distribution.
Reinforced Plastics Volume 59, Number 3 May/June 2015
FEATURE
Vibration based
y−coordinate [m]
1 0.2
0.5 0
0.1 −0 .5 0
0
0.1
0.2
0.3
−1 FEATURE
x−coordinate [m] Normalized mode shape curvature [ ] in x-direction of the pristine structure.
1
y−coordinate [m]
Ooijevaar’s SHM research was vibration based: ‘‘A structure vibrates differently if it has damage. You can look at dynamic properties, like Eigenfrequencies, mode shapes, derivatives of mode shapes, damping. . . You have to look at the potential damage locations. And then you have to understand the effect of damage on the dynamic behavior. In a skin with stiffeners, the most critical part is the skin-stiffener interface.’’ Structures are developed to have some damage tolerance before safety and performance are affected; production can already cause miniscule cracks or air inclusions. A composite will often fail because of delamination, which usually starts with micro cracks you don’t have to detect because they hardly affect the structural integrity. The measurement technique should warn you when the cracks have grown to the point where the part is close to failing. So the method must be sensitive enough, but not so sensitive that it identifies noise as damage, generating false positives. In dynamic methods the sensitivity is dependent on the frequency: methods that use high frequencies, like ultrasonic scanning, will detect very small damage because of their short wavelength. ‘‘I looked at the effect of damage only in the skin, between the stiffener and the skin, and at the end of the stiffener,’’ Ooijevaar says. ‘‘I used a linear dynamic method, based on the change of the curvature of the vibration, and a nonlinear dynamic method: a crack can open and close if it vibrates.’’ In Ooijevaar’s tests, a shaker was used. But you can also use operational vibrations from the drive system and/or environment. It is even possible to use the piezoelectric sensors as actuators. You can run a measurement cycle by using each sensor in turn as the actuator, while the others act as sensors. You can then look at the frequency content and attenuation of the signal. Damage is a discontinuity: if you send a wave through the material, it will be altered by the discontinuity (reflected, attenuated, dispersed, etc.). Ooijevaar tested the undamaged panel to get a baseline. The panel was damaged by dropping a weight onto it, simulating a bird strike or tool drop. Then the damaged panel was tested. The data were compared to try and detect, localize and characterize the damage. That’s a difficult problem, as Loendersloot explains: ‘‘It’s not too difficult to predict the effect of certain damage. But if you measure an effect, it’s much more difficult to identify the damage, because then you’re trying to solve an inverse problem. A different type of damage in a different location can often produce the same effect. So you have to dig deeper: do more math or physics to really understand the problem. Maybe if you measure in one frequency you cannot tell two forms of damage apart because of symmetry, so you have to measure with a second frequency as well. Usually it is even more complicated.’’ Ooijevaar discovered that a method based on the curvature of the mode shape is very effective for monitoring damage in the stiffener-skin interface. But it does not work for detecting damage in the skin between the stiffeners. ‘‘Especially in delamination you will see large discontinuities in the curvature,’’ says Ooijevaar. ‘‘Compare it to a guitar string. If it is locally damaged, the fundamental frequency will change and the string will vibrate in a slightly different shape. The derivative of that shape is more sensitive. You can also compare it to a bathroom tile. Many people will tap it and listen to check if it is loose. We are using the change in the shape in which it vibrates.’’
0.2
0.5 0
0.1 −0 .5 0
0
0.1
0.2
0.3
−1
x−coordinate [m] Normalized mode shape curvature [ ] in x-direction of the damaged structure.
Levels of SHM SHM methods can be classified in different levels: 1. Detecting the presence of damage. 2. Localizing the damage. 3. Characterizing the extent and severity of the damage. 4. Making a prognosis of the lifespan the component has left. Many SHM methods are now capable of level 1. Localizing and characterizing the damage is more difficult. So if the system detects damage, more investigation will still be required, with an ultrasonic scanner for example. ‘‘We have to go to more difficult structures and improve performance,’’ says Ooijevaar. ‘‘And different environments have to be taken into account: do I still understand my structure at 40 or +40 8C for example, or is the environment causing false positives?’’ To make SHM really effective, you would like it to reliably achieve level 4 and have it integrated into the design process of the structure. Then you can design a more critical, thus lighter, structure. It will take much more research. But the scale of the test pieces has gone up. The technique is at the stage that testing is starting on realistic components. ‘‘We found there is not a single method optimal for every structure,’’ Ooijevaar says. ‘‘There is a plethora of SHM techniques and, especially in composites, damage can be very complex. 141
FEATURE
Reinforced Plastics Volume 59, Number 3 May/June 2015
FEATURE Relation between CBM (Condition Based Maintenance), SHM and PHM (Prognostics and Health Management).
