2.1 Modeling ram extruder
The proposed ram extrusion was designed to effectively control the flow of printing material via two modes: either by (1) pressure or (2) volumetric flow control. When the ram extruder is under pressure control, the volumetric flow rate can be monitored. Vice versa, if the ram extruder is in volumetric flow control, the pressure inside the syringe can be monitored. The actuator, for the ram extrusion, is commanded by a control signal and can provide sufficient force to extrude the material out of the syringe. In this case, a permanent magnet DC motor is preferred, compared to a stepper motor. As a result, the motor model is less complex, and the flexible control mode can be implemented, as described in this section.
The voltage of a DC motor can be determined using the following equation, indicating that the current is dependent on both the voltage and the rotational speed of the motor:
$$V={L}_{m}\dot{I}+ {K}_{V}\omega + {R}_{m}I$$
(1)
where
V denotes the voltage;
I denotes the electrical current, which flows through an armature coil; and
ω represents the motor speed.
In the modeling of the ram extruder, the following parameters and variables are used, as shown in Tables
1 and
2.
Rm | 4 | Ω | Motor resistance |
Lm | 99.5 | µH | Motor inductance |
KV | 0.0261 | V·s/rad | Motor velocity constant |
KT | 0.0261 | N·m/A | Motor torque constant |
J | 0.001 | kg·m2 | System moment of inertia |
f | 30 | N | System friction |
r | 0.001 | m | Screw radius |
α | 4.55 | deg | Lead angle |
µ | 0.05 | N/A | Screw friction coefficient |
D | 0.0148 | m | Syringe inner diameter |
M | 0.08 | kg | Carriage mass |
m | N/A | kg | Material mass |
k | N/A | N/m | Material spring constant |
c | N/A | N·s/m | Material damping coefficient |
X1 | m | Displacement of piston |
X2 | m | Displacement of material entering nozzle |
θ | rad | Angular displacement of screw |
ω | rad/s | Angular velocity of screw |
V | V | Voltage |
I | A | Current |
T | N·m | Motor torque |
F | N | Screw load |
P | Pa | Syringe pressure |
Q | m3/s | Volumetric flow rate |
The resulting torque of the permanent magnet DC motor is proportional to the current
I that flows through the motor coil. The torque
T generated by the motor can be derived from the motor’s electrical current
I as:
If the applied voltage
V is commanded and the motor speed
ω is monitored in real-time, the output torque
T can be approximated by the mathematical model, as described in Eqs. (
1) and (
2).
The motor is designed to run at high speed but only provides a small output torque. Transmission is required to provide mechanical advantages such that the speed is lowered and output torque is increased having the same power rating. In this case, a mechanism to convert the rotational motion of the motor into linear motion that can drive the piston is needed. It is noted that syringes are cylindrical and their pistons move in pure translation. For the proposed design, a ball screw was selected since it can provide mechanical advantages and can convert rotational motion into linear motion. To connect the output shaft of the motor to the ball screw, a flexible coupling is used. The coupling frees other axes except the rotational axis where both shafts are interfaced. In this design, the ball screw ensures rigid transmission between the shaft of the motor and the linear motion of the piston. The feed position of the piston
x1 can be derived by the kinematic relationship of the ball screw [
24]:
$$x_1=r\;\tan\left(\alpha\right)\theta$$
(3)
Although the screw enables a favorable over-motion conversion and mechanical advantage, it exhibits unavoidable friction between the relative surfaces of the nut and screw. Such friction has asymmetric properties, is hard nonlinear, time-variant, and sensitive to disturbance. The point at which friction occurs is between the rotational and translational domains; thus, the complication is amplified.
