Integration of an In-Situ Consolidation Mechanism into the Fused Deposition Modeling Process for Low Porosity Continuous Fiber-Reinforced Thermoplastic Structures

(1)

1

Integration of an

In-Situ Consolidation Mechanism

into the

Fused Deposition Modeling Process

for Low Porosity

Continuous Fiber-Reinforced

Thermoplastic Structures

Nick Willemstein, MSc.

Faculty of Engineering Technology

(2)

(3)

Integration of an In-Situ Consolidation Mechanism

into the Fused Deposition Modeling Process

for Low Porosity

Continuous Fiber-Reinforced

Thermoplastic Structures

PDEng Thesis

to obtain the degree of

Professional Doctorate in Engineering (PDEng) at the University of Twente,

on the authority of the rector magnificus,

prof.dr. T.T.M. Palstra,

on account of the decision of the graduation committee,

to be defended

on Thursday the 18th of February 2021 at 14:30 hours

by

Nick Willemstein

born on the 30th of January 1991

in Sliedrecht, The Netherlands

(4)

(5)

Acknowledgements

In February 2017, I started my PDEng journey, which has been a great learning experience and for which I am grateful for having had this opportunity. It enabled me to develop my skills, both theoretical and practical, while also enjoying the journey through the challenges and people that I got to know along the way.

Firstly, I would like to thank my supervisors, Tom & Ismet, for their helpful feedback and the freedom they gave me during the PDEng project. Even though my planning suffered some setbacks, their feedback ensured that I did not stray too much from the path towards timely completion.

Secondly, I would like to thank the DPM workshop staff (especially Norbert, Theo, Joop, Thomas, Simon, and Dennis) without whom I would have been lost during the manufacturing phase of my research. In particular, their patience with teaching me how to use the machines and coping with my large number of laser cutting requests. Also, I want to express my gratitude to the Production Technology lab staff, both Bert and Nick, who were always very helpful in preparing/using samples and machine and the usage of their valuable lab space (even though at times my setup was quite messy).

Thirdly, I also wish to express my appreciation to my fellow PDEngs (Sam & Logen) with whom I have had fruitful discussions and a good time overall within the lab. Additionally, I want to thank my office mates of WH251 (Monique, Katja, Merel, Gisela, Roland, Chanmi, and Ilknur) for the sometimes (not so) deep conversations. The same appreciation goes out to the people who were present at the daily lunch break, such as Robert-Jan, Patrick, Roberto, and many more. Thanks for the many laughs and discussions over the years.

Lastly, I would like to thank my parents and younger brother, Edwin, for their support in pursuing my ambitions and believing in me.

(6)

(7)

Abstract

Additive manufacturing of continuous fiber-reinforced thermoplastics (C-FRPs) enables engineers to realize complex structures with a high strength-to-weight ratio and resistance to corrosion and impact. Recently, fused deposition modeling (FDM) has become a prominent candidate due to its low cost and ability to manufacture complex structures. Although some commercial FDM printers for manufacturing C-FRP structures exist (such as the MarkForged & Orbital Composites), these have been shown to often contain porosity content in the range of 10%. Whereas the high-performance C-FRP applications requires a much lower porosity. This constraint is due to the negative effect of porosity on the mechanical performance of the C-FRP structure, which includes a reduction of flexural, transverse tensile, and interlaminar shear strengths. Thus, to improve the application range of printed C-FRP structures the reduction of the porosity content is an important step.

The high porosity content in printed structures is mainly due to the absence of consolidation pressure during printing. In conventional C-FRP manufacturing processes (such as tape laying), a process-specific in-situ consolidation (ISC) mechanism is integrated to supply this consolidation pressure. The integration of an ISC mechanism into the FDM process would be a key milestone to realize printed C-FRP structure with low porosity. However, the ISC mechanisms of conventional processes cannot be directly copied to the C-FRP FDM process, due to the absence of a mold and quick solidification. This thesis aimed to realize an FDM-specific ISC mechanism by investigating the associated design problem of ”The integration of a consolidation (sub)process into the FDM process” & the development of an FDM platform for further research.

A filament production & FDM printer with an integrated ISC mechanism were developed within this work. The former was realized with a pultrusion setup, which converted a glass-fiber-reinforced polypropylene commingled yarn with 2000TEX to a circular C-FRP filament. The filament was manufactured using a die opening larger than required to induce a higher initial void content. This approach for manufacturing low-quality filament enabled the validation of the ISC mechanism’s robustness.

For printing with the C-FRP filament, a custom FDM-based platform was developed. The key design tasks for this system were the integration of an ISC mechanism and an extendable control framework (for both process planning and control). The latter was realized by developing, a custom slicer, electronics, and software framework. The inherent flexibility of the developed framework allows for straightforward changes and/or expansion of the system’s behavior. These behavioral changes include adjustments to the path planning & additional inputs and outputs.

The developed FDM platform and pultrusion system were capable of fulfilling their respective

(8)

VIII ABSTRACT

tasks. The was validation was done by printing with the C-FRP filament and producing filament, which did show that they had some stability issues. These stability issues were often related to mechanical components. Within the pultrusion system, clogging of the infeed material was possible and leakage of the matrix material at the heated die. Whereas the cutter and printhead need improvement for the FDM system, to realize reliable automated cutting and more reliable extrusion. In contrast, the underlying software framework and electronics have demonstrated both robustness and simplicity in changing and/or adding functionality.

An ISC mechanism provides consolidation capabilities by melting and then solidifying the matrix-material under pressure. The developed ISC mechanism implements this behavior by using a long plate that incorporates a thermal gradient (for melting and solidification) and pre-loaded springs (for pressure). The former is realized by having a 90 mm long steel plate, with temperatures set to 200◦_{C and & 80}◦_{C at the front and end of the plate, respectively. This gradient has the intention}

to melt the polymer at the front and exit the plate solid & consolidated. The plate itself was moved with a speed of 0.5-0.8 mm/s over the printed structure (at least) once per layer to consolidate the current layer. During this motion, a consolidation force of around 80 N was exerted.

To validate the ISC mechanism multiple small beams (50x12x4 mm3_{) were printed with different}

process settings. The 80 N exerted by the ISC mechanism, led to consolidation pressures of 0.133 MPa and 0.266 MPa (when moving twice over half the structure). Analysis of micrographs of the printed beams indicates that the developed ISC significantly reduces the porosity while also achieving a good matrix/fiber distribution when compared to unconsolidated structures. Therefore, it was concluded that the developed ISC mechanism is a viable option for integrating consolidation capabilities into FDM printers.

Although the ISC mechanism has demonstrated potential, further research is needed to optimize the processing parameters & characterize the effect on mechanical performance. Firstly, the slow consolidation speed (0.5-0.8 mm/s) compared to printing speed (2 mm/s) used in the current iteration is a considerable drawback. This speed discrepancy can be reduced by increasing the maximum temperature and further investigation of the effect of a higher consolidation speed on the porosity content. Secondly, further reduction of the porosity content requires a higher consolidation pressure, which requires an increase in force output. However, the simplicity and robustness (to processing low-quality filament) make the developed ISC mechanism an interesting option if higher consolidation speeds and pressure can be realized.

