Polyphase Induction Motor


Thread Starter

phil dryburgh

G'Day List

We have 200 large 415vac 50hz 3 phase induction motors employed at a coal prep plant.

3 core + E Aluminum core cables are run to these motors, sizes vary 185mm to 300mm and some drives have two cables in parallel. Over the last 10 years, possibly 10 drives have had overheating connections at the motor, causing damage to the
motors and cables, plus plant downtime.

In this time a program was taken to open each motor j/b, inspect for damage, re-crimp each lug, this is time consuming. The substations are inspected by thermal imaging on a 12 month
period and proved a godsend to minimising problems. This problem is raising its ugly head again, the problem is we can't or won't remove the covers with the motors running, so any ideas to address the possibly of locating hot connections before they destroy the tails inside the j/b would be appreciated.


Phil Dryburgh

[email protected]

Steve Bailey

Two thoughts:

1. Compare the current draw in all phases. The hot connection should be drawing more than the rest.

2. A thermocouple in the JB compared to one mounted somewhere else on the motor.
Two methods come to mind.

I'm going to go into a bit of detail on a related test in suggestion #2 because it may be of general use to electrical troubleshooters who may
happen upon this thread, so please bear with me.

1. Measure the connection temperature by clamping a thermocouple onto the lug of each of the phase leads, and sending the TC signals into either a
multi-channel TC alarm setpoint unit, or to three discrete, inexpensive PID controllers set up for on-off (not PID) operation. Usually these types of
controllers have at least one and possibly two alarm outputs as well as the setpoint output, so they can be set up to give a 'yellow' alert (connection hotter than normal, but still survivable), 'red' alert (starting to cook, shut down the machine soon!), and 'shut down now, dammit!' (place the contacts in the emergency stop circuit, or elsewhere to stop machine before
flames erupt ...).

While this is probably best, if the idea of typing TCs directly onto the phase leads doesn't appeal to you then, depending on how the leads are
arranged in the motor conduit box, it may be possible to place infrared 'thermocouple' units to 'look' at the connections in question without coming into physical contact with live parts. That's the pro, but two cons are that these critters are probably 4 to 8 times more expensive that a standard thermocouple, and their measurement accuracy is affected by placing them in a hot environment, and it will probably be necessary to offset their readings to get the 'true' connection temperature.

2. This is a somewhat speculative solution - I've never built the circuit described, but it should work dandy.

One of the ways I check for poor connections in motor power circuits is to measure the voltage drop across various points within a particular phase, measuring the line current, and calculating the wattage dissipated between
the these points.

For instance, say we have a typical motor circuit that has line fuses, a motor starter (with thermal overloads), and connections to the motor.

In this procedure I place one meter lead against the wire leading into the line side of fuse holder terminal block, and the other against the wire on the load side of the motor starter overload relay going to the motor, then measure the voltage drop (which should be less than a volt). I then measure the line amps, calculate the wattage, and repeat this for the other two
phases. The resulting voltage drops and wattages should be almost identical for each phase - in this case, if they are not, then the phase (or phases) with significantly higher drops have poor connections somewhere - in a terminal block, where the terminal blocks are bolted to the fuse holder clamps, perhaps within the fuse itself, a loose wires under one of ther terminal connections, a failing motor overload element, or failing motor starter contacts.

Once I find a suspicious phase then I narrow the scope of the test, perhaps checking the voltage drop only across the motor starter contact set, fuse block, thermal overload, etc. until the source of high voltage drop (hence high resistance connection) is located.

Now to the suggestion ... depending on how far you want to take this, it should be possible to 'automate' this troubleshooting procedure to give
on-line diagnostic and alarming capabilities. What I'm thinking is bolting a small lead - say, 12 AWG, onto the side of your connection electrically 'closest' to the motor (and do it for all three phases), and bringing these leads up to the control cabinet.

