LokAlp: A Reconfigurable Massive Wood Construction System Based on Off-Cuts from the CLT and GLT Industry

Abstract

This paper presents LokAlp, a modular timber construction system invented and developed by the authors, inspired by the traditional Blockbau technique, and designed for circularity and self-construction. LokAlp utilizes standardized interlocking blocks fabricated from CLT and GLT off-cuts to optimize material reuse and minimize waste. The study explores the application of massive timber digital materials within an open modular system framework, offering an alternative to the prevailing focus on lightweight structural systems, which predominantly rely on primary engineered wood materials rather than reclaimed by-products. The research evaluates geometric adaptability, production feasibility, and on-site assembly efficiency within a computational design and digital fabrication workflow. The definition of the LokAlp system has gone through several iterations. A full-scale demonstrator constructed using the LokAlp final iteration (Mk. XII) incorporated topological enhancements, increasing connection variety and modular coherence. Comparative analyses of subtractive manufacturing via 6-axis robotic milling versus traditional CNC machining revealed a >45% reduction in cycle times with robotic methods, indicating significant potential for sustainable industrial fabrication; however, validation under operational conditions is still required. Augmented reality-assisted assembly improved accuracy and reduced cognitive load compared to traditional 2D documentation, enhancing construction speed. Overall, LokAlp demonstrates a viable circular and sustainable construction approach combining digital fabrication and modular design, warranting further research to integrate robotic workflows and structural optimization.

1. Introduction

Self-construction, traditionally linked to informal building practices, is being reinterpreted through emerging modular construction systems that invite user participation via explicit assembly rules. Enabled by computational design and digital fabrication, this approach allows individuals to interact with discrete, reconfigurable, and sustainable components.

Recent architectural research has introduced the concept of open modular systems, where components are no longer function-specific but behave as digital materials [1]. Unlike traditional modularity—based on static elements assigned to predefined roles—digital materials are uniform units without intrinsic function. Through a limited set of reversible connections, these units aggregate topologically and support the addition or removal of parts without disrupting systemic coherence, consistent with the principles of open modularity [2]. This model departs from classical modular design, where architectural function is embedded in the component. In systems built from digital materials, functionality emerges from each unit’s position and relationships within the aggregate [3,4]. This enables high adaptability using a minimal set of standardized elements. Architectural customization is achieved not by diversifying parts, but by combining identical elements into varied topological configurations.

This paradigm aligns with Design for Disassembly and Reuse (DfDR) principles, promoting a circular and adaptive view of the built environment. It allows materials to shift from static end-products to reconfigurable resources, extending their life cycle, facilitating reuse, and reducing the environmental costs of demolition and new production [5].

1.1. State of the Art and Limitations of Current Open Modular Systems

In the past 15 years, open modular wood-based construction systems have become a major topic of speculation and research, particularly concerning the discretization of architectural space through new modular logics enabled by computational design. Architectural practice has shown a strong interest in this direction, especially in East Asia, due to its longstanding tradition of timber-based construction techniques.

Most open modular wood-based construction systems have been realized as temporary ephemeral structures, such as pavilions. However, some permanent buildings also exist [6,7]. Generally, temporary pavilions have aimed to discretize architecturally complex forms [8,9,10]. Only rarely have such constructions employed components characterized by substantial material mass [11,12], and in most cases, they have remained within sub-story height scales—with notable exceptions [13].

The above-mentioned open modular wood-based construction systems, developed and built at full 1:1 scale, have generally exhibited recurring design configurations in both typological and structural terms. In particular, these systems tend to adopt configurations definable as lightweight structural systems, i.e., linear skeletal systems in which loads are transmitted exclusively through linear elements [14]. These structures are characterized by high spatial permeability and low material density, where voids prevail over mass, although some exceptions exist where the system is inherently massive [12] or where massive elements are integrated into a generally lightweight structural system [11].

At the academic level, research on such discrete construction systems has been extensively examined by Retsin [15], who highlighted that the development of these innovative construction strategies—often driven by computational design and digital fabrication—has primarily been demonstrated through complexly shaped pavilions, which rarely reach architectural scale. Nevertheless, some exceptions addressing applications at architectural scale have also been tested and demonstrated [16].

The primary objective of these construction systems, however, concerned mainly the demonstration and validation of parallel and complementary technologies rather than the direct application of these systems in conventional building contexts.

Different discrete systems have demonstrated, in fact, the use of augmented reality for assisted assembly [17], the employment of virtual environments for simulating construction processes [15], or the integration of modular robots for automated assembly [18], focusing future outlooks on the merits of these technologies rather than on the projection of construction systems into permanent building contexts. Since the aim of this line of investigation is to enable knowledge transfer from experimental prototypes to the construction industry, it becomes essential to assess how such systems can be meaningfully deployed within real-world architectural frameworks. In this perspective, future research should explore how these systems might evolve to meaningfully articulate the spatial configurations typical of everyday living environments consistent with architectural practice rather than focusing mainly on formal or technological experimentation.

