1. University of Minnesota
Investigation of Intake Concepts for a Formula SAE Four-Cylinder Engine Using 1D/3D (Ricardo WAVE-VECTIS) Coupled Modeling Techniques
Mark Claywell
Donald Horkheimer
Garrett Stockburger
University of Minnesota

2. Agenda Background Motivation Design Method
Simulation Methods and Assumptions
Grid Convergence Study
Results
Flow Visualization
Improved Understanding Through Issues Raised By Simulation
Conclusion

3. University of Minnesota SAE Engine
Background
Student Design Competition
Events in America, Australia, Brazil, Germany, Italy, Japan, United Kingdom
200+ Universities involved
Team score based on sales presentation, cost report, design quality, acceleration time, fuel economy, skid-pad, auto-cross and endurance race
University of Minnesota SAE Engine
Yamaha YZF-R6, Four Cylinder, Four stroke
600cc Displacement
15,500 rpm redline
Bore = 65.5mm, Stroke = 44.5mm
4-2-1 Exhaust Header
Sequential Port Fuel Injection (student calibrated)
DOHC, 4 valves per cylinder
Compression Ratio = 12.4:1
Fuel – Gasoline, 100 Octane

4. Motivation – Where to begin?

5. Design Process Main Focus of Paper Generate Concepts Evaluate & Select
State
Needs
Define
Specifications
Generate
Concepts
Evaluate
& Select
Detailed
Design
Manufacture
& Test
State
Needs
Define
Specifications
Generate
Concepts
Evaluate
& Select
Detailed
Design
Manufacture
& Test

6. Concepts vs Designs
Concepts
Designs

7. Making Concepts Comparable
Geometric Similarities
Inlet box to diffuser exit is identical
Restrictor geometry identical
Plenum volume kept constant
Runner length, diameter, and taper kept constant
Packaging bend angle held at 55°

8. Ricardo WAVE and VECTIS Simulation Software
WAVE (1D)
VECTIS (3D)
Intake to Tail-Pipe Engine Code
Easily provides realistic boundary conditions to CFD solver
Uses simple models to analyze complex problems
Provides actionable engine performance information
Quick simulation time
Off the shelf
Computational Fluid Dynamics (CFD) Code – More Accurate Flow Results
Integrated pre/post-processing and solver
Automatic mesh generator works with CAD derived geometry
Automotive specific solver modules
Easy to implement parallel solver
Off the shelf
Guessing at CFD boundary conditions is no good!
WAVE makes the use of VECTIS for intake design worthwhile
No code coupling = Questionable fidelity

9. Why Not a Steady State CFD Approach?
Agreement between flow solutions is poor
Steady state cylinder balance didn’t match
Steady state didn’t result in shocks, unsteady did
Finding non-tuning design improvements with steady state CFD may still be possible

10. Simplifying Assumptions
WAVE-VECTIS junctions placed in 1D flow areas
No throttle body
No fuel spray particles in CFD domain
k-ε turbulence model
Inlet Box

11. Grid Convergence Study
Grid convergence studies
ASME, AIAA, and others require it. Good practice.

12. Results – Total Volumetric Efficiency Predictions
Differences in total VE from concept to concept is small
VE curves can be made similar by varying intake dimensions

13. Results – Volumetric Efficiency

14. Results – Absolute Average Deviation of Volumetric Efficiency (I)
Total volumetric efficiency hides the superiority of the best intake concept
Individual cylinder to cylinder imbalance needs to be measured to identify best concept

15. Results – Absolute Average Deviation of Volumetric Efficiency (II)
Conical-Spline Intake Concept (With Straight Runners)

16. Results – Improvements in Calibration Process and Radiated Sound

17. Results – Choked Flow Insights and Post Diffuser Total Pressure Recovery
Lower AAD results in more regular pressure pulses at throat and lower time of choked flow
Beyond a certain diffuser length/area ratio total pressure recovery is limited
Diffuser Exit
Side Entry Intake
Conical Intake

18. Flow Visualization – Enhanced Understanding
Look at air and fuel cylinder to cylinder stealing
Identify regions of pressure loss and flow separation

19. Flow Visualization – Enhanced Understanding II
11,500 RPM
11,500 RPM
14,000 RPM
14,000 RPM

20. Time Averaged Velocity – 14,000 RPM
ISE pic is at 14,000 RPM

21. Velocity Normal to Plane – Time Averaged 14,000 RPM

22. Flow Visualization – Flow Dynamics Through Animation
14,000 RPM
14,000 RPM

23. Conclusion
Looked at how plenum geometry determines performance using WAVE-VECTIS
Found grid convergence studies essential for good CFD
Conical intake stood out as best
Smallest cylinder to cylinder imbalance
Better AFR control and acoustic characteristics
Regular pressure pulses at throat reduce choked flow
Adding bent runners for realistic packaging hurt performance, but only slightly
Improved understanding of fluid flow and dynamics

24. Questions? Acknowledgements
Ricardo Sponsorship and Support - Patrick Niven & Karl John
University of Minnesota Supercomputer Institute - Dr. H. Birali Runesha and Support Staff
University of Minnesota SAE Chapter - Dr. Patrick Starr and Dr. David Kittleson
Minnesota State University, Mankato - Dr. Bruce Jones

25. Improved Understanding Through Issues Raised By Simulation
Why would a rise in total pressure occur?
It could happen at low Re (<100) and areas of high normal stress
Changes in mesh size near large flow gradients
Any kind of flow averaging of unsteady non-homogenous transonic flow is going to introduce an error
Area-averaging can over-predict total pressure