- Email: [email protected]
Active control of a smart composite with shape memory alloy sheet using a plastic optical fiber sensor
Accepted Manuscript Title: Active Control of a Smart Composite with Shape Memory Alloy Sheet using a Plastic Optical Fiber Sensor Author: K.S.C. Kuang S.T. Quek W.J. Cantwell PII: DOI: Reference:
S0924-4247(13)00312-9 http://dx.doi.org/doi:10.1016/j.sna.2013.06.024 SNA 8384
To appear in:
Sensors and Actuators A
Received date: Revised date: Accepted date:
25-1-2013 26-6-2013 26-6-2013
Please cite this article as: K.S.C. Kuang, S.T. Quek, W.J. Cantwell, Active Control of a Smart Composite with Shape Memory Alloy Sheet using a Plastic Optical Fiber Sensor, Sensors and Actuators: A Physical (2013), http://dx.doi.org/10.1016/j.sna.2013.06.024 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Active Control of a Smart Composite with Shape Memory Alloy Sheet using a Plastic Optical Fiber Sensor
K.S.C. Kuang1*, S.T. Quek1 and W.J. Cantwell2
2
Department of Civil Engineering and Environmental Engineering, Blk. E1A, #07-03, National University of Singapore, 117576, Singapore
ip t
1
Department of Aerospace Engineering, Khalifa University of Science, Technology & Research (KUSTAR) P.O.Box 127788, Abu Dhabi, United Arab Emirates
us
cr
Emails: [email protected]; [email protected]; [email protected]
Abstract
This paper investigates the morphing characteristics of a fiber metal laminate (FML) based on shape memory
an
alloy (SMA) skin and a carbon fiber reinforced epoxy composite. Prior to manufacture, the SMA surface layer was trained by heating them to 500oC in a furnace and quenching rapidly in a water bath. The SMA skin was
M
then bonded to the composite and a thin silicon heater was attached to the SMA skin. During activation, the deflection of the beam was detected via a plastic optical fiber (POF) sensor bonded to the upper surface of the
d
smart FML. The output from the POF sensor was monitored using a photodetector which in turn was connected to a control unit. This closed-loop arrangement enables the operator to introduce the prescribed set-point
te
deflection values of the beam. The desired deflections of the smart FML are achieved by the control unit
Ac ce p
switching on, and subsequently switching off, the silicon heater using the POF to provide the feedback signal. Tests have shown that this smart FML is capable of accurately achieving prescribed values of beam deflection based on the POF sensors signal. The repeatability of the system has been demonstrated by systematically increasing and then decreasing the prescribed deflection and monitoring the subsequent response. In addition, the structure has been shown to be capable of accommodating unwanted disturbances, such as those associated with external loading regimes. During initial studies, using an on-off control strategy, the beam tended to overshoot the required deflection before stabilising at the set-point value some tens of seconds later. This effect was reduced through the tuning of a proportional-integrative-derivative (PID) control setting on the control unit. It is believed that the findings outlined in this paper demonstrate the potential benefits offered by smart hybrid structures. Keywords: shape memory alloy, plastic optical fiber, sensors, smart structures, fiber metal laminates, active control *Corresponding author
|Email: [email protected] |Tel: +65-65164683 | Fax: +65-67791635 Pg 1
Page 1 of 19
1. Introduction Fiber metal laminates (FMLs) are manufactured from alternating layered arrangement of composite materials metal alloy sheets. This hybrid material system offers a range of attractive features including outstanding
ip t
specific strength, high stiffness and superior fatigue properties associated with the fibrous phase as well as the intrinsic machinability and toughness associated with most metals. Previous studies have highlighted the
cr
excellent quasi-static and dynamic properties of FMLs, demonstrating their superiority in structural performance
us
compared to conventional plain aluminum alloys [1-3].
In recent years, there has been a growing interest in the morphing characteristics of aircraft wings and other
an
primary structures to achieve significant changes in aerodynamic form during flight. The use of shape memory alloys (SMAs) has been highlighted as a possible approach to introduce morphing capabilities in conventional aerospace materials [4-7]. As a result of their ability to undergo shape recovery while exhibiting a large
M
recovery stress, shape memory alloys have been selected for use in many applications. SMAs exhibit many attractive features including a shape memory effect (SME) and super-elastic behavior resulting from thermally-
d
induced and stress-induced martensitic transformations, respectively. In order for the former to occur, the shape
te
memory alloy must be heated above its activation temperature. Above this temperature, the martensitic phase will undergo a reverse transformation into an austenitic phase. A comprehensive review of the mechanics of
Ac ce p
shape memory alloy is given elsewhere in the literature [8, 9].
The SMA wires, with typical diameters ranging from 0.02 mm to 12 mm, are available commercially and their integration into composite materials has received much attention. Very little work has been reported, however, on the use of SMA sheets as actuators and as a structural member in conjunction with advanced composites. The embedding of SMA wires for strain recovery applications where the diameters of the wires are typically that of 10-order of magnitude larger than the reinforcing fibers, may lead to degradation of the structural properties of the host composites. It has been reported, however, that the introduction of super-elastic SMA wires (without actuation) embedded in fiber-reinforced composites would lead to potential reduction of the damage in the host material under low-velocity impact conditions compared to a conventional fiber-composite system [10, 11]. To achieve the improvement in impact response, however, it was critical to ensure a high level of impregnation of the SMA wires as well as an appropriate choice of volume fraction of SMA wires in the composite. Another drawback in embedding an array of SMA wires within composites is that this arrangement could be expected to
Pg 2
Page 2 of 19
further compound the loss of structural integrity leading to a structurally compromised material system. In addition, due to its greater degree of freedom compared to a rectangular sheet, placement and fixation of these wires to the host composite requires careful attention during specimen preparation and manufacture due to the difference in activation temperature of the SMA wire and the consolidation or curing temperature of the
ip t
polymer matrix.
The concept of using thin SMA sheets bonded to composite prepregs in the same fashion as conventional fiber
cr
metal laminates offers a viable alternative without the risk of degradation in structural properties in addition to the comparative ease of manufacture. The resulting material system termed as a smart FML, offers actuation
us
capability in addition to superior structural characteristics compared to conventional materials. Previous studies on the fracture properties of a fiber metal laminate with SMA sheet has shown attractive mechanical properties
an
as a result of incorporating shape memory alloy sheet as one of the constituent composite materials [12].
In order for an SMA to be used as an actuator, it is necessary to train the SMA sample to impart the desired
M
memorized shape and degree of strain recovery by subjecting it to a training regime which involves holding the specimen at an elevated temperature (400-500oC) for a short duration of time, from 3 – 30 minutes and
d
subsequently quenching it in a bath, typically cold water. The training involves the reordering of the dislocation
temperature.
te
into stable position such that the shape memory alloy reverts to this position when it is heated to its austenitic
Ac ce p
In control applications, the use of a suitable feedback sensor could lead to a significantly more accurate control of the parameter of interest. Due to the non-linear and hysteretic response exhibited by SMA actuators i.e. the deflection traces generated by the heating and cooling cycles do not coincide, thus precise positional control of the SMA, or the host specimen actuated by an SMA, is a challenge. Designs based on linear control are not suited to tackle this type of response and hence compensation techniques involving proportional and derivative control are often employed to overcome the hysteretic and non-linearity associated with SMA actuators. Song et al. used signals obtained from a laser sensor set-up in conjunction with a programmable power supply to control the displacement of the tip of a composite beam embedded with SMA wires [13] and successfully showed a high level of positional control.
Optical fiber sensors have received much attention in recent years for measuring a variety of parameters which include strain, temperature, deflection, pressure and others. Recent studies utilizing silica-based fiber Bragg grating sensors have demonstrated the possibility of using such sensors to monitor and control the condition of Pg 3
Page 3 of 19
an advanced-composite adaptive wing embedded with SMA wires [14, 15]. Research using intensity-based plastic optical fiber sensors on the hand, have shown promising results for a number of structural monitoring applications [16-20]. Compared to fiber Bragg grating sensors, the interrogation of intensity-based POF sensors are considerably easier and can be achieved using inexpensive and ruggedly packaged photodetectors. Another
ip t
advantage of using an intensity-based approach is the ease of interfacing between the photodetector units with other electrical devices. In this study the output of the photodetector unit is in the form of electrical voltage
cr
which is readily received by most PID controllers. It is the aim of this paper to investigate if the intensity-based POF sensors proposed in here can used to monitor a smart fiber metal laminate specimen effectively and with
us
ease.
