Development of Acoustic Simulations using Parametric CAD Models in COMSOL

(1)

Linköping University | Department of Management and Engineering Master’s thesis, 30 ECTS | Aeronautical Engineering Spring term 2019 | LIU-IEI-TEK-A-19/03468-SE

Development of Acoustic

Simulations using Parametric

CAD Models in COMSOL

Antoine Bouilloux-Lafont Rubén Noya Pozo

Supervisors: Anton Wiberg, Linköping University Ulf Sellgren, KTH University

Jakob Nyström, SCANIA CV AB Dario Vignali, SCANIA CV AB Examiners: Mehdi Tarkian, Linköping University

Ulf Sellgren, KTH University

Linköping Universitet SE-581 83 Linköping

(2)

AUTHORS:

Rubén Noya Pozo

M.Sc. Student in Machine Design KTH University | Sweden

Antoine Bouilloux-Lafont

M.Sc. Student in Aeronautical Engineering Linköping University | Sweden

SUPERVISORS:

Jakob Nyström

Exhaust Systems V8 Engines (NXDE) Scania CV AB | Sweden

Dario Vignali

Fluid Dynamics and Acoustics Simulation (NXPS) Scania CV AB | Sweden

Anton Wiberg

Division of Machine Design Linköping University | Sweden

Ulf Sellgren

Department of Machine Design KTH University | Sweden

EXAMINERS:

Ulf Sellgren

Department of Machine Design KTH University | Sweden

Mehdi Tarkian

Division of Machine Design Linköping University | Sweden

(3)

(4)

Abstract

With constantly changing regulations on emissions, heavy commercial vehicles man-ufacturers have to adapt for their products to preserve their quality while meeting these new requirements. Over the past decades, noise emissions have become a great concern and new stricter laws demand companies to decrease their vehicle pass-by noise target values.

To address the requirements from different disciplines, Scania follows a simulation driven design process to develop new concept models EATS. The collaboration among engineers from different fields is thereby necessary in order to obtain higher perfor-mance silencers. However, the pre-processing step in terms of acoustic simulations is time-consuming, which can slow the concept development process.

In this thesis, a new method was introduced to automate the pre-processing of si-lencer acoustic models and allow for design optimisation based on acoustic perfor-mance results. A common Scania product study case was provided to several theses within the NXD organisation. The collaboration among the master thesis workers aimed to demonstrate the benefits of KBE and MDO and how they can be integrated within Scania’s current concept development and product introduction processes. The performed work was divided in the following steps: data collection, method development and concluding work. The first step consisted in gathering sufficient knowledge by conducting a thorough literature review and interviews. Then, an ini-tial method was formulated and tested on a simplified silencer model. Once approved and verified, the method was applied to the study case EATS.

The study case showed that a complex product can have its acoustic pre-processing step automated by ensuring a good connectivity among the required software and a correct denomination of the geometrical objects involved in the simulations. The method investigated how morphological optimisations can be performed at both global and local levels to enhance the transmission loss of a silencer. Besides opti-mising the acoustic performance of the models, the method allowed the identification of correlations and inter-dependencies among their design variables and ouput pa-rameters.

Keywords: parametric CAD models, KBE, acoustic simulations, optimisation, col-laboration, MDO

(5)

(6)

Acknowledgement

The work performed in this thesis was conducted at Scania CV AB in collaboration with Linköping University and KTH Royal Institute of Technology during the spring of 2019. The carried out work has been challenging but also educational, giving us valuable knowledge regarding acoustics and the truck industry. In addition, the col-laboration with other master thesis workers allowed us to understand the complexity in designing a model satisfying requirements across several disciplines.

First and foremost, we would like to sincerely thank all the people at Scania CV AB who helped us achieve the goals of this master thesis successfully and made us feel welcomed from the very first day. We would also like to give a special thank you to the Exhaust After-treatment Performance and Exhaust After-treatment Design departments at Scania CV AB for their continuous support and feedback. We are also especially grateful to Kim Petersson, who ensured the best conditions for our thesis and personal development within the company.

We would like to give a big thanks to our supervisors, Dario Vignali and Jakob Nyström at Scania CV AB, who guided us since the very beginning and throughout the entire thesis. It has been a pleasure to have you as supervisors and your guidance has been crucial to our work.

We would like to also thank our respective universities supervisors and examiners, namely Anton Wiberg, Mehdi Tarkian and Ulf Sellgren, as well as our opponents Rubén Brau, Víctor Pérez and Matas Buzys, providing us constructive feedback. Lastly, we would like to thank the other master thesis workers for the focus group meetings and great company.

Södertälje, June 2019

(7)

(8)

Nomenclature

Abbreviations and Acronyms

Abbreviation Meaning

ASC Ammonia Slip Catalyst

CAD Computer aided design

CF Content Freeze

CFD Computational fluid dynamics

DA Design Automation

DF Design Freeze

DG Design Generation

DOC Diesel Oxidation Catalyst

DOE Design of Experiment

DPF Diesel Particulate Filter

EATS Exhaust After-treatment System

FD Finite Difference

FE Finite Element

FEM Finite Element Method

FV Finite Volume

HCV Heavy Commercial Vehicle

IOSH Institution of Occupational Safety and Health

IL Insertion Loss

KBE Knowledge Based Engineering

KTH Kungliga Tekniska Högskolan

LiU Linköping University

MDO Multidisciplinary Design Optimisation

NM Engine Development

NX Emission Solutions Development

NXD Exhaust System Design

NXDE Exhaust System V8 Engines

NXDX Exhaust System Inline Engines

NXP Exhaust After-treatment Performance NXPS Fluid Dynamics and Acoustics Simulation

PCF Preliminary Content Freeze

RR Result Review

SCR Selective Catalytic Reduction

SHERPA Simultaneous Hybrid Exploration that is Robust, Pro-gressive and Adaptive

SPL Sound Pressure Level

TL Transmission Loss

UN United Nations

UNECE United Nations Economic Commission for Europe

VG Verification Generation

(9)

Latin Symbols

Symbol Description Units

E Element [−] F Flexibility [−] G Operator [−] L Lift [N ] p, A, B Acoustic pressure [P a] P Acoustic power [W ] q Volume velocity m3_/s R Robustness [−] S Design Space [−] T Transfer Matrix [−] U n-dimensional Space [−]

V Dimensionless Mean Design Space [−]

V Volume m3

W Weight [N ]

Greek Symbols

Symbol Description Units

∆ Dimensionless variable range [−]

σ Bending Stress [N · m]

Subscripts and superscripts

Abbreviation Meaning a inlet b outlet C feasible i running number j running number max maximum min minimum ref reference

(10)

(11)