Connect matching leads at the motor starter load terminals, and route the wires into signal conditioners to give sufficient electrical isolation, and amplify the millivolt signal to something more usable (0-10 VDC, or 4-20 ma), feed these into a PLC or discrete controllers, and do something along the lines of the alarming described above in the temperature controller
example. The motors would need to be fully loaded (pulling a lot of amps) to determine good alarming and shutdown millivolt drop setpoints.

This could be taken even further with a bit more sophisticated control ... measure the millivolt drop between the motor lead, and the top of the fuse block, which would useful information about the motor starter, fusing, and wire connections between them all, but it would be necessary to isolate the sensing circuit shortly before turning off the motor starter, and only making it active shortly after pulling in the motor starter, or likely the signal isolator would go kerplunkt.

Problem with both approaches is initial cost, but you could probably make a case for them if the downtime cost is significant.

That's the nice thing about the voltage drop test - it is still effective "manually" if checked on a monthly basis, and merely requires a good DVM
(and the 'sense' lead brought out from the motor connections to somewhere close to the motor starter), but (and this is the hard part) requires
scrupulous record-keeping, and probably 15 minutes per motor.

Robert R. Stephens Pennzoil Products

CSI, now a division of Emerson makes a device called a Status Monitor that you place on the motor frame and it monitors performance, vibration, heating, HP and lots more. I have used them at a refinery with over 1500 motors and
found them very helpful in finding and eliminating the "Bad Actors" in the motor population. They are reasonably priced and you can reconfigure them and move them to another motor. Here's a link to their site:

Brainstorm ideas: (From crude to exotic!)

1) Thermally couple, (but insulate for > 5kV!), a lo-temp thermostat to each lug and wire to covered socket on j/box lid. Can then test status at any time with continuity tester.

2) Install a "Filled System" temp sensor into each j/box with gauge outside to give continuous reading of internal temp.

3) Epoxy T/Cs, (Or RTDs), in good thermal contact with termination lugs and connect to covered sockets on j/box lids so that you can read temps with handheld meter. (Needs > 5kV insulation),

4) Install quartz windows, (transparent to IR), in j/box lids so that you can do IR audit at any time without removing covers,

5) Permanently install Fiber Optic temp probe to each lug, read with portable instrument, ("Luxtron" Temp Measurement - expensive, maybe US$15-20 for the instrument. Probe has fluorescent material on tip, that when excited by a pulse of light from the instrument, fluoresces and then exponentially decays at a rate proportional to the temperature at the tip. The instrument excites and converts this decay into temp units).

Personally favor 3) or 4) as being both practical and low cost.
Hope this helps,

Tony Firth, Electrical Eng,
Quester Technology Inc.,Fremont, CA

George M. Dulik

Due to the large sample size (~200 units) you may be able to develop a statistically valid predictive model. You may be able to do non-intrusive temperature measurements of the junction boxes, feeder conduits, ambient location, and motor frame, and possibly power (amperage) drawn by the motor.

Arranging this data in a spreadsheet or database and evaluating for baseline and anomalies could have significant payoffs.

George M. Dulik
Maintenance Supervisor
Plant Engineering
Lawrence Livermore National Laboratory
Hello again, Phil

1) Measure surface temp of junction box, and motor's endbells. If j/b is higher than endbells by more than 2 deg C, then connections are suspect.

2) Measure phase currents and obtain average. If individual phase is higher than average by 2%, then connections are suspect.

Other Questions & Comments:

A) What is motor terminal voltage? Supply bus voltage?

B) Max Ambient Temp? Exposure to sunlight?

C) What is current capacity of 185mmq? Typically, 185mmq (Al-cond, plastic-insulated) is rated about 350 A, if 3-1/2c or 3c+e. Motor nominally rated at about 280 A. Thus not much margin for temp-rise. Above amp figure ignores derating for multiple cable installations.

D) Are motors Wye-Delta started?

E) Are motors TEFC, non-ventilated, open drip-proof, other?

The above should give you some food for thought.

Phil Corso, PE
Trip-A-Larm Corp

C S Chandrasekhar

There is a method of finding hot-connections using
Non-Contact type temperature measurement, using
Raytek's Pistol-grip IR thermometer. PC interface
facility is also present. Can try this...in most of the cases some flashes have been prevented using this.


