Home   |    International   |    Contatti   |    Info   |   

Engine Innovation - Engine Layout definition through modeFRONTIER/WAVE automatic optimization

The definition of thermo-fluid dynamics behavior of new engine layout related to achieving of specific target is the first step of development of new product. It involves both aspect, the first is the pre-design of engine and the second one is the simultaneously design of several components.
In this job it has been built a whole parametric model of four-stroke 4V engine of scooter by means of one-dimensional WAVE code, calibrating by comparison with experimental data.
Once the parametric model was available, modeFRONTIER – optimization software - has been coupled with it in order to manage high number of variables, complex relationship amongst several part of engine, non-linear responses of the components involved in.
modeFRONTIER can handle in automatic way the variable parameters changing them to reach efficiently the best configuration of the engine, improving performances and reducing development time, thanks to the use of DOE and Multi Objective Genetic Algorithm.

modeFRONTIER coupled with one dimensional WAVE parametric model lead to a new approach compared to traditional trial-error approach, which allow to define/optimize the entire layout of engine and to optimize each component of it.
Among the several outputs deriving from WAVE analysis the most significant ones have been chosen as objectives of the optimization: the torque and power diagram respect to rpm changing has been considered and exactly the maximum value of torque and power (Cmax and Pmax)in that range. At the end of optimization processes the predicted value of Cmax and Pmax have been increased of 30% and 23% respectively, and at the same time a considerable time saving has been achieved. The results have been finally checked by means of experimental test, which confirmed the improvement of performances

Altri case studies sullo stesso argomento

Altri case studies sullo stesso argomento

Calendario 2018
Corsi di formazione sulle tecnologie software

Formazione personalizzata
Percorsi formativi personalizzati e traning-on-the-job: contattaci!

Workshop e seminari
Visita la pagina degli eventi: potrai partecipare alle iniziative EnginSoft ed essere informato sui più importanti appuntamenti europei

Newsletter EnginSoft
Leggi l'ultimo numero ed Abbonati alla rivista!


Chiedi all'esperto!
Articolo completo

The definition of the thermo-fluid dynamic behaviour is the first step of the development of a new engine layout to achieve the required performances in compliance with the related specification. The reference is to the optimization process necessary for the definition of the layout of a four-stroke engine of a Piaggio scooter.
The specific objective pursued has been the engine performances optimization in terms of torque/power, in accordance with the physical and constructive constraints fixed by the specification, such as airbox volume, maximum lengths of the intake and exhaust ducts in relation to dimensions problems, timing diagram for valve/piston interference.
In order to achieve such results, once the parametric model was available, we used modeFRONTIER, as optimization environment, coupled with WAVE, as software for engine one-dimensional thermo-fluid dynamic analysis.
The WAVE outputs chosen as main objectives were the torque and the power curves in relation to the rpm variation. At the end of the optimization process the achieved results were a remarkable time saving due to process automation as well as an improvement of a 30% and 23% respectively in Cmax and Pmax. Such results have been then validated through experimental tests, which confirmed the performance-increasing trend already underlined.
modeFRONTIER coupled with WAVE computation code, for the management of the engine parametric model, has led to a new problem approach, instead of the traditional trial&error one, allowing the definition/optimization of the whole engine layout and the optimization of specific components, as well as the management of several variables and constraints which are necessary for the description of the complex connections among the different engine parts.

The thermo-fluid dynamic definition of a one-dimensional four-stroke engine for a scooter was achieved using the modeFRONTIER optimization code, applied to the 1D WAVE simulator, aiming at the identification of the optimum configurations in terms of performances, in accordance with the precise targets defined by Piaggio specification.
The hypothesis has been therefore to change the induction line, the exahust system, the internal head pipes, the laws of valve lift and timing.

piaggio multi objective optimization

Fig. 1: Engine layout and mono-dimensional model

Engine 1D model and Engine experimental validation
The first activity carried out with the WAVE 1D model has been the engine layout, as the following figure (Fig. 1) shows:

Afterward the 1D model has been validated by means of experimental measurements. The figure 2 here below shows how the model is representative of the real engine thermo-fluid dynamic behaviour, underlining its predictive elements.

Parameterization for modeFRONTIER application
The adjusted model has been parameterized and used in modeFRONTIER in order:

  • To define the geometric components together with most interesting input variables to create a generalized model, free from the starting configuration;
  • To obtain a 1D WAVE model which can de better managed by the optimizer;
  • To simplify some components to reduce the design computation time.
piaggio multi objective optimization

Fig. 2: Experimental validation – intake pressure value

piaggio multi objective optimization
Fig. 3: modeFRONTIER workflow

Parametric expressions have been worked out according to realistic schemes (both geometrically and physically) for the following components:

  1. airbox,
  2. silencer,
  3. head internal pipes,
  4. throttle body.

