FlowLab Implementation at Cornell University

1
FlowLab Implementation at Cornell University

• Rajesh Bhaskaran and David Caughey
• Sibley School of Mechanical
• Aerospace Engineering
• Cornell University
• Ithaca, New York

2
FlowLab in a Lab Course

• MAE 427 Fluids/ Heat Transfer Laboratory
• Required class for Mechanical Engineering seniors
• About 110 students, 2 professors, 6 TAs
• Taught in groups of 6-8 students
• Approach
• Perform simulation corresponding to experiment
• Compare experimental and simulation results
• Experiments considered Heated pipe flow, Flow
  over an airfoil

3
Heated Pipe Flow Experiment

• Instructor Prof. Elizabeth (Betta) Fisher
• Raw measurements Wall and gas temperatures
  pressure drops power to heater

4
Heated Pipe Flow Experiment

• After number crunching, obtain
• Reynolds number, Re
• Friction factor, f (dimensionless drag)
• Nusselt number, Nu (dimensionless heat transfer
  coeff.)
• To make room for simulation, dropped measurements
  at several unheated conditions (Re and f only)

5
Heated Pipe Flow Simulation

• FlowLab Interface

6
Heated Pipe Flow Simulation

• FlowLab output Temperature contours

7
Heated Pipe Flow

• Mode of FlowLab use
• While waiting for experimental set-up to reach
  steady state, students walk over to adjacent
  computer lab
• T.A. provides hands-on introduction to pipe flow
  template with nominal experimental data
• Students repeat the simulation for their
  experimental data outside of class
• They calculate friction factor and Nusselt number
  manually and compare with experiment
• FlowLab available in a 24-hour lab

8
Heated Pipe Flow Results
9
Heated Pipe Flow Handout

• Developed 8-page handout
• Strategy of CFD
• Steps in CFD solution process

10
Heated Pipe Flow Handout

• Provides brief explanation for each of the steps.
  For instance
• Geometry Why rectangle? Dimensions of rectangle
• Physics Flow properties, viscous model, boundary
  conditions (inflow conditions, heat input at the
  wall etc.)
• Turbulence modeling background
• Grungy operating details
• Available on FlowLab website

11
Heated Pipe Flow

• The simulations help students
• Gain a better understanding of the experiment
  (through contour and vector plots) than is
  possible from a few point measurements.
• Confirm some assumptions made in the data
  processing for the experiment (e.g. that mixing
  region is long enough).
• Compare numerical approach with experiment and
  correlation.
• Classroom observations
• Small groups well suited to introducing students
  to the use of FlowLab and CFD background
• Enthusiasm for learning more about CFD.

12
Heated Pipe Flow

• FlowLab implementation was facilitated through
  close interaction with Fluent Inc.
• Enhancements to the standard pipe flow template
• Troubleshooting poor agreement with experimental
  results
• Bugs and suggestions
• Problems were addressed quickly by Fluent
• This interaction was key to success of the
  implementation

13
Heated Pipe Flow Student Feedback

• Students were asked to rate statements such as
  As a result of my learning in the CFD Labs, I am
  able to
• Present results from CFD simulations in written
  and graphical form.
• Run Flowlab and implement CFD process for
  laminar and turbulent flow.
• The rating scale ranged from "strongly agree"
  (scored as 6) to "strongly disagree" (scored as
  1).

14
Heated Pipe Flow Student Feedback

• 80 responses
• Overall average was 4.16 (SD 1.15) indicating
  that students, on average, "mildly agreed" with
  these statements.
• Strongest agreement (M4.63, SD 1.16) was for
  I am able to appreciate that simulation
  involves approximations and tradeoffs.
• Least agreement (M3.68, SD 1.20) was for I
  am able to evaluate iterative convergence through
  setting iterative convergence criteria and
  analysis of solutions residuals.

15
Heated Pipe Flow Student Feedback

• 58 responses to What are the best things about
  learning in the CFD lab?
• Visualization of results
• Increases in understanding and knowledge
• Hands-on experience with software
• 41 responses to What needs to be improved in
  the CFD lab to maximize its value to you?
• Instruction Better trained TAs, more time spent
  on CFD, etc.
• Technical aspects of FlowLab such as bug fixes
• Use FLUENT rather than FlowLab

16
Airfoil Experiment

• Effort led by Prof. David Caughey
• Flow past NACA0012 airfoil in a wind tunnel
• Rec 54,000
• Wall effects are significant

17
Airfoil Simulation

• FlowLab gives non-physical result Discontinuity
  in lift curve at zero incidence

18
Airfoil Simulation

• FLUENT results 
• Steady solution diverges even for zero incidence
• This suggests flow is unsteady
• Results differed for the segregated and coupled
  solvers
• Current FlowLab template designed for steady
  flows Not suited for unsteady, laminar,
  separated flow over an airfoil with wall effects

19
Conclusion

• FlowLab facilitates better pedagogy
• Enables simulation use in a lab setting
• Contributes to improved understanding of the
  experiment
• Helps students confirm some assumptions made in
  data reduction
• Pooling of resources by universities for
  curriculum development is essential
• Partnership with Fluent has been crucial

20
Acknowledgements

• Professor Elizabeth Fisher
• NSF Department of Undergraduate Education
• Swanson Engineering Simulation Program,
  Mechanical Aerospace Engineering, Cornell
  University