Mahera Musallam

Real-Time Compact Thermal Models for Health Management of Power Electronics

Implementation of real-time health assessment and thermal management of power electronic devices require real-time electrothermal models that can be used to predict the temperatures of device junctions, interfaces, etc., which cannot ordinarily be measured during service. This paper presents a real-time reduced-order compact thermal model, which is incorporated in a pulsewidth modulation and current controlled full bridge. An accurate representation of the dynamic thermal behavior was obtained experimentally and converted into a simplified multiexponential form then combined with lookup tables that provide estimates of the device losses based on measured values of the phase current. For interfaces away from the surface, such as solder layers, a validated Flotherm model is used to predict the temperatures of the hidden layers. Comparison of the real-time temperature estimates with IR measured values obtained from a high-speed IR camera showed that the reduced-order model was capable in estimating the modules temperatures over a range of modulation conditions. This real-time model is well-suited to the continuous monitoring of the internal behavior of the electrothermal effects within power electronic modules and can thus be used as part of a prognostic tool to provide knowledge through thermal cycling by combining with thermomechanical wear out models for health management of power electronics.

Mission Profile-Based Reliability Design and Real-Time Life Consumption Estimation in Power Electronics

Power electronics are efficient for conversion and conditioning of the electrical energy through a wide range of applications. Proper life consumption estimation methods applied for power electronics that can operate in real time under in-service mission profile conditions will not only provide an effective assessment of the products life expectancy but they can also deliver reliability design information. This is important to aid in manufacturing and thus helps in reducing costs and maximizing through-life availability. In this paper, a mission profile-based approach for real-time life consumption estimation which can be used for reliability design of power electronics is presented. The paper presents the use of electrothermal models coupled with physics-of-failure analysis by means of real-time counting algorithm to provide accurate life consumption estimations for power modules operating under in-service conditions. These models, when driven by the actual mission profiles, can be utilized to provide advanced warning of failures and thus deliver information that can be useful to meet particular application requirements for reliability at the design stage. To implement this approach, an example of two case studies using mission profiles of a metro-system and wind-turbines applications is presented.

A Physics-of-failure based Prognostic Method for Power Modules

This paper describes a physics of failure (PoF) based prognostic method for power electronic modules. This method allows the reliability performance of power modules to be assessed in real time. A compact thermal model was firstly constructed to investigate the relationship between the power dissipation and the temperature in the power module. Such relationship can be used for fast calculation of junction temperature and the temperatures at each interface inside power modules. The predicted temperature profile was then analyzed using a rainflow counting method so that the number of thermal cycles with different temperature ranges can be calculated. A reduced order thermo-mechanical model was also constructed to enable a fast calculation of the accumulated plastic strain in the solder material under different loading conditions. The information of plastic strains was then used in the lifetime prediction model to predict the reliability of the solder interconnect under each regular loading condition. Based on the linear damage rule and the number of cycles calculated from the rain flow counting algorithm, the accumulated damage in the power module over the whole period of usage can be predicted. As a demonstration, this method has been applied to a typical IGBT half bridge module used in aircraft applications.

Real-time life consumption power modules prognosis using on-line rainflow algorithm in metro applications

A real-time prognostic tool to predict life-time of IGBT power modules in a metro application is presented. Applying conventional life models (e.g. Coffin-Manson) for real applications is infeasible because these models are only applicable to cyclic data. Use of off-line rainflow algorithm is common solution but cannot be applied in real-time in its original form. This paper presents on-line life-estimation of the power modules using real-time rainflow coding algorithm. This technique is applied to an example metro application that requires use of cycle counting for an arbitrary load profile. The proposed method uses a stack-based implementation which employs a recursive algorithm to identify full and half cycles of the temperatures obtained as outputs from real-time compact thermal models. This then allows life-time models to be used to provide life consumption estimates. This method provides less complexity and more accurate on-line prediction for the studied modules failure mechanisms.

A prognostic assessment method for power electronics modules

This paper describes a prognostic method which combines the physics of failure models with probability reasoning algorithm. The measured real time data (temperature vs. time) was used as the loading profile for the PoF simulations. The response surface equation of the accumulated plastic strain in the solder interconnect in terms of two variables (average temperature, and temperature amplitude) was constructed. This response surface equation was incorporated into the lifetime model of solder interconnect, and therefore the remaining life time of the solder component under current loading condition was predicted. The predictions from PoF models were also used to calculate the conditional probability table for a Bayesian Network, which was used to take into account of the impacts of the health observations of each product in lifetime prediction. The prognostic prediction in the end was expressed as the probability for the product to survive the expected future usage. As a demonstration, this method was applied to an IGBT power module used for aircraft applications.

