Article

Induction-Heated, Unrestricted-Rotation Rectangular-Slot Hot

End for FFF
Miguel Rodríguez 1 , David Blanco 1, * , Juan Antonio Martín 2 , Pedro José Villegas 2 , Alejandro Fernández 1
and Pablo Zapico 1

1 Department of Construction and Manufacturing Engineering, University of Oviedo, 33204 Gijón, Asturias, Spain
2 Department of Electrical, Electronic, Communications, and Systems Engineering, University of Oviedo,
33204 Gijón, Asturias, Spain
* Correspondence: dbf@[Link]

Abstract
This work presents a fused-filament fabrication (FFF) hot end that combines an unrestricted-
rotation C-axis with a rectangular-slot nozzle and an induction-heated melt sleeve. The
architecture replaces the popular resistive cartridge and heater block design with an external
coil that induces eddy-current heating in a thin-walled sleeve, threaded to the heat break
and nozzle, reducing thermal mass and eliminating wired sensors across the rotating
interface. A contactless infrared thermometer targets the nozzle tip; the temperature is
regulated by frequency-modulating the inverter around resonance, yielding stable control.
The hot end incorporates an LPBF-manufactured nozzle, which transitions from a circular
inlet to a rectangular outlet to deposit broad, low-profile strands at constant layer height
while preserving lateral resolution. The concept is validated on a desktop Cartesian
platform retrofitted to coordinate yaw with XY motion. A twin-printer testbed compares
the proposed hot end against a stock cartridge-heated system under matched materials and
environments. With PLA, the induction-heated, rotating hot end enables printing at 170 ◦ C
with defect-free flow and delivers substantial reductions in job time (22–49%) and energy
per part (9–39%). These results indicate that the proposed approach is a viable route to
higher-throughput, lower-specific-energy material extrusion.
Academic Editor: Fulvio Lavecchia

Received: 3 November 2025

Keywords: fused filament fabrication; induction heating; rotating nozzle
Revised: 4 December 2025
Accepted: 10 December 2025
Published: 13 December 2025

Citation: Rodríguez, M.; Blanco, D.; 1. Introduction

Martín, J.A.; Villegas, P.J.; Fernández, Fused-filament fabrication (FFF) is the dominant material extrusion (MEX) desktop
A.; Zapico, P. Induction-Heated,
modality. It employs a heated nozzle to melt and extrude unfilled or composite thermo-
Unrestricted-Rotation
plastic filaments to form continuous strands according to programmed toolpaths [1]. As of
Rectangular-Slot Hot End for FFF. J.
Manuf. Mater. Process. 2025, 9, 409.
2024, FFF represents approximately 45–50% of desktop-printer unit sales [2]. Yet, despite
[Link] its growing role in prototyping and short-run production [3,4], protracted manufacturing
jmmp9120409 times and high specific energy prevent it from being a cost-effective solution [5–7]. In fact,
Copyright: © 2025 by the authors.
a modest 100 cm3 part in polylactic acid (PLA) can require up to 2.4 kWh of energy [8].
Licensee MDPI, Basel, Switzerland. The situation worsens when processing higher-temperature polymers [9,10]. Therefore, to
This article is an open access article compete in medium and large batches, faster and lower energy printing is critical [5,11].
distributed under the terms and To optimize consumption, statistical, physics-guided or hybrid models have been
conditions of the Creative Commons
proposed to relate geometry and printing parameters to energy use [12–16]. All of them
Attribution (CC BY) license
show that printing speed and layer height are the primary drivers of energy demand,
([Link]
licenses/by/4.0/).
with temperature being a distant third [17,18]. Therefore, three sorts of strategies arise to

J. Manuf. Mater. Process. 2025, 9, 409 [Link]

J. Manuf. Mater. Process. 2025, 9, 409 2 of 22

minimize consumption, but all of them impose quality penalties: (i) increasing the layer
height, which degrades surface finish and interlayer strength [15,17], (ii) raising the feed
rate, which promotes shear-induced defects such as “shark-skin” [19] and (iii) boosting
volumetric flow by adopting larger-diameter circular nozzles (at the expense of lateral
resolution and corner fidelity) and often driving conventional cartridge heaters beyond
their practical thermal delivery capacity [20,21].
An alternative route to higher productivity is to widen the extruded road while
maintaining a constant layer height. This strategy preserves the nominal vertical resolution
and staircase pitch in the z-direction yet increases the in-plane deposition rate and reduces
build time by covering more area per pass. A rotating rectangular slot in the nozzle
can deposit anything from a knife-thin trace to a broad “ribbon”, since the projection
of the slot in the build plane changes with the yaw angle. First sketched in industrial
patents, yaw was coordinated on the fly to print wide tracks for infill and narrow ones for
detail [22]. Subsequent experimental studies have tightened the yaw mechanics and control
loops [23,24], quantified how orientation dictates strand geometry and accuracy [25,26]
and extended the idea to multi-axis platforms with rotation-aware tool paths [27].
Even if the longer edge of the slot stays tangential to the tool path (no width modula-
tion), a rectangular nozzle can still dispense strands several times wider than the typical
0.4 mm circular counterpart. Yet, two practical obstacles appear. On the one hand, sus-
tained rotation twists the filament, the wiring and the cooling lines. This hampers smooth
re-orientation around closed contours or sharp corners, degrading surface finish and dimen-
sional fidelity [27]. On the other hand, the higher volumetric flow overwhelms conventional
cartridge heaters [28], while thermal inertia lengthens the warm-up times [29], inflates idle
power draw [30] and harms the response to flow transients [31].
Induction heating offers a viable alternative to the thermal limitations by delivering
contactless, volumetric heat with low thermal inertia, thereby improving the thermal re-
sponse while reducing parasitic losses. Prototype studies have achieved millisecond-scale
temperature transients and more homogeneous melt zones by optimizing coil geome-
try, incorporating high-frequency in situ sensing, and implementing closed-loop power
control [32–34]. In addition, the only inductive hot end currently on the market, to our
knowledge, for MEX reports substantially lower thermal inertia and a reduced moving
mass, compared with conventional resistive designs [35]. However, to the best of our
knowledge, there is currently no hot end design that simultaneously combines unrestricted
C-axis rotation, a rectangular-slot nozzle for wide strands and a low-inertia induction-
heated melt zone with closed-loop temperature control. In addition, the consequences of
such an architecture for printing time and energy use have not yet been systematically
evaluated relative to conventional cartridge-heated hot ends.
The larger cross-section of rectangular-slot strands increases the thermal load and
demands a fast, well-controlled heat input to maintain a homogeneous melt at higher
volumetric flow rates; in conventional cartridge-heated blocks, especially when mounted
on a rotating head, this leads to heavy, thermally inert assemblies that are difficult to wire
and limit both throughput and the practicality of unrestricted C-axis rotation. Induction
heating is particularly attractive in this context because it concentrates heat generation
in a thin metallic sleeve with low thermal inertia, enabling a rapid thermal response
and efficient power delivery while keeping the inductor and power electronics stationary.
The combination of a rectangular-slot nozzle with an induction-heated sleeve is therefore
synergistic: the former exploits wider strands to reduce the toolpath length, while the
latter provides the responsive, lightweight heating required to sustain such deposition in a
rotating architecture.
J. Manuf. Mater. Process. 2025, 9, 409 3 of 22

Here, both techniques are integrated into a single structure: a rectangular nozzle is
fitted to a continuously rotating C-axis with unlimited yaw control, and a non-contact
induction heater is designed with closed-loop infrared thermometry. Working in concert,
broad, low-profile strands are placed at a constant layer height, mitigating the need for
multi-pass perimeter deposition. At the same time, warm-up delays and standby energy
losses are reduced significantly. The concept is validated on a desktop Cartesian FFF printer
retrofitted with a prototype of the new Unrestricted-Rotation Hot End (URHE), which
is benchmarked against an unmodified machine. Test parts are manufactured with PLA.
Comparative performance is quantified by the percentage reduction in manufacturing time
and the percentage of plug-level energy savings per part.

