1. Introduction
Prototyping is a critical activity within the new product development process (Wall et al. Reference Wall, Ulrich and Flowers1992), leading to improved products, stakeholder alignment and innovation (Camburn et al. Reference Camburn, Viswanathan, Linsey, Anderson, Jensen, Crawford, Otto and Wood2017; Menold et al. Reference Menold, Jablokow and Simpson2017) and reduced risk (Houde & Hill Reference Houde, Hill, Helander, Landauer and Prabhu1997). A considerable body of research has focused on the improvement of prototyping via processes and media, including the study of activity purpose (Ulrich & Eppinger Reference Ulrich and Eppinger2016), the role of prototypes (Houde & Hill Reference Houde, Hill, Helander, Landauer and Prabhu1997; Lauff et al. Reference Lauff, Kotys-Schwartz and Rentschler2018; Petrakis et al. Reference Petrakis, Hird and Wodehouse2019; Real et al. Reference Real, Snider, Goudswaard and Hicks2021), audience (Bryan-Kinns & Hamilton Reference Bryan-Kinns and Hamilton2002; Crilly et al. Reference Crilly, Moultrie and Clarkson2004), scope (Ulrich & Eppinger Reference Ulrich and Eppinger2016), functionality (Pei et al. Reference Pei, Campbell and Evans2011), structure (Jones et al. Reference Jones, Gopsill, Real, Snider, Felton, Kent, Goudswaard, Gebler and Hicks2024) and fidelity (McCurdy et al. Reference McCurdy, Connors, Pyrzak, Kanefsky and Vera2006; Sauer & Sonderegger Reference Sauer and Sonderegger2009; Liker & Pereira Reference Liker and Pereira2018). Further work has considered heuristic prototyping frameworks that help direct the overall approach and prototyping effort (Camburn et al. Reference Camburn, Dunlap, Gurjar, Hamon, Green, Jensen, Crawford, Otto and Wood2015; Menold et al. Reference Menold, Jablokow and Simpson2017) – what methods to apply, when and how – and considered adaptation of prototyping strategies based on context, culture and accessibility across stakeholders (Deininger et al. Reference Deininger, Daly, Lee, Seifert and Sienko2019; Rodriguez-Calero et al. Reference Rodriguez-Calero, Coulentianos, Daly, Burridge and Sienko2020).
As an activity focused on learning (Goudswaard et al. Reference Goudswaard, Real, Snider, Camargo, Zamora and Hicks2023), it is important that prototyping and prototypes allow exploration. Significant barriers here then exist due to the cost and time of fabrication (Otto & Wood Reference Otto and Wood2001; Viswanathan & Linsey Reference Viswanathan and Linsey2010; Camburn et al. Reference Camburn, Dunlap, Gurjar, Hamon, Green, Jensen, Crawford, Otto and Wood2015). Thomke (Thomke Reference Thomke1998) observed that it is essential to prototype effectively by reducing cost and time, as these factors then limit prototype use (Otto & Wood Reference Otto and Wood2001; Christie et al. Reference Christie, Jensen, Buckley, Menefee, Ziegler, Wood and Crawford2012; Camburn et al. Reference Camburn, Dunlap, Gurjar, Hamon, Green, Jensen, Crawford, Otto and Wood2015). Managing resources to reduce the cost and time of prototyping is crucial for the success of the product development process. It enables earlier design evaluation and more frequent prototyping iterations (Camburn et al. Reference Camburn, Viswanathan, Linsey, Anderson, Jensen, Crawford, Otto and Wood2017), which in turn may stimulate innovation (Viswanathan & Linsey Reference Viswanathan and Linsey2010), reduce design fixation (Youmans Reference Youmans2011), accelerate the design process (Neeley et al. Reference Neeley, Lim, Zhu and Yang2014) and produce better outcomes (Yang Reference Yang2008).
More generally, improvements to prototyping can be realised via two mechanisms: activity improvements and tool improvements. Activity improvements focus on frameworks that guide designers on when and how to implement prototyping (e.g. Christie et al. Reference Christie, Jensen, Buckley, Menefee, Ziegler, Wood and Crawford2012; Camburn et al. Reference Camburn, Dunlap, Gurjar, Hamon, Green, Jensen, Crawford, Otto and Wood2015; Menold et al. Reference Menold, Jablokow and Simpson2017). Tool improvements involve the development of methods and technologies to reduce fabrication time, cost, complexity and skill requirements, (e.g. Hildebrand et al. Reference Hildebrand, Bickel and Alexa2012; Mueller et al. Reference Mueller, Im, Gurevich, Teibrich, Pfisterer, Guimbretière and Baudisch2014a; Goudswaard et al. Reference Goudswaard, Hicks, Gopsill and Nassehi2017; Ranscombe et al. Reference Ranscombe, Zhang, Rodda and Mathias2019). Many novel methods have been proposed for such barriers, including editable models (Lennings et al. Reference Lennings, Broek, Horváth, Sleijffers and de Smit2000), the reuse of models (Camburn et al. Reference Camburn, Viswanathan, Linsey, Anderson, Jensen, Crawford, Otto and Wood2017; Real et al. Reference Real, Snider and Hicks2022), digital augmentation (Cox et al. Reference Cox, Gopsill, Snider and Hicks2024; Snider et al. Reference Snider, Kukreja, Cox, Gopsill and Kent2024) and the use of construction kits (Boa et al. Reference Boa, Mathias and Hicks2017). However, most of them concentrate on singular methods of prototype construction, addressing one of these mechanisms. A few researchers have considered both (e.g. (Mueller et al. Reference Mueller, Mohr, Guenther, Frohnhofen and Baudisch2014b; Song et al. Reference Song, Deng, Wang, Dong, Li, Fu and Liu2016; Mathias et al. Reference Mathias, Hicks and Snider2019a) via the utilisation of mixed-media prototyping with a design process.
