|| FPA |
Fast Panel Analysis
CDI's Fast Panel Analysis, (FPA), is a revolutionary new boundary element based aerodynamic design tool for the rapid prediction of steady subsonic potential flows about complex 3D configurations.
Boundary element methods (or panel methods) have been in use for over three decades, playing a key role in the preliminary design process for aerospace, marine, automotive,
and a wide range of other industries. Unfortunately, the panel codes in use today are based on 30 year old technologies. CDI's new Fast Panel Analysis (the FPA code)
integrates state of the art fast multipole methods, hierarchical data structures, and iterative solution techniques leading to more than an order of magnitude reduction
on both the computation time and storage requirements for boundary element simulations. With the FPA code designers can routinely perform design trade studies on complex
configurations with more that 100,000 panels on workstation class machines.
Space shuttle surface Cp distribution using 35,022 panels. Solution time 36 min.†
Civil transport surface Cp distribution using 12,542 panels. Solution time 13.5 min.†
This tremendous reduction in simulation time gives designers the opportunity to introduce accurate aerodynamic modeling earlier in the preliminary design process,
to explore a wider range of design configurations, and to integrate new high risk technologies into the product line earlier. Further, the speed and low memory requirements
of CDI's FPA code opens the way for application of high fidelity aerodynamic modeling in a wide range of emerging multidisciplinary applications, including aerodynamic
devices using smart materials, morphing technologies, and advanced flow control. Overall this new technology has the potential for reducing design cycle times, improving
product performance, and reducing design costs in an increasingly competitive market place.
Panel methods offer the following advantages for modeling subsonic flows:
The main drawback of conventional panel codes is that the computational cost, measured in terms of both storage and CPU, rises rapidly with the number of panels, thus severely
restricting the problem size or in a design environment the design envelope which may be explored. CDI's FPA code integrates a fast multipole method with hierarchical data
structures and an innovative iterative solution algorithm to drastically reduce these computational costs. Formally, for N panels, both storage and CPU are reduced from O(N2)
to O(N log N). In practical terms, the computational cost for full aircraft configurations (5,000 to 100,000 panels) tend to scale as 1 min. CPU time per 1000 panels and 1 MB
memory per 1000 panels†.
- Only requires a surface panelization. (No volume grid generation problems.)
- Robust formulation which can handle discontinuous surface panel distributions and small gaps in the surface description.
- Implicit and accurate far field boundary conditions.
- Higher accuracy than FV, FD or FE-based CFD codes for same surface resolution.
This new design tool represents the culmination of over ten man-years of activity in fast panel and fast modeling at CDI, and incorporates technical
developments from a variety of NASA- and company-sponsored initiatives. CDI is committed to the long term support and development of fast boundary element methods and to meeting
the needs of individual customers using this technology.
- More than an order of magnitude reduction in computation times and computational memory requirements compared to standard panel codes. Memory requirements and CPU
costs are observed to scale respectively as 1 MB / 1000 panels and 1 min. / 1000 panels†.
- A truly unstructured and flexible geometry description. All that is required is the number of panels and the coordinates of the panel corners. Panel connectivity and other
necessary geometric information is developed internally using fast search procedures.
- Wake specification options include: automatic generation of wakes off all downstream sharp trailing edges; panel index-based specification of wake release edges and geometric
definition of wake release lines.
- Linear subsonic flow modeling.
- Optional reconstruction of linear order doublet distributions for superior predictions of the near surface velocities.
- Portability over multiple platforms.
† Silicon Graphics, Inc.™ Origin R10,000/195 MHz processor.