Project

# ANALYST : EM compatibility ANALYsis & Statistical Techniques in aeronautics

Goal: In the context of Electric Wiring Interconnection System (EWIS), the installation directives and specifications are generally provided by airframers to suppliers. They comply with recent certification specification and implicitly guarantee the reliability of the system. These specifications are coming from a long term experience in aeronautical industry. However, the applied good practice design and installation rules suffer from fast technological evolution in aeronautical industry (lightweight composite materials, generalization of electrification of functions on board aircraft, etc.) and industry evolution itself (development cycle, productivity, reduction of the number of physical prototypes, etc.).

These rules concern the allocation of cables in bundles (segregation in routes depending on criticity of electrical functions), the relative geometry of routes of bundles (spacing and height between routes...), the size of bundles, the cable function and loads. Those rules depend on the airframer and generally on the aircraft.

Controlling EWIS reliability in today aeronautical industry cannot be managed by good practice rules only. Aircraft industry needs reliable Tools and Methods for managing the wiring complexity of the wiring installed in new generation of aircraft.

In this context, the ANALYST’s objective is double.

In the first hand ANALYST aims at developing and validating electromagnetic (EM) numerical Tools and Methods for optimizing a very important installation concern: the minimum distance between bundles. This distance must guarantee the absence of EM susceptibility for an acceptable set of bundle parameters (constitution and geometry, crossing angle...). From a mathematical perspective, the distance must be determined with the highest possible level of confidence..

On the other hand, in some particular configurations, the minimum distance is not achievable because of the limited room available. It is then mandatory to derogate to this absolute parameter. ANALYST aims at evaluating the level of confidence associated with this derogated distance.

The minimum distance rule must be numerically evaluated on a generic configuration. In both situations, absolute and derogated, the minimum distance will be determined on a simple configuration of two straight bundles having the same length. Then the robustness of this rule will be evaluated on complex and realistic cable-bundles.

This project has received funding from the Clean Sky 2 Joint Undertaking under the European Union’s Horizon 2020 research and innovation programme under grant agreement No 821128

Date: 1 November 2018 - 1 November 2020

0 new
3
Recommendations
0 new
0
Followers
0 new
26
2 new
284

## Project log

Citation: Morio, J.; Junqua, I.; Bertuol, S.; Parmantier, J.-P. Optimisation of segregation distances between electric cable-bundles bundles embedded in a structure. Appl. Sci. 2021, 1, 0. https://doi.org/ Received: Accepted: Published: Publisher's Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. possible open access publication under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecom-mons.org/licenses/by/ 4.0/). Abstract: This paper presents the optimisation of the segregation distance between two electric 1 cable-bundles installed in an aircraft structure under electromagnetic compatibility constraints. We 2 first describe the problem formulation where a probabilistic constraint has to be verified during the 3 optimisation process. To overcome the nonlinearity of the constraint function and guarantee the 4 algorithm convergence, we propose a joint approach between Monte Carlo sampling and Kriging 5 surrogate to estimate the optimum distance with a low computational cost. This methodology is 6 tested on a realistic use-case of distance segregation between cable-bundles. 7
Due to the large amount of electrical equipment aboard modern aircraft, several EMC problems appear which can be tackled by filtering disturbances at the equipment inputs and shielding cable links (braiding, covering, over-braiding). Among the others, the segregation between interfering cables or bundles of cables becomes crucial to ensure that coupling constraints will not exceed the equipment susceptibility threshold. The aim of this work is to statistically characterize the coupling between cable bundles and its minimization by optimization under stochastic constraints. The Monte Carlo method is used, and an interpolation scheme based on Smolyak grids is investigated to accelerate the computation.
Modern aircraft hosts a large number of electric systems. Due to the transition to a More Electrical Aircraft, future platforms will use even more electric systems. This increases the complexity of the EWIS design from an EMC point of view. When possible, segregation between potentially interfering cables or bundles is a convenient solution to avoid EMC problems. Due to the huge number of installation configurations, the evaluation of the minimal segregation distance between two interacting bundles can be unfeasible. To overcome this situation, a specialized tool has been developed, based on a statistical approach for the analysis of the interaction between couples of bundles. In the paper, the different modules needed for this analysis are detailed and described. Some canonical analysis are then reported to show the capabilities of the tool.
Due to the large amount of electrical equipment aboard modern aircraft, several EMC problems appear which can be tackled by filtering disturbances at the equipment inputs and shielding cable links (braiding, covering, over-braiding). Among the others, the segregation between interfering cables or bundles of cables becomes crucial to ensure that coupling constraints will not exceed the equipment susceptibility threshold. The aim of this work is to statistically characterize the coupling between cable bundles and its minimization by optimization under stochastic constraints. The Monte Carlo method is used, and an interpolation scheme based on Smolyak grids is investigated to accelerate the computation.
Segregation rules are one of the three solutions to overcome EMC problems together with filtering of disturbances at the equipment inputs and shielding of cable links (braiding, covering, over-braiding, etc.). These rules enable mass reduction in the selection of cables and optimal route definition in which compatible cables (having similar levels of susceptibility and emissivity) can be installed inside the same bundles or nearby bundles, in the aircraft. The main parameters influencing the design of these "routes" are the distance that separate the routes from each other and the distance from the electrical structure, in order to avoid the effects of friction, electrical arcing (in the event of damage the insulators) and electromagnetic (EM) stresses [1]. Compatibility to environmental constraints [2] is usually obtained by the installation of filters or EM shields. Internal EMC constraints between the "routes" are solved by the application of a minimum distance ensuring that EM coupling constraints will not exceed equipment susceptibility thresholds. We suggest that, to the first order, this minimum distance is determined by modeling a pair of routes of similar length, adjusting and optimizing the segregation distance between them as shown in Figure 1. In this problem, this distance will vary mainly with the nature of the routes (EM sensitive versus EM not-sensitive) but also with the distance of each "route" from the nearest electrical reference, the coupling length and the crossing angle between two routes (as the routes can travel in parallel or cross each other with an angle). Figure 1. Principle for determining the minimum distance between two "routes" Two questions result from this approach: • How to determine this minimum distance considering the large number of configurations that are required? • If it is not possible to respect this minimum distance in the real aircraft installation, what level confidence can we give to lower distances in order to have their installation accepted? The second question is motivated by a significant number of requests for exemption from the cable design teams who have to deal with modifications of the 3D model of the aircraft and from the installation teams who cannot install the harnesses as specified, generally for mechanical reasons. To this extend, we propose to generate segregation distance tables when designing these routes to assess the level of acceptance of exemption distances.
The ANALYST project requires to solve high dimensionality stochastic problems. The ANALYST team invetigates several possible efficient solutions. One of them is the Smolyak method. This method is based on sparse grids that allow to efficiently represent and interpolate functions on multidimensional hypercubes.
The Lagrange interpolation technique is applied for computing the interpolation coefficients. Smolyak grid points are constructed within a normalized multidimensional hypercube along with corresponding Smolyak basis functions, then the values of the true function are interpolated at the grid points through the basis functions.
In order to improve the accuracy of approximation, the number of grid points and basis functions must equally be increased for all variables, which may be computationally costly for high-dimensional systems. To overcome these limits, an anisotropic Smolyak grid has been implemented for an asymmetric treatment of variables by allowing to separately set the accuracy level for each dimension and to increase the quality of approximation.