1.2. Reuse of CLT and GLT By-Products in Open Modular Systems

Open modular wood-based construction systems developed to date predominantly employ first-processing engineered wood [6,7,8,9,11,13]. In contrast, the reuse of production residues from mass timber manufacturing—such as those generated during CLT and GLT fabrication—offers promising opportunities for the development of engineered wood-based structural building components. Recent research on CLT production residues has demonstrated that these materials can achieve mechanical performance comparable to standard CLT while potentially contributing to lower environmental impacts and production costs. This approach provides a concrete pathway to enhance the sustainability of timber-based construction within a circular economy framework [19].

This application domain, still scarcely investigated in the scientific literature on open modular construction systems, constitutes a strategic nexus for transitioning toward more sustainable and resilient construction models.

In the industrial production process of CLT, substantial waste is generated, mainly from the re-cutting required to create openings for doors and windows. In the European context, such by-products can represent up to 20% of the total panel volume processed [20], while in the United States, CNC waste and end cuts represent approximately 15% of the oven-dry mass of 1 m³ of CLT ([21], Table 3), calculated as 72.06 kg out of 472.44 kg. Currently, these materials are predominantly destined for energy recovery processes or incineration, since the presence of industrial adhesives prevents landfill disposal and hinders biodegradation [22]. These valorization methods clearly represent a downcycling process, as energy recovery yields products with much lower added value than the original material [9]: a material with high mechanical performance and significant technological complexity is downgraded to a low-value energy source, resulting in loss of the embedded value within the production chain and the generation of climate-altering CO₂ emissions.

For GLT beams, technical studies have highlighted the systematic presence of off-cuts generated during distribution [23]. GLT beams are commonly produced in standard lengths between 14 and 17 m and are subsequently cut at distribution centers to meet the dimensional requirements of residential building applications, which often require shorter lengths. This process frequently leaves residual pieces with typical cross-sections of 130 × 160 mm and lengths under 4 m, which are not destined for specific structural use. However, these off-cuts still retain the mechanical properties of the original product.

Although some virtuous case studies confirm the technical feasibility of employing by-products derived from CLT production within building systems [24], such approaches do not strictly adhere to the discrete logic—i.e., the use of standardized, universal components enabling systematic assembly and reuse—that is central to this study. Rather, they rely on principles of mass customization of individual components, thus forsaking one of the fundamental premises of open modularity: the replicable and combinatorial standardization of construction units. Within research on discrete modular systems, the use of industrial residues has been primarily limited to lightweight, non-structural three-layer panels employed in lightweight discrete structural systems configurations [25], without fully exploring the potential offered by reusing massive structural elements such as CLT or GLT.

2. Methodology

2.1. Research Objectives

As analyzed in the previous Section 1.1, nearly all timber digital materials developed to date have been conceived according to structural logics attributable to lightweight structural systems. Such systems have generally been tested exclusively at the pavilion scale, significantly limiting the exploration of their use in ordinary architectural-scale building contexts, as observed by Oosterhuis [26].

This study is guided by three principal objectives, which serve as the foundation of the methodological workflow followed throughout the research and conceptually illustrated in Figure 1:

(i)

Paradigm shift: Development of a timber construction system within discrete architecture that critically departs from the prevailing approach by proposing a massive paradigm inspired by traditional European Blockbau techniques [27]. This European massive timber construction system can be partially interpreted considering the principles of the digital material concept, as its components are characterized by an intrinsically informative geometry that enables their deconstruction and subsequent reuse without loss of integrity. A paradigmatic example of this logic is the 1966 deconstruction and reconstruction of a Blockbau building originally constructed in the 15th century [28]. This intervention was executed without damaging the wooden elements, which were disassembled, transported, and reassembled at a new site located at a higher altitude, demonstrating the system’s full reversibility and reusability.

(ii)

Architectural application: Unlike multiple studies focusing on discretizing complex geometries [8,9,10,12], the system proposed here aims to reproduce ordinary multi-story living spaces and to construct building components such as load-bearing walls, columns, beams, and floors. By reinterpreting the Blockbau construction technique through computational design and digital fabrication tools, the developed system must guarantee the possibility to generate continuous load-bearing walls free from material discontinuities while preserving the discrete logic of assembly.

(iii)

Waste valorization: demonstrating the feasibility of a more efficient and sustainable use of off-cuts generated by the CLT and GLT industry, integrating them within the context of discrete architecture. Considering that digital materials operate as architectural voxels for space discretization and thus, by definition, possess dimensions smaller than conventional building elements, it is possible to reuse cutting waste produced during the processing of CLT panels (particularly off-cuts generated by door and window openings) and GLT beam off-cuts for the fabrication of such construction components. This approach aims to preserve the added value of engineered wood in terms of structural performance, avoiding the typical downcycling associated with energy recovery.

