Cost Model Framework for Pieces Additively Manufactured in Fused Deposition Modeling for Low to Medium Batches

Abstract

The cost impact of implementing additive manufacturing (AM) in the fabrication process is nowadays an issue. The scope of this research is to establish a cost model framework that can predict the cost of a piece in a low to medium batch considering fused deposition modeling (FDM) printing parameters. Every enterprise wants to increase its internal capabilities for tools, prototypes, and the production of pieces for maintenance purposes. FDM is an AM technology increasingly used in aerospace, automotive, and many other sectors. The research methodology consists of developing a cost model based on the extrusion-type AM process for any given machine characteristics and comparing the cost per piece based on diverse lot sizes and raw materials. Two test cases were simulated to show the usefulness of the cost model, one with a conventional polymer material (acrylonitrile butadiene styrene) and another with a high-performance material (polyetheretherketone); materials with very different costs, machine technical requirements, and energy consumption. The framework could be used to predict the best machine size and material type that could be suitable for a certain situation. The strength of our approach lies in the energy cost calculus, which is dependent on machine capabilities and size.

Introduction

Additive manufacturing (AM) is an effective fabrication approach for converting virtual models into physical 3D models that join materials layer upon layer,¹ reducing time and fabrication expenses. AM fabrication methods are used in many fields such as end-use parts, biomedical applications, aerospace, and automotive. AM can be classified according to the raw material state: liquid, filament, powder, and solid layer. The most common methods for processing liquids are stereolithography, multijet molding, and liquid thermal polymerization.

The methodology used for the filament raw material was fused deposition modeling (FDM). Powder raw materials can be classified into powder beds and spray powder processes. The powder bed fabrication methods are selective laser sintering (SLS), laser powder bed fusion, and electron beam melting (EBM). For the spray powder process, the methods are direct energy deposition and 3D printing. Finally, for the solid layer process, the AM method is termed laminated object manufacturing (LOM).² In this study, the FDM method was selected to evaluate the cost model for diverse types of materials and manufacturing scenarios.

FDM technology is characterized by the application of heat by a hot end to reach the extrusion temperature of a filament material.¹ FDM machines are capable to extrude diverse types of materials from amorphous to semicrystalline polymers; Table 1 shows the most common ones used for FDM and machine temperature control for each material.

Table 1.

Commonly Used Fused Deposition Modeling Raw Materials

Name	Abbreviation	Performance	Type	Temperature control
				Build plate	Chamber
Polylactic acid	PLA	Commodity	Amorphous
Polypropylene	PP	Commodity	Semicrystalline	x
Polycarbonate	PC	Engineering	Amorphous	x
Acrylonitrile butadiene styrene	ABS	Engineering	Amorphous	x
Polyesters	PHA	Engineering	Semicrystalline	x
Polyetherimide	PEI	High	Amorphous	x	x
Polyetheretherketone	PEEK	High	Semicrystalline	x	x

Adapted from Refs.^50,66

This study was conducted to make a theoretical contribution to AM cost models through different cost-estimation approaches that are oriented to companies looking for a cost estimation depending on their variable requirements, such as spare parts on demand, prototyping, repairing, and refurbishment, usually low to medium production volumes, to demonstrate the cost difference among the material selection, machine requirements, and process parameters.

FDM process parameters may affect the mechanical properties of the manufactured part: extrusion temperature, preheated build plate, isolated build chamber, and cooling systems.³ Consequently, the cost per piece may vary. The variety of materials that FDM technology can extrude makes it affordable for various applications, even aerospace and automotive.⁴ The scope of this research is to establish a cost model framework that can predict the cost of a piece in a batch, considering several FDM printing parameters for different thermoplastic materials. This could become a managerial tool for cost estimation between several scenarios, with the purpose to increase the “in-house” capabilities of a company to produce tools, prototypes, and remanufactured parts for maintenance.