In this work, a plastic optical fiber (POF) sensor has been attached to the surface of the advanced composite
an
beam host to provide the feedback signal to the controller unit to control the amount of heating necessary to achieve the required deflection in the smart FML. At the present form, the sensor is not suitable for embedding
M
due to the low operating temperature of PMMA POF. Additionally, the diameter of the bare POF used in this study is relatively large (1 mm) and may degrade the mechanical properties of the host in the embedded configuration. Smaller diameter POF (0.25 mm) may be embedded in composite systems with low processing
d
temperatures (< 90oC) and will be studied in future research. Other possibilities of integrating POF sensors to
te
the host structure include mounting the sensor to the interior surface (i.e. located within the cavity of the aircraft
Ac ce p
or turbine wing aerofoil), for example, if mounting on the external surface is not an option.
To achieve better positional control of the deflection of the specimen, in this study a direct measure of the physical state (i.e. deflection) of the beam was used as the feedback signal instead of temperature, which is the default feedback signal for the temperature controller unit. Although the temperature at which the SMA is subjected is related to the deflection of the beam, this relationship may not be readily formulated due to the effects of both hysteretic and non-linear thermomechanical response of the SMA. Temperature monitoring via a thermocouple provides an indirect measurement of the physical response of the SMA whereas monitoring the deflection (or strain) itself, via the POF sensor, gives a direct real-time physical response of the beam allowing a more precise control of the specimen.
This paper reports a study into the use of a nickel-titanium shape memory alloy sheet and fiber-reinforced composites to produce a novel smart fiber-metal laminate. A POF sensor was used to monitor the deflection of
Pg 4
Page 4 of 19
the specimen while providing the feedback signal to the controller which powers a thin silicon heater bonded to the SMA sheet.
2. Experimental Procedure
ip t
2.1. Specimen Fabrication The investigation focused on studying the morphing characteristics of a 1/1 FML consisting of an SMA sheet
cr
bonded to a carbon fiber reinforced composite. Clearly, this type of bi-material construction is impractical from an engineering standpoint but it is useful for demonstrating the morphing ability of SMA-based FMLs. The
us
SMA sheet was a 0.8 mm thick nickel-titanium (Ni-Ti) alloy type M, having an activation temperature (austenite finish), Af of 65oC. The SMA was supplied by Memory-Metalle GmbH, Germany. In preparation for
an
training, the 400 x 55 mm SMA sheets were rolled up and inserted into a steel tube with an inner diameter of 120 mm x thickness of 5 mm x width of 55 mm. The training regime involved heating the Ni-Ti SMA in a preheated furnace at a temperature of 500oC. A thermocouple was attached to the SMA specimen by inserting the
M
probe between the steel tube and SMA sheet to monitor the temperature experienced by the SMA. The temperature experienced by the SMA specimen increased rapidly and was held at 500 oC for two minutes before
d
being quenched in water whilst remaining constrained in the circular tube This treatment cycle was repeated
te
three times in order to ensure that significant strain recovery was achieved. It should be noted that three cycles
Ac ce p
may not be the optimum number for full strain recovery, but was proved sufficient for our purpose. The carbon fiber composite used to produce the inner core of the FML was supplied by Gurit AG, Switzerland (EP121-C15-53). This woven fiber system was cured for one hour at 125oC and then bonded to the Ni-Ti sheets using a two part epoxy adhesive. A thin silicon heater (with thickness, length and width dimensions of 0.4, 250 and 55 mm respectively) was then attached to the surface of the SMA sheet using a room-temperature vulcanising adhesive (1430RED7871 Innerbond from Inland Inc).
2.2 POF Sensor Design and Controller Construction The POF used in this study was based on a 2.2 mm diameter multimode, step-index fiber (PGU-CD1001-22E) from Toray Industries Inc. supplied with a tight polyethylene outer jacket to protect the inner fiber. Two POFs were used to construct the sensor. For each fiber, a section measuring approximately 20 mm in length of the jacket was removed to expose the bare PMMA fiber of 1 mm in diameter. The bare sections of the fibers were Pg 5
Page 5 of 19
inserted into the two ends of a glass tube to align the bare fibers as shown in Figure 1. A gauge length of 80 mm was used when installing the POF sensor to the FML beam. A calibration curve of the POF sensor used is given in Figure 2, relating the normalized POF signal loss to the change in the distance between the two cleaved fiber surfaces. This curve allows the change in relative movement of the two fibers to be determine from the optical
ip t
signal and therefore the estimated strain experienced by the host substrate. The POF sensor used throughout this investigation was based on a modification from a previous design [15].
cr
The basic POF sensor functions by measuring the relative separation of two cleaved fiber surfaces located within the glass tube. The output signal from the sensor is proportional to the relative separation of the two fiber
us
ends. Since the POF sensor was attached to the upper surface of the FML in this investigation, the two fiber ends in the sensor approach each other during activation of the SMA (the SMA curled to its trained shape),
an
giving an increase in the level of light transmitted along the fiber. Here, POF reader consisting of the light source and photodetector, given in Figure 2, was used to monitor the POF output signal, the magnitude of which
M
is relayed as a feedback signal to a control unit, which in turn control the power delivered to a thin silicon heater. The light source consisted of four stable high-intensity light-emitting diodes (LEDs) while the narrowband photodetector was capable of producing a voltage that is proportional to the intensity of the transmitted
te
fluctuations.
d
signal. Repeated testing demonstrated that the photodetector was stable and did not exhibit power source
Ac ce p
The silicon rubber heater used in the study was powered using a microprocessor-based temperature controller (Watlow Series SD 1/16 DIN). The controller was wired and configured to accept either a K-type thermocouple and a 0-5 V (dc) POF signal as the process variable - this could be toggled easily by keying in the appropriate parameter in the controller panel. In this study, the POF signal was assigned as the process variable to the controller. A solid-state relay (SSR-240-10A-DC1) was wired to the output of the controller to power the silicon rubber heater. The control algorithm was based on a simple on-off control initially and subsequently a proportional-integral control was introduced. The completed lay-up of the test specimen is shown in Figure 3. The silicon rubber heater layer is shown exposed for clarity.
3. Results and Discussion
Pg 6
Page 6 of 19
Initial attention focused on establishing the ability to control the hybrid structure by imposing different deflection settings to the smart FML via the controller unit. Here, three beam deflections were programmed into the controller unit as set-points (SP) and the response of the beam was tracked by the optical fiber assigned as the process variable (PV). It should be noted that these three positions were chosen in an arbitrary manner and
ip t
could, if necessary, be re-programmed to achieve any position within the range of the recovery strains that the SMA layer is capable of achieving. Figure 4 shows the resulting normalized response of the POF as well as the
cr
temperature within the sample. It should be noted that the temperature was measured via a small thermocouple position under the heating element. The horizontal lines as indicated by (A), (B) and (C) in Figure 4 correspond
us
to the three displacement settings that were inputted into the controller unit.
Consider first the initial heating cycle in Figure 4 i.e. that required to achieve Position A. Following the
an
application of heat via the surface-mounted silicon rubber heating pad, the initial portion of the deflection trace (indicated by the POF sensor) exhibits a steep climb. An examination of the thermocouple output shows that the temperature cycle mapped closely onto that of the sensor data. The beam deflection clearly overshoots the
M
desired response and the POF sensors force the controller unit to switch off the power supplied to the heater, leading to a drop in temperature. This then allowed the beam to relax and move downwards as the heater cooled
d
and finally settling at Position A after approximately 50 seconds. The evidence in the figure suggests that the
te
measured temperature overshoots by approximately 25o C, which in turn resulted in the unwanted overshoot in deflection. A larger displacement was then introduced by setting a higher set-point via the controller. The
Ac ce p
temperature and beam deflection increased rapidly once more before overshooting once again and ultimately stabilizing at the new position, Position B. The final deflection position was subsequently keyed into the controller’s input panel and a similar response of the SMA was observed, with the overshoot and final recovery to Position C. In the cooling phase following Position C, the previous set-point used to arrive at Position A was entered into the controller and was achieved after approximately 200 seconds without undershoot. A second set of heating cycle was carried out to assess the repeatability of the result and it is clear from Figure 4 that the setup allows the FML beam to achieve all three desired positions with a good degree of accuracy, albeit the overshoots which occurred in all five heating regime. For the final set of heating cycle, the beam was activated from room temperature condition and it was of interest to see the responsiveness of the smart FML beam and whether the desired position could still be achieved given a larger step-change in the beam deflection. The result showed that Position C was again achieved accurately following the overshoot. Closer examination suggests that the overshoot tend to increase as the prescribed
Pg 7
Page 7 of 19
displacement increases. The subsequent rise time, i.e. the time taken for the signal to increase to 100% of the steady state value, was found be approximately 10 seconds, associated with a heating rate of 6.5oC per second and a significant overshoot. The recovery rate was determined and found to be approximately 0.1% strain per second, highlighting the relatively high responsiveness of the SMA. From Figure 4, it is evident that the smart
ip t
hybrid system offers a high degree of repeatability with the system adopting the same set-point positions during different heating/cooling cycles. For example, Position C was assigned three times and in each case the beam
cr
deflected to the desired position with a high degree of precision.