Figure 1. Conceptual workflow adopted in this study. From left to right: (1) Initial inputs and research objectives; (2) Design system: development and iteration of 15 prototypes based on computational design; (3) Design demonstrator: architectural-scale design through generative design methods; (4) CNC production of demonstrator components via cutting center; (5) Robotic simulation of a parallel process for the same components; (6) Output: on-site self-assembly of the demonstrator supported by AR glasses.

Figure 1. Conceptual workflow adopted in this study. From left to right: (1) Initial inputs and research objectives; (2) Design system: development and iteration of 15 prototypes based on computational design; (3) Design demonstrator: architectural-scale design through generative design methods; (4) CNC production of demonstrator components via cutting center; (5) Robotic simulation of a parallel process for the same components; (6) Output: on-site self-assembly of the demonstrator supported by AR glasses.

This research carries both theoretical and practical implications. Theoretically, it challenges the prevailing dominance of lightweight structural logics in timber digital materials by proposing a massive timber paradigm, thereby opening new avenues for discrete architecture research focused on constructive continuity and non-permeable solutions at the architectural scale.

Practically, the study advances the feasibility of integrating waste valorization strategies within digital fabrication workflows, promoting sustainability through the reuse of industrial off-cuts. By targeting the construction of ordinary buildings rather than purely experimental pavilion-scale structures, the proposed system aims to bridge the gap between computational design innovation and real-world architectural applications, offering a potentially scalable solution aligned with industry demands.

2.2. The LokAlp System

LokAlp is a design and construction system based on standardized modular wooden blocks assembled using a dry-assembly method without adhesives or mortar to enable easy disassembly and reuse. This series of components enables the creation of various building elements—such as beams, columns, walls, and floors—through aggregative logics that ensure compositional versatility and geometric coherence. The morphological and functional definition of the discrete “LokAlp” block was guided by a set of integrated design parameters, which accounted for both the compositional potential of the system and the constraints deriving from the use of wood off-cuts such as CLT panels and GLT beams.

The block design was conceived following a spatial discretization logic based on a cubic matrix, assuming that system modularity could emerge from the repetition and combination of regular units. Within this logic, the block’s short side was defined based on the maximum thickness recoverable from available wood off-cuts, generally ranging between 100 and 160 mm, in accordance with the average thickness values of load-bearing CLT walls used in buildings from three to eight stories [29]. The overall block length was determined as a multiple of the short side, thus maintaining coherence with the three-dimensional cubic grid and ensuring the spatial composability of the system, as shown in Figure 2.

During the design phase, the main challenges involved defining formal and technical solutions to guarantee the realization of 90-degree corner nodes without protruding elements, configuring three-way joints for beam–column integration, and enabling openings within walls for door and window insertion. Another key design issue concerned optimizing the number and type of possible connections for each block. Excessive junction variety would have certainly increased compositional freedom but would have also significantly raised fabrication waste, reducing the efficiency of reusing engineered wood by-products.

The maximum block weight was set at 10 kg, in compliance with the ISO 11228-1:2003 standard [30] regulating manual load handling according to frequency, duration, distance, and operational conditions. According to these standards, a 10 kg load can be manually handled up to a maximum frequency of approximately seven lifts per minute, only if the activity duration is intermittent and ranges between one and two hours. Considering a reasonable transport distance of four meters—a representative value for movement between the component storage area and the assembly site—the maximum load that can be manually handled over an eight-hour workday is estimated at around 10,000 kg, with a handling frequency not exceeding two cycles per minute.

The entire design process led to the definition and prototyping of 15 constructive variants of the system, named “Mark” and numbered progressively (Figure 3). Their main morphological and functional characteristics are summarized in

Table 1. Comparative overview of the 15 LokAlp prototypes. The table highlights the structural, geometric, and functional evolution of the system across iterations, based on the longest structural unit (Figure 3). Dim. 1 and Dim. 2 indicate, respectively, the block’s cross-sectional size and total length. Conf. refers to the total number of configurations per block. Cubic Matrix indicates compliance with a cubic spatial grid. The labels 3-axis knot, Openings, and Angle refer to the ability to form three-way joints, integrate openings (windows/doors), and create orthogonal wall connections without protrusions.

Model	Dim. 1 [mm]	Dim. 2 [mm]	Conf.	Cubic Matrix	3-Axis Knot	Openings	Angle	Weight [kg]	Waste [%]
Mk. I	200	600	15	Yes	No	No	Yes	8.1	35.7
Mk. II	200	600	23	Yes	No	No	Yes	6.8	18.6
Mk. III	200	600	37	Yes	No	No	Yes	6.1	16.8
Mk. III/A	15	75	22	Yes	No	No	Yes	6.3	20.5
Mk. III/B	10	40	10	Yes	No	No	Yes	1.5	20.8
Mk. IV	15	75	38	No	No	Yes	Yes	5.8	27.8
Mk. V	15	75	40	No	Yes	Yes	Yes	5.5	28.2
Mk. VI	15	90	44	Yes	No	Yes	No	6.8	25.7
Mk. VII	16	96	44	Yes	Yes	Yes	No	8.6	22.6
Mk. VIII/A	16	96	44	Yes	Yes	Yes	No	7.7	30.1
Mk. VIII/B	16	96	44	Yes	Yes	Yes	No	7.8	29.5
Mk. IX	16	96	44	Yes	Yes	Yes	No	8.4	24.2
Mk. X	16	96	40	No	No	Yes	No	8.4	23.9
Mk. XI	16	96	44	No	No	Yes	Yes	8.7	21.8
Mk. XII	16	96	44	Yes	Yes	Yes	Yes	7.6	30.1

