Simulation - Acsystème

Simulation : what for ?

Even if the majority of industrialists trust the power of simulation when it comes to designing and optimising structures, there are still some sectors where this approach is not in use, sometimes owing to the unawareness of potential benefits.

Thanks to the improvement of computation performance, and particularly with numerical computation, modelling and simulation have become key elements when studying a system. Through simulation, we are able to explore a lot of situations in a relatively short time. According to the need, modelling can cover several shapes (knowledge-base model, black box…) with different levels of detail (global model, simplified model, finite element…).

In this article, we will present a (non-exhaustive) overview of the modelling approach in industry. It is illustrated with examples that are derived from Acsystème experience in which simulation is employed for:

operational and behavioral analysis,
sizing and optimising process.

Behavioral analysis

In this case, simulation is used to check some aspects of system principle. Thus, EDF designs control systems for automatons based on numerical models that will help to expand its hydro-electrical facilities. These models also guarantee safety and security standards.

Simulation is also applied to elaborate strategies in case of an incident: it will validate the quality of corrective and preventive measures under non-replicable scenarios (10-year flood…) or requiring expensive or dangerous tests.

Design, optimisation

To design and optimise, it is possible to create a model and simulate it according to different operational scenarios. Afterwards, we know information like the efficiency of a machine, the energy cost to manufacture a product, the equipment wear… Then, the designer will be able to change (manually or with a specific algorithm) several parameters so as to optimise system performance.

For instance, if we consider an industrial furnace, simulation can help to improve product quality and productivity while limiting energetic consumption or pollutant emissions.

Knowledge-based model and behavioral model

In any case, modelling may be performed in various manners. We are used to hearing about knowledge-based or behavioral models.

A knowledge-based model is drawn from physical links between system quantities. These relations are built by reasoning from specific hypothesis and general laws of physics applied to the system. So, to model the movement of a mechanical element, we can assimilate it to its gravity center, neglect friction and apply fundamental principles of dynamics.

On the other hand, behavioral models constitute a purely mathematical description of the relations observed between input (causes) and output (effects) quantities, without relying on system physics. For instance, the characteristic curve of a piezoelectric injector is a complete behavioral model. Identification techniques like neural network or parametric estimation are a good way to obtain more complex behavioral models.

Practical application

Automotive

Industry

Autonomous vehicles in our ports

BA Systèmes and Gaussin: two French companies tackling port terminals to develop automated solutions for container logistics management.

Industry

Agri-food

Optimisation of the level control for an ammonia evaporator

Yara, a chemical company producing fertilizer, wished to improve the level control dynamic behaviour of the workshop producing ammonium nitrate. A strategic project due to the precision required for the chemical formulas.

Automatic control

Model-Based Design

Control expertise

Simulation