Design improvements of an automotive air intake system

(1)

Design improvements of an automotive air intake system

Cite as: AIP Conference Proceedings 2233, 020008 (2020); https://doi.org/10.1063/5.0001440 Published Online: 05 May 2020

Han-Bo Ronald Gan, Noor Zafirah Abu Bakar, Nur Fadzilah Shaikh Dawood, and Muhammad Adam Rosli

ARTICLES YOU MAY BE INTERESTED IN

Effect of different fibre treatment of oil palm fibre reinforced polyester composite AIP Conference Proceedings 2233, 020002 (2020); https://doi.org/10.1063/5.0003796 Circularly polarized cylindrical dielectric resonator antenna for 5G applications

AIP Conference Proceedings 2233, 030002 (2020); https://doi.org/10.1063/5.0002355 Effect of titanium metal addition on the properties of zirconia ceramics

AIP Conference Proceedings 2233, 020001 (2020); https://doi.org/10.1063/5.0001504

(2)

Design Improvements of an Automotive Air Intake System

Han-Bo Ronald Gan

¹

, Noor Zafirah Abu Bakar

^1,a)

, Nur Fadzilah Shaikh Dawood

²

, Muhammad Adam Rosli

²

1School of Engineering, Taylor’s University Lakeside Campus, No.1 Jalan Taylor’s, 47500 Subang Jaya, Selangor D.E, Malaysia.

2Perusahaan Otomobil Nasional, PROTON Holdings, HICOM Industrial Estate, Batu 3, 40918 Shah Alam, Selangor D.E, Malaysia.

a) Corresponding author: [email protected]

Abstract. This project was conducted in conjunction with Proton R&D department to study and improve the air intake system (AIS) for the current Proton Iriz vehicle. The AIS comprises of few parts: snorkel, airbox, air filter, and zip tube.

The function of an AIS is to provide a continuous flow of clean air to the engine for combustion. The pressure drop will occur within the airbox due to turbulent flow caused by the geometry of the airbox and air filter which would lead to a decrease in the performance of the engine. Simulations were conducted based on the exact AIS model in ANSYS Fluent from the point where the air enters the airbox to the point where it leaves the airbox to the engine while any pressure loss within this flow was observed and analysed. The CFD simulation model was validated with the experimental data provided by Proton. The percentage of the difference between the experimental data and the simulation results were within an acceptable range, hence the simulation model was validated. Improvement designs were conceived after analysing and studying the fluid characteristics that are occurring inside the AIS. To solve the pressure drop problem inside the airbox, the hypothesis made was that the geometry of the AIS is the root cause of pressure drop and it occurs where the flow begins to transform from laminar to turbulent. Three design improvements have been made to the existing AIS design and were simulated. The simulation results show an improvement of about four percent for the first design, three percent for the second design and two percent for the third design improvement.

INTRODUCTION

This project was conducted in conjunction with Proton R&D department to study and improve the air intake system for the current Proton Iriz vehicle. The function of an air intake system (AIS) is to provide a continuous flow of clean air to the engine for combustion [1]. The AIS is a crucial component to allow the engine to operate as it collects air and directs it to the air intake manifold that directs air to the pistons of the engine where the combustion of air-fuel mixture occurs. The AIS comprises of snorkel, airbox, air filter, resonator and zip tube as shown in FIGURE 1 [1].

The air intake system’s snorkel is usually located away from the engine to avoid drawing in hot air from the surrounding atmosphere which reduces the efficiency of the engine as the particles in lower temperatures are denser, thus richer in oxygen which is required for combustion. The snorkel would usually be located near the fender or grille of the engine [2]. The cold air drawn in would eventually allow more complete combustion which increases the vehicle’s engine performance and fuel efficiency [3, 4].

The resonator is a chamber that is installed on the side of the AIS to dampen out the vibration and noises produced during operation [5]. In this project, the resonator was excluded as the fluid flow disturbance caused by it is minimal, hence negligible according to Proton. On the other hand, the air filter plays a major role in the entire AIS as it significantly affects the fluid flow in the airbox [6]. The air filter allows the engine to inhale clean air which is required for engine combustion. The air filter’s main role is to filter out dirt from the air particles from the snorkel. Dirt and foreign particles may cause severe damage to the engine pistons if entered [7]. Next, the airbox which is the most important part of the AIS functions to contain the air filter and provide a constant flow of air to the engine via the zip

(3)

tube. The zip tube function as a connector that connects the airbox to the engine intake manifold. Pressure loss occurs throughout the AIS and mainly within the airbox due to the complex geometry.

FIGURE 1. Components of the Proton Iriz air intake system modelled in Solidworks.