Figure 5 shows the variation of the thermocouple reading with POF reading (effectively the beam deflection) for
us
the series of cycles shown previously in Figure 4. The graph clearly highlights the presence of both a relatively linear as well as a non-linear response from the FML. The data from the cooling portions of the thermo-
an
mechanical cycles appear to collapse onto a unique curve, whereas an examination of the heating traces suggests the dependence of response of the beam on the initial temperature from which the heating cycle begins. The larger the difference between the initial process value and the set-point, the more pronounced the hysteresis
M
loop, as evidenced by the relatively larger hysteresis loop for the 3rd cycle compared to the earlier cycles in Figure 5. It is evident in this figure that if temperature readings were used as the PV and SP values, this would
d
lead to ambiguous deflection positions - e.g. based on Figure 5, if Position C is the desired set-point
te
(corresponding to POF reading of approximately 0.062) by heating from room condition, a temperature set-point of 90oC could lead to an erroneous deflection position (one which corresponds to POF readings of both 0.062
Ac ce p
and 0.082). It is clear from the figure that for a given temperature set-point, two different beam deflections are possible i.e. one in the heating phase, the other in the cooling phase, since the heating and cooling traces do not overlap each other. The results in this study clearly demonstrates that using deflection as the process variable input and set-point value for control of SMA actuation offers a more attractive alternative compared to temperature to achieve accurate control of beam deflection. It is also interesting to note that the desired beam deflection positions were achieved following an overshoot in the deflection in each case, inherent to the simple on-off control adopted thus far. PID controls may, in turn, be introduced to eliminate the overshoot and subsequently achieve the desired steady-state condition. When used in real engineering applications, smart structures are likely to be subjected to external loads, such as those associated with wind buffeting or impact loading. In this study, the ability of the POF to detect an external disturbance and to auto-correct its configuration was also assessed. Here, a weight was added to the free end of the cantilever to introduce a pronounced displacement perturbation (i.e. process disturbance) and the subsequent
Pg 8
Page 8 of 19
response is shown in Figure 6. With the introduction of the perturbation, the POF sensor detects the change in the deflection of the beam which in turn triggers the controller unit to initiate the heating pad. In turn, the SMA was activated forcing the beam back to its predetermined position. This is evident in the figure where an initial weight was applied after 120 seconds causing a reduction in the POF signal (the beam was being forced
ip t
downwards). The structure then responds, recovering its initial position in approximately ten seconds. A second weight was applied at 220 seconds which again triggered the heating element to activate the beam to recover.
cr
Finally, when the two set of weights were removed from the end of the beam, it spring upwards, reflecting the trained shape of the SMA. The temperature is evidently reduced, allowing the beam to return to the original
us
position as evidenced by the POF plot. Here, the recovery time was relatively long, with the beam taking approximately 100 seconds to recover to the initial preset value. The recovery time can be reduced by means of
an
assisted cooling e.g. by the use a fan to bring down the temperature of the specimen through better heat exchange with its surrounding. From the plot, it is observed that the set-point position was achieved during the
M
cooling phase in each case and no overshoot was exhibited in the last cycle of test in contrast with previous two cycles. It is also interesting to note that in each case, overshoot only occur during the heating phase of the control while none occurred during the cooling phase. It may be argued that the lower rate of change in the
te
off control.
d
temperature during the cooling phase allows a more precise and controlled response despite using simply an on-
Ac ce p
In the final part of this study, the possibility of reducing or even eliminating the overshoot observed during the heating phase of each cycle in Figure 4 was investigated. Here, a PID control was introduced to minimize the overshoot. During the heating phase of the cycle, the proportional control reduces the average power sent to the heater as it approaches the set-point based on the setting of the proportional band. The tuning of the parameter associated with the proportional band in effect was to obtain the optimum heater on-off time ratio proportional to the difference in the values between the set point and the process input. The three PID settings were adjusted via trail-and-error to achieve a fast control and satisfactory steady-state stability. In this part of the study, the same heating and cooling sequences were adopted as those used in Figure 4. The effect of introducing a PID control to the system can be seen in Figure 7 where, with the adoption of suitable settings for the control value, it was possible to eliminate the large overshoot. The results obtained, however, lead to a relatively modest oscillatory response which could be reduced by further tuning for the proportional band to achieve the desired level of control. Figure 8 shows the variation of the thermocouple reading with POF reading for the series of cycles shown in Figure 7. Compared to Figure 5, it is apparent that the hysteresis loop is moderately smaller due Pg 9
Page 9 of 19
to the elimination of the overshoot. Similarly to Figure 5, the cooling traces for all 3 cycles overlap each other well. The result in the figure again suggests the inherent difficulty in using temperature as the process variable for deflection control of SMA actuators while highlighting the effectiveness of adopting low-cost sensors such
ip t
as POF for accurate monitoring and control of smart FMLs.
4. Conclusions
cr
A novel hybrid laminate based on a nickel-titanium shape memory alloy has been developed and its activation capabilities have been investigated. The results have highlighted the controllability of this structure through the
us
introduction of a POF sensor which was used to monitor the deflection of the beam. This signal was then used as feedback to the controller unit. Tests were also conducted to assess the auto-correcting capability of the
an
hybrid using the POF sensor signal. Here, process perturbations in the form of simple weights, were added to the end of the beam forcing the deflection away from the desired set-point forcing the controller unit to activate the
M
heating element and return the beam to the desired configuration
Tuning of the PID parameters on the controller unit has shown that it was possible to eliminate the large overshoot initially observed resulting in a better control of the beam deflection for the entire duration of the test.
d
It is expected that the oscillatory behavior could be further minimized through further tuning of the PID control.
te
The results have also highlighted that the incorporation of POF sensors into the laminates can successfully
Ac ce p
measure the out-of-plane response of these novel hybrid structures.
ACKNOWLEDGEMENTS
The authors acknowledge the funding (R-534-000-007-414) from the Minerals, Metals and Materials Technology Centre (M3TC) of the National University of Singapore.
Pg 10
Page 10 of 19
References N Tsartsaris, M Meo, M Dolce, U Polimeno, M Guida and F Marulo, “Low velocity impact behaviour of fiber metal laminates”, Journal of Composite Materials Vol.45 (2011) pp.803-814.
[2]
S Krishnakumar, “Fiber Metal Laminates: The synthesis of metals and composites”, Materials and Manufacturing Processes Vol.9 (1994) pp.295-354.
[3]
LB Vogelesang and AVlot, “Development of fiber metal laminates for advanced aerospace structures” Journal of Materials Processing Technology Vol.103 (2000) pp.1-5.
[4]
R Loewy, “Recent developments in smart structures with aeronautical applications”, Smart Materials and Structures Vol.6 (1997) R11-R42.
[5]
DJ Hartl and DC Lagoudas, “Aerospace applications of shape memory alloys”, Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering Vol. 221 (2007) pp.535-552.
[6]
J Kudva, “Overview of the DARPA smart wing project”, Journal of Intelligent Material Systems and Structures Vol.15 (2004) pp.261-267.
[7]
B Sanders, R Crowe and E Garcia, “DARPA-Smart Materials and Structures demonstration program overview”, Journal of Intelligent Material Systems and Structures Vol.15 (2004) pp.227-233.
[8]
V Birman, “Review of mechanics of shape memory alloy structures”, Applied Mechanics Reviews Vol.50 (1997) pp.629-645.
[9]
K Otsuka and CM Wayman (Eds.) Shape memory materials, Cambridge University Press, Cambridge, 1999.
[10]
M Meo, F Marulo, M Guida and S Russo, “Shape memory alloy hybrid composites for improved impact properties for aeronautical applications” Composite Structures Vol.95 (2013) pp.756-766.
