Download a printable 72k pdf version

Abstract: In the first paper[i], the keys to the implementation of a technology for predictive maintenance and reliability and the ability to estimate how much time is left prior to equipment failure was outlined.  The first part of the study was to establish a baseline for electric motor life time estimation using motor circuit analysis techniques was presented.  In the second part of the study, variations of operating conditions including environment, load, cycling, and others are evaluated to fine-tune the estimation.  The focus of the study is low voltage, random wound machines in a variety of environments.  Additional studies are in progress for medium voltage rotating machinery.

INTRODUCTION

Electric motor winding failure occurs over time based upon a variety of conditions.  In the first paper, the baseline for estimating time to failure was determined for a standard motor condition.  In this paper, several of the general causes for adjustments to the time to failure estimations will be included.

Considerations include: Environmental conditions; Load; Load variations; Severity of fault; Power supply condition; and, Mechanical condition.  Estimations are based upon a survey of tested motors, and changes to condition over time that has identified faults using MCA.   In this paper, we will review the environmental conditions, load and load variations with the remainder coming in following papers.

Motor Circuit Analysis (MCA) involves low voltage measurements of resistance, impedance, inductance, phase angle, current:frequency response and insulation to ground testing.  Winding condition is determined by comparing similar windings, such as one phase to another in a three phase motor, with variations identifying defects.  Resistance is used to identify loose or broken connections, impedance and inductance are compared in order to identify winding contamination or overheated windings, phase angle and current:frequency are used to identify inter-turn insulation degradation, and insulation resistance is used to identify insulation to ground faults.

WINDING CONTAMINATION

Winding contamination is identified by comparing the pattern of impedance and inductance.  Impedance will decrease, and fall towards inductance, as the insulation system begins to degrade due to contamination.  However, in most cases, the insulation life may be extended by removing from service, cleaning, dipping and baking (re-insulating) for a limited time after detection.  As the fault progresses, it will graduate to an insulation to ground fault or winding short.  The progress is, normally, insulation degradation, changes to phase angle or I/F, then failure.

Table 1: Winding Contamination Rating to Hours (example)

The ambient environment effects the progression of winding contamination degradation.  Table 1 identifies a rating system applied to the detection of winding contamination when testing monthly (see table in Paper 1).  Statistically, it was found that the mean time to failure after fault detection progressed by a simple natural log multiplier:

Equation 1: Winding Contamination Multiplier

e^-m * Hb

where m = multiplier and Hb = Base Hours

The multiplier was based upon the ambient conditions (See Table 2) and the base hours on the base rating tables in Paper 1.

Table 2: Ambient Conditions

WINDING SHORTS

A variety of conditions effect the time to failure following the detection of winding shorts.  Statistically, the variations in time to failure followed a simple logarithmic scale.   The dominant conditions include, for the purpose of this paper:

ü     Motor loading

ü     Start/Stop cycling

Equation 2: Log Scale Used

M = e^-b

Where M = multiplier and b = condition variable

Motor Loading

In Paper 1, the load was based upon an average of 75%.  Conditions vary, depending upon the application of the electric motor.  For instance, many fan and pump applications are only loaded to about 50% of load.  In other cases, the manufacturer may have included the service factor in the design of the equipment.  The average motors evaluated were NEMA Design B, 1.15 service factor, TEFC. 

Table 3 indicates the condition variable value applied to Equation 2 for motor loading.  For instance, if a motor is loaded at 25% of nameplate rating, the resulting multiplier (M_L) would be M_L = e^-(-0.5) = 1.64.

Table 3: Load Conditions (Condition Variables)

Motor Start/Stop Cycling 

The number of starts and stops that a motor is capable of sustaining in an hour is primarily based upon the rotor design.  In the case of estimating time to failure, the start/stop frequency is based upon the stresses to the winding, both mechanically and electrically.  The majority of motors reviewed in this survey were random wound.  While the impact did appear to vary by motor size, for time and simplicity, we will consider a conservative cycling at 50 horsepower in an across the line starting condition.