The pressure loss for this project will be calculated using a computational fluid dynamic (CFD) program. It is an industrial practice that CFD is to be used for fluid flow analysis rather than theoretical or experimental approaches to reduce cost or time required since computing cost had significantly reduced over the past decades. Thus, to study the fluid flow through the AIS, an exact model will be drawn in a solid modelling computer-aided design software before simulating it in the real-world environment through the CFD program [8]. The CFD simulation will begin from the point where the air is sucked into the airbox to the point where it leaves the airbox to the engine while any pressure loss within this flow will be observed and analysed. The CFD simulation model will be validated with the experimental data provided by Proton which is one of the objectives of this project. A conceptual improvement design could be conceived after analysing and studying the fluid characteristics that are occurring inside the AIS. The hypothesis made is that the geometry of the air intake system is the root cause of pressure drop and it occurs where the flow begins to transform from laminar to turbulent.

All in all, pressure drop reduction in the AIS would lead to an increase in performance of the engine with the least amount of cost required as compared to improving the performance by using performance chipsets, adding turbo or superchargers, or changing to aluminium engine pistons [9].

RESEARCH METHODOLOGY

This project is solely software-based; hence a physical prototype will not be manufactured. Figure 2 shows the simplified flow chart for the project. The flow chart is repeated for every design iteration starting with the original design. The geometry file was prepared in Solidworks before exporting into ANSYS for meshing and Fluent CFD simulation. The results were obtained through Post-CFD and analysed. The analysis of the fluid flow patterns inside the AIS provides ideas for design improvements. The design improvements are then drawn and simulated before being compared with the original design. If the design shows deterioration, the entire process restarts.

FIGURE 2. Simplified flow chart for the project.

(4)

Geometry

The CAD geometry file was prepared in SolidWorks according to the exact AIS dimensions. The external features of the AIS such as bolts, nuts, brackets and ribs were removed as it was not required for CFD simulation while the internal features such as ribs and fillets were maintained. CFD simulation only requires the fluid part of the geometry which is the internal part only. To obtain the internal part of the AIS, a negative Boolean was conducted on the model.

The Boolean Subtract method is to allow extraction of an internal volume of a part [10]. To do so, the assembly of the AIS was saved as a part file. An additional solid was extruded through the entire model without merging the results.

A hole was cut at the inlet and the outlet as to allow SolidWorks algorithms to know that there are inlets and outlets.

The ‘Intersect’ feature is used to extract the internal volume of the flow [11]. After obtaining the internal volume, the file was exported as a Parasolid file. This step is required as ANSYS does not support later versions of SolidWorks file format. Parasolid file provides lesser geometrical issues during the conversion as compared to IGS and STEP file formats [10, 12].

The geometry file for the design improvements was also prepared in Solidworks, following the same procedure as the original design. Figure 3 shows four different designs of the AIS after negative Boolean was conducted.

FIGURE 3. The extracted volume of four different design of the air intake system.

Design Modeler

During the conversion of file formats, geometrical errors could occur especially for large and complex assembly models such as the AIS. The Parasolid file format was first imported into ANSYS Workbench before Design Modeler was used to check and repair geometrical problems. Faulty faces, discontinued edges, silvers and holes were fixed before defining the parts. Since the model was imported as a whole, the ‘Slice’ tool was used to separate them into three different parts: Top Part, Filter, and Bottom Part. The top part consists of the top part of the airbox and the zip tube which connects the airbox to the engine. The bottom part consists of the bottom part of the airbox, filter piece holder, the connecting tube, and snorkel. This step is required to separate the filter and the rest of the parts. It was required to let ANSYS know that the filter is a porous material while the other faces could be assumed to be a wall.

The porosity of the filter would be defined later in the Fluent setup. Before proceeding into meshing, the entire geometry was defined as fluid in the tree outline section.

Meshing

Meshing is a process of discretizing the component or geometry into a number of elements and nodes. This is to allow a load to be uniformly distributed around the component or geometry [13]. Fine mesh sizes are required for a more accurate analysis however requires more computing power and time [14]. To find a balance between accuracy and computing resources, grid independence testing (GIT) was conducted. GIT is a procedure whereby a graph of a

(5)

number of nodes against pressure difference was plotted. As the mesh sizes increases, the number of nodes increases.

The higher the number of nodes, the higher the accuracy of the analysis [14]. However, there would be a point where the graph converges, and the difference would not make a significant difference. Hence, at this point, the mesh size was optimal to be used on all models.

Before meshing the geometry, named selection was defined. This would allow Fluent to automatically recognize the defined selection. The snorkel’s inlet surface was named as ‘inlet’, the zip tube’s outlet surface was named as

‘outlet’ while the filter piece was named as ‘porous’. The non-mentioned faces will automatically be assumed as walls by Fluent.