Every situation requires its own approach. People are always talking about Big Data. Conversely, we did not try to generate as much data as possible. Instead we tried to reduce the amount of data needed, by studying the physics of the damage and the material. If you understand that, you can choose the smartest way to measure and process the data, so it gives you exactly the information you need. In the end I think the solution will be to use a combination of methods: maybe different analytic methods, different sensor types, different locations. . . And maybe to use one method for detection, and another for characterizing. If all methods say there is damage, then you have a high reliability.’’
142
Just on the horizon SHM is coming, offering economic, ecologic and safety gains. But in order to really exploit it, you need to know exactly the hotspots, the failure behavior and the damage criterion of your structure. ‘‘Maybe the biggest challenge is that it requires a multidisciplinary approach,’’ says Ooijevaar. ‘‘The type and place of the sensors, data acquisition, failure mechanisms, wiring, energy supply to the sensors, where to process the data, etcetera. . .’’ It might still take a decade before SHM will be used in commercial aircraft, because of all the tests and certification in the aircraft industry. In wind turbines it should happen faster. There are fewer hot spots in a wind turbine blade.
www.reinforcedplastics.com
FEATURE
Structural health monitoring: composite skins are getting a nervous system Django Mathijsen Django Mathijsen discusses structural health monitoring (SHM) and how it is entering real world applications. Imagine not needing testing or measurement equipment to assess how much wear and damage a composite structure has sustained. Just plug your laptop into the component and its built-in sensors tell you if it needs replacing. That’s what structural health monitoring is promising for the future. But it takes a lot of research to make it work. Structural health monitoring (SHM) is integrating sensors into a composite component so you can real time measure if it is still functional and safe. The concept of SHM was first applied in civil engineering: attaching sensors to steel and concrete structures like bridges and buildings, especially in earthquake sensitive areas. The field is shifting to composites, and applications like aerospace and wind turbines. But the technology cannot easily be transferred. In composites the damages are of a different order. And the structures tend to be more complex, as well as the damage: it can be internal and is influenced by fiber orientations. Still, the technology is close to leaving the laboratory and entering real world applications.
Fast and effective damage analysis ‘‘You can measure anything that will give you information about the performance of a structure,’’ says Richard Loendersloot of the University of Twente about SHM. ‘‘Temperature in bearings for example. But we are especially looking at dynamic behavior, measured for example with strain gauges.’’ The university is participating in the project Wibrate (http:// wibrate.eu/), integrating SHM in a helicopter rotor and in milling machines for the automotive industry. They are cooperating with NLR (the Dutch National Aerospace Laboratory) to develop SHM for fixed wing aircraft. And they are looking at implementing SHM in composite aircraft panels with Airbus in the project Saristu (http://www.saristu.eu/). ‘‘The aim is to get to a higher technology readiness level,’’ Loendersloot says. ‘‘So we are initially limiting the level of SHM. For example I am investigating a monitoring system for
the composite parts around a door surround in an aircraft (together with the partners of the Saristu project). There is a lot of traffic around the doors when an aircraft is stationary. So the impact risk is high. If an impact happens, you need fast and effective damage analysis.’’
Detecting invisible damage Composites are being used more in airplanes, like the Boeing 787 and Airbus A350. ‘‘Composites have many advantages, like low weight and high stiffness and strength, but a disadvantage is that you cannot see internal damage on the outside,’’ says Ted Ooijevaar who earned his PhD at the University of Twente for an SHM system to monitor composite skin-stiffener structures, a common construction in the aircraft industry. Aircraft maintenance crews use the ‘‘Barely Visible Impact Damage’’ criterion: if you can’t see any damage under typical lighting conditions from a distance of five feet, the structure is still good to fly. Fine for metal structures, but not for composites. ‘‘Internal damage, invisible from the outside, can have significant consequences,’’ says Ooijevaar. ‘‘Delamination for example between a stiffener and the skin. So one reason for developing SHM is to be able to guarantee the integrity of the structure. A second reason is: if you exactly know when a part will fail, you can go from maintenance at fixed periods to maintenance as needed. My research was financed by the European project Cleansky (http://www.cleansky.eu/), which focuses on the development of breakthrough technologies to improve the environmental performance of airplanes. That’s the third reason: if I can measure the behavior of the structure, I may be able to lengthen its working life.’’