The remainder of this article is organized as follows. The theoretical background of the AM cost models is discussed in the Literature Review on AM Cost Models. In research methodology and model, the methodology adopted for this study, the cost model formulation, and the framework are presented. The results of the model test and analysis are presented in the Results. Finally, discussion and conclusion with and overview of the main findings are presented.

Literature Review on AM Cost Models

A considerable number of cost models are addressed for different types of AM methods. Most of the cost models were tested using SLS^5–8 for thermoplastic materials, followed by SLA.^5,8–10 Moreover, for metal raw materials, various AM technologies in cost models were considered: EBM, direct metal laser sintering (DMLS), and LPBF.^8,11–14 Finally, for FDM, the cost model proposed for acrylonitrile butadiene styrene (ABS) filament⁵ and the one proposed in a previous study,¹⁵ as far as our knowledge goes, are the only proposed models for this AM technology. In Table 2, a list of cost models is preset for various AM methods and raw materials.

Table 2.

Cost Models Proposed by Year, Technology, and Raw Material Used for Each One

Cost model proved for
Reference	Year	FDM	SLA	SLS	SLM	EBM	DMLS	LPBF	Raw material
Hopkinson and Dickens5	2003	X	X	X					ABS, Epoxy, and Nylon
Ruffo et al.6	2006			X					Duraform PA
Campbell9	2008		X						Not specified
Atzeni et al.7	2010			X					PA
Rickenbacher et al.11	2013				X				Not specified
Baumers et al.12	2016					X	X		Ti-6Al-4V and 17–4 PH
Fera et al.8	2017		X	X		X			Ti-6Al-4V and 17–4 PH
Yang and Li10	2018		X						LS600 Resin
Colosimo et al.13	2019							X	Cobalt chrome, Stainless steel, and Ti-6Al-4V
Lei and Yang14	2020						X		316L, AlSi10Mg, and Ti-6Al-4V
Yi et al.15	2021	X	X						Not specified

DMLS, direct metal laser sintering; EBM, electron beam melting; FDM, fused deposition modeling; LPBF, laser powder bed fusion; SLA, stereolithography; SLM, selective laser melting; SLS, selective laser sintering.

The cost models proposed in Table 2 were based on various technologies, which implies that different parameters were considered for each one. In FDM,⁵ analyzing the following sources of cost: machine, labor, and material of this technology,¹⁵ an app was proposed for extrusion and vat photopolymerization-type AM machines.

Previous studies have shown that different FDM parameters may affect the cost per piece, such as build orientation, layer height, raster angle, and air gap.^16,17 Filament-type material to be extruded for a build is relevant.^3,18 The diversity of materials that could be used for the FDM technology, the characteristics of the FDM printer to process the filament, and the end-use a company gives to this technology are aspects to be rigorously analyzed for the general performance of a company. Therefore, the cost performance of a company depends on the purpose of FDM.

For other AM technologies, the formulated cost models consider the material cost, machine operation cost, and direct labor-oriented cost for setup and postprocessing activities. The proposed model in this study includes these sources of cost and other sources for better estimation.

AM parameters for the cost model

Previous studies have emphasized parameters such as build time, design for AM (DfAM), energy consumption, and other factors that could affect the cost per piece. Hereafter, the parameters to be considered for a cost model in FDM technology is build time, which is one of the tradeoffs in AM technology. According to a previous study,¹⁹ the build time and cost per piece are dependent parameters in every AM technology due to the fabrication process. Several authors have proposed build time estimators for different AM technologies^9,20 to close this gap. Currently, the software is capable to translate an STL file into a G-code file and, with some information, predict the printing time.²¹

On previous study,⁷ from DfAM perspective, authors suggest the redesign of plastic parts that are suitable for AM fabrication and make it less expensive by reducing the material per build. From other perspective, authors in²² analyzed the design of lightweight components and their functional improvements versus the environmental impact in AM manufacturing.

