Design and fabrication of an actuation module using integrated pneumatic technology

In this thesis, xurography techniques are used to create an integrated pneumatic card (IPC) by integrating fluidic channels inside the electronic circuitry of their control system. In the simplest design of an IPC, there are three main layers: the top layer made of printed circuit board (PCB), the middle layer made of double-sided tape with air channels xurographically cut in it, and a bottom layer of PCB. Flow enters through an opening in the top layer of PCB and is routed through valves and to outlets in the bottom PCB by the air channels in the tape. The surfacemount valves are attached pneumatically and electrically to the first layer of PCB. By using PCB as the substrate, the circuitry for controlling the device can be integrated directly onto the same material as the valves. A prototype was device created for testing purposes to verify that the device could be operated within specifications. It consists of an air inlet and eight three-way valves that route air to four bladders, three pinch points and one flow-through channel. The dimensions of the entire device measure 2.4 x 3.6 x 1 in excluding the protruding wire attachment. The modeling of this device was done using a one-dimensional flow assumption and was used to predict the time constant to inflate a bladder. Key calculations of this model include: an equation to calculate bladder pressure as a function of time and flow rate, and an equation for flow rate as a function of pressure. Also, equations to predict the acceleration of air in the channel were developed. Data for channel expansion due to expansion of the acrylic adhesive are used directly in the model. These calculations were made iteratively using a spreadsheet. The results were compared to experimental results and shown to be an adequate representation of reality.

PNEUMATIC TECHNOLOGY by John Nathan Maxwell The University of Utah partial fulfillment of the requirements for the degree of December 2007 DESIGN AND FABRICATION OF AN ACTUATION MODULE USING INTEGRATED PNEUMATIC A thesis submitted to the faculty of in Master of Science Department of Mechanical Engineering University of Utah Copyright © John Nathan Maxwell 2007 All Rights Reserved T H E U N I V E R S I T Y U T A H G R A D U A T E S C H O OL APPROVAL satisfactory. THE UNIVERSITY OF UTAH GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a thesis submitted by John Nathan Maxwell This thesis has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory. k;0~ K.L De ries /O/I/b2 ; ? Kuan Chen T H E U N I V E R S I T Y U T A H G R A D U A T E S C H O OL F I N A L R E A D I N G A P P R O V AL I have read the thesis of John Nathan Maxwell in i t s f m a l f o r m 7. MA/ 2077 Supervisory Committee Department Chapman THE UNIVERSITY OF UTAH GRADUATE SCHOOL FINAL READING APPROVAL To the Graduate Council of the University of Utah: in its final form and have found that (1) its format, citations, and bibliographic style are consistent and acceptable; (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the supervisory committee and is ready for submission to The Graduate School. 2 Ilbv 2{)07 Date Chair: SupervisetY Committee Approved for the Major Department Approved for the Graduate Council David S. Chapm Dean of The Graduate School ABSTRACT surface-mount It results were compared to experimental results and shown to be an adequate representation of reality. In this thesis, xurography techniques are used to create an integrated pneumatic card (IPC) by integrating fluidic channels inside the electronic circuitry of their control system. In the simplest design of an IPC, there are three main layers: the top layer made of printed circuit board (PCB), the middle layer made of double-sided tape with air channels xurographically cut in it, and a bottom layer of PCB. Flow enters through an opening in the top layer of PCB and is routed through valves and to outlets in the bottom PCB by the air channels in the tape. The surface­mount valves are attached pneumatically and electrically to the first layer of PCB. By using PCB as the substrate, the circuitry for controlling the device can be integrated directly onto the same material as the valves. A prototype was device created for testing purposes to verify that the device could be operated within specifications. consists of an air inlet and eight three-way valves that route air to four bladders, three pinch points and one flow-through channel. The dimensions of the entire device measure 2.4 x 3.6 x 1 in excluding the protruding wire attachment. The modeling of this device was done using a one-dimensional flow assumption and was used to predict the time constant to inflate a bladder. Key calculations of this model include: an equation to calculate bladder pressure as a function of time and flow rate, and an equation for flow rate as a function of pressure. Also, equations to predict the acceleration of air in the channel were developed. Data for channel expansion due to expansion of the acrylic adhesive are used directly in the model. These calculations were made iteratively using a spreadsheet. The TABLE OF CONTENTS 1 1 Microfluidics 3 Experiment # 1 - Time Constant Determination 18 Experiment #2 - Fatigue Testing 24 25 28 28 30 Contact 35 37 Spreadsheet 37 Results 39 39 of Model 40 Recommendations and 41 ABSTRACT .... ....... .. ............................ .......................... ............ ............. ..... .... ..... ...... .. .... .. ....... ..... iv ACKNOWLEDGMENTS ... ... ... ....... ....... .. .... ... .......... ..................... ................ ............ ... ... ... ......... vii Chapter I - INTRODUCTION ... .............. .. ...................... .. ....... .. ...... ............................................ ...... I Scope of Work ........ ... ................ ......................... ..... ..... .... ..... ......... .... .......... .............. ... 2 Microfluidics ....... .. ... .. ...... .. .............................................................................. .... ......... 3 Printed Circuit Board (PCB) and Fluidics ..................................................................... 4 Xurography .................................................. .. ........................................... ..... ..... .... ....... 6 Overview of Thesis ........... ... .................................. .. .... .......... .. ... .. ...... .... ... .................... 7 2 - DESIGN AND CONSTRUCTION ................. .... .... .. ........................................... .. .... ...... 8 Layer by Layer Design ...... .......................................................................... .. ....... ......... 8 Bladders ............. ................................. .. .. .................. ......... ...... .. ... ... ... ... ....... ......... ...... 15 Electrical Connections ........... ......... ............... .... ..... ....... ... ... ..... ..... ..... .. .. .. ........ ...... ..... 16 Conclusion .................. .... .................................... ... ... ......... ... ... ...... ..... .... .. .. ...... ..... ...... 16 3 - TESTING AND RESULTS ....... .. ........................ ... .... ................... .. ................. ....... .. ..... 18 ...................... .... ............... .. ... ....... ..... ......................................... .. .............. ..... .. .... ..... .... .... Experiment #3 - Tape Expansion .............................................. .... ...... ..... ........... ........ 25 Conclusion ........ ...... .......... ...... .................................... ... ... ... ...... .. ........... ... ... .... ... ...... .. 27 4 - MODELING ........ ............... ........ ..... ............. .......... ...... ..... .......... ...... ... ... ............... ........ 28 Pressure Derivation .... ....... ....... .. ..... ............ .......... ................ ............. .. ........................ 28 Velocity Model .... ......................................... .... ........................ ... ...... .. ............... ......... 30 Area of Contact. ....... ............................ ...... ..... ... .... ........ .. .......................... ......... .... .. ... 35 Channel Expansion .. ........ ... ......... ... ....................................... ...... ............. .. ......... ..... ... 37 Overview of Spreadsheet. ....... ....... .. ............................... ........ ... ..... ... ..................... ... .. 37 Results and Analysis of Velocity Model .............. ............................... .......... .. ............ 39 Acceleration Model .......... ... ..................................................... ....... .. ........ ... ....... ... ..... 39 Results and Analysis of Acceleration ModeL ........... ....... .... .... ...... .. ...... .. ...... .. .......... 40 Recommendations and Conclusion ........................ ... ............. ................. .. .. ................. 41 5-CONCLUSION EXPERIMENTAL REFERENCES 50 vi 5 - CONCLUSION ........ ... .. .. ............... ................... ............... .. ........ ................... .. ......... ..... .. .... 42 Improvements .. ................................ .... ..... .... .... ...... .... .................. ...... .................... ..... 42 Future Work ................. ..... ....... ... .. .. ... ... .... ....... ... .............. ... ... ............ .. .... ... ................ 44 Summary ....................................... ..... .. ...................... ................. .. .................... ..... ...... 47 APPENDIX: TIME CONSTANT EXPERIMENT AL DATA ............ .. .. ..................................... 48 REFERENCES .. ..... ....... .. ......... ... .. ....... ........ ..... .... ........ ..... ..... .......... ...... ....... ...... .. ... .. .. ......... ..... .. 