(9)

List of Figures

1.1 MarkForged X7 [1] (left) and Orbital Composites printer [2] (right). . . 2 1.2 Process-ability versus performance for different fiber types [3]. . . 2 1.3 Micrograph from the cross-section of a multi-layer 3D printed C-FRP structure, with

alternating layers of matrix-only and fiber-reinforced material [4]. . . 4 1.4 C-FRP parts printed with a MarkForged MarkOne with inter-track voids [4]. . . 5 1.5 Schematic view (left) and micrographs for ABS filament (right) of inter-layer voids

adapted from [5] and [6], respectively. . . 6 1.6 The increase of thickness due to the deconsolidation of a fiber-reinforced PPS

(Polyphenylene sulfide) laminate exposed to a range of temperatures [7]. . . 6 1.7 Schematic structure of the thesis. . . 8 2.1 Functional flow of the desired FDM process for C-FRPs, pictures taken from [8] [9]

[10] (left to right). . . 9 2.2 Common methods to extrude C-FRPs in the FDM process. . . 10 2.3 A schematic view of a pultrusion setup [11] and commingled yarn [12]. . . 11 2.4 A representative image of pultruded filament (microscopic image and impregnation

gauge), adapted from [13] (left) and [14] (right) . . . 12 2.5 Temperature and radial expansion versus time for a reheated pultruded sample [15]. 12 2.6 An example of a no-cut strategy using hatch spacing [16]. . . 13 2.7 Representative images of pre- (left) and post-printhead (right) cutting processes. . 14 2.8 Evolution of microstructure for increasing consolidation time, adapted from [17]. . . 15 2.9 Schematics of existing ISC methods. . . 16 2.10 Printhead with multiple pultrusion stages to cyclically de-consolidate and consolidate

[15]. . . 17 2.11 Printhead with pressure roller [18]. . . 18 3.1 The functional flow of the pultrusion system with requirements allocation (images

A, B, and C from [12], [14], and [8], respectively). . . 23 3.2 The architecture for the pultrusion setup with functional allocation. . . 23 3.3 Schematic with all important parts of the FDM process. . . 25 3.4 Functional flow of the FDM process for C-FRPs (boat images from [19] & filament

from [8]). . . 25 3.5 The consolidation process in functional expression. . . 27 3.6 The functional flow with consolidation integrated (boat images from [19] & filament

from [8]). . . 28 4.1 Categorization of all subsystems in terms of control, yellow are part of the pultrusion

setup. . . 32

(14)

XIV LIST OF FIGURES

4.2 Abstract description of thermal control loop. . . 33

4.3 Sensing electronics for thermal control, thermistor from [20]. . . 34

4.4 Actuation electronics for thermal control. . . 34

4.5 Approach used for computing and using state-space controllers. . . 35

4.6 Abstract depiction of motion system. . . 36

4.7 Driver electronics for motion system. . . 36

4.8 ATMEGA328P motion control with RPi commands. . . 37

4.9 Global layout of the pultrusion system, with all subsystems indicated (including relevant subsection). . . 38

4.10 Functional description of Infeed subsystem. . . 39

4.11 Considered solutions for the infeed subsystem. . . 39

4.12 Infeed - Concept (left) and CAD model (right). . . 40

4.13 Realized Infeed subsystem, with spool (left) and alignment (right) section. . . 41

4.14 Functional description of heated die subsystem. . . 42

4.15 Considered solution for the heated die. . . 42

4.16 Two-Plate Mold - Concept. . . 43

4.17 Sandwich Iron with mounting - the entire subsystem (left) and internal die (right). . 44

4.18 Functional description of cooling system subsystem. . . 45

4.19 Considered solution for cooling module subsystem. . . 46

4.20 Cooling Module - Concept (left) and CAD model (right). . . 46

4.21 Realization of the Cooling Module. . . 47

4.22 Functional description of outfeed subsystem. . . 48

4.23 Outfeed - concept (left) and CAD model (right). . . 49

4.24 Realized system of the pulling mechanism/outfeed subsystem. . . 50

4.25 The fully realized pultrusion system, which also shows the integration of power supplies and electronic boxes (grey, subject of next chapter). . . 50

4.26 Functional analysis of XY-Table. . . 52

4.27 CAD model of the XY-Table. . . 53

4.28 The realized XY-Table subsystem, digitally enhanced to focus on important parts (the big box on the right contains electronics). . . 53

4.29 Functional analysis of Gantry subsystem. . . 54

4.30 CAD model (left) & realized system (right) of the gantry. . . 55

4.31 Functional analysis of the Printhead. . . 56

4.32 The two types of extrusion (left, adapted from [21]) and the Titan Extruder from E3D [22] (right). . . 57

4.33 The heating section of a printhead into four parts. . . 57

4.34 The realized printhead, which includes heating block and extruder subsystems. . . 58

4.35 Functional description of the printbed. . . 59

4.36 Concept of printbed. . . 60

4.37 Realized system of the printbed, with off the shelf components indicated. . . 61

4.38 Circuitry used for the load cell, using a HX711 and an Arduino Uno. . . 61

4.39 The functional analysis of the cutting mechanism. . . 62

4.40 Sketch of the Kyros Module with the coupler and spine indicated. . . 63

4.41 Realized system of Kyros’ motion system. . . 63

4.42 Considered solutions for cutting of the C-FRP filament. . . 64

4.43 Sketch of the wire cutter-side of the Bowden system. . . 65

(15)

LIST OF FIGURES XV

4.45 Control circuit for Kyros’ cutting mechanism. . . 66

4.46 Realized system of both the stepper and wire cutter side. . . 67

4.47 The functional analysis of the ISC mechanism. . . 68

4.48 Considered solutions for the thermal gradient component of the Typhon module. . 68

4.49 Concept of the Typhon module. . . 69

4.50 Concept of the Typhon module’s consolidation process. . . 70

4.51 Scarf to preserve geometric accuracy after consolidation. . . 70

4.52 Realized Typhon Module. . . 71

4.53 Ideal force evolution for the Typhon Module. . . 72

4.54 Force evolution for the Typhon Module, per the pressure experiment with lowpass filtering. . . 73

4.55 Force evolution for the Typhon Module for a small beam leading to a triangular shape. 73 4.56 The AM system fully realized, with all the power supplies and electronic modules integrated. . . 74

5.1 Set of PCBs with controlled subsystems indicated. . . 78

5.2 Simplified controller board design. . . 79

5.3 Simplified I/O layout of the Gantry PCB. . . 80

5.4 Realized PCBs for microcontroller (left) & gantry (right). . . 80

5.5 Sequence diagram of the interface design, for the AM system. . . 82

5.6 Finite state machine for the pultrusion setup. . . 82

5.7 Finite state machine of AMCore, with initialization phase in blue. . . 83

5.8 Hierarchical FSM with ”air-gap” between AMCore’s coordination and subsystem I/O. . . 84

5.9 Simplified quick control of selected subsystems. . . 84

5.10 Interaction between AMCore and subsystem. . . 85

5.11 Functional description of slicer for generating the deposition path. . . 86

5.12 Activity diagram of the PolySlicer algorithm with important intermediate points labeled. 87 5.13 PolySlicer output for labels A & B in the flow diagram. . . 87