Although limited, academic studies have shown great potential in coupling different prototyping media to leverage their complementary affordances while concomitantly mitigating their limitations. Examples include using 3D wire printing in combination with laser cutting (Mueller et al. Reference Mueller, Im, Gurevich, Teibrich, Pfisterer, Guimbretière and Baudisch2014a; Beyer et al. Reference Beyer, Gurevich, Mueller, Chen and Baudisch2015) and the combination of LEGO and 3D printing (Mathias et al. Reference Mathias, Snider, Hicks and Ranscombe2019b. These approaches demonstrated that considerable benefits can be realised in terms of reduced fabrication time, reduced material use, increased prototype flexibility and improved prototype functionality. This class of mixed-method prototype is here defined as hybrid prototyping (HP).
While HP is a relatively recent term, there are several reported examples involving combinations of prototyping techniques that could be considered hybrid. Beyer et al. (Reference Beyer, Gurevich, Mueller, Chen and Baudisch2015) combined laser-cut sheet plastic and low-cost 3D printing. This accelerated low-fidelity fabrication allowed the fabrication of large-scale 3D objects (Song et al. Reference Song, Deng, Wang, Dong, Li, Fu and Liu2016 and housed functional components within printed parts (Gao et al. Reference Gao, Zhang, Nazzetta, Ramani and Cipra2015). Similarly, a few researchers combined construction kits (in the form of LEGO) with low-cost 3D printing (Mathias et al. Reference Mathias, Hicks and Snider2019a; Mueller et al. Reference Mueller, Mohr, Guenther, Frohnhofen and Baudisch2014b to reduce fabrication times. Mathias et al. (Reference Mathias, Snider, Hicks and Ranscombe2019b) showed that HP could achieve a 45% reduction in fabrication time and a 57% saving in resource cost in a simple case.
These examples suggest the disruptive potential of HP in terms of delivering step changes in cost, time and ease of prototype fabrication. However, while HP does occur in practice, implementations remain largely informal and ad-hoc. Research has similarly been limited, mostly exploring at a conceptual level, and failing to consider that HP may be employed in varying ways and thus may realise different degrees of benefit.
Recognising that prototyping with different media brings different benefits (Pei et al. Reference Pei, Campbell and Evans2011; Camburn et al. Reference Camburn, Viswanathan, Linsey, Anderson, Jensen, Crawford, Otto and Wood2017; Kent et al. Reference Kent, Snider, Gopsill and Hicks2021; Snider et al. Reference Snider, Goudswaard, Ranscombe, Hao, Gopsill and Hicks2023), this paper considers that further benefits may be realised by varying the way in which HP is implemented across its media in a given case (here termed the HP approach). These approaches would be chosen according to the intent and goals of the specific prototyping activity. As such, this paper moves beyond the establishment of the basic benefit of HP in isolated cases as performed in other works (including those of the authors (see Mathias et al. Reference Mathias, Snider, Hicks and Ranscombe2019b), instead considering hybridisation approaches and the interrelationship between benefit, intent and approach by which HP is employed with consideration of real-world prototyping goals.
To do so, the paper proposes three hybridisation approaches for the combination of Lego and 3D printing, each aligned with a different design intent. It then quantitatively investigates the relative benefits of each across three simulation cases, before exploring and verifying emergent benefits for a real product prototype. In addition, the paper considers the constraints and potential good practice guidance for HP implementation, providing the foundations for a first-of-a-kind methodology and HP support tool. It follows that the novelty and contribution of this paper with respect to previous works lies in the elicitation of good practice guidance, the development and evaluation of hybridisation approaches, the investigation and quantification of relative benefits between approaches and their demonstration via a real-world case.
The paper is organised as follows. Section 2 presents the three proposed approaches for hybridisation concerning mixed fidelity, parallelisation and prototype reuse. To investigate the feasibility and relative benefits of these approaches, practical implementation considerations for hybridisation are developed and incorporated into a prototyping support tool in Section 3. The developed HP tool is then used to simulate the application of the hybridisation approaches to three artefacts, exploring time/cost benefits and trade-offs in Section 4 before further verification and validation through application to a real-world design project in Section 5. Finally, the relative advantages and limitations of the hybridisation approaches are discussed alongside the generalisability of the method and future research directions in Section 6.
2. Hybrid prototyping with LEGO and 3D printing
In considering the coupling of prototyping techniques, it is important to appraise the relative properties and characteristics of each so that they can be coupled to maximise overall benefit. Taking an assumption that similar techniques would result in largely similar benefits, it stands that broader benefits may be realised through the coupling of dissimilar techniques. In line with this reasoning, this paper considers the hybridisation of construction kits (LEGO) with additive manufacturing (3D Printing) to: a) maximise potential benefits that may be identified and b) consider the hybridisation of two intuitively less compatible techniques.
Based on the beneficial characteristics of each media (Mathias et al. Reference Mathias, Hicks and Snider2019a, Reference Mathias, Snider, Hicks and Ranscombeb ), the fundamental approach for LEGO/3D printing HP is depicted in Figure 1. Here, LEGO occupies the internal volume of a prototype with 3D-printed components providing a high-fidelity outer surface, itself affixed to the standard LEGO interface. This method utilises LEGO to realise rapid assembly, reducing print time and doing so with reusable parts. Such an approach can be considered analogous to CNC machining, where the LEGO is a ‘rough cut’ that forms the quick, approximate shape and the 3D printing is a ‘finishing pass’ that creates high fidelity detail. For the purpose of this paper, the standard LEGO library of bricks (c.f. part (a) of Figure 3) was considered. In Mathias et al. (Reference Mathias, Snider, Hicks and Ranscombe2019b), this approach is shown to provide up to a 45% time saving over 3DP alone, including assembly time.
Figure 1. An illustration of LEGO and 3D printing implementation of HP.
As shown in Table 1, these two prototyping methods can be considered to concomitantly complement and mitigate one another, giving rise to notable opportunities and synergies that include:
-
1. Coupling low-cost 3D printing and LEGO introduces a level of fidelity unachievable by LEGO alone while maintaining the flexibility and reconfigurability of a LEGO construction kit (Boa et al. Reference Boa, Mathias and Hicks2017).
