DE Actuated Automotive HVAC Louvers

Transcription

1 DE Actuated Automotive HVAC Louvers Nick M Manzek Tizoc Cruz-Gonzalez Jonathan Luntz GM/UM Smart Materials and Structures Collaborative Research Lab Diann Brei ABSTRACT Automotive manufacturers are always looking to add features to their vehicles to improve the user experience. Being able to electronically actuate the HVAC vent louvers can allow the user to have total control of air flow as well as increase the system efficiency by better distributing air flow throughout the cabin. This actuation cannot be achieved using traditional actuators due to space and energy constraints. Smart materials such as dielectric elastomers (DEs) are being explored for this application. A DE is an electrostatic actuator based on a thin silicon film under tension with carbon nanotube electrodes over its surface. When high voltage is applied to the electrodes the film is compressed in thickness expanding in-plane. This type of actuation allows DEs to produce large strokes in compact and energy efficient packages. This paper presents the design process and testing of a DE controlled HVAC vent. A theoretical model of the system was created by characterizing DEs based on the Gent Strain Energy Model for hyper elastic materials coupled with a kinematic model of the existing HVAC vent. Using this model, design parameters were manipulated to create a system that maximized the angular range of motion of the louvers within the physical limits of the DEs. From this design, a physical system was built, tested and compared to the predicted results of the theoretical system. The physical system was able to produce a stroke that corresponded to 70 degrees of motion. This proves that a high degree of air vent control can be achieved with a design that falls within size constraints and is energy efficient.

2 PURPOSE Being able to actively control automobile HVAC louvers provides significant advantages for automobile manufacturers. Being able to control where the air flow is going can increase the energy efficiency of the HVAC system by being able to target hot spots within the vehicle. It can also improve the user experience because features such as sweeping air flow can create a more relaxing environment for the passengers. Unfortunately, traditional actuators cannot be used because they are too expensive and will not fit within space constraints. In addition, traditional actuation uses too much power and can create electromagnetic interference for the other systems in the vehicle. Smart materials such as dielectric elastomers (DEs) are being explored to actuate HVAC vents because they are inexpensive, come in a small package, use very low power and create no electromagnetic interference. DE OPERATION A DE is a squishy stretchy capacitor made from an elastomer with two compliant electrodes on the top and bottom. When high voltage is applied to the electrodes, the difference in charge attracts the bottom surface to the top surface. A Maxwell stress is created and this stress causes the DE elastomer to compress in thickness and expand in plane. The behavior of a DE in the force length space follows a hyperelastic curve (Fig. 1). There are three regions: and initial stiff region, plateau region and a final stiff region where the DE reaches the hyperelastic limit. When an external load of constant force is applied and the DE is turned off it reaches a voltage off equilibrium position. When the voltage is turned on, the DE lengthens and a new equilibrium position voltage on equilibrium is found. By turning the voltage on an off, the DE can switch between these positions and create a large stroke. However, in practice it is hard to apply a constant force. A spring can be used as a bias force except due to the stiffness some of the stroke is lost. Figure 1: Stroke prediction using constant force. 1

3 Antagonistic Design Instead of a spring, we can use another DE actuator to create the bias force against the first actuator. When actuator 1 is turned on and actuator 2 is off, equilibrium position 2 is reached. When the voltage is flipped, equilibrium position 3 is reached (Fig. 2). Cycling between these two positions creates a stroke similar in size to a stroke obtained using constant force. However, this does not utilize the full length of the hyperelastic curve. Figure 2: Stroke prediction using antagonistic design. Reverse Bias Design To utilize the full length of the hyperelastic curve, a reverse bias design features two actuators that pull against each other through a mechanism. The moment arms that the DEs pull on change as the mechanism moves through it motion. The change of the moment arm is designed to create more leverage when it is needed. This design takes the old curves from the antagonistic design and shifts them to make full use of the hyperelastic curves (Fig. 3). 2