. The 15 developed 3D prototypes, along with detailed descriptions of their assembly potential, are openly accessible via the repository indicated in the Data Availability Statement to allow verification, reproduction, and reuse for scientific purposes [31].

2.3. Assembly Logic of the LokAlp Mk. XII System

The LokAlp Mk. XII system is based on two modular types of building blocks, designated A and B (Figure 4). Both blocks share a square base cross-section but differ solely in length. Block A has a length equal to six times the side of the square, while block B measures three times the same side, exactly half the length of block A. The geometric proportions between blocks A and B must remain constant regardless of the initial side length chosen as the reference unit.

Block assembly occurs through specific connectors called biscuits (components C–I) (Figure 4), which constitute the sole means of creating connections between modular elements. Blocks A and B can be interconnected in any combination using the biscuits (A–A, B–B, A–B, B–A), but biscuits cannot connect to each other. Biscuits C–G perform the primary junction function between blocks, whereas biscuits H and I act solely as passive locking elements for biscuits C–G and do not independently enable block-to-block connections.

Each block A can accommodate up to 22 simultaneous connections with other blocks A or B (Figure 5), while block B can accommodate up to 12. The grooves present on all six faces of blocks A and B, characterized by variable but internally uniform inclinations and depths, directly influence the geometry of the biscuits but do not affect the number or type of possible connections. Therefore, parametric variation of the grooves does not alter module compatibility nor the combinatorial flexibility of the system.

2.4. Computational Design

Starting from the LokAlp Mk V block version, a parametric design system based on the Grasshopper3D framework [32] was introduced with the goal of optimizing the management of geometric variables during iterative development phases. The integration of computational design tools played a strategic role, particularly in enhancing the efficiency of the transition from modeling to production. The ability to modify the generative parameters of the block in real time made the operational workflow highly responsive, following a file-to-factory logic that directly connects the digital 3D model with physical fabrication processes.

In addition to supporting 3D printing of 1:10 scale models during prototype validation stages, the parametric system facilitated coordination with the partner carpentry company, helping to manage technical uncertainty related to tool availability. Specifically, the type of dovetail cutter mounted on the machining center to be used later was not definitively determined until the final phases of the project, necessitating a sufficiently flexible modeling approach able to adapt to different tool geometries. The dynamic nature of the computational model enabled direct modifications to the milling cutter profile within the parametric environment without interruptions to the production flow, including the possibility of radical tool geometry substitutions during machining phases.

The parametric code was developed to operate without external input data and to guarantee an accessible user interface even for non-expert users. Its modular structure allows rapid generation of construction variants tailored to specific production constraints, configuring a highly resilient and replicable design environment. The parametric code and the detailed user manual are openly accessible via the repository indicated in the Data Availability Statement to allow verification, reproduction, and reuse for scientific purposes [31].

2.5. Design of the Architectural-Scale Demonstrator

To test the mechanical validity of the LokAlp system and verify its effectiveness during assembly, an architectural-scale demonstrator was designed and constructed, and subsequently subjected to physical assembly tests. The configuration included elevated portions and cantilevered parts, aiming to explore different structural conditions. These specific conditions were selected because they represent critical load scenarios: elevated elements test vertical load transfer and overall stability, while cantilevered parts impose bending stresses and highlight the mechanical performance of connections under challenging moments. The design process was carried out using the WASP plug-in for Grasshopper3D [33], specifically oriented toward modeling discrete aggregations.

WASP represents a flexible solution for generating open modular aggregations within computational design. The framework allows explicit definition of both the geometry and topology of discrete units, establishing connection rules customizable by the designer. These rules are then used to automatically generate discrete aggregations based on predefined geometric constraints and assembly rules; the system also integrates tools to avoid collisions between elements.

In the case of the LokAlp Mk XII block “A” (Figure 4), the algorithm was programmed to recognize 44 theoretically possible combinations starting from a single generative element. However, once geometric and topological restrictions were applied, only 22 combinations were admissible for simultaneous aggregation.

For generating the final architectural demonstrator, spatial constraints defined through the Constraints module of the WASP plug-in were employed. These constraints can be formulated as geometric entities of different types—lines, planes, surfaces, and volumes—and act as spatial archetypes guiding the aggregation process. The computational system interprets these constraints as containment and orientation conditions, attempting to fill the available space as closely as possible through iterative distribution of the discrete elements constituting the digital material. The aggregation result obtained through the application of constraints is shown in Figure 6.