Next, the meshing method used for this geometry is the automatic method. The automatic method control functions by first using sweep method for solid models and uses quadrilateral element generation for surface body models.

However, if the solid models could not be swept, the body will be meshed using the tetrahedrons method [15, 16].

Newer version of ANSYS would automatically select the best mesh option for each part. For instance, the rectangular filter piece was meshed using square-sweep mesh while the other complex geometry parts were meshed using triangular-tetrahedral meshes.

The model was meshed at different element sizes based on the GIT requirements. GIT requires that the number of nodes increases by two-fold for every increment. However, there are no settings in ANSYS Meshing to set the desired number of nodes as the outcome, hence a rough estimate of element size was used to begin. For the original design, it was meshed at 2.10 mm, 1.85 mm, 1.35 mm and 1.00 mm which provides 300k nodes, 750k nodes, 15 million nodes and 30 million nodes which fits the requirement of the GIT whereby the number of nodes increases by two-fold.

Fluent CFD Simulation

The continuity and momentum equations are solved by Fluent to obtain the CFD simulation results. The continuity equation follows the mass conservation law whereby energy may neither be created nor destroyed [13]. This means that the rate of increase of a mass of a fluid element must equal to the net rate of the mass flow into the fluid element, giving Eq. (1) [16].

߲ߩ

߲ݐ൅ ݀݅ݒሺߩݑሻ ൌ Ͳ (1)

Whereby:

డఘ

డ௧ = rate of change of density with respect to time

݀݅ݒሺߩݑሻ = net mass flow out of the element across its boundaries

The momentum equation which satisfies Newton’s second law of motion states that the summation of forces which are acting on the fluid element must equal to the product between its mass and acceleration of the fluid element.

Through this principle, Eq. (2) to Eq. (4) was obtained for the x-direction, y-direction and z-direction [13, 16, 17].

For the x-direction:

ߩܦݑ

ܦݐ ൌ߲ሺെ݌ ൅ ߬_௫௫ሻ

߲ݔ ൅߲߬_௬௫

߲ݕ ൅߲߬_௭௫

߲ݖ ൅ ܵ_ெ௫ (2)

For the y-direction:

ߩܦݒ ܦݐ ൌ߲߬_௫௬

߲ݔ ൅߲൫െ݌ ൅ ߬_௬௬൯

߲ݕ ൅߲߬_௭௬

߲ݖ ൅ ܵ_ெ௬ (3)

For the z-direction

ߩܦݓ ܦݐ ൌ߲߬_௫௭

߲ݔ ൅߲߬_௬௭

߲ݕ ൅߲ሺെ݌ ൅ ߬_௭௭ሻ

߲ݖ ൅ ܵ_ெ௭ (4)

Whereby:

ߩ = density ߬ = shear stress

݌ = pressure S = source term

஽௨

஽௧ǡ^஽௩

஽௧ǡ^஽௪

஽௧ = velocity in the x, y and z-direction

For the source term S, if gravity was acting in the negative y-direction, then,

ܵ_ெ௫ ൌ Ͳǡ ܵ_ெ௬ൌ െߩ݃ǡ ܵ_ெ௭ൌ Ͳ

(6)

Before beginning the setup for Fluent, several parameters should be known: mass flow rate, material properties and porosity data.

TABLE 1. The mass flow rate at different engine speeds [18, 19].

Engine Speed (rpm) Air Flow (kg/s)

5500 0.067831

4000 0.047703

2000 0.021992

In this project, three different speeds were analysed. The three speeds that will be simulated are the speed at when power is at maximum (5500 RPM), torque is at maximum (4000 RPM) and during normal driving (2000 RPM). From Table 1, it can be seen that the higher the engine speed, the higher the mass flow rate.

Table 2 shows the materials data for the AIS. The data is provided by Proton whereby the air properties was obtained based on the environment when the experiment was conducted.

TABLE 2. Material and fluid properties [20, 19].

Material Density Specific Heat Thermal Conductivity Dynamic Viscosity

Ȁ^ଷ Ȁ ή Ȁ ή Ȁ

Air 1.185 1006.3 0.026 1.844E-05

To determine the type of flow for this simulation, the Mach number equation was used. The Mach number can be determined using Eq. (5).

ǡ ൌ

(5)

If the value of Mach number is lower than 0.3, the flow is considered as incompressible flow; If the value of Mach number is in between 0.3 to 1, the flow is said to be subsonic flow while supersonic flow occurs when the Mach number is higher than 1 [21].