Another parameter associated with a cost per piece is the energy invested in the AM process,²³ including the energy consumption of a DMLS system for multiple part production in the cost estimation of a piece. To achieve transparency in terms of energy and financial inputs of the manufacturing process through AM adoption, on the contrary,^8,14 both authors considered the energy invested, to build parts with various AM technologies, in their cost models. Both the authors considered the energy consumption of a build using different methods.

In the literature, the layer thickness parameter has been studied to determine the minimum value for a cost-effective build and to maintain mechanical performance.^10,24 However, in a previous study,²² the layer thickness parameter was analyzed, together with other parameters, for the environmental and economic impact of AM technology.

Finally, the waste produced in the AM fabrication process is considered a parameter because of the necessity of a support structure in certain pieces.²⁵ The cost model for AM metal fabrication was proposed to evaluate the economic impact of scrap fractions in the AM process.¹³

These AM parameters were analyzed for various AM technologies and can be extrapolated to FDM technology. Table 3 lists the parameters analyzed by the time and authors.

Table 3.

Additive Manufacturing Parameters Authors Considered for Economic Implications

Authors	Year	Build time	Piece orientation/design	Energy consumption	Layer thickness	Scrap/waste
Ruffo et al.20	2006	X
Campbell9	2008	X
Atzeni et al.7	2010		X
Lindermann et al.67	2012	X
Rickenbacher et al.11	2013		X
Baumers et al.23	2013			X
Kellens et al.22	2017		X	X	X
Fera et al.8	2017			X
Yang and Li10	2018				X
Colosimo et al.13	2019			X		X
Medina-Sanchez et al.35	2019	X
Yang et al.17	2019	X
Urbanic and Saqib34	2019	X
Di and Yang14	2020			X
Lawand et al.24	2020	X			X

Throughout the literature, the discussion goes on how each analyzed technology can replace or adapt to a traditional manufacturing environment. Several authors have proposed batch fabrication of the same component with different AM methods^26–28 due to the geometrical freedom of the AM fabrication process, parts distribution in a build is not a problem. Other authors have proposed cost model parameters for spare parts for low to medium production batches.^14,29,30 The authors identified other parameters, such as service location, energy invested in postprocess activities, and mixed production strategies. Moreover, they described other capabilities such as AM performance, design freedom, various components in each production batch, in-house production instead of outsourcing, and the benefits of its adoption.

Some authors have studied the cost of AM products from a life-cycle perspective,^22,24,31,32 and how AM could be beneficial to the overall cost performance from a DfAM perspective. Among the benefits found are the repair/remanufacturing opportunities, environmental performance compared with traditional machining and redesign.³³

AM build time estimation has been well studied in the literature. Some authors have proposed build time estimation techniques for various AM technologies^6,8,11,34,35 to calculate the time an AM machine dedicates to the production process, and therefore, the machine cost involved in each production batch. Besides,³⁶ the AM cost has been estimated using machine learning to predict the time and cost based on historical data, and¹⁷ ways have been explored to optimize FDM extrusion parameters, such as tensile strength and surface roughness, to reach less build time.

Moreover, the AM energy performance in the AM fabrication process is a recent research trend. Several authors have analyzed the energy consumption in AM technologies for metal products because the AM fabrication process requires a considerable amount of energy. The analysis also allows the estimation of the environmental impact^37,38 and cost-effectiveness.³⁹ Moreover, the energy performance of the FDM technology was analyzed and compared with other AM machines for plastic parts,^40–43 design performance,^44–46 and traditional machining.⁴⁷ Description and classification the energy demand in the FDM technology using three different types of FDM machines.⁴⁸ It was stated that the energy demand is concentrated in the machine warmup phase.

Finally, the material selection influences not only the mechanical properties of a part but also the fabrication cost. First, research describes the environmental impact between material selection and the AM method⁴⁹ compared with traditional machining using an eco-indicator. Following the same research line,⁵⁰ the environmental and economic impact of four AM methods using PA 12 in the form of powder and filament is summarized.

