Requirements for Certification of Composites

Currently, composite structures are made and accepted according to a mix of requirements, some relate to the process history and some relate to measurable final outcomes. When I started my Ph.D., I was not familiar with the accurate and detailed certification process of composite parts in the aerospace sector, but I found that to be one of the main bottlenecks for innovation. While traditional material requirements are generally based on final outcomes, the question arises that why do not use the same procedure for composites and would this decision lead to more efficient manufacturing and/or structural performance? Here, I am going to answer these questions and explain how we can improve our current workflow to foster innovation and growth in composites manufacturing. Therefore, first, I distinguish process requirements and structural requirements. Then, I explain the characteristics of composite precursors and manufacturing processes, which demands process specifications. Next, I explain the uncertainty and variability in composite materials and how it connects with material equivalency conditions. Finally, I present tools that can help us to have a more efficient and reliable manufacturing and structural performance, although process control is always necessary for path-dependent materials.

For composite parts, the range of response to factory system attributes defines the possible outcomes, which can be divided into three categories: process parameter outcomes, material structure outcomes, and material performance outcomes ¹. The necessity to define requirements as an ad hoc mix of process and final outcomes is deeply rooted in the current practice for composite manufacturing. The “know-how” approach to composite processing dictates the necessary tests and their procedures to be based on a mix of science, pragmatism, and experience accumulated over time. The current process is designed to reduce risk, but it is not optimized to minimize costs and time to completion.

The process history parameters infer the state of the material during the manufacturing process, e.g. the temperature history as seen in the material. AC 23-20 ² states that if a fabrication process requires close control to produce “consistently sound structures”, the process must be done under an approved process specification. The final outcomes either specify the arrangement of components within the material, e.g. void ratio, in different scales or the final geometry and in-service behaviour of the part, e.g. part thickness or modulus of elasticity. Material and process specifications used to produce composite materials must contain adequate information to ensure that critical parameters in the manufacturing process are identified to facilitate production and adherence to standards in the final part. This is very important because the integration of design, materials, and processing method creates a reproducible product.

The material properties of fibre-reinforced composites are manufactured into the structure as part of the production process (process-intensive material) ². Resin is a path-dependent material and changing the cure cycle will directly affect the final part properties, such as the modulus of elasticity and the gelation temperature. While part of the cure path-dependency has been understood with experiment-based mathematical models, we have not yet established a full understanding of all of the effects of the process on the final geometry and behaviour of the material. For example, using the temperature cycle, part and tool arrangement, and thermal properties, we can create models and find the thermal profile and degree of cure for any point in the structure. However, the same is not true for the mixed volatiles or the heat transfer boundary conditions for parts inside autoclaves. Our understanding of their interactions during the manufacturing process and their effects on the final part is still limited. The lack of understanding is another reason for including a precise description of the whole process in certification documents.

The lack of knowledge of the underlying physics or the inability to measure the physical parameters in a composite system is a type of epistemic uncertainty. The lack of knowledge extends from the thermal management parameters, e.g. the boundary conditions, to the dimensional control management parameters, e.g. the part/tool interaction, and quality management parameters, e.g. porosity. The question is whether we can define all of the measurable final outcomes that are important for the integrity of the structure and whether we can measure them efficiently and with sufficient accuracy. The proper method of measurement has not yet been defined for some of the processes, such as bonding of composite parts, which requires defining a detailed recipe and close monitoring of the process.

Moreover, due to the statistical variability inherent in composites, even subtle changes to production processes can result in unintended changes to the product attributes ³. A list of more than 60 sources of variability for the autoclave and resin transfer molding has been provided in ⁴. The variability is usually addressed with probabilistic modeling in the literature. In the composite industry, the lack of knowledge compounded with the parameter variability led to the “material equivalency” concept in the building block approach.