50 Vi challenge and for somehow making me equal to the task. ACKNOWLEDGMENTS Many people are deserving of sincere thanks for their help with this work. I would first like to thank Kirk Ririe and Idaho Technologies for their funding and input on this project. Appreciation is due to Dr. Bruce Gale for many reasons, some of which include directing me to the master's program, giving me guidance along the way, and most of all for taking a chance on a kid with a business degree and little engineering background. I am grateful to the Mechanical Engineering Department at the University of Utah for their constant efforts to recruit and retain not only great researchers, but also excellent teachers and staff. Professors like Dr. Larry DeVries, Dr. Tim Ameel, Dr. David Hoeppner and Dr. Raymond Cutler who take pride in their teaching do a tremendous job of stretching the abilities of students and I am grateful to them for their influence on me and those around me. John Brady was a tremendous help in getting this project started and building many of the early prototypes. Without his help completion of this project would have been much more difficult. I would like also to thank others who worked with me and around me, particularly Himanshu Sant and Scott Sundberg for a great deal of help setting up valid experiments and giving of their time to help me work through models. These acknowledgments would mean nothing if I did not thank my wife Kristen who dared me to do something I was afraid of doing and cheered me on through every success and every failure. Without your support, I never would have made it, or even started it. Finally, I would like to thank God for believing in me enough to let me struggle through such an excellent CHAPTER 1 biocard CHAPTER} INTRODUCTION In pneumatic applications today, systems are often designed which have need for several hoses and valves which route air pressure to specific locations. In many of these applications, the network hoses and valves can take up a great amount of space and can become cumbersome especially when compact geometries such as hand-held devices are desired. A pneumatic system of this sort, created by Idaho Technologies as part of a portable biosample analysis system, is often carried in soldiers' backpacks, so reductions in volume are greatly beneficial. The device contains a pneumatic system of hoses and valves that are used to route air to an array of bladders and pinch points mounted on an actuation layer as shown in Figure 1. The valves in this system are controlled with nearby electronic circuitry and the three part system (actuation layer, pneumatics, and electronics) is used to control fluid flows in an adjacent biocard. if) u 'c o I... -t) Q.) Q.) Figure 1. A sample pneumatic device controlled by electronics, pneumatics, and an actuation layer 2 Scope of Work in detail in Chapter 2. An inflated bladder on the actuation layer compresses a bladder on the biocard which moves the biosample through a channel to next bioanalysis region where another testing agent can be added. The channels in the biocard are sealed off during each test by the pinch points, which simply apply a pressure to the channel, thus sealing it. The goal of the present research is to integrate the actuation layer, pneumatics and electronics all into one compact device which will perform the same functions as its predecessor. The device is given a 20 psi input pressure and must be able to inflate bladders at 15 Hz so two or more bladders can be used to perform mixing operations by moving a fluid back and forth at 15 Hz. In this thesis, xurography techniques are used to create an integrated pneumatic card (IPC) that integrates fluidic channels inside a printed circuit board (PCB) substrate. Valves are mounted to the PCB alongside the controlling circuitry and this comprises the pneumatics and electronics portion of the device. The actuation layer is created by stacking other substrate layers to the pneumatics and electronics portion of the device. In the simplest design of an IPC, there are three main layers which consist of the top layer made of PCB, the middle layer made of double-sided sticky tape with air channels xurographically cut in it, and a bottom layer of PCB. Flow enters through an opening in the top layer of PCB and is routed through valves and outlets in the bottom PCB by air channels in the tape. The surface-mount valves are attached pneumatically and electrically to the first layer of PCB as shown in Figure 2. Together these components create a total thickness of about 1.5 cm (0.58 in) and comprise the electronics and pneumatics portions of the device. The actuation layer is attached to this device using an additional tape layer as shown in Figure 3 and will be discussed 3 Valves electronics 1 Air Inlet Air Outlets f Figure 2. A simple integrated pneumatic card ui o 3 pneumatic Figure 3. A complete device consisting of an IPC stacked onto an actuation layer and biocard. Because the IPC is easily customizable, implications of its design go beyond solving one company's design requirements and can be applied to many systems in which simplification of valve and hose networks would create a simpler, more user friendly product. For example, many liquid handling or biotechnology instruments are designed for working with hundreds or thousands of samples simultaneously and typically require complex hose assemblies and connections. This design could quickly eliminate many of the challenges associated with high-throughput pneumatic systems. Microfluidics This design employs a less conventional application of microfluidics by using xurographically cut fluidic channels to direct air flow, while most microfluidic applications are designed to perform complicated and precise mixing for medical, chemical and biological analytical procedures [1]. In most of these more common applications, one key benefit of the - ~ r PCB T~ ~13 pneumati .. electronics 'I ~~---------- biocard Microfl uidics non-reactive IPC. The machinability of PCB using traditional machining techniques such as drilling or sawing also microfluidics is that the small volumetric flow rates are easier to control with precision. Use of microfluidics in this device was not based on these generally accepted scaling advantages for microfluidics, but was used because of advantages such as volume reduction and relative ease of manufacture. 4 Prior microfluidics designs have used many different materials and techniques to create channels. Early designs were done by etching channels into silicon using traditional micromachining techniques. However, the high costs of silicon make it a poor choice for microfluidics, which generally require a much larger surface area than traditional MEMS devices. Glass is a traditionally preferred material for chemistry applications because it is non­reactive and transparent. Methods for creating glass microfluidics have been developed and are commonly used in many applications today. However, due to the brittle nature of glass, attaching valves to it or using it in a hand-held device creates several obvious problems. Polymers such as polydimethylsiloxane (PDMS) have also been used in microfluidics, but their manufacturing methods are quite slow and still being established. Printed Circuit Board (PCB) and Fluidics A less common microfluidic substrate is printed circuit board (PCB), which is most commonly used to connect components in electronic devices. PCB consists of an electrically insulating substrate which is covered on both sides by a conductor such as copper. In a common processing technique, copper is patterned and etched away leaving only traces of copper on the surface, which act as wires connecting electrical components such as transistors, resistors, and capacitors, as shown in Figure 4. The insulating substrate of the PCB is generally made of either fiberglass or a similar composite material, which makes it very durable and ideal for handheld devices such as the IPc. make it a more desirable substrate for manufacturers than silicon, glass or PDMS. Figure 4. Printed circuit board with attached electrical components The first devices to integrate fluidics into PCB were created in Germany in 1998 [2]. This integration was beneficial because researchers were able to create fluidic systems inside the techniques of the PCB fabrication industry, fluidic channels were etched into the copper layer of one side of the PCB while the circuitry design was patterned onto the opposite side. The side of the PCB with the fluid channels etched in it was then bonded to another layer to seal the channel as shown in Figure 5. of this etched fluidic channel copper, J M ^1 (a) (b) 0 MPa T Figure 5: Assembly of a fluidic channel in a printed circuit board using an applied adhesive. 5 attal ctrical l2]. creat,e PCB, which also externally held the electronic controls for these systems. Using established Several devices have been created which take advantage of the capabilities of integrating PCB and microfluidic technology. These include a resistive heating channel, pneumatic actuator, capacitance channel, bubble detector, valves, micropumps, pH-regulation systems, fuel cells and oceanographic salinity sensors [2-7], which all demonstrate the potential ofthis technology. Another benefit of this technology is that it uses an adhesive so two dissimilar materials can be easily bonded together, which is not the case with many common bonding techniques in t 10 MPa copper approximately Xurography of thin microfluidics. This allows a piece of glass to be bonded to a more machineable substrate so experimental results can be observed visually. 6 One difficulty that researchers have encountered with present designs is that the adhesive bonding causes a slight reduction in the width of the fluidic channel due to the glue resin settling when the two layers are compressed together while the glue sets, as shown in Figure 5(b). This problem is exacerbated by the fact that patterned PCB has surface imperfections which require a 10 MPa (1450 psi) pressure to be applied for channels to seal completely. This problem is further complicated by the fact that adhesive strength decreases significantly as adhesive thickness decreases below the ideal adhesive thickness of approximately 0.02 in (0.5mm) [8]. With an applied pressure of 10 MPa, it can be assumed that most of the adhesive is pressed out from between the contact surfaces, leaving only a very thin adhesive layer with very little strength. This result seems to imply that most of the bond strength in the constructed devices actually comes from the glue resin that has been pressed into the fluid channels as opposed to the glue on the contact surfaces. Therefore, for PCB systems using an adhesive for assembly, the problem of reduced channel width can only be minimized, but cannot be entirely eliminated due to the nature of the adhesive process used. This problem greatly complicates modeling of such devices as the random distribution of displaced adhesive leaves the actual geometry of the cross section of fluid channels as an unknown factor at every point throughout the device. The proposed solution for the adhesion problem is to instead use xurographically cut double sided tape to bond the substrate layers and create the channels. Xurography is a relatively new technique used in the field of micromanufacturing to create different types of microstructures in a variety ofthin films. A knife plotter, which has been traditionally used in the sign industry, is used to cut a computer generated design into films, acting similar to a printer with a razor blade 7 [9]. In microfluidics, these designs are the borders of fluidic channels which are vacated by removing the excess material with tweezers. Double-sided sticky tape with fluidic channels xurographically cut in it provides an ideal alternative to gluing etched PCBs together. While providing all the same benefits of the glued PCB technology, it provides several additional advantages for this application. It not only eliminates the problem of incidental channel filling and