Chandrasekhar CS

Alastair Fordyce

The quality of the crimp joint between the cable end and the cable lug is VERY important. The quality of the termination (bolting) onto the motor terminals is also important once you have crimped the cable lug correctly. Unfortunately aluminium cables are more likely to fail at joints than equivalent rated copper cables because of the difficulty of getting a good termination. You need to be very careful with cleaning of the bare cable end to remove oxide compounds and to use a quality jointing compound and a good quality bimetal type lug. And then after you have completed the repair the
aluminium cable work hardens with vibration and the joint resistance goes up and so does the temperature = failed joint again. Can you terminate in a separate junction box and run some
copper cables to the motor? Can you talk to a cable jointer/ tradesman of the old school who has plenty of experience with aluminium cables?

Best regards,

Alastair Fordyce TM.IPENZ,REA, NZCE,CQA
Systems Engineer
Bremca Industries Limited www.bremca.co.nz
PO Box 7169
tel +64 (3) 332 6370
fax +64 (3) 332 6377
e-mail [email protected]
Further to my Fri, May 12, 2000, 12:00pm, reply to Phil Dryburgh:


If a single connection is at fault, then, technically speaking, its phase will show a lower measurement than the unaffected phases. My reasoning is based on modeling the "poor" connection as a series resistor in the faulted phase. Sensitivity analysis reveals that the difference in measurement is statistically insignificant. In other words, the 2% delta value I stated is unlikely! Also, such measurement will reveal nothing if all three phases are similarly

Therefore, since contact resistance of an Al/Cu connection is exacerbated by both temperature cycling and vibration, I recommend that you address the connection method.

If space is a problem, then increase the size of the j/b, or add an external one. For the connection use aluminum setscrew-type lugs
on the cable, and copper ones for the motor's pigtails. The lugs should then be torque-bolted together, or to a piece of copper bus bar, using Belleville washers.

The method above was used many years ago to resolve a similar problem. It is probably commercially available, today. Try Connector & Terminating Device manufacturers such as T&B, Burndy, Alcoa, Gould-Shawmut, etc.

Based on your observation of cable insulation failure every 3-4 years, I estimate that the resultant temperature-rise caused by the poor connection is 15-25 °C greater than design.

In closing, I apologize for "illuminating" the wrong path.

Phil Corso, PE
Trip-A-Larm Corp

Bob Desrochers

Something doesn't make sense here.

Why are there repeated failures at the doghouses to begin with ?

Could you post typical motor and cable specs on this job ? Perhaps describe the method of termination and thermal OL sizing as well.

Sorry to disagree, Ricardo, but instead of adding more of an (obviously) bad stuff, why not just to change to old reliable cooper cables ?


Vitor Finkel [email protected]
P.O. Box 16061 tel (+55) 21 285-5641
22222.970 Rio de Janeiro Brazil fax (+55) 21 205-3339
Instrumenting a PWM synthesized sinusoidal waveform can be a little dicey from a few different perspectives. First, the fast switching currents create a great source of noise. This is sure to wreak havoc with the signals you are interested in viewing. Second, the bus from which these waveforms are developed is running at about 600VDC. By the time you include ringing on a set of motor leads 30m long, you are likely going to see voltages on the order of 1000-1400 volts depending on how the leads are run (twisted or in conduit or ???) Measuring delta voltages sufficient to cause localized heating is difficult at best . . . not to mention

Better to stick with non-contact, non-conductive instrumentation methods, these will be cheaper and more direct not to mention less of a risk.

Ken Brown
Applied Motion Systems, Inc.

Phil Dryburgh

In answer to Bob, regards Hot joints in J/B's.

1 All large motors are cabled with Aluminium 3 C+E cables, Terminated with Bi-Metal Lugs with jointing compound then double crimped. Motor connector cables are silicon glass insulated copper cable. The motors are connected in star.