-
2. Potential reduction in fabrication time of 45% and resource cost saving of 57% (Mathias et al. Reference Mathias, Snider, Hicks and Ranscombe2019b.
-
3. Use of similar and compatible materials – that is low-cost thermoplastics – providing a more seamless or integrated prototype finish.
-
4. The manufacturing capability (i.e. precision) of 3D printers is adequate to reproduce and extend upon LEGO’s patented interface.
-
5. Beyond standard 3D printing hardware, no requirement for bespoke tooling or hardware is imposed.
Table 1. Relative and generalised properties of LEGO and 3D printing for prototyping
Table 1 presents counterpoint (relative) benefits of LEGO and 3D printing as prototyping media. Depending on the specific case, the degree of fidelity and flexibility may be important elements of the prototyping process. This variance in requirement, depending on the case, is exactly what enables the later-proposed hybridisation approaches.
Of note is that higher 3D printing skill refers to the requirement for CAD modelling prior to printing, and the need for technical knowledge in print setup to ensure quality (i.e. overhang management, part orientation, machine parameters). While construction kits are not low-skill by definition, the authors argue they often require a narrower and less technical knowledge base when applied by the generalist and in the general case.
The concept of HP with LEGO and 3D printing has been considered by a number of researchers (Mueller et al. Reference Mueller, Mohr, Guenther, Frohnhofen and Baudisch2014b; Mathias et al. Reference Mathias, Snider, Hicks and Ranscombe2019b, positing benefits (Boa et al. Reference Boa, Mathias and Hicks2017; Camburn et al. Reference Camburn, Viswanathan, Linsey, Anderson, Jensen, Crawford, Otto and Wood2017), technical operation of a tool, for example, for constructing a freeform prototype from LEGO (Mueller et al. Reference Mueller, Mohr, Guenther, Frohnhofen and Baudisch2014b, or positing benefits and feasibility based on specific media characteristics (Mathias et al. Reference Mathias, Hicks, Snider and Ranscombe2018; Mathias Reference Mathias2020). These works reveal a potentially disruptive capability but do so in isolation and theoretical or ad-hoc cases, thereby showing only fundamental technical feasibility of the concept. They do not consider that in practice, hybridisation may occur through varying approaches, the constraints of real-world implementation, or that these will impact the relative benefits of HP in terms of time/cost saving.
It is important to note that HP may integrate many different materials, and that mixed-media and mixed-fidelity prototyping are used in industry, and, commonplace in software engineering (de Sá et al. Reference de Sá, Carriço, Duarte and Reis2008). However, there are few formalisations or investigations that specifically consider the use and approach for mixed-media prototyping, and the relative benefits that mixed-media itself may bring in terms of fabrication time and cost. As previously discussed, this work selects distant media with what can be considered to be opposing strengths and weaknesses, and formalises the approaches and benefits that such an approach provides. While a specific study of other combinations is required, this work illustrates the degree of potential benefit (cost and time saving) that may be realised by careful combination, and discusses the generalisable principles that this specific combination presents.
2.1. Proposed hybridisation approaches
In basic form, HP using LEGO and 3DP occurs via internal brickpacking to which an external 3DP skin is affixed. However, the manner and degree to which this approach maximises benefit will likely depend upon the intent of the prototyping activity and accompanying design process drivers. While previous work has typically not considered that the implementation of HP may vary (i.e. see (Mueller et al. Reference Mueller, Mohr, Guenther, Frohnhofen and Baudisch2014b; Mathias et al. Reference Mathias, Snider, Hicks and Ranscombe2019b), this work posits that the varying properties of the materials and their fabrication processes create opportunities to streamline HP fabrication towards activity goals, thus maximising benefit. The use of prototyping approaches to control process and maximise benefit is well-studied in literature, considering, for example, the media used (Hannah et al. Reference Hannah, Michaelraj and Summers2009; Camburn et al. Reference Camburn, Viswanathan, Linsey, Anderson, Jensen, Crawford, Otto and Wood2017, Reference Camburn, Dunlap, Gurjar, Hamon, Green, Jensen, Crawford, Otto and Wood2015; Mathias et al. Reference Mathias, Hicks, Snider and Ranscombe2018), prototyping techniques (Christie et al. Reference Christie, Jensen, Buckley, Menefee, Ziegler, Wood and Crawford2012; Hansen & Özkil Reference Hansen and Özkil2020), techniques for identifying prototyping purpose (Menold et al. Reference Menold, Jablokow and Simpson2017), prototyping guidelines (Ahmed & Demirel Reference Ahmed, Demirel, Marcus and Rosenzweig2020) and affordances of different prototyping techniques (Snider et al. Reference Snider, Goudswaard, Ranscombe, Hao, Gopsill and Hicks2023).
Based on the taxonomy of prototyping purposes and, in particular, the communication, evaluation, cost and design stage factors established by Hannah et al. (Reference Hannah, Michaelraj and Summers2009), three interrelated attributes are proposed that impact the design and fabrication of hybrid prototypes and the magnitude of potential benefit. These are: fidelity of overall form, speed of fabrication and reusability of components (parts). For example, a form prototype may not require high fidelity in all areas; a highly iterative design task might require a prototype that can be modified and evolved rapidly and a prototype for communication might need to be fabricated as quickly as possible. Aligning hybridisation with such goals imposes conditions on prototyping such that different approaches are required, which may induce varying benefits and impose other limitations. When applying hybridisation, designers must therefore be aware of the impact that it will have on prototyping and, as such, there is a need to investigate the variance in benefit resulting from differing approaches.
Here, and aligned with the above interrelated properties, three HP approaches are proposed for investigation. These approaches are not intended to be exhaustive, but rather allow investigation of HP applied in the context of differing prototyping scenarios and relative benefits due to the way in which HP is implemented.
2.1.1. Approach 1: adaptive fidelity
Fidelity as a concept may refer to many aspects of a product, including visual, functional breadth and depth, interactivity (McCurdy et al. Reference McCurdy, Connors, Pyrzak, Kanefsky and Vera2006), operational, environmental and psychological (Cox et al. Reference Cox, Hicks and Gopsill2022). Here, fidelity is intended to refer to geometric precision, which in turn leads to visual and aesthetic fidelity, and functionality, where this is directly related to geometry (i.e. motion).
