design optimisation of a line-start pmsm...
TRANSCRIPT
Design Optimisation of a Line-start PMSM Considering Transient and Steady-state Performance Objectives
Albert Sorgdrager – Sasol Technology
Prof. Roger Wang - US
Dr. Andre Grobler - NWU
Place additional logo here
SAIEE Rotating Machine Technical Forum – 24 August 2017
Overview
• Introduction
• Preferred LS PMSM Design Approach
• Study Objective
• Proposed Optimisation Framework
• Design Criteria
• Implementation of the Optimisation Framework
• Experimental Investigation
• Conclusion
2
3
Introduction • LS PMSM Overview
• Machine Optimisation Overview
Introduction: LS-PMSM
4
Radial-type Topology Spoke-type Topology
Introduction: LS-PMSM
5
Synchronous Torque Curve
Introduction: LS-PMSM
6
Torque vs Slip
Introduction: Optimisation
7
Introduction: Optimisation
8
Parameter Performance
9
LS PMSM Design Approach • Design Optimisation Flow Diagram
• Limitation
LS PMSM Design Approach
10
LS PMSM Design Approach
11
Preferred design strategy:
Global vs Robust optimum:
LS PMSM Design Approach
12
4-Quadrant performance model:
13
Objectives
Develop an optimisation framework that: • Considers both transient and steady-state objectives. • Realises a stable machine design. • Has the ability to realise a balanced design. • Can be used for Parameter design and Sensitivity analysis.
Devine a transient performance objective that is: • Quantifiable. • Computationally inexpensive. • Easily implemented.
15
Proposed Optimisation Framework • Taguchi Method Overview
• Taguchi Based Regression Rate Framework
• Analytical Synchronisation Criteria
• Overall Evaluation Criteria
• Pareto Optimum
Taguchi Method Overview
• Fractional factor optimisation method • Form part of the DOE approach methodology. • Mainly used in quality control and manufacturing applications.
17
Design Breakdown: • Select arrays • Parameter range • Compile trail array • Set MSD • Set S/N plot • Set ANOVA
Main array
Noise array
Bigger-is-Better Smaller-is-Better Nominal-Target
S/N Plots
ANOVA Calculations
Parameter Interaction
Parameters Sum-of-Squares
Optimum Conditions
Confirmation Test
Adjust main array range
Taguchi Method Overview
Using the Taguchi method in a fully automated iterative optimisation framework requires: • The exclusion of the designer’s intervention during the
process. • A sub-method is required to adjust the range of a design
parameter for the next iteration.
A Taguchi based regression rate method proposed by W. Weng may be used to address the issue.
19
Taguchi Based Regression Rate Method
20
• Initial level difference:
• Ith+1 level difference:
• Termination criteria
Analytical Synchronisation Criteria
Method: Time-domain approach using a non-linear partial differential equation system The load equation for any candidate machine is the same however the maximum synchronisable load inertia for each candidate has to be determined. Maximum load inertia Critical machine inertia
21
maxlcr
rotor
JxJ
=maxl cr rotorJ x J=
Overall Evaluation Criteria
22
𝑂𝑂𝑂𝑂𝑂𝑂 = 𝑓𝑓 𝑤𝑤1,𝑤𝑤2 = 𝑤𝑤1𝑆𝑆𝑆𝑆
𝑆𝑆𝑆𝑆𝑚𝑚𝑚𝑚𝑚𝑚+ 𝑤𝑤2
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚
𝑂𝑂𝑂𝑂𝑂𝑂 = 𝑓𝑓 𝑤𝑤1,𝑤𝑤2 = 𝑤𝑤1𝑃𝑃𝑃𝑃
𝑃𝑃𝑃𝑃𝑚𝑚𝑚𝑚𝑚𝑚+ 𝑤𝑤2
𝑥𝑥𝑐𝑐𝑐𝑐𝑥𝑥𝑐𝑐𝑐𝑐 − 𝑚𝑚𝑚𝑚𝑚𝑚
With 𝑤𝑤1 + 𝑤𝑤2 = 1
𝑃𝑃𝑃𝑃𝑚𝑚𝑚𝑚𝑚𝑚 = 𝑓𝑓 1,0 and 𝑥𝑥𝑐𝑐𝑐𝑐 − 𝑚𝑚𝑚𝑚𝑚𝑚 = 𝑓𝑓(0,1)
Pareto Optimum
23 Objective 1
Obj
ectiv
e 2
24
Implementing the Optimisation Framework • Machine Specifications
• Topology
• Results
• Identifying the Balanced Design
• Verification of Transient Performance
25
Machine Specifications
Objective: Develop an LS PMSM rotor capable of synchronising
with a cooling fan. Load 14Nm, Jl of 0.18 kg.m2
Base machine: 4-pole, 2.2kW, W22 IM. line-to-line 525V.
25
26
Rotor Topology
26
Radial-type PM Duct Block-type slot
Optimisation Results and Design
27
Balance Design
Optimisation Results and Design
28
Balanced Optimum Design:
2D Time-step FEM Verification:
Jcr -5%
Jcr +5%
29
Experimental Investigation • Selected Design
• Test Set-up Steady-State Fan Synchronisation Critical Inertia Validation
• Test Results Validate Steady-state Model Validate Load Specific Design Approach Validate Critical Inertia Index
Selected Rotor Design
30
Design Variants:
3 sets of PM with different
widths:
M1 – 20mm
M2 – 30mm
M3 – 40mm
Test Set-up
31
Steady-State: • Power Factor • Efficiency
Transient: • Fan Synchronisation
Critical Inertia: • Load • Inertia
Test Results: Steady-State
32
Performance comparison between calculated and measured outputs
Test Results: Validating xcr
33
Test Results: Validating xcr
34
35
Conclusion • The use of the TBRR method for a multi objective design
optimisation of LS-PMSMs considering both steady-state and transient performances is valid.
• It has also been found that there exists a competing relationship between the PF and Jcr
• The TBRR method can accurately identify both global and robust optimum designs on the Pareto front.
• The analytical calculation of Jcr shows good agreement with that of the measured one close to rated load conditions.
Thank you for listening Questions?
Email: [email protected]