2.9 Discretisation Criteria . . . 19 2.10 Chapter Summary . . . 19 3 Methodology 21 3.1 Data Collection . . . 21 3.1.1 Former Work . . . 21 3.1.2 Interviews . . . 22 3.1.3 Literature Study . . . 23 3.2 Method Development . . . 24 3.2.1 Method Formulation . . . 24 3.2.2 Risk Analysis . . . 24 3.2.3 Model Preparation . . . 24 3.2.4 Model Generation . . . 24 3.2.5 Model Simulation . . . 25 3.2.6 Model Evaluation . . . 25 3.3 Concluding Work . . . 26 3.3.1 Method Finalisation . . . 26 3.3.2 Method Documentation . . . 26 3.4 Chapter Summary . . . 26

(12)

4 Results 28

4.1 Interview Results . . . 28

4.1.1 Product Development Process . . . 28

4.1.2 Design and Simulation departments interaction . . . 29

4.2 Method Application . . . 30

4.2.1 HEEDS - Optimisation Software . . . 31

4.2.2 EXCEL - Connecting Software . . . 32

4.2.3 CATIA V5 - Modelling Software . . . 33

4.2.4 COMSOL Multiphysics - Simulation Software . . . 34

4.3 Proposed Method . . . 35

4.3.1 Common General Method . . . 35

4.3.2 Pre-CAD . . . 36

4.3.3 CAD Model Generation . . . 36

4.3.4 Analysis Setup . . . 37

4.3.5 Optimisation Setup . . . 41

4.3.6 Result Review . . . 42

4.4 Case Study . . . 43

4.4.1 Theses work collaboration . . . 43

4.4.2 First box of a two-box EATS . . . 43

4.4.3 Define Design Scope . . . 44

4.4.4 CAD Model Generation . . . 47

4.4.5 Analysis Setup . . . 52 4.4.6 Optimisation Setup . . . 56 4.4.7 Result Review . . . 58 4.5 Chapter Summary . . . 61 5 Discussion 63 5.1 Assumptions . . . 63 5.2 Simplifications . . . 63

5.3 Delimitations and Limitations . . . 64

5.4 Accuracy Results . . . 64

5.5 Work reflection and potential alterations . . . 65

6 Conclusions 67 6.1 Research Question 1 . . . 67

6.2 Research Question 2 . . . 67

6.5 Method implementation within Scania . . . 69

7 Future Work 71

Appendices 74

A First Appendix - Risk Analysis A0

B Second Appendix - Interview Guide A1

(13)

D Fifth Appendix - Simplified one-box silencer A6

(14)

List of Figures

1 Partial organisation chart of NX at Scania . . . 2

2 Range of Human audibility with normal hearing [8] . . . 6

3 Normal Equal Loudness Contour curves (ISO 226:2003) [8] . . . 7

4 Sound level weighting curves across the frequency range [8] . . . 7

5 View of an Euro 6 engine and the Exhaust After-treatment System (Scania CV AB, 2018) . . . 9

6 Transmission Loss Schematic - Illustration by Munjal (2014) . . . 10

7 Acoustical two-port schematic of a EATS - Illustration based on Åbom (2010) . . . 11

8 Design paradox: the MacLeamy Curve. The relation among freedom of action, product knowledge and modification cost (Illustration by Overbey, 2018) . . . 12

9 Product development at Scania [3] . . . 12

10 KBE system offering different variants of a product [20] . . . 13

11 Benefits of using KBE in the product development process [20] . . . 14

12 Different stages of Morphological (left) and Topological (right) trans-formations [21] . . . 16

13 Design space of the geometrical model [21] . . . 17

14 Thesis Methodology Flow Chart . . . 21

15 Flow overview of the iterative loop . . . 25

16 Method decomposition architecture . . . 30

17 Common Method flowchart proposal for NX . . . 35

18 Define Acoustic Design Scope . . . 36

19 Acoustic CAD Geometry Modelling . . . 37

20 Selections process for getting the denomination and the number for each surface . . . 39

21 Types of denomination COMSOL asks as input . . . 39

22 Acoustic Analysis Setup . . . 41

23 Acoustic Optimisation Setup . . . 42

24 Acoustics Result Review . . . 42

25 Exploded view of the first box of the two-box case study EATS acous-tic model - modelled in CATIA V5 . . . 45

26 Master Skeleton first box of two-box case study product - in CATIA V5 47 27 Active External References using Publications (left) and case study EATS references and copied solids (right)- in CATIA V5 . . . 48

28 Outer Volume (left) and Outer Turn (right) of the case study first box two-box EATS acoustic model - modelled in CATIA V5 . . . 49

29 Substrates surfaces and pole points of the first box of the two-box case study EATS - in CATIA V5 . . . 50

30 Substrate source and destination poles schematic . . . 51

31 Poroacoustics Wool Area indicated in blue color - in COMSOL Mul-tiphysics . . . 53

32 Interior Hard Soundary Wall indicated in blue color - in COMSOL Multiphysics . . . 54

(15)

33 Cylinder where perforated plate condition is applied - in COMSOL Multiphysics . . . 57 34 TL plots for the first box of the two-box case study EATS, from A to

B (top) and B to A (bottom) - in COMSOL Multiphysics . . . 58 35 Correlation matrix for the case study EATS without applied local

optimisation - in HEEDS MDO . . . 59 36 Correlation matrix for the case study EATS with applied local

opti-misation - in HEEDS MDO . . . 60 37 Local optimisation for L_test at the perforated plate . . . 61 38 CAD model geometry of the simplified one-box EATS (without

extru-sions) - in CATIA V5 . . . A6 39 Discretised simplified one-box EATS (without extrusions) - in

COM-SOL Multiphysics . . . A7 40 TL plot of a simplified one-box EATS - in COMSOL Multiphysics . A8

(16)

List of Tables

1 Limit values for pass-by noise of road vehicles (Miloradovic et al., 2017) 8 2 Case study first box components used for the acoustic simulations . . 44 3 Design Scope Summary of case study first box for automatic acoustic

simulations . . . 46 4 Object denomination examples of first box of the two-box case study

EATS . . . 51 5 Default discretisation setting applied to the study case EATS . . . . 54 6 Global optimisation settings . . . 56 7 Local optimisation settings . . . 57 8 Correlation matrix parameter correspondence . . . 60 9 Thesis Project Risk Analysis Table . . . A0 10 List of simplified silencer parts included in the acoustic simulations . A6 11 List of parameters of the EATS utilised in acoustic optimisation loop A8 12 Object denomination of first box of the two-box case study EATS . . A9

(17)

(18)

1 Introduction

This master thesis was performed in collaboration with the departments of Fluid Dynamics and Acoustics Simulation of Exhaust After-treatment System (NXPS) and Exhaust System V8 Engines (NXDE) at Scania CV AB (herein referred to as "Scania"), Linköping University and KTH University.

With the combined knowledge from the students having different backgrounds in aeronautical engineering and machine design, a new method is proposed for con-ceptual design of Exhaust After-treament Systems (EATS) for trucks, buses and industrial applications in order for the Computer-Aided Design (CAD) models to be automatically prepared for acoustic simulations and be optimised based on the obtained simulation results.