The issue raised in the ANALYST project may be modelled as a reliability based design optimization problem. It consists thus in a scalar optimization with reliability constraint. On the one hand, the optimization part can be handled with a constrained optimization scheme by a linear approximation algorithm (COBYLA) or with a simple dichotomy if some monotony assumptions are verified. On the other hand, assessing the reliability of a complex system with uncertainty propagation consists in estimating its probability of failure. Common sampling strategies for such tasks are notably Monte Carlo methods but they are well suited to characterize events for which associated probabilities are not too low with respect to the simulation budget. We propose here an approach that combines Kriging (a type of surrogate model) and Monte Carlo called AK-MCS to estimate the probability of failure and avoid the costly time due to the evaluation by a complex simulation code. The combined use of COBYLA and AK-MCS should enable to find a robust and reliable optimum of the ANALYST problem.

At the design phase of a harness, the positions of the cable bundles inside the 3D structure are not defined and there is a significant incertitude on the exact running of the routes of bundles. This incertitude defines a kind of space reservation that can be shared by several routes. In addition, even if the constitution of the routes is quite known in terms of cables, the end loads, the EM susceptibility levels of the function links and the perturbation sources of those functions are roughly known.
The problem to be solved in ANALYST has been made more precise in WP2. It consists in determining the minimum distance acceptable between two routes (or bundles) regarding the above mentioned imprecisions.
For this, we define a simplified geometrical configuration made of two routes of equal length over an infinite and perfectly conducting plane, with a set of limited random parameters; the length of the routes, the heights of each route, the crossing angle between the routes (when the routes are not parallel). The problem to be solved is an optimization problem: the determination of the minimum distance between two routes, guarantying an absence of EM susceptibility with a given level of confidence.
The concept of ANALYST consists in applying this minimum (optimized) distance between routes, determined on this simplified configuration, all over a real harness installation.
In WP2, test-cases of increasing complexity on which the ANALYST tools will be developed and applied have been defined by SAFRAN. On the one side, they concern generic test-cases with a low number of parallel routes and cables and, on the other side, complex test-cases with a large number of parallel routes and cables. A third type of test-case concerns a real branched cable network involving several routes and several equipment pieces, for which measurements will be made available by SAFRAN at the end of the project for the validation of the ANALYST concept.