Generative Manufacturing: A requirements and resource-driven approach to part making

Abstract

Advances in CAD and CAM have enabled engineers and design teams to digitally design parts with unprecedented ease. Software solutions now come with a range of modules for optimizing designs for performance requirements, generating instructions for manufacturing, and digitally tracking the entire process from design to procurement in the form of product life-cycle management tools. However, existing solutions force design teams and corporations to take a primarily serial approach where manufacturing and procurement decisions are largely contingent on design, rather than being an integral part of the design process. In this work, we propose a new approach to part making where design, manufacturing, and supply chain requirements and resources can be jointly considered and optimized. We present the Generative Manufacturing compiler that accepts as input the following: 1) An engineering part requirements specification that includes quantities such as loads, domain envelope, mass, and compliance, 2) A business part requirements specification that includes production volume, cost, and lead time, 3) Contextual knowledge about the current manufacturing state such as availability of relevant manufacturing equipment, materials, and workforce, both locally and through the supply chain. Based on these factors, the compiler generates and evaluates manufacturing process alternatives and the optimal derivative designs that are implied by each process, and enables a user guided iterative exploration of the design space. As part of our initial implementation of this compiler, we demonstrate the effectiveness of our approach on examples of a cantilever beam problem and a rocket engine mount problem and showcase its utility in creating and selecting optimal solutions according to the requirements and resources.

Figures (17)

• Figure 1: Our proposed concept of Generative Manufacturing: requirements and resource driven design.
• Figure 2: Our proposed framework. 1) The user inputs the problem domain and boundary conditions and a set of manufacturing method and material combinations. 2) An initial probe of the supply chain using system generated representative part guesses helps swift removal of infeasible combinations. This process also helps establish the relationship between different requirements (mass, compliance, lead time, cost) for each of the current suppliers, gives an approximate range of values for each of the requirements that help the user determine the constraints for mass, lead time and cost for performing topology optimization in the design generator and also helps determine the best supplier to achieve Pareto optimal solutions. 3) One optimized design corresponding to each input configuration (defined by boundary conditions, manufacturing method, material and supplier) is output by the design generator and then passed through the supply chain scheduler to get the lead time and cost. 4) All the designs are then evaluated and visualized using the Explainable AI and Results Interface, where trade-offs are explored, which helps in user feedback to the system and selection of the most effective part.
• Figure 3: Decentralized Coordination with Supply Network
• Figure 4: The design generator. 1) The problem domain defines the coordinates input into the neural network (NN) that outputs the density at each of these coordinates. 2) The density values define the topology of the part on which the quantities in the loss function depend. 3) The material, manufacturing method, supplier probing and constraints help define the loss function. 4) The NN learns the weights (using backpropagation and gradient descent) to output the optimal topology that minimizes the loss function.
• Figure 5: The decision tree explainer. 1) All of the designs generated so far are used as inputs into a decision tree learner, implemented with the sklearn library. 2) The output from this is post-processed to provide a cleaner display of the decision logic and relevant information about the design space covered by each subtree. Subtrees containing designs from the current iteration are highlighted in green.