On the contrary,^3,16 a review of the material used for FDM and the relationship between the mechanical properties, material selection, and printing parameters is presented; the parameters of FDM technology are reviewed.³ For high-performance materials, several studies have stated that the temperature regulation in the bed and chamber affects the tensile strength of the piece. Temperature regulation may imply variation in the energy consumption of an FDM machine. On the contrary,¹⁶ the printing parameters in the FDM process that may affect the quality and efficiency of the printed part are reviewed and described.

Research Methodology and Model

The research methodology consists of developing a cost model based on the extrusion-type AM process, specifically for thermoplastics, and comparing the costs based on diverse lot sizes and printing conditions. The framework may be suitable for various extrusion-type materials after modification under the given conditions.

Cost model formulation and framework

Various cost-estimation methods were considered to fit our study matter^51,52:

• parametric estimation-based method

• feature-based costing

• activity-based cost.

The cost model was divided into two sources: direct and indirect. For direct costs, material, labor, and machine costs at a machine level were considered. Due to the system size, for indirect costs the production infrastructure and administrative overheads were considered at a factory level. A three-level cost schema is presented in Figure 1 using a breakdown structure. Only direct costs were considered for this framework because they are direct to the FDM printing process and were used to compare the results between printing parameters.

FIG. 1.

Cost breakdown structure of three levels.

For our framework, the cost calculation is articulated as follows:

The total cost of manufacturing in FDM is expressed as follows (Eq. 1):

Direct cost is expressed as the sum of the costs directly invested in the material, labor, and machine (Eq. 2):

The material cost and the sum of the probability of custom failures (Eq. 3). The custom failures are warping and overhang.

C_procurement is the material cost per kilogram multiplied by kilograms to build a piece. C_waste is the result of multiplying the C_procurement and the probability fixed at 4%, according to previous studies.^53,54 It is assumed that the final pieces with these failures were discarded. It could be interpreted as an opportunity cost to build using FDM.

Labor cost (C_labor) is the result of the hourly rate multiplied by the time dedicated to the piece in Eq. (4).

The following activities are considered in the dedication time of an operator to a piece: machine setup and piece postprocessing activities. The time assumed for these activities is 5 min for the machine setup, which includes the time consumed preparing the 3D model information into the machine. Ten minutes of postprocessing activities included the removal of pieces from the build plate.

Machine cost (C_machine) is given in Eq. (5):

Certain assumptions were made for the C_machine listed below:

Machine depreciation is linear and shared in 5 years.

Machine maintenance is 9% of the annual machine depreciation. In the present work, AM equipment maintenance is proportional to the machine price. The percentage was obtained from the authors who considered it in their cost models.^12,15

The energy cost is due to the energy consumed by the FDM machine solely.

The C_acquisition is described in Eq. (6):

where I_usage is an index of time between the time spent for a single build and the total time the FDM machine is available, considering 24 h a day for 1 year.

The C_maintenance is expressed in Eq. (7):

C_energy includes the energy needed for the preheating process, heat for the material extrusion, machine idle state, and axis movement during the build process multiplied by the kilowatt per hour price (Eq. 8).

E_preheat is the energy utilized to start heating the chamber and build plate. If the system has some heat loss, E_heat is the energy necessary to maintain the temperature in the chamber and build plate during the whole printing time. E_idle, E_mov, and E_cooling are the energy invested by the FDM machine in the standby mode, axis and extruder motors, and chamber cooling system (fan), respectively. Finally, E_material is the energy required to reach the melting point of the material.

The idle-state mean power is estimated as 1% of the machine mean power based on a previous study.⁴⁷ For this study, a 10-min pause between prints was considered.

According to some authors, in which a similar FDM printer was used, three quarters of the mean power consumed was invested in the preheating process.^45,48 For this study, it is assumed that for the energy invested in the preheating process, three quarters of the mean power of the machines selected in Eq. (9):

where E_{total_heat} is the sum of the energies invested in the preheat and maintains the temperature in the system.