2. Unrestricted-Rotation Hot End: Design and Optimization

2.1. Hot End System Architecture
The basic architecture of the URHE (Figure 1) integrates four subsystems: (i) an
unrestricted-rotation C-axis for continuous yaw alignment of a rectangular nozzle, (ii) an
induction-heated melt sleeve driven by an inverter, (iii) contactless infrared temperature
sensing and control, and (iv) annular, quasi-isotropic part cooling.

Figure 1. Overview of the induction-heated rotating hot end. Cutaway rendering, showing the
different subsystems in the proposed hot end design.

2.1.1. Hot-End Stack and Rotation (C-Axis)

As in common commercial designs (Figure 2a), the hot end stack features a nozzle, a
thin-walled thermal barrier or heat break, a finned heat-sink and a guidance polytetrafluo-
roethylene (PTFE) tube.
It differs, however, in the heating system. The resistance heater cartridge embedded
in an aluminum block has been substituted by a non-rotating external coil which induces
eddy currents in a hollow rotating component or sleeve (Figure 2b), into which both the
nozzle and the heat break are threaded.
Heat is generated in situ by Joule heating uniformly along the sleeve. Through conduc-
tion, the filament melts within the central channel, establishing a symmetric liquefaction
zone and allowing more homogeneous fusion of the material. The arrangement preserves
nozzle interchangeability, eliminates all the wires in the rotating interface and reduces the
overall thermal mass.
J. Manuf. Mater. Process. 2025, 9, 409 4 of 22

(a) (b)

Figure 2. Hot-end architectures: (a) conventional FFF hot end with aluminum heater block and
resistive cartridge; (b) proposed URHE with an external coil inducing eddy-current heating in a
thin-walled sleeve.

The interchangeable nozzle was designed with an upper 3.00 mm cylindrical inlet
that morphs into a 1.20 × 0.40 mm rectangular outlet (Figure 3a,b). The 0.40 mm mi-
nor dimension was preserved so that the effective lateral resolution matches that of a
conventional 0.40 mm nozzle. To promote stable flow and reduce pressure drop, the
circular-to-rectangular transition was modeled as a smooth, curvature-continuous profile,
generated by a lofted cut between intermediate cross-sections (Figure 3b). Because this
geometry is difficult to reproduce reliably at this scale with conventional machining, the
nozzle was manufactured by laser powder bed fusion (LPBF) in maraging steel M300 and
then threaded for interchangeability (Figure 3b). Upstream, a Ti-6Al-4V heat break is co-
bored and chamfer-matched to the nozzle to form a continuous flow channel, minimizing
ledges at the cold–hot interface.

(a) (b) (c)

Figure 3. Nozzle orifice. (a) Design overview. (b) Dimensions of the LPBF-manufactured nozzle.
(c) Micrograph of the M300 LPBF-fabricated tip, showing the rectangular-slot outlet.
J. Manuf. Mater. Process. 2025, 9, 409 5 of 22

The finned heat-sink serves as the structural hub of the C-axis transmission, acting as
the bearing journal and torque carrier for the rotating subassembly. The set is supported
by a single-row deep-groove ball bearing (Figure 1), which promotes concentric rotation
about the filament axis with low runout during sustained yaw. Upstream of the bearing,
a GT2 timing pulley is assembled on the C-axis shaft and driven by a NEMA-8 stepper
(Nanotec Electronic GmbH & Co. KG, Kapellenstr, Bavaria, Germany.) via a belt (Figure 1).
It provides the required reduction while it isolates the motor mass and vibration from
the rotating hot end components. Finally, the heat-sink is cooled by a fan placed on the
stationary portion of the extruder head.

2.1.2. Induction Heating Assembly

For inductive heating, the sleeve acts as the core of a solenoid. The concentric, external
winding placed around it is excited by a sinusoidal current which produces a magnetic
field of the same frequency. As a result, eddy currents are induced on the core’s outer
surface, with a penetration or skin depth δ (Equation (1)) that decays exponentially in radial
direction and depends on the current frequency ( f ) and on the electric conductivity (σ) and
magnetic permeability (µ) of the material.

1
δ= p (1)
π f σµ

A suitable combination of excitation frequency, geometry and sleeve material sets the
eddy-current path resistance and produces Joule heating in the hot end. Because induction
is confined to a thin surface layer, we adopt a cylindrical heater sleeve with a wall thickness
of a few δ at the operating frequency to maximize efficiency, minimize thermal mass, and
maintain uniform azimuthal coupling. The sleeve simultaneously serves as the mechanical
interface for the threaded nozzle and heat break that encloses the melt channel that acts as
a liquefaction chamber; accordingly, the electromagnetic design employs an axisymmetric,
concentric induction coil to establish a uniform circumferential field with strong coupling.
An effective sleeve must concentrate magnetic flux (high permeability with shallow
skin depth), enable strong eddy-current heating at tens of kilohertz (appropriate resistivity),
conduct heat to the melt channel (adequate thermal conductivity) and remain ferromagnetic
across the operating window (Curie point well above extrusion setpoints); taken together,
these requirements make AISI 420 martensitic stainless steel (1.4021 (X20Cr13)) a balanced
choice. Ferromagnetic, with a moderate electric resistivity (55 µΩ·cm) [36], a thermal
conductivity of 24.9 W·m−1 ·K−1 [36] and a Curie point above 600 ◦ C [37], provides a
convenient electric resistance in the eddy current path, an adequate heat diffusion and
stable magnetic performance at typical temperatures of PLA extrusion (170–230 ◦ C) [38].
To size the induction sub-system, we estimated the required thermal power under a
deliberately conservative assumption. Because the thermophysical properties of the se-
lected PLA were not precisely known, we adopted worst-case values corresponding to fully
crystalline PLA (100% crystallinity) from Pyda et al. [39] and a density of 1.290 g·cm−3 [40].
For a representative deposition condition (0.2 mm layer height, and 50 mm·s−1 print speed)
assuming the rectangular slot design (1.2 mm strand width), the theoretical heat input to
raise the polymer from 20 ◦ C to 220 ◦ C is 6.2 W. An iterative mechanical design fixed the
heater as a cylindrical induction sleeve (external diameter: 8 mm, internal thread (through
hole): M6; axial length: 20.5 mm). For this geometry, parasitic losses by conduction, con-
vection and radiation were estimated at 5.6 W. Accordingly, the total estimated power
requirement is 11.8 W.
Additionally, we performed thermal simulations using ANSYS Electronics (2022) to
corroborate the analytical estimates. Because the electromagnetic skin depth is small, the
J. Manuf. Mater. Process. 2025, 9, 409 6 of 22

induced Joule heating was modeled as an equivalent surface heat flux, rather than a volu-
metric source. In the simulation, the thermal properties of the materials used (previously
mentioned) were considered, along with an ambient temperature of 25 ◦ C and the simula-
tion of a 2.2 cfm fan flow. Radiation is neglected as a second-order effect for the temperature
range considered. The mesh consisted of more than 1.5 million elements (tetrahedral), with
a maximum element size of 1 mm and a minimum of 0.05 mm. A k–ε RNG turbulence
model was used.
Figure 4 presents the simulated temperature field for the same operating conditions
used in the theoretical calculations.

Figure 4. Thermal simulation of the hot end: temperatures reached with a heat power of 11.8 W for
extruding PLA with a 50 mm·s−1 printing speed.

The simulation predicts a maximum nozzle temperature of 228.6 ◦ C (Figure 5), which
is in good agreement with the 220 ◦ C target (relative error 3.9%). Deviation likely arises from
modeling simplifications in the theoretical estimate, like idealized geometry, temperature-
independent properties, and approximate boundary/contact conditions. Based on these
results, we specified a design margin and sized the induction subsystem to deliver up to
15 W, providing adequate headroom over the 11.8 W estimate.