4 Figure 3: Stroke prediction using reverse bias design. DESIGNING WITH DEs Kinematic Model of HVAC Vent The key to a reverse bias design is the moment arms. To design properly, a mechanism model was created to predict the size of the moment arms as the mechanism moves through its motion. Figure 4 shows the physical geometry of the HVAC louvers with l "# and l 2θ being the lengths of the moment arms of actuators AB and CD, respectively. Using the equations in figure 5, a model can be created to relate the length of the two DEs to the louver angle. Since AO, BO, CO, and DO are fixed lengths due to the vent geometry, the length of the actuators is a direct result of the angle of the vent louvers ( θ ). At any given value of θ, the length of both actuators is known and therefore the force can be calculated using the the Gent Strain Energy Model. This model relates the force of a DE to its length, whether it has no voltage or a positive voltage applied. Figure 4: Diagram of louvers. Figure 5: Mechanism equations 3

5 Gent Strain Energy Model In the force length space, DEs follow a hyperelastic curve which can be modelled using the equation in figure 6. The left side of the equation contains the terms that relate to the stretch ratios of the elastomer. The right side of the equation includes the terms that relate to the Maxwell stress that occurs when a voltage is applied. As the voltage is applied the curve is shifted down (Fig. 7), due to the right side of the equation. λ " = l L where l is the active length of the DE and L is the initial length, both in meters. W is the width of the DE in meters. The thickness of the DE is t + in meters. μ(v) is the shear modulus in Pa. λ 0 is the transverse stretch ratio of the DE. V is the voltage being applied. J 2 is the asymptotic stretch limit. ε is the permittivity of vacuum. Figure 6: Gent Strain Energy Equation. Figure 7: Hyperelastic curves of DE, voltage on (3100V), voltage off (0V). 4

6 Contour Plots Using the geometry and material model, we can predict the equilibrium positions over a range of design variables. Given space constraints, we were only able to easily change the position of the DEs attached to the HVAC vent walls. Figure 8 shows the equilibrium position that can be achieved when actuator 1 is on and actuator 2 is off. By moving the attachment points (y1,y2), different equilibrium points, as function of the louver vent angle can be reached. With y1=6mm and y2=6mm and equilibrium position of about 75 degrees can be reached. Coupling this plot with a similar one, when actuator 1 is off and actuator 2 is on, can be used to narrow down the design variables to create a design with favorable stroke. Figure 8: Contour plot showing rotation angles that can be achieved by varying y1 and y2. Torque Theta Plot Once a design is picked, we can examine it further considering the torque theta space instead of the force length space. Figure 9 shows the equilibrium position of the two actuators. Moving between positions 2 and 3 can create a stroke that creates a louver angle sweep of 79 degrees. 5

7 Figure 9: Predicted stroke using design variables chosen from the contour plots. MANUFACTURING PROCESS Dielectric Elastomer Actuator After a design was chosen, the DEs were made. First, Silpuran 6000/10 part A and B were mixed. 2.5mL of the mixture was dropped onto a 2 inch by 2 inch slide of glass. The slides were air dried for 48 hours and then cured at 175 degrees Celsius for 2 hours. After the silicon films were cured, they were peeled off the glass slides and stretched onto a manufacturing frame which puts 100% axial and 150% transverse pre stretch into the silicon. Plastic masks were placed on the top and bottom of the film to block out the area that would become the electrodes. Carbon nanotubes were sprayed on the masks to create the active area of the DE. Conductive leads and double sided Kaptan tape were added to create the external electrical and mechanical connections. The DE actuator was then released from the frame and tested for proper function. The average stroke for a properly working DE is about 2-3mm. Figure 10: Parts of DE actuator. 6

8 HVAC Vent Assembly Once the DEs were made, they were attached to the vent. First, the 5 fins were removed from the vent enclosure to gain easy access to the attachment points. Within the vent enclosure, lines were drawn to specify the exact attachment points of the DE on the vent wall. Each DE was required to be a certain distance from the axis of the fin to meet the specifications of the chosen design. To accomplish this, calipers were used to mark the fin where the DE was supposed to be attached. This allowed for an easy and accurate application process. The Kaptan tape was that used to create the structure of the DE came with a plastic coating to protect the adhesive from dust and debris during storage of the DE. This coating was removed and the DEs were attached to the fins on their respective sides. Once both DEs were attached to the fins, they DE-fin assembly was slid back into the vent enclosure and snapped into place. Lastly, the other sides of the DEs were attached to the walls of the enclosure, finalizing the assembly. Figure 11 demonstrates that the vents are hardly noticeable and do not block the air flow through the vent. Figure 11: HVAC vent with DEs attached. PHYSICAL TESTING Louver Angle Sweep A video was taken of the louver motion obtained by switching between the equilibrium positions using the two DEs. Two video frames, at the leftmost and rightmost point of motion were analyzed further to determine total stroke. Using optical software to count pixels and translating those into distance measurements, it was calculated that the stroke produced a louver angle sweep of 17 degrees. This is significantly less than the models predicted. When the vent louvers were moved by hand, with the DEs attached, they would stick in positions that the models did not predict as equilibrium positions. 7