difficulty depositing a uniform layer of adhesive on a relatively large area, but the acrylic adhesive on the tape expands as pressure is applied and increases channel width allowing for better flow at higher pressures. Fluid channels are fully sealed with only hand-applied pressure and drying time is only necessary to strengthen bonds, which are already quite strong immediately after initial adhesion. Although a variety of applicable research has been documented which could be applied to create a similar device, to this writer's knowledge no device has been documented that can route air flow to several locations in a minimal size envelope. The present design utilizes the integration advantages of a PCB substrate, while solving many of the issues associated with PCB-based designs by using xurographically cut tape as an adhesive. Overview of Thesis The following chapter of this thesis describes the design and construction of a multilayer IPC containing several bladders and pinch points. Key manufacturing and assembly details are given for PCB machining, xurography, bladders and pinch points. Chapter 3 explains testing procedures used to determine the time constant and actual operating frequencies at different applied pressures. In Chapter 4, the model that was derived to predict losses through the system is presented and compared to the results from experiments. Finally, conclusions from this work will be drawn and future work will be recommended including improvements and other applications of the technology. PCB­based different DESIGN AND CONSTRUCTION IPC,prototype. for PCB layers are explained in the section which discusses layer 1, and xurography procedures discussed only as to their functions. To ensure the accuracy and alignment of fluid channels in the tape with the vias in PCB IPC Displacement CHAPTER 2 CONSTRUCTION This chapter discusses the design and construction of the :WC,prototype. Design requirements are listed in Table 1 and the accomplishment of these goals is shown in later chapters. Each layer of the device is discussed individually with the specific functions of each layer being addressed. Manufacturing techniques for different materials are discussed only in the section on the first layer made of that material. For example, details of manufacturing methods are given in the section dedicated to layer 2, but manufacturing techniques for subsequent layers made of PCB and tape are omitted because the same techniques are used and these layers are Layer by Layer Design layers, a CAD drawing of each part was created in SolidWorks® and the layers were assembled digitally to ensure proper alignment. The corresponding drawings created for each layer were then used directly to cut tape layers and as shop drawings as described below. Table 1: Design requirements for the :WC prototype. Operating Frequency 15 Hz Max Input Pressure 20 psi Min Component Displacement 0.08 in Pinch point force 1.0 Ibf Lifetime 1,000,000 cycles 9 Layer 1 - PCB The first layer of the device is the top layer, which houses the valves and air inlet. In future designs, this layer will also house all the controlling electronics. The layer is made of PCB and contains valve holes for eight valves, two alignment holes, an inlet hole and three vent holes as shown below in Figure 6. Valve holes are 0.04 in in diameter and the five-hole pattern for each valve shown in the figure correspond to the three air ports down the center of the valve, and two guide pins on each side. The dotted line represents the approximate size of one valve. The valves used are SMC S070 solenoid valves available from www, smciisa.com. These are three-way valves designed to be bolted on, but a quick drying epoxy was used instead because the use of glue eliminates the need for bolt holes that would pass through the tape layer which already has limited space available for air channels. Adhesion is improved by roughing the surface of both the PCB and valves using medium-grit sandpaper. The glue is applied directly to the PCB by hand using a fine tip spatula, taking special care to get glue around each hole without getting it inside the hole. A thicker layer (about 0.04 in) of glue is then applied to the remainder of the surface area represented with the dotted line in Figure 6 and the valve is mounted to the Layer 1 - PCB The first layer of the device is the top layer, which houses the valves and air inlet. In future designs, this layer will also house all the controlling electronics. The layer is made of PCB and contains valve holes for eight valves, two alignment holes, an inlet hole and three vent holes as shown below in Figure 6. Valve holes are 0.04 in in diameter and the five-hole pattern for each valve shown in the figure correspond to the three air ports down the center of the valve, and two guide pins on each side. The dotted line represents the approximate size of one valve. The valves used are SMC S070 solenoid valves available from www.smcusa..COJ!}. These are three-way valves designed to be bolted on, but a quick drying epoxy was used instead because the use of glue eliminates the need for bolt holes that would pass through the tape layer which already has limited space available for air channels. Adhesion is improved by roughing the surface of both the PCB and valves using medium-grit sandpaper. The glue is applied directly to the PCB by hand using a fine tip spatula, taking special care to get glue around each hole without getting it inside the hole. A thicker layer (about 0.04 in) of glue is then applied to the remainder of the surface area represented with the dotted line in Figure 6 and the valve is mounted to the Figure 6: Top layer of PCB showing machined holes and etched copper pattern. Layer 2 - Tape 100/mi. 200/mi PCB by inserting the guide pins into their corresponding holes and clamping the valve down, concentrating the pressure on the manifold end of the valve to assure proper sealing. 10 The alignment holes [0.08 in diameter] are in every layer of the device and are used during assembly to line up each layer of the device. This technique employs the use of stainless steel precision fit pins of the same diameter and a finishing drill bit, which machines the holes to a tighter tolerance with the pins. Because of the tight fit of the pins and the difficulty of exactly matching hole locations in each layer, it is a good idea to machine the PCB and Plexiglas layers to the desired size first and then clamping them all together and drilling the guide holes at the same time. Other holes in the layer include the air inlet hole [0.1 in], which has a brass nozzle soldered to it to allow for an external pressure source to attach to the device, and the vent holes [0.08 in], which provide a pressure release for the bladders. The PCB is cut to dimensions using a band saw and finished using a belt sander. Once cut to proper dimensions, holes are drilled as laid out on drawings using a milling machine with a digital readout to ensure precision. A centerdrill must be used as well because the drill bits tend to wobble out of place when machining PCB. A copper pattern is used to create a common connection for ground wires of valves and can be seen in Figure 6. It was created using a solvent resistant masking tape to create a mask on the PCB, then immersing it into a ferric chloride PCB etching solution and using a mixer to accelerate the etch rate. This is the only layer having copper traces left on the PCB; all others have the copper etched away completely. Tape The tape used in this device is the 444PC from 3M which comes in a 4 in wide roll and has a nominal thickness of lOO.um. To improve flow through the device, the tape was doubled onto itself for layers containing fluid channels to create a 200.um thick air channel. The width of pressure loss through the system comes from friction through the channels as discussed in 11 each fluidic channel is 0.06 in and the geometry of the channels is designed to eliminate sharp turns by using a large radius of curvature whenever possible. Designing tape channel layers is not a trivial task and careful consideration must be made to avoid allowing channels to become too close to each other or intersect. For this device, the location of outlet holes was constrained by the desired location of bladders and pinch points. It is not always possible to include all fluid channels in only one tape layer without creating excessively long channels. Long channels should be avoided in design because about 40% of Chapter 4. For that reason, most of the fluid channels were created in layer 1, but it was necessary to also include an additional channel in layer 4. The design of this tape layer is shown in Figure 7. Valve holes 1-3 shown in the figure correspond to the holes from layer 1 and by inspection of the other 7 valve locations it can be seen that valve hole 1 is connected to the inlet pressure in each case. Hole 2 of each valve is connected by a channel to a hole through layer 3 which leads to an outlet. Valve hole 3 is connected to channels that lead to pressure release holes in other layers (one of the alignment holes is used as a vent in this design). Figure 7: First tape layer (layer 2) with fluid channels xurographically cut in it and showing several marks from layer 1 for comparison purposes. SolidWorksVi 0.01mm, default 4 5 Q uncut portions 12 When a valve is in rest position, valve hole 1 is closed and holes 2 and 3 are connected. If there is pressure in the channel connected to hole 2 it will be released through hole 3 to a vent. When the valve is actuated, holes 1 and 2 are connected, allowing input pressure to reach bladder outlets in subsequent layers. A Graphtec Cutting Pro FC5100-75 sign cutter is used to xurographically cut the tape to desired dimensions. The sign cutter operates using Adobe Illustrator, but drawings from a CAD program such as SolidW orks® can be exported as a .dxf file and imported into Illustrator. A tape guide was made to hold the tape during cutting using 8V2 by 11 in sheet of clear plastic and cutting a rectangle out of its center large enough to leave about Vz in of extra tape on all sides of the cut tape. The sign cutter settings are set to: Condition: 5, Speed: 1, Force: 8 (single sheet) or 12 (double-thick), Quality: 1, Step Size: O.Olmm, and the rest to the default settings. A 45Q blade was used and the remaining steps follow the OEM's standard operating procedures. The sign cutter only cuts the geometry borders into the tape so the excess material must be removed using precision tweezers. A great aide in assembly is to leave uncut portions in the tape as shown in Figure 8. This allows for some excess tape outside the patterned portion of the tape to be used as hand-holds during alignment. This also allows the tape backing to be removed completely so the transparency of the tape can be exploited to visually align the tape to the stainless steel guide pins which have already been inserted into layer 1. Also, removing the backing actually allows the guide pins to pass through their corresponding holes, as the knife plotter does not cut through the Cut tape with fluidic channels "'U"CU. port;O". ~ Figure 8: Tape with uncut portions to aide in assembly. 