Bird Strike Simulation for an Aero Engine using LS-DYNA

Aim:

1) To create the simulation of bird strike on the Aero Engine from the given FE model.

2) Following are the information and conditions required to model the phenomenon

3) The blades should rotate at a constant velocity but the casing should remain stationary.

4) The cylindrical bird model should travel along its own axis and hit the blades.

5) Use elastic material for the bird(2000MPa) and the casing(200GPa) and the material model for the blades are given.

6) The velocity of the engine blades and the birds can be chosen so that blade failure can be seen within a short span of time.

7) To follow a consistent numbering approach 100000+ for nodes, 500000+ for elements, and 1000+ for the parts. All other keywords should be numbered within 10000-19999 while following a range for each one.

8) The bird, casing and the blades should be indifferent input files as well as control cards and boundary conditions and there should be one main file referencing all the input files.

9) The reference includes path should be such that the folder can be copied anywhere and it can correctly reference all the file and run the simulation.

10) The Unit System should be in kg-mm-ms only.

Theory:

The collision of birds on an aircraft while cruising in the air is called the Bird Strike event and it is one of the most dangerous accidents that often occur. Therefore, certification of a bird strike is one of the important processes in aircraft design. In particular, the areas where bird strikes occur a lot are on the front leading edge of the wing, the cowling part of the engine, rotating engine fan blades, the cockpit transparency and the landing gear part. This project is a classic nonlinear transient dynamics problem similar to a car crash and mobile drop. While accurate modelling of the problem requires advanced techniques such as SPH, this problem can be solved using a generic explicit solver.

Procedure:

The given FE model of Aero Engine assembly is opened in LS-PrePost and each part is assigned with section property and saved with ‘.k’ extension. The given FE model consists of parts such as bird, blade, hub and casing.

1. Part Definition keyword files:

Parts Section Definition:

1) Bird:

The given FE model of the dummy bird is retained and all other parts are deleted. The part is assigned with section shell properties with ELFORM = 16 and thickness = 2.5 mm. The keyword file is saved with a suitable name using the ‘.k’ extension.

2) Blade:

The given FE model of blade is retained and all other parts are deleted. The part is assigned with section shell properties with ELFORM = 16 and thickness = 1.2 mm. The keyword file is saved with a suitable name using the ‘.k’ extension.

3) Hub:

The given FE model of the hub is retained and all other parts are deleted. The part is assigned with section solid properties with ELFORM = 4. The keyword file is saved with a suitable name using the ‘.k’ extension.

4) Casing:

The given FE model of the casing is retained and all other parts are deleted. The part is assigned with section shell properties with ELFORM = 2 and thickness = 2 mm. The keyword file is saved with a suitable name using the ‘.k’ extension.

Parts Material Definition:

The MAT_ELASTIC material card with mid 11001 is assigned to part with PID 1001 i.e., bird.

The MAT_ELASTIC material card with mid 11003 is assigned to parts with PID 1003 and 1004 i.e., hub and casing.

The MAT_PIECEWISE_LINEAR_PLASTICITY material card with mid 11002 is assigned to part with PID 1002 i.e., blade. The data of the material that we have input to the blade material card through the curve is defined and referenced using the *Define_Curve with LCID 13001.

Assigning Material and Section to Parts:

All the Materials and Sections were assigned to their respective parts i.e. Hub, Casing, Blade, and Bird.

2. Boundary Conditions:

Boundary SPC:

As stated in the problem the casing has to be stationary, hence the degrees of freedom of nodes in the casing is constrained in all direction.

Initial Velocity:

1) The velocity of the engine blades and the birds are chosen so that blade failure can be seen within a short span of time. Hence, the angular velocity of the blade is assumed as 0.5 rad/ms in the *Initial_Velocity_Generation card.

2) The velocity of the bird striking the blade is assumed as 20 mm/ms. The cylindrical bird model is made to travel along its own axis and hit the blades.

3. Contact Cards:

1) In the case of bird impacting the fan blades directly, it is important to assign the proper contact between,

• Bird and fan blades,
• Fan blades and hub,
• Fan blades with other blades,
• Blades with casing.

2) Contact between bird and fan blades was defined by using the *Contact_Automatic_Nodes_to_Surface where the bird is the nodes and blades are surfaces. The impact of the bird with a blade induces reaction loads that counter the rotational forces of the blade thus deforming the blade enough or fracture, portions of the same blade may even come into contact with the remaining blade and casing.

2) The hub and blade are held together using the *Contact_Tied_Surface_to_Surface.

3) The contact between fan blades with other blades is defined using the *Contact_Automatic_Single_Surface.

4) The contact between damaged fan blades with the casing is defined using the *Contact_Surface_to_Surface.

4. Control Cards:

1) The *Control_Energy card is used to calculate hourglass energy, stonewall energy, sliding interface energy and Rayleigh energy.

2) The *Control_Termination is used to set the termination time which is set as 4 ms in this case.

3) The *Control_Hourglass is used to control the hourglass effect if present in the model during simulation.

5. Database Cards:

1) The time step value of 0.1 ms is given for the BINARY_D3PLOT and in the DATABASE_ASCII option for GLSTAT, MATSUM, RCFORC and SLEOUT.

