“Data is the new plastic” - Data algorithms in ….. Plastic Materials Selection

Hello and welcome to this blog post entitled “data is the new plastic”.

“In God we trust, all others must bring data.” W. Edwards Deming

As data science evolves throughout the plastics industry, materials, moulds and machines information come closer together by means of digital services. Following, are the three main business interfaces where data science is playing or will play a crucial role:

1. Resin suppliers and their customers: Better decision making together with the customer with regards to material proposal and selection (optimal material at the right time for the right price).
2. Plastics processing: Increased efficiency and streamline processes. For example, Toray Plastics optimized their manufacturing operational performance by using GE Digital’s Big Data Analytics [9].
3. Product designers and tool makers: Better prediction of which applications will evolve due to touch points with customer (e.g. industrial marketing offers by GadflyZone [10]) and better product design using data and simulation like CFD and FEA

As data analytics and data mining operations find their way through the plastics business value chain, business operations such as marketing, sales, social selling and customer relationship management (CRM) evolve into new disciplines - Plastics Information Technology (PIT). Data-driven marketing actions ensure a better customer journey.



Figure 1: Data generated from polymeric materials, machines and moulds is the foundation for Plastics Information Technology.

In this post, I focus on point 1) from above and thus on data driven decision making on the optimal material for a certain application. I will not cover the conventional selection methods, since most of us are familiar with them. Especially for automotive industry, I would like to refer to Paul F. Kusy’s material selection guide [11]. Basically, we always have a material screening phase and a material comparing phase. In the first one, elimination of the obvious unsuitable materials takes place. The second phase involves a multi-level decision process which finishes with a material decision step.

Al-Oqla et. al. [8] explains that the design of an engineering part has three elements: specifying a shape, selecting a material, and choosing a proper manufacturing process. The “selecting the material” element will be supported more and more by algorithms.

Figure 2: Overview of the data science approach for polymeric materials selection.

There are three main groups of algorithms which can be applied to the screening phase of materials selection and comparison phase (Figure 2):

1) Artificial intelligence (AI) methods: AI is used for solving complex problems by implementing intelligence for processing unstructured knowledge. Looking back, an engineer could inform himself in engineering handbooks, scientific journals and own experience. As the amount of polymeric materials and their compounds are increasing, an efficient managing of engineering data is necessary. Here, AI methods can play a key role. AI can access over the network (and stored in the cloud) all kind of information and does not need to make mistakes to learn. Following are some methods used in AI for materials selection:

• Computer-aided materials selection systems [1]
• Knowledge-based systems [6]
• Case-based reasoning [2]
• Neural networks [3]

2) Genetic algorithms combined with neural network: Connecting the desired mechanical properties with the polymer properties is enabled by genetic algorithms and neural network. It was found that neural network reduces the time and costs for selecting the optimum polymeric material. Al-Oqla et al. [8] has shown this by applying genetic algorithms and neural network for selecting the optimum composite material for different applications. In this case, fuzzy oriented models were used to select the relationship between performance requirements and material properties.

3) Optimization: three optimization techniques are shown for enabling better materials selection using an optimization algorithm.

• Mathematical programming [5]
• Computer simulation [7]: Simulation tools like for example Moldflow and Sigmasoft have an extensive database on materials which allow you to test materials of different costs in a virtual domain.  It is possible to couple the results to a finite element analysis to simulate the mechanical performance too.
• Genetic algorithm [2]

The result of your screening process will be two to three materials which can do the job. Following, is the second phase of the material selection. This itself is a multi-level decision making process to come to a definite outcome. There are the following methods available which are supported by algorithms:

• Multiple attribute decision making methods (MADM) [12]: Focus on making the decision based on the attributes of the material which can be expressed as exact numeric values, Boolean values or just as ranking (best-good-bad).
• Multiple objective decision-making methods (MODMM) [12]: Focus on making a material decision considering different desirable objectives. The design engineer has the objective to develop the lightest and/or cheapest solution considering the boundary conditions.
• Polymer processing decision solver [4]: The main element of this program is a decision matrix which has as input criteria and options. As a result, the user will get a weighting of the options with points. The options are the different polymers to be decided upon. 

Let’s start harvesting insights from your data generated by our plastics products and get a competitive edge by proposing the optimal material!

Thanks for reading and till next time!

Greetings, Herwig Juster

New to my Find Out About Plastics Blog – check out the start here section.

Literature:

[1] Jahan, A., Ismail, M. Y., Mustapha, F., & Sapuan, S. M. (2010a). Material selection based on ordinal data. Materials & Design, 31, 3180–3187.

[2] Amen, R., & Vomacka, P. (2001). Case-based reasoning as a tool for materials selection. Materials & Design, 22, 353–358.

[3] Goel, V., & Chen, J. (1996). Application of expert network for material selection in engineering design. Computers in Industry, 30, 87–101.

[4] H. Juster, Strategic decision-making in the product development of polymer components, Master thesis, University Leoben, Austria, 2015 

[5] A. Jahan, M.Y. Ismail, S.M. Sapuan, F. Mustapha, Material screening and choosing methods – A review, Materials & Design, Volume 31, Issue 2, 2010, Pages 696-705, ISSN 0261-3069, https://doi.org/10.1016/j.matdes.2009.08.013.(http://www.sciencedirect.com/science/article/pii/S0261306909004361)

[6] Sapuan, S. M. (2001). A knowledge-based system for materials selection in mechanical engineering design. Materials & Design, 22, 687–695.

[7] SIGMASOFT® Virtual Molding: The best Ally in Resin Selection

[8] AL-Oqla, F. M., & Sapuan, S. M. (2015a). Polymer selection approach for commonly and uncommonly used natural fibers under uncertainty environments. Journal of the Minerals Metals and Materials Society, 67(10), 2450–2463.

[9] https://www.featuredcustomers.com/vendor/ge-digital/customers/toray-plastics-inc

[10] http://gadflyzone.com/

[11] Kusy, P.F., Plastic Materials Selection Guide, 1976

[12] AL-Oqla, F. M., & Omar, A. A. (2015). An expert-based model for selecting the most suitable substrate material type for antenna circuits. International Journal of Electronics, 102, 1044–1055.