Depending on the elements of focus within the prototype, its form may have different fidelity requirements. This hybridisation approach aims to partition the prototype form into regions of interest (ROI), each with targeted fidelity requirements dependent on the prototype’s purpose. These regions are decomposed, printed and assembled according to their fidelity requirements, as shown in Figure 2, using 3D printing for high-fidelity geometry and LEGO, where low fidelity is sufficient. The potential benefit of this approach is that the approximate overall form of the prototype can be constructed quickly out of LEGO with potentially only a small amount of 3D printing required to achieve the desired overall form fidelity, thereby reducing printing time.
Figure 2. Hybrid prototypes with adaptive fidelity from high (left) to low (right).
2.1.2. Approach 2: parallel fabrication
As hybridisation requires the decomposition of the prototype form into components, there is an opportunity to leverage the modularity of a LEGO–3D-printed hybrid prototype to decrease overall fabrication time by distributing the printing of parts across multiple printers. Similarly, for larger prototypes it might be possible to distribute the assembly of the fabricated parts across a team of designers. The limitation of this approach lies in a potentially increased quantity of 3D-printed material via the increased surface area of 3D-printed components, increased assembly time with increasing numbers of components, and the introduction of splits and interfaces on the prototype that may interfere with learning.
2.1.3. Approach 3: component reuse
Hybridisation via the use of LEGO and printed media creates an opportunity for increased reuse between successive prototype iterations. In the case of minor changes, the new printed parts can be assembled onto the existing underlying LEGO structure (effectively swapping out parts or collections of parts). This also means that only the printed parts that change between iterations need to be printed, significantly improving the reusability (all the unchanged parts can be reused) and potentially reducing the overall times for prototype fabrication across a project or set of similar projects.
3. Practical implementation of hybridisation
As it requires the interfacing of dissimilar media, it is important to establish how HP may be physically implemented in practice. Where previous studies have not considered practical constraints such as ensuring strong interfaces, feasible assembly and good prototyping practice, such properties may impact the implementation of approaches and hence the magnitude of benefit that may be realised.
To ensure hybridisation as studied accounts for practical implementation, this section develops knowledge and capability to create hybrid prototypes via principles of hybrid construction, good practice and an outline of HP methodology. These are then used to develop an HP support tool that, in turn, enables the investigation of relative benefit for differing approaches.
3.1. Construction principles
Construction principles comprise the practical considerations that must be met to ensure hybrid prototypes can be fabricated. These include integrity (joining of parts), the need for the entire system to be self-supporting, the requirement for manual assembly/disassembly by the designer, and the use of standard parts (LEGO) in the presented HP approach. It is noted that prototype purposes (Hannah et al. Reference Hannah, Michaelraj and Summers2009; Petrakis et al. Reference Petrakis, Hird and Wodehouse2019; Snider et al. Reference Snider, Goudswaard, Ranscombe, Hao, Gopsill and Hicks2023) will also influence these aspects, such as the level of desired user interaction and the ability to assemble/disassemble a prototype.
3.1.1. Integrity (joining)
Due to the standard interface of LEGO, constraints on hybridisation are introduced by the need for 3D-printed components to align with the LEGO interface method. In this paper, the standard LEGO library was used (shown in Figure 3 top left) alongside Ultimaker 2+ 3D printers. The Ultimaker 2+ printer has X, Y and Z resolutions of 12.5, 12.5 and 5 μm respectively (Ultimaker nd) – equating to achievable layer heights of 0.02 mm. This level of precision, combined with a redesign of the female interface (Figure 3 lower left), was found to be sufficient to 3D print LEGO-compatible parts that can adequately interface with commercial injection moulded bricks. Experimental studies (Luo et al. Reference Luo, Yue, Huang, Chung, Imai, Nishita and Chen2015; Mathias Reference Mathias2020) revealed that a separation force of around 0.7 N is typically required to separate a single LEGO stud. The redesigned interface was refined in terms of interference to achieve a separation force between 0.5 and 1 N.
Figure 3. LEGO kit and redesign interface. (Top Left) Library of standard LEGO parts. (Top Right) Possible decomposition of 3D-printed portion of a hybrid prototype. (Lower Left) Redesigned female stud interface for 3D-printed parts. (Lower Right) Assembly sequence for a hybrid prototype.
In addition to the design of the stud interface, form and structural integrity of a LEGO model are highly dependent on sufficient overlap of bricks and alternating orthogonal orientation of consecutive layers of bricks. Correspondingly, maximising overlap, alternating orthogonal orientations and satisfying the condition of a two studs or more overlap must be incorporated into a hybridisation method.
3.1.2. Assembly planes
As the LEGO interface exists only on the upper and lower faces of the bricks and for a single build plane, orthogonal assembly and overlapping parts are required. The implication of using LEGO with a single explicit build plane is that assembly constraints of ‘building out and then up or down’ must be accommodated in the method.
3.1.3. Decomposition planes
Decomposition planes concern the 3D-printed portion of the prototype, and in particular, how the 3D-printed portion(s) are decomposed over and above the minimum necessary to realise differing approaches and enable assembly. For example, the hollow shell in Figure 3, top right, could be assembled regardless of whether it is split into two parts (a top and bottom section) or a large number of parts (see decomposed shell). Considering manual assembly by a designer, the latter is more time-consuming and embodies precedence in the order of assembly, see Figure 3 (lower right). Decomposition planes are also impacted by parallelisation of fabrication, that is multiple 3D printers and the need to realise overlapping interfaces between adjacent components.
3.2. Principles of good practice
In the context of HP with LEGO and 3D printing, good practice considerations include aspects of design, manufacture and assembly that must be considered to ensure prototype performance and to maximise benefit. The complete set of Design for Fabrication and Assembly (DfFA) rules has been published by the Authors (Mathias et al. Reference Mathias, Hicks, Snider and Ranscombe2018, Reference Mathias, Snider, Hicks and Ranscombe2019b), and correspondingly, only the aspects not discussed in the preceding subsections are here summarised.
