• International Journal of Technology (IJTech)
  • Vol 13, No 8 (2022)

Fabrication of Rigid Polyurethane Foam Lumbar Spine Model for Surgical Training using Indirect Additive Manufacturing

Fabrication of Rigid Polyurethane Foam Lumbar Spine Model for Surgical Training using Indirect Additive Manufacturing

Title: Fabrication of Rigid Polyurethane Foam Lumbar Spine Model for Surgical Training using Indirect Additive Manufacturing
Asriyanti, Kushendarsyah Saptaji, Nisa Khoiriyah, Muhammad Satrio Utomo, Made Subekti Dwijaya, Farid Triawan, Muhammad Hanif Nadhif

Corresponding email:

Cite this article as:
Asriyanti, Saptaji, K., Khoiriyah, N., Utomo, M.S., Dwijaya, M.S., Nadhif, M.H., Triawan, F.,  2022. Fabrication of Rigid Polyurethane Foam Lumbar Spine Model for Surgical Training using Indirect Additive Manufacturing. International Journal of Technology. Volume 13(8), pp. 1612-1621

Asriyanti 1. Mechanical Engineering Department, Faculty of Engineering and Technology, Sampoerna University, Jl. Raya Pasar Minggu No. Kav. 16, Jakarta 12780 Indonesia, 2. Department of Aerospace and Mechanic
Kushendarsyah Saptaji Mechanical Engineering Department, Faculty of Engineering and Technology, Sampoerna University, Jl. Raya Pasar Minggu No.Kav. 16, Jakarta 12780 Indonesia
Nisa Khoiriyah 1. Mechanical Engineering Department, Faculty of Engineering and Technology, Sampoerna University, Jl. Raya Pasar Minggu No.Kav. 16, Jakarta 12780 Indonesia, 2. Department of Aerospace and Mechanica
Muhammad Satrio Utomo 1. National Agency for Research and Innovation, Kawasan PUSPIPTEK, Banten 15314 Indonesia, 2. Faculty of Medicine, University of Indonesia, Jakarta 10430 Indonesia
Made Subekti Dwijaya National Agency for Research and Innovation, Kawasan PUSPIPTEK, Banten 15314 Indonesia
Farid Triawan Mechanical Engineering Department, Faculty of Engineering and Technology, Sampoerna University, Jl. Raya Pasar Minggu No.Kav. 16, Jakarta 12780 Indonesia
Muhammad Hanif Nadhif Faculty of Medicine, University of Indonesia, Jakarta 10430 Indonesia
Email to Corresponding Author

Fabrication of Rigid Polyurethane Foam Lumbar Spine Model for Surgical Training using Indirect Additive Manufacturing

Lumbar model is an artificial bone that is commonly used in surgical training to simulate working with the human-like bone for the trainer. The common lumbar model is made of rigid polyurethane (PU) foam and is produced using casting. However, the current lumbar model is expensive and has limitations in representing the real human lumbar, especially in geometry, visuals, and haptics. Therefore, an alternative method of fabricating lumbar models made of rigid polyurethane for surgical training using indirect additive manufacturing will be investigated in this paper. The proposed indirect additive manufacturing is a combination of 3D printing and casting methods. The main process of this method is started by fabricating a mold made of polyvinyl alcohol (PVA) using fused deposition modeling (FDM) 3D printing and subsequently casting PU foam material into the 3D printed PVA mold. Accordingly, the aim of this study is to find the optimized casting process parameters, especially for injecting the material into the mold, to achieve a better quality of lumbar model. The study was conducted using a Design of Experiment (DoE) Taguchi Orthogonal Array to optimize the casting process. The geometrical measurements of middle end-plate depth, upper end-plate width, spinal canal width, spinal canal depth, and lower pedicle length show the error ranged from 0.14% to 0.85%. The average porosity, measured from the body, lamina, and spinous, was found to be non-uniform. It is ranged from 19.58% to 21.94% on the middle part and 39.78% to 45.41% on the subsurface of lumbar model. The density was increased by 64.89% compared to the reference open molded PU foam.