The coulomb friction model is used for representing the friction in a screw. The angle of friction
γ can be written in the form of the coulomb coefficient
µ as:
$$\gamma=\left\{\begin{array}{c}-\text{tan}^{-1}\left(\mu\right);backward\;drive\\\text{tan}^{-1}\left(\mu\right);forward\;drive\end{array}\right.$$
(4)
It is acknowledged that
γ can be varied by following the driving direction of the mechanism. Hence, the equation of motion of the carriage can be defined as:
$$\begin{array}{c}\left(Mr\;tan\left(a+\gamma\right)+\frac1{rtan\left(a\right)}\right){\ddot x}_1\\=T-Fr\;tan\left(a+\gamma\right)\end{array}$$
(5)
If the motion of the system is determined by the effort from the rotation of the screw, the direction is called forward drive. Vice versa, if the effort from the nut drives the system, the direction is backward. Mechanical advantage varies depending on the driving direction, whereas backward driving delivers more mechanical advantage. The reason behind this effect lies in the interaction of the screw’s nonlinear behavior and the friction. By operating in a backward drive, the negative value of \(\gamma\) reduces the value of \((\alpha +\gamma )\) inside the tangent function. This reduction reduces the influence of load \(F\), generating a higher mechanical advantage. As for a screw with a small lead angle \(\alpha\), the effect of negative \(\gamma\) is large enough such that \((\alpha +\gamma )\) is zero or negative. This outcome prevents the effect of the action on the nut side to the system, which is called non-back drivable. The screw that possesses this behavior has self-locking ability.
Loaded material is typically viscoelastic such that its behavior is a blend between viscous fluid and elastic solid. To precisely control the flow of such material, two material properties are implemented. The first one is elastic modulus, which explains the elastic response of the material. The second one is viscosity, which describes the resistance of a fluid to flow. The fluid aspect of viscoelastic material that is mostly used in LDM is a non-Newtonian fluid; the relationship between shear rate and shear stress is nonlinear [
2]. Thus, the nonlinear resistance of the flow is exhibited.
As for the aspect of modeling, the lumped element model is used to describe the material. Elasticity is represented by a spring whose spring constant k corresponds to the elastic modulus of the material. The viscous element can be modeled as a damper with a nonlinear damping coefficient \(c(\dot{{x}_{2}})\), corresponding to the viscosity, where \({x}_{2}\) is a displacement of material that enters the nozzle. It is assumed that the elastic behavior is dominated only in the syringe, while the viscous effect is mainly presented in the nozzle. A small amount of extruded mass is neglected. The spring is connected to the damper in series; a mass of the material that is inside the nozzle m is attached between them.
Let the displacement of material that is entering the nozzle be
\({x}_{2}\). The load that acts as a piston can be defined by the spring effect inside the syringe as:
$$F=f+k({x}_{1}-{x}_{2})$$
(6)
where
f is the overall friction, which resists the system’s motion. The equation of motion can then be derived as follows:
$$m{\ddot{x}}_{2}={\text{k}}\left({x}_{1}-{x}_{2}\right)-{\text{c}}(\dot{{x}_{2}})\dot{{x}_{2}}$$
(7)
where
\(c(\dot{{x}_{2}})\) is a nonlinear damping coefficient:
$$c(\dot{x_2})=c_1+c_2\;sgn\;(\dot{x_2})\dot{x_2}$$
(8)
When all the governing Eqs. (
1)–(
8) have been presented, a stability analysis of the system is carried out. Lyapunov’s second method can be used to guarantee the stability of nonlinear dynamical systems. Subsequently, the Lyapunov candidate function
\(\Upsilon \left(X\right)\), as represented by the summation of the kinetic energy of the screw and material, electrical energy in the motor, plus the energy from viscoelastic material deformation, yields [
25]:
$$\Upsilon \left(X\right)=\frac{k\hspace{0.17em}{\left({x}_{1}-{x}_{2}\right)}^{2}}{2}+\frac{{L}_{m}\hspace{0.17em}{I}^{2}}{2}+\frac{{m}_{t}\hspace{0.17em}{ \dot{{x}_{1}}}^{2}}{2}+\frac{m{\dot{{x}_{2}} }^{2}}{2}$$
(9)
Such a function is in a quadratic form, holding the condition at stable point: \(\Upsilon \left(0\right)= 0\), and the positive definition becomes \(\Upsilon \left(X\right)= 0\) if \(X \ne 0\).