-
• Technical Constraints – These include the need for the prototype to be assembled and disassembled by hand (see Figure 3 (lower right)) and the availability of LEGO parts (no. of and type).
-
• Process checks – These include the size and number of 3D-printed parts and the location of planes along which the prototype is split – that is in some instances, planes may intersect an important feature of the prototype, and this must be checked by the designer.
-
• Design Checks – These include overall dimensional constraints, such as the limits imposed by the smallest LEGO part and the maximum volume of a 3D printer, and good practice for FDM part design, such as those described by Redwood (Redwood et al. Reference Redwood, Schöffer and Garret2017).
3.3. HP guidance
Via the investigated hybridisation approaches and the above considerations, Table 2 outlines foundational considerations for HP that may be used to guide practical implementation. This guidance has been used to develop a HP support tool to investigate the benefit of hybridisation approaches in this work. In turn, this tool enables the automated application of the proposed principles to three design cases and the associated exploration of the influence of the three approaches on the resultant hybrid prototypes.
Table 2. Foundational considerations for the creation of a hybrid prototyping method, derived from preceding sections
3.4. HP support tool
Hybridisation of LEGO and 3D-printed materials will, by its nature, require the creation of 3D geometry beyond that of the prototype alone. To enable this process in a consistent and comparable manner, an HP support tool that can be integrated into the 3D printing workflow – sitting between geometry creation in CAD and generation of printer settings and G-code in slicing software – has been created by and is available through the authors. The tool was created as an add-on for Blender version 2.79, an open-source 3D creation suite, selected because of its Python-based API enabling access to ray intersection, 3D volume calculations, Boolean operations on 3D objects and functions for handling mesh data, manipulating geometry, checking printability and generating G-code. A screenshot of the GUI is shown in Figure 4.
Figure 4. (Left) A screenshot of the custom blender user interface for the digital hybrid prototyping tool (right) A flow diagram of the user workflow when creating hybrid prototypes using the LEGO and 3D printing implementation.
The workflow of this tool embodies the guidance within Table 2. It enables the generation of a hybrid prototype via 11 steps, depicted in Figure 4 (Right), summarised in Table 3 and illustrated in Figure 5 through a worked example.
Table 3. Hybridisation tool workflow
Figure 5. Worked example of hybrid prototype generation.
Demonstrating the fundamental benefit of hybridisation and validity of the approach in line with previous research (Mathias et al. Reference Mathias, Hicks and Snider2019a, Reference Mathias, Snider, Hicks and Ranscombeb ), this tool and process were used to create the computer mouse shown in Figure 5. This example took 197 minutes to print 12 parts on two desktop FDM printers, 8 minutes to assemble and included 24 LEGO bricks. For comparison, printing the mouse as a single part on a single printer took 246 minutes. Both cases used the same printer with identical printer settings, including material, layer height, print speed, shell thickness and infill density, thus allowing relative comparison of fabrication times. The HP version thus resulted in a baseline reduction in fabrication and assembly time of 17%.
Note that as this paper focuses on exploring the benefits of different HP approaches in terms of fabrication time and cost, the time to learn and apply the tool and the user journey in so doing is excluded from the results. This tool is considered a research tool to actuate the approaches and thereby allow their comparison.
Via consideration of practical implementation of HP, this section has established principles of construction and good practice, and used these to devise hybridisation guidance and a software-based support tool. Utilising this tool, the following sections investigate the relative benefits that may be realised through the three proposed hybridisation approaches of adaptive fidelity, parallel fabrication and component reuse.
Hybridisation approaches were first investigated through a simulation study across three artefacts (Section 4), followed by validation across three iterations of a real-world product development process (Section 5).
4. Simulation study
The simulation study used the approach employed by Mathias et al. (Reference Mathias, Snider, Hicks and Ranscombe2019b) to investigate the fabrication time and material use of HP under each hybridisation approach. In the previous work by Mathias et al., hybridisation was explored as a basic concept. There was no consideration of good practice, different hybridisation approaches, or applied benefit in a real-world context. While previous work has shown that HP is fundamentally viable, this work significantly extends and contributes by exploring different HP approaches, and how benefits vary and may be maximised by consideration of good practice.
The coupling of 3D printing and LEGO is a deterministic problem with no variance in simulation results due to repetition, requiring each simulation to be run only once. Simulations were applied using the tool and method described in Section 3, aligned with each hybridisation approach developed in Section 2. Excepting those specifically mentioned (i.e. number of printers), all print settings are maintained between cases to allow comparison.
Recognising that the performance of each approach may vary across products, dependent on their geometric features, each approach was applied to three artefacts: a computer mouse, a video game controller and a digital camera. These are shown in Figure 6. As previous work has considered artefact scale relative to construction kit scale (Mathias Reference Mathias2020), those used here are chosen to be of similar scale. As specific benefits will always be dependent on geometry and designer intent, artefacts are chosen to present a range of properties that affect hybridisation. The mouse provides a simple geometry with minimal interface complexity and an overall shape that suits internal brick-packing. The controller has a significant interface and ergonomic focus, with relatively simple object geometry yet dimensions (i.e. thickness, horizontal scale) that make internal brick-packing challenging. The camera has a more complex total geometry with discrete functional modules, and a large internal volume that lends itself to internal brick-packing.
Figure 6. The three artefacts used in the simulation study.