13 outlet holes and the slightly larger alignment holes. 65 °Layer 3 - PCB Layer 3 is made of a completely etched PCB and is machined in the same manner as layer 1. The eight 0.08 in holes seen in Figure 9 are the pressure outlets that eventually lead to Layer 4 - Tape Layer 4, shown in Figure 10, is a channeled tape layer so its design and manufacture are created in the same manner as layer 2. This layer contains an additional channel that provides final routing of air that could not be provided by layer 2. The holes correspond to the holes tape backing. To maximize adhesion, the tape should be allowed to cure for 24 hours at room temperature or at 65°C for 1 hour. PCB I. the bladders. This layer attaches to layer 2 with the aid of the alignment holes and guide pins. Tape Figure 9: Layer 3, which is a PCB layer, showing Figure 10: Layer 4, a tape layer, showing large holes for pinch points, alignment holes, and bladder holes PMMA tape^ PCB 0 | J O o 0 O ( o 0 o Layer 5 - Actuation Layer PCB and PMMA LA 14 through layers 3 and 5. Enlarged holes in the figure are the bladder connections for pinch points. More will be said of this in the following sections. Layer The actuation layer consists of three layers together: 1) a 0.35 in thick PMMA layer, 2) a tape layer, and 3) a final PCB layer. These three layers form the housing for pinch points and are shown in Figures 11 and 12. The holes in all three layers are concentric and the tape layer and bottom PCB layer have identical hole layouts. In Figure 12, the PCB'and tape are behind the PMMA and the dark regions represent the visible portion of those layers. The PCB and PMMA layers are manufactured using the same traditional methods described previously but with the additional step of machining pinch point holes. The pinch point holes are made by drilling a ',4 in hole through the PMMA, then milling the large recess to a depth of 1.38 in using a 1 in end mill. The pinch points are made by creating a piston as shown in 13 and 14 that functions like a hydraulic cylinder to concentrate pressure to the seal-off point on the biocard. Pinch points are _Iu 10 Figure 11: Side view of a portion of the actuation layer. o o o o o Figure 12: Actuation layer made in PMMA with exposed portions of PCB and tape represented by the darker shading. Figure 13: Pinch point piston 1/8 tubing and is stiff enough to lift the piston from the compressed position but not so stiff as to greatly impede actuation pressure from the bladder. The bladder is used for actuation and is attached directly to the larger holes in layer 4. The larger holes correspond to the diameter of the inlet holes of the bladders. aligning each bladder hole with the hole in the tape and manually applying pressure. Springs and pistons were positioned into place in the actuation layer and the guide pins were then used to Bladders together in a specified pattern and attaching an inlet nipple. These bladders were modified by 15 made by attaching 118 in copper tubing to the center of a 3/4 in diameter copper head using JB Weld adhesive. The spring shown in Figure 14 has a slightly larger diameter than the copper Assembly of the actuation layer was completed using the alignment holes and guide pins as described previously. The bladders for the pinch points were then attached to layer 4 by attach the actuation layer to layer 4. The entire device was clamped together using a mild pressure and placed in a 65°C oven for 1 hour to strengthen adhesion. The bladders are made by Idaho Technology by heat sealing two pieces of plastic removing the nipple leaving only the hole. For the external bladders, tape was cut to the shape of each bladder and was attached by visual alignment. Figure 14: Cross section of a pinch point 16 The negative wire for each valve was soldered to the U-shaped portion of the copper on the perimeter of layer 1 shown in Figure 6. An additional wire was attached to the copper and connected to one pin of an empty 9-pin connector along with the eight positive wires of each valve. The other end of the 9-pin connector was connected to an eight switch array and 12 V power supply, which was used during testing. For a manufactured product, both valve wires would be connected to the top layer of PCB, which would also contain all the circuitry for the device. air inlet and eight three-way valves which route air to four bladders, three pinch points and one flow-through channel. The dimensions of the entire device measure 2.4 x 3.6 x 1 in excluding the protruding wire attachment and inlet hose. All bladders and pinch points were fully functional and no air leaks were detected when the device was attached to a low volume flowmeter. Figure 15: Top view of prototype showing Figure 16: Bottom view of prototype showing bladders, pinch points and air outlet. Electrical Connections val ve. Conclusion The prototype created for testing purposes is shown in Figures 15-17 and consists of an functional flowmeter. air inlet and valves. ure 17 Figure 17: Side view of prototype. device CHAPTER 3 TESTING AND RESULTS The primary objectives in the experiments conducted on the gevice were to better understand its functionality by establishing data for comparison with theoretical data and to ensure that the device could meet minimum lifetime requirements. To this end, three experiments were conducted. The first experiment was designed to directly measure force from a pinch point as a function of time in order to determine operating frequency for the device. The second experiment was a fatigue test designed to ensure that the device could meet the minimum lifetime of approximately one-million cycles and the third experiment measured the channel expansion when the device is connected to a pressure source. These experiments are explained in detail as follows. Experiment #1 - Time Constant Determination This experiment was designed to test for two design constraints. In order for the bioanalysis device to function properly, the pinch points must produce a force of at least 1 pound at a displacement of 0.8 in and must have the capacity to operate at 15 Hz with a 20 psi input pressure. An experiment was designed to measure the output force of the pinch point relating to channel number three as shown in Figure 18. This channel was hypothesized to have the greatest losses due to its length and curvature, so establishing that this channel can operate at 15 Hz at 20 psi validates the other channels in the system as well. Experimental data was collected by using 19 7 * ® - 3 8 • ® 2 ®5 * 4 an IESP-12 force sensor from HWV Technologies (www.hwvtech.com) attached to a Plexiglas manifold designed to keep the device and sensor in place. A schematic of this setup can be seen in Figure 19, which shows the device clamped into place on the Plexiglas manifold with the sensor in between them. The actual manifold is shown in Figure 20 with the testing device detached. The IESP-12 'force sensor' is in reality a displacement sensor (which corresponds to a force) and therefore testing for one pound of force was achieved simply by fixing the sensor at a distance where the fully displaced pinch point pressed it down to the position, d, corresponding to one pound of force as seen in Figure 21. Hence, used in this manner, the sensor could not 19 ® Figure 18: Layout of numbered channel geometries. an IESP-12 force sensor from HWV Technologies attached to a Plexiglas manifold designed to keep the device and sensor in place. A schematic of this setup can be seen in Figure 19, which shows the device clamped into place on the Plexiglas manifold with the sensor in between them. The actual manifold is shown in Figure 20 with the testing device detached. The IESP-12 'force sensor' is in reality a displacement sensor (which corresponds to a force) and therefore testing for one pound of force was achieved simply by fixing the sensor at a distance where the fully displaced pinch point pressed it down to the position, d, corresponding to one pound of force as seen in Figure 21. Hence, used in this manner, the sensor could not Sensor Figure 19: Schematic of device clamped into Plexiglas manifold showing the force sensor and steel plate. •plllp 4 Figure 20: Photographs of the Plexiglas manifold with the testing device detached, showing the 21: 10k£2 20 force sensor and steel plate. Figure 21 : Schematic showing the displacement, d, of the force sensor which corresponds to one pound of force and the maximum displacement of the pinch point measure the total force from the pinch point, but was simply used to indicate when the minimum force of one pound was reached. The sensor was placed a distance of approximately 0.001 in from the unactuated pinch point and attached to a 5.2 V power source in series with a lOkD. resistor as shown in Figure 22. Voltage readings from the force sensor were collected by a PCI 6023E DAQ card connected to Lab View 6.1. This setup was used to gather voltage and time data for the pinch point at its maximum displacement of 0.08 in. . S.2V - Figure 22: Circuit used to evaluate the pinch point data. 21 (C) E3611A power supply and (E) the National Instruments PCI 6023E DAQ card and computer. shows the manifold with the testing device clamped in place. The eight switch unit that controls the eight valves is shown at Point B, and Point C shows a flow meter connected between the air source and testing device and the 30 psi pressure gauge used to accurately determine the input pressure into the system. Point D shows the HP E3611A power supply and Point E shows the DAQ card connected to Lab View. 1) other connections to Lab View are made as described above. In order to get small time steps, Lab View was set to take three samples per reading at a rate of 10,000 samples/sec. once the data began to appear visually in LabView. 3) The valve was then opened and remained open until a steady state value was reached. The entire experimental setup is shown with its five components in Figure 23. Point A E36llA LabView. Pinch point data were collected using the following procedure: Input pressure was adjusted to the desired value and attached to the testing device. All LabView LabView 2) The data collection began with the valve in the open position. The valve was then closed Figure 23: Experimental setup showing (A) the Plexiglas manifold with the testing device clamped into it, (B) the switch array, the flowmeter and pressure gauge, (D) the HP E36llA R,=R2-^-, (1) Y IN V\ where the force sensor is R} and the lOkQ resistor is R2. Resistances were converted to forces using manufacturer's data. Each of the five resulting force curves was evaluated to find the first value greater than 1 lbf. The data collected at 15 psi are shown in Figure 24 and data from other input pressures are found in the Appendix. The time value corresponding to the first force greater than 1 lbf was subtracted from the time value corresponding to the final time that showed zero force. The resulting value gave the time constant for that pressure. Five values were collected for each pressure and these time constants were then inverted to give the frequency at which the device could be operated. 