Figure 5. A simulation of the power generated in the hot end with a current of 3.5 A RMS at 75 kHz.
J. Manuf. Mater. Process. 2025, 9, 409 7 of 22

Power to be transferred by magnetic coupling was analyzed with finite-element elec-

tromagnetics (FEMM 4.2 and ANSYS Electronics 2022). To size the inductor, we designed a
13-turn solenoid wound with Litz wire (170 × 0.08 mm), mean diameter 15 mm and axial
length 20 mm, arranged to couple uniformly with the sleeve along the melt zone. The coil
is wound around a flanged PEEK bushing, whose section has an inner diameter of 14 mm,
so the radial distance to the outer surface of the sleeve is 3 mm.
The operating frequencies (>66 kHz) are high enough for efficient eddy-current heating,
yet low enough to simplify the resonant driver and limit electromagnetic interference. With
the core installed, the measured coil inductance is L1 = 2.23 µH at 70 kHz. A 24 V
dc half-bridge inverter drives a series-LC tank formed by the coil, an external inductor
L2 = 10 µH, and a capacitor C = 470 nF, setting the resonance near 66 kHz. The
inverter operates slightly above resonance to achieve soft switching and to limit the peak
tank current.
At 75 kHz (well above resonance), the current is 3.5 A RMS, which FEMM predicts
corresponds to 3.8 W of sleeve heating (Figure 5). A frequency-control loop tunes the
inverter: lowering the drive frequency toward resonance increases the series-LC current
and thus, the delivered power. To prevent excessive hot end temperatures near resonance,
the maximum heating power is software-limited to 15 W, reached at 68.5 kHz.
An electronic prototype and a simplified hot end (without the rotary assembly) were
built to validate the inductive heating design. The inverter used two IRF540 MOSFETs
driven by an IR2111 gate driver. At 75 kHz, the system delivered a quasi-sinusoidal
3.5 A RMS current to the heater winding. The measured inverter input power was 5.4 W,
in reasonable agreement with the theoretical prediction (Figure 5) after accounting for
semiconductor conduction losses and winding resistance.

2.1.3. Contactless Infrared Thermometry and Frequency-Modulated Control

Reliable temperature feedback is essential for stable extrusion, but the unrestricted
rotation of the head rules out wired junction sensors (thermocouples/thermistors). Slip
rings were considered and rejected because rotating leads and junction are susceptible to
electromagnetic interference from the induction field, parasitic heating, and wear.
The adopted solution is a contactless infrared radiation (IR) thermometer MLX90614-
DCI (Melexis, Ieper, Belgium) placed on the stationary case at a shallow angle so that its
field of view intersects the nozzle tip (Figure 6). This sensor operates from 3.3 V over an
I2 C/SMBus digital link, reporting object and ambient temperatures. Narrow field of view
optics (5◦ ) and an internal chopper provide adequate immunity to the local electromagnetic
environment. A two-point calibration (170 ◦ C and 220 ◦ C) against a contact reference
established an effective emissivity ε = 0.80 for the nozzle. After calibration, residual error
during static operation was within ±2–3 ◦ C.
Because inductive power is a strong function of inverter frequency, temperature control
is implemented using this parameter. For the initial warm-up, the inverter operates near
resonance, 68.5–69.7 kHz, providing a large amount of power for rapid transients. Upon
reaching the setpoint, the corresponding frequency is stored as f m . Then, closed-loop
adjustments control the frequency above f m with proportional steps to maintain error
below ±2 ◦ C. Low-temperature variations (|∆T | > 5 ◦ C ) trigger a safeguarded fallback
(fixed high-power frequency), while overshoot momentarily inhibits drive to prevent
runaway. The loop updates every second, filtering the IR signal with a moving average to
suppress jitter.
J. Manuf. Mater. Process. 2025, 9, 409 8 of 22

(a) (b) (c)

Figure 6. Temperature control: (a) IR sensor orientation; (b) simulated velocity field of the ad hoc
designed shroud. (c) Shallow water flow visualization beneath the annular shroud, showing an
approximately axisymmetric disturbance of the surface under the cooling-air jets.

2.1.4. Azimuthally Uniform Layer Cooling

In conventional FFF heads, one or two blowers emit lateral jets toward the nozzle
tip. As these jets come from a fixed azimuth, cooling and therefore solidification of the
strands depend on its alignment with the air flow. This asymmetry is amplified when using
a freely rotating head and a non-circular outlet, because their relative orientation changes
continuously.
To avoid cooling directionality effects, the design incorporates an annular layer-cooling
shroud (Figure 1). It delivers uniform airflow from the azimuth independently of the
yaw angle. The shroud is fixed to the non-rotating portion of the head, surrounds the
nozzle and incorporates evenly spaced radial outlets, aiming at the strand just below.
A circumferential plenum is internally tapered to compensate pressure drops along the
manifold, so that all ports discharge at nearly the same velocity and flow rate. The duct
was dimensioned for fabrication by FFF and evaluated via steady-state computational fluid
dynamics (SolidWorks Flow Simulation 2022). The inlet condition was set to 80% of a
24 V dc blower nominal flow (5.36 cfm). Simulations (Figure 6b) show matching exit speeds
across all ports, which indicate uniform cooling regardless of nozzle yaw. Nevertheless, in
the present work, the predicted outlet velocities and flow distribution are obtained from
CFD and are not yet supported by dedicated quantitative measurements.
To obtain a rapid, physically intuitive picture of the airflow pattern, a shallow tray of
still water was placed directly beneath the stationary nozzle while the blower operated at
its nominal printing speed. The water surface acted as a simple flow visualizer: the jets
emerging from the annular shroud produced an approximately axisymmetric disturbance
on the surface, indicating a nearly uniform circular distribution of cooling air (Figure 6c).

2.2. Synchronized C-Axis Toolpath Control

Generating toolpaths for a continuously rotating, non-circular nozzle introduces new
challenges. Control adaptations are experimentally validated with Creality HP PLA fil-
ament (Shenzhen Creality 3D Technology Co., Ltd., Shenzhen, China), with a 1.75 mm
nominal diameter.
The first trials updated the nozzle yaw at discrete reference points (segmented “orient”
commands), causing intermittent in-place rotations and leading to local over-extrusion and
surface artifacts (Figure 7a). To eliminate long gaps in rotations, yaw was incorporated
directly into G1 linear moves that drive X–Y motion. In this synchronized scheme, to
obtain smooth evolution in orientation throughout the deposition, the C-axis angle (θ)
is updated continuously along the arc length (s) of each segment. As rotational and
translational components are kinematically linked, abrupt yaw changes can perturb the feed
J. Manuf. Mater. Process. 2025, 9, 409 9 of 22

rate. Consequently, the incremental yaw was limited to tenths of a degree per planner step,
allowing the rotational component to remain negligible with respect to the commanded
tool speed.

(a) (b)

Figure 7. Toolpath control. (a) Over-extrusion in corners, due to the nozzle yaw at the vertex.
(b) Integration of the rotation in the displacement movement.

A further refinement introduced the lead distance (L Lead ): instead of executing a yaw
change exactly at a vertex, the nozzle begins reorientation slightly in advance, interpolating
between incoming and outgoing segment angles (Figure 7b). When the space between
vertices is shorter than L Lead , explicit pre-rotation is skipped, and the yaw is blended with
the concurrent X–Y move. This avoids static rotations over a single point and preserves
near-constant head velocity on finely segmented or curved paths.
Empirically, the synchronized + lead strategy reduced discontinuities and over-
deposition on the strand, with smoother surfaces. Nevertheless, sharp corners and intricate
details require adequate L Lead values to prevent residual artifacts, achieving a noticeable
quality improvement (Figure 8).

Figure 8. Variations in vertex quality as a function of different LLead values (from 0 mm to 10 mm).
J. Manuf. Mater. Process. 2025, 9, 409 10 of 22

2.3. Practical Considerations

This section summarizes the prototype adjustments and implementation details that
improved performance and reliability.

2.3.1. Stable IR Reading

Because the IR sensor is fixed while the nozzle/sleeve rotates, the temperature trace
shows periodic oscillations as high as ±6 ◦ C in early builds that vanish when the rotation
stops; this is evidence of a systematic, not random, error. True circumferential gradients are
unlikely, given inductive-heating symmetry; instead, changing surface is normal within
the sensor’s field of view modulate emissivity, producing a cyclic bias. To mitigate this
effect, we implemented three modifications: (i) rounding the nozzle outlet edges to prevent
glints; (ii) applying a uniform matte-black circumferential coating at the measurement
zone and (iii) low-pass filtering the signal (1 Hz, five-sample moving average). These
measures reduced uncertainty to below ±2 ◦ C and provided stable feedback during the
continuous rotation.