Table 4: Start/Stop

Other Conditions 

A variety of other conditions exist that create variations in the time to failure estimation.  These conditions will be included in future papers.

Winding Short Multiplier

Using the days/hours found in Paper 1 as the base system (1 year = 4000 hours), an estimated time to failure can be calculated.

Equation 3: Winding Short Formula

M = M_L * M_S * Hours

Where M_L is the Motor Loading multiplier and M_S is the Motor Starting/Stopping multiplier.

Using this algorithm, a time estimation can be considered for trended equipment, in various loads and stop/start cycling environments, that has identified winding shorts using MCA.

TEST CASES

Following are several test cases using the system described in this paper.

Winding Contamination (10 horsepower)

A 10 horsepower motor in a plywood application (screen motor), operating 6,000 hours per year in a damp environment tested monthly.  Winding contamination indicated in February of 2003.  Applied a condition of 3 from Table 2.

Monthly time to failure estimation from Figure 1 of the first time to failure paper shows a potential life of 2000 hours.  Applied to Equation 1, the estimated life of this application was:

Example 1: Winding Contamination

e^-0.75 * 2000 = 945 hours

The motor was removed, cleaned, dipped and baked during the following month.  Winding contamination was visually confirmed.

Winding Short (150 horsepower)

A 150 horsepower motor in a paperboard application tested quarterly.  A turn to turn short developed with both phase angle and I/F changing by more than 3 points between tests.  The motor is started and stopped approximately twice per hour and is running at about 50% load.  The motor had been tripping off-line randomly.  Operates 4,000 hours per year.  The base hours, from Table 3 of Paper 1 show 335 hours as a baseline.

Example 2: Winding Short

e^-(-0.25) * e^-1.5 * 335 hours = 96 hours

The effective time to failure was estimated to occur in just over 1 week.  The motor catastrophically failed in operation within 3 weeks.

Winding Contamination (100 horsepower)

Winding contamination detected in a 100 horsepower motor which is tested every six months.  The motor operates 2,000 hours per year and the insulation resistance dropped to 45 MegOhms.  Ambient conditions indicated a condition 2 environment.

Example 3: Winding Contamination

e^-0.5 * 1,000 hours = 606 hours

The motor remained in operation for close to one year prior to being removed for rewind repair.

CONCLUSION

Time estimation of electric motor life following the detection of winding contamination or developing shorts, using motor circuit analysis measurements of resistance, impedance, inductance, phase angle, current/frequency response and insulation to ground can be performed.  Evaluation must be based upon averages with multipliers that indicate operating conditions.

The purpose of the estimation is to allow the diagnostic user to plan shutdowns, repairs and/or replacements.  The estimation should be considered as the average time to failure and any action should be taken well within the estimated period.

Additional work on low and medium voltage electric motors continues based upon real-world research and development and equipment history.  Uses of these formulae are at the risk of the user.

ABOUT THE AUTHOR

Dr Penrose is the General Manager of ALL-TEST Pro, A Division of BJM Corp.  BJM Corp is a manufacturer of motor circuit analysis instruments and electrical submersible pumps.  Dr Penrose is presently the Vice Chair of the Connecticut Section of IEEE and holds other elected and volunteer offices with IEEE and other professional organizations.

Contact:
Howard W Penrose, Ph.D.
General Manager, ALL-TEST Pro
A Division of BJM Corp
123 Spencer Plain Rd
Old Saybrook, CT 06475
Ph: 860 399-5937 ext 123
Fax: 860 399-3180
Email: hpenrose@alltestpro.com

Web: www.alltestpro.com

Cell: 860 575-3087

ACKNOWLEDGEMENTS

A large number of motor circuit analysis users submitted tens of thousands of sets of data, findings and history as part of these studies.

BIBLIOGRAPHY

Penrose, Howard W, Ph.D., “Estimating Motor Life Using Motor Circuit Analysis Measurement,” Proceedings: 2003 EIC/EMCW Conference, IEEE.