5.14 PolySlicer output for labels C (lower right) & D (lower left) in the flow diagram. . . . 88

5.15 Scanline algorithm example. . . 88

5.16 Simplified activity diagram for the G-Code construction in PolySlicer. . . 89

5.17 Plot of the G-code wherein the blue lines represent extrusion (infill) and the walls (greenish). . . 90

5.18 User interface for the AM with red, blue and yellow indicating boxes that can be set, user feedback, and functional blocks. . . 92

5.19 User interface for the manual control of the AM system with red, blue and yellow indicating boxes that can be set, user feedback, and functional blocks. . . 92

5.20 Example of real time tracking of the temperature over time. . . 93

5.21 Leakage between die halves (top) and effect of clamping pressure on deformation (bottom). . . 95

5.22 Filament on spool. . . 96

5.23 Test print with unreinforced filament with printed (left) and CAD model (right). . . . 97

5.24 Beam structure made of C-FRP filament. . . 98

5.25 Slivers near Titan Extruder after printing with C-FRP filament. . . 100

(16)

XVI LIST OF FIGURES

6.2 Pultruded filament in which the fibers are accumulated in the center surrounded by

the PP-matrix. . . 106

6.3 Micrographs of the experiments at a die temperature of 175◦_{C. . . 107}

6.4 Micrographs of the experiments at a die temperature of 180◦_{C). . . 107}

6.5 Micrographs of the experiment at a die temperature of 185◦_{C. . . 108}

6.6 Activity diagram for computing the eccentricity and major/minor diameter of the filament samples. . . 108

6.7 Typhon module process parameters. . . 111

6.8 Illustration of processing with the Typhon module before, during, and after consolidation.112 6.9 Printed beam after printing & consolidation with general features indicated. . . 113

6.10 Micrograph for the beam without consolidation. . . 113

6.11 Micrograph for the beam with ’Default’-parameters. . . 114

6.12 Micrograph for the beam with ’Low Temperature’-parameters. . . 115

6.13 Micrograph for the beam with ’High Speed’-parameters. . . 115

6.14 Micrograph for experiment with ’High Pressure’-parameters. . . 116

7.1 Hybridization of 3D printing and conventional C-FRP processes. . . 123

A.1 Concept for Typhon version 2.0. . . 136

A.2 A schematic of the desired system with workspace and mounting points. . . 137

A.3 Sketch of the XY mechanism . . . 138

A.4 CAD model for the XY stage, the link with a hexagon at the output stage. . . 138

A.5 CAD model for the rotational joint. . . 139

A.6 Sketch for the flexure for Z-translation . . . 140

A.7 Schematic of the parallel flexure from [23]. . . 140

A.8 CAD model for Z-stage . . . 141

A.9 Displacement of Z-translation stage for a load of 80 N . . . 142

A.10 Stress analysis from Solidworks model . . . 142

A.11 Motion Controller Version II, circuit design . . . 143

A.12 Bed meshing principle & incorporation into software. . . 144

A.13 The Helios Build Platform Concept. . . 145

A.14 A concept for an improved heated die. . . 146

A.15 Functional analysis of the infeed subsystem with integrated pre-heating. . . 147

A.16 A concept for the inside of a pre-heated infeed (top view). . . 147

B.1 Activity diagram of the PolySlicer-algorithm with relevant section indicated. . . 149

B.2 Faces, CL & vertices of a tetrahedron. . . 150

B.3 An example of a pre-processed mesh. . . 151

B.4 Processing a tetrahedron using the amount of unique z-values and current height zk.152 B.5 Effect of a small change in the local CL. . . 153

B.6 Set of 2D polygons that make up the shape (left) and the unified polygon (right). . 154

B.7 Computing an expansion of a polygon. . . 155

B.8 The NOT-operation and its result. . . 157

B.9 Computed Trajectories for the skirt (left) and brim (right). . . 158

B.10 Infill polygon with all scanlines plotted. . . 160

B.11 Trajectory generation using a piecewise approximation for walls. . . 160

B.12 Trajectory generation using a piecewise approximation for infill. . . 161

(17)

LIST OF FIGURES XVII

B.14 Difference between current and proposed trajectory planner. . . 163

B.15 Example of a structure that requires multiple skin fills. . . 164

B.16 Computation of a fillet (left) and effect (right). . . 165

B.17 Process to introduce complex infill patterns. . . 167

B.18 Seed line based generation of scanlines, with relevant points indicated. . . 168

B.19 Output of seed line approach, for angles of 0, 0.25π, and 0.5π radians. . . 169

B.20 Activity diagram for constructing Voronoi Diagrams using Euclidean Allocation. . . 170

B.21 An example of using Euclidean allocation. . . 171

B.22 G-Code linking based on output from the scanline algorithm for a single layer. . . . 173

C.1 Standard circuit for the PWM-controlled power MOSFETs. . . 179

C.2 Sensing circuitry for a thermistor. . . 181

C.3 Integration of the DRV8825 stepper driver onto the PCB. . . 183

C.4 Heated bed MOSFET circuitry (right) and the 5V-to-12V converter (left). . . 184

C.5 Fan control of the Sanyo Denki Fan used in the Typhon module, left (control circuitry) and right (internal circuitry [24]). . . 185

C.6 Example of PCB for expanding functionality of AM system. . . 186

D.1 Exothermic energy versus time, for experiment 1. . . 189

E.1 The multi-bolt concept with mount (left) and dies (right). . . 192

E.2 The realized two bolt system. . . 192

E.3 A concept to enable blending-preserving re-spooling. . . 193

E.4 Analysis of extruder design for the extrusion of C-FRP filament. . . 195

E.5 White residue after printing with C-FRP filament. . . 195

F.1 An example of a PRBS signal, generated using MATLAB’s idinput-function. . . 198

F.2 Activity diagram for I/O data acquisition. . . 199

F.3 Hankel singular values for the original 42-state H∞-controller. . . 201

F.4 Class diagram of the StateSpace controller-class. . . 203

F.5 Activity diagram for using the state space controller inline. . . 204

F.6 Activity diagram for the step-wise reference signal generator. . . 205

G.1 Activity diagram for finding eigenvalues. . . 209

G.2 Activity diagram for dimensioning the heated die/cooling system. . . 211

G.3 Schematic of the stresses on a spooled filament. . . 212

G.4 The microscopic void model. . . 215

G.5 The ABB IRB460 [25] (left) and a wire model approximating its geometry (right). . 220

G.6 Two poses of the Talos manipulator in SolidWorks (left) and realized module (right). 221 G.7 Sketch of Kyros mechanism . . . 221

G.8 Class diagram for a generic module, which shows the name, parameters, functions (from top to bottom). . . 223

G.9 State diagram for TCUs (heated bed and block). . . 224

G.10 State diagram for the Typhon module (ISC). . . 224

G.11 State diagram for the Kyros module. . . 225

G.12 State diagram for the axis controller with a single axis. . . 225

(18)

(19)