1.1 Background

Founded in 1897 in Södertälje, Sweden, Scania is today a global company active in more than 100 countries manufacturing trucks, buses and combustion engines for heavy-duty vehicles, marine, as well as general industrial applications. Following an environmentally friendly development process that aims at minimising waste and improving resources efficiency, the goal at Scania is to offer sustainable modular products having a minimal footprint, while achieving higher efficiency [1]. These cutting edge high quality products are the reasons why Scania is a front-runner within its field. With evolving emissions regulations, Scania has to constantly adapt to stricter environmental laws in order to keep its position on the market. The design of EATS is therefore crucial to comply with the air pollutant and noise emission standards.

The concept development process at Scania follows an interdiscipline iterative loop method, connecting design, simulations and evaluations [2]. Design engineers modify and improve the created CAD models based on the feedback from the different departments involved in the development process. The objective of the company is to enhance this process by being able to connect and combine the work performed at the different departments. Thus, offering superior products while ensuring a reduction in cost [3].

The development of EATS at Scania is performed within the Emissions Solutions Development (NX) organisation. Figure 1 shows the organisation chart of NX and its sub-divisions and departments. Note that all departments from these organisa-tions have not been listed. NXPS belongs to the Exhaust After-treatment Perfor-mance (NXP) organisation, responsible for performing simulations of the different exhaust system products whereas NXDE belongs to the Exhaust System Design (NXD) organisation, in charge of the development and design of exhaust systems for engines. Note that the mission at NXDE is the maintenance and development of after-treatment systems for medium and large silencer. The collaboration between these departments is essential to conceive a silencer that meets all the requirements.

(19)

Figure 1: Partial organisation chart of NX at Scania

The automation of smart parametric CAD models has been proven to be efficient for the personnel at Scania across different departments, allowing fast manipulation of geometrical topologies, and it has improved the performance of EATS to meet the ongoing requirement changes in terms of emissions.

Design evaluation time has been significantly decreased by implementing parametriza-tion and Design Automaparametriza-tion (DA) within various fields at Scania such as turbine houses, inlet ports and more recently in EATS. This reduction in time allows for a deeper exploration of the design space and perform further evaluations [4].

In order for DA to be implemented, the parametric CAD models need to be developed in such a way that the failure rate upon parameter changes remains minimal. The robustness of a model describes its quality by generating a multitude of models having different design parameters to see if it transforms successfully or not. The exploration of the design space consists of determining a set of suitable parameters by a process called Design of Experiments (DOE) [5].

1.2 Purpose and Goals

The variation of models among the different disciplines at Scania creates a long and expensive development process. The aim of various ongoing projects is to have a common model on which the different departments are able to perform simultaneous simulations and evaluations as well as design changes based on their results. The main purpose of this thesis was to simplify the coupling between the design and acoustic simulations divisions.

A previous project aimed at introducing the acoustic simulations into the iterative loop process as other disciplines have been implemented in the past. By automating the acoustic model preparation, the most time-consuming step of the design au-tomation process was greatly reduced. This method was shown to be efficient and possible to incorporate within the current product development process. However, the method requires a series of steps and the use of various software to prepare the files for acoustic simulations [3].

(20)

The objectives of this thesis were therefore to first simplify the preexisting automated method used to prepare the CAD models for acoustic simulations. This new method should also account for transmission loss. Then, after applying the method to a simplified EATS and having validated the results with existing models, the level of complexity was increased. While ensuring the robustness and flexibility of the CAD model, this represented a more realistic design. This method was also made applicable to special cases, such as adding a poro-acoustic material in the muffler. Lastly, general guidelines summarising the steps of the proposed methodology were created in order to implement it in real internal projects at Scania.

1.3 Research Questions

To meet the objectives previously stated and fulfil the requirements from Scania and both universities, the following research questions were posed:

RQ1 - How can a parametric CAD model of an EATS be modelled to auto-mate simulations of acoustic performance?

RQ2- How can the developed method for the simplified case be applied to a complex geometry in the industry?

RQ3 - How efficient is the proposed method in terms of time expenses com-pared to the methodology currently in use?

RQ4 - How well does the proposed method allow for design optimisation in terms of acoustics?

To address the questions formulated above, the thesis outline had to be defined. The work was divided into three stages: data collection, methodology development and concluding work. The first stage consisted of gaining insight on the topic of acoustics: a literature review on acoustics and previous theses performed at Scania were studied, as well as conducting interviews with specialists at the company regarding this topic. Then, with the gained knowledge, a first draft of the new method to be applied was created, assessed and tested on a simple case. Once the method verified, the complexity of the model was increased and a documentation for the proposed method was written. Further details on the conducted methodology can be found in Section 3 of the report.

(21)

1.4 Delimitations and Limitations

This thesis aimed at automating the iterative loops of the concept development of a silencer regarding the acoustic performance. That is, from the preparation of the parametric CAD model to the acoustic simulations. This process belongs to different departments in the company and therefore the main objective was to make this process less time consuming and hence, more effective.

This project is the continuation of a previous thesis carried out during last year at Scania. Aside from CATIA V5 and COMSOL Multiphysics, the thesis involved the use of ANSA software and NASTRAN files for the discretisation phase and preparation of the model for simulations within COMSOL Multiphysics. The aim of this thesis was to develop a new simpler and efficient method that could be compared to the one currently in use. Therefore, CATIA V5 and COMSOL Multiphysics were kept as the main software to be studied.

From an industrial point of view, the main requirement from Scania was that all the performed work involved the compatibility with other software available within the company. In other words, the requirements from other disciplines had to be taken into consideration when developing the methodology.

One limitation regarded the time constraint to perform the work as the master thesis was intended to be completed within a 20-week timeframe. The iterations involved simplifications in the first CAD parametric models. This delimitation was set to ensure correct acoustic simulations of the models. Consequently, complexity added into the design at a later stage limited detail design changes due to the time constraint. To reduce the computational costs, other design simplifications consisted of the omission of mechanical parts not having a significant acoustic impact on the simulations.

(22)

(23)

2 Theoretical Framework

This chapter focuses on the theoretical background required to conduct the thesis work. Note that some information might only be mentioned or referred to the previ-ous thesis on acprevi-oustics simulations. The chapter includes a brief content on acprevi-oustic theory as well as its implementation within the current concept development process at Scania.