Indirect additive manufacturing; Lumbar spine model; Rigid polyurethane (PU) foam; Surgical trainingSurgical training


     Lumbar is located at the lower part of the spine and has a function to support the upper body and protect the spinal cord (Frost et al., 2019). Due to its heavy functions, lumbar is prone to be injured and surgery is required to restore its function. As a result of the increasing use of lumbar surgery, a growing demand for lumbar spine model used in surgical training is also increasing (Lewandrowski et al., 2020). Moreover, post-surgery monitoring is necessary to ensure the success of the surgical and implantation processes, such as in the dental implant system (Genisa et al., 2020).
      There are several challenges in creating a lumbar spine model, and all the challenges lead to one fundamental issue where the models cannot accurately replicate the genuine parts in terms of visual, geometric, and haptic feedback. Various fabrication methods have been used to produce the lumbar spine model, such as machining, casting, and additive manufacturing. Machining is a traditional or subtractive manufacturing method used to make prostheses (Rani et al., 2017). This method is popular due to its low cost, high surface quality, and ease of use. However, this method is limited in the shapes that can be produced since it only moves in three axes for conventional machining and five axes for computer numerical control (CNC) machining (Kong et al., 2020). In addition, the machining process can also induce residual stress on the machined workpiece, which can initiate failure (Saptaji et al., 2019). The casting process is also widely used due to its ability to produce complex shapes with a variety of materials. However, the casting process is time consuming and has low dimensional accuracy, resulting in a low geometrical representation of the lumbar model (Lyashenko et al., 2018). In recent study, in order to improve the process efficiency and accuracy, additive manufacturing or 3D printing is being introduced due to its cost effectiveness, customizability, and able to build complex shaped model (Bai et al., 2019; Hanon et al., 2021).  Fused Decomposition Modeling (FDM) is the most popular 3D printing method, in which the object is built layer by layer by using the extrusion method to melt the raw material in the form of filament, commonly made of polymer, through the nozzle (Hadisujoto et al., 2021; Mwema & Akinlabi, 2020). However, FDM is limited in the material used because the method does not support the printing process of the widely used material for the lumbar model (Clifton et al., 2019). Meanwhile, the material for the bone structure such as lumbar spine must have similar structure and properties (Saptaji et al., 2022). Rigid polyurethane (PU) foam is the most frequently used material. It is widely used in lumbar spine models due to its similarities with lumbar properties (Shim et al., 2012). Computational validation on thermoplastic polyurethane with lattice structure for intervertebral disc replacement also showed that the material is a suitable candidate (Nadhif et al, 2021). However, the fabrication of PU foam to become a lumbar model is only possible through the casting method, which is inefficient and has poor accuracy (Gama et al., 2018).
      Due to all the challenges faced in fabricating lumbar model, an indirect additive manufacturing approach can be used to obtain a better quality of the lumbar spine model. Indirect additive manufacturing can be performed by combining FDM and casting processes to produce the part (Montero et al., 2020). The combination can be implemented since the desired material to fabricate the part is only possible through the casting process, meanwhile traditional casting process is not recommended due to its low accuracy and a long period of time required for mold preparation. Therefore, FDM can be used to produce the mold to improve efficiency and accuracy. The challenges in utilizing indirect additive manufacturing are located in determining the printing parameters and the process of injecting the material into the mold.
      There are no studies reporting efforts to investigate and solve the optimization of the casting process, especially in the lumbar spine model fabrication. One of the challenges in casting PU foam is due to structure stabilization during the foaming and curing processes and its tendency to shrink (Rampf et al., 2011). The study in finding the optimal indirect additive manufacturing parameters, particularly during the casting process, is required in order to produce an excellent quality lumbar model. Therefore, the objective of this study is to find the optimized casting process parameters, especially for injecting the material into the mold. It is expected that a better quality of lumbar model can be achieved by evaluating the properties of a PU lumbar model fabricated using indirect additive manufacturing. The experimental work was conducted using Taguchi Orthogonal Arrays Design of Experiment (DoE). This method is commonly used to optimize the fabrication process by making some assumptions about factors and levels that have a significant effect on the issues (Mondal et al., 2020). The DoE was divided into two stages which are the optimization of printing parameters and casting processes.