1.5 - 0 10 20 30 40 50 time [sec] 15, 22 4) The valve was then closed for at least two seconds, then opened again until reaching equilibrium. This was repeated until 5 curves were obtained at the input pressure. 5) These steps were used to collect data for pressures of 10, 12.5, J 5, 17.5, and 20 psi. The resulting data were imported into a spreadsheet and voltage values were converted to resistance values using the voltage divider shown in equation 1, (1) where the force sensor is RJ and the lOill resistor is R2. Resistances were converted to forces using manufacturer's data. Each of the five resulting force curves was evaluated to find the first value greater than 1 lbf. The data collected at 15 psi are shown in Figure 24 and data from other input pressures are found in the Appendix. The time value corresponding to the first force greater than 1 lbf was subtracted from the time value corresponding to the final time that showed zero force. The resulting value gave the time constant for that pressure. Five values were collected for each pressure and these time constants were then inverted to give the frequency at which the device could be operated. 0.25 . ~ 30 40 50 tirre[sec] Figure 24: Force data collected from pinch point at an input pressure of 15 psi. Results Analysis and Sources of Error 15, Lab View -, 140 - - (Hz) - requency 80 - 60 - Li. q _| , .. ^ , . . . . . . | . .. . r , j .j 7.5 10 12.5 15 17.5 20 Input Pressure (psi) 23 The results from the time constant experiments are shown in Figure 25 . The time constants have been inverted in order to give results in terms of operating frequency of the device. Error A key concern for the device, and the reason for determining time constants, was that the channel geometry was assumed to produce a large pressure drop so the need existed to establish that the device could meet design specifications. As seen in Figure 25, although there is scatter in the data, it is clear that the device meets the 15 Hz requirement at an input pressure of 20 psi and could even perform faster if needed. Error was minimized by the use of computerized data collection which gave time steps ranging from 7 to 14 milliseconds. This minimum time step became increasingly significant as input pressure was increased and time constants for IS, 17.5 and 20 psi gave average values of 17, 14, and 10 milliseconds, respectively. The greatest error was produced in the 20 psi data in which all data points are less than one LabView time step, so the operating frequency is most 160 140 120 ~ 100 J >- g 80- Q) ::I ~ 60- LL 7.5 15 17.5 Input Pressure (psi) Figure 25: Working frequencies for pinch point based on 1 lb force output. from the widely used pressure supply were likely. functionality. It This frame provided a 0.08 likely even higher than represented by this data. The resulting scatter in the data is evident in Figure 25, which shows increasing scatter with increased pressures. 24 Another source of error was due to a varying pressure source. The pressure source in the lab where these experiments were conducted is part of a system which is run throughout an entire building and could fluctuate based on usage in other labs. This error was minimized by connecting a 0-30 psi pressure gauge to the system and carefully monitoring the pressure during experiments. Because experiments generally lasted only 1-2 minutes for a data set at a given pressure, this error likely did not have a large impact on the data although minor pressure changes Overall the data collected in this experiment were considered to be valuable in establishing that this particular device was able to exceed design requirement of 15 Hz at 20 psi and indicated that if desired, the device could be operated at a much higher frequency or lower input pressure. In Chapter 3, these data will be compared with a mathematical model. Experiment #2 - Fatigue Testing Critical to the success of the project is the lifetime of the device. Based on estimated usage of this device, it was determined that each bladder would need to last for about one million cycles. The fatigue test described in this section was designed to show that the device would meet the lifetime requirements while maintaining full functionality. A second testing device was constructed using a simple IPC construction as explained in Chapter 2. consisted of two machined PCB's with a tape 0.2 in from the edge of the PCB. Three valves were attached to the device which controlled three bladders on the opposite side of the IPC as shown in Figure 26. Also shown in the figure is a frame that was constructed to maintain a counterpressure against the inflated bladders to simulate actual operating conditions. in clearance for the bladders to inflate. 1 -i-lape bladdefs ^H^ 1O( 2U8 WR) = J l, 512,000cyc/« • (2) Results and Analysis significant JUM thick r frame 25 valves air inlet ttape Figure 26: Schematic of device used for fatigue testing. The device was connected to a 20 psi pressure source and each valve was operated at 15 Hz for 28 hours producing about 1.5 million cycles as shown in equation 2. Experiments were conducted at room temperature with no attempt to control humidity or other environmental factors. ( IS cycteS](3600sec](28hour) 1,512,OOOcyctes' sec hour (2) Analysis After 28 hours of operation at 15 Hz, the device showed no signs of leaks or failures of any kind and continued to run with no problems. This experiment showed that the double sided tape and valves could be relied upon for a final product and that the method of attachment for the valves and bladders was also reliable for one million cycles. The possible fluctuations in the source pressure mentioned in the analysis section of experiment #1 were also a possible source of error for this experiment. However, these fluctuations vary between 1-2 psi at any given time which is not likely to have had a significant impact on this particular experiment. Experiment #3 - Tape Expansion The tape used in the device is 3M 444PC double sided tape. As illustrated in Figure 27(a), each of the two tape layers consist of a 13 ,Lim thick layer of PET, and 36 and 51 ,Lim thick P (a) (b) Figure 27: A cutout view of a sample cross-section of an air channel, (a) The channel with zero applied pressure, (b) The same channel with an applied pressure showing the elongation of the acrylic layers. (Drawings are not to scale) •:jP'^:ft:- significant 10 jum Results and Analysis 26 coatings of adhesive on the two sides of the PET. During early experiments, this adhesive was observed to be highly elastic and it was hypothesized that when channels are under a significant pressure that the tape, which is has a thickness consisting of 87% adhesive, would expand slightly and increase the cross-sectional area of the air channel as seen in Figure 27(b). Later research showed that the adhesive in the tape is 3M's high strength acrylic adhesive 300. Acrylic adhesives behave similar to elastomeric materials and have stress/strain curves similar to those shown in [10]. Because little is known about the acrylic adhesive used in the tape, expansion could not be predicted for the model so it had to be measured. Digital calipers with an accuracy of 1 ° .um were used to take thickness measurements at input pressures of 0, 10, 15 and 20 psi. Because of the touch sensitivity and accuracy of the calipers, five measurements were taken at each pressure and averaged. These values were then subtracted from the nonpressurized thickness to obtain the channel expansion at each pressure and are shown in Figure 28. Analysis The data follow the general trend seen in stress/strain curves for elastomeric materials such as an acrylic adhesive that have a steep initial slope followed by a region of reduced stiffness. If higher pressures were applied, it is probable that the slope would increase once again (b) Figure 27: A cutout view of a sample cross-section of an air channel. (a) The channel with zero applied pressure. (b) The same channel with an applied pressure showing the elongation of the acrylic layers. (Drawings are not to scale) 27 Figure 28: Channel Expansion vs. Input Pressure. Data points are averaged values from measurements, while the curve was generated to show the general trend of the data. as in the example curves but these higher pressures were not in the operating range of the device The three experiments were critical to the characterization of the IPC prototype. The determination of time constant at each pressure done in experiment one shows that the device is able to operate at the specified design parameters and gives data that will be compared to a mathematical model in the next chapter. The fatigue experiment showed that the tape layer and other components are able to survive under a constant amplitude cyclic loading having the same number of cycles as the required design lifetime. This was by no means an all-inclusive fatigue test for the tape, but showed that under the safe-life approach, this particular prototype has a safe life greater than 1.5. The tape expansion experiment is valuable because its data give an indication of how the a significant effect on air flow through the IPC and data from this experiment will be used directly in the mathematical model. ,.' ! 