online monitoring and sensorless temperature measurement...

P. Jenopaul et. al., International Journal of Research in Engineering and Social Sciences, ISSN 2249-9482,

Impact Factor: 6.301, Volume 07 Issue 05, May 2017, 0-18

http://indusedu.org Page 10

This work is licensed under a Creative Commons Attribution 4.0 International License

Online Monitoring and Sensorless

Temperature Measurement of Electrical

Machines

P. Jenopaul1, Jikku James Kurian

2, Barvinjegan.P

3, and Sreedevi.M

4

1,2,3,4(Department of Electrical and Electronics Engineering, St. Thomas College of Engineering and

Technology, Kerala, India)

Abstract: AC motors are the main workhorses in process industries. Their malfunction may not only lead to repair or replacement of the motor, but also cause significant financial losses due to unexpected process

downtime. Reliable thermal protection of AC motors is crucial for reducing the motor failure rate and

prolonging a motor’s lifetime. It is no secret that heat kills electric motors. A sensor less internal temperature

monitoring method for induction motors is proposed in this project. This method can be embedded in standard

motor drives, and is based on the stator windings resistance variation with temperature.

A small low frequency AC signal is superimposed to the power supply current and is injected to the

motor, in order to measure the stator resistance online. The proposed method has the advantage of requiring a

very low-level monitoring signal; hence the motor torque perturbations and additional power losses are

negligible. Furthermore, temperature estimations do not depend on the knowledge of any other motor

parameter, since the method is not based on a model. This makes the proposed method more robust than model-

based methods.

The measurement technique, which uses ac signals instead of DC signals, allows a significant reduction of the Monitoring Signal amplitude without affecting its accuracy, even in presence of high offset,

drift, or noise in the analog circuits. Hence, the proposed measurement technique solves the main problem

related to dc signal injection.

Keywords: Online Monitoring, Sensor less Temperature Measurement, Stator Resistance Analysis

I. INTRODUCTION 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 (1). 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 (2, 3, and 4).

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. Although induction

motors are rugged (5,6) and reliable, the stator winding insulation failure is potentially destructive(7,8). 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.

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].


Impact Factor: 6.301, Volume 07 Issue 05, May 2017, Page 10-18



II. PROPOSED METHOD The resistivity dependence on temperature of many materials is traditionally used as a temperature

transducer. It is used to transform the physical variable temperature into a parameter that can be electrically

measured. The most common devices are thermistors with Positive Temperature Coefficient (PTC) or Negative TC (NTC). These sensors are widely used in electric machine temperature measurement due to their robustness

and simplicity. The method proposed in this paper uses the resistivity variation with temperature of the stator

windings, which are usually made of copper. This variation is approximately linear, and can be described by the

following equation:

(1.1) Where α is the temperature coefficient, approximately equal to 3.82 × 10−3 1/◦C for copper. The

variables R and T represent the copper resistance and temperature, respectively, and the subindex 0 indicates the

initial state. Modern electric drives require an initial configuration procedure before the first operation. In this

initialization, the motor electric characteristics are entered by the user and the electric drive performs an

automatic parameter measurement (self-commissioning).

The proposed method requires a manual measurement of the motor temperature (T0), while the motor

drive measures the corresponding stator resistance (R0). This temperature can be obtained by measuring, after a

long-enough time with the motor turned off, the room temperature near the motor. Using (4.1) and the copper temperature coefficient, the measurement accuracy requirements can be computed taking into account the

desired temperature estimation accuracy. For instance, for overheating protection purposes 10 ◦C is an

acceptable precision.

This represents less than 10% the stator windings maximum temperature variation, from 25◦C (room

temperature) to 130◦C (maximum recommended temperature for class B insulation). Considering the minimum

required precision, the stator resistance should be measured with the following maximum error:

(1.2) The measurement method must guarantee that the RS measurement error does not exceed this value.

Measurement Method

As previously mentioned, proposals that measure RS online by injecting a small dc current are found in

literature. The main problem of using small dc signals is the analog circuits’ offset and drift. These factors

reduce the measurement precision if the signal levels are too low. A simple solution to this problem is the

utilization of ac instead of dc, hence offset and drift effects do not affect the measurements and the signal level

can be significantly reduced. As the signal level is reduced, its negative effects (e.g., power losses and torque

distortion) are also reduced. A small ac signal is used to measure the stator resistance online. The measurements are performed by

using the synchronous detection technique, used in LIAs [8]. LIAs are measurement instruments designed to

measure small ac signals in presence of noise with much higher levels. An LIA uses a sinusoidal reference

signal rX(t), and its quadrature version rY (t)

rx (t) = sin(2π f0t) (1.3)

ry(t) = cos(2π f0t). (1.4)

The input signal i(t) is composed by a sinusoidal signal of frequency f0 added to a generic function that

represents noise and harmonic distortion called n(t)

i(t) = Asin(2πf0 t + θ) + n(t). (1.5)

A digital LIA amplifies and digitizes this signal. The processing software multiplies the input signal by the in-phase and quadrature (shifted 90◦) reference components

Figure 4.1 Induction motor steady-state equivalent circuit with two voltage sources.





