DSToolbox

DSToolbox is a Matlab toolbox for analyzing and modelling of unsteady fluid dynamics experiments.

Installation

Clone the repository in the folder of choice on your computer, for example Documents/MATLAB. Open a Terminal window and type:

cd Documents/MATLAB
git clone git@github.com:lucasschn/dstoolbox

You can now make sure that you have a new folder called dstoolbox at the specified location.

Organization

The src folder contains the source code and the script folder contains scripts and examples that make use of the toolbox functionalities. The srcfolder is further divided in three subfolders:

common contains the classes definitions for the core-objects of the toolbox, e.g. Airfoil, AirfoilMotion, or SteadyCurv that are used independently of the model.
lib contains a library of useful functions, not classes, also common to all models.
model contains functions that are model-specific, e.g. only used for Sheng and Expfit models

Usage

The repository consists in a collection of objects, such as airfoils or typical motions, that can be created and on which functions can be applied. Scripts can then be written where these objects are created, such as in the example below:

airfoil = Airfoil('naca0012',0.5) % creates an Airfoil object with name naca0012 and 0.5m chord length
airfoil.steady = SteadyCurve(alpha,CN,13) % creates a SteadyCurve object

The created steady curve is assigned as the steady/static curve to the airfoil, with angle of attack alpha, normal coefficient CN, and static stall angle 13°. Different methods apply to steady curves, for additional computations or for plots.

airfoil.steady.plotCN() % plots the normal coefficient as a function of the AoA
airfoil.steady.fitKirchhoff() % fits a Kirchhoff curve to the static stall curve

An other object is needed to represent the dynamic stall experiment. It can be a ramp-up motion with constant pitch rate:

ramp = RampUpMotion('r',0.01,'V',0.5) % creates an ramp-up object with reduced pitch rate 0.01 and incoming flow velocity 0.5m/s.

a sinusoidal motion with constant frequency:

pitching = PitchingMotion('alpha',alpha,'CN',CN,'k',red_freq) % creates a pitching motion object with angle of attack vector alpha, normal coefficient CN and reduced frequency red_freq.

or a general motion with custom angle of attack history:

motion = AirfoilMotion('alpha',alpha,'CN',CN)

RampUpMotion and PitchingMotion both inherit from AirfoilMotion, meaning that all properties and methods of AirfoilMotion also apply to RampUpMotion and PitchingMotion. Howvever, RampUpMotion and PitchingMotion both individually have properties and methods that AirfoilMotion does not, such as the reduced pitch rate r and the reduced frequency k respectively.

All three airfoil motions accept name-value pair arguments when constructed. This means that you can pass any 'name',value pair as an argument when creating the object to automatically assign the value value to the property name to the object, as long as the property name exists for this object.

Different methods can be applied to a newly created ramp object, such as setCL() for setting the experiment lift coefficient corresponding to this ramp manually. A convenient function loadRamp(casenumber,filtered) sets up the experimental data to the ramp automatically from the server data.

ramp = loadRamp(22,false);
ramp.setPitchRate(airfoil);
ramp.findExpOnset()

The convenient function ramp=loadRamp(c,filtered) runs the labbook, loads the data, zeroes the data correctly and filters it if filteredis true. It then isolates the part of interest of the experiment, namely the ramp itself and a bit after it, and returns a RamUpMotion object ramp with the experimental force fields filled. Here the number 22 defines the experimental case number c corresponding to the desired experiement. All case numbers are defined in the labbook (labbook.min the repository). setPitchrate(airfoil)must be executed independently because it requires an airfoil object as an argument (in order to define the reduced pitch rate, the chord length is required). This will also set the convectime time vector, which allows findExpOnset() to be run. It is recommended to take the habbit to declare a ramp using this three methods before any usage.