The derivative form of the candidate function is derived as follows:
$$\dot{\Upsilon }\left(X\right)={J}_{t}\hspace{0.17em}\ddot{{x}_{1}}\hspace{0.17em}\dot{{x}_{1}}+k\hspace{0.17em}\left( \dot{{x}_{1}}-\dot{{x}_{2}}\right)\hspace{0.17em}\left({x}_{1}-{x}_{2}\right)+m\hspace{0.25em}\ddot{{x}_{2}}\hspace{0.17em}\dot{{x}_{2}}+{L}_{m}\hspace{0.17em}I\dot{\hspace{0.17em}I}$$
(10)
Substituting Eqs. (
1)–(
7) in the derivative form of the candidate function and voltage
V =
0 results in:
$$\dot{\Upsilon }\left(X\right) =-{R}_{m}{I}^{2}-c(\dot{{x}_{2}}) \hspace{0.17em}{\dot{{x}_{2}}}^{2}-f\dot{{x}_{1}}-\left(\frac{K\hspace{0.17em}I \dot{{x}_{1}}}{r\hspace{0.17em}\mathit{tan}\left(\alpha \right)}-\frac{K\hspace{0.17em}I\hspace{0.17em}\dot{{x}_{1}}}{r\hspace{0.17em}\mathit{tan}\left(\alpha +\gamma \right)}\right)$$
(11)
To prove Lyapunov’s stability [
25], the state must converge at a stable point so that
\(\dot{\Upsilon }\left(X\right) < 0\). It is noted that in the case of asymptotic stability, such conditions must hold true for
\(X \ne 0\).
The first term
\({R}_{m}{I}^{2}\) is a power loss by the resistance of the motor, which is always positive. The second term
\(c(\dot{{x}_{2}}) \hspace{0.17em}{\dot{{x}_{2}}}^{2}\) is a loss in the nonlinear damper:
$$c(\dot{x_2})\dot{x_2}^2=(c_1+c_2\;sgn\;(\dot{x_2})\dot{x_2})\dot{x_2}^2$$
(12)
By expanding
\(sgn (\dot{{x}_{2}})\), the term can be written as:
$$c(\dot{{x}_{2}}){\dot{{x}_{2}}}^{2}={c}_{1}{\dot{{x}_{2}}}^{2}+{c}_{2} \frac{{\dot{{x}_{2}}}^{4}}{\left|\dot{{x}_{2}}\right|}$$
(13)
and proves to be positive at
\(\dot{{x}_{2}}\ne 0\).
When considering friction loss
\(f\dot{{x}_{1}}\), the direction of friction always resists the motion of
\({x}_{1}\). Thereby, coulomb viscous friction occurs and is defined as:
$$f=sgn\left(\dot{{x}_{1}}\right)\left|f\right|=\frac{\dot{{x}_{1}}}{\left|\dot{{x}_{1}}\right|}\left|f\right|$$
(14)
Consequently,
\(f\dot{{x}_{1}}\) can be written as:
$$f\dot{{x}_{1}}=\frac{{\dot{{x}_{1}}}^{2}}{\left|\dot{{x}_{1}}\right|}\left|f\right|$$
(15)
This term is always positive. The fourth and fifth terms of Eq. (
11) can be expressed as follows:
$$\frac{K\hspace{0.17em}I\dot{x_1}}{r\hspace{0.17em}\mathit{tan}\;\left(\alpha\right)}-\frac{K\hspace{0.17em}I\hspace{0.17em}\dot{x_1}}{r\hspace{0.17em}\mathit{tan}\;\left(\alpha+\gamma\right)}=\frac{K\hspace{0.17em}I\dot{x_1}}{r\hspace{0.17em}}\cdot\left(\frac{\mathit{tan}\;\left(\alpha+\gamma\right)-\mathit{tan}\;\left(\alpha\right)}{\mathit{tan}\;\left(\alpha\right)\;\mathit{tan}\;\left(\alpha+\gamma\right)}\right)$$
(16)
According to the direction of the driving, i.e., forward and backward drive, it is seen that the friction angle
\(\gamma\) varies. Direction can be assumed to be dependent on the energy flow of the screw. The power of the motor
\(K\hspace{0.17em}I \dot{{x}_{1}}\) can be an indicator of the driving direction. If
\(K\hspace{0.17em}I \dot{{x}_{1}}\) is positive, the motor gives power to the system, and the screw is in forward drive. Thus, the friction angle can be written case-wise:
$$\gamma = \left\{\begin{array}{c} -{\gamma }_{abs} ; K\hspace{0.17em}I \dot{{x}_{1}} < 0\\ {\gamma }_{abs} ; K\hspace{0.17em}I \dot{{x}_{1}} > 0\end{array}\right.$$
(17)
In the case of
\(K\hspace{0.17em}I \dot{{x}_{1}} < 0\)$$\left(\frac{\mathit{tan}\left(\alpha -{\gamma }_{abs}\right)-\mathit{tan}\left(\alpha \right)}{\mathit{tan}\left(\alpha \right)\mathit{tan}\left(\alpha -{\gamma }_{abs}\right)}\right) < 0$$
(18)
and when
\(K\hspace{0.17em}I \dot{{x}_{1}} > 0\),