The slip is different for each frequency. Filtering-out the ac components and keeping the average value

(dc signal), two signals with estimations of the in-phase (real) and quadrature (imaginary) components of the

input signal are obtained

(1.6) The measurement method proposed in this paper consists of injecting a very low-frequency signal to the

stator. An embedded digital LIA measures the current and voltage of this signal. These measurements are used

to compute the stator resistance and then to estimate the motor temperature.

Figure 4.1 shows an induction motor steady-state equivalent circuit. The proposed method does not use

this model to estimate temperature; the equivalent circuit is only used in this section to study the method. The

stator is fed by two voltage sources with different frequencies, the power supply (VP ) and the Monitoring Signal (MS) Voltage Source (VM ). In order to make most of the stator current flow through the magnetizing inductance

(LM ), the MS frequency must be low enough.

Mathematically,

𝑗𝜔_𝑀 𝐿_𝑀 ≪ 𝑅𝑅

𝑆𝑀+ 𝑗𝜔𝑀𝐿𝐿𝑅 (1.7)

Where ωM is the MS frequency, LM is the magnetizing inductance, 𝑅𝑅 is the rotor resistance, LLR is the

stator leakage inductance, and 𝑆𝑀 is the normalized slip frequency measured with respect to the MS. This

relation can be satisfied by selecting a very low 𝜔𝑀 , making its validity independent of the motor parameters. If

this condition is satisfied, the equivalent stator impedance at frequency 𝜔𝑀 , seen at the motor terminals, is as

follows:

(1.8) Where,

(1.9)

The adequate ωM value depends on the machine electrical parameters, and in general, a correct choice is

0.1 Hz or lower. Figure 4.2 shows the 𝑅𝑆 measurement error using the approximation (4.6) as a function of the

MS frequency and the normalized slip (i.e., different rotor speeds) for a motor. For dc injection

the error is zero for any slip. For very low frequencies (i.e., 𝜔𝑀< 0.1 Hz), the error is low enough for temperature monitoring applications (<1%). If the MS frequency is lower, the measurement time is

higher and the time response can be too slow to detect fast temperature changes.

Figure 4.2 Expected RS measurement error as a function of slip (normalized with respect to ωM , with

different MS frequencies (from 0 to 0.5 Hz). Lighter lines indicate higher frequencies. The elected

frequency is indicated with a thicker line.

Figure 4.3: Proposed temperature monitoring system block diagram





A block diagram of the proposed scheme is shown in Fig.4.3. The lock-in amplification technique is

used to measure the voltage and current signals at frequency 𝜔𝑀 . From these values, the stator resistance is

computed, and the average stator windings temperature is estimated. The stator resistance is computed as

follows:

𝑅𝑒 =𝑉𝑋 𝐼𝑋 +𝑉𝑌 𝐼𝑌

𝐼𝑋2 +𝐼𝑌

2 (1.10)

Where the sub-indexes X and Y represent the real and imaginary parts of the signals, 𝑉𝑀 and𝐼𝑀 are the MS voltage and current measurements, respectively, obtained by using the lock-in technique.

If the MS frequency is low enough, the imaginary parts can be neglected and (4.9) can be

approximated to reduce the computational requirements. In the prototype presented in Section III, the MS

frequency is 0.1 Hz and it was experimentally verified that (12) is valid. Injected Signal Effects Any injected

MS used to monitor the motor temperature has negative incidence in motor operation. The main effects are

additional power dissipation in the motor windings, torque ripple and braking torque. Considering the equivalent

circuit of Fig. 4.1, these effects can be predicted.

In Fig.4.1, the stator is fed by two voltage sources, the power supply (Vp ) and the MS source (Vm ). At

constant speed, the equivalent circuit is a linear system. Hence, superposition can be applied to analyze individually the power supply and MS currents.

The stator current is given by,

(1.11)

Where IS and VS are the stator current and voltage, respectively. The ac component produces torque ripple. This ripple can produce a significant speed ripple if the load inertia or the ripple frequency is not high

enough. The dc component is composed by the sum of the torques produced by both frequencies. If the machine

is working as a motor, the low-frequency current has a negative slip.