2 There are no connection block in J/B, tails from motor and cables are bolted together (12mm Bolts) empire tape, scotchcast putty tape, then
ranchem heat shrinks over connection. The idea being a spec from the engineer, that problems and inspection would be minimal, as well as providing insulation as barrier against possiable
moisture if it enters box. Not so, it takes time to strip back, possiably workmanship, when the place was built, crimps not up to scratch perhaps.
A program several years ago, all boxes were opened, all connections checked visually and recrimped. Re-insulated, motor direction tested.

3 Overloads are Sprecher CET3 EOL, ulitising CT's matched to the rating of each motor, they are outdated, parts are becoming a problem. Features are over temperature, 2 phasing, inverse current time constants for overload, locked rotor. Thermistors are fitted to each motor as additional protection, separate to the Electronic overload .

Example Recently a 132kw 415v 214amp pump motor, star conn, blew one phase lead off in the J/B, AL cable150(sq)mm 3 C+E, cause unknown, pumps running
at 85% of its rating but does have a habit of cavitating, due to the water balance in the sump, also these pumps are not in a good environment and have a habit of having a buildup on the motor, they do have roofs over motor for
protection. Well that's the facts, if you have any suggestions or comment, it would be

Phil Dryburgh
In response to Vitor's response to Ricardo's response to Phil Dryburgh's request for help, this is what I talked about many months ago in my
blurb about engineers who often "catch the mice, but let the elephant's go free."

Having investigated several power feeder (both cable and bus duct) "failures" attributed to the use of Aluminium, I can attest to the fact that in the preponderance of cases, the termination method is at fault. And, almost every failure resulted from connecting Al and Cu.

In this specific case, it is silly to enlarge or replace (unless the respondents were being facetious) Aluminum cables, when all it takes is
attention to the connection method.

The most common causes of splice or termination failures are: 1) insufficient care taken in the preparation of cable ends and/or bus to
eliminate oxide formation; 2) improper wrapping of the joint to exclude moisture penetration; and 3) omission of Belleville washers.

An aside, Belleville washers were invented by the Frenchman Julien F. Belleville in 1867!

Phil Corso, PE
Trip-A-Larm Corp
Updating earlier comments/suggestions to resolve Phil Dryburgh's problem:

Rather than current averaging, I suggest the use of a single-phase clamp-on kW-meter. Measuring one phase at a time will yield more positive results.

Single-phase power measurement is more sensitive to phase power value differences because of the current 'squaring' effect. Two-phase or three-phase kW measurement will not work, because they are 'totalizers.' Calculations show that this method will yield meaningful results to about 200 Hp, based on accuracy of the kW meter. Fortunately, the test can be carried out at the starter without having to open the motor's
junction box.

This method will also work for multiple outgoing feeder leads (eg, used for wye-delta starting or increasing feeder ampacity). In this case the
phase conductors, either individually or as a pair, must be accommodated by the 'window' of the clamp-on meter.

However, before you go thru an exercise in futility try it out in your electrical shop or your rewind vendor's facility.

Phil Corso, PE
Trip-A-Larm Corp
Honest, this is my last comment on the subject:

Analysis revealed that the parameter most sensitive to contact resistance is phase power factor. Thus, make sure that the clamp-on
Watt-meter can also provide a p.f. measurement.

If interested in the mathematical analysis contact me.

Phil Corso, PE
Trip-A-Lam Corp
I know I promised an end to my comments on this thread, but following is a brief response to specific questions:

When I referred to phase power factor sensitivity, I really meant the differences (from norm, from average, or from min/max values) of the per phase values of Amperes, Volts, Watts (kW), reactive Volt-Amperes (kVAR), and power factor (pf). Thus, a clamp-on meter having the ability to measure per phase values is essential. Connection of voltage leads and current direction through the ammeter "window" is also critical,
since they establish rotation sequence.

Manufacturers of meters having the capabilities noted above are:

Sorry for the persistence.

Phil Corso, PE
Trip-A-Larm Corp
(Deerfield Bch, FL)