A case study in CAD design automation

Abstract

Computer-aided design (CAD) software and other product life-cycle management (PLM) tools have become ubiquitous in industry during the past 20 years. Over this time they have continuously evolved, becoming programs with enormous capabilities, but the companies that use them have not evolved their design practices at the same rate. Due to the constant pressure of bringing new products to market, commercial businesses are not able to dedicate the resources necessary to tap into the more advanced capabilities of their design tools that have the potential to significantly reduce both time-to-market and quality of their products. Taking advantage of these advanced capabilities would require little time and out-of-pocket expense, since the companies already own the licenses to the software. This article details the work of a small research team working in conjunction with a major turbine engine manufacturer endeavoring to make better use of the underutilized capabilities of their design software. By using the scripting language built into their CAD package for design automation, knowledge-based engineering applications, and efficient movement of data between design packages, the company was able to significantly reduce design time for turbine design, increase the number of feasible design iterations, increase benefits from relational modeling techniques, and increase the overall quality of their design processes. The design of turbine engines involves creating, modeling, and documenting the development of airfoil geometry for turbine, impeller, and compressor blades. This process is highly iterative due to the circular revisions made between design and analysis groups chasing the optimal airfoil shape and performance. Airfoil blades are a crucial component within a turbine engine, and their design covers many engineering disciplines such as thermodynamics and statics. For both analysis and manufacturing, these airfoils are modeled in a CAD system. However, the complex shapes of airfoils make this difficult. They are typically modeled using b-splines or NURBS, and the development of methods to do this has been ongoing for decades (Corral, Roque, Pastor, & Guillermo, 2004; Korakianitis & Pantazopoulos, 1993)). After revisions are made, geometric data are often reengineered and recreated within the CAD system. This process ranges from hours to days because the current methods of creating the airfoil models in the CAD system are not parametric, (i.e., the geometry is not associated with the engineering definition of the airfoil after the model is created). A turbine engine can contain as many as of 20 different airfoils, so any improvement in the time for one design iteration will have a beneficial effect on the total design process. In addition, additional benefits can be realized depending on whether a turbine, compressor, or fan blade is being designed, as the geometric complexity of each part varies from relatively simple to highly complex. According to O’Brien et al. (2006), the knowledge-based engineering (KBE) techniques can make a substantial impact in the design of engineering products. It is gaining prominence as a major tool to speed up product development by capturing knowledge from engineers and designers and embedding that into software configuration (Bermell & Fan, 2002; Prasad, 2005; Rosenfeld, 1995). This knowledge is then used to assist designers while they create products within the CAD system (Hunter, Rios, Perez, & Vizan, 2005). KBE systems are used to automatically create objects (Clark, 2001; Sekiya, Tsumaya, & Tomiyama, 1998), assist designers while they create objects (Carleton, 2005), and compare the cost versus efficiency of created objects (Susca, Mandorli, & Rizzi, 2000). The industrial research partner in this project does its CAD design in Siemens PLM NX and ports their models into assorted versions of ANSYS and various other in-house applications for analysis. At the beginning of the project the design process was almost totally manual – aerodynamics engineers would pass point cloud data representing turbine airfoils to modelers who would spend one or more full workdays constructing a CAD model from the data. This time encompasses only the airfoil itself and not any of the turbine wheel attachment points or internal cooling geometry. There were no standards in place, so each modeler created their T h e J o u rn a l o f Te c h n o lo g y S tu d ie s A Case Study in CAD Design Automation Andrew G. Lowe and Nathan W. Hartman 2