The speed of sound at room temperature was given as 343 m/s while the speed of flow was obtained from Table 1.

To convert the mass flow rate to velocity, Eq. (7) was used.

ܯܽݏݏ݂݈݋ݓݎܽݐ݁ǡ ݉ሶ ൌ ߩܸܣ (6)

The density of the fluid which is air is given as 1.1845 kg/m³ while the diameter of the intake tube was 0.054 m.

Taking engine speed at 5500 RPM with a mass flow rate of 0.0678 kg/s, ͲǤͲ͸͹ͺ ൬

൰ ൌ ͳǤͳͺͶͷ ൬

^ଷ൰ ൈ ቀ

ቁ ൈɎ ൈ ͲǤͲͷͶ^ଶ

ܸ݈݁݋ܿ݅ݐݕǡ ܸ ൌ ʹͷȀ Ͷ

(7)

Hence

,

ܯ݄ܽܿܰݑܾ݉݁ݎǡ ܯܽ ൌ ʹͷ

͵Ͷ͵

ൌ ͲǤͲ͹͵

The calculation for Mach number confirms that the flow was incompressible as the value is lower than 0.3.

Besides that, another important aspect is Reynold’s number. The Reynolds number can be mathematically defined as the division of inertial forces of the fluid over the viscous forces of the fluid which gives a dimensionless quantity [22]. To determine whether the flow is laminar or turbulent, Reynold’s number was calculated. Eq. (8) shows the formula for Reynold’s number.

ܴ݁ݕ݊݋݈݀^ᇱݏܰݑܾ݉݁ݎǡ ܴ݁ ൌߩܸܦ ߤ

(8) Using the parameters obtained in Eq. (7) and Table 2,

(7)

ܴ݁ݕ݊݋݈݀^ᇱݏܰݑܾ݉݁ݎǡ ܴ݁ ൌሺͳǤͳͺͶͷȀ^ଷሻሺʹͷȀሻሺͲǤͲͷͶሻ ሺͳǤͺͶͶȀሻ

ൌ ͺ͸͹ͳ͹Ǥ͹͵͵

The Reynold’s number obtained for the flow at an engine speed of 5500 RPM was 86717.733. For an incompressible fluid flow in a pipe, if Reynold’s number is less than or equals to 2300 the fluid was in the laminar regime while for any number above 4000, the fluid was in the turbulent regime [23].

Next, several turbulence models such as k-epsilon, Transition SST and Reynolds Stress equations were simulated to determine which provides the most accurate results as compared to the experimental results. Laminar flow models will not be used as the flow through the AIS was confirmed to be turbulent. Turbulence modelling was conducted on the three types of models however only k-epsilon was able to converge within a reasonable amount of time. Transition SST was not able to converge even after 5000 iterations with a computing time of roughly 24 hours. On the other hand, Reynold’s Stress model was far worse. It took more than 48 hours to reach 5000 iterations and does not show signs of converging. Hence, the k-epsilon model was selected to be conducted on the other two mass flow rates. The main reason for the model to not converge was due to the mesh method. Tetrahedral mesh has contrasting element size, large face angles and high vertex degrees of mesh which is a serious limitation when directional flow field was present such as in this case [24]. For complex geometries such as the AIS, the only mesh method that could be used was the tetrahedral mesh method.

Energy equation was switched off as thermal effects shall be neglected in this simulation. The properties of fluid were entered and the cell zone conditions of the entire AIS are set to fluid. For the filter, it was defined as ‘Porous Zone’ and the values in Table 3 were entered. The viscous resistance is denoted as tensor d1, d2, and d3 while the inertial resistance was denoted as tensor f1, f2, and f3. The viscous resistance and inertial resistance are constant at all three directions, as the filter was assumed to be symmetrical inside out.

TABLE 3. Porosity data [19].

Porosity Viscous Resistance (tensor d) Porous Inertial Resistance (tensor f)

^ିଶ ^ିଵ

ͳǤͶͺͺ ൈ ͳͲ^଼ ͻǤͲ͵͵

Next, in the boundary condition settings, the inlet was chosen to be mass flow inlet. The mass flow rate from Table 1 was entered for each engine speeds. The walls of the AIS were set to no-slip condition (which was by default) as to simulate a realistic condition [25].

The following step was to run hybrid initialization. Hybrid initialization solves the Laplace equation which was formed by a collection of parameters and boundary interpolation methods [16]. After running the hybrid initialization, the calculation was allowed to run. The number of iterations was set to 5000 as it should be sufficient. To know whether 5000 iterations were sufficient, the scaled residual graph plotted by Fluent can be analysed. All values on the graph should converge to 0.001 residual value; if the graph does not converge to 0.001, more iterations are required [16].