Experimental Methods

2.1. Materials

 Materials used in this study are water soluble polyvinyl alcohol (PVA) filament as the casting mold and rigid polyurethane (PU) foam as the material of the lumbar spine model. Rigid PU foam consists of part A (polyol) and part B (diisocyanate) with a ratio of 1:2 and a mixing time of 2.5 minutes (Sidek et al., 2017).
2.2. Tools
    Several tools were used during this study, including software such as SolidWorks, Meshmixer, FlashPrint slicing software, ImageJ, and FlashForge Creator Pro 3D Printer. SolidWorks was used to design the samples. Meshmixer was used to modify the stereolithography (STL) file obtained from the patient’s lumbar computerized tomography (CT) scan. FlashPrint was used to generate g-code that could be printed. ImageJ was used during the porosity characterization.  A FlashForge Creator Pro 3D Printer with a nozzle diameter of 1.75 mm is used to print the PVA mold. The drilling tool is used to make holes in the mold for injection and excess PU foam during casting.

2.3.  Experimental Procedures
2.3.1. Casting process optimization

The experiment was carried out based on the Design of Experiments (DoE) using Taguchi Orthogonal Arrays to determine the optimized parameters for the casting process. Preliminary findings and analysis show partially filled mold cavities occurred during casting of the PU foam due to the number of injections and excess (surplus) holes introduced to the PVA mold issue. Therefore, to ensure that PU foam can completely fill the mold cavity, the focus in DoE was to determine the number of injection and excess holes on the mold.
       The experiment was initially performed by modifying the lumbar CT scan image data to become a lumbar mold using Meshmixer. CT scan image data is also useful in the analysis of a femoral bone fracture in the case of a sideways fall accident (Izmin et al., 2020). An initial experiment was performed prior to the 3D printing process of the mold to identify the best 3D printing parameter variations, particularly mold thickness and travel speed (Haque, 2020; Azhikannickal & Uhrin, 2019).The best travel speed was determined to be 70 mm/s, and the mold thickness was determined to be 2.4 mm. Meanwhile, the other printing parameters were determined based on the literature (Montero et al., 2020; Tagami et al., 2017), such as  printing speed of 50 mm/s, extruder and platform temperatures of 205 0C and 45 0C, infill of 0%, layer height of 0.2 mm,

Table 1 Standard orthogonal array of DoE







Number of injection holes (In)





Number of excess holes (Exc)





Then, the lumbar mold made of PVA was sliced and printed. Subsequently, the printed lumbar mold was drilled to add injection and excess holes. The number of injection and excess holes varied from anterior body (In), spinous (In), articular process (Exc), and transverse process (Exc) based on the standard orthogonal array shown in Table 1. The injection and excess hole positions for the four variations are shown in Figure 1. The casting process was subsequently completed by injecting 9 mL of PU foam. The casting samples were then cured and undergone dissolution process to remove the PVA mold from the beaker glass to obtain the lumbar model. The workflow of the experimental setup is shown in Figure 2.

Figure 1 Locations of injection (In) holes and excess (Exc) holes for (1) Variation 1, (2) Variation 2, (3) Variation 3, (4) Variation 4 (Bozdag & Karaman, 2021)

Figure 2 Experimental setup of lumbar model fabrication

2.3.2. Open molded PU foam

An open molded PU foam was fabricated as the reference properties of PU foam. It was used to compare the effect of indirect additive manufacturing on the PU foam properties. The specimen was prepared in an open aluminum cup that was used as a mold and allowed to foam freely. The procedure is shown in Figure 3.

