Structural Design Optimization of 25 MVA 132 kV Power ...€¦ · • Customer's requirement ......
Transcript of Structural Design Optimization of 25 MVA 132 kV Power ...€¦ · • Customer's requirement ......
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Structural Design Optimization of 25 MVA 132 kV
Power Transformer Tank Assembly
- Shantanu R Torvi, Prachi Khullar, Deosharan Roy
CG Global R&D Centre
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Contents
• Introduction
• Problem Description
• Parameter Optimization of transformer tank
• APDL macro and Optimization variables
• Integration with OptiSlang
• Results and Discussion
• Conclusion
• Future Scope
• References
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Introduction
• Customer's requirement - High Quality design, High performance, low cost.
• Optimization process is a complex process.
• FEA packages and optimization softwares enhanced the designer’s capabilities.
• Structural optimization and design improvements of transformer tank assembly
using ANSYS11 and OptiSlang software.
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Problem Description
• Power transformers
• The pressure and vacuum tests
• The criteria for passing the tests
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Problem Description (contd..)
• The objective is to minimize the weight of transformer tank keeping it structurally safe at
the same time. This has been achieved by effective usage of ANSYS and OptiSlang
capabilities.
• Formulation of problem includes:
A. Parameterization
B. State variables and objective function
C. APDL macro
D. Integrated use of ANSYS & OptiSlang
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Flowchart showing Automation of Optimization
Parameterization of Problem - Defining Design
variables, state variable and objective function
Design variables: Stiffener
thickness, width, breadth,
stiffener height etc.
State variables:
Maximum equivalent stress,
Maximum deflection
Objective Function:
Tank Mass
Defining range/limit for design variables
Constraint on permissible value of state variable
Solving the Problem for optimization
using suitable algorithm
APDL code for FE analysis
Integration with OptiSlang
Formulation of Problem
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
A. Parameterization
• The step on side wall was removed after optimization inputs from electrostatic analysis.
• The parameters which defined basic dimensions of transformer tank (i.e. length, height
and tank wall thickness) are left unaltered.
• Only stiffener arrangement and its geometry are the parameters which have been defined
as design variables.
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
A. Parameterization (contd..)
The design parameters which would be varied i.e. the driving
parameters are listed below:
• tb: stiffener thickness
• bw: stiffener breadth
• bh: stiffener height
• sl: stiffener length
• n: number of stiffeners present in
span A
• x1: distance of stiffener from side
• n2: number of stiffeners present in
span B
• x3: distance of stiffener in span B to last
stiffener in span A (as shown in figure)
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
B. State Variables and Objective Function
State Variable-
• The limitation on mechanical strength (stress limitation parameter)
• Instability (deflection limitation parameter)
Objective Function-
• Total mass of tank assembly
Note:
• The existing tank design is fabricated and passed the required quality tests and checks
• Hence, the vacuum test simulation results for existing tank are taken as benchmark for
comparison.
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
C. APDL Macro
• APDL macro has been used for modeling
transformer tank and stiffener arrangement in
ANSYS, applying boundary conditions, loads
and solving for vacuum test simulation.
• Boundary conditions and loads include:
- Full vacuum (101325 Pa) on tank inner walls
- Base and top of transformer tank constrained
in all directions
• APDL also writes final results-state variable
and objective function values in an output
text file. This is required for integration with
OptiSlang.
APDL Code
Output
From
APDL
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Input and state variables are needed to be re-defined in OptiSlang.
1. The range/limit in which the input design variables can vary are defined.
2. Constraint on permissible value of state variables:
• Maximum equivalent deflection < 10 mm
• Maximum equivalent stress < 688 MPa
• bw (stiffener width) => bh (stiffener height)
D. Integrated use of ANSYS with OptiSlang
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
3. OptiSlang uses non deterministic optimization method such as genetic algorithm and
evolutionary strategy for providing global optimum solutions with specified constraint
functions with combination of existing finite element analysis.
D. Integrated use of ANSYS with OptiSlang (contd..)
Evolutionary Algorithm
method has been used for
calculating optimized design
parameters.
Parameterization of input & state variables
from APDL code of baseline design
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Results and Discussion
Existing Design
Optimized Design
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Results and Discussion
Evolutionary Algorithm method used for calculating optimized tank design parameters.
Result of optimized design set:
Graph of objective function (tank mass in kg) along Y axis against the corresponding design set number along X axis
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Results and Discussion
State Variables for best design Input Variables for best design
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
FEA Simulation Results: Equivalent Stress plot
Existing Design
Modified Design
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
FEA Simulation Results: Equivalent deflection plot
Maximum deflection observed is 9.86 mm
Maximum deflection observed is 9.81 mm
Existing Design
Optimized Design
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Results Summary
Properties Existing tank design Optimized tank design Improvement
Equivalent stress (MPa) 688 568.9 17.31 %
Equivalent deflection (mm) 9.86 9.81 0.5 %
Tank mass (kg) 5370.82 4838.57 9.91 %
Reduction in tank mass (kg) 532.25
• Tank was further optimized by changing the profile used on short sides and inputs from
electrostatic and electromagnetic optimization.
• Optimized transformer tank and cover design has been validated under all mandatory loading
conditions.
• Recommended optimized design achieved an overall weight reduction of 15.6 % of existing tank
weight or 1037 kg.
• It also resulted in saving 2706 liters of oil.
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Conclusion
• Successful optimization of design parameters subject to constraints by adjusting 8
parameters of the tank geometry is obtained.
• Optimization of box stiffener geometry and its arrangement was achieved thereby
reducing the tank weight by 9.91 percent.
• The total displacements on the side walls and maximum equivalent stress were reduced.
• Fully automated optimization process for transformer tank is established.
• This methodology can be used for other products also with proper parameterization of problem.
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
Future Scope
• Objective function can be modified so as include the cost consideration.
• For instance if welding cost is also to be considered, then its value as a function of
stiffener length and numbers can be defined
Name : Shantanu R Torvi, Prachi Khullar, Deosharan Roy Division: CG Global R&D Centre Category: Confidential
References
• Oliver Konig, Marc Wintermantel. CAD-based Evolutionary Design Optimization with
CATIA V5. EVEN - Evolutionary Engineering AG, Zurich, Switzerland.
• ANSYS 11.0 Help
• OptiSlang 3.1.0 Help
• Prashant Bardia, Optimal Support Locations and Component Placement for
Electronic PCB Assembly towards minimizing Vibrations. October 2010 ANSYS India
Conference
• P Radhaswamy, P Velrajan, P Duraimurugan and T. Andrew. Automation of Aircraft
Component Roof Fitting Clamp Optimization using ANSYS Workbench 11. October
2010 ANSYS India Conference
• Fernando Batista, Hélder Mendes, and Emanuel Almeida. Structural Optimization of
Power Transformer Tank Covers. 3rd International Conference on Integrity,
Reliability and Failure, Porto/Portugal, 20-24 July 2009.