15 j 10 +-----------------------j 5+---------------------------------------------------------~ o+-------~--------~------~------~--------~------~--~ o 0.04 0.00 0.00 0.1 0.12 01arT1eI Bcparsion (rrnl) from so it was deemed an unnecessary risk to the device to continue to add pressure. Conclusion tape layer expands under pressure. This expansion increases the width of the air channel and has CHAPTER 4 would Pressure Derivation MODELING The system was characterized by creating a model that woulq accurately predict the operating frequency of the bladders at various input pressures. This mathematical model consists of three key calculations. First, an equation was derived that could predict the pressure in a bladder as a function of air velocity and time. Next an equation to predict the air velocity based on head losses through the system was developed to be used with the pressure equation. Because the velocity equation is a function of pressure and the pressure equation is a function of velocity, an iterative technique is used to predict losses over time. The measured output is force so calculations are also made to estimate the surface area of the bladder in contact with the pinch point piston at full displacement. In the sections that follow, the pressure derivation, velocity model, acceleration model, area of contact and the iterative spreadsheet setup are all described in detail. Following these computations, the results of the mathematical model are compared with actual test data for validation. Because the head loss equations did not provide a completely satisfactory result, an acceleration model was also added and is described following presentation of the head loss model. The first step in creating the model is to derive an equation for the pressure change inside the pinch point bladder. This derivation assumes that the air flowing through the system is an ideal gas, which can be characterized by the ideal gas law: W . n = ^- . (4) Wair p, V. Volume can be separated to a volumetric flow rate, Q, multiplied by time, t, and the volumetric flow rate is converted to an average velocity, V , by dividing out the cross-sectional area A of the Wmr =pV=pQt = (5) n = , (6) (MWair) py= p V A t RT. (7) nRT . (3) T, p, R, number W • , from its molecular weight MWair 29 PV = nRT. For the present problem we will assume that temperature, density, and universal gas constant, are all constants and that the most significant data will come from changes in pressure, P and the number-of moles, n. Volume, V, is also assumed constant because although the bladder volume is changing, a constant volume of air is required to fully inflate the bladder. The number of moles, n, can be converted to a more useful form by separating the weight of air, Wair ' MWair n = ----W-"a'i-r- --- (MWair ) (4) Equation 4 can broken down further by setting Wair equal to density, times volume, t, flow rate is converted to an average velocity, V by dividing out the cross-sectional area A of the flow channel. Wair = pV = pQt = pVAt. Equation 5 is then substituted into equation 4 giving pVAt n =---'---- (MWair ) , which is then substituted into equation 3 yielding PV = pVAt RT. (MWair ) (5) (6) d p = JTpVA_d V(MWall.) P2, V'f R T ^ V A d t . (9) MW„,) missing from equation 10 is a more detailed equation for the velocity V , which must be Laminar Model l 1]. A m 2 = (11) 32LjU AP L length,, Dh ju 30 The derivative of pressure with respect to time is then taken and rearranged to give dP = RTpVA dt. air ) (8) which is then integrated between two times and two pressures P2 12 fdP = f RTpVA dt. PI tlV(MWair ) (9) Solving for P2. the solution P = P. RTpVA A 2 1+ ut. V(MWair ) (10) is given, which can be used to iteratively solve for pressures over time. A key element that is still . computed at each time step. Velocity Model Model Because preliminary estimates for Reynolds numbers in the system ranged between 700 and 1800, the initial assumption was a laminar flow as based on the critical Reynolds number of 2300 for macroscale duct flow[ 11 ]. For fully developed laminar flow in a horizontal pipe, the average velocity V through the channel is given as _ MD 2 V = h 32Lfl (11) where /J.p is the pressure drop, is the channel length, , D" is the hydraulic diameter and f1 is the dynamic viscosity of the fluid. D„A 02) Ac P aforementioned straight IPC. Turbulent Model turn It turn (Recr) Dh, 31 This model indicated that the device should be able to be run at frequencies of 129 Hz and 588 Hz for input pressure of 10 and 20 psi respectively. Experimental data taken at the same pressures indicated operating frequencies of about 5 and 40 Hz, respectively, so the laminar model was shown to be a poor model in spite of the indication by the Reynolds number and reasons for turbulence needed to be investigated. Several researchers have noticed similar turbulence in microchannels, and although the actual value is not well-defined, most researchers agree that for microchannels, the critical Reynolds number CReer) is closer to 1000 [12,13] with microchannel being defined as a channel with a hydraulic diameter, Dh, less than 0.04 in. The hydraulic diameter of the IPC channel was determined using, D = 4A( h P (12) where Ac is the cross-sectional area of the channel and is the perimeter of the cross-section. The hydraulic diameter for the microchannel portion of the device is less than 0.0157 in, which indicated that for the current device a turbulent model is justifiable. This aforementioned research was also based on flow through microchannels, and gives no indication of Reynolds number values for the complicated flow path in the IPe. The turbulence that occurs in this system could be due to flow through its microchannels, but is more likely due to tripping of turbulence through a serpentine flow path which is modeled as follows. Model The flow of air experiences similar conditions as it passes through the channels to each bladder. As illustrated in Figure 29, the air enters through a large diameter hole and makes a 90° tum into a tape channel with small cross-sectional area. then makes another 90° tum into the valve which has a large cross-section at the inlet, a small cross-section near the valve seat, and 32 -valve- 1 turn. turn turn additional bend as discussed in Chapter 3). flow form P 2 P = h Ljotal ' (13) and by assuming AX - OC2 = 1, V, ~ V2, z, ~ z2 and renaming px and p2 to pin and pb respectively, equation 13 becomes, PIN ~ PB P h L,total ' (14) pin pb p hL t o t a l is the total head loss through the system. f---vaive--1 Figure 29: Example of the path of air through an idealized two-dimensional channel in the IPC returns to a large cross-section after making the 180° tum. As it exits the valve, it makes another 90° tum into a small diameter tape channel where it travels until it finally makes its last 90° tum into the large diameter hole that leads to the bladder or ambient air (the channel for valve 6 has an Each flow path was, in fact, more complex than this simplified two-dimensional flow channel, having curves in the tape channels and direction changes throughout the system, but the assumption was made that by designing curves with large radii of curvature that these losses would be minimized and become insignificant compared to other losses in the system. In modeling this two-dimensional system, the modified energy equation was used, starting with the form and by assuming a l = a 2 = 1, VI "" V 2 , ZI "" Z2 and renaming PI and P2 to Pin and Ph respectively, equation 13 becomes, Pin - Ph h = tolal' p where Pin and Ph are the inlet pressure and bladder pressure respectively, is the air density and hUo/al is i system, and valve. It should be noted that the contraction and expansion shown inside the valve is due to are listed in Figure 30 only for the sake of geometric completeness. L V2 h L , b e n d = f ^ - > d5) LJD velocity through the tape channel. 33 Head loss was determined by superposition of the components shown in Figure 30 consisting of (a) six 90 bends, (b) three expansions and contractions, (c) frictional losses through the tape channel, and (d) frictional losses through the larger diameter holes in the non-tape layers the geometry of the valve itself and is unrelated to the microchannels in the tape. Also, the larger diameter holes in the non-tape layers have negligible losses and are ignored in calculations; they For each bend in a channel, the head loss is determined by h - f Le V2 L.bend - D 2 ' (15) where/is the friction factor, LID is the dimensionless equivalent length, and V is the average Head loss for the expansions and contractions (b) and for straight channels (c) was determined similarly using, n I In I I---valve---l (a) J rllL 1 + (b) ~~ --...JL- ----..r- L- -...r- ~---.r- ~ (e) + + (d) I I I I Figure 30: Breakdown of head loss components in the system. (a) 90 bends, (b) expansions, contractions, (c) tape channel, and (d) larger diameter holes in the non-tape layers and valve. K exp/ cont = K V (16) h L,straigh Du 2 solving V = f'/he K L --+ -+ f 2D 2 2DhJ It 1 = -1.81og + • 6 ^ found in Table 2. It should be noted that loss coefficients are multiplied by N, the number of times the geometry is seen in a channel. Constants used in equations 10,18, and 19 are listed in Table 3. 34 il 2 h K- L,expi cont 2 ' and L il 2 hL,straighl == f --, Dh (17) respectively, where K is the loss coefficient, L is the length of a channel, and V is the average velocity. For the expansions and contractions, V is the average velocity through the smaller of the cross-sectional areas. Plugging equations 15-17 into equation 14 and sol ving for V yields V= ( / Le K L J p --+-+/- 2 D 2 2Dh (18) The friction factor/at each time step was computed using the Haaland equation [14], a modified version of the Colebrook equation. is defined as _1_ = -1.810g[[el Dh )1.11 + 6.9]. f1 3.7 Re (19) Equations 10,18, and 19 are the key equations for modeling flow through the IPC and predicting pressure inside the pinch point bladders as a function of time. These equations were modeled using a spreadsheet and solved at time steps on the order of about 0.05 milliseconds. The constant values relating to losses in channel #3 for the model are 35 Table 2: Loss constants used for each type of geometry including the number of each type of geometry, N, which is the multiplier for each equation. Le/D K L [mm] Dh [mm] N - - - - - - - - 3 m3 m2 R K/mol K absolute viscosity, /j 1.82e-05 m2 Dh molecular air, MWair surface roughness, e fjm This concludes the derivation as far as pressure prediction is concerned. However, of interest in this model is the prediction of a time constant for the pinch point associated with channel #3 to reach a force of 1.2 lb, because that is what was measured in experiments. order to predict the output force, an analysis of bladder geometry must be made in pinch point piston. Only then can the minimum pressure needed for the system be found by dividing the 1.2 lb force by this area. This may initially seem to be a straightforward calculation of dividing the force by the surface area may be in contact with the pinch point piston as illustrated in Figure 31. the bladder dynamics was simplified as follows. The bladder was analyzed from its two- Le/D K L [mm] Dh [mm] N Bend Expansion Contraction 60 0.5 0.8 6 3 3 Tape channel 40 0.391 2 Table 3: Physical constants used in the model bladder volume, V density, p channel cross-section, A universal gas constant, R temperature, T absolute viscosity, fJ hydraulic diameter, Dh wt. MWair surtaceroughness, e 2.37e-07 m3 1.0176 kg/m 3 (varied) m2 8.314 J/Klmol 294 1.82e-05 Ns/m2 (varied) mm 0.02897 kg/mol 1.15 fJm Area of Contact In order to make an accurate prediction of the amount of bladder area actually in contact with the cross-sectional area of the bladder, but unfortunately as the bladder inflates, less of its surface Because only an approximate solution was needed for the model, an in-depth analysis of 36 the pinch point piston. Pb, Pb=2d + 2h = 2d0. where d is the bladder diameter and h is the height of deflection. When there is no pressure in the system, h = 0, so the perimeter becomes a known constant equal to two times the uninflated diameter, d0. Solving equation 20 for d yields = d 0 h which gives the diameter at a known height. 