[11]
S Pappada, R Rametta, A Largo and A Maffezzoli, “Low-velocity impact response in composite plates embedding shape memory alloy wires” Polymer Composites Vol.33 (2012) pp.655-664.
[12]
P Cortes, WJ Cantwell and KSC Kuang, “The fracture properties of a smart fiber metal laminate”, Polymer Composites, Vol.28 (2007) pp.534-544.
[13]
G Song, B Kelly and B N Agrawal, “Active position control of a shape memory alloy wire actuated composite beam” Smart Materials and Structures Vol.9 (2000) pp.711-716.
[15]
[16] [17]
cr
us
an
M
d
te
Ac ce p
[14]
ip t
[1]
J A Balta, F Bosia, V Michaud, G Dunkel, J Botsis, JA Manson, “ Smart Composites with embedded SMA actuators and fiber Bragg grating sensors: activation and control” Smart Materials and Structures Vol.14 (2005) pp.457-465. M Mieloszyk, L Skarbek, M Krawczuk, W Ostachowicz and A Zak, “Application of fiber Bragg gratings sensors for structural health monitoring of an adaptive wing”, Smart Materials and Structures Vol.20 (2011) p.125014 (12pp) KSC Kuang, ST Quek and M Maalej, “Assessment of an extrinsic polymer-based optical fiber sensor for structural health monitoring” Measurement Science and Technology Vol.15 (2004) pp.2133-2141.
KSC Kuang, M Maalej and ST Quek, “An application of a plastic optical fiber sensor and genetic algorithm for structural health monitoring” Journal of Intelligent Material Systems and Structures Vol.17 pp.361-379.
[18]
KSC Kuang, CY Tan, SH Chew and ST Quek, “Monitoring of large strains in submerged geotextile tubes using plastic optical fiber sensors” Sensors and Actuators A: Physical Vol. 167 pp.338-346.
[19]
KSC Kuang, ST Quek and M Maalej, “Plastic optical fiber sensor for damage detection in offshore structures” Proceedings of the SPIE Vol.7522 (2010) art. no. 75223A.
[20]
KSC Kuang and WJ Cantwell, “The use of plastic optical fibers and shape memory alloys for damage assessment and damping control in composite materials” Measurement Science and Technology Vol. 14 (2003) pp.1305-1313
Pg 11
Page 11 of 19
Illustrations
sensor gauge length
adhesive
2.2 mm
outer jacket light in
ip t
80 mm
housing
cr
light out
ID 1.06 mm
substrate
1 mm bare POF
us
OD 2.0 mm
Ac ce p
10.0
te
d
M
an
Figure 1 Schematic of the POF sensor used in the study.
Extension (mm)
8.0
POF reading Curve fitting
6.0
4.0
2.0
0.0 0.0
0.1
0.2
0.3
0.4
0.5
0.6
Normalised POF Loss
Figure 2 Plot showing the response of POF sensor to externally applied extension
Pg 12
Page 12 of 19
ip t cr us an
M
Figure 3 Schematic of the experimental test set-up.
te
0.09 0.08
POF Sensor
90
Ac ce p
Thermocouple
0.07
80 70
(C)
0.06
60
0.05
50 (B)
0.04
40
0.03
30
0.02
20
(A)
0.01
Temperature oC
Normalised POF Reading
100
d
0.1
10
0
0 0
500
1000
Time (seconds)
1500
2000
Figure 4 Plot showing the controllability of the smart FML beam using POF sensor.
Pg 13
Page 13 of 19
0.1
1st Cycle 2nd Cycle
0.08 0.07
ip t
3rd Cycle Position (C)
0.06
cr
0.05 Position (B)
0.04 0.03 0.02
us
Normalised POF Reading
0.09
Position (A)
an
0.01 0 20
40 60 Temperature (oC)
80
100
M
0
d
Figure 5 Plot showing the thermomechanical response of the smart FML.
Pertubation introduced here Weights added at free end of beam
0.06
More weights added at free end of beam
80 Controlled position of beam
70 60
0.05
50
0.04
40
0.03
30
0.02
POF Sensor Thermocouple
Weights removed from free end of beam
0.01
Temperature (oC)
Ac ce p
0.07
Normalised POF Reading
90
te
0.08
20 10
0
0 0
100
200
300
400
500
Time (sec) Figure 6 Plot showing the self-correcting response of the SMA beam based on the POF sensor signal.
Pg 14
Page 14 of 19
0.1
100 POF Sensor Thermocouple
0.08
90 80
0.07
70
ip t
(C)
0.06
60
0.05
cr
(B)
us
0.04 0.03
an
0.02 0.01 0 500
1000 Time (sec)
M
0
(A)
50 40 30
Temperature (oC)
Normalised POF Reading
0.09
20 10
1500
0 2000
te
d
Figure 7 Plot showing the response of the beam using PID control to reduce the overshoot observed previously.
0.1
1st Cycle
Normalised POF Reading
Ac ce p
0.09
2nd Cycle 3rd Cycle
0.08 0.07
Position (C)
0.06 0.05
Position (B)
0.04 0.03
Position (A)
0.02 0.01 0 0
20
40 Temperature (oC)
60
80
Pg 15
Page 15 of 19
Ac ce p
te
d
M
an
us
cr
ip t
Figure 8 Plot showing the thermomechanical response of the smart FML under PID control.
Pg 16
Page 16 of 19
Paper: Active Control of a Smart Composite with Shape Memory Alloy Sheet using a Plastic Optical Fiber Sensor Ms. Ref. No: SNA-D-13-00077
ip t
Figures Caption List Figure 1: Schematic of the POF sensor used in the study
cr
Figure 2 : Plot showing the response of POF sensor to externally applied extension Figure 3: Schematic of the experimental test set-up
us
Figure 4: Plot showing the controllability of the smart FML beam using POF sensor Figure 5: Plot showing the thermomechanical response of the smart FML
an
Figure 6: Plot showing the self-correcting response of the SMA beam based on the POF sensor signal
M
Figure 7: Plot showing the response of the beam using PID control to reduce the overshoot observed previously
Ac ce p
te
d
Figure 8: Plot showing the thermomechanical response of the smart FML
Pg 17
Page 17 of 19
Authors Biography
cr
ip t
Kevin S. C. Kuang is an Assistant Professor at the Department of Civil and Environmental Engineering and is active in the research areas of smart structures, structural health monitoring and st sensors technology. He obtained his PhD and BEng (1 Hons) degrees from the Universities of Liverpool and Leeds, United Kingdom respectively. He has commercialized an optical fibre sensor educational kit and has a patent granted on optical fibre sensor (U.S. Patent Number 7496247). He has published over 40 technical papers and has been invited to give talks in the Institute of Engineers Singapore, SPRING Technical Conference, and Singapore Science Centre. He was also an invited speaker at the 20th IPOF Conference in Bilbao (Spain) 2011 and the JEC Innovative Composites Summit 2011.
M
an
us
Ser-Tong Quek is currently Professor and Vice Dean of Graduate Programmes and Director of the Global Engineering Programme in the Faculty of Engineering and Professor in the Department of Civil and Environmental Engineering at the National University of Singapore (NUS). He completed his Ph.D. in 1986 at the University of Illinois at Urbana-Champaign, Illinois, under Professor Alfredo Ang before returning to NUS to resume his academic career. His research encompasses structural health monitoring, structural reliability and mechanics and innovative structural composites. He has published more than 150 international journal papers. He was a recipient of Institution of Engineers Singapore KYD Gin Prestigious Publication Award and Defense Technology Prize for Research in 1990 and 2006, respectively.
Ac ce p
te
d
Wesley Cantwell is a Professor of Aerospace Engineering at Khalifa University of Science Technology and Research in the United Arab Emirates where he is on leave from the University of Liverpool. Wesley joined Khalifa University in October 2012 and is the Director of the Aerospace Research and Innovation Center (ARIC). In Liverpool, he leads a group of ten researchers investigating the fracture properties of lightweight materials and structures for use in the aerospace industry. Much of his research focuses on developing sandwich structures for energy absorption, understanding the blast response of composite materials, characterizing novel metal composite hybrids and producing and characterizing smart aircraft structures.
Pg 18
Page 18 of 19
Highlights (4 highlights; 85 characters max. per highlight)
pt
ed
M
an
us
cr
ip t
SMA sheet was integrated to FMLs and activated using a thin silicon heater POF sensor allows control of the FML deflection accurately and good repeatability POF data is used as PV to control the deflection of FML instead of temperature. Tests showed set deflections for FML could be achieved accurately using POF sensor.