4.1. Metrics
Known relationships exist between number of prototypes and performance (Neeley et al. Reference Neeley, Lim, Zhu and Yang2014), particularly in the context of iteration (Dow Reference Dow2011; Camburn et al. Reference Camburn, Viswanathan, Linsey, Anderson, Jensen, Crawford, Otto and Wood2017), emphasising tool-focused improvement of prototyping on fabrication time and resource use (Christie et al. Reference Christie, Jensen, Buckley, Menefee, Ziegler, Wood and Crawford2012; Hildebrand et al. Reference Hildebrand, Bickel and Alexa2012; Mueller et al. Reference Mueller, Mohr, Guenther, Frohnhofen and Baudisch2014b; Goudswaard et al. Reference Goudswaard, Hicks, Gopsill and Nassehi2017; Ranscombe et al. Reference Ranscombe, Zhang, Rodda and Mathias2019). In accord with the context of this work, which considers fabrication time and cost for prototypes, it stands that measurement of improvement to prototyping may be inferred via two metrics: (i) the reduction in prototyping time and (ii) the reduction of material through increased quantity of LEGO parts and decreased filament consumption. This is not intended to imply that these metrics or the reduction of fabrication time or cost alone provide a full picture of how prototyping may be improved.
The metrics employed are summarised in Table 4. Of note is that in the results of the simulation studies, only fabrication time is considered, as this has been shown to be dominant, with material use following a similar trend. In the real-world case (Section 5), material use is presented for completeness. While material use is not the only factor in cost, the quantity of non-recoverable material used is a major component of the cost of an individual prototyping episode (Ajay et al. Reference Ajay, Rathore, Song, Zhou and Xu2016).
Table 4. Performance metrics (relative benefits)
The calculation of fabrication rates followed the process of Mathias et al. (Reference Mathias, Snider, Hicks and Ranscombe2019b). Print time per unit volume was linearly interpolated from a measured dataset of print times for a range of different 3D-printed objects. The default print settings used in these studies were 0.15 mm layer height, 18% infill and 60 mm/s print speed, with values kept constant to allow comparison. The print rate, as time per unit volume, was found to be 8.328×10−2s/mm3.
The total artefact print time was calculated using:
$$ {T}_p=\left({V}_o-{V}_B\right){R}_p $$
Where Tp is the object print time, Vo is the volume of the object, VB is the volume of all bricks used in the prototype and Rp is the print rate.
The total fabrication time for an artefact is the sum of all print and assembly times across LEGO and 3D-printed parts. In Mathias et al. (Reference Mathias, Snider, Hicks and Ranscombe2019b), assembly rates for LEGO bricks were calculated to be 18.33 s/brick by averaging the construction times of 14 participants in a simple assembly task. Combining this value with object print time allows the calculation of total fabrication time Tf:
$$ {T}_f={T}_B+{T}_p $$
where TB is the total brick assembly time calculated via the number of bricks multiplied by the assembly rate.
4.2. Analyses of hybridisation approaches
Each hybridisation approach was applied to varying degrees for each artefact to better map variance in relative benefits. In practice, the degree to which an approach is applied is a designer’s decision. To validate the cases, all the approaches were compared with a version printed as a single part on a single printer. The cases considered to realise the three approaches and their analyses are presented one by one.
4.2.1. Adaptive fidelity
For this approach, three levels of surface geometric fidelity were considered for each artefact, namely, full fidelity (100% 3D-printed shell), medium fidelity (approximately even split of media across surface) and low fidelity (majority LEGO). Only a single printer was considered in simulations for all artefacts to ensure independent variables remained only the artefact and fidelity level. In practice, the choice of which design elements should be of which fidelity will be highly designer and use-case specific (McCurdy et al. Reference McCurdy, Connors, Pyrzak, Kanefsky and Vera2006; Liker & Pereira Reference Liker and Pereira2018). To approximate designer intent, regions of interest were here selected to either maximise fidelity of the user interface and controls, or of ergonomic features (i.e. grips).
The level of fidelity was measured and normalised as the percentage of the exterior surface area that was printed (at full fidelity). Table 5 shows fidelity percentages across artefacts for each level, while Figure 7 illustrates the composition and fidelity for each artefact across the three levels.
Table 5. Degree of fidelity by surface area
Figure 7. (Left) Full fidelity, (mid) medium fidelity and (right) low fidelity models studies for each artefact.
The fabrication time for each level of fidelity for each version of the artefact is shown in Figure 8 (Top). Results show that fidelity must be reduced to decrease overall fabrication time below that of a purely printed version. This is due to the need to offset the increased print time for hybrid parts created by the increased total surface area of the outer shell components.
Figure 8. (Top) Fabrication times for each artefact at three levels of adaptive fidelity. (Bottom) Normalised fabrication time against fidelity level.
To quantify the results, the potential difference/savings in fabrication time at differing levels of fidelity (% in Table 5) is extrapolated using linear regression, as shown in Figure 8 (bottom). It can be inferred from the regression analysis that if required fidelity is below 78.7%, the fabrication time would be expected to become less than that of time to fabricate a single part.
4.2.2. Parallel fabrication
To apply a parallel fabrication approach, each version of the artefact was decomposed into varying quantities of parts to allow distribution of printing across a set of printers ranging from 1 to 20. To ensure consistency in decomposition, higher decomposition levels were controlled by increasing the number of vertical split planes (see Figures 3 and 5). It should be noted that the location of splits would typically be guided by design intent, that is, with splits located away from important geometric details to avoid interference with critical surfaces.
A bin packing algorithm was used to evenly distribute print times across a fixed number of printers, with the overall print time being the total print time of the printer with the most significant total allocation. The overall fabrication time is then given by the following equation:
$$ {T}_f=\underset{n\in N}{\max}\sum \limits_{i\in {P}_n}{P}_{ni}+{T}_a $$
Where Tf is the fabrication time, Ta is the assembly time, N is the number of printers and Pni is the individual part print time for printer n.
Jobs were allocated to each printer using a greedy algorithm that iteratively allocates the currently longest print time to the printer with the shortest queue until all prints are allocated.
Figure 9 shows the variation in fabrication time as the number of printers increases for each artefact. Each line represents a different number of vertical cuts, where higher values create more decomposed parts. In all cases, the fabrication time ceases to decrease when the number of printers exceeds the number of parts.
Figure 9. (Top) Fabrication times for parallelisation with increasing decomposition for: Left – Computer mouse, Mid – Video game controller, Right – Digital camera. (Bottom) Reduction in fabrication time with increasing numbers of printers compared to: Left – a basic hybrid prototype (see Section 2), Right – a prototype printed as a single part.