2.1 Acoustics

Sound can be described as the audible vibrations propagating through a transmission medium; a fluid or a solid body for instance. Acoustics represents the science that studies production, control, transmission and reception of sound waves (Merriam-Webster’s collegiate dictionary, 2019). The energy transferred by a sound wave can be defined as the pressure difference over a specific period of time, where the sound pressure defines the amplitude of the wave. Pressure is an important property of sound, as some magnitudes can result in pain experienced by humans. The human audibility frequency range varies from 20 to 20 000 [Hz]. It should be noted that the frequency of the sound waves does not determine the level of pain, if any is observed. In fact, according to Assistant Professor Matthias Möbius, pain threshold is almost independent of frequency, as shown in Figure 2 below. [7]

Figure 2: Range of Human audibility with normal hearing [8]

Noise can have health effects on the human body that can be both physical and psychological when regularly exposed to consistent elevated sound levels. According to the Institution of Occupational Health and Safety (IOSH), the most well-known effect of noise is hearing impairment but can also lead, amongst others, to cardio-vascular effects, irritation and annoyance, stress and other effects on psycho-social well-being. Thereby, the study of acoustics is necessary to be included into the development process of a product in order to minimise those effects. [9]

(24)

The sound level represents the relative intensity of sound and is expressed by the dimensionless unit [dB]. The human ear is particularly sensitive in the frequency range of 300 to 6000 [Hz] but has a lower sensitivity at lower and high frequencies, as seen in Figure 3. The contour curves can be inverted at a particular intensity level, which give the relative frequency response plot of the human ear. Each contour is defined as a "phon" level, which corresponds to a unit for loudness equal to the sound pressure level in decibels and where a phon represents a 10 [dB] step and is perceived approximately as twice as loud as the previous level [8]. The loudness levels in phons correspond to the sound pressure levels (SPL) at 1000 [Hz] [10].

Figure 3: Normal Equal Loudness Contour curves (ISO 226:2003) [8]

Frequency weighting filters are used in sound level meters to attenuate the sound signals. The different A, B, C and D weighting curves can be observed in Figure 4. The A weighting curve, also denoted dB(A) curve, is the most commonly used filter for background noise and represents a baseline for noise emissions regulations.

(25)

2.2 Acoustics of heavy commercial vehicles

As a Heavy Commercial Vehicle (HCV) European manufacturer present all over the globe, Scania must comply with the emission regulations from the United Nations (UN). In Europe, these are written by the European regional commission: the United Nations Economic Commission for Europe (UNECE). The European Commission classifies vehicles on emission standards, where category M describes vehicles having at least four wheels and used for the carriage of passengers, and category N for power-driven vehicles having at least four wheels and used for the carriage of goods [3] [11]. Each category is divided in sub-categories based on the mass and/or power of the vehicle. Table 1 below describes the noise limits for vehicles used for the carriage of goods, the category of interest in this thesis. Note that Phase 1 to Phase 3 represent the limit values applicable for new vehicles types from July 1st of 2016, 2022 and 2026 respectively.

Table 1: Limit values for pass-by noise of road vehicles (Miloradovic et al., 2017)

Vehicle Category

Vehicle used for the carriage of goods Phase 1 [dB(A)] Phase 2 [dB(A)] Phase 3 [dB(A)] N1 mass ≤ 2500 [kg] 72 72 69 2500 [kg] < mass ≤ 3500 [kg] 74 73 71 N2 rated engine power ≤ 135 [kW ]_{rated engine power > 135 [kW ]} 77₇₈ 75₇₆ 74₇₅

N3

rated engine power ≤ 150 [kW ] 79 77 76 150 [kW ] < rated engine power ₈₁ ₇₉ ₇₇

≤250 [kW ]

rated engine power > 250 [kW ] 82 81 79

The acoustic performance of HCV involves different complex systems where four major noise sources can be defined: the engine, the air-intake system, the exhaust system and the tyres. Standard methods are implemented by the UNECE in Eu-rope to test the vehicle pass-by noise emissions. The original method, denominated method A, demonstrated that the power train was the main contributor to this noise pollution. The test consisted of measuring the noise emitted by the vehicle at wide open throttle (WOT), i.e. at maximum acceleration. The microphones are placed at the middle of the test track, therefore the air-intake is dominant at the beginning of the test, followed by the power train at the middle of the test and the exhaust noise being dominant at the end of the track, all three masking tyre noise [13]. This method can also be referred as the two or multiple microphone method. However, urban traffic noise is not characterised by full accelerating vehicles and therefore tyre noise has a greater contribution in urban centres. According to method B, the new approved but not yet implemented method performed at lower speeds, tyre noise is dominant over power train noise for speeds between 35 and 80 [km/h] [14].

In order to keep its position in the market, Scania must adapt to the ever changing regulations and hence keep reducing the noise emissions from the EATS, thereby comply with the defined noise regulations.

(26)

2.3 Exhaust After-treatment System

Figure 5: View of an Euro 6 engine and the Exhaust After-treatment System (Scania CV AB, 2018)

In Figure 5 a schematic view of an Euro 6 engine and the exhaust after-treatment system (EATS) is shown. The main purpose of the whole system is to fulfil the emission regulations [15].

It is known that the emissions requirements not only depend on the EATS but also on the fuel composition used in each vehicle. In general, there are two major issues in regards to the combustion process in the engine: the release of carbon dioxide (CO2) and oxides of nitrogen (NOx), both harmful for the environment and health.

Moreover, the incomplete combustion of diesel may also produce carbon monoxide (CO) and hydrocarbons (HC), therefore EATS are needed to reduce the amount of dangerous chemicals and thereby fulfil the requirements. [3] [15]

The EATS contains five major components, named substrates. The first component in the system is the Diesel Oxidation Catalyst (DOC), which has the main function of oxidising carbon monoxide and hydrocarbons. The second system that comes into play is the Diesel Particulate Filter (DPF), a filter made of porous material which hinders the particles to passing through it. The following step in the system is the addition of urea. This is decomposed into ammonia and isocyanic acid, which in turn reduces the concentration of NOx and creates water, CO2 and nitrogen in the

Selective Catalytic Reduction (SCR) by means of a chemical reaction. Incomplete chemical reactions might lead to remains of ammonia in the exhaust gases, which are retained by the Ammonia Slip Catalyst (ASC) before the gases are released into the environment. [3] [15]

A common problem with the DPF is the build-up of soot in the filter. This can be partially fixed by oxidising this soot into ash, which reduces the build-up and thereby the back-pressure. Nevertheless, this ash may also build-up and increase the back-pressure with time and hence a regular change of the filter is necessary. The urea may also cause build-up of soot due to an incorrect vaporisation caused by the decrease of temperature in the SCR. Note that this built-up increases the back-pressure and therefore decreases the fuel efficiency. [15]

(27)

2.4 Transmission Loss

The performance of a EATS or silencer can be evaluated by measuring the trans-mission loss and insertion loss. The transtrans-mission loss (TL) represents the difference between the incident acoustic energy into a system and the transmitted acoustic energy into an anechoic – free from echoes and reverberations – environment after the system (Figure 6). In other words, it describes the sound intensity or the power carried by sound waves being emitted from the silencer to an anechoic termination. [16]

Figure 6: Transmission Loss Schematic - Illustration by Munjal (2014)

In Figure 6, the acoustic pressure An of the incident wave from the inlet reaches

the system, represented by a filter, or resonator in this example. This larger volume is used in order for the propagating sound waves to hit the walls of the filter and be reflected, which in turn allows the sound waves to cancel each other. A reduced acoustic pressure A1 of the transmitted wave reaches the anechoic termination while

a reflected acoustic pressure Bn travels back towards the inlet. The length of the

system is determined in order to cancel certain ranges of sound frequencies generated in the EATS.