Finally, E_cooling for this framework is a constant value and could be pointed as the multiplication of the mean power of the cooling system, for our purpose 0.5 kW by 30 min to cool down the system. The mean power depends on the machine size and available build volume.^46,55

Model and Case Study Assumptions

The proposed model is designed for companies that evaluate the implementation of the FDM technology in their manufacturing process; for low to medium production batches defined by short production series and low to medium build volumes used normally for spare parts, tooling, and personalized products to support production activities. Practitioners evaluate cost performance and seek to minimize technology costs versus the mechanical characteristics, and end-use of the part. Due to the variety of FDM 3D printers in the market, we look forward to comparing a desktop printer with the one capable to process high-performance materials.

The material selection is based on the capabilities of FDM 3D printers: ABS in Ultimaker S5⁵⁶ and polyetheretherketone (PEEK) in CreatBot PEEK-300.⁵⁷ Both 3D printers have a similar printing area available for the batch planned for the cost model test.

Ultimaker S5 is a conventional desktop FDM printer that is capable of processing diverse types of plastic polymers. CreatBot PEEK-300, also an FDM printer, due to its characteristics is capable of processing high-performance thermoplastics, such as PEEK, ULTEM, and PEI. Moreover, a comparison between the cost of each specimen and the specifications of each FDM machine was performed.

To evaluate the effectiveness of the cost model, a type 1 tensile test specimen⁵⁸ was simulated on CURA software Figure 2. Tensile test specimen type 1 according to the 2016.²¹ Some tests were performed using the same specimens. The FDM machines were compared for different lot sizes in a build plate, and to test the cost-effectiveness, they were measured in various scenarios. The purpose of the test is to evaluate the effectiveness of the model by comparing the differences between the selected batch, materials, and printers. The differences should be proportional to the lot size and the type of material in which the specimen is printed.

FIG. 2.

Tensile test specimen type 1 according to the 2016 ASTM D638-140 standard.

Certain assumptions were made for the model formulation and are listed below:

• The model was performed for a short production lot size from the same reference.

• Custom failure, warping, and overhang were considered in the cost model as possible sources of waste.

• Into the machine cost are considered the transportation fees and equipment installation.

• No warehousing, disposal, or stocking cost was considered.

• One nozzle extrusion for one type of material was considered.

• One operator for the preparation process and postprocessing with a 10 € hourly rate was considered according to a previous study.⁵⁹

During AM processes, it is necessary to engage an operator. The operator is involved in preprocessing, such as machine setup and filament feed, and introduces the STL file into the FDM machine. After building the part, the operator removes it from the build plate and performs the necessary postprocessing operations. For this study, the postprocessing operation, to simplify, is simply part removal from the build plate.

Results

Model test

One of our goals is to analyze how the proposed cost model can predict the cost of a piece in a batch. For cost model tests A and B, the following data were used (Table 4).

Table 4.

Data Type and Values Used

Item	Test A	Test B
Material
Name	PEEK	ABS
Material price	700 €/kg	30 €/kg
Machine
FDM machine	CreatBot PEEK-300	ULTIMAKER S5
Machine cost	12,000	6,000
Depreciation	5 years	5 years
Maintenance	9%	9%
Availability	8760 h
Build volume (mm)	300 × 300 × 400	330 × 240 × 300
Bed temperature*	139°C	100°C
Chamber temperature*	185°C	N/A
Hot-end temperature	373°C	220°C
Preheating temperature	160°C (bed) and 90°C (chamber)	N/A
Preheating time	30 min	N/A
Cooling time	30 min	N/A
Printing speed	25 mm/s
Idle time	10 min
Mean power	2.25 kW	0.17 kW
Labor
Operator salary (€/year)	20,800
Time dedication	15 min/batch

N/A, not applicable.