9 Design Variable Sensitivity Analysis One possible explanation for the lower than expected stroke was the accuracy to which the prototype was manufactured. Working in the small spaces of the vent enclosure made it hard to attach the DEs with high accuracy. Upon measuring the actual build attachment points, it was determined that the build only had an accuracy of only ±1mm. Figure 12 is the torque theta diagram of the initial design. By changing the model attachment points by ±1mm, the torque theta diagram changes to that in figure 13. It is evident that this design is very sensitive to the accuracy of the manufacturing process, which is why 29 degrees of the intended louver angle sweep was lost. That is loss of 37%. Figure 12: Predicted stroke of initial design Figure 13: Predicted stroke with design variables changes by ±1mm Vent Mechanism Friction Analysis Another explanation for the decreased stroke was friction within the fin mechanism. When one actuator is contracting it must overcome the force of the other actuator as well as the force of friction within the mechanism. To determine whether the vent mechanism had enough friction to decrease the stroke significantly, an experiment was performed. The vent was rotated 90 degrees and held stationary in a vertical position. String was attached to one of the fins and then attached to a load cell above the vent. String was attached to another fin and weights were hung to vary the force being put on the mechanism. The load cell was hooked up to a power source and an oscilloscope to capture its readings. By moving the fins through their motion the force of friction can be plotted (Fig. 14). The difference between the upper and lower peaks corresponds to two times the force of friction of the mechanism. This is calculated to be equivalent to a 3x10-4 N-m torque on the mechanism. In figure 16, the +3100V curve for actuator 1 is shifted up by a 3x10-4 N-m torque and the 0V curve for actuator 1 is shifted down by a 3x10-4 N-m torque, due to the friction of the mechanism. Friction has decreased the louver angle sweep by 19 degrees which is 24% less than the designed stroke in figure 15. 8

10 Figure 14: Friction of louver mechanism through its motion. Figure 15: Predicted stroke without friction. Figure 16: Predicted stroke with friction, arrows show shift in curves due to friction. Revised Design To try to reduce the effect that poor manufacturing accuracy and friction have on the louver angle sweep, a new design was created. Figure 17 depicts the new torque theta curves. While the new design predicts a stroke of only 41 degrees, it is much less sensitive to manufacturing and friction. When friction is incorporated in the model the louver angle sweep is only decreased by 22% which is 9 degrees of motion (Fig. 18). When the DE attachment points are changed by ±1mm, the louver angle sweep is only decreased by 2 degrees which is only a loss of 5% (Fig. 19). 9

11 Figure 17: Torque vs. theta curves of revised design. Figure 18: Revised design with friction, arrows show shift in curves due to friction. Figure 19: Sweep with manufacturing tolerances 10

12 CONCLUSIONS AND RECOMMENDATIONS DE actuators can be paired with automotive HVAC vents in a very attractive package. They do not block air flow and use very little energy. They are also easily integrated into the structure of the vent and can be attached with simple adhesive tape. However to gain the desired amount of stroke, the specific characteristics of the mechanism (such as friction) need to be carefully considered during the design process. Because the difference between the on and off hyperelastic curves of the DE is so small, the stroke is very easily affected by friction. Small shifts in the curves due to friction cause large changes in stroke. Within the vent enclosure its hard to attach the DEs accurately. One way to mitigate this is to attach the DEs to a thin structure on the inside that can be moved along the walls of the enclosure to fine tune the placement to make sure it coincides with the design specifications. One way to remove a significant amount of the friction from the mechanism would be to create a solely antagonistic design where the DEs are attached to just one fin of the louver mechanism (Figure 13). This isolates the middle fin and does not create unwanted friction because it does not put unwanted forces on the other fins and allows them to move more freely. A design similar to this could create the largest stroke with a high chance of success. Figure 13: Proposed DE placement to reduce friction. 11