Embedded sensors Maintenance crews need specialist knowhow and equipment (like ultrasonic scanners) for assessing internal damage. SHM means
0034-3617/ß 2015 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.repl.2015.02.013
139
FEATURE
Reinforced Plastics Volume 59, Number 3 May/June 2015
FEATURE
them for quality control during and after production,’’ Ooijevaar adds. Don’t the embedded sensors weaken the part? ‘‘Not if you do it right,’’ Loendersloot answers. ‘‘In composites it’s a matter of not disturbing the intended path of the fibers. If the sensor pushes fibers aside, especially if they are under an angle, you have a problem.’’ Critics have other qualms about integrated sensors. They increase the cost of a component. And if a sensor breaks, should you replace the whole component? ‘‘No manufacturer wants an aircraft with sensors everywhere,’’ says Ooijevaar. ‘‘So I think SHM will mainly be used for hot spot monitoring.’’ From the design you know where the part is likely to fail critically first: that’s where you want sensors. Chiefly piezoelectric elements are used for the strain gauges. They require wires coming out of the component: creating a robust connection is a challenge. ‘‘A wire can break, especially if it is long,’’ Loendersloot says. ‘‘You might prefer to process the data in a small processor near the sensors, and maybe send the results wireless to a central unit. We are working with such a wireless node in the Wibrate project. The processor can also accept commands from the central unit. Wire sensors on the other hand give you more freedom with algorithms. So SHM requires an integrated design process.’’ They are also looking at using SHM in wind turbine blades to prevent blade breakage. There, wireless data transfer from the rotating to the static part has big advantages. Going wireless offers its own challenges, like the power supply: do you install a battery or can you harvest energy? Another way of reducing the number of wires is to use an optical fiber Bragg grating: a fiber that can contain multiple sensors. It has a local periodical variation in the refractive index of the fiber core, acting as a wavelength-specific mirror. The reflected frequency changes as the fiber is strained. These fibers offer other challenges. integrating sensors into the structure so you only have to plug in a laptop, which will run a measurement cycle and tell you if the structure is healthy. Integrating the sensors instead of sticking them on increases their chemical and mechanical resistance. ‘‘And you can use
Skin-stiffener structures, a common construction in the aircraft industry. 140
A skin-stiffener damage index distribution.
Reinforced Plastics Volume 59, Number 3 May/June 2015
FEATURE
Vibration based
y−coordinate [m]
1 0.2
0.5 0
0.1 −0 .5 0
0
0.1
0.2
0.3
−1 FEATURE
x−coordinate [m] Normalized mode shape curvature [ ] in x-direction of the pristine structure.