The building block approach is used as a validation process to assess the risk incrementally in the scale-up of part size and complexity. Material equivalency is based on the commonality of the material at all levels of the building block ¹. The minimization of the process variability is possible upon identification and quantification of the variability sources in each level. In the building block approach, material and process variabilities are evaluated at lower scales (eg. coupon level), while at higher scales (e.g. production scale), structural designs and load responses are tested.

Lack of knowledge about the underlying physics in manufacturing processes obliges the regulator and manufacturer to protect the working practice thoroughly to keep consistency and reproducibility. For example, the FAA procedure for material qualification explicitly states that “All critical prepreg constituent(s) or constituent manufacturing process must remain unchanged” ⁵. When the process parameters are altered, the requalification should be performed to show the product equivalency. Material variability and lack of design-allowables block the growth of composite material technologies ⁶.

Significant improvement in efficiency, effectiveness, and robustness requires a deep understanding of the physics of the problem. Understanding the knowledge, “know-why” approach, provides the opportunity to advance the current practice, design the process based on the final requirements, and upscale it upon request. Increasing the understanding of associated physics through science-based approaches will help us to reduce the risk of scalability and novelty as well.

Science-based approaches support engineers to identify effective physics and measure or track the important parameters with sufficient accuracy. For example, using the proper model, we would be able to capture and quantify the change in the structure’s pointwise degree of cure upon deviation from a given cure cycle. The science-based approaches can be used to directly support manufacturing decisions at all stages of the development of the manufacturing process to achieve the desired outcomes.

The science-based methods can turn practice from a prescriptive-driven process, demanding a definitive description of each stage, to a performance-driven one, based on how the part of interest would behave. For example, in the current practice, we determine the location of sensors for temperature measurements through a mix of experience and science and we protect that throughout the up-scaling to reduce risk. The better practice for thermal assessment would be to identify the critical zones using the mathematical modeling and simulations tools and decide about the required measurements, position, and numbers of sensors based on the obtained insight.

Experts argue that the use of mathematical models can accelerate the certification process. The potential for computational models to speed and assist process-structure-property optimization is not a new concept. The most direct way to reduce the cost and time of product certification is to first certify the simulation tools. Engineers can use the certified simulation tools expecting that, within the bounds of the certification, no further verification or validation would be necessary. Certification of composites simulation tools might not be a simple undertaking. Before certification of simulation tools, their maturity level must be assessed. Cowles et al. ⁷ defined five levels of maturity for simulation tools. Pipes ⁸ adds another two levels. In the final level, the simulation tool can be used individually to predict variability distribution without testing. The simulation tools give us the ability to investigate the robustness of each process. For example, engineers can place deviations, e.g. wrinkles, areas with insufficient dimensional control, and porosity, in regions and find the effects of them on strength, stiffness, or reliability.

The efficacy of simulation tools and science-based approaches also depends on the size of the manufacturing unit since smaller manufacturers have less capability to find and use the accumulated knowledge and accept and manage the uncertainty associated with innovation while moving from the “know-how” to “know-why” ⁶. The reluctance to accept a more efficient method of control and validation of the product can be helped with “Knowledge Mobilization” by research centers. Research centers can also change the tacit knowledge, gained through personal expertise, to explicit knowledge that is easy to codify and transfer. The current advances in big data and machine learning will also give us a valuable opportunity to discover the underlying physics through exhaustive data generation and data processing. The effective physics-based big-data models can potentially consider sources of variability in their prediction and enough training would increase our confidence in their results. However, for a complex model, reliability is always a big issue that makes it harder to debug, codify, and standardize the procedure.

In conclusion, it is possible to relax part of the costly material equivalency requirements by increasing the knowledge of the process. However, we cannot omit all of those process requirements because either we need them as process control parameters for monitoring the path-dependent constituents or are still not confident about all of their effects. Research centers play an important role in advancing the current “know-how” procedure to a science-based culture, investigate the physics behind bottleneck operations in the certification process, and perform the “Knowledge Mobilization” to establish the connection between the research community and policy-makers.

References