A review of geometric dimensioning and tolerancing (GD&T) of additive manufacturing and powder bed fusion lattices

In manufacturing and product design, geometric dimensioning and tolerancing (GD&T) plays an important role to describe the product and to facilitate communication between stakeholders involved in the process from conceptual development to manufacture. GD&T is a set of standardized symbols and rules used to communicate the design process and product description [1]. GD&T describes the nominal geometry of the product and the allowable variation of the geometric features. Realizing the functional nature of the geometrical measurements, Stanley Parker was credited being one of the first to develop the foundations of GD&T in 1938, at the beginning of World War II by developing the concept of “True Position” while working in a munition facility referring to the tolerances for the first time [2]. The first standard document related to GD&T was published by the British Standards Institution (BSI) as BS 308:1943, Engineering drawing office practice [3], followed by the US Army 30-1-7:1946, Mil-Std-8:1949 and American Society of Mechanical Engineers (ASME) Y14.5-1957. The latest version of ASME standards for general GD&T is Y14.5-2018 which includes stakeholder groups such as the American Standards Association (ASA), United States of America Standards Institute (USASI) and American National Standards Institute (ANSI) [2]. The International Organization for Standardization (ISO) also developed a set of standards including ISO 1101:2017, ISO 14405-1:2016, ISO/TS 17863:2013, and ISO 16792:2021 for GD&T and design specifications, covering conventional product design to Computer-Aided-Design (CAD) systems for additive manufacturing (AM) [4]. ISO 17296:2014 and ISO 52902:2019 refer to the test methods of geometrical characteristics and test artifacts of the parts made by AM. ASME published and updated Y14.41-2019 and Y14.46-2017 standards for digital product description data sets and drawing requirement in digital format for AM [5]. A list of existing standard documents for GD&T is presented in Table 1.

The ability to achieve predictable and repeatable shapes via AM is critical. To optimize the design of an additive manufactured product, tolerancing is a key issue [8] for defining, communicating, and assessing the dimensional and geometric accuracy of parts [9]. Although GD&T is a mature field in conventional manufacturing industries, digital and smart manufacturing processes have created a profound need for standardisation [10]. For this study, the underlined issues in GD&T benchmarking for PBF techniques have been presented from a critical review of current literature of general GD&T for AM, and the GD&T benchmarking approach of using artifacts.

The layer-by-layer bottom-up approach of AM causes rough surfaces known as stair-stepping that may potentially lead to geometrical inaccuracy. These geometrical errors also depend on factors such as part shrinkage, material properties, process parameters, support structures and surface approximation errors due to slicing techniques. Another important factor is the difference between the CAD and the toolpath model that creates a geometrical mismatch of the printed parts. The reason is that the printed part is not solely replicated from the original CAD model due to data adjustments made during the slicing stage to create a toolpath model that is dependent on various slicing parameters. Rupal et al. [11] addressed this issue and proposed a novel reverse CAD model algorithm that can convert the sliced file back to a CAD model. The reverse CAD model approach was able to assess the geometric and mechanical behaviours of the printed part while incorporating the effect of the slicing parameters. Some research works [12, 13] addressed the variation in printed part, especially for metal AM, that influences the part quality. Moges et al. [14] identified the sources of uncertainties of laser powder bed fusion (L-PBF) process chain that includes modelling uncertainty, parameter uncertainty, numerical uncertainty, and measurement uncertainty. They developed a methodology of quantifying uncertainties by case studies of semi-analytical and FEM-based L-PBF melt pool models. Another quantification approach using Isotherm Migration Method (IMM) model [15] by choosing melt pool width as the output quantity of interest was proposed by Lopez et al. [16], considering four sources of uncertainties corresponding to modelling, simulation, and measurement processes of L-PBF.

The isotropic layered structures, differential shrinkage of the parts, stair-stepping, and effect of support structure result into poor surface finish that cause non-conformance of the geometry. Many researchers and users of AM technologies realised that the GD&T of AM finished part does not fully comply with the existing ISO standardised GD&T process flow. AM yet not a standalone manufacturing option to produce higher precision surface finish of the printed part which still depends on the post-processing subtractive manufacturing technology such as CNC machining [17]. Rupal [18] mentioned that the two mostly adopted approaches for the geometric conformance or tolerance quantification for AM, (1) experimental methods based on geometric benchmark test artifacts (GBTA) and (2) predictive methods such as numerical analysis, are not in compliance with ISO 1101 standard. His work points that the limitations in GBTA design, guidelines in terms of geometric conformance, linkage of features to GD&T and their characterisation, parametric optimisation GD&T are yet to be addressed to fully realised the GD&T for AM assembled parts or moulds. Specifically in L-PBF, GD&T based datasets are crucial to consider the effect of removal of base plate. A Horizon 2020 project Computer Aided Technologies for Additive Manufacturing (2018) developed a metrology workflow including point cloud-based analysis that adds machining allowance to AM parts to achieve quality surface finish accomplished by post CNC machining [19].