2.3.2. Runout and Concentricity

Localized wall defects—lateral offsets between consecutive layers at specific
azimuths—were observed in early trials. XY positioning errors can produce similar arti-
facts; however, inspection of the G-code and motion logs ruled out software or kinematic
causes. The most probable mechanism was non-coaxiality between the heat-sink shaft and
the nozzle orifice. Measurement of nozzle tip runout with a 0.01 mm resolution dial indica-
tor yielded a total indicated runout of 0.46 mm (orbital radius of 0.23 mm), comparable to
the 0.40 mm slot height and therefore sufficient to generate visible offsets. Such a level of
runout would be unacceptable in a production system and should be viewed as a limitation
of this prototype, rather than as an inherent restriction of the URHE concept. Future itera-
tions will require a stiffer and more accurate rotary assembly, together with tighter control
of the nozzle and sleeve geometry, to reduce radial error to the level typically expected from
conventional hot ends. Nonetheless, after mechanical correction, confirmation prints were
executed, which retained faint misalignments, indicating a residual concentricity error.
A secondary contributor was identified in the outlet geometry: the LPBF-fabricated,
nominally rectangular orifice exhibited slight distortion (Figure 3b), consistent with sen-
sitivity to powder size and local sintering conditions, introducing an additional angle-
dependent displacement component. To mitigate this, the post-processor records the yaw
at the first visit to each XY coordinate and enforces the same yaw on subsequent layers;
applied to outer perimeters, this eliminated residual offsets and restored surface fidelity in
the validation prints.

2.3.3. Extrusion Temperature

The initial screening showed that the induction-heated hot end delivers stable, defect-
free extrusion at the lower bound of temperatures that are commonly reported for PLA
in FFF (170 ◦ C) [38], and below typical manufacturer recommendations. Because 170 ◦ C
lies above PLA’s typical melting range (150–160 ◦ C) [41], this relatively low setpoint is
mainly attributed to an improvement in thermal geometry with closer-to-the-target heat
production. Accordingly, 170 ◦ C was adopted as the process standard for all subsequent
experiments with the URHE.
J. Manuf. Mater. Process. 2025, 9, 409 11 of 22

3. Materials and Methods

3.1. Materials
Two Cartesian 3D printers (Creality Ender-3, Shenzhen Creality 3D Technology Co, Ltd.,
Shenzhen, China) were selected for the comparison (Figure 9a). One serves as the reference,
using a conventional cartridge-heated hot end with a circular 0.40 mm nozzle (CREALITY
hot end). The other supports the proposed induction-heated URHE (Figure 9b) with a
1.20 mm × 0.40 mm rectangular slot. The unmodified printer runs on the original firmware
(Marlin), while the modified one runs Duet firmware. The modified printer integrates a
Duet 3 6HC, which allows for control of the rotary axis (C-Axis). Aside from the hot end
assembly and C-axis hardware, both platforms remain indistinguishable.

(a) (b)

Figure 9. Experimental setup for the twin-printer testbed: (a) the printer on the left is equipped with
the URHE, while the printer on the right uses the CREALITY hot end. (b) Close detail of the URHE.

Both printers are installed side-by-side to withstand the same ambient temperature,
relative humidity and any other environmental condition (Figure 9a). They are also lev-
eled using the same protocol: Z-offsets are adjusted to a common feeler-gauge standard.
Additionally, the PLA filament is drawn from the same lot and conditioned identically.
In the trials, the same jobs are launched simultaneously for both benchmarks. This twin
design isolates the effect of the hot end architecture while holding the slicing, the motion
constraints, the environment and the material constant, thereby supporting the consistency
of the comparative analysis of time and energy measurements to produce the same parts
with different architectures.

3.2. Methods
3.2.1. Time and Energy Comparative Test
Four different geometries are selected to cover a wide range of complexity: three
hollow pieces with different shapes and a Moai statuette containing infill (Figure 10).
In every case, the same STL file is duplicated, so the two versions of a given part
are dimensionally identical. Process parameters that could bias the total manufacturing
time were kept identical: the layer height was fixed at 0.20 mm for the simpler shells and
0.15 mm for the more intricate models, and all hollow parts shared the same wall thickness
(1.2 mm).
J. Manuf. Mater. Process. 2025, 9, 409 12 of 22

Figure 10. Benchmark parts. From left to right: cylindrical cup, Moai, regular vase and twisted vase.

However, certain parameters of the process are necessarily different. The URHE op-
erates at 170 ◦ C (see Section 2.3.3), while the conventional one requires 210 ◦ C to achieve
adequate flow and bonding. Printing speeds could also differ between machines. However,
in this case, any of these differences consistently works against the URHE, since its deposi-
tion speed is intentionally limited in certain scenarios to ensure process stability. Perimeter
speeds are kept identical, ensuring that geometric fidelity and surface finish remain com-
parable. To avoid any speed-induced bias, the URHE infill was intentionally limited to
25–35 mm·s−1 (versus 40–60 mm·s−1 in the reference setup), so differences, where present,
are confined to infill speeds. Therefore, no speed-related advantage is introduced that
might artificially shorten the manufacturing times of the URHE.
Table 1 summarizes the main configuration of the parameters for each part and each
printer, highlighting the critical settings.

Table 1. Printing configurations for comparison tests.

Printing Speed Temperature

Geometry Hot End
Layer Height Nozzle Nozzle Size [mm·s−1 ] [◦ C]
[mm] Shape [mm]
Perimeters Infill Hot End Bed
CREALITY 0.2 Circular 0.4 50 50 210 60
Cylindrical cup
URHE 0.2 Rectangular 1.20 × 0.40 50 25 170 60
CREALITY 0.2 Circular 0.4 50 40 210 60
Regular vase
URHE 0.2 Rectangular 1.20 × 0.40 50 30 170 60
CREALITY 0.15 Circular 0.4 60 40 210 60
Twisted vase URHE 0.15 Rectangular 1.20 × 0.40 60 30 170 60
CREALITY 0.15 Circular 0.4 60 60 210 60
Moai
URHE 0.15 Rectangular 1.20 × 0.40 60 35 170 60

To track the energy usage of each printer, a consumer-grade Ankrs plug-in power
meter [42] records, among other parameters, total energy consumption (kWh) and peak
current draw. Operating time is registered with a digital chronometer. For each trial, a
single video simultaneously records the kinematic behavior of both printers, the real-time
and cumulative energy readings of each power meter, and the total elapsed time from the
onset of the job until each machine completes it (Video S1).

3.2.2. Steady-State Hot-End Power Measurement

Hot-end hold power PHold is measured under static conditions (no filament
motion), with the heat-sink fan ON and heat bed OFF. For both heads, setpoints
TSet = 170, 180, 190, 200, 210, 220 ◦ C are commanded. After the temperature first enters
J. Manuf. Mater. Process. 2025, 9, 409 13 of 22

TSet , the system is allowed to stabilize for ≥1 min. Afterwards, PHold is computed as the
mean of a 60 s measurement.

3.2.3. Thermal Transients

Thermal transient tests were defined for both heating and cooling operations. Time-to-
setpoint, t T , is defined as the lapse between the moment the order is issued and the moment
a given temperature, T, is first reached and maintained for more than 1 s. Cool-down time
t50 is defined as the interval from Tset to 50 °C, measured at the shroud sensor location with
part-cooling off. For this test, the inductive head operates near resonance (68.5–69.7 kHz).
Temperature and power are logged at 1 Hz.

3.2.4. Sustained Flow Rate

Maximum sustainable volumetric flow is determined using a straight toolpath 50 mm
in length with a fixed layer height of 0.2 mm. Tests start at Q = 10 mm3 ·s−1 and increase in
∆Q = 0.5 mm3 ·s−1 steps. Acceptance requires no visible gaps, and no feeder “clicks” or
missed steps for each Q.