32(a). Using this assumption, the contact diameter is approximated by subtracting the length h/2 from both sides of the diameter which gives d = d0-2h. area, h/2 area, Figure 31: Illustration of change in cross-sectional area of bladder in contact with dimensional cross-section and assumed to be rectangular with a perimeter, Ph, of (20) do. d = do - h (21) This approximation is still quite rough and is refined by assuming that the radius of curvature at each end of the rectangle is equal to one-half the deflection height as shown in Figure hl2 = do -2h. d d ! +"-----/---- (a) (b) Figure 32: Estimation of contact area. (a) Radius of curvature of hl2 used to estimate the diameter of the contact area. (b) A more accurate representation of the cross-section (22) and this is assumed to extend outward, broadening the radius of curvature as shown in (b). The diameter in contact with the pinch point piston remains unaffected. Using this approach with a bladder diameter d0 of 0.56 in and deflection height of 0.08 in, the contact diameter d was calculated to be 0.4 in which gave a contact area of 0.126 in2 or 8.17 x 10"5 . lb in2 This pressure corresponds to the closing of the pinch point and time constants in the model are based on this pressure. Bladder volume was also calculated to be 2.34 x 10"7 using the average diameter of 0.48 in and height of 0.08 in. Channel Expansion from model for values of 10, 15 and 20 psi. Predictions for 12.5 and 17.5 psi were obtained by fitting values to the experimental data curve. upper section is comprised of constants and initial conditions for the spreadsheet all with the exception of the boxed frequency value, which is a result of calculations. Input pressure, P^, volumetric flow rate, Q and the time step, delta t, are the inputs to the spreadsheet and all other values are either constants or calculations based on these inputs. The value for Q is an initial guess based crude experimental data of flow through an open air channel from a similar experiment. Because the open air channel has a different geometry than the tested flow path its estimate is inaccurate, but the velocity resolves itself by the second time step so this error is rendered negligible. It is clear in Figure 32(a) that there is more length on the square corner in than in the rounded, unaffected. 37 do in2 10.5 m2 • The force of 1.2 Ib was divided by the area of 0.126 in2 to give a pressure of 9.5 psi. 10-7 m3 Expansion of fluid channels due to internal pressure could not be predicted from available data so the data collected in experiment 3 shown in Figure 28 was used directly in the Overview of Spreadsheet Shown in Figure 33 is a sample from the spreadsheet for the 20 psi calculation. The Pin, Volume 2.34E-Q7 rri"3 1 _j e/Dh = 0.002 Pin 20 psi h e -- 1.15E-OB Pa w m kg/m"3 MW_air kg/rnol 4.86E-07 mn2 mu 1.82E-05 Ns/mA2 8.314 J/K/mol Dh 5.33E-04 m V1 = 30.86 hO 0.0002 d e l t a j delta_h 0.000124 m = com bined time Re f Velocity P2(psi) freq 0 0 0 0.0001 919 607 0.069 87 14 0.23 0 0.0002 2596 5 0.048 101.68 1555 0.49 0 0.0003 3029.62 0 046 103 04 0.76 ' 0.0004 3070.21 0 046 102 51 1.02 0.0005 3054 3 0 046 101 73 1.28 0.0006 3031.17 0.046 Tool! 8854" 1.55 " 0 Re step, / i s Columns Pin rho Area R T delta t Q V1 = 0.0001 sec I Freq = 238.095 Hz Turbulent and Laminar combined Velo P1 P2 30.86 0 r 8714 0 1555 · 101.68 1555 3370 3370 5209 0 5209 , 7039 0 101.73 7039 8854 0 100.93 8854 10656 0 Figure 33: Sample of the spreadsheet used in time constant calculations 38 0.9 A summary of the columns is as follows: the time column is the iterative time which is based on the time step. is the Reynolds number which is computed using the velocity from the previous time step. lis the friction factor and is calculated using equation 19. Velocity is the turbulent velocity equation 18 where the pressure difference is calculated using the input pressure and the pressure from the previous time step. Colunms PI and P2 are respectively the initial and final bladder pressures for the time step. PI is equal to the P2 from the previous time step. All of these pressures are computed in Pascals and column P2(psi) is the conversion to psi. The frequency column is set to zero for pressures less than 9.5 psi in the P2(psi) column and for values greater than 9.5, it is computed as the inverse of the time in seconds, yielding a value in Hertz. This is the previously mentioned value that is boxed in the upper section and represents the maximum operating frequency for the device at a given input pressure. 250 t 0 H 1 1 1 1 1- 7.5 10 12.5 15 17.5 20 It were initially assumed, and an accounting of these losses should be made. F = ma. (23) multiplying 39 Results and Analysis of Velocity Model The turbulent model predicted much faster flow rates than were observed in experiments shown in Figure 34. is obvious from these results that more losses exist in the system than an Acceleration Model A key assumption in the previous model is instantaneous acceleration to the turbulent velocity. In other words, as seen in Figure 33, the initial velocity when the valve is opened is assumed to be the fully developed velocity based entirely on the pressure difference in the system and losses due to channel geometry. The reality is that once the valve is opened, the pressurized air must accelerate the non-pressurized air in the previously sealed cavity. This air in the sealed cavity has a mass that is the product of the channel volume and air density. This mass can then be directly used in Newton's second law: F=ma. Force can be computed simply by mUltiplying the channel pressure difference by the cross-sectional area of the channel. Solving for acceleration yields • Experiment Data ~ 200 - Turbulent Model N ~ >- 150 u cQ5 )- 100 • ••• ~ LL ~ 50 ~ ~ ~ • * Input Pressure (psi) • • Figure 34: Comparison of experimental data with results from the model. Shown are individual data points from experiments and the model line. a = (24) m a Pin Pb A m V2=V1+ a(At). 40 where is the acceleration of the air, Pin and Pb are respectively the input and bladder pressures, is the cross-sectional area of the channel, and is the air mass. The air velocity is computed directly from the acceleration at each time step using (25) Equation 10 was used to compute the bladder pressure at each time step. The acceleration model was then combined with the turbulent model using the same iterative procedure previously described. This iterative model, however, computes velocity based on both the head loss model and the acceleration model. The model then selects the smaller velocity of the two (head loss only or acceleration only) to use with equation 10 to compute the pressure at that time step. Consequently, this technique uses equation 25 as the air is accelerating, but once the air reaches a velocity at which head loss will have a significant effect, it makes pressure computations based on head loss velocity. Results and Analysis of Acceleration Model The results of the combined acceleration and head loss model are shown in Figure 35. As shown in Figure 35, the acceleration model made some improvements to the turbulent head loss model, but the results are still not a complete representation of the observed behavior of the IPC prototype. This is likely due to additional losses that were not accounted for in the system. The assumption that most likely causes the greatest discrepancy between the model and the data is the assumption that was made of incompressible flow. This is often a reasonable assumption in fluid dynamics problems, especially when the medium is a liquid. The very operation of the IPC prototype, however, depends on compressible flow. Pressurized air flows through a channel into a bladder. This channel and bladder are not vacant, 41 250 t 0 i 1 1 1 1 1 7.5 10 12.5 15 17.5 20 during It much margin above the operating requirement. 250 • Experiment Data 200 - Turbulent Model '"N"' - Acceleration Model ~>- 150 • 0 t: • '/ Q) • ;j 100 0- ~ • ~ "",j" : LL 50 • * 0 7.5 10 12.5 15 17.5 20 Input Pressure (psi) Figure 35: Plot of acceleration model compared to experimental data and the turbulent model. but initially contain air at atmospheric pressure, which must be compressed if the system is to reach equilibrium. Clearly, this is what happens in the IPC prototype, and it is very likely that much of this compression takes place in the air channels at the location where the greatest pressure difference is found, thus resulting in a compressible flow condition. Recommendations and Conclusion The model presented does not fully characterize the operation of the IPC prototype, but is a good first estimate, especially given the inability of LabView to collect data fast enough during experiments as mentioned in Chapter 3. is likely that given a more accurate method of data collection, measured time constants would decrease even further and the model would fit the data better. Additionally, a model using compressible flow would likely yield better results, but is far more complex than an incompressible flow model and is outside the scope of this project. This project simply specified the capacity to operate at 15 Hz with an input pressure of 20 psi. Experimental results show that the flow channels allow for operating frequencies over 100 Hz so proving the 15 Hz operation using a mathematical model is of lesser concern because there is so CHAPTER 5 This work has detailed the motivation, design, fabrication and characterization of an desired bioanalytical assays. While offering similar functionality as its predecessors, this device has several improvements over them, which will be discussed and followed by a recommendation of future work that is necessary to move the device through the development stage. Improvements is paramount. in3. IPC in3 , computer software and hardware. Other designs that had been considered involved a fluidic manifold similar to the tape layer, but machined from aluminum. This would not only require the use of a CNC machine to create consistent fluidic channels in the aluminum, but would also CHAPTERS CONCLUSION anc~ integrated pneumatic card used as a miniature hydraulic module for a biocard analysis device. It has been shown that this device is able to provide the operating conditions necessary to run the The most significant improvement in the IPC over its predecessors is the size reduction. This was the primary goal of the funding for this research, so establishing that this goal was met The original device consisting of hoses, valves and an actuation layer could be stowed in a carrying case with dimensions of approximately 3.5 x 6 x 8 in, resulting in a volume of 168 in3 . The LPC device used in this paper had dimensions of 2.4 x 3.6 x 1 in resulting in a size envelope of 8.64 in3 , less than 5.2% the size of the original device. Using xurography to create the fluids channels in the tape allows the IPC to be easily machined using only a milling machine, band saw and a knife plotter with the appropriate solution, but adds manufacturing steps and parts, which both increase manufacturing cost. It also does not solve the problem of having to tolerance the aluminum to a specific surface finish so it will properly seal against another layer. A seal layer could be used to solve this problem, but once again, additional parts mean additional manufacturing cost. The tape layer used in the IPC has the dual benefit of providing the functional air channel layer and also acting as a seal between other layers of the device. It carrying could operate with as little as 15 psi input pressure. Displacement and force of 0.08 in and 1 lbf respectively were met as design constraints, and fatigue was shown to have a lifetime of at least 1.5 times the required lifetime of one million cycles. 