There are several ways to determine TL but at NXPS the acoustic power on the surface at the inlet and outlet are measured and inputted into Equation 1. Extrusions at both ends of the silencer are created to simulate the multiple microphone method. The values obtained are then compared to experimental measurements performed at the same location. It should also be noted that the simulation is also performed twice, with sound waves starting at the inlet and outlet, respectively, to verify the obtained results. The model setup is further explained in Section 4.

T L = 10 · log₁₀ Pn P1

!

, (1)

Where P is the acoustic power at the given surface and the subscripts refer to the points shown in Figure 6.

(28)

2.4.1 Multi-port concept

Modelling sound generation and transmission of sound waves in ducts is performed utilising acoustical multi-ports. According to Åbom (2010), a multi-port can be defined as a system where a casual relation exists between a set of input and output variables (x and y, respectively). Note that it is normally assumed the input and output variables observe the same dimensions and their relation is mathematically described as the existence of an operator (G) such that:

y = G [x]. (2)

This approach can also be referred to a "black-box" model, as the internal workings, operations and implementations are applied without full knowledge of the inner prop-erties of the system. In acoustics, the input and output variables studied are often pressure, force and displacement. [17]

2.4.2 Transfer Matrix

A silencer can be assumed to be a two-port model coupling acoustic pressure (p) and volume velocity (q) as both the input and output variables. The coupling results from the conservation of mass and momentum over a small control volume surrounding the interface, assuming continuity of pressure and volume velocity, and no fluid flow. The coupling leads to the transfer matrix (T ), which is a formulation employed to describe the relationship between inlet acoustic pressure and volume velocity with the outlet acoustic pressure and volume velocity, as seen in the equation below: [17]

" ˆ pa ˆ qa # = T " ˆ pb ˆ qb # = " T11 T12 T12 T12 # " ˆ pb ˆ qb # , (3)

Where the circumflex accent represents the Fourier transform of the variable. A silencer can thereby be represented as a acoustic two-port schematic, as seen in Figure 7 below. Note that a and b represent the inlet and the outlet of the EATS, respectively.

Figure 7: Acoustical two-port schematic of a EATS - Illustration based on Åbom (2010)

(29)

2.5 Product development

The design of a product follows a series of phases from creation to manufacturing: preliminary design, conceptual development design and detailed design. Although the denomination of these phases and their sub-parts might differ according to the company and the industry, the same structure can be observed. This is the engineer-ing design process [18]. The preliminary phase consists of establishengineer-ing the require-ments of a new product based on the needs of the customer, which will define the design space of the product. In the concept development phase, different combina-tions of parameters are tested based on the given requirements. The proposed models go through a concept screening phase where simulations and evaluations regarding different disciplines are conducted. Finally, the detailed design phase involves the specification of the different components and final parameters of the product [3] [19].

Figure 8: Design paradox: the MacLeamy Curve. The relation among freedom of action, product knowledge and modification cost (Illustration by Overbey, 2018)

The latter described phases can be related to the design paradox as seen in Figure 8, where the possibility of introducing design changes reduces as these modifications become costly and, on the other hand, the knowledge on the product increases as the project evolves.

(30)

At Scania, the current product development process follows three phases (Figure 9). The first phase, yellow arrow or concept development, emphasises on solving current problems and integrating new technological feasible solutions into the current business model. This phase is based on the need to improve existing designs or new demands arising such as new design requirements. Green arrow or product introduction is the following phase and deals with the refinement of the concept until the product manufacturing and launch. Note that both yellow and green arrow phases are based on iterative loops from CAD modelling to simulation and evaluation of results. Once the product is well established into the market, the red arrow or product follow-up phase begins, in which solutions are provided for deviations of the product. [1] [3]

2.6 Design Automation

2.6.1 Knowledge Based Engineering

Knowledge Based Engineering (KBE) can be defined as integrating technology knowl-edge and engineering design processes within the concept development process of a product [20]. This KBE system can then use the gathered information for the de-velopment of similar models with a new set of input specifications, as seen in Figure 10.

Figure 10: KBE system offering different variants of a product [20]

The benefits of KBE include the reduction of time expenses in regards to routine tasks and enhancing the proportion of creative tasks. According to Devaraja Holla V (2018), an organisation can observe a reduction of 20 to 40 % in cycle time as well as effort in the development process. An outcome of KBE is Design Automation (DA), where KBE methods are employed through modern CAD systems allowing the flexibility of geometric models [21]. The morphology and topology of high level CAD models can be modified through the use of rules, relations and facts. The influence of KBE on product development can be seen in Figure 11.

(31)

Figure 11: Benefits of using KBE in the product development process [20]

The great qualities of KBE come however at a cost and some of the critical factors include having a structured developed methodology and literature, as well as good management support [3] [20].

2.6.2 Multidisciplinary Design Optimisation

Optimisation can be defined as "the act, process, or methodology of making some-thing (such as a design, system, or decision) as fully perfect, functional, or effective as possible" (Merriam-Webster’s collegiate dictionary, 2019). In other words, it is the exploration of the design space and search to determine design parameters to iden-tify solutions, optimal designs, fulfilling the requirements imposed and respecting the pre-defined constraints [3].

Multidisciplinary optimisation (MDO) describes a complex optimisation connecting different disciplines with one of more objective functions and/or constraints. An example in aeronautical industry could be to have two functions f and g, representing the weight (Wwing) and lift (Lwing) of a wing, respectively. The wing is subjected to

bending stress (σ) and must contain a minimum fuel volume (Vf uel). The objectives

would therefore be to minimise the weight and maximise the lift, while ensuring minimal/maximal values requirements, or in other words, comply with the given constraints. In this case, the fuel volume and bending stress must be less than the wing volume (Vwing) and the stress limit (σlim), respectively. Equation 4 summarises

(32)

max

x f (x) where f (x) = Lwing min

x g(x) where g(x) = Wwing (4)

s.t.

σ < σlim Vwing> Vf uel

with respect to: x.

Developing detailed models can be difficult when knowledge about the product is limited. The collaboration among disciplines leads to more expert knowledge. If this occurs during the earlier stages of the design process, time expenses can be reduced by decreasing the cycle time in the development of complex systems. Note that the word collaboration is defined as the work performed jointly among experts on different disciplines (Merriam-Webster’s collegiate dictionary, 2019). Collaborative MDO is considered to be a new approach to deal with complex multidisciplinary systems in the design phase of a product, which integrates high fidelity design tools in its process [24]. At Scania, joint efforts can be observed among the design, simulation and test engineers for instance, where experts are involved earlier in the process to better conceptualise the design, allowing for its optimisation.