4. Results and Discussion

4.1. Manufacturing Time and Energy Comparative Test
Figure 11 presents the eight benchmark parts used for manufacturing time and energy
comparison. The observable print quality was comparable within each pair, with sur-
face finish and dimensional integrity preserved across both systems. The most noticeable
difference is a visible seam artifact on the parts printed with the URHE (Figure 11b,d).
This was anticipated, as the wider strand produced by the rectangular nozzle exagger-
ates the start–stop overlap, and no dedicated seam-hiding strategy was implemented in
this prototype.

(a) (b) (c) (d)

(e) (f) (g) (h)

Figure 11. Representative geometries manufactured for the comparison tests: (a) Cylindrical cup
printed with the CREALITY hot end. (b) Cylindrical cup printed with the URHE. (c) Regular vase
(CREALITY hot end). (d) Regular vase (URHE). (e) Twisted vase (CREALITY hot end). (f) Twisted
vase (URHE). (g) Moai (CREALITY hot end). (h) Moai (URHE).
J. Manuf. Mater. Process. 2025, 9, 409 14 of 22

In contrast, the peripheral definition was slightly better in the URHE prints for parts
with radial symmetry (e.g., the cylindrical cup and regular vase). The URHE-printed
twisted vase, however, (Figure 11f) showed a lower surface fidelity than its conventional
counterpart (Figure 11e), exhibiting a shallow, rotation-locked waviness along the helical
ridges. This texture is consistent with the slot’s directional strand geometry, combined with
residual nozzle–axis runout, finite-step yaw interpolation and the characteristics of the
torsional geometry.
The results for build time and cumulative energy consumption are summarized in
Table 2.

Table 2. Results of the comparison tests.

Energy
Energy
Extruder Manuf. Time Time Energy Volume Reduction
Geometry Consumption
Head [min] Reduction * Savings * [cm3 ] Per PLA
[Wh]
Deposited
CREALITY 47 82 7.3
Cylindrical cup 49% 39% 41%
URHE 24 50 7.5
CREALITY 88 151 19.7
Regular vase 39% 27% 25.7%
URHE 54 110 19.3
CREALITY 271 468 48.7
Twisted vase 35% 18% 14.3%
URHE 175 386 46.9
CREALITY 54 99 7.3
Moai 22% 9% 18.6%
URHE 42 90 8.1
* Reductions and savings are relative to CREALITY (baseline).

Results of manufacturing times show that the URHE achieves a significant reduction.
This outcome aligns with expectations and is primarily attributed to the decrease in the
number of passes required to deposit the outer shell, thanks to the wider strand extruded
by the rectangular nozzle.
The reduction is especially pronounced in parts that do not require infill, reaching a
maximum of 49% in the case of the cylindrical cup. Even in the more complex Moai model,
which includes internal infill, the build time was reduced by 22%. These time savings
translate directly into lower cumulative energy consumption, with reductions ranging from
9% (Moai) to 39% (cylindrical cup).
Overall, the comparative tests demonstrate that the proposed design delivers a sub-
stantial reduction in build time and appreciable energy savings, although the magnitude
of these benefits varies depending on part geometry and toolpath structure. As expected,
the reduction time is much greater in parts that have extensive perimeters and reduced
or no infill. After all, the number of passes required to print the perimeters of the parts is
reduced by one third, whereas the number of infill passes remains unchanged. While the
exact contribution of each subsystem (geometry, heating method, thermal mass) remains
coupled in the current design, the results suggest that revisiting URHE can yield gains
that are not achievable through isolated parameter tuning. The variation in savings across
parts also reinforces the need for geometry-aware optimization, in which nozzle shape,
orientation, and thermal response are co-designed for specific toolpath patterns.
To illustrate differences in the overall geometric quality, the vase specimens have
been digitized using a handheld laser scanner (FreeScan Combo, SHINING 3D, Hangzhou,
China), and the resultant point-clouds are compared with Geomagic Studio 2013 (3D
Systems, Rock Hill, SC, USA). Results are presented in Figure 12, where the color-mapped
deviation field between the URHE part and the nominal CAD surface (right) shows that,
apart from the layer-seam region, where a slightly thicker ridge is visible, the deviation
J. Manuf. Mater. Process. 2025, 9, 409 15 of 22

map shows no systematic bulging, flattening or twisting. The average positive deviation
is 66 µm, while the negative one is −228 µm, with a standard deviation of 167 µm. These
results do not reveal a significant difference between both specimens, even when no specific
dimensional calibration has been performed.

Figure 12. Digitized point-clouds for the CREALITY-manufactured regular vase (left), the URHE-
manufactured regular vase (center) and the color-mapped deviations in mm (right).

Additionally, an illustrative analysis of surface quality differences has been conducted

over the vase surface. A small patch has been acquired from each specimen with a ConoScan
4000 conoscopic holography system (Optimet Metrology, Jerusalem, Israel). Then, each
digitized surface patch was first fitted with a linear interpolation model in MATLAB R2025b
(The MathWorks Inc., Natick, MA, USA), which allows for obtaining the Z coordinate for
any X–Y combination within the digitized area. From this model, a Z matrix was generated
on a regular X–Y grid and exported in the SPIP format. The resulting point cloud was
then loaded into SPIP (Image Metrology A/S, Hørsholm, Denmark), where it was filtered
using both an S-filter (to remove high-frequency/short-wavelength noise) and an L-filter
(to remove low-frequency/long-wavelength curvature).
Following the common recommendations, the L-filter cutoff was set to 1 mm, ap-
proximately five times the wavelength of the coarsest texture of interest (≈0.220 mm),
while the S-filter cutoff was set to 0.025 mm, a value greater than three times the capture
resolution (0.012 mm), but still low enough to avoid excessive smoothing. After filtering,
the areal surface parameters were computed using the processed dataset. Figure 13 shows
a comparison between both patches.
The areal roughness parameters for the URHE and reference hot ends are of the
same order of magnitude and fall within the range expected for MEX parts printed with a
0.20 mm layer height. The URHE patch shows a slightly higher roughness (Sa = 22.5 µm,
Sq = 27.8 µm) than the CREALITY patch (Sa = 18.3 µm, Sq = 22.2 µm), together with a
larger total height Sz (261 µm vs. 189 µm). These differences indicate a moderately rougher
surface for the URHE specimen, but in both cases, the overall values are consistent with the
expected staircase- and bead-related topography. The skewness and kurtosis (Ssk ≈ −0.13,
Sku ≈ 3.1 for URHE; Ssk ≈ 0.28, Sku ≈ 2.6 for CREALITY) are close to Gaussian in
both cases, suggesting a similar texture character with only a slight tendency of the URHE
surface toward deeper valleys.
J. Manuf. Mater. Process. 2025, 9, 409 16 of 22

(a) (b)

(c) (d)

Figure 13. Digitized surface patches: (a) original patch—CREALITY. (b) Filtered patch—CREALITY.
(c) Original patch—URHE. (d) Filtered patch—URHE.

4.2. Steady-State Hot End Power Measurement

Focusing specifically on the power drawn by the hot end in steady state (PHold ) con-
firms that the inductive design delivers a clear energy advantage over the cartridge-heated
head throughout the 170–220 ◦ C window. As Figure 14 shows, the URHE requires only
8.5 W at 170 ◦ C and 12.1 W at 220 ◦ C, whereas the conventional assembly draws 16.1 W and
21 W, respectively. Differences vs. theoretical expectations arise from geometric/thermal
boundary simplifications and component losses.

25
Creatity
21
20.1
19.2 URHE
20 18.1
17
16.1
PHOLD [W]

15
12.1
11.1
9.7 10.3
9.3
10 8.5

0
160 170 180 190 200 210 220 230
Temperature [°C]

Figure 14. PHold comparison.

Considering that the URHE employs a reference of 170 ◦ C to provide the best dimen-
sional fidelity and surface finish, while the resistive cartridge requires 210 ◦ C, inductive
solution cuts hot end energy consumption by approximately 58% (8.5 W vs. 20.1 W).
J. Manuf. Mater. Process. 2025, 9, 409 17 of 22

Nevertheless, the hot end represents just one contributor to the total power draw; motion,
bed heating and electronics also consume significant energy.

