Process Based Modeling of Energy Consumption for Multi-material FDM 3D Printing


 Fused deposition modeling (FDM) is one most cost-effective 3D printing technique for forming complex 3D components based on thermoplastic materials. The energy consumption analysis is one criterion to determine the capacity and sustainability of the FDM. In this paper, the energy consumption of a dual- extruder FDM is studied by differentiating the whole multi-material printing process into independent operation modes, which characterize the thermal behaviors of the printing parameters. By investigating the execution instruction which describe the tooling plan of the FDM, the nozzle temperature distributions with different filament materials are measured and simulated. The energy consumption details can be accurately captured in our work and therefore we are able to predicate the power demand changes with different working processes of the multi-material FDM 3D printing. This work will be beneficial for optimization of 3D printer design and manufacturing sustainability in next.


Introduction
Fused deposition modeling (FDM) is widely considered as one most cost effective additive manufacturing (AM) technique for building a 3D complex structure by extruding the liquified thermoplastic or thermoset filaments through a small diameter nozzle [1], for which energy consumption analysis is one prerequisite to evaluate its printing capacity and manufacturing sustainability [2]. For a FDM printing process, Peng [3] proposed a two-level energy analysis method based on the energy flow during the printing process. The primary energy is the thermal energy required for melting the printing filaments, and the secondary energy is the auxiliary energy input to drive the mechanical movements, auxiliary temperature fields, etc.
Ajay etc. [4] obtained the impacts of printing accuracy, printing speed, and temperature on the energy consumption and found that the heat energy consumption in the printing process accounts for about 40% of the total power consumption. Therefore, Le Bourhis etc. [5,6] analyzed each printing parameter comprehensively and established a prediction model on the power factor of the consumable filament material.
From 3 different FDM printers, Clemon etc. [7] found that most of the energy actually went to the heating elements during the printing. Therefore, Mongol etc. [8] began to explore the impact of other in-process printing parameters on energy consumption and noticed that the placement of parts would also impact the final printing power consumption. Balogun etc. [9] conducted a unified analysis of three different FDM printers, and summed up the power consumption characteristics of the main working stage of the printer. They found that the printer reduced the loss of thermal energy with continuous printing under a full task requirement [10,11]. Lately, when the effect of layer thickness and packing density on the mechanical properties of parts were studied [12], the researchers also found that, compared with traditional subtractive manufacturing, higher dimensional accuracy did not necessarily need higher power consumption. For a high-precision energy consumption prediction, Yosef etc. [14] recently divided the FDM printing process into different working stages, and obtained the energy consumption model for each stage through experiments. Therefore, Ajay etc. [15] analyzed the printer motion execution instructions, and proposed turning off the other axis drive motor to achieve energy saving in a single axial movement. As a result, energy was saved by 25% with new printing strategy.
The existing researches only focused on the FDM 3D printing of a single type of material and the temperature field during printing can be assumed as constant.
However, FDM printer with multiple extruders have overwhelmed in both academic and industrial sectors [16], in that it can directly fabricate a complex functional 3D part [17]. The variety of materials in multi-material FDM 3D printing may own very different melting temperatures and as a result frequent power switching (on, off, increase, decrease) was required during the printing of a multi-material part [18,19].
These influences would cause the complexity with evaluating the electrical energy consumption. Therefore, this paper will draw on how to model the power necessary to drive a dualextruder FDM 3D printer, which will help with optimizing multi-material FDM design and also other extensive types of AM.    For multi-material FDM 3D printing, we designed a printable cylindrical "Yin" and "Yang" 3D part shown in Figure 3 with the height 2.5mm and diameter set as 30mm, in Chinese philosophy which represents two principles: one negative, dark, and feminine (Yin); and one positive, bright, and masculine (yang). The part was first sliced into printing layers at a thickness of 0.2mm During printing of each layer, the ABS filament was first extruded out from "ABS" nozzle and deposited in the "Yin" zone, by ramping the nozzle temperature from 200 to 240 °C. For "Yang" zone, the PLA filaments were extruded out from "PLA" nozzle and deposited by heating temperature as maintained at 175°C. Recorded were the driving current and voltage datum to the computer in real-time during the printing process.

Results and discussion
The power consumption of the FDM printer at the idle mode was first calibrated by measuring the required minimum power to drive the peripherical module, such as the control motherboard, the LCD, and the cooling fan. The power at the idle mode was constant in this study. Secondly, preheating mode of FDM was switched on, in which the building plate was heated up to its pre-set value, and then the printing nozzle started to be heated for melting the filament. For the FDM 3D printer, its "ABS" nozzle was directly heated up to 240 °C, slightly higher than the melting temperature of ABS material. This nozzle temperature was maintained whenever the whole ABS material was printed. When the two materials ABS and PLA were printed, both "ABS" and "PLA" nozzles were heated to their own standby temperatures, i.e. 200°C for ABS and 175°C for PLA, respectively.  Therefore, according to Ref [14], the overall power consumption of the system can be calculated by where is the power consumption at idle mode; is the required power to heat the build plate; and represent the electrical power added on the "ABS" nozzle and "PLA" nozzle; is the power to drive three-axis movements of the FDM 3D printer; and are the power used to drive the extruder motors to fuel the filaments into the heating nozzle. Table 2 gives the required power for each printing parameter as well as the related Gcode execution instruction. For example, when a nozzle was set to maintain at 220°C for preheating the ABS, an electrical power input of 24W was in need. And this calibration was done by us to run the Gcode command of "M109 S220".  Once Table 2 and other possible printing parameters under investigation were obtained, Figure 6a demonstrated the power curve for the build plate heated from room temperature to 80° C by running the Gcode. There, the maximum power for heating the build plate was 144w, which determined the higher mean power at the preheating stage. The power of the building plate still occupied more than 50% of the total power as consumed during the whole printing. The heating power consumption of the nozzle was then recorded by ramping the temperature from the room environment to 220°C, as shown in Figure 6b. A simple control strategy was deployed for this FDM 3D printer, in Figure 6c, such as: when the temperature was raised, it was heated at a full power input; and when the temperature was lowered down, the heating power was completely turned off until the preset temperature was reached, and then it was turned to a low-power mode holding constant temperature. In final, the power consumption of the motion motor module was shown in Figure 6d. printing can be identified and actually calculated by directly extracting the relevant commands in the Gcode file, shown in Table 3. Since the overall movement distance of the printer is limited, the movement of G1 was assumed to be variable acceleration. Therefore, the time for taking the printer to execute this instruction was: For the command of G1, the running duration was： = # (7) where was the distance of each segment, and was the speed.   The overall trend based on theoretical estimation performed in consistance with the experimental measurements as shown in Figure 10. There was a small margin of error in time synchronization for estimation. The small difference might lie in that: first, the operating environment, including room temperature and cooling fan installation, could cause less control on the air convection and therefore the heating efficiency; secondly, the heating efficiencies of the two nozzles somehow distinguished from each other. In Figure 10, although there is a certain deviation between the predicted value and the actual value, the predicted power map fully reflects the fluctuation amplitude and time range of the power change during the bi-material printing process.