Base Voltage = 4600V

Table1: Influence of the MS amplitude

MS Voltage Braking torque Torque ripple Power loss

0.05 pu 5.62E-3 pu 0.2 pu 5.22E-2 pu

0.005 pu 5.62E-5 pu 0.02 pu 5.22E-4 pu

Figure 4.4 Simulation of the electromagnetic torque with different MS amplitudes

The torque produced by the MS is negative and it counteracts the positive torque produced by the power

supply current. This property is used as braking technique in many applications. The additional power





dissipation generated by the MS is another negative effect of signal injection methods. The MS stator current

increases the stator winding losses. Furthermore, the negative torque component produces a mechanic power

dissipated in the rotor resistance. Considering that the MS frequency is very low, the additional power dissipated

by the MS in the stator and rotor resistances is as follows:

∆𝑃 = 𝐼𝑅𝑀2 𝑅𝑠 + 𝐼𝑅𝑀

2 𝑅𝑅 (1.12)

Where 𝐼𝑆 and 𝐼𝑅 are given by (4.10), and the M subindex indicates the MS components of both currents.

In Table I and Fig. 4.4, the effects for different MS amplitudes are compared. The MS voltage used in the

implementation presented in this paper is 0.005 pu, the value used in implementations that use dc injection is

generally higher than 0.05 pu.

Figure 4.4 shows a simulation of the electromagnetic torque in steady state using a model of the machine described in the Appendix. If the MS amplitude is not low enough, the torque ripple can be significant

(0.2 pu peak). In Table I, the braking torque, torque ripple, and additional power losses are compared. Power

losses are normalized with respect to the total resistive power losses without signal injection. Another secondary

effect produced by the injected signal is an eventual increment in the magnetic saturation of the machine. This

saturation cannot be predicted using the linear induction motor model. However, if the MS level is low enough

(i.e., its current is much lower than the nominal current), its effects on machine saturation are negligible. The

arguments presented in this section show the necessity of reducing the MS amplitude in order to minimize its

negative effects.

III. REAL IMPLEMENTATION Current and Voltage Measurements

The current and voltage measurement subsystem has central importance in all measurement systems.

Fast response to electromagnetic transient, high accuracy, good linearity, small offset and thermal drift in

current and voltage measurements are desired for the purpose of accurate estimation of the stator winding

temperature. In addition, proper signal conditioning techniques must be applied to suppress the noise and high

frequency interference inherent in the raw measurements. Fast analog to digital conversion is also a necessary

step in the data acquisition process. This A/D conversion enables the storage of measurement data on computer

hard drive or other data storage devices for the subsequent data analysis and visualization. In this section, the

hardware platform for the current and voltage measurements are first described. Then the software program,

which is written to facilitate the configuration and control of the data acquisition devices, is briefly discussed and its major functions are outlined.

Closed loop Hall effect transducers are used in the current and voltage measurement subsystem, as

shown in Figure 5.1(b). Compared to its open loop counterpart (Figure 5.1(a)), which amplifies the Hall

generator voltage to provide an output voltage, the closed loop Hall effect transducers use the Hall generator

voltage to create a compensation current in a secondary coil to create a total flux, as measured by the Hall

generator, equal to zero. Operating the Hall generator in a zero flux condition eliminates the drift of gain with

temperature. An additional advantage to this configuration is that the secondary winding will act as a current

transformer at higher frequencies, significantly extending the bandwidth and reducing the response time of the

transducer

Figure 5.1: The Hall Effect transducers





The microcontroller that has been used for this project is from PIC series. PIC microcontroller is the

first RISC based microcontroller fabricated in CMOS (Complementary Metal Oxide Semiconductor) that uses

separate bus for instruction and data allowing simultaneous access of program and data memory.

The main advantage of CMOS and RISC combination is low power consumption resulting in a very

small chip size with a small pin count. The main advantage of CMOS is that it has immunity to noise than other

fabrication techniques. Various microcontrollers offer different kinds of memories. EEPROM, EPROM, FLASH etc. are some

of the memories of which FLASH is the most recently developed. Technology that is used in pic16F877 is flash

technology, so that data is retained even when the power is switched off. Easy Programming and Erasing are

other features of PIC 16F877.

Figure 5.2: Over all circuit diagram of the proposed system

Figure 5.3: Hardware implementation of proposed system





IV. EXPERIMENTAL RESULTS Output of proposed system

The proposed system will continuously monitor the parameters (voltage, current, temperature). If the

value exceed the limit it will safely remove supply to motor. And the monitored parameters will send wireless through Bluetooth module and can be monitored using Bluetooth device, and analyses the parameters. The

experimental results show that the proposed method is able to accurately estimate the stator temperature in the

whole motor speed range. It measures the instantaneous current and the temperature calculated. The measured

values were displayed on the lcd display. If the temperature rises above 45 degree the controller will trip the

relay and remove the motor from the supply.

Figure6.1: wireless output displayed

The Figure 6.1 shows the system parameters displayed on the Bluetooth device wirelessly. The instantaneous

values were transmitted the through Bluetooth module. The speed, voltage, current, temperatures were

displayed.