Figure 3 Illustration on the fabrication process of open molded PU foam

2.4.  Characterization
2.4.1. Morphometry analysis

A morphometric analysis was conducted on evaluate the percent error to the desired dimension. A Vernier caliper with 0.05 accuracy was used to do the characterization. The dimensions of lumbar model between the CT scan data and fabricated lumbar model were compared in several parts, shown in Figure 4, including middle end-plate depth (EPDm), upper end-plate width (EPWn), spinal canal width (SCW), spinal canal depth (SCD), and lower pedicle length (PLI). The requirements of the error based on ASTM F1839 must be less than 5% (ASTM, 2014).

Figure 4 Locations of morphometry analysis performed for lumbar model (Bozdag & Karaman, 2021)
2.4.2. Porosity

    The porosity of the lumbar model was assessed to evaluate the effect of the indirect additive manufacturing method on the uniformity of PU foam distribution. The sample preparation was based on ASTM F1839. A Motic BA310 Microscope with 5x magnification was used to capture the pore distribution in the subsurface and middle part of the lumbar body, lamina, and spinous. The lumbar model that undergone this measurement was limited to those who passed the morphometry analysis. The data obtained from the microscope was exported to ImageJ to measure the pore area. All the data at each location are summed and divided by the cross-sectional area to obtain the porosity.
2.4.3. Density measurement

    Density measurement was conducted based on ASTM 1622 (ASTM, 2020) with the same samples characterized in porosity. It was used to see the effect of the indirect manufacturing process toward the change in the density of PU foam. The density was measured using the density formula, where the mass and volume must be obtained for each specimen. The mass was measured using a digital mass balance. Meanwhile, the volume was determined using the Archimedes method.

Results and Discussion

    There were 12 lumbar models with 3 replications for each variation. The replications were produced in order to examine the consistency of the selected parameters. Figure 5 depicts a representative of each fabricated lumbar variation. During the casting process, it was found that PU foam was expanded throughout the excess holes. It shows that the mold has been completely filled with PU foam material, particularly in the pedicle area.. However, after the dissolution process to remove PVA mold, it was found that there are still some missing parts that have not formed in the lumbar model, even though the PU foam has raised throughout the excess holes. In general, variation 1 produced the perfect shape of the model with no missing parts as compared to other variations.

Figure 5 Superior view of PU foam lumbar model for (1A) Variation 1, (2A) Variation 2, (3A) Variation 3, (4A) Variation 4
3.1.  Morphometry Analysis

   The result of the morphometry analysis of 12 lumbar models, especially for EPDm, EPDu, SCW, SCD, and PLI, is shown in Figure 6. Based on the diagram, the percentage error of the lumbar model compared to the CT scan data is varied. In general, the percent error of variation 1 in every location is always below the percentage error limit, which ranges from 0.14% to 0.85%. In contrast, variation 2, variation 3, and variation 4 have some dimensions that exceed the error limit, ranging from 0.19% to 14.68%, 0.55% to 18.68%, 0.37% to 10.05%, respectively. Variation 2 is exceeded the error limit in EPDm and EPWu. Meanwhile, variation 3 surpasses the error limit in EPDm, EPWu, and SCD. Lastly, variation 4 exceeds the limit on EPWu and SCD.

Figure 6 Dimensional error of fabricated lumbar model relative to CT scan data

    The phenomenon where some variations exceeded the error limit are due to the missing parts on the lumbar model. For instance, in variation 2, there are some missing parts for every repetition on the lumbar body as shown in Figure 7. This missing part significantly increases the error in the EPWu dimension compared to the CT scan data. In this experiment, the casting process was performed by injecting PU foam to the mold from the anterior side vertically. As a result, the material was poured directly into the mold's bottom and raised to the interior side. However, due to the number and location of excess holes in the bottom, during the foaming process the material was prone go to the excess holes massively, resulted in mold pressure reduction (Lyashenko et al., 2018) and interparticle bonding strength between the material. Consequently, some parts of the lumbar model are fractured and collapsed during the dissolution of PVA mold.
The repeatability test was also performed on this characterization by considering the relative standard deviation (RSD). For all variations, the value of RSD is below 10% indicating that all variations are likely to have the same dimensions for every repetition. Accordingly, from this morphometry analysis, variation 1 shows the best result, in which the percent errors in dimension for every repetition are not exceeding the 5% limit.