Search

Arm® technology is becoming increasingly important in HPC. Recently, Fugaku, an Arm®-based system, was awarded the number one place in the Top500 list. Raspberry Pis provide an inexpensive platform to become familiar with this architecture. However, Pis can also be useful on their own. Here we describe our efforts to configure and benchmark the use of a Raspberry Pi cluster with the HPX/Phylanx platform (normally intended for use with HPC applications) and document the lessons we learned. First, we highlight the required changes in the configuration of the Pi to gain performance. Second, we explore how limited memory bandwidth limits the use of all cores in our shared memory benchmarks. Third, we evaluate whether low network bandwidth affects distributed performance. Fourth, we discuss the power consumption and the resulting trade-off in cost of operation and performance.

Experimental data availability is a cornerstone for reproducibility in experimental fracture mechanics, which is crucial to the scientific method. This short communication focuses on the accessibility and long term availability of raw experimental data. The corresponding authors of the eleven most cited papers, related to experimental fracture mechanics, for every year from 2000 up to 2016, were kindly asked about the availability of the raw experimental data associated with each publication. For the 187 e-mails sent: 22.46 % resulted in outdated contact information, 57.75 % of the authors did received our request and did not reply, and 19.79 replied to our request. The availability of data is generally low with only 11 available data sets ( 5.9 %). The authors identified two main issues for the lacking availability of raw experimental data. First, the ability to retrieve data is strongly attached to the possibility to contact the corresponding author. This study suggests that institutional e-mail addresses are insufficient means for obtaining experimental data sets. Second, lack of experimental data is also due that submission and publication does not require to make the raw experimental data available. The following solutions are proposed: (1) Requirement of unique identifiers, like ORCID or ResearcherID, to detach the author(s) from their institutional e-mail address, (2) Provide DOIs, like Zenodo or Dataverse, to make raw experimental data citable, and (3) grant providing organizations should ensure that experimental data by public funded projects is available to the public.

With this research, an implementation of an overlay and abstracting RESTful API (application programming interface) for 3D printers is proposed to expose these resources to the Internet for utilization within and for cloud services. This is to abstract the underlying communication structure and means for accessing and controlling a 3D printer resource in one of three ways. The first way is a proprietary protocol or a 3D printer driver in Microsoft Windows. The second way is the control via a USB-serial connection between a controlling computer and the printer resource. This protocol can either be in a proprietary format or based on open standards like G-Code (ISO 6983-1:2009). The third way of control is based on physical storage devices attached to the printer with machining instructions stored on them. This research excludes the communication and control means involving proprietary protocols or drivers due complexity restrictions within the implementation. The approach is designed with extensibility in mind so that future access to proprietary protocols can be added to the control API. 3D printer resources with only the third control method available are also excluded from this research due to their lack of remote controllability. This work describes the design and implementation of an abstraction API layer between varying software and hardware components with an extensible architecture for future hardware and software components for within the domain of additive manufacturing (AM). With this research, the connection to further cloud services as 3D printing resources as well as a cloud printing service for usage and control of this API is demonstrated. This enables the use of AM machinery within cloud or business process-oriented architectures as the AM machinery and the associated software are exposed in an abstract and unified way and usable as services.

This research evaluates the geometric fidelity of 3D printed objects by utilizing digital image analysis techniques on consumer grade 3D printers. ISO 527-2 Type B specimen are modeled using Houdini Software, then positioned on the virtual print-bed and sliced in 0.3 mm thickness layers utilizing MiracleGrue slicer. The specimens are printed on a Makerbot Replicator2X 3D printer with an X-Y resolution of 0.011 mm. The test set is divided into three where the first object set (Test set A) is printed with their longest side parallel to the X axis, the second object set (Test set B) with their longest side oriented at an angle to the X axis and the third object set (Test set C) as a control with a different material type and structure. The test includes objects with parallel patterns of 0, 5, 10, 30, 45, 60 and 90 degree resulting in toolpaths along the parallel lines. For the second test set the orientation angle is chosen to match the parallel pattern degree. The objects are analyzed with a softwar...

The intelligent exploitation of sensor data and order parameters from modern production systems is one of the biggest challenges in the context of Industry4.0. Currently, data from single machines are processed individually and not integrated with upstream or downstream processes. This case is prevalent in automotive assembly lines. Here, numerous machining tools from different vendors prohibit a smooth collaboration. However, only the aggregation of the entirety of available data sources permits a comprehensive and intelligent analysis and optimization of production lines. This approach leads to the regulation and behavior prediction of single components and finally of whole production systems. Such an intelligent assessment can be realized by Smart Services which are self-contained application containers allowing for efficient data analytics in modern production lines. The SePiA.Pro project develops and investigates a self-describing and secure packaging format for Smart Services facilitating their automatic provisioning. The project implements an open, standard- and cloud-based platform consisting of a modelling environment for Smart Services; a repository for the exchange of Smart Services; and a provisioning engine for automated deployment of Smart Services. Said platform opens up modern data analytics capabilities for anyone, as both customers and suppliers of analytics services. Use cases from automotive manufacturing demonstrate the value of the developed solution.