Furthermore, the influence of every single modification of the output has been estimated during the whole parameterization process, with the aim of avoiding any mistake in the generalization/ simplification.

modeFRONTIER automatically manages the engine geometry through the outputs (variables, constants, expressions) and the transfer functions that define the single elements, that is the “bricks” of the one-dimensional code. The input variables control:

  1. the pipes length and diameter;
  2. the connections volumes and diameters;
  3. lifts and timing diagrams.

In order to make the engine layout fully parametric and manageable in modeFRONTIER, we used:

  • 20 inputs variables,
  • 18 constant inputs,
  • 8 « expression » inputs,
  • 44 transfer functions.

The choice of using some inputs as constants was due to its versatility; the simple switching into variables allows to carry out a new optimization with no need for a further workflow and pointing to be rebuilt. Furthermore, new constraints concerning dimensions problem and other technical constraints were included.
In the workflow main diagram here below, each sub-model represents a specific component with its associated variables, constraints and constants. The example refers to the exploded view of the induction line subsystem.

The engine performance values (power and torque), in relation to the rpm, were identified as relevant outputs. According to them, the following objective were defined:

  1. to optimize power and torque for fixed rpm values;
  2. to model the power curve in order to guarantee a vehicle that can be easily driven.

Optimization strategies
Considering the huge amount of variables and their wide range of possible applications, leading to something like a 1013 possible configurations to be evaluated, it was finally decided to work out a DOE of about 9000 cases, by means of a Sobol Sequence algorithm. The idea was to obtain an as much uniform database distribution as possible for a exhaustive investigation of the configuration space.

The MOGAII genetic algorithm was then applied, considering its robustness and good convergence capacity; the on-line post-processing has allowed a better comprehension of the results and the identification of the suitable moment to stop the optimization, with reference to the improvements achieved between the different generations (Fig. 4).

piaggio multi objective optimization piaggio multi objective optimization
Fig. 4: optimization history diagram
Fig. 5: modeFRONTIER scatter chart – Pareto frontier highlighted

The parallel computation on 4 processors generated 53400 cases in 7 days (an average of 4 cases per minute). The optimization process reached a good convergence after a certain number of generations, with no need for further MOGA II generations. The most remarkable results have been:

I. scatter chart of the solutions and the Pareto frontier (red solutions), whose concentration was, as expected, in the last 5000 generated configurations, proving the convergence to the frontier solutions,

piaggio multi objective optimization

Fig. 6: modeFRONTIER correlation chart

II. 3D correlation chart that shows:

  • the correlations between the different input variables. Their analysis allows to understand not just their correlations, but also their relation with the objectives, for a more efficient management of the problem,
  • the influence of a single input variable on the achievement of the objectives.

III. parallel chart of the power curves that allows:

  • to have a database with the whole set of configurations,
  • to process the thousand of possible solutions in a faster and more efficient way,
  • to directly work on outputs, taking advantage of the active “filter” functions, so to obtain a real-time engine configuration in accordance with the fixed objectives.

modeFRONTIER, the multi-objective computation code, was used to define the layout of a four-stroke engine.
In order to do that, a fully parametric WAVE model was first created, where the following components were treated as variable parameters:

  1. airbox volume
  2. diameters and lengths of the induction ducts
  3. type of throttle body
  4. sizing of internal head pipes
  5. distribution (timing and lifts)
  6. length and diameters of the exhaust manifold
  7. silencer volume
piaggio multi objective optimization piaggio multi objective optimization

Fig. 7 : modeFRONTIER parallel chart avoid on-line filter of solutions

Fig. 8: engine power – baseline (red) and optimazed (green) solution

The coupling of the optimizer with the 1D WAVE computation code has allowed:

  • to predict the achievement of the required power and torque curves,
  • to define the sensitive variables and their correlations,
  • to work out a first identification of the design decisions making,
  • to have a real-time and user-friendly database of the engine configurations,

in a short setting up phase and with an innovative and automatic approach, instead of the traditional trial&error method.
The procedure diagram and the model parameterization allow the application of modeFRONTIER to both the engine layout definition and the single components development/optimization.


Articolo pubblicato sulla Newsletter EnginSoft Anno 5 n°1

Di Palma Stefano, Matteucci Luigi, Simoncini Marco
Piaggio&C. S.p.A

EnginSoft SpA | © All rights reserved | VAT nb IT00599320223
Questo sito utilizza cookie tecnici e cookie analytics, che raccolgono dati in forma aggregata | Informativa completa sulla privacy e sui cookie