The results were analysed in Post-CFD. The pressure contours, streamlines and pressure values at the inlet and outlet were determined. The pressure at the inlet and outlet was obtained through the probe feature. The location of the probe was provided by Proton which was used to obtain the experimental results. The results were compared with experimental results to determine the validity of the CFD which was the first research objective.

RESULTS AND DISCUSSIONS Mesh Statistics

Table 4 to Table 7 shows the mesh statistics for the original design, improvement 1 to improvement 3. The tabulated data were used to plot the GIT graphs shown in Figure 4 to Figure 7. Based on the graphs, it can be seen that the results converge at roughly 1.3 mm element size for all designs. To conduct turbulence modelling, the 1.3 mm element size mesh was used, however, for all the design iterations, the graph does not converge for Reynold’s stress method and takes more than 24 hours to reach 5000 iterations without converging for Transition-SST method, thus k-epsilon method was used for all the design iterations. Since turbulence modelling succeeded only with the k-epsilon method, the lowest element size which is 1 mm for all designs were used instead of using 1.3 mm element size as the result were already present.

(8)

TABLE 4. Table of mesh statistics for the original design.

Original Design Element

Size (mm)

Number of Nodes

Number of

Element Skewness Orthogonal Quality

Pressure Difference (Pa)

Time Taken (hr)

1.00 3098148 12075816 0.213 0.869 -786.85 5.82

1.35 1544695 6286380 0.217 0.866 -788.13 3.15

1.85 767355 3259034 0.223 0.862 -793.97 0.89

2.15 356123 1235612 0.231 0.857 -799.80 0.47

FIGURE 4. Grid Independence Testing Graph for the original design.

TABLE 5. Table of mesh statistics for Improvement 1.

Improvement 1 (Removed Bottom Section) Element

Size (mm) Number of

Nodes Number of

Element Skewness Orthogonal

Quality Pressure

Difference (Pa) Time Taken (hr)

1.00 2979726 11461906 0.2121 0.7877 -756.67 4.44

1.32 1542625 6229626 0.2164 0.7828 -757.97 2.72

1.85 729377 3064111 0.2223 0.7764 -777.46 1.45

2.10 315811 1421467 0.2256 0.7716 -782.00 0.55

FIGURE 5. Grid Independence Testing Graph for Improvement 1.

785.00 790.00 795.00 800.00 805.00

0.00 100.00 200.00 300.00 400.00

Number of Nodes (x1000)

Grid Independence Testing

750.00 755.00 760.00 765.00 770.00 775.00 780.00 785.00

0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00

Grid Independence Testing

(9)

Improvement 2 (Removed Top and Bottom Section) Element Size

(mm) Number of

Nodes Number of

Quality Pressure

Difference (Pa) Time taken (hr)

1.00 3062901 10880712 0.21063 0.87046 -763.04 10.14

1.30 1520354 6064170 0.21496 0.86732 -764.01 8.50

1.75 771378 3174491 0.21894 0.86419 -770.83 0.96

2.10 354807 1627941 0.21997 0.86201 -775.00 0.51

Improvement 3 (Spherical Bottom) Element

Size (mm) Number of

Nodes Number of

Quality Pressure Difference (Pa)

Time Taken

(hr)

1.00 3044825 11805932 0.211 0.788 -771.72 5.30

1.35 1517044 6153474 0.215 0.784 -773.12 3.78

1.85 753249 3189131 0.221 0.777 -782.17 1.51

2.15 351489 1787398 0.231 0.770 -793.99 0.85

760.00 765.00 770.00 775.00 780.00

0 50 100 150 200 250 300 350

Grid Independence Testing

770.00 775.00 780.00 785.00 790.00 795.00 800.00

0 50 100 150 200 250 300 350

Grid Independence Testing

(10)

Fluent Pressure Contour

FIGURE 8. Collage of pressure contours of the four designs at 5500 RPM.

Figure 8 shows the pressure contour for four designs at 5500 RPM. The pressure contour for all the design shows similar features whereby the zip tube has the lowest pressure. On the contrary, the highest pressure appeared to be at the inlet. This proves that pressure loss occurs as the fluid flows through the intake tube to the airbox to the zip tube.

TABLE 8. Table of results for the original design.

Original Design

RPM Inlet (Pa) Outlet (Pa) Pressure Drop (Pa)

5500 -42.17 -829.01 -786.85

4000 -149.47 -566.10 -416.62

2000 -145.86 -258.88 -113.02

TABLE 9. Table of results for Improvement 1: Removed Bottom Section.