The selected characteristics for the material extrusion are based on each material datasheet.^60,61 However, most of these parameters are not critical for a cost comparison. For consistency, the selected machines had the same depreciation time and 9% of the annual depreciation for the maintenance fee. For the machine operation, 7.5 min for machine setup and 7.5 min in batch postprocessing activities were considered.

For Test A, 12 printing cases were considered, with an increasing number of parts. A summary of the results for each lot size is presented in Table 5.

Table 5.

Results on Test A

Specimen per build	Time × build (h)	Cmaterial (€ per piece)	Clabor (€ per piece)	Cacquisition (€ per piece)	Cmaintenance (€ per piece)	Cenergy (€ per piece)	Cost per piece (€)
1	1.03	4.79	2.50 €	0.28	0.03	0.48	8.08 €
2	2.05	4.79	1.25 €	0.28	0.03	0.40	6.75 €
3	3.08	4.79	0.83 €	0.28	0.03	0.37	6.30 €
4	4.10	4.79	0.63 €	0.28	0.03	0.36	6.08 €
5	5.25	4.79	0.50 €	0.29	0.03	0.36	5.96 €
6	6.27	4.79	0.42 €	0.29	0.03	0.35	5.87 €
7	7.30	4.79	0.36 €	0.29	0.03	0.35	5.81 €
8	8.32	4.79	0.31 €	0.28	0.03	0.34	5.76 €
9	9.35	4.79	0.28 €	0.28	0.03	0.34	5.72 €
10	10.48	4.79	0.25 €	0.29	0.03	0.34	5.70 €
11	11.52	4.79	0.23 €	0.29	0.03	0.34	5.67 €
12	12.53	4.79	0.21 €	0.29	0.03	0.34	5.65 €

The C_material values in the 12 cases analyzed in Test A were equal because the volume of the specimen did not change. C_labor is proportional to the number of printed batches. The operator dedication times in the setup and postprocessing activities were distributed between the number of pieces in the batch. Moreover, C_acquisition and C_maintenance remain similar between batches (Fig. 3a).

FIG. 3.

(a) A bar chart indicates the percentage weight of the material, labor, acquisition, maintenance, and energy cost; and how they change between the batches in Test A. (b) A Pareto chart for eight piece batch in Test A.

In the eight piece batch, among all the cost sources, C_material is proportionally the highest with 89% of the total cost, followed by C_energy with ∼6% (Fig. 3b).

The cost per piece variation between batches tends to decrease in the larger batches. The material remains constant and is independent of the lot size; however, the other cost sources are dependent on the lot size, as shown in Figure 4.

FIG. 4.

Bar chart of the cost per piece between the 12 batches. Every bar expresses the material, labor, acquisition, maintenance, and energy cost per batch in Test A.

Besides, the results for Test B are shown in Table 6.

Table 6.

Results on Test B

Specimen per build	Time × build (h)	Cmaterial (€ per piece)	Clabor (€ per piece)	Cacquisition (€ per piece)	Cmaintenance (€ per piece)	Cenergy (€ per piece)	Cost per piece (€)
1	1.00	0.24 €	2.50 €	0.14 €	0.01 €	0.03 €	2.92 €
2	2.03	0.24 €	1.25 €	0.14 €	0.01 €	0.02 €	1.67 €
3	3.05	0.24 €	0.83 €	0.14 €	0.01 €	0.02 €	1.25 €
4	4.07	0.24 €	0.63 €	0.14 €	0.01 €	0.02 €	1.04 €
5	5.08	0.24 €	0.50 €	0.14 €	0.01 €	0.02 €	0.91 €
6	6.12	0.24 €	0.42 €	0.14 €	0.01 €	0.02 €	0.83 €
7	7.13	0.24 €	0.36 €	0.14 €	0.01 €	0.02 €	0.77 €
8	8.15	0.24 €	0.31 €	0.14 €	0.01 €	0.02 €	0.73 €
9	9.17	0.24 €	0.28 €	0.14 €	0.01 €	0.02 €	0.69 €
10	10.20	0.24 €	0.25 €	0.14 €	0.01 €	0.02 €	0.66 €
11	11.22	0.24 €	0.23 €	0.14 €	0.01 €	0.02 €	0.64 €
12	12.25	0.24 €	0.21 €	0.14 €	0.01 €	0.02 €	0.62 €

The weight between the cost sources for Test B using ABS is presented in Figure 5a, and the Pareto analysis from the eight piece batch shown in Figure 5b.