1
y−coordinate [m]
Ooijevaar’s SHM research was vibration based: ‘‘A structure vibrates differently if it has damage. You can look at dynamic properties, like Eigenfrequencies, mode shapes, derivatives of mode shapes, damping. . . You have to look at the potential damage locations. And then you have to understand the effect of damage on the dynamic behavior. In a skin with stiffeners, the most critical part is the skin-stiffener interface.’’ Structures are developed to have some damage tolerance before safety and performance are affected; production can already cause miniscule cracks or air inclusions. A composite will often fail because of delamination, which usually starts with micro cracks you don’t have to detect because they hardly affect the structural integrity. The measurement technique should warn you when the cracks have grown to the point where the part is close to failing. So the method must be sensitive enough, but not so sensitive that it identifies noise as damage, generating false positives. In dynamic methods the sensitivity is dependent on the frequency: methods that use high frequencies, like ultrasonic scanning, will detect very small damage because of their short wavelength. ‘‘I looked at the effect of damage only in the skin, between the stiffener and the skin, and at the end of the stiffener,’’ Ooijevaar says. ‘‘I used a linear dynamic method, based on the change of the curvature of the vibration, and a nonlinear dynamic method: a crack can open and close if it vibrates.’’ In Ooijevaar’s tests, a shaker was used. But you can also use operational vibrations from the drive system and/or environment. It is even possible to use the piezoelectric sensors as actuators. You can run a measurement cycle by using each sensor in turn as the actuator, while the others act as sensors. You can then look at the frequency content and attenuation of the signal. Damage is a discontinuity: if you send a wave through the material, it will be altered by the discontinuity (reflected, attenuated, dispersed, etc.). Ooijevaar tested the undamaged panel to get a baseline. The panel was damaged by dropping a weight onto it, simulating a bird strike or tool drop. Then the damaged panel was tested. The data were compared to try and detect, localize and characterize the damage. That’s a difficult problem, as Loendersloot explains: ‘‘It’s not too difficult to predict the effect of certain damage. But if you measure an effect, it’s much more difficult to identify the damage, because then you’re trying to solve an inverse problem. A different type of damage in a different location can often produce the same effect. So you have to dig deeper: do more math or physics to really understand the problem. Maybe if you measure in one frequency you cannot tell two forms of damage apart because of symmetry, so you have to measure with a second frequency as well. Usually it is even more complicated.’’ Ooijevaar discovered that a method based on the curvature of the mode shape is very effective for monitoring damage in the stiffener-skin interface. But it does not work for detecting damage in the skin between the stiffeners. ‘‘Especially in delamination you will see large discontinuities in the curvature,’’ says Ooijevaar. ‘‘Compare it to a guitar string. If it is locally damaged, the fundamental frequency will change and the string will vibrate in a slightly different shape. The derivative of that shape is more sensitive. You can also compare it to a bathroom tile. Many people will tap it and listen to check if it is loose. We are using the change in the shape in which it vibrates.’’
0.2
0.5 0
0.1 −0 .5 0
0
0.1
0.2
0.3
−1
x−coordinate [m] Normalized mode shape curvature [ ] in x-direction of the damaged structure.
Levels of SHM SHM methods can be classified in different levels: 1. Detecting the presence of damage. 2. Localizing the damage. 3. Characterizing the extent and severity of the damage. 4. Making a prognosis of the lifespan the component has left. Many SHM methods are now capable of level 1. Localizing and characterizing the damage is more difficult. So if the system detects damage, more investigation will still be required, with an ultrasonic scanner for example. ‘‘We have to go to more difficult structures and improve performance,’’ says Ooijevaar. ‘‘And different environments have to be taken into account: do I still understand my structure at 40 or +40 8C for example, or is the environment causing false positives?’’ To make SHM really effective, you would like it to reliably achieve level 4 and have it integrated into the design process of the structure. Then you can design a more critical, thus lighter, structure. It will take much more research. But the scale of the test pieces has gone up. The technique is at the stage that testing is starting on realistic components. ‘‘We found there is not a single method optimal for every structure,’’ Ooijevaar says. ‘‘There is a plethora of SHM techniques and, especially in composites, damage can be very complex. 141
FEATURE
Reinforced Plastics Volume 59, Number 3 May/June 2015
FEATURE Relation between CBM (Condition Based Maintenance), SHM and PHM (Prognostics and Health Management).
Every situation requires its own approach. People are always talking about Big Data. Conversely, we did not try to generate as much data as possible. Instead we tried to reduce the amount of data needed, by studying the physics of the damage and the material. If you understand that, you can choose the smartest way to measure and process the data, so it gives you exactly the information you need. In the end I think the solution will be to use a combination of methods: maybe different analytic methods, different sensor types, different locations. . . And maybe to use one method for detection, and another for characterizing. If all methods say there is damage, then you have a high reliability.’’
142
Just on the horizon SHM is coming, offering economic, ecologic and safety gains. But in order to really exploit it, you need to know exactly the hotspots, the failure behavior and the damage criterion of your structure. ‘‘Maybe the biggest challenge is that it requires a multidisciplinary approach,’’ says Ooijevaar. ‘‘The type and place of the sensors, data acquisition, failure mechanisms, wiring, energy supply to the sensors, where to process the data, etcetera. . .’’ It might still take a decade before SHM will be used in commercial aircraft, because of all the tests and certification in the aircraft industry. In wind turbines it should happen faster. There are fewer hot spots in a wind turbine blade.