Ac ce
Page 19 of 19
S0924-4247(13)00312-9 http://dx.doi.org/doi:10.1016/j.sna.2013.06.024 SNA 8384
To appear in:
Sensors and Actuators A
Received date: Revised date: Accepted date:
25-1-2013 26-6-2013 26-6-2013
Please cite this article as: K.S.C. Kuang, S.T. Quek, W.J. Cantwell, Active Control of a Smart Composite with Shape Memory Alloy Sheet using a Plastic Optical Fiber Sensor, Sensors and Actuators: A Physical (2013), http://dx.doi.org/10.1016/j.sna.2013.06.024 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Active Control of a Smart Composite with Shape Memory Alloy Sheet using a Plastic Optical Fiber Sensor
K.S.C. Kuang1*, S.T. Quek1 and W.J. Cantwell2
2
Department of Civil Engineering and Environmental Engineering, Blk. E1A, #07-03, National University of Singapore, 117576, Singapore
ip t
1
Department of Aerospace Engineering, Khalifa University of Science, Technology & Research (KUSTAR) P.O.Box 127788, Abu Dhabi, United Arab Emirates
us
cr
Emails: [email protected]; [email protected]; [email protected]
Abstract
This paper investigates the morphing characteristics of a fiber metal laminate (FML) based on shape memory
an
alloy (SMA) skin and a carbon fiber reinforced epoxy composite. Prior to manufacture, the SMA surface layer was trained by heating them to 500oC in a furnace and quenching rapidly in a water bath. The SMA skin was
M
then bonded to the composite and a thin silicon heater was attached to the SMA skin. During activation, the deflection of the beam was detected via a plastic optical fiber (POF) sensor bonded to the upper surface of the
d
smart FML. The output from the POF sensor was monitored using a photodetector which in turn was connected to a control unit. This closed-loop arrangement enables the operator to introduce the prescribed set-point
te
deflection values of the beam. The desired deflections of the smart FML are achieved by the control unit
Ac ce p
switching on, and subsequently switching off, the silicon heater using the POF to provide the feedback signal. Tests have shown that this smart FML is capable of accurately achieving prescribed values of beam deflection based on the POF sensors signal. The repeatability of the system has been demonstrated by systematically increasing and then decreasing the prescribed deflection and monitoring the subsequent response. In addition, the structure has been shown to be capable of accommodating unwanted disturbances, such as those associated with external loading regimes. During initial studies, using an on-off control strategy, the beam tended to overshoot the required deflection before stabilising at the set-point value some tens of seconds later. This effect was reduced through the tuning of a proportional-integrative-derivative (PID) control setting on the control unit. It is believed that the findings outlined in this paper demonstrate the potential benefits offered by smart hybrid structures. Keywords: shape memory alloy, plastic optical fiber, sensors, smart structures, fiber metal laminates, active control *Corresponding author
|Email: [email protected] |Tel: +65-65164683 | Fax: +65-67791635 Pg 1
Page 1 of 19
1. Introduction Fiber metal laminates (FMLs) are manufactured from alternating layered arrangement of composite materials metal alloy sheets. This hybrid material system offers a range of attractive features including outstanding
ip t
specific strength, high stiffness and superior fatigue properties associated with the fibrous phase as well as the intrinsic machinability and toughness associated with most metals. Previous studies have highlighted the
cr
excellent quasi-static and dynamic properties of FMLs, demonstrating their superiority in structural performance
us
compared to conventional plain aluminum alloys [1-3].
In recent years, there has been a growing interest in the morphing characteristics of aircraft wings and other
an
primary structures to achieve significant changes in aerodynamic form during flight. The use of shape memory alloys (SMAs) has been highlighted as a possible approach to introduce morphing capabilities in conventional aerospace materials [4-7]. As a result of their ability to undergo shape recovery while exhibiting a large
M
recovery stress, shape memory alloys have been selected for use in many applications. SMAs exhibit many attractive features including a shape memory effect (SME) and super-elastic behavior resulting from thermally-
d
induced and stress-induced martensitic transformations, respectively. In order for the former to occur, the shape
te
memory alloy must be heated above its activation temperature. Above this temperature, the martensitic phase will undergo a reverse transformation into an austenitic phase. A comprehensive review of the mechanics of
Ac ce p
shape memory alloy is given elsewhere in the literature [8, 9].
The SMA wires, with typical diameters ranging from 0.02 mm to 12 mm, are available commercially and their integration into composite materials has received much attention. Very little work has been reported, however, on the use of SMA sheets as actuators and as a structural member in conjunction with advanced composites. The embedding of SMA wires for strain recovery applications where the diameters of the wires are typically that of 10-order of magnitude larger than the reinforcing fibers, may lead to degradation of the structural properties of the host composites. It has been reported, however, that the introduction of super-elastic SMA wires (without actuation) embedded in fiber-reinforced composites would lead to potential reduction of the damage in the host material under low-velocity impact conditions compared to a conventional fiber-composite system [10, 11]. To achieve the improvement in impact response, however, it was critical to ensure a high level of impregnation of the SMA wires as well as an appropriate choice of volume fraction of SMA wires in the composite. Another drawback in embedding an array of SMA wires within composites is that this arrangement could be expected to
Pg 2
Page 2 of 19
further compound the loss of structural integrity leading to a structurally compromised material system. In addition, due to its greater degree of freedom compared to a rectangular sheet, placement and fixation of these wires to the host composite requires careful attention during specimen preparation and manufacture due to the difference in activation temperature of the SMA wire and the consolidation or curing temperature of the
ip t
polymer matrix.
The concept of using thin SMA sheets bonded to composite prepregs in the same fashion as conventional fiber
cr
metal laminates offers a viable alternative without the risk of degradation in structural properties in addition to the comparative ease of manufacture. The resulting material system termed as a smart FML, offers actuation
us
capability in addition to superior structural characteristics compared to conventional materials. Previous studies on the fracture properties of a fiber metal laminate with SMA sheet has shown attractive mechanical properties
an
as a result of incorporating shape memory alloy sheet as one of the constituent composite materials [12].
In order for an SMA to be used as an actuator, it is necessary to train the SMA sample to impart the desired
M
memorized shape and degree of strain recovery by subjecting it to a training regime which involves holding the specimen at an elevated temperature (400-500oC) for a short duration of time, from 3 – 30 minutes and
d
subsequently quenching it in a bath, typically cold water. The training involves the reordering of the dislocation
temperature.
te
into stable position such that the shape memory alloy reverts to this position when it is heated to its austenitic
Ac ce p
In control applications, the use of a suitable feedback sensor could lead to a significantly more accurate control of the parameter of interest. Due to the non-linear and hysteretic response exhibited by SMA actuators i.e. the deflection traces generated by the heating and cooling cycles do not coincide, thus precise positional control of the SMA, or the host specimen actuated by an SMA, is a challenge. Designs based on linear control are not suited to tackle this type of response and hence compensation techniques involving proportional and derivative control are often employed to overcome the hysteretic and non-linearity associated with SMA actuators. Song et al. used signals obtained from a laser sensor set-up in conjunction with a programmable power supply to control the displacement of the tip of a composite beam embedded with SMA wires [13] and successfully showed a high level of positional control.
Optical fiber sensors have received much attention in recent years for measuring a variety of parameters which include strain, temperature, deflection, pressure and others. Recent studies utilizing silica-based fiber Bragg grating sensors have demonstrated the possibility of using such sensors to monitor and control the condition of Pg 3
Page 3 of 19
an advanced-composite adaptive wing embedded with SMA wires [14, 15]. Research using intensity-based plastic optical fiber sensors on the hand, have shown promising results for a number of structural monitoring applications [16-20]. Compared to fiber Bragg grating sensors, the interrogation of intensity-based POF sensors are considerably easier and can be achieved using inexpensive and ruggedly packaged photodetectors. Another
ip t
advantage of using an intensity-based approach is the ease of interfacing between the photodetector units with other electrical devices. In this study the output of the photodetector unit is in the form of electrical voltage
cr
which is readily received by most PID controllers. It is the aim of this paper to investigate if the intensity-based POF sensors proposed in here can used to monitor a smart fiber metal laminate specimen effectively and with
us
ease.