List of Tables

4.1 Analysis of solutions for supplying pressure component oin the Typhon module. . . 69

5.1 Requirement evaluation for the pultrusion system. . . 94

5.2 Evaluation of operation requirements for the pultrusion setup. . . 96

5.3 Requirement Evaluation for the process level requirements. . . 99

5.4 Evaluation of operation requirements for the AM system. . . 101

6.1 Parameters for pultrusion experiment . . . 105

6.2 Estimated values for the pultrusion experiment, eccentricity, diameter and missing area . . . 109

6.3 Experimental definition of consolidation experiments. . . 111

B.1 Commands used in PolySlicer & AMCore. . . 172

B.2 Initialization G-Code G-Init. . . 173

B.3 Normal X-Move G-Code (no CFRP), G-NormX-NoCFRP. . . 173

B.4 Extruding with Typhon G-Code, G-ExtrTyphon. . . 174

B.5 No Extrusion X-Move G-Code, G-NoExtrX. . . 174

B.6 Change ScanLine G-Code, G-ChngScnLn. . . 174

B.7 Typhon Positioning for Multi-Track Consolidation G-Code, G-TyphonConsol. . . 175

B.8 Finish up G-Code, G-FinishUp. . . 175

B.9 Change height G-Code, G-ChngZ. . . 175

D.1 Parameters within the DSC experiments. . . 189 G.1 Coupling of process parameters to the quality metrics (a x means most likely none). 217

(20)

(21)

Nomenclature

1D One-dimensional 2D Two-dimensional 3D Three-dimensional

ABS Acrylonitrile Butadiene Styrene ADC Analog-to-digital Conversion AM Additive Manufacturing ATP Automated Tape Laying

C-FRP Continuous Fiber-Reinforced Thermoplastic CAD Computer-aided Design

CAM Computer-aided Manufacturing

CFRTS Continuous Fiber-Reinforced Thermosets CL Connection List

CLT Classical Laminate Theory DIP Dual In-line Package DoF Degree of Freedom

DSC Differential Scanning Calorimetry

EEPROM Electrically Erasable Programmable Read-only Memory FDM Fused Deposition Modeling

FEM Finite Element Method FSM Finite State Machine GUI Graphical User Interface HMI Human-Machine Interface I/O Input/Output

I2_{C Inter-Integrated Circuits}

(22)

XXII NOMENCLATURE

IC Integrated Circuit

ILSS Interlaminar Shear Strength IP65 Ingress Protection Rating 65 ISC In-situ Consolidation

LED Light Emitting Diode LKF LQI with Kalman Filter LMI Linear Matrix Inequality LQI Linear-Quadratic-Integral

MCLV Motion Control Loop Validation

MOSFET Metal-Oxide-Semiconductor Field-effect Transistor MIMO Multiple Inputs - Multiple Outputs

NTC Negative Temperature Coefficient OS Operating System

PCB Printed Circuit Board PEEK Polyether ether ketone PLA Polylactic Acid

PP Polypropylene

PPS Polyphenylene sulfide

PRBS Pseudo-Random Binary Signal PSU Power Supply

PWM Pulse Width Modulation RPi Raspberry Pi

SISO Single Input, Single Output SPI Serial Peripheral Interface

TCLV Thermal Control Loop Validation TCU Thermal Control Unit

UCF User Control Framework UD Uni-Directional

UI User Interface

(23)

Chapter 1

Introduction

This chapter provides an introduction to the work presented in this thesis. Firstly, an overview of 3D1_{printed fiber reinforced thermoplastic structures is provided in Section 1.1. Subsequently,}

the sociotechnical potential of continuous fiber-reinforced thermoplastics (C-FRPs) is explored in Section 1.2. Afterwards, the effects and origin of the porosity content in printed C-FRP structures are discussed (1.3), which leads to the design problem that this work focused on (1.4). Lastly, the methodology and structure of this work are presented in Section 1.5.

1.1 Fused Deposition Modeling of Fiber-Reinforced

Thermoplastics

Over the last decade, additive manufacturing (AM) technologies have matured significantly in terms of the underlying processes, materials, and applications. One AM technology that has received a considerable amount of attention is fused deposition modeling (FDM), which combines a simple process and cheap materials with the ability to produce complex-shaped structures. Although the FDM process has existed since the early 1990s [26], the integration of continuous fiber reinforcements has been introduced more recently. The first fused filament fabrication2

printer that used C-FRPs was released by MarkForged in 2014 [27]. Since then other C-FRP printers have been released such as the printer from Orbital Composites [28] and the MarkForged X7, which are both shown in Figure 1.1. The reason for this interest in C-FRP is their potential to enable the manufacturing of strong structures with a high strength-to-weight ratio with resistance to both corrosion and impacts. These properties make C-FRPs a promising candidate for usage in a broad spectrum of load-carrying applications in industries such as aerospace, automotive, robotics, and prosthetics.

1_{Three-dimensional}

2_{Fused filament fabrication describes the same process as FDM but the latter is a copyrighted term.}

(24)

2 CHAPTER1. INTRODUCTION

Figure 1.1: MarkForged X7 [1] (left) and Orbital Composites printer [2] (right).

C-FRPs consist of a fiber to provide stiffness, and a polymer (often referred to as the matrix/resin). The matrix must hold the fibers together and transfer mechanical loads to them. There are two major types of fiber-reinforced filament in 3D printing applications, namely; short and continuous fibers. These primarily differ in the length of the fibers, which in turn presents a trade-off between mechanical performance and processability, as seen in Figure 1.2. This graph indicates that although C-FRPs are more difficult to process, they offer higher mechanical performance.

(25)

1.2. THEUNTAPPEDSOCIOTECHNICALPOTENTIAL 3

To put the potential of printed C-FRP structures into perspective, Dickson et al. [29] reported that printed structures consisting of nylon with 33vol% glass fiber can achieve tensile strengths that are 33% higher than Aluminum 6061-T6. Whereas their densities are 1786.2 and 2500 kg/m3_,

for the composite3 _{and Aluminum (alloys [32]), respectively. These results highlight the ability}

of fiber-reinforcements to attain both a higher tensile strength (+33%) and lower density (-28%) compared to their metal counterparts.

In the past five years, research in this field has become very active. This research is driven by the mechanical performance of C-FRPs combined with the FDM process’s inherent low cost, handling of complex geometries & rapid prototyping capabilities. These properties make the combination of C-FRPs and the FDM process, a potent combination for a wide range of applications.

1.2 The Untapped Sociotechnical Potential

C-FRPs have the potential to reduce weight and improve lifetime performance while preserving performance for a wide variety of applications. This combination of useful traits has led to the prediction that the usage of composites (including C-FRPs) will increase by 10-15% in the next ten years [33].

Two industries that are increasingly looking into C-FRPs for next-generation applications are; automotive [34], and aerospace [35]. One reason for this interest is that C-FRP’s low weight can aid in reaching the increasingly more stringent demands on fuel efficiency. Additionally, there have been case studies towards using printed C-FRP parts in robotics [36], civil engineering [37], and prosthetics [38]. These industries can benefit from C-FRP’s properties, which can (for example) improve user-friendliness (prosthetics), reduce part count, and inertia (robotics). The low cost of FDM hardware is slowly making C-FRPs a viable option for industries besides the ”conventional” industries of aerospace and automotive.

Currently, conventional production methods are used to manufacture high-performance products such as airplane fuselages [35] and structural components of cars [34]. These methods include, but are not limited to, pultrusion, stamp forming, and tape laying. Although these methods are in general faster than FDM, they are limited in terms of the achievable geometries due to their reliance on a mandrel/mold. Additionally, these molds increase the costs of both prototyping and manufacturing. Thus, a production method capable of highly complex geometries and/or (cheap) prototyping is lacking, which is a role that the FDM process can play.