43 require a method to seal the layer against leaks. Bolting the layers down seems to be a simple Another cost-saving technique was to use PCB as the substrate for the IPC. In order to have electronic controls in the device, PCB would have to be purchased, whether used as a substrate or not. would also have to been processed, machined and etched, most likely by an outsourcing company that specializes in PCB processing. Modifications to a PCB design to accommodate IPC controls consist only of some additional holes drilled in the PCB and some additional planning by circuit design engineers to ensure electrical components fit on the PCB with the valves. Once the initial circuit design is complete, the additional holes drilled in the PCB would amount to only a few cents added to the present PCB manufacturing cost. So the design of the IPC not only led to nearly a 95% reduction in size, but the manufacturing cost for such a device is greatly reduced due to the use of xurography and the reuse of electronic cauying PCB as a substrate. In fact, the only significant costs for the device are the valves and the electronics, which are both outside the scope of the present research. Accomplishment of the frequency, pressure, displacement, force and fatigue design requirements shown in Table 1 have all been established in the preceding chapters. These chapters have shown that the device could be operated at frequencies up to about 35 Hz with the specified 20 psi input pressure, or that at the specified operating frequency of 15 Hz, the device I Fatigue temperatures and in various levels of humidity. One fatigue test was already conducted as part of this research, but fatigue is largely stochastic and many tests must be conducted in order to have any reasonable predictability of fatigue life, especially given the range of temperatures at which the device could be operated. Fatigue tests for the 444PC tape should be conducted both at room temperature and at elevated temperatures and S-N curves developed at several temperatures and similar load conditions in order to accurately predict the useful lifetime of an IPC. Valves further. 44 Future Work This work has successfully presented the proof of concept for an IPC-type device, but in order to move the device through the development stage and into a production stage, additional studies should be conducted. The IPC's intended use requires pressurization and subsequent depressurization of air channels, putting cyclic loads on the tape layer, which holds the device together. Also, the proposed use of the device requires it to be carried in backpacks of soldiers, possibly in high IPc. Another task that would be meaningful in the development of an IPC-device would be to find an alternative to the off the shelf valves used in the device. Valves cost approximately $20 each, and when six valves are mounted on a 50 cent actuation layer, an eyebrow of concern must be raised and the question arises of how to eliminate or lower the valve expense. Many MEMS-type valves have been developed in the past several years [1], and some have even been developed for use with a PCB-microfluidic device [2]. If a simple valve could be developed for use with the IPC, the cost could be lowered further. normally •Ell BWBBiliii Fig ~~ MM The valve is opened by some actuator acting through the small hole in the top of the PCB, which IPC. 200/^m 45 One design, that has some merit is shown in Figure 36 and was built and tested as a side project of this research. The thinnest layers in the figure are tape layers, and the slightly thicker layers are made of a thin silicone sheet. The valve seat is made of a small plastic washer that is prestressed against the upper silicone layer and gets further closing force from the input pressure. deflects the plastic washer enough to break the seal and allow flow. The valve can also be used as a check valve as shown in the 'normally open' diagram in the figure. Tests on this valve showed a great deal of potential for use with the IPc. These tests were conducted by actuating the device manually and required only a deflection of about 200,um lally closed open normally open Figure 36: A valve design based on IPC technology. It is normally closed in one direction, but normally open in the opposite direction acting as a check valve that can be actuated. Tape Expansion Fluid Modeling for full actuation. Potential means of actuation include thermal actuation using paraffin wax or the use of a small solenoid. 46 Parts for this valve have the similar low cost of other materials in the IPC, so the development of an actuator for this or a similar valve has the potential to greatly reduce the cost of an IPC device because, like other MEMS devices, many valves could be produced with only one production step. Expansion In order to be able to predict flow rates and time constants in future designs, more information must be made available about the expansion of the tape channels. A complete analysis should be made on the tape to develop a more accurate and complete stress/strain relationship for the adhesive. Analysis should be done not only to establish the increase in height of a pressurized channel, but should also include the change in width of the channel as illustrated in Figure 27. Because the channel height was the critical dimension for this device, minute changes to width would not have a large effect on the cross-sectional area and flow. For example, with dimensions of 1500 x 200 microns, a 50 micron change to the height increases cross-sectional area by 25% while the same increase to the width increases the area by only 3.3%. Although these changes were considered negligible in this analysis, they could very well be a reason the model is slightly conservative and should be studied in more detail, especially for devices having width as a critical dimension. Modeling The fluid model was shown to have much room for improvement. Several assumptions were made that affect the data, but the assumption that most likely will affect the model the most is the assumption of incompressible flow. If operating frequencies drastically increase for the Summary significantly 47 device, a compressible flow model should be developed for design purposes. However, the SMC solenoid valves have a maximum operating frequency of about 30 Hz, so any design that pushes the limits of a 100 Hz operating frequency will be limited first by the SMC valves. In conclusion, a device was created that meets design requirements and significantly reduces size and cost of its predecessor. Operating frequencies were determined experimentally and yielded results which verified that the design requirement of 15 Hz at 20 psi had been met. This work could be improved and further developed conducting fatigue analysis on the tape layers to give an accurate estimation of operating lifetime. Integration of inexpensive valves would reduce costs of the most expensive components in the device and further development of a more exact fluid model would allow for design of more complex IPC systems. APPENDIX TIME CONSTANT EXPERIMENTAL DATA 1.25 0.75 0.5 0.25 0 pinch point force for 10 psi 10 20 30 time [sec] • -I mm , m ^» m>-r- 40 50 1.5 1.25 1 0.75 -§ 0.5 0.25 0 0 pinch point force for 1Z5 psi 20 30 40 time [sec] 50 60 The following figures represent the data collected in Experiment #1 that was a collection of time constant data at different input pressures. Figures are listed in order from 10 psi to 20 psi. 1.5 C. ~ 0.75 0.5 0.25 0 0 20 3) tirre[sec] 12.5 :c' c:. i 0 10 tirre3 fsec] 40 1.5 1.25 „3d? 1 i 8 0.75 0.5 - 0.25 I 0 -I pinchpoint force for 15 psi o 10 20 time3 0[s ec,] 40 50 60 1.5 1 0.75 0.5 0.25 pinch point force for 17.5 psi tirrefsec] 1.5 0.75 0.5 0.25 pinch point force for 20 psi 10 20 30 40 time [sec] 50 60 49 pird ~nt 1.25 :c' 1 c::. B 0.75 0.5 0.25 0 0 20 tirre3f'sec] 40 1.25 :c' =:. B 0.75 0.5 0.25 0 0 5 10 15 tirrefsec] 25 30 35 40 1.25 1 :c' c:. i 0 0 20 30 40 tirre[sec] I] Foundations of MEMS. [2] T. Merkel et al, "A new technology for fluidic microsystems based on PCB technology," Sensors and Actuators, vol. A77, pp. 98-105, Feb. 1999. [3] T. Merkel et al, "A microfluidic pH-regulation system based on printed circuit board technology," Sensors and Actuators, vol. A84, pp. 230-235, Sep. 2000. [5] A. Wego et al, "Fluidic microsystems based on printed circuit board technology," J. [6] R. O'Hayre et al., "Development of portable fuel cell arrays with printed-circuit Industrial Electronics Society, 2005, 32nd Annu. Conf. of IEEE, pp. 2352-2356. al, ASTM [9] D.A. Bartholomeusz et al, "Xurography: rapid prototyping of microstructures using a cutting plotter." Journal of Microelectromechanical Systems, vol. 14, pp 1364-1374, Dec. [10] Griffith Polymers Inc.(2007, June 2). Stress/Strain Properties of Polyurethane. Website. Available: http://www.polyurethane-l.com/tensile-strength.htm [ I I ] Fox et al., Introduction to Fluid Mechanics, 6th Ed., New York: John Wiley & Sons, 2004. Microscale Thermophysical [13] N.T. Obot, "Toward a better understanding of friction and heat/mass transfer Conf. Transport Phenomena in Microscale, Banff, Canada, 2000, pp. 54-64. http://en.wikipedia.org/wiki/Colebrook_equation. REFERENCES [1] C. Liu, New Jersey: Pearson Education, Inc., 2006. ai., 1999. aI., [4] N.T. Nguyen and X. Huang, "Miniature valveless pumps based on printed circuit board technique," Sensors and Actuators, vol. A88, pp. 104-111, Feb. 2001. aI., Micromech. Microeng., vol. 11, pp. 528-531, Jun. 2001. ai., technology," Journal of Power Sources, vol 124, pp. 459-472, Jun. 2003. [7] D. Fries et al., "PCB MEMS for environmental sensing systems," in Proc. of Industrial 32nd Con! [8] G.E. Wheeler et aI., "Fracture mechanics applied to adhesive joints," Journal of ASTM International, vol. 2, no. 5, pp. 1-18, May 2005. at., 2005. Website . http://www.polyurethane-comltensile-1 J] R. at., [12] G. Morini, "Laminar-to-turbulent flow transition in microchannels," Microscale Thennophysical Eng., Vol. 8, Num. 1, pp. 15-30, Mar. 2004. in microchannels - a literature review," Proceedings of the Int. Con! on Heat Transfer and [14] Wikipedia, The Free Encylopedia. (2007, May 17). Colebrook Equation. Available: equation.