FIG. 5.

(a) A bar chart indicates the percentage weight of the material, labor, acquisition, maintenance, and energy cost; and how they change between the batches in Test B. (b) A Pareto chart for eight piece batch in test.

According to Figures 3a and 5a, all cost sources except C_material tend to decrease proportionally to the number of pieces, and the relationship between cost sources remains similar in both tests. Besides, in Figures 3b and 5b in the Pareto analysis, for Test A C_material represents >80% of the cost sources found; on the contrary, for Test B C_labor and C_material represent ∼76% of all the cost sources. In both analyses, the eight piece batch is shown. The relationship between the cost sources and lot size in Test B is shown in Figure 6.

FIG. 6.

Bar chart of the cost per piece between the 12 batches. Every bar expresses the material, labor, acquisition, maintenance, and energy cost per batch in Test B.

The results now provide evidence that cost savings can be measured using our cost model framework. The percentage of cost savings found between one specimen and different lot sizes is shown in Figure 7.

FIG. 7.

Line graph plotting the percentage of savings between one piece batch to the rest of the 11 batches in Tests A and B.

A trend to a constant value of cost per piece can be observed as the lot size increases. These results indicate that the proposed model predicts the possible cost scenarios from different scheduled lot sizes.

Energy consumption

Currently, the energy cost involved in manufacturing is gaining relevance.⁶² Research in energy goes through the identification of new sources and environmental compatibility; meanwhile, energy price forecasts have shown an overall upward trend.⁶³ A sensitivity analysis was performed in both tests to prove the energy cost effective in the proposed model using eight piece batch (Fig. 8).

FIG. 8.

(a) A double ring chart of the cost weight distribution in Test A for an eight piece batch. (b) A double ring chart of the cost weight distribution in Test B for an eight piece batch.

The inner ring in Figure 8a and b indicates the proportion of cost with an energy cost per kilowatt-hour of 0.08 €; on the contrary, the outer ring indicates the proportion of cost with an energy cost per kilowatt-hour of 0.21 €.⁶⁴

For Test A, the difference in energy cost diminishes the C_material weight proportionally by 7% between the other costs. On the contrary, for Test B, all the costs change proportionally by 1%, except for C_energy, which increases by 4%.

The cost structures in Tests A and B were relatively different. For Test A, C_material is proportionally the most relevant in the cost structure, followed by C_energy, possibly for the relation between the material cost and the energy necessary to process the PEEK material. Moreover, the cost structure in Test B is relatively distributed among all the cost sources, except C_maintenance and C_energy. For Test B, ABS does not require special features, such as PEEK. Besides, C_labor is proportionally the highest cost in Test B, which could impact the final product cost.

Discussion

This study was conducted to make a theoretical contribution to AM cost models through different cost-estimation approaches. From other AM technologies in the market, FDM was considered because of its versatility, the range of raw material costs, and the lack of contribution in the literature from a cost model estimation approach.

Other studies have shown a similar cost structure with other AM technologies⁶ and a cost model based on continuous production.⁵ However, when comparing our model with those of older studies, it must be noted that our framework considers the energy performance of the FDM machine based on the mean power and material selected. To the best of our knowledge, this is the first cost model in FDM technology that considers energy consumption for a build cost estimation.

