Skip to main content

Physical Architecture Optimization System

Periodic Reporting for period 1 - PHAROS (Physical Architecture Optimization System)

Reporting period: 2019-10-01 to 2021-09-30

Modern engineering products such as aircraft consist of a multitude of interconnected subsystems such as engine, control systems with sensors, actors and displays with switches and so forth. The design of an aircraft includes therefore the design of all these interconnections between these subsystems. More specifically, electrical subsystems need to be interconnected using an electrical wire harness (by a so-called "3D-routing" task) and the hydraulic subsystems need to be interconnected by a hydraulic pipe network (by a so-called "3D-piping" task). Since the start and endpoints of both networks depend on the chosen locations of the subsystems inside the product (i.e. the so-called "3D-packaging" task), all the interconnections of a subsystem need to be updated each time when the package of the system is changed.

In the aircraft industry, this sequence of 3D-packaging, 3D-piping and 3D-routing represents the key issue in the so-called "physical system architecture" design and is executed multiple times (in the course of so-called "architecture trades") in order to find a feasible, a good or even an optimal solution in respect to the given constraints. Since in today's European aircraft industry, physical system architecture design is a design task executed manually by human experts, the design and all the subsequent modifications to achieve an optimal physical system architecture accounts for many tedious design loops which may easily take days, weeks or even months. The fact that the task of physical system architecture design is multi-disciplinary by nature and is governed by internal company design codes as well as by international and national design standards and regulations which underlines the immens manual and intellectual effort behind.

Europe's competitiveness in the aircraft sector relies on technological innovations driven by hardware, software or process innovations in both design and manufacturing. Despite the permanent increase in product complexity and the rising demand in product mass customization, the efforts of reducing cost, shortening time to market and increasing product quality are the key factors for the commercial success of Europe's aerospace industry facing global competition. Automating complex design tasks by machine-executable workflows consisting of sequences of "smart algorithms" borrowed from graph theory and artificial intelligence embedded in a software optimization framework (i.e. the so-called PHAROS software stack) allows the automated exploration of design alternatives for 3D-packaging, 3D-piping and 3D-routing. This results in enormous savings of design time and cost while still increasing product quality by first-time-right designs and collapsing the former manual design time of weeks to months down to the runtime of the smart algorithms, usually only hours or days.

The "Physical Architecture Optimization System" (PHAROS) project aimed therefore at the development of a fully automated software solution for physical system architecture design under special emphasis on developing an algorithmic solution for 3D-packaging, 3D-piping and 3D-routing embedded in an optimization loop. The elimination of tedious manual routine work allows to shift the engineering effort to higher levels in the product value chain and to concentrate more on engineering system performance and system architecture trades instead. The performance of the developed PHAROS software stack is illustrated in a public demonstrator in form of a wing system architecture optimization, and a confidential industrial demonstrator in form of an aircraft A320 series landing gear bay provided by the industrial partner AIRBUS.
The project started with the collection of requirements for physical system architecture design, most notably with requirements for 3D-routing and 3D-piping. As examples, minimal bending radii, minimum distance to other cables or to other system components or structural parts are typical design constraints for the routing algorithm during computation. Similarly, manufacturing constraints for pipe bending machines include maximum angles and necessary straight lines in between two angles are typical constraints for the 3D-piping algorithm.

Concerning the public demonstrator, the following digital models were manually created in the PHAROS project: 1) a CAD-file containing a wing structure. The wing structure offers several passages which can be opened or closed in order to provide several different topological configurations of the design space, 2) a MODELICA-file which contains a system model of the electric and the hydraulic system, and a bill of material (BOM) with qualified CAD-Parts of the necessary connectors and fixings. The public demonstrator was used to execute an in-depth design-of-experiment (DoE) study in order to explore the influence of the different topological choices of passages and component positions inside the wing and yielded an optimized wing system architecture with minimal weight. One single architecture generation run (3D-packaging, 3D-piping and 3D-routing) inside the DOE study in the cloud took about 1,5 hours.

Concerning the confidential demonstrator, AIRBUS provided CAD-data of an AIRBUS A320 series landing gear bay as an industrial use-case. Using this A320 series CAD data, an automated 3D-piping was executed in the PHAROS software stack and yielded a piping solution with about 10% less weight and length at the cost of a 40% increase of the number of bends. These results were generated fully automatically in about 23,5 hours on a desktop PC and were very positively evaluated by AIRBUS system experts.

Towards the end of the PHAROS project, 2 papers were successfully placed in peer-reviewed conferences, the EASN 2021 conference and the DLRK 2021 conference. At the time of this writing, the reviewer decision whether the DLRK2021 paper is invited to be extented into a journal paper, is still pending. Based on the promising intermediate results, the two PHAROS project partners University of Stuttgart and IILS were capable to win another research grant in the ERA-NET MANUNET program to built a digital factory for the manufacturing simulation of wire harnesses. Several other oral presentations of the project at BOSCH and AIRBUS and the creation of a PHAROS project flyer available at the websites of the project partners IILS and NOESIS for download complemented the exploitation and dissemination efforts of the PHAROS project.
The PHAROS software stack with its capability to automate physical system architecture design in the form of smart algorithms for automated 3D-packaging, 3D-piping and 3D-routing and optimize either for minimum weight or manufacturing cost exhibits several features well beyond the state of the art. It furthermore reduces design times by at least one order of magnitude and allows the exploration of alternative topological architecture design solutions at almost no manual effort. Due to the algorithmic solution, designs are first-time-right which helps to save cost and time and avoids scrap and waste in the production after. Understanding the underlying complexity of such an automation approach both from an engineering and a computer science perspective strengthens Europe's engineering position and helps to further push engineering design productivity to new levels in order to justify higher wages for multi-disciplinary educated and trained specialists.
Picture shows the optimized physical system architecture of the public demontrator