Thus, realizing the FDM of C-FRP structures would outfit multiple industries with a process that allows for (cheap) rapid prototyping and manufacturing of complex C-FRP structures. Thereby allowing C-FRP to be viable in a much broader spectrum of applications. However, to realize this step the ability to print C-FRP structures with good mechanical properties needs to be realized first.

(26)

1.3 Porosity & 3D Printed Continuous Fiber-Reinforced

Structures

Currently, one of the major issues of printed C-FRP structures is the presence of a high porosity content [4] [39]. To put this content into context, C-FRP structures with more than 2% porosity content are considered a reason for rejection in aerospace applications [40]. Whereas Blok et al. [4] indicated that porosity content of more than 10% is commonly present in printed C-FRP structures, implying that significant progress still needs to be made.

The porosity content is a cause of concern, as it decreases the mechanical performance of C-FRP parts. For instance, the interlaminar shear strength (ILSS), which describes the resistance of the structure to transverse loads, is lowered by the porosity. In Caminero et al. [41], it was observed that increasing the fiber content did not improve the ILSS in the same proportion. They concluded that this discrepancy was due to the higher frequency and size of voids, which increased with the number of fibers. This decrease in ILSS is in line with observations in the field of continuous fiber-reinforced thermosets (CFRTS), which have shown the same correlation [42] [43]. A porosity of 10% would equate to a decrease of the ILSS of 25-38% for CFRTS based on [42] [43]. Due to the similarities between C-FRPs and CFRTS, a similar correlation should be expected.

Besides the ILSS, there is also a strong correlation between voids and a reduction in flexural strength. Hagstrand et al. showed that the flexural strength decreases with 2% per 1% of additional void content [44] . Similarly, Olivier et al. [45] showed that both the longitudinal and transverse tensile strengths decrease for an increasing void content in CFRTS. Although they focused on CFRTS structures, similar correlations are to be expected for C-FRPs.

To illustrate the porosity content in printed structures a micrograph of a cross-section is shown in Figure 1.3, which shows voids as dark spots. It can be observed that a significant amount of voids are present in the printed structure both between and within layers.

Figure 1.3: Micrograph from the cross-section of a multi-layer 3D printed C-FRP structure, with

(27)

1.3. POROSITY& 3D PRINTEDCONTINUOUSFIBER-REINFORCED STRUCTURES 5

The origin of voids in the printed structure can be traced back to two major causes, namely:

1. Input material

• Voids present in the (input) filament, which can grow when the matrix is molten due to the absence of pressure after deposition.

• Inconsistencies in the input material, which can lead to voids when there is an insufficient amount of matrix material.

2. The deposition process

• Inter-track voids, which originate due to insufficient (flow of the) matrix material in a layer. This type of void is especially noticeable turning around corners as seen in Figure 1.4. Blok et al. [4] noted that the printer software sometimes recognized the holes and partially filled it with unreinforced nylon. This mitigation does introduce weak points by having matrix-rich regions.

• Inter-layer voids, which are due to the transition from a circular filament to a flat line. As illustrated in Figure 1.5, the compression of the circular filament will induce diamond and/or triangular voids at the intersection of individual tracks and layers even for un-reinforced filament. These can be mitigated by tuning the interface [46] or using skewed patterning [6]. Additionally, a decrease in layer height could also mitigate this type of void, but this approach increases manufacturing time and compression of a C-FRP filament can be challenging.

(28)

Figure 1.5: Schematic view (left) and micrographs for ABS filament (right) of inter-layer voids

adapted from [5] and [6], respectively.

All these types of voids can grow due to the absence of pressure after deposition, which is called deconsolidation. The effect of deconsolidation is illustrated in Figure 1.6, which shows how the thickness of a plate exposed to different temperatures increases due to deconsolidation. This increase in thickness (up to ≈22.5%) can largely be attributed to the growth of voids [7]. Thereby implying that voids affect not only the mechanical properties but also the geometry of the structure. Although processing times of an individual track in FDM are on a much shorter timescale, the iterative nature of the process can lead to the accumulation of deconsolidation.

Figure 1.6: The increase of thickness due to the deconsolidation of a fiber-reinforced PPS

(Polyphenylene sulfide) laminate exposed to a range of temperatures [7].

Thus, deconsolidation can be seen as a major challenge in the FDM process, which can reduce the mechanical performance significantly and is initiated by the absence of pressure when the matrix material is molten. Thus, the logical next step to integrate C-FRPs into the FDM process is to introduce a consolidation mechanism. This mechanism would supply pressure when the matrix material is molten to reduce the porosity content. The development of such a system is a key step towards printed C-FRP structures with high mechanical performance.

(29)

1.4. THEDESIGNPROBLEM 7

1.4 The Design Problem

Although FDM’s potential to manufacture complex C-FRP structure is recognized, the presence of high porosity/void content in printed C-FRP structures is a notable weakness. This thesis strives to aid research/industry through tackling the design problem of;

Integration of a consolidation (sub)process into the FDM process with minimal impact on other relevant aspects of the printing process.

This design problem is related to the broader technological deficiency of:

The need to improve the mechanical performance of printed C-FRP structures.

Developing a solution to tackle the design problem requires the usage of a sound methodological approach, which will also function as the basis for structuring the thesis.

1.5 Structure of Thesis

The focus of this thesis is on integrating an in-situ consolidation mechanism into the FDM process. The overall approach that was used within this thesis is the V-model, as shown in Figure 1.7. There were three reasons for choosing the V-model:

• Design is structured from an architectural viewpoint (i.e. from system to subsystem and back). Due to the large number of interacting elements in the FDM processes, such a structuring should improve the design flow.

• The inherent coupling of requirements and functions on different detail levels (subsystem or system) allows for a structured evaluation throughout the development cycle.

• Proactive defect discovery by separating the validation of the system and subsystems. The V-model allows the pinpointing of a problem to a subsystem or an interface.

(30)

Figure 1.7: Schematic structure of the thesis.

In Chapter 2 a literature survey is provided focusing on the three major ingredients needed to print high-performance C-FRP structures, which are filament, printer, and consolidation. The former two are discussed in the context of the integration of C-FRP material into the FDM process. Whereas the latter discusses existing solutions for in-situ consolidation in conventional C-FRP manufacturing methods.

Chapter 3 discusses the methods used for establishing the foundation of both systems. By integrating knowledge from literature and the design problem, a system architecture will be defined and requirements set. Subsequently, Chapter 4 focuses on detailing the subsystems for both the filament production setup & AM system. The subsystem designs will start from a functional analysis and discuss the approach taken to design the mechanical, electrical, and software components. Subsequently, the methods used for the validation of the individual subsystems are discussed.

In Chapter 5 the system integration & validation approach is described. This integration is performed through the design of the electronics, user interface, process planning, and control. Subsequently, the integrated systems are verified with regards to the requirements that were set in Chapter 3. Afterwards, in Chapter 6 the output of both systems is analyzed. This validation step will include an investigation of the filament quality and the effect of the developed in-situ consolidation mechanism. Within this chapter, the emphasis is on the evaluation methodology and analysis of the resulting microstructure.