This study confirmed the differences between the costs of a build with different parameters. Material-type selection is a source of cost itself, but when the machine type and parameters to be used are considered, the cost difference is considerable. Our results show that the cost of a single specimen in Test A is 2.8 times more than that of a single specimen in Test B. Moreover, for a 12 piece batch the cost per piece in Test A is 9.10 times more than the cost per piece in Test B. The material cost for both tests is a fixed cost based on grams of materials used per piece. However, the costs related to machine operation vary because of the energy required to process each material.

To process the PEEK material for each specimen, the energy required in the eight piece batch is 16 times more than the energy required to process the ABS for each specimen. Energy consumption is relatively low; this is consistent with the findings of previous studies.^12,65 However, energy consumption increases proportionally to the technology necessary to process the selected material. According to the literature, the energy cost per kilowatt-hour is increasing, making it important to be considered for the investment in AM technology.

These results go beyond previous reports, showing that this framework can be adapted to any printer and any type of filament for FDM machines. Moreover, the cost model schema considers the direct cost for FDM-type printer parameters, to the best of our knowledge, not considered before, such as the material waste, energy consumption using the machine mean power, and the maintenance cost as a percentage of the annualized depreciation. It could be noted that our framework considers, instead of other approaches, that the effective energy invested in the printing process is dependent on the machine characteristics and material type. The proposed cost model provides information to the managerial team for material-type selection and the machine type necessary to process each type of material and its influence on cost performance.

After six to seven pieces per lot, the cost savings were not significant in either test. By using the proposed framework, it is possible to calculate the minimum lot size for the optimal economic profit level. This implies that for users, the market, and companies, a way to calculate the cost savings is to take advantage of the machine size.

Conclusion

In this article, an innovative cost model framework for FDM technology was proposed. The framework proposed considered direct and indirect costs. For the direct costs, the model considers the cost of machine, labor, and material. Other cost models were previously proposed for FDM, according to the literature review presented, although low production volumes and printing parameters of the technology itself were not specifically considered. Major part of the existing analysis is centered in a full productivity scenario, where the intended use for the printing machine is to maximize its use.

Nevertheless, many companies start to use AM for specific tasks such as spare parts production, tooling, or R&D activities; tasks characterized by low to medium production batches and low to medium build volumes where the main goal is to increase the production flexibility. This is the reason to simulate two cases based on the use of printers with a limited printing area and cost, in comparison with full productivity scenarios with large printing area machines and high cost. These two cases also allow comparing the use of AM for conventional polymer materials with high-performance polymers.

The key conclusion of our results that can be pointed are as follows:

• The cost per piece is affected by the machine size, heated build plate, isolated chamber, and material type, not considered together before.

• The framework applied, for both test cases, can show the different specific weight of material adoption in the cost structure.

• The filament selection and machine characteristics are dependent on the desired mechanical properties. Therefore, the framework can be used to predict the best machine size and material type that are suitable for certain situations.

• Energy invested in the process depends directly on the machine type and desired material.

• Waste due to support structure, warping, and overhang is a source of cost, and depends proportionally on the build shape, size, and position of the printed parts.

It is also interesting to note that in both example tests, there is not much improvement in the cost per part when printing more than eight parts per lot; that is, ∼65% of the printing area. Therefore, this could justify the use of AM even in not fully production scenarios.

Finally, although FDM is not an AM process so energy intensive as others, it is shown that test case A (PEEK material) is quite sensitive to energy costs compared with case B (ABS material), or conversely, the printing of conventional polymer materials is not very sensible to the energy cost.

The results provide evidence to open new research lines related to the optimization of each parameter of the proposed framework. Future research should expand the framework to consider other dimensions, such as logistics cost in distribution, spare part inventory versus AM production in-house, and sustainability.

Authors' Contributions

M.P.L. provided conceptualization, methodology, and data curation. J.S. contributed to conceptualization and methodology. J.R. performed methodology, supervision, and validation. J.A.A. contributed to supervision, formal analysis, and writing—reviewing and editing. L.N. accomplished writing—reviewing and editing. M.E.H.K. contributed to writing—original draft preparation, data curation, and visualization.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.