2) *DATABASE_EXTENT_BINARY card with STRFLG =1, is used to compute the elastic strain in the model.

6. Including all the data in the Main file:

The bird, blade, hub, casing, material, boundary condition, contact and control database are in different input files. Hence to reference these input files to one main file, Open LS-PrePost, in the *INCLUDE card type the input keyword file names and insert it and save them as the main keyword file with the ‘.k’ extension. The reference includes a path that is such that the folder can be copied anywhere and it can correctly reference all the files and run the simulation.

7. Model Checking:

The model checking is done in the main file to find out for any errors or warnings in the model. It was found there are no errors or warnings.

8. Runumbering:

Renumbering is done by selecting individually the keywords like nodes, elements, parts etc and entering the Start ID value and click Set and Renumber. A consistent numbering approach is followed i.e., 100000+ for nodes, 500000+ for elements, and 1000+ for the parts. All other keywords are numbered within 10000-19999 while following a range for each one. These requirements are necessary for a professional setting when dealing with large models and multiple people working with the same model simultaneously.

Results and Discussion:

1. Animation Contours for Vonmises and Effective Plastic Strain:

From the simulation, it is observed that the bird striking the blades of the Aero Engine gets damaged. The stress and strain induced in the impacting zone are more such that the blades are broken and gets permanently deformed.

2. Maximum Vonmises Stress Plot:

An element 507697 of the blade at the impact zone is taken to plot the graph of v-m stress. From the graph, it is observed that the v-m stress induced in the element is maximum and reaches a value of 0.0702 GPa at around 3.1 ms which is more than the yield stress value of 0.013362 GPa of the blade material. Hence, the blade is damaged.

3. Maximum Effective Plastic Strain:

From the graph, it is observed that the Effective plastic strain induced in the element 508593 is maximum and reaches a value of 0.0875 at time 1.7 ms during the impact and remains the same till the end of the simulation. Hence, the blade is deformed.

4. Energy Plots:

From the energy plot graph, it is observed that the kinetic energy is reduced during the time of the bird striking the blade of Aero Engine. The internal energy is increased due to the decrease in kinetic energy. The hourglass energy remains 0 during the entire time of the simulation since we are using the ELFORM 16 (Fully Integrated Element Formulation) for the blade and bird also we are using the *Control_Hourglass to reduce the hourglass effect. The total energy remains constant throughout the simulation.

5. Mass Scaling:

1) Without Mass-Scaling:

The estimated time required to complete the simulation was shown as 1 hrs 33 mins but actually, it took only 40 mins to complete the task. A mass scaling approach can be adopted to reduce the runtime.

2) With Mass-Scaling:

Trial 1: TSSFAC = 0.9 and DT2MS = -6.14E-5

For, DT2MS = -6.14E-5, the estimated clock time to complete the simulation for TSSFAC= 0.9 is 1 hrs 33 mins. The percentage increase in mass is 0%.

Trial 2: TSSFAC = 0.9 and DT2MS = -8.00E-5

For, DT2MS = -8.00E-5, the estimated clock time to complete the simulation for TSSFAC= 0.9 is 2 hrs 48 mins. The percentage increase in mass is 0.00027193%.

Trial 3: TSSFAC = 0.9 and DT2MS = -1.00E-4

For, DT2MS = -1.00E-4, the estimated clock time to complete the simulation for TSSFAC= 0.9 is 1 hrs 7 mins. The percentage increase in mass is 0.0090455%.

Trial 4: TSSFAC = 0.9 and DT2MS = -2.50E-4

For, DT2MS = -2.50E-4, the estimated clock time to complete the simulation for TSSFAC= 0.9 is 27 mins. The percentage increase in mass is 8.3602%. 

As per industry standards, for a dynamic simulation, 10 % is the limit set for mass scaling. After simulating the current case I didn't see any significant changes between the one with mass scaling and the one without mass scaling. I also understand that we are doing mass scaling to only decreasing the simulation time for the current model.

6. Debugging:

For the initial simulation, the following parameters were assumed,

1. Poisson’s ratio for bird material = 0.475
2. Density = 1.00 e-06 kg/mm^3
3. ELFORM=2
4. Initial velocity of bird = 20 mm/ms
5. Termination time = 4 ms

Upon running the simulation a faced an error which was the frontal part of the bird was penetrating the blade as shown in the image given below.

Hence to resolve this issue, the keyword file was debugged by running the file for several iterations of assumed parameters with the help of a literature survey. The issue was resolved by changing the value of the density of material for the bird and also changing the elform to 16.

Conclusion:

1) The keyword file to simulate bird strike on an Aero Engine is created by following the necessary conditions and requirements.

2) In the simulation, the blades at the impact zone are damaged and deformed.

3) Mass scaling approach was adopted to reduce the runtime with the optimum value of DT2MS = -2.50E-4.

4) The initial keyword files were debugged to get a realistic simulation of the bird strike phenomenon.

5) Learned to create and organise as well as debug the input keyword files similar to a professional setting when dealing with large models and multiple people working with the same model simultaneously.