2.6. Production Using Hundegger CNC Work Center

The production phase of the LokAlp block was carried out in collaboration with a company specialized in timber carpentry for construction and experienced in processing GLT and CLT elements.

The blocks were manufactured using a Hundegger K2i machining center [34], a CNC machine designed for automated processing of architectural-scale wooden elements. This cutting center enables complex carpentry operations thanks to its technical equipment, which includes circular and horizontal saws (up to 13 kW) and a universal 5-axis milling machine with an industrial 35 kW spindle. The machine is designed for continuous production across three shifts, including handling large elements (up to 12 m in length), making it suitable for large-scale prefabrication.

The LokAlp blocks were produced from spruce GLT beams with a square cross-section of 160 mm and a density of 450 kg/m³.

The entire production process represented a significant test bench to evaluate the limits and potentialities of employing an industrial cutting center in a non-serial, non-standardized context. The production facility and part of the dovetail milling operations are shown in Figure 7.

2.7. Robotic Simulation

In parallel with production using an industrial cutting center, a feasibility study was conducted on the use of industrial robotics as an alternative to conventional machinery (Hundegger machining centers).

The fabrication process was based on the use of a six-axis ABB IRB 4600 robotic arm, featuring a maximum payload of 20 kg and a reach of 2.5 m. Motion programming was developed within the Grasshopper3D environment, enabling automatic generation of trajectories required both for pick-and-place operations and for sequential milling of all six faces of the block.

The robotic cell was designed, including a centering plane, the ABB robotic arm, and a cutting plane, as shown in Figure 8. The centering plane is a three-axis inclined surface that allows the block to slide towards a fixed geometric reference, ensuring precision in the robot’s grasp. Knowing the exact position of the block within the robot’s coordinate system allows highly accurate operations without the need for further calibrations.

Once the robot grips the block, it transports it to the cutting plane following an optimized trajectory scripted in Grasshopper3D. After milling one face, the block is repositioned on the inclined plane and rotated by 90 degrees to allow machining of the next face. This cycle repeats until all six surfaces are completely processed, without operational interruptions or human intervention.

2.8. Assembly in Self-Build Context via Augmented Reality

The assembly of the discrete modular system was carried out through the implementation of an augmented reality (AR)-based workflow, using HoloLens 2 Industrial Edition smart glasses [35]. The integration was achieved via the “Fologram” [36] plug-in for Grasshopper3D, which enables real-time synchronization between the parametric modeling environment and the AR interface.

Through this system, all 367 assembly steps were encoded into an interactive digital interface: each new block or connector to be placed was highlighted using selective white illumination, while already positioned elements were displayed in a transparent mode.

3. Results

The final tectonic configuration developed for the production and assembly phases consisted of an architectural composition partially inspired by traditional blockbau timber systems, recalling the structural articulation of pillars and floor frames with primary and secondary beams. While this reference maintained realistic proportions for beams and columns, it was not pursued as a literal interpretation but rather as a formal suggestion integrated into the discrete construction language. The resulting aggregation comprised a total of 120 units: 110 LokAlp Mk XII blocks of type “A” and 10 blocks of type “B” (Figure 9). Assembly required a total of 237 connectors, including 206 hexagonal biscuits of type “E”, 24 of type “F”, and 7 of type “I”.

The production of the 110 “A” and “B” blocks was carried out using the Hundegger K2i CNC machining center. The machine’s operational cost was estimated at approximately EUR 200/hour, according to data provided by the industrial partner company located in Northern Italy. The full production cycle of a single type “A” block required an average of 40 min, excluding about 20 min of occasional spindle cooling downtime occurring every several blocks due to thermal load. While not systematic, this highlights a potential area for process optimization. Under these conditions, the unit cost for manufacturing a single LokAlp type “A” block was quantified at approximately EUR 130. In parallel, a fabrication process using a six-axis robotic arm was simulated, demonstrating that at the same cutting speed as the Hundegger, a single LokAlp “A” block could be produced in 22 min. However, this result is based solely on simulation and requires validation under real-world conditions to confirm actual performance, including thermal behavior, downtime, and precision.

The entire assembly process was carried out at the partner company’s production facility, simulating a realistic self-build context with a single operator and the use of a mobile scaffold for assembling elevated and cantilevered portions (Figure 10). The prefabricated components, produced via the Hundegger center, were placed on pallets located about 4 m from the construction area to ensure realistic working conditions and handling times. To facilitate spatial orientation during execution, three A4 sheets displaying unique QR codes—generated using the Fologram plug-in—were placed at strategic points of the construction site to enable accurate positioning of the structure.