Lastly, in Chapter 7 conclusions are drawn concerning the developed system and consolidation performance. Afterwards, directions for future work are discussed.

(31)

Chapter 2

Literature Survey

Within this chapter, a review of the literature is discussed. This review will focus on relevant aspects of the functional flow of the desired AM process, which is shown in Figure 2.1. Although the manufacturing capabilities of FDM printers primarily stem from a combination of a printhead and 3D motion, the integration C-FRPs in the FDM process requires several modifications. In this chapter, the changes to the deposition method (Section 2.1), production of the filament (2.2), and strategies for handling the -continuous- fibers (2.3) are discussed. Afterwards, in Section 2.4, existing consolidation mechanisms for conventional C-FRP manufacturing methods are discussed. Lastly, Section 2.5 provides a summary of this chapter.

Figure 2.1: Functional flow of the desired FDM process for C-FRPs, pictures taken from [8] [9]

[10] (left to right).

2.1 Deposition Methods for Reinforced Thermoplastics

FDM manufactures complex geometries by depositing molten polymer in a pre-defined shape layer-by-layer. The deposition of molten thermoplastic is achieved by pushing a thermoplastic filament through a heated block (liquefier) and pressing it on top of the layer below. The liquefier

(32)

10 CHAPTER2. LITERATURESURVEY

melts the polymer, which provides the thermal energy to realize bonding. This simple process has already been demonstrated to be a valid method for C-FRPs. For the printing of C-FRP structures there are two common methods for deposition, namely;

1. Co-Extrusion [16], Figure 4.32 (left) - Deposits the fiber and polymer separately by controlling their individual extrusion rates.

Advantage:Allows variation of the matrix fraction by increasing/decreasing its extrusion rate relative to the fiber.

Disadvantage:The in-situ impregnation entails complex flow dynamics during the extrusion. 2. Extrusion of pre-impregnated filament. Herein the printhead extrudes a C-FRP filament, which bonds to previously deposited polymer material using one or multiple printheads. Prominent examples of this approach include dual extrusion (used in the MarkForged printers [4]) and single extrusion [13], Figure 4.32 (middle and right).

Advantage: De-couples the impregnation and printing process. Additionally, the deposition of matrix-rich and fiber-reinforced areas is possible (dual-only).

Disadvantage:Restricts the range of volume fractions and requires a custom filament.

Figure 2.2: Common methods to extrude C-FRPs in the FDM process.

It can be observed that both options have their drawbacks and advantages. In this work, single extrusion will be used, as it allows for the decoupling of filament production and the FDM process. Effectively, this approach removes the transients due to discontinuous extrusion from the FDM process. Additionally, it will require only a single printhead. A pultrusion system was developed to produce the filament, as it had already demonstrated to be a feasible option [13].

2.2 Filament Production

The “single extrusion”-method requires filament that consists of a polymer pre-impregnated with fibers, which -unfortunately- is not readily available for purchase. This filament is, in essence, a cylinder with a length of multiple meters long. For C-FRPs, one dominant manufacturing method capable of producing long structures is pultrusion. Literature shows that pultrusion speeds of 165

(33)

2.2. FILAMENTPRODUCTION 11

mm/s [12] can be reached for a PP-glass fiber rod with a diameter of 2 mm. Thereby, making it a potent process for producing a large amount of filament in a small amount of time. In essence, pultrusion consists of four stages, as seen in Figure 2.3. A concise description of this process is: An input material consisting of fibers and a polymer is pulled through a heated die and formed into the desired shape.The base material in (thermoplastic) pultrusion systems are commingled yarns [11]. These yarns consist of threads of polymer and fiber, which is shown in Figure 2.3. Usage of this material as a base material has already been shown to be feasible for pultrusion/printing combinations [13] and is -relatively- low cost.

Figure 2.3: A schematic view of a pultrusion setup [11] and commingled yarn [12].

During the pultrusion process, the input material undergoes four sequential processes, namely:

1. During the infeed, the commingled yarn is aligned to ensure proper orientation.

2. The pre-heating stage increases the speed of the process by heating the yarn to slightly below its melting point [14]

3. To achieve a high-quality structure the thermoplastic matrix must redistribute to properly impregnate the fibers, which is the role of the (tapered) heated die.

4. To consolidate the part, it is cooled under pressure.

Although these are the four general stages for pultrusion lines, both the pre-heating and cooling steps can be omitted. This reduction was shown to be a viable approach in Rietema [13]. Although a thorough quantification of the effect on filament quality was not performed, it did show that proper impregnation of the commingled yarn is possible at low pulling speeds (≤ 4 mm/s). An example hereof is shown in Figure 2.4, which displays both a microscopic image from [13] and a gauge of impregnation from [14].

(34)

Figure 2.4: A representative image of pultruded filament (microscopic image and impregnation

gauge), adapted from [13] (left) and [14] (right)

Although Miller et al. [12] were much faster (165 mm/s versus 4 mm/s), they incorporated a much higher pulling force (525 N (Miller et al.) versus less than 100 N (Rietema)). Additionally, Miller et al. included, a pre-heater and cooling system followed by a second die making their setup significantly more complex. Therefore, using a setup similar to Rietema would reduce complexity while still allowing for the production of enough C-FRP filament for printing purposes.

Although pultrusion can produce low void content filament, the extrusion inherent to the FDM process will induce significant void growth. Eichenhofer et al. [15] demonstrated this void growth by pultruding samples that were, subsequently, reheated while the radial expansion due to void expansion was measured. Their result (shown in Figure 2.5) clearly shows that the void content rapidly grows shortly after reaching the melting temperature. It can be observed that the void content stabilized shortly after reaching the steady-state temperature of 200◦_{C implying that the}

final void content is dependent on the processing temperature.

(35)

2.3. FIBERHANDLINGSTRATEGIES 13

2.3 Fiber Handling Strategies

The advantage of the high tensile strength of the fibers is that it allows for processes such as pultrusion (i.e. pulling-based processes). However, in the FDM process, this continuity is a problem, as the “normal” extrusion process implicitly assumes that the material flow is completely controllable by stopping/starting the extruder. This assumption is not valid for C-FRPs due to the continuous nature of the fibers, which therefore requires the integration of a fiber handling strategy. In the literature, there are two strategies for handling the fibers. The first method is the “no-cut” strategy. Within this method, the trajectories are planned such that a continuous length of fiber can be used, as shown in Figure 2.6. An advantage of this method is that no time is wasted on cutting. The downside is that this approach limits the printable geometries, as it requires a continuous path to accommodate the fibers. Additionally, it can lead to defects related to fiber steering, which in itself is a complex topic even for conventional C-FRP methods (the interested reader is referred to [47] [48] [49]).

Figure 2.6: An example of a no-cut strategy using hatch spacing [16].

A less complex strategy, from a path planning perspective, is to use intermediate cuts. These cuts can be made, for instance, if the angle between two printing lines is too large [13] or after laying the fibers in a pre-defined concentric ring (MarkForged, see for instance [4]). Deciding on the cutting locations can be integrated into the trajectory generation logic. To realize the cutting of the fibers the common approach is to either cut the fiber pre- or post-printhead as shown in Figure 2.7. These two possibilities and their (dis-)advantages are:

1. Pre-Printhead, which is used in the MarkForged, Figure 2.7 (left) Advantage: No additional time needed for cutting.

Disadvantage: Larger minimal deposition length of fibers required and a mechanism to re-feed the filament to the printhead.