Table2: Output Readings

SPEED CURRENT TEMPERATURE

1000 1.5 35

1100 1.8 36

1200 2 40

X-axis 10 div =0.5 A

Y-axis speed 10 div=50 rpm

Figure6.2: curve between speed and current

950

1000

1050

1100

1150

1200

1250

0 0.5 1 1.5 2 2.5

SP

EE

D

CURRENT

SPEED vs CURRENT

SPEED





From the graph it is clear that the current increases with the rise in speed. When we set different speed

the current will change. The temperature is calculated displayed.

X-axis 10 div= 50 rpm,

Y-axis 10 div=1 degrees

Figure6.3: Curve between speed and temperature

Figure6.3 shows the relationship between the temperature and speed. The temperature will increase

with the speed increases. The current will increase with the rise in speed.

X-axis 10 div=0.5 A

Y-axis 10 div=5 degrees

Figure6.4: Temperature v/s Current graph

Figure6.4 shows the temperature v/s current characteristic. The temperature increases with the increase in current. As the temperature increases the winding resistance increases and the current will increases.

V. CONCLUSION In this paper the internal temperature of induction motors, without using additional sensors, was

proposed in this project. The method injects a low-level signal to the stator windings in order to measure their

electric resistance. The proposed method has advantages with respect to the other methods found in literature.

The measurement technique, which uses ac signals instead of dc signals, allows a significant reduction

of the monitoring signal amplitude without affecting its accuracy, even in presence of high offset, drift, or noise

in the analog circuits. Hence, the proposed measurement technique solves the main problem related to dc signal injection. It is mathematically demonstrated that the MS level reduction allows a significant decrement in torque

distortion and additional power losses. The experimental results show that the proposed method is able to

accurately estimate the stator temperature in the whole motor speed range, even in current and speed transients,

and in field weakening region (very high speed). The mathematical analysis and the experimental prototype

presented in this paper are applied to induction machines.

However, the proposed technique is applicable to other types of AC machines, such as synchronous or

brushless machines. It is worth noticing that although in this paper the signal injection method is implemented

using the motor drive inverter; the same technique can be applied to line-connected motors by using external

3435363738394041

950 1000 1050 1100 1150 1200 1250

TEMP

SPEED

SPEED V/S TEMPERATURE

TEMP

0

10

20

30

40

50

0 0.5 1 1.5 2 2.5

TE

MP

ER

AT

UR

E

CURRENT

TEMPERATURE V/S CURRENT

temperatu…





devices. The only modification to the external device proposed in and can be implemented by software. Instead

of injecting a dc offset in one motor phase, the injected component must be a low-frequency ac. The digital

signal processing complexity required for the technique implementation is low, and it does not need high-speed

microprocessors. This project can be modified as a protective device for all electrical machines in future.

VI. REFERENCES [1] H. Bonnett and G. C. Soukup, “Cause and analysis of stator and rotor failures in three-phase squirrel-cage induction motors,”

IEEE Trans. Ind. Appl., vol. 28, no. 4, pp. 921–937, Jul. 2012.

[2] R. Krishnan, Electric Motor Drives: Modeling, Analysis and Control. Englewood Cliffs, NJ: Prentice-Hall, 2011.

[3] J. T. Boys and M. J. Miles, “Empirical thermal model for inverter-driven cage induction machines,” Inst. Electr. Eng. Proc.

Electr. Power Appl.,vol. 141, no. 6, pp. 360–372, 2014.

[4] Z. Lazarevic, R. Radosavljevic, and P. Osmokrovic, “A novel approach fortemperature estimation in squirrel-cage induction

motor without sensors,”IEEE Trans. Instrum. Meas., vol. 48, no. 3, pp. 753–757, Jun. 2016.

[5] S. B. Lee and T. G. Habetler, “A remote and sensor less thermal protection scheme for small line-connected AC machines,”

IEEE Trans. Ind. Appl.,vol. 39, no. 5, pp. 1323–1332, Sep./Oct. 2013.

[6] P. Milanfar and F. H. Lang, “Monitoring the thermal condition ofpermanent-magnet synchronous motors,” IEEE Trans. Aerosp.

Electron.Syst., vol. 32, no. 4, pp. 1421–1429, Oct. 2016.

[7] Ch. Kwon and S. D. Sudhoff, “An on-line rotor resistance estimator for induction machine drives,” in Proc. IEEE Int. Conf.

Electr. Mach. Drives,2015, pp. 392–397.

[8] M. O. Sonnaillon and F. J Bonetto, “A low cost, high performance, DSPbasedLock-In amplifier capable of measuring multiple

frequency sweeps simultaneously,” Rev. Sci. Instrum., vol. 76, pp. 024703-1–024703-7,2015.

online monitoring and sensorless temperature measurement...

Documents