The lower portion of Figure 9 shows the percentage decrease in fabrication time of increasing parallelisation by comparing a hybrid prototype fabricated as a simple brick-packed shell (see Section 2) and a prototype printed as a single part. For both hybrid and standard prototypes, fabrication time decreases rapidly up to around six printers, with approximately diminishing returns beyond 10.
These results show that parallel fabrication holds significant potential to reduce prototype fabrication time, with two printers providing approximately 46% reduction for a hybrid prototype vs 35% reduction for a non-hybrid prototype. While parallelisation thus creates benefits regardless of hybridisation, HP allows the relative savings to be increased.
It is noteworthy that parallelisation will create interfaces between parts that will affect the visual fidelity of the part, and that these may be undesirable depending on the use case and context. It will always be a matter of designer intent and priority whether the potential savings to be gained from parallelisation are sufficient to offset potential visual defects. Further, parallelised prototypes will require assembly. While assembly time has previously been shown to be of the order of 5% of fabrication time (see Section 3), this will vary depending on prototype geometry and complexity.
4.2.3. Component reuse
The goal of the component reuse approach is to maximise the reuse of components between prototype iterations and hence generate the hybrid prototype in such a way that only parts of the form that are modified will require refabrication. This is a particular strength of construction kits over 3DP, which will often require refabrication, whereas LEGO only requires reassembly, thus potentially holding substantial time savings. This approach thus leverages hybridisation over multiple artefact iterations.
For each artefact, two consecutive design iteration scenarios were considered: Large and general changes that affect the whole prototype, followed by local changes that affect small regions only.
For the first case, the form of each artefact was iterated from a simplified form to the complete version. For the second case, localised changes were added, with the remainder of the geometry kept constant. Figure 10 shows the geometry for each case for each artefact.
Figure 10. (Left) simplified forms, (mid) general forms and (right) localised changes for each artefact.
For LEGO bricks, the comparison between versions was measured via differences in the contained bricks. To consider the reuse of 3D-printed parts, the surface area and volume of each part were calculated with a threshold used between parts of different iterations to infer if the part was present in each. While it is feasible that parts between iterations will hold identical surface areas and volumes while not being identical, this is unlikely due to the constraints and complexity of the prototype geometry.
Figure 11 compares fabrication times across artefact iterations. Replicating an iterative process, it compares the fabrication time and cost of a subsequent iteration if completed using no HP approach (outer form printed and internal structure comprised of LEGO, see Section 2.1) and component reuse-focused hybridisation. The results show that for general changes to the form, there is little potential for saving from reuse-focused hybridisation. For local changes to the geometry, however, substantial savings may be realised. It is interesting to note that as iterations continue and the design becomes more stable, this effect will compound the cumulative time savings. Further, the savings are more pronounced for the largest of the studied objects, reflecting that local changes require a far smaller proportion of the object to be refabricated and that total savings are higher for larger artefacts.
Figure 11. Fabrication times and time saving for levels of component reuse.
4.3. Comparison of hybridisation approaches
The preceding analysis shows that each hybridisation approach may generate savings, but the scale of savings depends on the degree to which the approach is applied. Through comparison of the application of the three hybridisation approaches to the three cases, the following observations can be drawn:
-
1. For adaptive fidelity to yield benefits in terms of time savings, the proportion of prototype form reproduced at a high fidelity must be less than 78% with an approximate reduction in fabrication time corresponding to around 57–58%. The relationship arises as a function of the print volume of parts and assembly time for LEGO bricks.
-
2. For parallel fabrication of the forms of prototypes significant benefit is realised with parallelisation across six printers, beyond which savings in fabrication time reduce considerably, plateauing at around 10–12 printers. This effect arises due to the relative volume of the prototype and the offsetting of reduced print times for smaller parts with increased assembly time.
-
3. A component reuse approach is of greater benefit where iterations of prototypes manifest more local changes to the form and features. Perhaps not unsurprisingly, significant changes allow fewer parts to be reused, requiring a greater proportion of the volume and shell to be printed for each iteration.
5. Case study – See Sense
To further examine and validate the aforementioned observations, a real-world product development project was analysed. By considering design iterations, this allowed replication of prototyping decision-making embedded with designer intent. The project concerned the design of a retrofitted automatic light bulb unit – See Sense – which aimed to create affordable smart lighting that would react to the presence of a person or turn on at specified times. The designs included buttons and interfaces for the user to control how the light behaved, from adjusting motion sensitivity to setting timers. For the original product, the prototyping process used foam models for early concept generation before using 3D-printed prototypes to elicit stakeholder feedback and improve the design.
5.1. Method
In line with previous sections, the prototyping process performed for See Sense was replicated as simulation studies using each hybridisation approach, to establish whether benefits could have been realised in a real-world case. Simulation inputs used digital models of the prototype iterations produced during the See Sense project. This replication approach was employed rather than the direct application of HP by the designer in order to allow specific focus on technical fabrication time and cost, without the confounding impact of factors such as user journey and tool interface.
The series of foam and 3D-printed prototypes created in See Sense is shown in Figure 12. The prototypes are approximately 80 mm wide and 70 mm tall, representing design iterations progressing from left to right. As the 3D-printed prototypes represent more detailed designs than the early-stage foam models, the iterations that they represent were modelled for this study. A standard light bulb screw thread (E27) was added to each digital model, as it was missing from the physical prototypes. The resultant digital models are shown in the lower portion of Figure 12.
Figure 12. (Above) Foam and 3D-printed prototypes produced during See Sense, showing design evolution from left to right. (Below) digital models of 3D-printed prototype iterations.
For the four iterations shown in Figure 12, hybrid prototypes were generated using the approaches and the developed tool detailed in Section 4, and compared to printing as a single part. Parallel fabrication initially utilised four printers, determined by design intent as further decomposition would interfere with part geometry. Adaptive fidelity reduced the volume of 3D printing by only printing elements with which the user interacts – the upper portion of the prototypes shown in Figure 12. This produced a fidelity ratio of 57.7%; below the 78% required to expect benefits to arise (see Section 4.3).