(36)

Advantage: Smaller minimum fiber length, which is 6 mm [13] vs. - at least - the print head length+ critical length (pre-printhead). The critical length is the minimal length for the fiber to stay bonded to the substrate, which is nonzero but there is no fixed value known at this time.

Disadvantage:Time lost on cutting and requires the integration of a mechanism to move the cutter.

Figure 2.7: Representative images of pre- (left) and post-printhead (right) cutting processes.

Integrating a fiber handling and an appropriate extrusion method allows for the usage of C-FRPs within the FDM process. Due to the small number of changes to use C-FRPs, one might expect only minor changes in the final quality. Unfortunately, as discussed in the previous chapter there is a high porosity content in printed C-FRP structures. Thus, to properly integrate C-FRPs into the FDM process a consolidation process needs to be integrated.

2.4 The Consolidation Process & Strategies

The presence of voids in a C-FRP structure can be traced back to either air inclusions in the filament or by the deposition process itself. These voids can grow if insufficient pressure is supplied when the matrix material is molten. Gil et al. [17] investigated the effect of consolidation time on the microstructure. Within their work, they deconsolidated a glass fiber-reinforced PP1

laminate at 180◦_{C and then cooled down the laminate for different amounts of time under pressure.}

The micrographs hereof are shown in Figure 2.8, which shows that void content is reduced when the time under pressure is increased with just 1.28% remaining after 300 seconds [17]. Thereby showing the importance of pressure, temperature, and time to realize proper consolidation.

(37)

2.4. THECONSOLIDATIONPROCESS& STRATEGIES 15

Figure 2.8: Evolution of microstructure for increasing consolidation time, adapted from [17].

In conventional C-FRP manufacturing processes, there are three common strategies to realize consolidation, namely: autoclave, in-situ, and combinations thereof. The autoclave is a process-agnostic strategy wherein the structure is put in a pressurized oven. By combining an elevated temperature and supplying (uniform) pressure to collapse the voids in the C-FRP structure.

However, the use of an autoclave will lead to additional processing time. In contrast, the other group of methods (in-situ consolidation (ISC)) integrates consolidation into the manufacturing process itself. This integration makes it possible to consolidate during the process itself, which reduces manufacturing time. Three ISC methods can be distinguished, namely (note that ”used in” only gives examples of applications);

1. Dual Dies, left part of figure 2.9, which is used in pultrusion [50] [14]

During the process, a commingled yarn is pulled through two dies. The first one melts the matrix and impregnates the fibers, which effectively removes/reduces the voids. Whereas the second stage ensures that solidification of the matrix occurs under pressure. This last step ensures proper consolidation.

2. Press, middle of Figure 2.9, which is used in stamp [51] & press forming [52].

In this method, the structure is formed by applying pressure and adding heat through a tool to melt the laminate. After sufficient time has passed the laminate is cooled down under pressure in the press, thereby consolidating the laminate.

3. Rollers, right of Figure 2.9, which is used in tape laying [53] [54] & filament winding [55] Right before deposition, the input material is heated above its melting temperature, which allows the incoming tape/filament to bond to the layer below it. Within these processes, ISC is achieved through a heated roller, which supplies the consolidation pressure during cooling.

(38)

Figure 2.9: Schematics of existing ISC methods.

Recently2_{, some methods for realizing an ISC strategy in the FDM process have been developed}

in the literature, which indicates that the topic is gaining some momentum. Within the literature two methods were found that are aimed to reduce the void content in the FDM process, namely: using multiple (de)consolidation cycles [15] and a pressure roller [18].

The former, developed by Eichenhofer et al., exploits cyclic softening [15]. Their method relies on the observation that performing multiple cycles of consolidation and de-consolidation will lead to a lower final void content. To validate their claims, they manufactured a printhead wherein the C-FRP material (PA12 with carbon fibers) was pulled through four pultrusion stages (see Figure 2.10). Within their printhead the diameter at the entrance of the next stage larger than the exit of the previous, which induces an initial de-consolidation phase. Their result indicates that significant reductions are possible with ≈ 6 % and ≈ 1.5% void content when deposited after one and four cycles at 1.25 mm/s, respectively. Their work did not evaluate the effect of prolonged exposure to the heating stages during travel, as these can negatively impact the growth of voids and/or degrade the polymer. It should be noted that due to their focus on the input material voids after extrusion, voids related to the deposition process were not characterized.

2_{The papers discussed in this paragraph were added for completeness but were published after the development of}

(39)

2.4. THECONSOLIDATIONPROCESS& STRATEGIES 17

Figure 2.10: Printhead with multiple pultrusion stages to cyclically de-consolidate and consolidate

[15].

Similar to the ”roller”-strategy, Zhang et al. [18] supplied pressure using a heated roller that can rotate around the printhead (see Figure 2.11). Their approach compressed the extruded filament using a heated roller right after deposition. Their printing material was PLA with a carbon fiber content of 10.3%3_{, which is much lower than Eichenhofer et al. (52% volume content).}

Other notable parameters are the low printing speed of 0.3 mm/s, which could negatively impact the overall void content. As the roller’s temperature (180◦_{C) will also affect the other tracks.}

They reported an increase of both tensile and bending strength for an increas in the amount of compression by the heated roller.

(40)

Figure 2.11: Printhead with pressure roller [18].

The conventional manufacturing methods show that the integration of an ISC is possible and can achieve low levels of porosity. Whereas 3D printing - effectively - only deconsolidates the C-FRP structure. Thereby leading to the growth of voids and -subsequently- limited mechanical performance.

2.5 Summary

Within this chapter, literature was presented concerning the realization of the ”deposition”, ”consolidation” and ”filament”-blocks of Figure 2.1. These results can be summarized as:

• Filament - To produce the C-FRP filament a pultrusion system was developed, which pulls a commingled yarn through a heated die that forms the desired shape.

• Deposition of C-FRP - Requires the integration of a fiber handling and extrusion method

– For the extrusion of C-FRP filament two common strategies exist: co-extrusion and

dual/single extrusion. The former extrudes the fiber and polymer separately, which enables control over the local matrix volume fraction. Whereas the latter extrudes a pre-impregnated C-FRP filament, which allows for the decoupling of filament production and the printing process.

– Fiber handling is required to compensate for the continuous nature of the fibers, which

can be done by computing a complex path or by cutting the fiber at intermediate points. The cutting of the fiber can be before or after the printhead, which both require the integration of an additional mechanism (either re-feeding filament or moving a cutter) and have different minimal extrusion lengths.

• Consolidation - Three ISC strategies exist (Dual Dies, Press, and Rollers), which all melt and then solidify the C-FRP under pressure. Providing a basis to solve the design problem of this thesis. Two existing approaches for the integration of consolidation in the FDM

(41)

2.5. SUMMARY 19

process were discussed, namely: a pressure roller and an extrusion method consisting of multiple stages. Both demonstrated their feasibility for integration into the FDM process and improvements on either the void content and/or mechanical strength.