Apply a dynamic stall model to an experimental case

Once the airfoil motion has been set up correctly, the corresponding aerodynamic normal coefficients can be predicted using a dynamic stall model. All dynamic stall models are methods that apply to motion objects. The general syntax for models is as follows:

ramp.BeddoesLeishman(airfoil,Tp,Tf,Tv,Tvl,'mode') % computes the aerodynamic loading experienced by an airfoil object describing the motion described by ramp according to Leishman-Beddoes model

The time constants Tp, Tf, Tv, and Tvl are necessary input arguments to Beddoes-Leishman model. Depending on the selected model the number of time constants can vary from 3 to 4. The 'mode' argument can be either 'experimental' or 'analytical' depending if the user wants numerical or analytical derivatives to be used. Alternatively, Sheng's model can be run on the same experimental data with the command:

ramp.BLSheng(obj,airfoil,Tf,Tv,Tvl,alphamode) % computes the aerodynamic load according to Sheng's version of LB model

In Sheng's model, there is no necessity to provide the Tpconstant, because the first delay due to airfoil unsteadiness is represented by the constant Talpha, which is determined based on experimental data. However, the prerequisite is that that constant has been already determined by running the script ShengCriterion2019.m. That one uses Sabrina's 2019 data. For Sabrina's 2018 data, see ShengCriterion2018.m.

ramp.BLSheng(obj,airfoil,Tf,Tv,Tvl,alphamode) % computes the aerodynamic load according to Sheng's version of LB model

You can verify if the necessary script has been run or not by checking for the presence of a linfit_flatplate.mat matfile in your repository (in case you are considering the flat plate airfoil).

Test files

A collection of scripts, such as testLB.m or testSheng.m contain all the necessary code for creating a ramp object from experimental data and apply one of the available dynamic stall models, depending on the test script. In testLB.m, the lines corresponding to the experiment number, associated with a certain pitch rate, and the time constants used in the call for BeddoesLeishman() method can be changed:

c = 71; % change this number to select the desired experiment (see labbook)

ramp.BeddoesLeishman(airfoil,3,3,1,1,'experimental') % change the four numbers corresponding to Tp, Tf, Tv and Tvl respectively

Parameter sweep

Part of this project is related to the analysis of the LB-prediction when the time constants are choosed randomly amongst a predefined population, with a large amount of samples. The script paramsweep.m runs the LB model using a user-defined number of samples among the user-defined range of Tp, Tf, Tv and Tvl. The script plotSweepResults.m allows for plotting the results of that parameter sweep using three different functions.

plotOneRate(res,rate,varx,vary,color_var) creates a scatter plot for the result mat-file loaded in the variable res, only for the experiments with pitch rate equal to rate. The x- and y-axis are defined by the variables varxand vary. The optional variable color_var allows for coloring the points on the scatter plot according to a third variable.
plotAllRates(res,varx,vary) creates a scatter plot for res, with x- and y-axis defined by the variables varxand vary. here no disction of pitch rate is made and all results are plotted.
plotHistogram(res,varx,vary,threshold) creates histograms for the result mat-file loaded in res with x-axis varx. The first one shows the distribution of all samples along the variable varx (in blue). The second one shows the distribution of samples meeting the criterion |vary|<threshold along the variable varx (in red). The two first ones are superimposed on the same axes. A third one in a new figure is then created showing the ratio between the samples meeting the criterion and the total number of samples for each bin of varx.

App

Before using the app, you have to run the script setPaths.mon your machine with the correct path to the folder where you want the produced figures to be saved.

Troubleshooting

When running a file always make sure that your current folder is the folder containing the file. For example, many scripts won't execute correctly if your current folder is not dstoolbox/scripts.

If you see the error message

Matlab couldn't read the experimental data. Are you sure you are connected to the server?

Make sure you are connected to the raw server. Otherwise, open labbook.m and make sure the path to the smartH folder is correctly set.

If Matlab stops responding when trying to load data from the server, first wait for at least 1min. The loading process of files up to 2GB has been observed to take around 30s on some configurations.

Then, check your firewall preferences. To make sure you can properly read a file from the server, browse to the file and try to manually open it by clicking on it.

On Windows machines, a permanent solution is to add the files and the raw servers to the list of sites considered as part of the local intranet. To do this, follow the instruction there

Please report any issue you may find using [Github's tool for issue reporting] (https://github.com/lucasschn/dstoolbox/issues)

License

MIT

lucasschn / dstoolbox