Improvement 1 (Removed bottom section)

RPM Inlet (Pa) Outlet (Pa) Pressure Drop (Pa) Percentage Difference (%)

5500 -70.15 -826.82 -756.67 3.84

4000 -162.65 -562.85 -400.20 3.94

2000 -146.39 -257.37 -110.98 1.81

TABLE 10. Table of results for Improvement 2: Removed top and bottom section.

Improvement 2 (Remove top and bottom section)

5500 -67.14 -830.18 -763.04 3.03

4000 -161.82 -564.53 -402.71 3.34

2000 -146.33 -257.72 -111.40 1.44

(11)

FIGURE 9. Graph of pressure drop across the AIS for the original design.

TABLE 11. Table of results for Improvement 3: Spherical Bottom.

Improvement 3 (Spherical bottom)

5500 -57.34 -829.06 -771.72 1.92

4000 -157.45 -564.47 -407.03 2.30

2000 -145.70 -257.89 -112.20 0.73

TABLE 12.Table of comparison of experimental results and original design simulation results.

RPM Experimental Results Original Design

Percentage Difference (%) Pressure Drop (Pa) Pressure Drop (Pa)

5500 -350.00 -786.85 -124.81

4000 -320.00 -416.62 -30.20

2000 -150.00 -113.02 24.65

Table 8 to Table 11 shows the results obtain from the inlet and outlet from the respective designs. The percentage difference was calculated in reference to the original design. FigureFIGURE 9 shows the graph of pressure drop across the air intake system for the original design and it could be seen that as the RPM increases, the pressure loss increases.

Among the three design iterations, it shows that improvement 1 has the highest percentage difference at almost 4%

improvement. TableTable 12 shows the percentage difference between the experimental result and the original design simulation result. The percentage of error ranges from 25 % to -125% which was a big variation. Despite having high deviation, the percentage of error for 4000 RPM and 2000 RPM are about 30% similar to the experimental results. A research done by Zhao Zhang et al. shows that a percentage of error within 10 % is considered extremely accurate while 20 to 30% was acceptable and common [26]. Thus, the simulation model can be said to be validated.

-900.00 -800.00 -700.00 -600.00 -500.00 -400.00 -300.00 -200.00 -100.00

0.00 Inlet Bottom Filter Upper Filter Outlet

Pressure (Pa)

Locations

Pressure drop across the AIS for the Original Design

5500 RPM 4000 RPM 2000 RPM

(12)

Streamline

FIGURE 10. Streamline view for the four designs.

Based on the streamline view shown in Figure 10, recirculation of flow occurs throughout the bottom airbox for the original design. To reduce the recirculation of flow at the bottom airbox, a section of the bottom airbox was removed. This significantly reduced the occurrence of recirculation of flow. However, it was noticed that the right side of the top airbox was not utilized as the flow tend to accumulate towards the left side of the top airbox. Since the fluid does not utilize the space, in the second design improvement, the section of void was removed. Despite having similar streamlines, the probed results show improvement 2 had a lower improvement percentage. Furthermore, improvement 3 streamline shows recirculation of flow at the bottom airbox but slightly more organized as compared to the original design. The probed results show a maximum of 2.3% improvement as compared to the original design. This observation shows that recirculation of flow was definitely a factor of the pressure difference. A more organized flow shows lower pressure difference as compared to a more scattered and chaotic recirculation of flow such as in the original design.

Design Discussions

The original design shows a high-pressure difference across the inlet to the outlet of the AIS. CFD simulation shows that the fluid flow was recirculating chaotically at the bottom airbox. The complex design of the AIS was the cause for the recirculation of flow as there are bends, and voids within the AIS. To redesign the AIS, several factors that must be considered are the total volume, structural integrity, and manufacturability. The total volume of the airbox must be at least 4.55 litres to avoid choking the engine as engine requires a certain amount of airflow every second.

On the contrary, the total volume of the airbox could not be larger than the original design (5.95 litres) as the space in the engine bonnet is packed.

In the first design improvement, a section of the bottom airbox was removed as the recirculation of flow occurs at that section. The total volume after reduction was 5.4 litres which are within the safe design limit of 4.55 litres. This design improvement shows roughly 4% improvement as compared to the original design. A problem arose in this design improvement as the airbox was not fully utilized. The fluid flow tends to accumulate at the left side of the airbox, leaving a void at the right side of the airbox. To counter this issue, the second design improvement was to remove the void at the top right of the airbox. After removal, the total volume was still within the safe design limits at 4.9 litres. The results, however, show a slight deterioration as compared to the first improvement but still had an improvement over the original design. There was nothing special about this design as the streamlines and pressure contour are very similar to the first design improvements besides the section of void. The first and second design

(13)

improvement was manufacturable according to Proton, hence it met the objective of the project whereby design improvements were made and were feasible. The third design improvement was made to study how much does the recirculation of flow affects the pressure drop. The third design was designed to have a spherical bottom to aid the recirculation of fluid flow. From the streamlines in Figure 10, it was seen that the recirculation of flow was more organized as compared to the original design. The results show there was an improvement by having a more organized recirculation of flow rather than a very chaotic recirculation of flow such as in the original design. The spherical bottom had only reduced the volume by 0.03066502 litres which are also within the safe design limits.