With slicing tools it is possible to convert digital 3D models into printing instructions for 3D printers. The general approach is: The model is cut into horizontal slices which are then used to create extrusion paths similar to milling paths in the traditional CNC field, which are then being filled with material, mostly plastic material.The goal of this study is to compare available slicing tools for 3D printers under defined aspects using different configurations. The main contributions of this study are: (1) collecting methods and tools to judge print results, (2) analysis of the available slicing tools with appropriate tests and comparisons, and (3) evaluating the slicing tools using the analysis as solid foundation.

The fourth industrial revolution is driven by the integration and analysis of a vast amount of diverse data. Thereby, data about production steps, overall manufacturing processes, and also supporting processes is gathered to enable holistic analysis approaches. These approaches promise to provide new insights and knowledge by revealing cost saving possibilities and also automated adjustments of production processes. However, such scenarios typically require analytics services and data integration stacks since algorithms have to be developed, executed and therefore be wired with the data to be processed. This leads to complex setups of overall analytics environments that have to be installed, configured and managed according to the needs of different analysis scenarios and setups. The manual execution of such installations is time-consuming and error-prone. Therefore, we demonstrate how the different components of such combined integration and analytics scenarios can be modelled in order to be reused in different settings, while enabling the fully automated provisioning of overall analytics stacks and services.

This dataset is the description of a curated list of companies involved in additive manufacturing in Germany. The companies included are of various categories, such as 3D printing providers, hardware manufacturers, software developers and vendors. The list was compiled through literature and Internet-based research, resulting in the compilation of information from a number of resources, such as the Bundesanzeiger (Federal Gazette), the Registergerichte (Register Courts), the respective websites themselves and a B2B marketplace (Wer liefert Was?). The aim of compiling this list is to provide information to researchers on the current situation of 3D printing in Germany.

In current industrial settings, data is dispersed on numerous devices, systems and locations without integration and sharing capabilities. With this work, we present a framework for the integration of various data sources within an industrial setting, based on a mediating data hub. Within the data hub, data sources and sinks for this industrial application are equipped with data usage policies to restrict and enable usage and consumption of data for shared analytics. We identify such policies, their requirements and rationale. This work addresses an industrial setting, with manufacturing data being the primary use-case. Requirements for these policies are identified from existing use-cases and expert domain knowledge. The requirements are identified as reasonable via examples and exemplary implementation.

In this paper we demonstrate and compare two complementary approaches to the automatic generation of production rules from a set of given graphs representing sample designs. The first approach generates a complete rule set from scratch by means of frequent subgraph discovery. Whereas the second approach is intended to learn additional rules that fit an existing, yet incomplete, rule set using genetic programming. Both approaches have been developed and tested in the context of an application for automated conceptual engineering design, more specifically functional decomposition. They can be considered feasible, complementary approaches to the automatic inference of graph rewriting rules for conceptual design applications.

The aim of this paper is to introduce emerging technologies for the implementation of graph-rewriting-based design automation applications that can be used in collaborative environments. The paper motivates the use of graph-rewriting for design automation and highlights an important issue: preservation of confidentiality. Approaches to the efficient derivation (using design knowledge) and rule induction (learning design knowledge) are discussed. The crucial feature of these approaches for confidentiality-preserving graph-rewriting is that knowledge-bases of contributing organizations do not have to be disclosed.

3D printers became more popular in the last decade, partly because of the expiration of key patents and the supply of affordable machines. The origin is located in rapid prototyping. With Additive Manufacturing (AM) it is possible to create physical objects from 3D model data by layer wise addition of material. Besides professional use for prototyping and low volume manufacturing they are becoming widespread amongst end users starting with the so called Maker Movement. The most prevalent type of consumer grade 3D printers is Fused Deposition Modelling (FDM, also Fused Filament Fabrication FFF). This work focuses on FDM machinery because of their widespread occurrence and large number of open problems like precision and failure. These 3D printers can fail to print objects at a statistical rate depending on the manufacturer and model of the printer. Failures can occur due to misalignment of the print-bed, the print-head, slippage of the motors, warping of the printed material, lack of adhesion or other reasons. The goal of this research is to provide an environment in which these failures can be detected automatically. Direct supervision is inhibited by the recommended placement of FDM printers in separate rooms away from the user due to ventilation issues. The inability to oversee the printing process leads to late or omitted detection of failures. Rejects effect material waste and wasted time thus lowering the utilization of printing resources. Our approach consists of a camera based error detection mechanism that provides a web based interface for remote supervision and early failure detection. Early failure detection can lead to reduced time spent on broken prints, less material wasted and in some cases salvaged objects.