During the first day of assembly, 245 operations were completed, corresponding to the placement of 90 LokAlp blocks and 155 connectors, within a standard 8 h work shift. On the second day, the remaining 122 operations (40 blocks and 82 biscuits) were executed within another 8 h shift. The reduced number of components installed on the second day (approximately –50%) is attributable to the different operational setup: the first day focused on assembling the central core of the aggregate, located at ground level, while the second day involved the cantilevered portion, fully elevated at approximately 2 m and assembled exclusively using the scaffold. On the second day, working exclusively at height resulted in reduced operator accessibility and mobility for block positioning, increased transport distances due to the need to place blocks atop the scaffolding prior to assembly, and generally more complex conditions for securing the blocks.

The entire process was conducted by a single operator without encountering significant operational issues, as demonstrated in the video available in the Supplementary Materials. The use of the scaffold ensured accessibility to elevated areas while maintaining continuous workflow. From a technological standpoint, the augmented reality system displayed through smart glasses (Figure 11) showed no critical malfunctions: with a single 10,000 mAh power bank and a partial recharge during the lunch break (approximately one hour), the AR environment remained active for the full 8 h working day without interruptions.

The results substantiate the study’s core objectives by demonstrating the feasibility of a timber construction system grounded in discrete architecture, which enables modularity and reversibility inspired by traditional Blockbau principles. The integration of computational design and AR tools ensures precise assembly and adaptability, supporting the architectural aim of constructing continuous, load-bearing elements within a discrete modular logic. Furthermore, the system’s compatibility with off-cut reuse from wood industries aligns with the circularity objective, promoting waste valorization and sustainable material cycles by preserving the structural value of engineered wood without resorting to downcycling. These findings collectively advance the paradigm shift toward a more sustainable and adaptable timber construction method.

4. Discussion

4.1. Comparative Analysis of LokAlp Systems

The LokAlp Mk. XII prototype was selected as the basis for architectural aggregation experimentation, as it represents the most advanced configuration among the 15 open modular systems developed. Compared to earlier iterations, the Mk. XII enables up to 44 potential connections per individual block while maintaining full compliance with key construction requirements: realization of orthogonal wall joints without protrusions, implementation of three-way joints for beam–column interconnections, and vertical arrangement of “A” and “B” blocks to define wall openings (doors and windows) (Figure 12).

Some of these features were already present in the Mk. V prototype. However, the latter exhibited a major limitation concerning its compatibility with a spatial discretization system based on whole cubic modules: the type “A” block, composed of five cubic units, required the “B” component to measure 2.5 units in length—precisely half. This fractional dimension compromised the system’s consistency with the founding principle of integer-based modularity, generating ambiguity in alignment and the repeatability of modules.

Similar issues were observed in prototypes Mk. X and Mk. XI, where the layout of the upper and lower milling patterns was offset by a quarter module relative to the reference cubic grid. This resulted in systemic misalignment of joints and prevented full interoperability between blocks.

However, the Mk. XII introduced a new geometric–structural limitation: the intrinsic asymmetry of “A” and “B” blocks. This design choice was driven by two main factors: the need to reduce machining waste—which already accounted for approximately 30% of the raw material volume—and the intention to preserve the mechanical integrity of the block ends, preventing further weakening due to symmetrical milling. Specifically, the wooden section between the two hourglass-shaped recesses proved to be the system’s most vulnerable point, fracturing on two occasions during assembly when struck with a rubber mallet to insert the “biscuit” connectors.

This asymmetry also introduced a margin of error during assembly, with occasional misorientations due to operator inattention.

Regarding material waste, which remains notably high in certain cases—such as the Mk XII model, where waste reaches 30%—different mitigation strategies could be implemented. From a production standpoint, approximately 8% of the total waste could be reduced by using a dovetail milling cutter with shallower and differentiated cuts, as demonstrated by the Mk XI model, which exhibits a reduced waste of 21.8%. However, the mechanical performance of the resulting wooden joint in this case would require experimental validation.

An alternative solution involves the selective elimination of certain milling operations. Although this approach would limit the potential for future reuse of the timber block in alternative configurations, it is noteworthy that milling on the upper and lower faces is rarely employed, typically reserved for openings such as doors and windows or three-way joints. Given their infrequent use, it could be reasonable to mill these faces only when strictly necessary.

According to this rationale, the majority of timber components would be milled solely with hourglass-shaped cuts on the four sides, reducing milling waste to approximately 14%, compared to the initial 30%. This strategy would also significantly shorten the production cycle time while simultaneously reducing material waste, albeit at the cost of a slightly diminished potential for reuse and reconfiguration.

4.2. Production Limitations of LokAlp Blocks on Hundegger Machining Center

In the production of timber components for the architectural-scale demonstrator, direct interaction with the manufacturing department proved crucial for optimizing the technical feedback cycle. Leveraging computational design tools enabled real-time adaptation of the LokAlp block’s generative parameters, allowing for customized modeling that maximized the capabilities of the available machinery.

A first limitation encountered in the production workflow using the Hundegger cutting center concerned the constrained nature of the machine’s operations. The proprietary software only permitted a predefined set of parametric operations, commonly referred to as “macros”. Although these macros could be adjusted in terms of depth, width, and angle, they did not support the implementation of fully customizable toolpaths. Consequently, it was necessary to sequence multiple distinct macros to approximate the desired geometries. Among the most relevant operations used was the dovetail milling macro.