2.7 Simulation Fidelity

Simulation fidelity can be defined as the extent or degree to which the simulation can replicate the actual environment. In other words, it refers to the level of "realness" of the simulation. An industrial example in the aviation industry can be flight training and simulation devices where the simulation fidelity represents the degree of realism in terms of looks, sounds, responds and manoeuvres compared to a real aircraft. In this thesis, the simulation fidelity regards the accuracy and precision of acoustic performance of an EATS determined by the employed simulation software. [25] It is natural to assume that higher fidelity simulation tools will grasp better the physics involved and return valid results. It should however be noted that the increase in fidelity implies gains in complexity and hence an overall increase in computational costs. The level of fidelity is also related to the knowledge of the physics involved in the simulation. Lower fidelity simulation might be preferred at earlier stages in product development process despite uncertainties regarding the accuracy of the results. On the other hand, high fidelity simulations are generally adopted towards the end of the development cycle of a product to be able to correctly verify the obtained results. [25]

(33)

2.8 CAD Modelling

CAD systems have been part of the design industry for more than 60 years and have proven to be essential engineering tools. The first parametric modelling techniques in the mid-80’s marked the beginning of the exponential use and possible outcomes by generating CAD models [1]. The aircraft Boeing 777 for instance, whose maiden flight was on the 12th _{of June 1994, is a great example of the extensive use of CAD}

models. In fact, the Boeing 777 was the first aircraft to have been entirely modelled in a CAD system: CATIA [26].

2.8.1 Smart Parametric Modelling

In order to increase the profitability of CAD systems, KBE methods are applied to create geometrical DA. This parametrisation of models allows for a flexible geome-try based on defined parameters. The transformation can be categorised as either morphological or topological.

The term morphology can be defined as the shape or form of an object. A mor-phological change therefore implies the modification of the geometry of an object by changing its dimensions. According to Amadori et al. (2012), morphological models can be identified as four different stages (Figure 12, left).

The first stage represents an object with given dimensions that cannot be modified, denoted Fixed Object. Then, the Parametrised Object stage allows for dimension changes by modifying the parameters of the object. The parametrisation of the object can also be based on mathematical relations (Mathematic Based Relation stage). Finally, the object geometry can follow a set of rules based on the input parameters: this is the Script Based Relation stage.

Figure 12: Different stages of Morphological (left) and Topological (right) transforma-tions [21]

(34)

The term topology refers to the location of features or objects in a geometrical model. A topological transformation involves therefore the addition, deletion or replacement of a geometrical feature or object, an "instance", in the geometrical model. Similarly to the morphological transformation stages, topological transformations observe four different stages (Figure 12, right). The first two stages are not context-dependent and represent the copy and paste, or "instantiation", of functions on various objects. The Automatic Instantiation stage differentiates from the Manual Instantiation stage by having the number of instances modified parametrically. The Generic Manual In-stantiationstage involves the instantiation of features or objects in different contexts manually, and the last stage, Generic Automatic Instantiation allows for this process to be performed automatically [21].

2.8.2 Model Flexibility and Robustness

When designing a parametric CAD model, two major requirements need to be taken into account: flexibility and robustness. According to Amadori et al. (2012), flexibil-ity "refers to the abilflexibil-ity to represent a wide range of different product configurations, arrangement and sizes". In other words, it describes the possibility of a CAD model to modify and adapt its geometry based on given input parameters. Robustness on the other hand can be defined as the quality of the CAD model when evaluating the errors or instability issues arising from the changing geometry. Therefore, the robustness is inversely proportional to the number of errors in the model.

Figure 13: Design space of the geometrical model [21]

To ensure that these quality properties are satisfied, they require to be measured quantitatively. Amadori et al. (2012) proposed a method that compares the number of successful model updates with the total number of updates in the geometry in relation with a specifically constrained design space based on the input parameters (Figure 13).

(35)

Following the proposed method, an overall n-dimensional space U comprises all the theoretically possible configurations where the input parameters can be modified over unlimited ranges. By assigning minimum and maximum values to these parameters, this design space can be reduced (S). Note that this space contains both feasible and unfeasible design. The design space comprising the totality of feasible designs (SC) is a union of sub-spaces (SCj), a disjointed union, in which only

morphologi-cal transformations can take place as topologimorphologi-cal transformations do not guarantee continuity. [3][21]

Considering one sub-space SCj and assuming that each variable xihas a defined

max-imum and minmax-imum values (ximax and ximin, respectively), a dimensionless variable

range ∆i can be obtained as follows:

∆i =

ximax− ximin

xiref

, (5)

Where ∆i is applicable for each parameter and xiref represents a baseline value

for the variable xi that must not be equal to zero. Note that the same baseline

value needs to be used among variants in order to allow for quantitative comparison. The dimensionless variable ranges allow then the obtention of a dimensionless mean design space [21]. VSCj = n Y i=1 ∆i. (6)

As previously mentioned, the robustness of a model can be measured quantitatively by comparing the number of successful updates with the total number of updates in the geometry. This can be translated into the following equation [21]:

RCj = 1 −

NF ailures

NU pdates

, (7)

Where NF ailuresrepresents the number of models whose parameters input values

lead to errors or instabilities in the geometry, and NU pdate defines the total number

of iterations. Finally, the flexibility of the model can be evaluated by factorising the dimensionless mean design space with its robustness [21]:

FCj = RCj· VSCj. (8)

Flexibility of a model is dependent on its mean design space and robustness. As the design space increases, so does the flexibility, the model is more prone to observe robustness issues. Inversely, high robustness in a smaller design space leads to a decreased flexibility [21].

(36)

2.9 Discretisation Criteria

Discretisation represents the approximation of continuous functions, models, vari-ables and equations into "discrete" quantities. This necessary step within the sim-ulation process to be able to evaluate functions and integrate governing equations at each element. Different methods exist to discretise geometrical object into sets of finite elements (FE), volumes (FV) or differences (FD) in order to perform simula-tions. Their differences include the ease of implementation and the type of equations defining these finite quantities. [27]

Acoustic simulations involve generally a set of FE, or a FE "mesh", to be evaluated. The mesh of a geometrical object has several properties affecting the accuracy of the simulation that can be modified. First, element size is an important parameter as it affects the mesh density. A denser mesh implies a higher fidelity and accuracy but also severely increases computational time. The mesh should therefore be denser in the zones of interest, and allow for a larger element size in the remaining geometry. Then, there are different mesh types that should be used accordingly and adapt to the shape of the geometrical object to avoid skewed elements that can reduce the accuracy of the results and destabilise the solution. [28]

2.10 Chapter Summary

This chapter has provided a review of the theoretical background for this thesis. First, acoustic theory about the human perception of sound was outlined. Then, the application of acoustic regulations on HCVs and the method employed measure pass-by noise emissions were described. The different components or substrates of a EATS were named and portrayed, and acoustic performance of a silencer was spec-ified by defining the transmission loss. Product development and its process within Scania were then detailed and illustrated. DA was next defined by introducing KBE and CDMO concepts, and smart CAD modelling was presented by differentiating morphological and topological transformations and introducing flexibility and ro-bustness. Finally, the accuracy of the predicted results was discussed by taking into consideration simulation fidelity and discretisation criteria and their effect on simulations.