4.3. Thermal Transients

Thermal transients consume a non-negligible share of a desktop FFF build cycle. On
the reference Creality Ender 3 with a resistive cartridge, the nozzle requires an average of
t170 = 81.3 s to reach 170 ◦ C. The URHE, on the other hand, requires t170 = 28.2 s. The
URHE’s more effective inductive heat delivery cuts warm-up time to about one third of
the time required by the resistive cartridge. This reflects the higher heating efficiency of
the induction system. Considering that the reference printer demands a 210 ◦ C setpoint to
achieve comparable printing consistency, it must be noted that it requires t210 = 135.3 s,
which is ≈4.8× longer than the URHE’s warm-up to its corresponding operating setpoint
(170 ◦ C).
The cool-down time to 50 ◦ C likewise improves from t50 = 120.7 s to t50 = 74.4 s,
consistent with the lower thermal mass of the inductive assembly. For clarity, these non-
productive transients are not included in the headline part-level time/energy comparisons
in Section 4.1. Nevertheless, they further enhance the process’ productivity by shortening
pre- and post-print dwell, thereby improving effective machine throughput.
The lower thermal mass of the URHE reduces thermal inertia [29]. This charac-
teristic, combined with the generation of heat directly in the sleeve, produces faster
transients—cutting warm-up and cool-down times by more than half—and thereby slashes
the non-productive dwell that brackets every build.

4.4. Sustained Flow Rate

The URHE sustains Qmax = 30 mm3 ·s−1 without observable under-extrusion, filament
grinding, or dimensional slumping. Under identical conditions, the reference CREALITY
hot end on the baseline printer reaches a practical Qmax of 11.5 mm3 ·s−1 , beyond which
insufficient melt delivery and intermittent feeder slip produce strand thinning. Crucially,
this throughput headroom is not exploited to set the print speeds used in Section 4.1.
The Qmax comparison is reported here to document scalability and the robustness of melt
delivery at elevated flow rates. These Qmax values are consistent with Go et al.’s module-
level analysis of FFF [28], which identifies conduction heat transfer into the filament core at
the heater/nozzle as a primary rate limit; increasing the available heat flux at the hot end
(and reducing losses) raises the steady-state volumetric flow ceiling.
Two design features were critical to this performance: the proposed design extends
the heater length to 20.5 mm (the equivalent length of the original hot end is 11.5 mm), and
the inductive heating deposits energy directly into the AISI 420 sleeve surrounding the
filament path, significantly reducing the radial heat-transfer distance.
Conventional guidance favors short melt zones to preserve control during extru-
sion/retraction, even though longer liquefiers are often linked to higher throughput [21].
In contrast, the URHE shows no loss of start/stop precision and no increase in stringing
at high flow. These results indicate that a longer, internally heated melt zone can sustain
high volumetric rates without sacrificing quality, challenging the usual hot end heuris-
tics [21]. With the heat source coaxial to the polymer path, the induction sleeve flattens the
axial temperature profile and confines thermal inertia largely to the molten polymer, so
residence time can be increased without slowing retraction response. Practically, the melt-
zone length becomes a tunable parameter for throughput, not a fixed limit from control
concerns. A longer channel also reduces the shear rate at a given flow, which may curb
thermal/mechanical degradation and expand the window for viscosity-sensitive materials.
J. Manuf. Mater. Process. 2025, 9, 409 18 of 22

More broadly, the data indicate that heating topology—not only power—sets the
trade-off between deposition rate and extrusion accuracy. When thermal uniformity and
low thermal mass are preserved, extending the melt zone need not compromise deposition
control, avoiding the usual escalation to heavier heaters and stiffer motion hardware [28].

4.5. Runout
As previously outlined, any eccentricity or angular misalignment—arising from the tol-
erances of the nozzle–hot end interface or from an imperfect non-circular outlet—displaces
the extrudate from the ideal path that is assumed to lie on the rotational axis. When the
head revisits the same XY location at a different yaw angle, this lateral offset accumulates
as shifts in the interlayer contours, being perceived as banding and edge waviness. In
the present prototype, mitigation by software (enforcing identical yaw at “analogous” XY
coordinates across layers) is proposed as a compromise solution, which suppresses visible
effects. However, it is true that it does not address the main cause of the problem and that
more sophisticated printers should consider accuracy in the design and the manufacturing.
From the experimental tests of this study, minimizing the runout of the nozzle tip
could be achieved by tightening the datum scheme and thread stack-up at the sleeve–
nozzle–heat-break interface, improving outlet fidelity for LPBF nozzles via post-processing
(micro-reaming/lapping) or providing an adjustable concentricity set. A complementary
path would be in situ calibration: printing a short diagnostic ring to estimate the eccentricity
vector and applying a slicer-level offset map. Implementing these measures would render
software workarounds unnecessary and keep printed dimensions closer to nominal while
preserving unrestricted rotation.

4.6. Temperature Measurement

Rotation-locked oscillations in the IR signal point to a geometric origin associated
with the faceted nozzle tip. To suppress this effect, future designs should replace the
local hexagonal flats in the line of sight with a narrow cylindrical band, concentric with
the rotation axis, optionally combined with a short baffle to reduce specular glints. The
controlled temperature is measured on the nozzle body, rather than directly at the melt
exit, so a finite axial gradient between the measurement location and the outlet is expected.
In conventional cartridge-heated hot ends, the sensor is typically embedded in the heater
block, farther from the nozzle orifice, so the corresponding axial gradient is expected to be
even larger than in the present URHE configuration.
In the present prototype, with a single miniature IR sensor integrated in the stationary
shroud, operation remained stable over more than 150 h of continuous printing. Nonethe-
less, single-channel sensing is inherently vulnerable to drift or dropouts. For safety, we also
recommend adding redundancy via a secondary independent IR channel, elevating the
system from fail-silent to fail-safe without compromising unrestricted C-axis rotation.

4.7. Limitations of This Work

This first assessment of the URHE is subject to several limitations. All experiments
were carried out with a single polymer (PLA) on one desktop Cartesian printer, so the
quantitative time and energy savings reported here cannot be directly generalized to other
materials or machine architectures. In addition, the prototype implementation exhibited
prototype-level nozzle runout that had to be mitigated by software compensation; this
is not acceptable for a production system and should be regarded as a limitation of the
current hardware, rather than of the URHE concept itself.
A further limitation of the present study is that part quality has not yet been assessed
in terms of mechanical performance or detailed microstructure. The benchmark geometries
were chosen as process demonstrators for time and energy comparisons and are not suitable
J. Manuf. Mater. Process. 2025, 9, 409 19 of 22

for standardized mechanical testing, and no cross-sectional analysis of internal porosity or

interlayer bonding was performed.
Future work will address these points by testing additional materials and printer
platforms and by adopting a mechanically refined rotating assembly with production-
grade runout. This will include mechanical testing of appropriately designed specimens
to quantify part performance. In addition, if this research is extended to metal/polymer
extrusion (MEX/M), microstructural investigations of internal porosity and interlayer
bonding will be required to verify that the wider URHE strand does not compromise the
structural integrity.