Furthermore, the system imposed a maximum of 130 macros per machining session, per unit. This constraint required the segmentation of production into successive phases, necessitating manual repositioning of the workpiece and resulting in increased cycle times. The intermediate scale of the LokAlp blocks—positioned between architectural elements and joinery-sized components—revealed additional limitations: the inadequacy of the integrated handling system forced operators to perform manual rotations of the pieces (up to 90°), and dimensional inaccuracies were observed in several operations, with precision deviations ranging from 2 to 5 mm.

Machine overheating, which required waiting periods of up to 20 min before resuming operation, was primarily attributed to the repeated use of the dovetail cutter. The machine was not optimized for such high-frequency sequences, exacerbating thermal stress and compromising workflow continuity.

4.3. Comparison Between Robotic and Hundegger-Based Production of LokAlp Components

The evaluation of a robotic cell as a production strategy for LokAlp components was prompted by the operational limitations observed in Hundegger cutting centers, which proved suboptimal for non-standardized, small-scale processes such as those required for the LokAlp block—particularly in terms of handling and cutting precision.

From a spatial standpoint, robotic cells require substantially less footprint than a Hundegger cutting center. However, the main advantage lies in their operational versatility. Cutting centers like Hundegger rely on a closed system of predefined operations, permitting only limited customization. These operations, implemented through standard “macros”, include typical joinery features such as dovetail cuts or edge milling, which are critical for conventional timber construction. While certain parameters—such as depth, width, and angle—can be modified, the creation of new macros is restricted. It requires direct intervention from the machine manufacturer and often entails extended waiting times (up to two to three months), with no guarantee of approval due to legal and safety liabilities.

Industrial robots, by contrast, are engineered for maximum reconfigurability and can be reprogrammed to perform a wide range of complex tasks in both manipulation and material processing. Six-axis robotic arms can execute diverse operations—including milling, drilling, pick-and-place, and cutting—with continuous operation and high positional accuracy. This adaptability effectively overcomes the dimensional precision limitations inherent to Hundegger systems, as also demonstrated in recent studies [37].

Regarding initial investment costs, the official price of the Hundegger K2i is not publicly available, mainly due to variability depending on the chosen setup and configuration. The same applies to the ABB IRB4600 robotic arm, whose pricing is generally provided only upon request. However, based on verified industry experience and quotations obtained from robotics companies, the price range for a 6-axis robot with an end effector can be estimated between EUR 50,000 and EUR 100,000. In contrast, a typical medium-sized CNC machining center usually exceeds EUR 400,000. This implies a roughly fivefold difference in cost, with the important distinction that an industrial robot is a quasi-general-purpose machine designed for multi-role applications, rather than being dedicated exclusively to a specific task such as CNC cutting. Additionally, the robotic arm requires system integration to achieve full operational capability, which is currently feasible through computational design workflows, as previously discussed.

The comparative analysis between the robotic and conventional systems highlighted a substantial performance improvement. The average production time per block decreased from approximately 40 min (using Hundegger) to 22 min with the robotic arm while maintaining identical cutting speeds. This reduction of over 45% in cycle time is particularly significant when contextualized within the hourly operating cost of the Hundegger system, estimated at around EUR 200. Nonetheless, it is important to highlight that such machines were not designed or intended for the serialized production of small-scale elements, such as the building blocks of the proposed system. The data underline the superior cycle-time efficiency of robotic production in non-serial manufacturing scenarios; however, the actual hourly operating cost of using the robotic system in place of the Hundegger remains to be empirically validated in order to enable a more precise comparison. Nonetheless, it is important to note that overheating issues similar to those encountered with the Hundegger system may also arise in robotic cells. Given the significant delays caused by thermal downtime, future developments should prioritize the optimization of thermal management strategies to ensure consistent productivity.

4.4. Benefits of Augmented Reality-Assisted Assembly

During the construction of the architectural-scale demonstrator, the assembly of 130 LokAlp blocks and 237 custom connection elements followed no repetitive pattern or recurring modular logic. Consequently, it was not feasible to define general rules or standardized sequences for the assembly process. A total of 367 discrete assembly steps were required, each involving non-repetitive operations. Conventional construction documentation methods—such as plans, sections, and elevations—proved entirely inadequate for communicating the complexity and order of operations in a two-dimensional format.

In response to these representational limitations, a digital instruction manual was developed, comprising 367 axonometric diagrams, conceptually inspired by LEGO^® instruction booklets. However, the implementation of this manual-based strategy revealed several operational inefficiencies. Each assembly step necessitated repeated consultation of the manual, which slowed down the process and increased cognitive load—particularly in the later stages of construction. Furthermore, due to the asymmetrical geometry of the blocks, frequent orientation errors occurred—such as unintentional rotations or inversions—which were typically detected only during subsequent assembly steps. These errors often required partial disassembly and reassembly of previously installed components, negatively affecting overall efficiency.