$$\left(\frac{\mathit{tan}\left(\alpha +{\gamma }_{abs}\right)-\mathit{tan}\left(\alpha \right)}{\mathit{tan}\left(\alpha \right)\mathit{tan}\left(\alpha +{\gamma }_{abs}\right)}\right) > 0$$
(19)
In these two cases, the sum of the two terms is always positive:
$$\frac{K\hspace{0.17em}I \dot{{x}_{1}}}{r\hspace{0.17em}\mathit{tan}\left(\alpha \right)}-\frac{K\hspace{0.17em}I\hspace{0.17em}\dot{{x}_{1}}}{r\hspace{0.17em}\mathit{tan}\left(\alpha +\gamma \right)} > 0$$
(20)
Overall, the four terms regarding the derivative of the Lyapunov candidate function are always negative. Therefore, at any initial state \(X\), the system tends toward stability. Thus, the energy that is stored in the system is always reduced.
After the stability of the system is discussed, the ability to sense and control both the pressure and volumetric flow rate of the material is presented. When in motion, both pressure P and volumetric flow rate Q are visible on the actuator’s side. Yet, when the motor and the piston stop, and flow ceases, the friction is in static friction. Friction behavior falls in the dead-band, and the pressure P is invisible on the actuator’s side.
The force pushing the material through the nozzle is analogous to the force exerted by the spring which is equal to
F −
f. The pressure of the material at the nozzle entrance
P can be estimated as:
$$P=\frac{4\left(F-f\right)}{\pi {D}^{2}}$$
(21)
Equations (
1)–(
5) show that the force at the piston
F can be calculated by the controllable voltage
V and the measured motor’s position
θ. The force can then be used in Eqs. (
6) and (
21) to obtain the volumetric flow rate
Q (directly proportional to
x2) and pressure
P, respectively. This proves that by measuring and controlling only two variables: the applied voltage
V and the motor’s position
θ, both the pressure
P and volumetric flow rate
Q of the system can be derived and manipulated at any given time. However, since both are directly affected by the same variables, only one of them can be controlled at a time:
$$P=\frac{\hspace{0.17em}4\left(\frac{{K}_{T}\hspace{0.17em}\left(V-{K}_{V}\omega \right)}{{R}_{m}}-f\hspace{0.17em}r\hspace{0.17em}{\text{tan}}\left(\alpha +\gamma \right)\right)}{{D}^{2}\hspace{0.17em}r\hspace{0.17em}\pi \hspace{0.17em}{\text{tan}}\left(\alpha +\gamma \right)}$$
(22)
Furthermore, at steady state, the elastic behavior of the material is disregarded, meaning that the rate of change for
x1 and
x2 are coupled. Thus, the volumetric flow rate
Q of the material can then be derived from the motor’s speed
ω via Eq. (
3). By applying the closed-loop controller over the measured motor’s position
θ, the steady-state volumetric flow rate
Q can be directly handled. Similarly, at steady state, only the voltage is required to control the pressure. It is important to note that the pressure inside the nozzle is the gradient from the pressure
P to the atmospheric pressure:
$$Q=\frac{\pi \hspace{0.17em}{D}^{2}\hspace{0.17em}\dot{{x}_{2}}}{4}$$
(23)
By rearranging all the equations, only the motor’s speed
ω is required to approximate the volumetric flow rate
Q at steady state, as follows:
$$Q=\frac{\pi \hspace{0.17em}{D}^{2}\hspace{0.17em}r\hspace{0.17em}\mathit{tan}\left(\alpha \right)\hspace{0.17em}\omega }{4}$$
(24)
Subsequently, the pressure P and the volumetric flow rate Q at steady state is thus determined via (1) the applied voltage V, (2) the time derivative of the motor’s position θ, and (3) the model.
It is noted that if both the pressure
P and the volumetric flow rate
Q are controlled, the printing process is enhanced [
22]. The technique of flexible actuation is an excellent candidate to control the flow of the material.