FIGURE 11.Drawing a control volume to act as a closed system.

TABLE 13. Area average pressure at the inlet and outlet of the control volume for the original design.

Original Design

5500 -294.01 -651.52 -357.50

4000 -277.53 -474.79 -197.26

2000 -174.25 -238.08 -63.83

TABLE 14. Area average pressure at the inlet and outlet of the control volume for Improvement 1.

Improvement 1

5500 -318.02 -651.88 -333.86

4000 -286.38 -474.51 -188.13

2000 -173.80 -236.77 -62.97

Improvement 2

5500 -317.56 -652.06 -334.50

4000 -285.97 -474.63 -188.66

2000 -173.69 -236.80 -63.11

Improvement 3

5500 -303.71 -652.09 -348.38

4000 -280.63 -475.02 -194.39

2000 -173.09 -237.04 -63.95

To understand the physics theory behind the three design improvements, a control volume was set such as shown in Figure 11. Since the volume in the space of interest for this analysis is only the airbox, a control volume was drawn to treat it as a closed surface or closed system. In close systems, Boyle’s law states that pressure at a given mass of an ideal gas is inversely proportional to the volume provided that the system is at a constant temperature. In the simulation, the environment was set to constant room temperature, thus it can be applied here [27, 28]. From the results shown in

(14)

Table 13 to Table 16, it was seen that as the volume decreases, the pressure increases except for the second design iteration. The second design iteration shows a lower pressure despite being 0.5 litres less in volume as compared to the first design improvement. The logical explanation is that the k-epsilon model may not be sensitive enough to predict the turbulent flow in this scenario. The k-epsilon model uses averaging procedure, unlike direct numerical simulation (DNS) which solves the time-dependent Navier-Stokes equation by capturing the large and small eddies directly which requires extremely high computational power. The averaging procedure basically takes the mean of the large and small eddies for modelling which significantly reduces computational power at a cost of less accurate and precise results [13, 29]. Furthermore, the third design improvement which 0.03066502 litres of volume had been reduced from the original design shows a higher pressure as compared to the original design which proves the theory of Boyle’s law.

CONCLUSION

The objective of this project was to design an air intake system with a lower pressure loss as compared to the original design. The objective was achieved through three different design iterations. Among the three design iterations, the first design improvement whereby a section of the bottom airbox was removed to guide the fluid flow upwards to prevent recirculation of flow. By comparing the pressure loss, there was roughly 4 % improvement as compared to the original design. The second design also showed improvement but slightly lower at 3.3 % improvement while the third design had the lowest improvement at 2.3 % only. Due to several design constraints such as volume, manufacturability and structural integrity, the design of the AIS could not be changed by a lot, the only design change could be inside the airbox, hence there is a limitation on how much it could be improved.

ACKNOWLEDGEMENTS

The author would like to thank and express gratitude to the School of Engineering, Taylor’s University Lakeside Campus, Malaysia for providing the computing resources required for this simulation project.

REFERENCES

1. A.S.Phulpaga and N. Gohel, "CFD Analysis of Air Intake System," International Journal on Theoretical and Applied Research in Mechanical Engineering, vol. 4, no. 2, pp. 1-4, 2015.

2. S. JohanneVevelstad, "Oxygen and Temperature Effect on Formation of Degradation Compounds from MEA,"

Energy Procedia, vol. 63, no. 1, pp. 957-975, 2015.

3. P.Baskar and A. Kumar, "Experimental investigation on performance characteristics of a diesel engine using diesel-water emulsion with oxygen enriched air," Alexandria Engineering Journal, vol. 56, no. 1, pp. 137-146, 2016.

4. M. M. Abdelaal, B. A. Rabee and A. H. Hegab, "Effect of adding oxygen to the intake air on a dual-fuel engine performance, emissions, and knock tendency," Energy, vol. 61, pp. 612-620, 2013.

5. C. Meriça, H. Erol and A. Özkan, "Design and application of a compact helical air intake system resonator for broadband noise control," Applied Acoustics, vol. 131, pp. 103-111, 2018.