To address the inherent limitations of traditional assembly methods, an augmented reality (AR)-based workflow was developed and implemented using Hololens 2 Industry smart glasses. This approach enabled a significant reduction in assembly errors by projecting the full-scale digital 3D model directly onto the physical construction site, with millimetric accuracy achieved through the use of floor-anchored QR codes as spatial reference markers.

Beyond substantially improving on-site assembly efficiency, the AR system also resulted in a marked reduction of cognitive load on operators, minimizing the need for sustained concentration during standard eight-hour work sessions. The presence of persistent, spatially coherent holograms eliminated reliance on printed or digital manuals, allowing for immediate and intuitive understanding of the required assembly actions.

5. Conclusions and Outlook

The LokAlp massive wood construction system has demonstrated, through the development of the Mk. XII prototype and the construction of a full-scale architectural demonstrator, a high degree of effectiveness in terms of compositional adaptability, executional accuracy, and compatibility with computational design processes. These results suggest that the use of massive timber components—integrating CLT and GLT off-cuts within the framework of open modular systems—could represent a viable alternative to the current state of the art, which predominantly focuses on lightweight structural systems based on primary engineered wood products.

Improvements introduced in the Mk. XII prototype—including the increased number of possible inter-block connections and alignment with a cubic modular logic—have revealed significant system evolution, while some localized weaknesses remain, such as the mechanical fragility of specific nodes and assembly errors related to the inherent asymmetry of the blocks. The mechanical weakness suggests that the structural performance of the joint could be improved by reducing the width of the hourglass-shaped interlock, thereby increasing the residual wood volume and enhancing the block’s resistance to localized stress. Alternatively, eliminating the end-face milling in cases where it is not functionally required could further reinforce the structural integrity of the block. Regarding the assembly errors, although only sporadic, they highlight the need to improve visual legibility or introduce clear orientation markings on the blocks to reduce ambiguity and minimize execution errors in future iterations.

Regarding production enhancement through a more flexible system based on robotics within the framework of digital fabrication, the comparison between traditional fabrication using a Hundegger cutting center and a robotic cell workflow has highlighted the reconfigurability and time optimization potential of the latter. Specifically, a reduction of over 45% in cycle times was observed. Nevertheless, it is important to note that the robotic simulations, while promising, have not yet been validated under actual manufacturing conditions. Potential issues such as tool overheating and cumulative tolerance deviations require targeted experimental verification. Concurrently, the augmented reality-assisted assembly workflow demonstrated substantial effectiveness in enhancing assembly precision and reducing operator cognitive load, thus overcoming the communicative limitations of conventional two-dimensional representation systems.

Future research developments are structured across several key areas. First, robotic fabrication processes must be validated in real-world operational contexts to identify critical constraints not captured during simulation, particularly concerning thermal dissipation during milling operations and process repeatability at production scale. Second, the integration of the LokAlp system with advanced structural analysis software is considered essential to optimize the design workflow and enhance digital interoperability. Further work is required to assess specific structural aspects, such as load transfer efficiency across discrete joints in beams and columns, the system’s performance under non-uniform loading conditions, and its behavior under dynamic and long-term loading. Regarding real-world adoption, scalability remains a critical challenge that requires thorough and systematic evaluation to ensure the proposed approach can be effectively implemented at larger building scales and across diverse contexts, including varying building codes and climate zones. Regulatory constraints must also be carefully addressed, as compliance with building codes and certification standards is essential for market integration. The prospects for adoption appear promising, as the base materials—namely, CLT and GLT—are already regulated by Eurocodes, and importantly, the related off-cuts maintain their mechanical, thermal, and fire resistance properties intact, unlike other wood by-products of the wood industry.

Building upon these potential perspectives, it is also strategic from an economic and production perspective to assess the feasibility of establishing a supply chain based on the reuse of CLT and GLT off-cuts, thereby contributing to a truly circular and sustainable manufacturing cycle. Future research should focus on conducting a Life Cycle Assessment (LCA) to quantitatively evaluate the potential environmental benefits, particularly in terms of avoided emissions, compared to the current practice of incinerating such off-cuts. This analysis is crucial to provide objective metrics that can support decision-making toward more sustainable wood waste management and validate the ecological advantages of the proposed reuse strategies.

In light of these results, LokAlp emerges as an experimental construction system with high potential, whose architectural scalability and applicability will depend on the continued technical refinement of its production, structural, and distributional processes.

6. Intellectual Property

The open modular construction system described in this study is protected under a Registered Community Design (RCD) filed with the European Union Intellectual Property Office (EUIPO), registration number 015040194 (001–009), filed on 7 November 2023 and valid until 7 November 2028. The rights holder is Politecnico di Milano, while the inventors correspond to the authors of this study. The patent filing and related costs were covered by Politecnico di Milano.