(37)

(38)

3 Methodology

The thesis methodology structure is summarised in the flow chart in Figure 14 below and consists of three main parts or steps: data collection, method development, being the longest one, and finally concluding work.

Figure 14: Thesis Methodology Flow Chart

3.1 Data Collection

The first phase consists of gathering technical information and gain knowledge on the given problem by conducting a literature study. Furthermore, the research performed through former master theses on acoustic simulations has been studied and used as a baseline for this thesis to formulate the objectives and expected outcomes.

3.1.1 Former Work

Following a similar methodology flow, the thesis conducted by Hellberg and Nyström involved gathering data, developing and finalising a method in the form of guide-lines to perform DOE studies of acoustic simulations of a EATS. Achieving a 98.75% robustness for 2000 tested designs allowing morphological transformations, the mod-els were automatically discretised within ANSA using a Python code. The use of a LiveLinkTM _{with the software MATLAB was used to run the simulations within}

COMSOL Multiphysics. The pre-process of CAD models for acoustic performance was automated and the proposed method was verified and validated to be able to be implemented into the product development process.

(39)

The thesis work underlined that the most time-consuming part for acoustic simula-tions is the pre-processing step for each design iteration. EATS are complex products and any minor design change might affect the simulation setup. Therefore, the time spent in running simulations, and most importantly, running the correct simulations on feasible designs is greatly reduced. The conducted work proposed the introduc-tion of the software ANSA, often used within NXPS, to keep track of the surfaces and other objects in the geometry. Thus, by recognising their correct denomination, the correct settings in the pre-processing step were automated. Defining the objects in the geometry within the modelling software based on their corresponding applied simulation settings was therefore crucial for the automation to be implemented. The connectivity among the involved software allowed the development of seven DOE studies. [3]

To achieve this automation, it was necessary to create a parametric acoustic model from the original geometry. The time estimation to generate such model was 35 [h]. Then, the pre-processing step and simulation setup were estimated to take around 35 [h] each. The design changes to be performed between DOE studies were also estimated to take about 35 [h]. Finally, the new pre-processing step and simulation setup were assumed to take the same amount of time as for the first DOE study. Overall, the performance of this automation offered a reduction of 10% compared to the general manual method applied at Scania.[3]

The conclusions from the thesis were that the implementation of the proposed method within NXD could greatly reduce the pre-processing step in the concept development phase. By automating the repetitive and time-consuming actions, the time between each iteration could be decreased. Additionally, the different devel-opment loops across the studied disciplines could be linked and favour informed decisions to reduce the cost of the new products development. [3]

3.1.2 Interviews

In order to get an insight on the involvement of the different departments in the design process of an EATS at Scania, a useful tool is to conduct interviews to gather information about expert knowledge, practices and experiences [1].

Several interviews were conducted with engineers from both the Exhaust System Design departments (NXDE, NXDX) and the Acoustics Simulations department (NXPS). In total, six people were interviewed on the following topics: DA, collabo-ration between departments, MDO, acoustics simulations, parametric CAD design, and CAD pre-processing for simulation in various disciplines. The conducted inter-views were of the type semi-structured, which can be defined as interinter-views following a guide with questions and topics to be covered. The aim of this type of interview is to collect information deeply into a specific topic [1]. Nonetheless, the questions are standardised and hence leading to a more conversational interview where the interviewer can modify or omit questions based on the conversation [29].

The interview guide followed for the interviews conducted at Scania, as well as a summary of the gathered information are documented in Appendices B and C, re-spectively.

(40)

3.1.3 Literature Study

To have a better understanding on the background of the thesis, a literature study has been conducted with the focus on product development and management within the industry and in particular at Scania, as well as technical subjects. The gathered information from the literature review that was necessary to conduct this thesis is documented in Chapter 2. The latter subjects involved the following general topics:

• Acoustics: physics, modelling, pre-processing and simulations • Emissions regulations

• Smart and high level CAD models • Parametric CAD Modeling

• KBE and DA • MDO

Note that throughout the duration of the thesis, additional literature was added and the gathered information reviewed. The literature has been primarily collected from e-resources from both universities libraries, institutes, and internal reports from Scania. The literature is also being shared among the other theses workers at Scania within the same department during weekly focus meetings regarding MDO. Scania has been conducting master theses on the latter topic since 2015, which allowed the access to numerous references on the studied topics.

(41)

3.2 Method Development

3.2.1 Method Formulation

The method development phase is the main body of the thesis in which the simu-lations and the optimisation varying the parameters are performed in an iterative loop process. This process starts by formulating a new method to be implemented. In other words, defining the processes and tools to be employed to reach the set objectives. This step was performed by analysing and evaluating the previously pro-posed method, as well as determining how it can be modified and improved. The new features of the software involved were investigated and tested to be integrated within a common framework.

3.2.2 Risk Analysis

Once formulated, the feasibility and applicability of the method to a case study need to be taken into consideration. The Risk Analysis, which can be found in Appendix A, is a crucial step as it ensures to have alternative scenarios to reach the objectives. The table summarises the likelihood of events as well as their severity affecting the work, and counter measures to mitigate their consequences.

3.2.3 Model Preparation

The following step is the pre-CAD phase, which consists of gathering all the neces-sary information of the product to be modelled, as well as the tools being utilised. The thesis aims at studying the first box of the two-box case study silencer in terms of acoustics, where the same product is used in related theses conducted at NXD (Sub-section 4.4.1). More details on the components and functionalities of the si-lencer can be found in Section 2.3. First, a simplified CAD is generated in order to complete the desired loop and once satisfying results are obtained, the CAD com-plexity is increased in order to have a more realistic model. The aim is then to be able to apply this methodology to an existing product created at Scania. The mod-elling and acoustic simulation software used at Scania are CATIA V5 and COMSOL Multiphysics, respectively. Therefore the same software were used for this thesis. This phase emphasises in determining the different parameters to be studied as well as the specific features of the EATS necessary to be modelled.

3.2.4 Model Generation

The CAD model requires to be parameterised in such way so that different designs can be simulated. Thus, allowing morphological changes in the design to perform model optimisation. Note however that the design and simulation of silencers at Scania is performed at different departments, hence manually. To automate this process, an optimisation software needs to be incorporated within the framework to define the lower and upper bounds of input parameters. These parameters shall automatically modify the geometry of the model by employing design procedures, rules and relations as described in section 2.8. The CAD models require also sufficient levels of robustness and flexibility.

(42)

3.2.5 Model Simulation

Similarly to the Model Generation step, the acoustic simulations should be able to be performed automatically for all imported CAD geometries. Format compatibility between the design and simulation software is fundamental in order not to loose details when importing the geometry. Also, as the geometry is flexible, it is necessary to keep track of the surfaces of the model in order to be able to apply the correct boundary conditions and desired descretisation specifications automatically.