In this work, a plastic optical fiber (POF) sensor has been attached to the surface of the advanced composite
an
beam host to provide the feedback signal to the controller unit to control the amount of heating necessary to achieve the required deflection in the smart FML. At the present form, the sensor is not suitable for embedding
M
due to the low operating temperature of PMMA POF. Additionally, the diameter of the bare POF used in this study is relatively large (1 mm) and may degrade the mechanical properties of the host in the embedded configuration. Smaller diameter POF (0.25 mm) may be embedded in composite systems with low processing
d
temperatures (< 90oC) and will be studied in future research. Other possibilities of integrating POF sensors to
te
the host structure include mounting the sensor to the interior surface (i.e. located within the cavity of the aircraft
Ac ce p
or turbine wing aerofoil), for example, if mounting on the external surface is not an option.
To achieve better positional control of the deflection of the specimen, in this study a direct measure of the physical state (i.e. deflection) of the beam was used as the feedback signal instead of temperature, which is the default feedback signal for the temperature controller unit. Although the temperature at which the SMA is subjected is related to the deflection of the beam, this relationship may not be readily formulated due to the effects of both hysteretic and non-linear thermomechanical response of the SMA. Temperature monitoring via a thermocouple provides an indirect measurement of the physical response of the SMA whereas monitoring the deflection (or strain) itself, via the POF sensor, gives a direct real-time physical response of the beam allowing a more precise control of the specimen.
This paper reports a study into the use of a nickel-titanium shape memory alloy sheet and fiber-reinforced composites to produce a novel smart fiber-metal laminate. A POF sensor was used to monitor the deflection of
Pg 4
Page 4 of 19
the specimen while providing the feedback signal to the controller which powers a thin silicon heater bonded to the SMA sheet.
2. Experimental Procedure
ip t
2.1. Specimen Fabrication The investigation focused on studying the morphing characteristics of a 1/1 FML consisting of an SMA sheet
cr
bonded to a carbon fiber reinforced composite. Clearly, this type of bi-material construction is impractical from an engineering standpoint but it is useful for demonstrating the morphing ability of SMA-based FMLs. The
us
SMA sheet was a 0.8 mm thick nickel-titanium (Ni-Ti) alloy type M, having an activation temperature (austenite finish), Af of 65oC. The SMA was supplied by Memory-Metalle GmbH, Germany. In preparation for
an
training, the 400 x 55 mm SMA sheets were rolled up and inserted into a steel tube with an inner diameter of 120 mm x thickness of 5 mm x width of 55 mm. The training regime involved heating the Ni-Ti SMA in a preheated furnace at a temperature of 500oC. A thermocouple was attached to the SMA specimen by inserting the
M
probe between the steel tube and SMA sheet to monitor the temperature experienced by the SMA. The temperature experienced by the SMA specimen increased rapidly and was held at 500 oC for two minutes before
d
being quenched in water whilst remaining constrained in the circular tube This treatment cycle was repeated
te
three times in order to ensure that significant strain recovery was achieved. It should be noted that three cycles
Ac ce p
may not be the optimum number for full strain recovery, but was proved sufficient for our purpose. The carbon fiber composite used to produce the inner core of the FML was supplied by Gurit AG, Switzerland (EP121-C15-53). This woven fiber system was cured for one hour at 125oC and then bonded to the Ni-Ti sheets using a two part epoxy adhesive. A thin silicon heater (with thickness, length and width dimensions of 0.4, 250 and 55 mm respectively) was then attached to the surface of the SMA sheet using a room-temperature vulcanising adhesive (1430RED7871 Innerbond from Inland Inc).
2.2 POF Sensor Design and Controller Construction The POF used in this study was based on a 2.2 mm diameter multimode, step-index fiber (PGU-CD1001-22E) from Toray Industries Inc. supplied with a tight polyethylene outer jacket to protect the inner fiber. Two POFs were used to construct the sensor. For each fiber, a section measuring approximately 20 mm in length of the jacket was removed to expose the bare PMMA fiber of 1 mm in diameter. The bare sections of the fibers were Pg 5
Page 5 of 19
inserted into the two ends of a glass tube to align the bare fibers as shown in Figure 1. A gauge length of 80 mm was used when installing the POF sensor to the FML beam. A calibration curve of the POF sensor used is given in Figure 2, relating the normalized POF signal loss to the change in the distance between the two cleaved fiber surfaces. This curve allows the change in relative movement of the two fibers to be determine from the optical
ip t
signal and therefore the estimated strain experienced by the host substrate. The POF sensor used throughout this investigation was based on a modification from a previous design [15].
cr
The basic POF sensor functions by measuring the relative separation of two cleaved fiber surfaces located within the glass tube. The output signal from the sensor is proportional to the relative separation of the two fiber
us
ends. Since the POF sensor was attached to the upper surface of the FML in this investigation, the two fiber ends in the sensor approach each other during activation of the SMA (the SMA curled to its trained shape),
an
giving an increase in the level of light transmitted along the fiber. Here, POF reader consisting of the light source and photodetector, given in Figure 2, was used to monitor the POF output signal, the magnitude of which
M
is relayed as a feedback signal to a control unit, which in turn control the power delivered to a thin silicon heater. The light source consisted of four stable high-intensity light-emitting diodes (LEDs) while the narrowband photodetector was capable of producing a voltage that is proportional to the intensity of the transmitted
te
fluctuations.
d
signal. Repeated testing demonstrated that the photodetector was stable and did not exhibit power source
Ac ce p
The silicon rubber heater used in the study was powered using a microprocessor-based temperature controller (Watlow Series SD 1/16 DIN). The controller was wired and configured to accept either a K-type thermocouple and a 0-5 V (dc) POF signal as the process variable - this could be toggled easily by keying in the appropriate parameter in the controller panel. In this study, the POF signal was assigned as the process variable to the controller. A solid-state relay (SSR-240-10A-DC1) was wired to the output of the controller to power the silicon rubber heater. The control algorithm was based on a simple on-off control initially and subsequently a proportional-integral control was introduced. The completed lay-up of the test specimen is shown in Figure 3. The silicon rubber heater layer is shown exposed for clarity.
3. Results and Discussion
Pg 6
Page 6 of 19
Initial attention focused on establishing the ability to control the hybrid structure by imposing different deflection settings to the smart FML via the controller unit. Here, three beam deflections were programmed into the controller unit as set-points (SP) and the response of the beam was tracked by the optical fiber assigned as the process variable (PV). It should be noted that these three positions were chosen in an arbitrary manner and
ip t
could, if necessary, be re-programmed to achieve any position within the range of the recovery strains that the SMA layer is capable of achieving. Figure 4 shows the resulting normalized response of the POF as well as the
cr
temperature within the sample. It should be noted that the temperature was measured via a small thermocouple position under the heating element. The horizontal lines as indicated by (A), (B) and (C) in Figure 4 correspond
us
to the three displacement settings that were inputted into the controller unit.
Consider first the initial heating cycle in Figure 4 i.e. that required to achieve Position A. Following the
an
application of heat via the surface-mounted silicon rubber heating pad, the initial portion of the deflection trace (indicated by the POF sensor) exhibits a steep climb. An examination of the thermocouple output shows that the temperature cycle mapped closely onto that of the sensor data. The beam deflection clearly overshoots the
M
desired response and the POF sensors force the controller unit to switch off the power supplied to the heater, leading to a drop in temperature. This then allowed the beam to relax and move downwards as the heater cooled
d
and finally settling at Position A after approximately 50 seconds. The evidence in the figure suggests that the
te
measured temperature overshoots by approximately 25o C, which in turn resulted in the unwanted overshoot in deflection. A larger displacement was then introduced by setting a higher set-point via the controller. The
Ac ce p
temperature and beam deflection increased rapidly once more before overshooting once again and ultimately stabilizing at the new position, Position B. The final deflection position was subsequently keyed into the controller’s input panel and a similar response of the SMA was observed, with the overshoot and final recovery to Position C. In the cooling phase following Position C, the previous set-point used to arrive at Position A was entered into the controller and was achieved after approximately 200 seconds without undershoot. A second set of heating cycle was carried out to assess the repeatability of the result and it is clear from Figure 4 that the setup allows the FML beam to achieve all three desired positions with a good degree of accuracy, albeit the overshoots which occurred in all five heating regime. For the final set of heating cycle, the beam was activated from room temperature condition and it was of interest to see the responsiveness of the smart FML beam and whether the desired position could still be achieved given a larger step-change in the beam deflection. The result showed that Position C was again achieved accurately following the overshoot. Closer examination suggests that the overshoot tend to increase as the prescribed
Pg 7
Page 7 of 19
displacement increases. The subsequent rise time, i.e. the time taken for the signal to increase to 100% of the steady state value, was found be approximately 10 seconds, associated with a heating rate of 6.5oC per second and a significant overshoot. The recovery rate was determined and found to be approximately 0.1% strain per second, highlighting the relatively high responsiveness of the SMA. From Figure 4, it is evident that the smart
ip t
hybrid system offers a high degree of repeatability with the system adopting the same set-point positions during different heating/cooling cycles. For example, Position C was assigned three times and in each case the beam
cr
deflected to the desired position with a high degree of precision.