In terms of GD&T for material extrusion (ME), Huang et al. [20] proposed a new method of identifying and predicting geometrical variation of ME parts using a “Skin Shape Model” to improve the geometrical quality of printed parts in the design phase, and validated by experiments with a Coordinate Measuring Machine (CMM) of a cylindrical test piece. Other researchers proposed steps to evaluate the geometrical accuracy based on conventional GD&T characteristics and test methods [21, 22], using Taguchi’s design of experiment (DOE) statistical approach to establish the relationship of the GD&T characteristics with 13 important process parameters of ME process which helped optimize the ideal parameters to improve the geometrical accuracy. The study extended a previous attempt [23] that proposing a design framework to build a test coupon with standard tolerance characteristics following ISO 1101:2005. They reported that four process parameters including component size, extruder temperature, print orientation and layer thickness are linked to the dimensional accuracy and geometrical tolerance. In terms of Vat Photopolymerisation (VP), a study was carried out to understand the effect of the slicing parameters on dimensional variation of a VP-printed part that could manipulate the horizontal dimensions of the unit printed-volume, known as a voxel [24]. The process resulted in printing with voxel dimensions below the size of the micromirrors in the VP process, thereby improving the GD&T of the printed parts. Powder bed fusion (PBF) has been widely used in tooling and biomedical sectors and the process uses a diverse range of materials especially metals, alloys, high strength polymers and ceramic. Based on the types of processing materials, fusion mechanism and heating source, PBF comprises a range of similar technologies including selective laser sintering (SLS), selective laser melting (SLM), electron beam melting (EBM), and high speed sintering (HSS). However, one of the biggest constraints towards industrial adoption of the PBF family is limited knowledge of the GD&T processes, as well as the lack of understanding of the functional relationships between materials, process parameters, support structure, part size and geometry. In the GD&T research for metal AM, laser-based powder bed fusion (L-PBF) process, especially SLM, paid more attention possibly because, it is easy and comparatively affordable to improvise and well-adopted by the industries. Zongo et al. [25] conducted a comparative study of GD&T calculations carried out both experimentally by CMM and using micro-computed tomography (µ-CT) and theoretically using ANSYS Additive Print software for a L-PBF process. The geometrical distortion of a topologically optimized part printed with an EOSINT M280 printer and AlSi10Mg alloy powder was measured with different support structures. The results found that the AP software predicted similar trend of distortion gained experimentally and the CMM technique showed higher accuracy in data acquisition than CT scans. Importantly, higher density of support structures influenced global geometrical distortion but unaffected the local deviation of highly strained zones. Other studies of L-PBF focused on developing a design framework of suitable Geometric Benchmark Test Artifacts (GBTA) for evaluating the geometric behaviours and features. Rupal et al. proposed a methodology based on features, that is a classification system based on geometric reasoning for designing the printed part [26]. The study was further extended for complex metal parts including assemblies, known as an Assembly Benchmark Test Artefact (ABTA) that include mating features to identify the assembly capacity and dependencies of the geometric tolerancing quantifiers [27]. The framework was proposed to estimate geometric tolerances based on the Skin Shape Model considering material shrinkage and validated through a case study [28], shown in Fig. 1.

To measure AM capabilities and qualify machines for maintaining geometrical accuracy and tolerancing, a 3D CAD model comprising of different shapes and features, called a Geometric Benchmark Test Artifact (GBTA) is used. Several studies have focused on designing test artifacts for evaluating the geometrical performance of AM processes [21], proposing a framework, design criteria [29], manufacturing methodologies [30] and guidelines towards developing standardised test artifacts [31]. The first GBTA for evaluating accuracy of AM process performance was proposed by Kruth in 1991 [32]. Rebaioli et al. [21] reviewed almost 60 types of such artifacts utilised in geometric benchmarking process of AM techniques. The artifacts created mainly for PBF is listed in Table 2 with corresponding GD&T characteristics for specific processes and materials. Aspects of linear accuracy, cylindricity and position were the most common features to be evaluated as compared to the level of assemblability and lattice structures that is not shown in a GBTA.

GD&T is a mature field for the manufacturing sector. However, for companies utilising AM, the geometric quality and structural integrity of the built part not only depends on the machine set-up, process parameters, built environment, but also the quality of the CAD model and the slicing techniques to convert the digital model into a toolpath planning and printing. In addition, the built position, orientation, support structure and materials influence the geometric quality of the built part [25]. Previous studies focused on optimising the mechanical properties of the built part and corresponding process parameters [45‐47]. Other studies [11, 21, 28, 48] only relate to geometric part qualification using common geometric features such as rectangles, squares, cylinders, holes, cubes, walls and slopes. They benchmarked GD&T characteristics of various artifacts produced from different AM technologies. The common GD&T characteristics such as flatness, cylindricity, parallelism, circularity were investigated, including the spatial repeatability of the features, surface finish and the minimum feature size [30]. Most GD&T work has focused on PBF techniques because of the excellent capability of the process to build various shapes, forms and complex features, including thin internal channels, infills, lattices and consolidated assembled parts [49]. However, there is a significant gap in the GD&T for complex features, especially for internal channels, lattice structures and consolidated parts. Handful studies developed tolerancing methods for lattice structures except for Ameta et al. [43, 44, 50], who proposed using a transition region known as TSS to represent the allowable variation in materials and geometry of a part with lattice structures and verifying the data using experimental metrology by CMM, XCT and FVM. However, their work is not fully applicable to different lattice structures shown in Fig. 3.

In summary, a literature review has been conducted on GD&T benchmarking for AM, and there is an extensive research gap requiring the need for GD&T benchmarking of PBF parts with complex features such as lattice structures. There is an urgent need for a standardised approach for GD&T artifacts encompassing lattice features and other freeform structures. Future work should also propose a GD&T workflow for designing artifacts with lattices structures.