Experiment 21: DTCFAM – Digital Twin For Continuous Fiber Additive Manufacturing

Motivation of the Experiment

The manufacturing/business problem being addressed is the time-consuming and labor-intensive manual defect-checking process in the production of lightweight composite structures using the automated fiber placement (AFP) process. The motivation behind the problem is the need to optimize production efficiency and reduce costs associated with human labor. While the AFP process automates the layup process, the inspection process is still done manually, requiring skilled operators to pause the process and mark down any defects on printed schematics, which takes about 10-15 minutes after each layer. With a typical part having about 20-30 layers, this inspection process becomes a significant bottleneck to production efficiency. While additional inspection systems can be purchased to perform the task, the proprietary sensors used in these systems are expensive, costing between €50,000 to €100,000, adding complexity with slow results, and have highly restricted inter-operability with third-party software, requiring additional license costs. Therefore, the motivation behind the problem is to develop an efficient and cost-effective solution that leverages HPC-based digital twin technology to automate the defect-checking process and optimize production efficiency.- Third party software compute defects after each layer is completed, taking 5-10 minutes to analyzethe data i.e. equivalent to manual labor checking time. All of the above mentioned issues have prevented adoption of this inspection process. Less than 1% presently installed AFP system use any inspection software.

Purpose of the Experiment

The goal of the experiment is to develop a digital twin using Addpath off-line programming software to detect defects in real-time from the streamed sensor data and present the defects to the operator for corrective action at the end of each layer layup. The use of a cloud/HPC platform is essential to leverage machine learning models that can be improved over time, thus enhancing the accuracy of defect detection and prediction. By automating the defect detection process, we can optimize production efficiency, reduce the need for skilled operators, and eliminate the need for expensive proprietary sensors. The AddPath software will be used to achieve the automated defect detection from the streamed sensor data, while machine learning models will be used for the defect prediction.

Technical Impact

The successful experiment is expected to have significant technological impact on both the manufacturing and IT aspects. From a manufacturing standpoint, the impact includes minimizing manual quality control and reporting by up to 40%, saving material by minimizing wastage (as 80% of faults occur in the first layer processing), increasing process throughput by 30-50%, identifying defects in real-time, and freeing up operator time to work on other value-adding tasks. The end part quality is consistently improved by capturing and analyzing data, thus enhancing overall production efficiency.

From a technical perspective, the impact includes the ability to provide a subscription-based solution to end-users, creating a new revenue source. The platform can be easily extended to multiple new use cases, such as robotic printing of metal, plastic, ceramic, etc. The developed software platform enables interoperability with FEA, allowing for a faster design cycle and integration into the user ecosystem. A reliable process will make the process more effective, expanding the usage into new markets, and a new business model with pay-per-use for defect detection is probable. Overall, the successful implementation of the experiment would lead to significant technological advancements and benefits to both the manufacturing and IT sectors.

Expected Economic Impact

The successful implementation of the experiment is expected to have significant economic impacts on both the end-user, Compoxi, and the independent software vendor, Addcomposites. For Compoxi, the impact includes the potential for more competitive production means, resulting in reduced product lead time, increased throughput, and the potential for gaining a larger market share. The increased production volume in Europe's growing satellite constellation systems could also lead to the stabilization of three full-time employees within the company. The technology's application to a wider range of products such as fuel tanks, antenna booms, and primary aerostructures could result in exponential growth with efficient processes and technologies. These economic impacts could lead to increased revenue and lower costs for Compoxi, as well as a reduction in waste and potential green impacts.
For Addcomposites, the impact includes the potential for capturing a significant market share through accessible overall solutions, user-friendly features, and the constant value added through digital twin for additive robotics processes. The estimated annual recurring revenue of $23 billion for the digital-twin software market for in-situ detection of defects in a production process could result in the capture of 10-15% of the market share, which is $2-$4 billion. The potential for automated quality control, SAAS business, and the software platform's growth could lead to significant economic impacts, including increased revenue and job creation. The DIGITbrain platform could also enhance and fast track the technology's impact. Overall, the successful implementation of the experiment could lead to significant economic impacts, including increased revenue, job creation, and potentially positive environmental impacts.

Project Partners: