Three-phase induction machines are used extensively in modern industry due to their cost effectiveness, ruggedness and low maintenance requirements. A single industrial facility may have thousands of induction motors operating along its assembly lines. As a result of this coordinated operation, malfunction of an induction motor may incur financial losses not only associated with the individual motor’s repair or replacement, but also losses associated with the down time of the entire assembly line and the loss of productivity. For this reason, reliable motor operation is crucial in many industrial processes. To ensure reliable motor operation, protection devices, such as thermal relays, are widely used in modern industry. Condition monitoring of induction machines, the underlying technology in motor protective devices, has experienced rapid growth in recent years

A major task of induction machine condition monitoring is to provide accurate and reliable overload protection for motors. According to IEEE Industry Applications Society (IAS) and Electric Power Research Institute (EPRI) surveys, 35–40% of motor failures are related to the stator winding insulation and iron core [1]-[3]. These failures are primarily caused by severe operating conditions, such as cyclic overload operation; or harsh operating environments, such as those in the mining or petrochemical industries [4]. Although induction motors are rugged and reliable, the stator winding insulation failure is potentially destructive. It often leads to stator winding burnout and even total motor failure. Protection of the stator winding from insulation failure is the main theme of this work.

Stator Winding Insulation Failure

The organic material used for insulation in stator windings of an induction motor must work below a certain temperature limit. Operating above this temperature limit for short durations does not seriously affect the life of the motor, but prolonged operations beyond the permissible temperature limit will produce accelerated and irreversible deterioration of the stator winding insulation material. Such deterioration often expedites the motor’s aging process and eventually reduces the motor’s life. As a rule of thumb, the motor’s life is reduced by 50% for every 10°C increase above the stator winding temperature limit.

Since excessive thermal stress is identified from industry practice as the primary cause of stator winding insulation degradation, especially for small-size mains-fed induction machines, the National Electrical Manufacturers Association (NEMA) has established permissible temperature limits for the stator windings of an induction machine based on its insulation class to ensure its continuous and reliable operation.

There are several conditions under which the temperature limit can be exceeded, resulting in acceleration of stator winding insulation degradation: transient overloads, running overloads and abnormal cooling conditions [4]-[6]

Transient overloads and running overloads are related to two regions of motor operation [6]. The first region of motor operation is the transient overloads with 250 to 1000% full-load current. These overloads include motor starting, wherein the motor draws up to 6 times its rated current during acceleration; motor stall, wherein the motor fails to accelerate the load to the desired speed during its starting phase; and motor jam, wherein the motor is stopped during its normal operation due to a sudden mechanical lock. In each of these scenarios, a significant amount of heat is generated by the large amount of inrush current in the stator winding due to the stator I2R loss. The lack of ventilation, caused by the slow or even complete halt of rotor movement, makes it difficult for the heat to be dissipated [7]. Therefore, the transient overloads can be regarded as adiabatic processes with very fast thermal transients. Normally it takes between 25 to 30 seconds for a typical motor stator winding to reach 140°C rise above its ambient temperature during a locked rotor condition [6].

In most applications, general and special-purpose NEMA T-frame motors may be considered to be protected at transient overloads when NEMA Class 20 overload relays are used. These relays allow 6 times full-load current to pass through the motor for 20 seconds [8].

The second region of motor operation is the running overload with 1 to 2 times the full load current. In this region the motor is continuously running, thereby providing a certain degree of heat dissipation for the internal losses, and resulting in a gradual increase of the stator winding temperature.

Unlike the motor operation in transient overloads, the internal heat is transferred to the motor ambient by means of conduction and convection during running overloads. Therefore, the thermal time constant under this type of motor operation is far larger than that under the transient overloads. This thermal time constant is determined by a number of factors, such as the motor design, the rotor speed and the temperature of the surrounding air. As a result, while a definite time relay can be used to protect the motor from transient overloads, a more sophisticated scheme is needed to protect the motor from running overloads. This defines the scope of the research presented in this work.

Abnormal cooling conditions are another possible cause of stator winding temperature rising beyond its limit. Typically the cooling ability of a motor is reduced due to a defect or fault in any of the components in the motor’s cooling system. This often leads to an abnormal motor temperature rise. For instance, when the fins or casing of the motor is clogged with dust or other particles, transfer of motor internal heat to its ambient is obstructed, and as the result the motor temperature increases. Another example is when the cooling of the motor is compromised due to high ambient temperature. Standard motors are designed to operate at an ambient temperature below 40°C, therefore the insulation life decreases significantly as the motor ambient temperature increases. There are even more serious situations in motor cooling, caused either by a broken cooling fan or accidentally blocked air vents or ducts. All of them decrease the motor’s cooling ability and lead to possible motor failure.

To safeguard the stator winding from insulation failure and extend the motor life, the stator winding temperature must be continuously monitored. Whenever the stator winding temperature exceeds the permissible limit, the motor should be shut down to avoid damage to its stator winding insulation materials. Many techniques have been developed for induction motor protection under overload conditions to guarantee reliable motor operation. These techniques can be classified into 3 major categories:

1) Direct temperature measurement

2) Thermal model-based temperature estimation

3) Parameter-based temperature estimation

Direct temperature measurement of the stator winding temperature is performed using embedded thermocouples, thermally sensitive resistors (thermistors), resistive temperature detectors (RTDs) or infrared cameras [9]. Such thermal sensors are capable of providing reliable temperature readings at their installed locations. However, since most thermal stresses lead to localized failures inside the stator winding, where these thermal sensors are not installed, the direct temperature measurement may not provide complete overload protection for the whole stator winding. In addition, direct temperature measurement is only considered a cost-effective method for large machines. The installation of thermal sensors in small machines is extremely difficult and costly.

Thermal model-based temperature estimation is the most commonly used technique in motor overload protection. Dual-element time-delay fuses, eutectic alloy overload relays and microprocessor-based motor protective relays are 3 major types of protective devices based on the thermal models of induction machines.

The dual-element time-delay fuse, which is the most extensively used device for motor overload protection due to its low cost, consists of a short-circuit element and an overload element [8]. A properly sized dual-element time-delay fuse can provide protection for both shortcircuit and running overload conditions. However, each time the motor is overloaded, the fuse needs to be replaced.

The eutectic alloy overload relays are another type of motor protective relays based on the emulation of the thermal characteristics of the stator winding. When coordinated with the proper short-circuit protection, this type of overload relays is intended to protect the motor against overheating due to excessive over currents. Nevertheless, the thermal discrepancy between the eutectic alloy overload relays and the motors makes it difficult to match both heating and cooling characteristics of the motor under all thermal conditions. As a result, the device often trips the motor based on an approximate estimate of the stator winding temperature, and spurious trips are common with these devices [10].

common with these devices [10]. Among all devices using thermal model-based temperature estimation techniques, the microprocessor-based motor protective relays represent the state-of-the-art in motor protection [6]. To provide an estimate of the motor’s stator winding temperature, the microprocessor-based overload relay first calculates the power losses from the current measurements at motor terminals based on the induction motor equivalent circuit. The relay then derives the stator winding temperature from a thermal model for the induction motor. Thermal model-based temperature estimation provides an accurate and reliable temperature estimate when compared to fuses or eutectic alloy overload relays, thus ensuring complete motor overload protection. In addition, it can be adjusted easily for different classes of motors due to its flexible software-based algorithm. However, similar to fuses and eutectic alloy overload relays, it cannot respond to changes in the cooling capability of a motor, which are often caused by either a clogged motor casing or a broken ventilation fan.

Parameter-based temperature estimation technique presents an alternative method in estimating the stator winding temperature. Since resistance is a direct indicator of temperature, this type of method provides superior performance over the thermal model-based temperature estimation. Besides the high accuracy associated with the estimated stator winding temperature in this method, it is capable of responding to the changes in the motor cooling condition because the temperature variation is reflected immediately on the stator resistance estimate. Compared with the direct temperature measurements from either thermocouples or RTDs, this method requires no temperature detectors, and is therefore non-intrusive in nature and inexpensive.

Reference [11] presents a detailed method of calculating the stator resistance, Rs, and the rotor resistance, Rr, from the induction machine equivalent circuit. However, as indicated in [12], a direct estimate of stator resistance at high speed operation is extremely difficult and susceptible to parametric errors from rotor resistance and motor inductances. To avoid the large error in the estimated stator resistance, one method assumes a fixed ratio between Rs and Rr [13]. Since Rr is strongly dependent on the rotor frequency due to skin effect, while Rs is uncorrelated to rotor frequency, the stator resistance estimate obtained in this manner is not the ‘true’ stator resistance, and consequently it is not a direct indicator of stator temperature. Other researchers propose dc injection method for line-connected and soft-started induction machines for parameter-based temperature estimation [14]-[16]. However, the major problem with using dc injection for Rs estimation is the torque pulsation and the negative torque induced by the dc current component [15].

In addition to the aforementioned variety of devices used for overload protection, bimetallic thermal protectors are also a popular type of temperature monitoring device. They are typically used on fractional to small integral-horsepower (up to 5 hp) ac induction motors to provide built-in overheating protection.

Detecting abnormal cooling conditions during motor operation is also one important aspect of induction machine temperature monitoring. In case of a cooling system fault, the motor may operate at a higher temperature under the same load or thermal condition compared to when the cooling system is healthy. This results in accelerated stator winding insulation degradation.

In references [17]-[18], methods for detecting abnormal cooling situations are proposed. By comparing the difference in temperature estimated from the thermal model and the temperature estimated from the resistance, the motor cooling system is monitored. If the difference is beyond a predetermined threshold value, a fault signal is generated to indicate a malfunction in the motor’s cooling system. The implementation of this scheme requires complete knowledge of the motor electrical and thermal models. Sophisticated signal processing techniques are necessary to unify these two models and produce a reliable estimate.

It was shown in the previous sections that temperature monitoring of the stator winding is crucial to protecting not only an individual motor but also the whole industrial process driven by motors. This work focuses on the development and implementation of a fast, efficient and reliable algorithm to estimate the stator winding temperature online with only voltage and current measurements from the terminals of small to medium size mains-fed induction machines. In addition, motor cooling system condition monitoring is also explored for complete stator winding protection. The ultimate goal of this work is to provide a comprehensive set of algorithms for motor overload protection to the next generation microprocessor-based protective relays.

The development of a thermal monitoring tool begins with a thorough investigation of state-of-the-art techniques for stator temperature estimation. The thermal model-based temperature estimation technique, though simple and reliable, suffers from inaccuracies in the thermal model parameters. These inaccuracies often lead to conservative estimates of stator winding temperature, resulting in spurious trips and unnecessary interruption of the whole manufacturing process. On the other hand, the parameter-based temperature estimation technique, though accurate, is highly susceptible to the errors in the induction machine electrical parameters.

Theoretically, estimation of Rs using the negative or zero sequence model is insensitive to motor parameter errors; however, continuous monitoring of Rs in practice is virtually impossible since small negative sequence or zero sequence current often causes singularity problems in signal processing. If the negative or zero sequence currents are intentionally injected into the machine to obtain an estimate of Rs, the inherent motor asymmetry in different phases may also cause large errors in the Rs estimate. Other problems associated with the current injection method include the deterioration of motor performance due to torque pulsations and motor internal heating. For example, the dc injection technique proposed in references [15]-[16] usually introduces undesired torque pulsations and motor performance deterioration.

Based on the analysis of the pros and cons of both the thermal model-based temperature estimation techniques and the parameter-based techniques, a new method is suggested in this proposal. First, a hybrid thermal model (HTM) is proposed to correlate the stator temperature with the rotor temperature. This model also accounts for the disparities in thermal operating conditions for different motors of the same rating. Then the rotor temperature, obtained from the rotor resistance estimation, is regarded as an indicator of the motor’s thermal characteristics. The rotor temperature is used to tune the parameters in the HTM to reflect the specific motor’s cooling capability. Finally the HTM is run independently after the tuning process to provide an accurate and reliable estimate of the stator winding temperature.

An abnormal cooling condition in motor operation, such as a clogged motor casing or a broken ventilation fan is also considered in this work. The entire algorithm is fast, efficient and reliable, making it suitable for implementation in real time for protection purposes.

A brief overview of the results of previous research related to stator winding temperature estimation is given in Chapter 2. Chapter 3 analyzes the induction machine thermal behavior via networks consisting of thermal resistors and thermal capacitors. Hybrid thermal models are also proposed in this chapter based on the analysis of design and thermal behavior of small to medium size mains-fed induction machines. As the first step in implementing the stator winding temperature estimation scheme via the hybrid thermal model, Chapter 4 gives detailed procedures to obtain the induction machine rotor resistance from only current and voltage measurements. A detailed analysis of the algorithm requirement for the motor operating points is also covered in this chapter. Chapter 5 derives the rotor temperature from the estimated rotor resistance and then gives the general rules to tune the thermal parameters in the hybrid thermal model, so that the true motor cooling capability is reflected by the tuned thermal model. To validate the proposed stator winding temperature estimation scheme, Chapter 6 shows the detailed experimental setup, including the hardware platform and the software used for data acquisition. The experimental results for the induction machine online thermal condition monitoring are given in Chapter 7. Chapter 8 summarizes this work with conclusions and contributions. Recommendations for future work on the algorithms for the online adaptive stator winding temperature estimator are also described to provide a more accurate and reliable estimation of the stator winding temperature for small to medium size mains-fed induction machines.