We develop a small and lightweight cloud based service for the utilization of 3D printer resources enabling users to collaborate on models. Users can collaborate by sharing model files, discussions on aspects of the printing process or using 3D printers as shared resources. This service consists of user, artefact and printer management building on existing web technology. It enables scheduling of printing jobs for artefacts and high utilization of 3D printer resources. This cloud based manufacturing (CBM) system enables 3D printers that are non-native networked to be used remotely by providing easily installable low cost networked computers or installable services. It focuses on the interface between the physical resources and their representation in software to form a cyber physical system (CPS). This service requires smart 3D printers and representation of technical capabilities of physical resources. This work is a research platform for smart machinery or the enhancement of machinery for smart control under the paradigm of Industry 4.0. We discuss the design and concept of this work in progress service and the distinctions from similar systems. Furthermore, the sharing requirements and capabilities of such a service are discussed with a focus on the data integrity and safety for sharing data among users.

This paper aims to present an overviewof the Thingiverse digital model repository and provides an analysis of a subset of randomly selected models of this site. To carry out this study, literature research was conducted to identify relevant works. Based on existing research, a subset of available models is acquired and analysed utilising existing open source software. Focusing on the geometrical details of the acquired models, an analysis is performed. Furthermore, statistical data such as download, view counters, and utilised licences of these models, is collected. For the literature review on this platform, 25 existing works were identified. From the 10,000 randomly selected ThingIDs, 3528 were existent and were acquired. Most metrics calculated are dominated by less than 20% of all the analysed models. Online repositories are a source for user-generated digital models. This paper suggests that the licence of choice is a creative commons – attribution – share alike licence.

In this work, the influence on and application of machine learning (ML) to the domain of Additive Manufacturing or synonymously 3D printing is reviewed. Existing literature is identified by a literature search and grouped according to its application in 3D printing. We provide insight into this research and the potential of ML, deep learning, and other related computational learning methods on additive manufacturing (AM) and its potential future development and embedding, such as cloud manufacturing or Industry 4.0. The application of ML is discussed to aid solving numerous problems from Additive Manufacturing, such as process control, process monitoring, and quality enhancement of manufactured objects. Furthermore, literature covering the generalities of the intersection of Additive Manufacturing and Machine Learning, reviews and future research questions are identified and presented herein. This work provides an overview of the benefits and drawbacks of combining Additive Manufacturing with Machine Learning.

The authors present a method to reverse engineer 3D printer specific machine instructions (GCode) to a point cloud representation and then a STL (Stereolithography) file format. GCode is a machine code that is used for 3D printing among other applications, such as CNC routers. Such code files contain instructions for the 3D printer to move and control its actuator, in case of Fused Deposition Modeling (FDM), the printhead that extrudes semi-molten plastics. The reverse engineering method presented here is based on the digital simulation of the extrusion process of FDM type 3D printing. The reconstructed models and pointclouds do not accommodate for hollow structures, such as holes or cavities. The implementation is performed in Python and relies on open source software and libraries, such as Matplotlib and OpenCV. The reconstruction is performed on the model's extrusion boundary and considers mechanical imprecision. The complete reconstruction mechanism is available as a RESTful (Representational State Transfer) Web service.

Cloud Manufacturing (CM) is the concept of using manufacturing resources in a service-oriented way over the Internet. Recent developments in Additive Manufacturing (AM) are making it possible to utilise resources ad-hoc as replacements for traditional manufacturing resources in case of spontaneous problems in the established manufacturing processes. In order to be of use in these scenarios, the AM resources must adhere to a strict principle of transparency and service composition in adherence to the Cloud Computing (CC) paradigm. With this review, we provide an overview of CM, AM and relevant domains as well as presenting the historical development of scientific research in these fields, from 2002 to 2016. Part of this work is also a meta-review on the domain to further detail its development and structure.

With the increased usage and application of 3D-printing or Additive Manufacturing (AM) the question arises of how content providers or creators can ensure their intellectual property on such model data. Similar to other digital media such object information that is represented in a number of file formats is easy to copy and reproduce lossless. This research contributes by a proposition of a watermarking schema for Fused Deposition Modeling (FDM) type 3D-printers. This system embeds information into the 3D-printed object without alterations to the structure or geometry by altering the entry points of each layer in a specific manner. With such a watermarking schema employed objects can be embedded with additional information such as a serial number or other traceable information.

In this paper the design and implementation of a sensor array suitable for 3D printers is presented. The sensor array includes sensors for motion/vibration, temperature, orientation and hygrometry. The sensor array is designed as an easily deployable, wireless sensor client-server system. Aggregated sensor data and print related data enable research on influencing ambient factors and quality control of the printing process. In future revisions this sensor system is intended as part of a closed-loop control system for 3D printers. The wireless connectivity enables the system to be incorporated in other machinery on moveable parts. In our research setting this is a CNC milling machine.