Figure 5 shows the variation of the thermocouple reading with POF reading (effectively the beam deflection) for
us
the series of cycles shown previously in Figure 4. The graph clearly highlights the presence of both a relatively linear as well as a non-linear response from the FML. The data from the cooling portions of the thermo-
an
mechanical cycles appear to collapse onto a unique curve, whereas an examination of the heating traces suggests the dependence of response of the beam on the initial temperature from which the heating cycle begins. The larger the difference between the initial process value and the set-point, the more pronounced the hysteresis
M
loop, as evidenced by the relatively larger hysteresis loop for the 3rd cycle compared to the earlier cycles in Figure 5. It is evident in this figure that if temperature readings were used as the PV and SP values, this would
d
lead to ambiguous deflection positions - e.g. based on Figure 5, if Position C is the desired set-point
te
(corresponding to POF reading of approximately 0.062) by heating from room condition, a temperature set-point of 90oC could lead to an erroneous deflection position (one which corresponds to POF readings of both 0.062
Ac ce p
and 0.082). It is clear from the figure that for a given temperature set-point, two different beam deflections are possible i.e. one in the heating phase, the other in the cooling phase, since the heating and cooling traces do not overlap each other. The results in this study clearly demonstrates that using deflection as the process variable input and set-point value for control of SMA actuation offers a more attractive alternative compared to temperature to achieve accurate control of beam deflection. It is also interesting to note that the desired beam deflection positions were achieved following an overshoot in the deflection in each case, inherent to the simple on-off control adopted thus far. PID controls may, in turn, be introduced to eliminate the overshoot and subsequently achieve the desired steady-state condition. When used in real engineering applications, smart structures are likely to be subjected to external loads, such as those associated with wind buffeting or impact loading. In this study, the ability of the POF to detect an external disturbance and to auto-correct its configuration was also assessed. Here, a weight was added to the free end of the cantilever to introduce a pronounced displacement perturbation (i.e. process disturbance) and the subsequent
Pg 8
Page 8 of 19
response is shown in Figure 6. With the introduction of the perturbation, the POF sensor detects the change in the deflection of the beam which in turn triggers the controller unit to initiate the heating pad. In turn, the SMA was activated forcing the beam back to its predetermined position. This is evident in the figure where an initial weight was applied after 120 seconds causing a reduction in the POF signal (the beam was being forced
ip t
downwards). The structure then responds, recovering its initial position in approximately ten seconds. A second weight was applied at 220 seconds which again triggered the heating element to activate the beam to recover.
cr
Finally, when the two set of weights were removed from the end of the beam, it spring upwards, reflecting the trained shape of the SMA. The temperature is evidently reduced, allowing the beam to return to the original
us
position as evidenced by the POF plot. Here, the recovery time was relatively long, with the beam taking approximately 100 seconds to recover to the initial preset value. The recovery time can be reduced by means of
an
assisted cooling e.g. by the use a fan to bring down the temperature of the specimen through better heat exchange with its surrounding. From the plot, it is observed that the set-point position was achieved during the
M
cooling phase in each case and no overshoot was exhibited in the last cycle of test in contrast with previous two cycles. It is also interesting to note that in each case, overshoot only occur during the heating phase of the control while none occurred during the cooling phase. It may be argued that the lower rate of change in the
te
off control.
d
temperature during the cooling phase allows a more precise and controlled response despite using simply an on-
Ac ce p
In the final part of this study, the possibility of reducing or even eliminating the overshoot observed during the heating phase of each cycle in Figure 4 was investigated. Here, a PID control was introduced to minimize the overshoot. During the heating phase of the cycle, the proportional control reduces the average power sent to the heater as it approaches the set-point based on the setting of the proportional band. The tuning of the parameter associated with the proportional band in effect was to obtain the optimum heater on-off time ratio proportional to the difference in the values between the set point and the process input. The three PID settings were adjusted via trail-and-error to achieve a fast control and satisfactory steady-state stability. In this part of the study, the same heating and cooling sequences were adopted as those used in Figure 4. The effect of introducing a PID control to the system can be seen in Figure 7 where, with the adoption of suitable settings for the control value, it was possible to eliminate the large overshoot. The results obtained, however, lead to a relatively modest oscillatory response which could be reduced by further tuning for the proportional band to achieve the desired level of control. Figure 8 shows the variation of the thermocouple reading with POF reading for the series of cycles shown in Figure 7. Compared to Figure 5, it is apparent that the hysteresis loop is moderately smaller due Pg 9
Page 9 of 19
to the elimination of the overshoot. Similarly to Figure 5, the cooling traces for all 3 cycles overlap each other well. The result in the figure again suggests the inherent difficulty in using temperature as the process variable for deflection control of SMA actuators while highlighting the effectiveness of adopting low-cost sensors such
ip t
as POF for accurate monitoring and control of smart FMLs.
4. Conclusions
cr
A novel hybrid laminate based on a nickel-titanium shape memory alloy has been developed and its activation capabilities have been investigated. The results have highlighted the controllability of this structure through the
us
introduction of a POF sensor which was used to monitor the deflection of the beam. This signal was then used as feedback to the controller unit. Tests were also conducted to assess the auto-correcting capability of the
an
hybrid using the POF sensor signal. Here, process perturbations in the form of simple weights, were added to the end of the beam forcing the deflection away from the desired set-point forcing the controller unit to activate the
M
heating element and return the beam to the desired configuration
Tuning of the PID parameters on the controller unit has shown that it was possible to eliminate the large overshoot initially observed resulting in a better control of the beam deflection for the entire duration of the test.
d
It is expected that the oscillatory behavior could be further minimized through further tuning of the PID control.
te
The results have also highlighted that the incorporation of POF sensors into the laminates can successfully
Ac ce p
measure the out-of-plane response of these novel hybrid structures.
ACKNOWLEDGEMENTS
The authors acknowledge the funding (R-534-000-007-414) from the Minerals, Metals and Materials Technology Centre (M3TC) of the National University of Singapore.
Pg 10
Page 10 of 19
References N Tsartsaris, M Meo, M Dolce, U Polimeno, M Guida and F Marulo, “Low velocity impact behaviour of fiber metal laminates”, Journal of Composite Materials Vol.45 (2011) pp.803-814.
[2]
S Krishnakumar, “Fiber Metal Laminates: The synthesis of metals and composites”, Materials and Manufacturing Processes Vol.9 (1994) pp.295-354.
[3]
LB Vogelesang and AVlot, “Development of fiber metal laminates for advanced aerospace structures” Journal of Materials Processing Technology Vol.103 (2000) pp.1-5.
[4]
R Loewy, “Recent developments in smart structures with aeronautical applications”, Smart Materials and Structures Vol.6 (1997) R11-R42.
[5]
DJ Hartl and DC Lagoudas, “Aerospace applications of shape memory alloys”, Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering Vol. 221 (2007) pp.535-552.
[6]
J Kudva, “Overview of the DARPA smart wing project”, Journal of Intelligent Material Systems and Structures Vol.15 (2004) pp.261-267.
[7]
B Sanders, R Crowe and E Garcia, “DARPA-Smart Materials and Structures demonstration program overview”, Journal of Intelligent Material Systems and Structures Vol.15 (2004) pp.227-233.
[8]
V Birman, “Review of mechanics of shape memory alloy structures”, Applied Mechanics Reviews Vol.50 (1997) pp.629-645.
[9]
K Otsuka and CM Wayman (Eds.) Shape memory materials, Cambridge University Press, Cambridge, 1999.
[10]
M Meo, F Marulo, M Guida and S Russo, “Shape memory alloy hybrid composites for improved impact properties for aeronautical applications” Composite Structures Vol.95 (2013) pp.756-766.
[11]
S Pappada, R Rametta, A Largo and A Maffezzoli, “Low-velocity impact response in composite plates embedding shape memory alloy wires” Polymer Composites Vol.33 (2012) pp.655-664.