6. A. K. Maddinenia, D. Das and R. M. Damodaran, "Numerical investigation of pressure and flow characteristics of pleated air filter system for automotive engine intake applications," Separation and Purification Technology , vol. 212, no. 14, pp. 126-134, 2019.

7. S. Kaur and E. R. Malik, "Design Optimization of Automotive Air Filter Housing for Minimum Pressure Drop,"

International Journal of Engineering Development, vol. 5, no. 4, pp. 1167-1170, 2017.

8. M. Zadravec, S.Basic and M.Hribersek, "The Influence of Rotating Domain Size in a Rotating Frame of Reference Approach for Simulation of Rotating Impeller in a Mixing Vessel," Journal of Engineering Science and Technology , vol. 2, no. 2, pp. 126-138, 2007.

9. L. Kerni, A. Raina and M. I. U. Haq, "Performance evaluation of aluminium alloys for piston and,"

materialstoday: Proceedings, vol. 5, no. 9, pp. 18170-18175, 2018.

10. M. Lombard, Mastering SolidWorks, John Wiley & Sons, Inc: Massachusetts, 2019.

(15)

11. G. Verma and M. Weber, SolidWorks 2017 Black Book, Rajasthan: CADCAMCAE Works, 2017 .

12. S. Marjudi, M. F. M. Amran and K. A. Abdullah, "A Review and Comparison of IGES and STEP," in Proceedings Of World Academy Of Science, Engineering And Technology Volume 62, Selangor, 2010.

13. J. Tu, G.-H. Yeoh and C. Liu, Computational Fluid Dynamics: A Practical Approach, Oxford: Elsevier Ltd, 2018.

14. H. Guo and H. Long, "Effect of Mesh Size in Numerical Simulation of Turbine Housing in Turbocharger," SAE Technical Paper, pp. 1-5, 2018.

15. I. ANSYS, ANSYS 19.2 Release, Pennsylvania: ANSYS, Inc, 2018.

16. I. ANSYS, ANSYS Fluent Theory Guide, Pennsylvania: ANSYS, Inc., 2013.

17. A. Bahatmaka and D.-J. Kim, "Numerical Approach For The Traditional Fishing Vessel Analysis of Resistance By CFD," Journal of Engineering Science and Technology, vol. 14, no. 1, pp. 207-217, 2019.

18. S. Kwan and N. Z. A. Bakar, "Design Improvement of an Automotive Air Intake System for 1.3L Passenger Vehicle," Taylor's University, Petaling Jaya, 2018.

19. N. S. b. Sepuan, "The Influence of Air Filter on the Pressure Drop Inside an Automotive Air Cleaner," Taylor's University, Petaling Jaya, 2018.

20. L. Clean and Science CO., "JAF-205 (NC) Specification Sheet," Clean and Science CO., Ltd., Seoul, 2004.

21. H.Guillard and B.Nkonga, "Chapter 8 - On the Behaviour of Upwind Schemes in the Low Mach Number Limit:

A Review," Handbook of Numerical Analysis, vol. 18, pp. 203-231, 2017.

22. GeoffBarker, "Pipe sizing and pressure drop calculations," The Engineer's Guide to Plant Layout and Piping Design for the Oil and Gas Industries, vol. 1, no. 1, pp. 411-472, 2018.

23. F. M. White, Fluid Mechanics 7th edition, London: McGraw-Hill, 2010.

24. R. Biswas and R. C.Strawn, "Tetrahedral and hexahedral mesh adaptation for CFD problems," Applied Numerical Mathematics, vol. 26, no. 1-2, pp. 135-151, 1998.

25. B. E.Rapp, "Chapter 9 - Fluids," Microfluidics: Modelling, Mechanics and Mathematics, pp. 243-263, 2016.

26. Z. Zhang, W. Zhang, Z. J. Zhai and Q. Y. Chen, "Evaluation of Various Turbulence Models in Predicting Airflow and Turbulence in Enclosed Environments by CFD," HVAC&R Research, vol. 13, no. 6, pp. 871-886 , 2011.

27. P. Poolchak and W. Kanchanapusakit, "An equipment design to verify Boyle’s law," Journal of Physics:

Conference Series, vol. 1144, pp. 1-3, 2018.

28. T. Limpanuparb, S. Kanithasevi, M. Lojanarungsiri and P. Pakwilaikiat, "Teaching Boyle’s Law and Charles’

Law through Experiments that Use Novel, Inexpensive Equipment Yielding Accurate Results," Journal of Chemical Education, vol. 96, no. 1, pp. 169-174, 2019.

29. J. Blazek, Computational Fluid Dynamics 3rd Edition, Oxford, United Kingdom: Butterworth-Heinemann, 2015.