3.2.6 Model Evaluation

Finally, the last phase regards the evaluation of the model, where the design is validated based on the results obtained from the acoustic simulation. The acoustic performance results should then give new ranges for the different design parameters the model is being optimised for. This step shall be performed within the introduced optimisation software. Note that simulation software often observe optimisation features that could also be implemented. However, as the simulation is run for each iteration, the in-built optimisation should be performed at local level. For instance, determining the adequate amount of poroacoustic material.

The process described above allows the formulation and documentation of a method to automatically simulate the acoustic performance of a parametric CAD model. As more knowledge and experience is acquired throughout the thesis time frame, more complexity in the CAD model is added, which requires iterating the loop multiple times to validate the results and deem the method efficient. A summary flow of the major steps of the method can be seen in Figure 15 below.

(43)

3.3 Concluding Work

3.3.1 Method Finalisation

The proposed method connecting the different software and allowing for an automatic acoustic performance simulation of parametric CAD models required to be functional and validated. The obtained results with the new method were compared to the one currently in use. The efficiency in terms of computational costs and accuracy of the results was assessed. Finally, the morphological optimisation on an EATS based on its acoustic performance was evaluated.

3.3.2 Method Documentation

The developed method was based on the gathered information obtained through the literature study and the conducted interview combined with the case study for this thesis. The collection of data allowed the gain of knowledge not only in acoustic simulations but also on the development process at Scania. The study case aimed at automating the CAD model preparation process for acoustic performance and perform the required simulations. The combination lead to creating a specific method between the design and simulation engineers and generalise a method that is not only applicable to the study case. The guidelines to this automation procedure is presented in Chapter 4.

3.4 Chapter Summary

This chapter described the followed methodology of this thesis defined in three steps: data collection, method development and concluding work. First, the necessary infor-mation to conduct the project was gathered through former theses work, interviews with Scania employees and a thorough literature study. Then, the method devel-opment listing the software and steps necessary to perform the optimisation was presented. The collaboration with other thesis workers and the common assembly product was then referred. Finally, the method validation comprising the tools to answer the research questions, as well as the documentation of the proposed method were listed.

(44)

(45)

4 Results

This chapter is divided in three sections: the Interview Summary presents the gath-ered information on the product development process and more specifically on the followed design and simulation procedures as well as the interaction between the respective departments. The Proposed Method is presented in the second section. Finally, the case study section shows the implementation and application of the developed method on an EATS.

4.1 Interview Results

The results from the conducted interviews are divided into several areas correspond-ing to the different topics discussed with the interviewees. Note that only the current development process at Scania, as well as the interaction between the design and simulation departments are presented in this section. Further information on the remaining discussed topics is documented in Appendix C.

4.1.1 Product Development Process

The product development process at Scania is divided into three main stages; namely yellow, green and red arrows. The first stage is the concept development phase that consist of a series of milestones, from the analysis of a need on the market to the approval to move on the product development phase, the green arrow stage. The yellow arrow therefore starts with the planning of a new project. Once approved, the correct budget is assigned and based on the new requirements, a set of concepts are created. A screening of the different concepts is then performed, where the risks and resources required are evaluated.

The concept development phase is performed through multiple iterative loops where the objective is to eliminate or minimise at best the risks regarding the different disciplines. Both morphological and topological changes in the design are investi-gated in this phase. The yellow arrow stage can last for a few years in total and finishes when a concept is deemed mature and promising. The product development phase represents the evolution of a concept from its selection to its detailed design and manufacturability readiness. The concept is iteratively modelled, simulated and evaluated for the different disciplines. Design changes in this stage are usually only morphological.

When a concept is deemed good enough, the design is "frozen", or in other words, a development generation (DG) freeze is performed. The DG is then virtually and physically tested and evaluated against acceptance criteria. A Verification Genera-tion (VG) freeze is then conducted and the concept undergoes thorough verificaGenera-tion testing. The VG is then certified and prepared for production. Once on the market, the product enters the last phase, the red arrow or follow-up stage. This phase em-phasises on finding solutions to deviations of the product. Note that as this thesis regards the concept development of a EATS, the red arrow stage was not further explored.

(46)

4.1.2 Design and Simulation departments interaction

The three stages mentioned earlier are applied to the different organisations within Scania but work processes and methods may vary across the departments working with product development. At NX, the process is as follows: the design engineer generates a model that is modified for the specific simulation it must be tested for and sends it to the simulation engineer. In the case of acoustic simulations, the assembled product is saved as a surface part. The latter person performs further modifications and simplifications in the geometry and discretises the model.

Once the model is simulated and evaluated, the required design changes are commu-nicated to the design engineer through a compiled result report. It should be noted that the "back-and-forth" actions between the two engineers varies in time based on the persons and type of product involved. Indeed, design modifications and discreti-sation procedure are relatively less time consuming for experienced engineers. As design and simulation engineers belong to different departments, the available resources can differ. In fact, design engineers do not have access to simulation tools and inversely for simulation engineers. Thereby, design engineers are unable for instance to perform quick tests in order to assess the quality of the model before it is sent to the simulation engineer for a thorough simulation.

Section 2.2 describes the four major contributors to the pass-by noise emitted by HCVs. The regulations apply however to the entire vehicle and hence the attributed vehicle noise pollution is divided among the different concerned departments. The desired performance is defined by development targets, or acceptance criteria that the compromise among the disciplines aims at fulfilling.

(47)

4.2 Method Application

In Section 3.2 an overview of method development is presented. The method appli-cation and more specifically the method finalisation belongs to the concluding work part since the process requires to be repeated multiple times until the method is deemed robust enough to be validated. This section emphasises on detailing the procedure performed in each phase and hence the different steps followed using the described software.

Figure 16: Method decomposition architecture

In Figure 16 the architecture of the method is presented. The simulation program HEEDS, developed by SIEMENS, provides an excellent way to automate processes, and is the main software in charge of running the iterative loops from the design phase to the optimisation phase, as well as determining the correlation of the design parameters with the acoustic results. A Microsoft Excel analysis is placed within HEEDS workflow and the worksheet parameters are tagged to the ones in the opti-misation software. The Excel node has two main functions: modify the geometry of the CAD model in CATIA V5 via macros scripted in the VBA Editor and trigger COMSOL Multiphysics via an Excel LiveLinkTM_{. This is to ensure the automatic}

start up of the application since the software HEEDS (version 2018.10) cannot be connected with COMSOL Multiphysics.

The following step is the import of the CAD model into COMSOL Multiphysics. This is made possible by having the Design Module licence and Import for CATIA V5 add on. These functionalities are necessary to keep track of the surfaces in the geometry and therefore apply the correct required mesh type based on the surface denomination (surface ID). Once the simulation is performed, the results will provide feedback to the Excel node (within HEEDS) and allow for the next iteration to start.

Development of Acoustic Simulations using Parametric CAD Models in COMSOL