Two sub-designs problems can be identified, namely: the design of a pultrusion system and an FDM printer that can handle and consolidate C-FRP filament. The latter requires the development of both components for the FDM process for using C-FRP filament (extrusion and cutting) and the integration of an ISC mechanism. To develop both systems the next step is to develop an appropriate architecture & an initial set of requirements.

(42)

(43)

Chapter 3

System Requirements &

Architecture

Within this chapter, the design problem is translated into a set of requirements and system architecture. To realize this translation, the systems need to be understood from a functional perspective. The methodology used to understand the systems consisted of four steps, namely:

1. Derive requirements from the perspective of the output. 2. Analyze the process to set system-level requirements.

3. Perform functional analysis to develop the architecture of the system.

4. Set operational requirements, to enable proper functioning of the system in its context. In the previous chapter, it was deduced realizing the FDM process with C-FRPs requires a C-FRP filament and a processing stage (3D printing with consolidation). For the deposition, the single extrusion method will be used as the basis for the AM system. This choice allows the inherent decoupling of the printing and filament production process. In turn, this separation meant that the design problem could be split into two smaller design tasks, namely: the filament production (i.e. pultrusion setup) and the AM system. These design tasks were split into two smaller tasks, namely: filament production & the AM system. For both systems, steps 1 to 3 are elaborated upon in Sections 3.1 & 3.2, respectively. The last step (operational requirements) is discussed in Section 3.3. Lastly, Section 3.4 provides a summary of both the system requirements and architecture.

3.1 Filament Production

The goal of the pultrusion system is to manufacture large quantities (i.e. several meters at a time) of C-FRP filament, faster than the AM system can consume it. To realize this behavior, a set of requirements for the filament and pultrusion process will be defined (Section 3.1.1). Subsequently, an architecture is derived based on existing pultrusion setups (3.1.2).

3.1.1 From Commingled Yarn to Filament

To produce filament a pultrusion setup needed to be developed. From a black-box perspective, this system needs to be capable to shape a commingled yarn into a round filament. The output

(44)

22 CHAPTER3. SYSTEMREQUIREMENTS& ARCHITECTURE

has two aspects that are of importance and defined by the process, namely: shape and void content. The shape and void content are important due to their effects on the FDM process. This reasoning leads to the following requirerement1_:

FIL.1 The filament shall have a consistent round shape.

FIL.2 The filament shall have a consistent fiber/matrix distribution. FIL.3 The filament shall have an initial void content.

Requirement [FIL.1] preserves the design freedom of the FDM process and should prevent clogging (due to for instance changes in the diameter). Whereas [FIL.2] should ensure that matrix material surrounds the fiber to protect from damage by external sources and consistent bonding. Both requirements [FIL.1] and [FIL.2] will be evaluated by analyzing micrographs of the cross-section for the fiber/matrix distribution and diameter, respectively.

Normally, the pultrusion process aims to reduce the porosity to zero. Paradoxically, the void content was intentionally kept higher in the filament for this thesis [FIL.3]. The goal of this ”designed” fault is to validate the consolidation mechanism and would allow for a more thorough evaluation of the ISC mechanism. Specifically, it enables the analysis of the ISC mechanism’s capability to cope with low-quality filament. Effectively, validating the design’s performance for both stabilizing & improving the mechanical properties of 3D printed C-FRP parts.

The requirements above define metrics for evaluating the process’ outcome. However, the system will serve as a supporting system, which means that it must serve the needs of the AM system. Therefore, the requirements on the system-level are focused on the interface with the AM system, which leads to the following set of requirements:

PUL.1 The system shall store the manufactured filament material for later use without damaging the filament (for at least one week).

PUL.2 The system shall produce material at a rate at least equal to the expected printing speed. PUL.3 The system shall allow for the production of ≥5 meters in a single run.

Preparing the filament before printing requires a storage method that does not damage the filament [PUL.1], which was identified as a potential issue in [13]. To support the (subsequent) FDM process it will be necessary to have the pultrusion system produce at rates of at least the printing [PUL.2]. Whereas ≥5 meters of material should allow for enough material to print while producing & storing new filament in parallel [PUL.3]. These requirements are the foundation to which the system should adhere, in terms of process and output. To finalize the foundation of the pultrusion system, the architecture needs to be designed to identify the major subsystems.

3.1.2 Architecture of Pultrusion Setup

Establishing a proper architecture for the pultrusion setup requires understanding the underlying process from a functional viewpoint. By following the path of the yarn in Figure 2.3, the functional flow shown in Figure 3.1 was acquired. It was decided to not include the pre-heater as early-stage testing indicated that it was not necessary (as also seen in [13]).

(45)

3.1. FILAMENTPRODUCTION 23

Figure 3.1: The functional flow of the pultrusion system with requirements allocation (images A,

B, and C from [12], [14], and [8], respectively).

Besides the functional capabilities, a major component in every design is the integration of an appropriate Human-Machine Interface (HMI). Employing an intuitive and simple HMI allows for control over all the process parameters and modules by the user. The addition of the HMI led to the architecture and functional allocation shown in Figure 3.2 (Tm refers to melting temperature

fo matrix material). The interfacing subsystem controls all the other subsystems but the arrows connecting it to other subsystems are omitted to improve clarity. Additionally, the storage and pulling functions were integrated into one subsystem. It can be observed that the envisioned pultrusion system consists of five subsystems, which will be further developed in Chapter 4. With the foundation of the filament production system developed, the next step is to establish a similar foundation for the AM system.

Integration of an In-Situ Consolidation Mechanism into the Fused Deposition Modeling Process for Low Porosity Continuous Fiber-Reinforced Thermoplastic Structures

Integration of an

In-Situ Consolidation Mechanism

into the

Fused Deposition Modeling Process

for Low Porosity

Continuous Fiber-Reinforced

Thermoplastic Structures

Integration of an In-Situ Consolidation Mechanism

into the Fused Deposition Modeling Process

for Low Porosity

Continuous Fiber-Reinforced

Thermoplastic Structures

PDEng Thesis

to obtain the degree of

Professional Doctorate in Engineering (PDEng) at the University of Twente,

on the authority of the rector magnificus,

prof.dr. T.T.M. Palstra,

on account of the decision of the graduation committee,

to be defended

on Thursday the 18th of February 2021 at 14:30 hours

by

Nick Willemstein

born on the 30th of January 1991

in Sliedrecht, The Netherlands

Acknowledgements

Abstract

Contents

List of Figures

List of Tables

Nomenclature

Chapter 1

Introduction

1.1 Fused Deposition Modeling of Fiber-Reinforced

Thermoplastics

1.2 The Untapped Sociotechnical Potential

1.3 Porosity & 3D Printed Continuous Fiber-Reinforced

Structures

1.4 The Design Problem

1.5 Structure of Thesis

Chapter 2

Literature Survey

2.1 Deposition Methods for Reinforced Thermoplastics

2.2 Filament Production

2.3 Fiber Handling Strategies

2.4 The Consolidation Process & Strategies

2.5 Summary

Chapter 3

System Requirements &

Architecture

3.1 Filament Production

3.1.1 From Commingled Yarn to Filament

3.1.2 Architecture of Pultrusion Setup