5. Conclusions
The proposed hot end (URHE) replaces the heater block with an induction-heated
sleeve, creating a low-inertia, internally heated melt zone; together with tip-aimed IR ther-
mometry and axially uniform cooling, it enables unrestricted C-axis rotation. Employing
a PBF-fabricated, interchangeable rectangular-slot nozzle permits wide, low-aspect-ratio
strands at a constant layer height, maintaining lateral resolution.
We benchmark the proposed system against a commercial FFF printer by fabricat-
ing identical parts and logging build time and energy consumption. Across four rep-
resentative geometries, the URHE shortens the manufacturing time by up to 49% and
reduces plug-level energy up to 39%, relative to a conventional cartridge-heated refer-
ence. These part-level savings are underpinned by lower steady hold-power (8.5–12.1 W
for 170–220 ◦ C versus 16.1–21.0 W) and substantially faster thermal transients (warm-up
and cool-down times reduced by 65.3% and 38.4%, respectively), which further improve
effective throughput.
Additionally, the URHE sustained a maximum volumetric flow of 30 mm3 · s−1
without under-extrusion or slumping, versus a practical 11.5 mm3 · s−1 for the baseline
platform. This achievement is related to the use of induction heating in a low thermal
inertia design, keeping the melt uniform and responsive to retraction and extrusion cycles.
For most parts, the quality was comparable to the reference; the most visible defect was
a seam accentuated by the wider rectangular strand. Nevertheless, the surface quality of the
twisted vase was clearly worse, showing shallow, rotation-locked waviness on the helical
ridges. This effect is plausibly caused by a combination of directional strand geometry,
small nozzle–axis runout, finite yaw discretization and the torsional toolpath.
This first assessment is limited to PLA on a single desktop Cartesian printer and
a prototype URHE with non-production runout mitigated by software compensation,
so the quantitative results should not yet be generalized to other materials or machine
architectures. However, the results indicate that the URHE provides a direct path to faster,
lower-energy FFF. However, since geometric accuracy is sensitive to runout and outlet
fidelity, future work will target mitigation via tighter tolerances and improved outlet
finishing. In addition, future efforts will incorporate the development of an advanced
toolpath-generation software that is capable of optimizing both perimeter toolpaths and,
especially, infill toolpaths. This enhancement is particularly relevant given that, in the
present study, these toolpaths were kept fixed. The planned optimization will help to
reduce manufacturing time and, consequently, energy consumption, increasing process
efficiency without compromising the mechanical properties of the part.
Moreover, beyond PLA and desktop duty cycles, the same heating topology and
kinematic freedom are transferable to higher-temperature engineering polymers where
cartridge heating is constrained by thermal mass and response. The same architecture
merits evaluation in MEX/M, where pairing wide-strand deposition with continuous yaw
control may enable step-change gains in the throughput and surface uniformity. Future
J. Manuf. Mater. Process. 2025, 9, 409 20 of 22

work will therefore include mechanical testing and microstructural analysis of URHE-
printed polymer and metal-filled parts to verify that the wider ribbon does not introduce
harmful inner voids and can be exploited for structurally improved infill strategies.

Supplementary Materials: The following supporting information can be downloaded at: https:
//[Link]/article/10.3390/jmmp9120409/s1. Video S1: Moai comparison.

Author Contributions: Conceptualization, D.B. and J.A.M.; methodology, M.R., D.B. and J.A.M.;
software, M.R. and A.F.; validation, M.R. and P.J.V.; formal analysis, M.R., P.Z.; investigation, M.R.,
P.Z. and A.F.; resources, D.B. and J.A.M.; data curation, M.R. and D.B.; writing—original draft
preparation, M.R., P.Z.; writing—review and editing, D.B., J.A.M. and M.R.; visualization, A.F., P.Z.
and P.J.V.; supervision, J.A.M.; project administration, D.B.; funding acquisition, D.B. All authors
have read and agreed to the published version of the manuscript.

Funding: This publication was supported by Project PID2023-146753OB-I00, funded by MI-

CIU/AEI/10.13039/501100011033 and by FEDER/UE. It was also supported by the Government
of the Principality of Asturias through the “Severo Ochoa” Programme of predoctoral grants for
research and teaching (BP21-043), as well as by the Council of Gijón through the University Institute
of Industrial Technology of Asturias (IUTA-22-GIJON-1-10).

Data Availability Statement: Data are available within the article.

Acknowledgments: The authors thank Ignacio Díaz Vigil (IDONIAL) for his support with nozzle-tip
design and manufacturing.

Conflicts of Interest: The authors declare no conflicts of interest.

Abbreviations
The following abbreviations are used in this manuscript:

ASTM American Society for Testing and Materials

AISI American Iron and Steel Institute
cfm Cubic Feet per Minute
IR Infrared
FFF Fused-Filament Fabrication
LPBF Laser Powder Bed Fusion
MEX Material Extrusion
MEX/M Extrusion of Metal/Polymer
µH Microhenry
nF Nanofarad
PEEK Poly(Ether Ether Ketone)
PLA Polylactic Acid
PTFE Polytetrafluoroethylene
RMS Root Mean Square
URHE Unrestricted-Rotation Hot End

References
1. ISO/ASTM 52900:2021; Additive Manufacturing—General Principles—Fundamentals and Vocabulary. International Organization
for Standardization: Geneva, Switzerland, 2021.
2. Desktop 3D Printing Market Size & Share Analysis—Growth Trends & Forecasts (2025–2030). Available online:
[Link] (accessed on 7 August 2025).
3. Fiedler, F.; Ehrenstein, J.; Höltgen, C.; Blondrath, A.; Schäper, L.; Göppert, A.; Schmitt, R. Jigs and Fixtures in Production: A
Systematic Literature Review. J. Manuf. Syst. 2024, 72, 373–405. [CrossRef]
4. Hernandez Korner, M.E.; Lamban, M.P.; Albajez, J.A.; Santolaria, J.; Ng Corrales, L.D.C.; Royo, J. Cost Model Framework for
Pieces Additively Manufactured in Fused Deposition Modeling for Low to Medium Batches. 3D Print. Addit. Manuf. 2024, 11,
287–298. [CrossRef] [PubMed]
J. Manuf. Mater. Process. 2025, 9, 409 21 of 22