Real-time compact electronic thermal modelling for health monitoring

Temperature monitoring and management is emerging as an important tool for through-life cost optimization of high reliability power electronic systems. When combined with physics of failure based reliability analysis, such techniques can be employed to provide functions such as life consumption monitoring and prognostic maintenance scheduling. Requisite elements of such schemes are real-time electro-thermal models that can be used to predict the temperatures of device junctions, interfaces etc. that cannot ordinarily be measured during service. This paper presents a real-time reduced-order thermal model and its application to a half-bridge IGBT module including its associated thermal management system. Real-time electro-thermal modelling of the junction temperatures is accomplished using simplified circuit representations for the thermal model and look-up tables for the device loss estimation. The model is applied to a PWM modulated, current controlled full-bridge and implemented in dSPACE using a multi-rate computational process to improve efficiency. Validation is accomplished by comparing the model results with temperatures obtained from a high speed infrared camera. Results are presented comparing the performance of the model for a range of PWM modulation inputs, including low frequency sine and square waves.

In-service life consumption estimation in power modules

Health management and reliability form a fundamental part of the design and development cycle of electronic products. In this paper compact real-time thermal models are used to predict temperatures of inaccessible locations within the power module. These models are then combined with physics of failure based reliability analysis to provide in-service predictions of crack propagation in solder layers and at the bond wire joints as a result of thermal cycling. The temperature estimates are combined with lifetime based reliability models to provide a tool for life consumption monitoring. Rainflow counting algorithms are applied to the temperature vs. time data to extract the occurrence frequencies of different thermal cycling ranges. Knowledge of the life consumed for each different cycle then allows the remaining life time to be estimated under arbitrary operational conditions. The technique can be employed to provide functions such as life consumption monitoring and prognostic maintenance scheduling.

Real-time life expectancy estimation in power modules

The environmental and operating conditions applied to power electronic modules, such as temperature changes, load cycling, vibration, etc., cause degradation and ultimately failure in particular at interfaces between dissimilar materials, such as wire bonds and in solder layers. In this paper a compact real-time thermal model is used to predict the temperatures of the active device junctions and inaccessible locations such as solder layers within the power module. The temperature estimates are combined with lifetime based reliability models to provide a tool for life consumption monitoring. A rainflow counting method is applied to the temperature vs. time data to extract the occurrence frequencies of different thermal cycling ranges. Knowledge of the life consumed for each different cycle then allows the remaining life time to be estimated under arbitrary operational conditions through application of the Palmgren-Miner accumulated damage rule. Example results for life consumption in power module substrate solder layers are presented.

Application of coupled electro-thermal and physics-of-failure-based analysis to the design of accelerated life tests for power modules

In the reliability theme a central activity is to investigate, characterize and understand the contributory wear-out and overstress mechanisms to meet through-life reliability targets. For power modules, it is critical to understand the response of typical wear-out mechanisms, for example wire-bond lifting and solder degradation, to in-service environmental and load-induced thermal cycling. This paper presents the use of a reduced-order thermal model coupled with physics-of-failure-based life models to quantify the wear-out rates and life consumption for the dominant failure mechanisms under prospective in-service and qualification test conditions. When applied in the design of accelerated life and qualification tests it can be used to design tests that separate the failure mechanisms (e.g. wire-bond and substrate-solder) and provide predictions of conditions that yield a minimum elapsed test time. The combined approach provides a useful tool for reliability assessment and estimation of remaining useful life which can be used at the design stage or in-service. An example case study shows that it is possible to determine the actual power cycling frequency for which failure occurs in the shortest elapsed time. The results demonstrate that bond-wire degradation is the dominant failure mechanism for all power cycling conditions whereas substrate-solder failure dominates for externally applied (ambient or passive) thermal cycling.

Estimation and control of power electronic device temperature during operation with variable conducting current

Frequent variations in device power loss cause corresponding changes in operating temperature, which may adversely affect device reliability. A method for reducing the device temperature variations is introduced. A simplified third-order thermal model of the device is evaluated in real-time to estimate the instantaneous device temperature. The estimated temperature is used in a temperature control loop to reduce temperature variations by adjusting the device switching frequency. In this way, changes in device conduction loss are counteracted by varying the switching losses, so that the overall losses are substantially constant. The principle is applied to a MOSFET switching a dc load current at random intervals.