[12]
P Cortes, WJ Cantwell and KSC Kuang, “The fracture properties of a smart fiber metal laminate”, Polymer Composites, Vol.28 (2007) pp.534-544.
[13]
G Song, B Kelly and B N Agrawal, “Active position control of a shape memory alloy wire actuated composite beam” Smart Materials and Structures Vol.9 (2000) pp.711-716.
[15]
[16] [17]
cr
us
an
M
d
te
Ac ce p
[14]
ip t
[1]
J A Balta, F Bosia, V Michaud, G Dunkel, J Botsis, JA Manson, “ Smart Composites with embedded SMA actuators and fiber Bragg grating sensors: activation and control” Smart Materials and Structures Vol.14 (2005) pp.457-465. M Mieloszyk, L Skarbek, M Krawczuk, W Ostachowicz and A Zak, “Application of fiber Bragg gratings sensors for structural health monitoring of an adaptive wing”, Smart Materials and Structures Vol.20 (2011) p.125014 (12pp) KSC Kuang, ST Quek and M Maalej, “Assessment of an extrinsic polymer-based optical fiber sensor for structural health monitoring” Measurement Science and Technology Vol.15 (2004) pp.2133-2141.
KSC Kuang, M Maalej and ST Quek, “An application of a plastic optical fiber sensor and genetic algorithm for structural health monitoring” Journal of Intelligent Material Systems and Structures Vol.17 pp.361-379.
[18]
KSC Kuang, CY Tan, SH Chew and ST Quek, “Monitoring of large strains in submerged geotextile tubes using plastic optical fiber sensors” Sensors and Actuators A: Physical Vol. 167 pp.338-346.
[19]
KSC Kuang, ST Quek and M Maalej, “Plastic optical fiber sensor for damage detection in offshore structures” Proceedings of the SPIE Vol.7522 (2010) art. no. 75223A.
[20]
KSC Kuang and WJ Cantwell, “The use of plastic optical fibers and shape memory alloys for damage assessment and damping control in composite materials” Measurement Science and Technology Vol. 14 (2003) pp.1305-1313
Pg 11
Page 11 of 19
Illustrations
sensor gauge length
adhesive
2.2 mm
outer jacket light in
ip t
80 mm
housing
cr
light out
ID 1.06 mm
substrate
1 mm bare POF
us
OD 2.0 mm
Ac ce p
10.0
te
d
M
an
Figure 1 Schematic of the POF sensor used in the study.
Extension (mm)
8.0
POF reading Curve fitting
6.0
4.0
2.0
0.0 0.0
0.1
0.2
0.3
0.4
0.5
0.6
Normalised POF Loss
Figure 2 Plot showing the response of POF sensor to externally applied extension
Pg 12
Page 12 of 19
ip t cr us an
M
Figure 3 Schematic of the experimental test set-up.
te
0.09 0.08
POF Sensor
90
Ac ce p
Thermocouple
0.07
80 70
(C)
0.06
60
0.05
50 (B)
0.04
40
0.03
30
0.02
20
(A)
0.01
Temperature oC
Normalised POF Reading
100
d
0.1
10
0
0 0
500
1000
Time (seconds)
1500
2000
Figure 4 Plot showing the controllability of the smart FML beam using POF sensor.
Pg 13
Page 13 of 19
0.1
1st Cycle 2nd Cycle
0.08 0.07
ip t
3rd Cycle Position (C)
0.06
cr
0.05 Position (B)
0.04 0.03 0.02
us
Normalised POF Reading
0.09
Position (A)
an
0.01 0 20
40 60 Temperature (oC)
80
100
M
0
d
Figure 5 Plot showing the thermomechanical response of the smart FML.
Pertubation introduced here Weights added at free end of beam
0.06
More weights added at free end of beam
80 Controlled position of beam
70 60
0.05
50
0.04
40
0.03
30
0.02
POF Sensor Thermocouple
Weights removed from free end of beam
0.01
Temperature (oC)
Ac ce p
0.07
Normalised POF Reading
90
te
0.08
20 10
0
0 0
100
200
300
400
500
Time (sec) Figure 6 Plot showing the self-correcting response of the SMA beam based on the POF sensor signal.
Pg 14
Page 14 of 19
0.1
100 POF Sensor Thermocouple
0.08
90 80
0.07
70
ip t
(C)
0.06
60
0.05
cr
(B)
us
0.04 0.03
an
0.02 0.01 0 500
1000 Time (sec)
M
0
(A)
50 40 30
Temperature (oC)
Normalised POF Reading
0.09
20 10
1500
0 2000
te
d
Figure 7 Plot showing the response of the beam using PID control to reduce the overshoot observed previously.
0.1
1st Cycle
Normalised POF Reading
Ac ce p
0.09
2nd Cycle 3rd Cycle
0.08 0.07
Position (C)
0.06 0.05
Position (B)
0.04 0.03
Position (A)
0.02 0.01 0 0
20
40 Temperature (oC)
60
80
Pg 15
Page 15 of 19
Ac ce p
te
d
M
an
us
cr
ip t
Figure 8 Plot showing the thermomechanical response of the smart FML under PID control.
Pg 16
Page 16 of 19
Paper: Active Control of a Smart Composite with Shape Memory Alloy Sheet using a Plastic Optical Fiber Sensor Ms. Ref. No: SNA-D-13-00077
ip t
Figures Caption List Figure 1: Schematic of the POF sensor used in the study
cr
Figure 2 : Plot showing the response of POF sensor to externally applied extension Figure 3: Schematic of the experimental test set-up
us
Figure 4: Plot showing the controllability of the smart FML beam using POF sensor Figure 5: Plot showing the thermomechanical response of the smart FML
an
Figure 6: Plot showing the self-correcting response of the SMA beam based on the POF sensor signal
M
Figure 7: Plot showing the response of the beam using PID control to reduce the overshoot observed previously
Ac ce p
te
d
Figure 8: Plot showing the thermomechanical response of the smart FML
Pg 17
Page 17 of 19
Authors Biography
cr
ip t
Kevin S. C. Kuang is an Assistant Professor at the Department of Civil and Environmental Engineering and is active in the research areas of smart structures, structural health monitoring and st sensors technology. He obtained his PhD and BEng (1 Hons) degrees from the Universities of Liverpool and Leeds, United Kingdom respectively. He has commercialized an optical fibre sensor educational kit and has a patent granted on optical fibre sensor (U.S. Patent Number 7496247). He has published over 40 technical papers and has been invited to give talks in the Institute of Engineers Singapore, SPRING Technical Conference, and Singapore Science Centre. He was also an invited speaker at the 20th IPOF Conference in Bilbao (Spain) 2011 and the JEC Innovative Composites Summit 2011.
M
an
us
Ser-Tong Quek is currently Professor and Vice Dean of Graduate Programmes and Director of the Global Engineering Programme in the Faculty of Engineering and Professor in the Department of Civil and Environmental Engineering at the National University of Singapore (NUS). He completed his Ph.D. in 1986 at the University of Illinois at Urbana-Champaign, Illinois, under Professor Alfredo Ang before returning to NUS to resume his academic career. His research encompasses structural health monitoring, structural reliability and mechanics and innovative structural composites. He has published more than 150 international journal papers. He was a recipient of Institution of Engineers Singapore KYD Gin Prestigious Publication Award and Defense Technology Prize for Research in 1990 and 2006, respectively.
Ac ce p
te
d
Wesley Cantwell is a Professor of Aerospace Engineering at Khalifa University of Science Technology and Research in the United Arab Emirates where he is on leave from the University of Liverpool. Wesley joined Khalifa University in October 2012 and is the Director of the Aerospace Research and Innovation Center (ARIC). In Liverpool, he leads a group of ten researchers investigating the fracture properties of lightweight materials and structures for use in the aerospace industry. Much of his research focuses on developing sandwich structures for energy absorption, understanding the blast response of composite materials, characterizing novel metal composite hybrids and producing and characterizing smart aircraft structures.
Pg 18
Page 18 of 19
Highlights (4 highlights; 85 characters max. per highlight)
pt
ed
M
an
us
cr
ip t
SMA sheet was integrated to FMLs and activated using a thin silicon heater POF sensor allows control of the FML deflection accurately and good repeatability POF data is used as PV to control the deflection of FML instead of temperature. Tests showed set deflections for FML could be achieved accurately using POF sensor.
Ac ce
Page 19 of 19