1. T. Guiard, Head of Energy-Saving Devices
Full-Scale Self-Propulsion Calculations Including Roughness Effects and Validation with Sea Trial Data
STAR Global Conference 2017 – Berlin
March 6 - 9
T. Guiard, Head of Energy-Saving Devices

2. Outline: Motivation & Background Test Case for Self-Propulsion
CFD Setup
Postprocessing and Results
Validation
Summary

3. 1) Motivation & Background
Who are we?
Founded in 1993 in Rostock
100% Owned by Becker Marine Systems (BMS)
Our main services for BMS:
Research & Development:
General CFD and FEM services for research and development.
Customization of rudders (powering, cavitation, maneuvering)
Energy Saving Devices:
Hydrodynamic design of ESDs
ESD product development

4. 1) Motivation & Background
Situation in ESD design:
The design of ESDs is strongly depending on the inflow field (ship wake).
Validation of CFD calculations for ship wake fields is typically limited to model scale.
A selected set of physical models, validated for model scale, is not automatically valid for full scale.
Full-scale wake, same ship & speed
Different model selection for turbulence and roughness
Question: What does the full scale wake look like, and what models should we use to reliably predict it?

5. 1) Motivation & Background
Validation studies for full-scale wake calculations:
Empirical scaling used as reference:
Validation with measured data:
Results: A combination of RST turbulence with a roughness height of ~0.2 mm shows the best correlation.
But: Using roughness is often said to lead to too high resistance.
Summary of validation studies in:
Guiard, T. (2014): ‘Challenges with Respect to Scaling and the Prediction of the Full-Scale Wake Field in the Context of the Becker Mewis Duct®’, Jahrbuch der Schiffbautechnischen Gesellschaft, 108. Band, pp , Hamburg, Germany

6. 2) Test Case for Self-Propulsion
In 2016 Lloyd’s Register was inviting to a workshop on full-scale simulation.
Test case: 16.9k DWT general cargo ship “REGAL” (IMO )
Hull, rudder and propeller geometry available from laser scan
Draught, trim and list, as well as water and air properties measured during the sea trial and made available for CFD.
Figure 1, Table 4, Table 7 from:
Ponkratov, D. (2016): ‘Workshop on CFD in Ship scale hydrodynamics: Description of cases’ Lloyds Register, June 2016

7. 3) CFD Setup - Basic Idea
Approach for the Workshop:
Using the Experience with wake prediction and propulsion modelling.
Concentrating on the calculation of propulsion and propulsive coefficients.
Gaining experience in calculating ship resistance, addressing the problem of too high resistance with roughness.
Problem: Which models to select for turbulence and wall modelling?
Solution: Splitting of the setup into resistance and propulsion.

8. 3) CFD Setup - Structure Setup 1 Resistance Setup 2 Prop Open Water
RANSE, SST k-ω, high y+
Roughness?
Free surface
Trim & sinkage
Setup 2 Prop Open Water
RANSE, RST, high y+
Smooth
Setup 3 Propulsion
RANSE, RST, high y+
Rough, ks = 188 µm
No free surface
No trim & sinkage
Propeller modelled as in setup 2.
“Setup” 4 Excel
Interpolating self-propulsion points

9. 3) CFD Setup – Resistance
General Setup:
RANSE, transient, SST k-ω,
Free surface, VOF
Trim & sinkage
Number of cells: 17 Mio
Wall functions, y+ ~ 100 to 500
Question: Should we use roughness for the resistance calculation?
How to find validation possibilities / references?
Reference to model ship measurements and corresponding trial predictions.
Reference to flat plate friction measurements with ‘ship-like’ surface conditions.

10. 3) CFD Setup - Resistance
Reference to ship model measurements and corresponding trial predictions:
Total viscous resistance comparison for two test cases:
very good correlation at model scale
Values scaled from model test are about 12 to 18% higher than calculated (SST k-ω, smooth).
A roughness of ks = 188 μm leads to ~ 25% increase in CV.

11. 3) CFD Setup – Resistance
Reference to flat plate friction measurements with ‘ship-like’ surfaces:
Table from: Schultz, Michel P. (2007): ‘Effects of coating roughness and biofouling
on ship resistance and powering’, Biofouling, 23:5,
Result: According to Schultz (2007) a value ks = 188 μm corresponds to a hull condition
between “Deteriorated coating…” and “Heavy slime”.
With respect to calculated viscous resistance and guidelines
ks = 188 μm seems realistic for a ‘mid-aged’ ship.

12. 3) CFD Setup – Propeller Open Water
General Setup:
RANSE, steady, RST, moving reference frame
Number of cells (propeller domain): 4.8 Mio
Wall functions, y+ ~ 100 to 450
Smooth
Problem: Propeller needs to run with RST turbulence for propulsion setup, SST k-ω would be preferred.
Question: How does the propeller perform with RST?
Result: Difference in open water data between RST and SST k-ω is considered acceptable.
RST is used for open water and propulsion without further corrections.

13. 3) CFD Setup - Propulsion
General Setup:
RANSE, transient, RST
Ship:
No trim and sinkage
Free surface approximated with a symmetry plane
Rough, ks = 188 μm
Wall functions, y+ ~ 150 to 700
Propeller:
Rotating SM
Propeller mesh identical to open water setup
Smooth
Wall functions, y+ ~ 100 to 600
Problem: Free Surface very close to (or actually intersecting with) the propeller.
Question: How should we position the symmetry plane?

14. 3) CFD Setup - Propulsion
Solution: Positioning the free surface in order to get the same nominal wake like in the resistance calculation.
Wake from resistance setup
SST k-ω, smooth
Wake from propulsion setup
SST k-ω, smooth
Wake from propulsion setup
RST, ks = 188 µm
9 kn:
14 kn:

15. 4) Postprocessing and Results
Preliminary Results
Postprocessing:
Get function RT(VS) from resistance setup
Get functions KT(J), KQ(J), η0(J) from open water setup
Get RT(VS) from propulsion setup without propeller
Get T(n, VS), Q(n, VS), RT(n, VS) from propulsion setup with propeller
Get t (from RT(n, VS), RT(VS), T(n, VS)), w and ηR
t, w, ηR assumed constant for small ΔVS (ΔCTH )
Interpolating on Rt(VS), Kt(J), Kq(J), η0(J) for self-propulsion point.

16. 5) Validation Sea trials performed under very good weather conditions
Very close agreement between CFD simulations and measurement (approx. within measurement accuracy)
Sea trial results & figure from:
Ponkratov, D. (2016): ‘Workshop on Ship Scale Hydrodynamic Computer Simulation: Draft Proceedings’ Lloyds Regiser, LR Reference: Ref. 8428

17. 6) Summary
Based on this limited set of full-scale results it seems that:
using ‘reasonable’ roughness values does not lead to too high resistance.
the SST k-ω model should be preferred for resistance, while the RST model should be preferred for wake. (which would be consistent with experience from model scale)
roughness should be taken into account with a ‘reasonable’ roughness value set. (which would be consistent with experience from model scale as ship models should be hydraulically smooth)
splitting the CFD setup can turn out to be beneficial by offering the possibility to use the ‘right tools’ for each individual job.

18. Thank You for Your Attention!