5. Jung, S.; Kara, L.B.; Nie, Z.; Simpson, T.W.; Whitefoot, K.S. Is Additive Manufacturing an Environmentally and Economically
Preferred Alternative for Mass Production? Environ. Sci. Technol. 2023, 57, 6373–6386. [CrossRef] [PubMed]
6. May, G.; Psarommatis, F. Maximizing Energy Efficiency in Additive Manufacturing: A Review and Framework for Future
Research. Energies 2023, 16, 4179. [CrossRef]
7. Su, J.; Ng, W.L.; An, J.; Yeong, W.Y.; Chua, C.K.; Sing, S.L. Achieving Sustainability by Additive Manufacturing: A State-of-the-Art
Review and Perspectives. Virtual Phys. Prototyp. 2024, 19, e2438899. [CrossRef]
8. Wichniarek, R.; Osiński, F. Assessing Energy Efficiency in Desktop-Size FFF 3D Printers. Appl. Sci. 2024, 14, 11819. [CrossRef]
9. Lunetto, V.; Priarone, P.C.; Galati, M.; Minetola, P. On the Correlation between Process Parameters and Specific Energy Consump-
tion in Fused Deposition Modelling. J. Manuf. Process. 2020, 56, 1039–1049. [CrossRef]
10. Kim, K.; Noh, H.; Park, K.; Jeon, H.W.; Lim, S. Characterization of Power Demand and Energy Consumption for Fused Filament
Fabrication Using CFR-PEEK. Rapid Prototyp. J. 2022, 28, 1394–1406. [CrossRef]
11. Kazmer, D.; Peterson, A.M.; Masato, D.; Colon, A.R.; Krantz, J. Strategic Cost and Sustainability Analyses of Injection Molding
and Material Extrusion Additive Manufacturing. Polym. Eng. Sci. 2023, 63, 943–958. [CrossRef]
12. Wang, Y.; Hu, C.; Wang, Z.; Lin, S.; Zhao, Z.; Zhao, W.; Hu, K.; Huang, Z.; Zhu, Y.; Lu, Z. Optimization-Based Non-Equidistant
Toolpath Planning for Robotic Additive Manufacturing with Non-Underfill Orientation. Robot. Comput.-Integr. Manuf. 2023,
84, 102599. [CrossRef]
13. Yan, Z.; Hui, J.; Lv, J.; Huisingh, D.; Huang, J.; Ding, K.; Zhang, H.; Liu, Q. A Hybrid Mechanism-Based and Data-Driven
Approach to Forecast Energy Consumption of Fused Deposition Modelling. J. Clean. Prod. 2023, 413, 137500. [CrossRef]
14. Manford, D.; Budinoff, H.D.; Callaghan, B.J.; Jeon, Y. Towards a General Model to Predict Energy Consumption for Fused
Filament Fabrication. Manuf. Lett. 2023, 35, 1358–1365. [CrossRef]
15. El Youbi El Idrissi, M.A.; Laaouina, L.; Jeghal, A.; Tairi, H.; Zaki, M. Modeling of Energy Consumption and Print Time for FDM
3D Printing Using Multilayer Perceptron Network. J. Manuf. Mater. Process. 2023, 7, 128. [CrossRef]
16. Gao, M.; Li, L.; Wang, Q.; Liu, C.; Li, X.; Liu, Z. Feature-Based Energy Consumption Quantitation Strategy for Complex Additive
Manufacturing Parts. Energy 2024, 297, 131249. [CrossRef]
17. Vidakis, N.; Kechagias, J.D.; Petousis, M.; Vakouftsi, F.; Mountakis, N. The Effects of FFF 3D Printing Parameters on Energy
Consumption. Mater. Manuf. Process. 2023, 38, 915–932. [CrossRef]
18. Zakaria, S.; Mativenga, P. A Scientific Base for Optimising Energy Consumption and Performance in 3D Printing. J. Clean. Prod.
2024, 482, 144227. [CrossRef]
19. Bakrani Balani, S.; Chabert, F.; Nassiet, V.; Cantarel, A. Influence of Printing Parameters on the Stability of Deposited Beads in
Fused Filament Fabrication of Poly(Lactic) Acid. Addit. Manuf. 2019, 25, 112–121. [CrossRef]
20. Cleeman, J.; Bogut, A.; Mangrolia, B.; Ripberger, A.; Kate, K.; Zou, Q.; Malhotra, R. Scalable, Flexible and Resilient Parallelization
of Fused Filament Fabrication: Breaking Endemic Tradeoffs in Material Extrusion Additive Manufacturing. Addit. Manuf. 2022,
56, 102926. [CrossRef]
21. Serdeczny, M.P.; Comminal, R.; Pedersen, D.B.; Spangenberg, J. Experimental and Analytical Study of the Polymer Melt Flow
through the Hot-End in Material Extrusion Additive Manufacturing. Addit. Manuf. 2020, 32, 100997. [CrossRef]
22. Lind, R.F.; Post, B.K.; Love, L.J.; Lloyd, P.D.; Carnal, C.L.; Blue, C.A.; Kunc, V. Variable Width Deposition for Ad-
ditive Manufacturing with Orientable Nozzle. U.S. Patent US20170320267A1, 9 November 2017. Available online:
[Link] (accessed on 9 September 2025).
23. Löffler, R.; Koch, M. Innovative Extruder Concept for Fast and Efficient Additive Manufacturing. IFAC-PapersOnLine 2019, 52,
242–247. [CrossRef]
24. Gharehpapagh, B.; Dolen, M.; Yaman, U. Investigation of Variable Bead Widths in FFF Process. Procedia Manuf. 2019, 38, 52–59.
[CrossRef]
25. Gharehpapagh, B.; Dilberoğlu, M.U.; Dölen, M.; Yaman, U. Use of a Nozzle with a Rectangular Orifice on a Hybrid FFF System. J.
Addit. Manuf. Technol. 2021, 1, 577. [CrossRef]
26. Gharehpapagh, B.; Dilberoglu, U.M.; Yaman, U.; Dolen, M. Adaptive Toolpath Generation for Material Extrusion Additive
Manufacturing Using a Nozzle with Rectangular Orifice. Addit. Manuf. 2023, 78, 103873. [CrossRef]
27. Gharehpapagh, B.; Dilberoglu, U.M.; Yaman, U.; Dolen, M. A Rotary Extrusion System with a Rectangular-Orifice Nozzle: Toward
Adaptive Resolution in Material Extrusion Additive Manufacturing. J. Intell. Manuf. 2025, 36, 1123–1139. [CrossRef]
28. Go, J.; Schiffres, S.N.; Stevens, A.G.; Hart, A.J. Rate Limits of Additive Manufacturing by Fused Filament Fabrication and
Guidelines for High-Throughput System Design. Addit. Manuf. 2017, 16, 1–11. [CrossRef]
29. Kühn-Kauffeldt, M.; Kühn, M.; Perrin, N.; Saur, W. Fused Filament Fabrication of Thermoplastics in High Vacuum without
Convective Heat Transfer. Sci. Rep. 2025, 15, 27497. [CrossRef] [PubMed]
30. Harding, O.J.; Griffiths, C.A.; Rees, A.; Pletsas, D. Methods to Reduce Energy and Polymer Consumption for Fused Filament
Fabrication 3D Printing. Polymers 2023, 15, 1874. [CrossRef]
J. Manuf. Mater. Process. 2025, 9, 409 22 of 22

31. Nzebuka, G.C.; Ufodike, C.O.; Rahman, A.M.; Gwynn, C.M.; Ahmed, M.F. Numerical Modeling of the Effect of Nozzle Diameter
and Heat Flux on the Polymer Flow in Fused Filament Fabrication. J. Manuf. Process. 2022, 82, 585–600. [CrossRef]
32. Oskolkov, A.A.; Trushnikov, D.N.; Bezukladnikov, I.I. Application of Induction Heating in the FDM/FFF 3D Manufacturing. J.
Phys. Conf. Ser. 2021, 1730, 012005. [CrossRef]
33. Oskolkov, A.; Bezukladnikov, I.; Trushnikov, D. Indirect Temperature Measurement in High Frequency Heating Systems. Sensors
2021, 21, 2561. [CrossRef]
34. Oskolkov, A.A.; Bezukladnikov, I.I.; Trushnikov, D.N. Rapid Temperature Control in Melt Extrusion Additive Manufacturing
Using Induction Heated Lightweight Nozzle. Appl. Sci. 2022, 12, 8064. [CrossRef]
35. Plasmics GmbH Key Features of the Plasmics INo Trident Induction Nozzle. Available online: [Link]
(accessed on 7 August 2025).
36. SIJ Metal Ravne. SINOXX 4021 (Mat. No. 1.4021; DIN X20Cr13; AISI 420)—PK3 Data Sheet; SIJ Metal Ravne d.o.o.: Ravne Na
Koroškem, Slovenia, 2016; Available online: [Link] (accessed on 5 May 2025).
37. Tavares, S.S.M.; Fruchart, D.; Miraglia, S.; Laborie, D. Magnetic Properties of an AISI 420 Martensitic Stainless Steel. J. Alloys
Compd. 2000, 312, 307–314. [CrossRef]
38. Wang, X.; Huang, L.; Li, Y.; Wang, Y.; Lu, X.; Wei, Z.; Mo, Q.; Zhang, S.; Sheng, Y.; Huang, C.; et al. Research Progress in Polylactic
Acid Processing for 3D Printing. J. Manuf. Process. 2024, 112, 161–178. [CrossRef]
39. Pyda, M.; Bopp, R.C.; Wunderlich, B. Heat Capacity of Poly(Lactic Acid). J. Chem. Thermodyn. 2004, 36, 731–742. [CrossRef]
40. Wang, L.; Qiu, J.; Sakai, E.; Wei, X. The Relationship between Microstructure and Mechanical Properties of Carbon Nan-
otubes/Polylactic Acid Nanocomposites Prepared by Twin-Screw Extrusion. Compos. Part A Appl. Sci. Manuf. 2016, 89, 18–25.
[CrossRef]
41. Pavon, C.; Aldas, M.; Samper, M.D.; Motoc, D.L.; Ferrandiz, S.; López-Martínez, J. Mechanical, Dynamic-Mechanical, Thermal
and Decomposition Behavior of 3D-Printed PLA Reinforced with CaCO3 Fillers from Natural Resources. Polymers 2022, 14, 2646.
[CrossRef]
42. Ankrs. Plug-in Electricity Meter—16A/3680W. Amazon (ASIN B0BGRV4P7V). Available online: [Link]
Consumption-Ankrs-Display-Backlight/dp/B0BGRV4P7V (accessed on 15 September 2025).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.