The three approaches were implemented such that there was little overlap and that their key traits could be investigated independently. In practice, the application of HP would likely involve a combination of these depending on the situation and the designer’s intent.
5.2. Results
Figure 13 (Left) shows the per-iteration fabrication time for each approach across the four iterations shown in Figure 12. The most significant benefits in terms of time and material usage over single printing of artefacts arose from parallel fabrication (56% reduction in fabrication time) and adaptive fidelity (76% reduction in material and 32% reduction in fabrication time). Component reuse had the opposite effect, increasing fabrication time by 27% but reducing material usage to a similar value to that of parallel fabrication.
Figure 13. A comparison of hybridisation principles. (Left) Fabrication times for each iteration of each approach. (Mid) Cumulative fabrication time for each approach. (Right) Cumulative printed material for each approach.
Figure 13 (Right) shows the cumulative printer material usage after each iteration for each approach. In all cases, hybridisation reduced material usage when compared to single prints, with an average reduction of 66%. Adaptive fidelity performed the best, requiring less than a quarter of the printed material.
There are several points to draw out from applying the different hybridisation principles. The first is the apparent poor relative performance of the component reuse approach. The greatest potential shown by this approach was for small, localised changes between iterations. It follows that, should speed be the primary driver, this approach is only beneficial for small changes across prototype iterations of product variants, where parts can be reused within and across projects respectively. In contrast, and for the case considered, there are large geometric changes to overall size and form – with the only constant part being the screw lightbulb fitting. However, this approach did lead to a slight increase in reusability – arising from the screw lightbulb fitting and some internal LEGO remaining constant between iterations. As such, this shows that should material usage be a primary driver, a component reuse approach can help to achieve this goal while maintaining full prototype fidelity.
Adaptive fidelity and parallel fabrication were the two approaches that outperformed a single FDM print. Adaptive fidelity showed strong performance, decreasing fabrication time below that of a single print and generating the greatest material saving – requiring only a quarter of the material. Both metrics could also be improved by further reducing fidelity. However, appropriate fidelity is a matter of intent, and further decreases in this case would affect elements with which the user interacts – the degree of benefit that may be realised is then dependent on designer discretion. Parallel fabrication showed a substantial reduction in fabrication time while still using only four printers. Here, an intent-driven trade-off again exists – the designer can create large reductions in fabrication time, but must balance this against the resulting prototype comprising many parts, which may impact prototype performance.
6. Discussion and conclusions
This paper has investigated the practical implementation of HP using LEGO and 3D printing via three hybridisation approaches aligned with different prototyping process goals. Construction principles and good practice guidance were established and applied as the basis for a HP support tool, which was then used to explore the relative benefits of each approach in a simulation study and a real-world design project.
The most significant benefits in terms of time and material usage arose from parallel fabrication (56% reduction in fabrication time) and adaptive fidelity approaches (76% reduction in material and 32% reduction in fabrication time). Interestingly, the simulated studies identify balance points for each approach, where certain conditions must be met to create benefits over the simple printing of a prototype as a single part. These depend on the approach selected, with adaptive fidelity requiring a high-fidelity proportion of less than 78% to outperform printing alone and parallel fabrication requiring approximately six printers to achieve significant benefit, which starts diminishing beyond 10. A component reuse approach is beneficial and depends on the prototype form and activity priorities, only saving time when changes are small, but it saves material across cases. These findings were validated and further elucidated in a real-life design project, showing similar levels of benefit to the simulation studies. Notably, the investigated approaches are not mutually exclusive, and benefit may be maximised by analysis of their combined application.
The focus of this study has been on hybridisation for the prototyping of consumer products (user-driven), where form and appearance are necessary for product differentiation. Consequently, being able to rapidly prototype form and assess user interaction and ergonomics in designing these products is crucial. With an ongoing rise in the use of 3D printing as the technology matures, this work shows that mixed media approaches present higher benefits when compared with 3D printing alone. While applied here to NPD, HP may offer similar time and material savings in domains where form-based prototyping and scale models of large products are created, including infrastructure, architecture, aerospace and marine. However, as the intent of prototyping in these domains may differ, approaches must be reconsidered to ensure they remain in alignment with overall process goals.
As such, while the HP approaches described here demonstrate realisation of a range of benefits, results are limited to only the combination of 3DP and construction kits. Further work should consider how benefits vary across different media combinations. Similarly, with the exact desired form of a prototype depending on designer intent and individual case, further work should also consider the variance in benefit that is realised across different scales, prototype forms, complexities and representations. In simulation studies, the process and results do not consider the additional decision-making present in HP. By allowing designers to, for example, choose between fidelities or introduce visual defects via split lines between parts, additional time and cost may be incurred during prototyping.
As design processes progress from conceptual design to detail and production, the prototyping requirements become more specific and targeted towards manufacturability and design optimisation. The prototyping techniques studied here cannot meet these requirements a priori. This is because production prototypes generally become more sophisticated and use materials, assemblies and techniques that prove the design can be produced – that is proof-of-production or integration prototypes. As such, the implementation of HP and the findings reported are more applicable to the early stages of the design process, and their utility cannot be assumed to be equivalent or applicable to later stages of the design process. Furthermore, it is in the early stages of design where the reduction of fabrication time and material costs is most desirable to enable quicker iteration, supporting a rapid ideation process and the larger design changes. Such benefits may not be considered substantive or desired during later stages.
Finally, the work here considers HP using two dissimilar media, partly demonstrating the feasibility of HP despite differences, and partly leveraging the distinct benefits each medium provides. Via the support tool produced and the three approaches, this work shows the viability of this approach, but it cannot be assumed to be general across media. Thus, further study should consider the implementation of HP approaches using other media, the benefit of such approaches in prototyping scenarios, and the alignment of different media and approaches with designer intent.
Acknowledgements
The work reported in this paper was conducted at the University of Bristol in the Design and Manufacturing Futures Laboratory (www.dmf-lab.co.uk), and was supported by EPSRC grants EP/W024152/1, EP/R032696/1 and EP/W020564/1.
Open access
