Soft-start or Variable Speed Drive? 2

[sry110]

I am a mechanical guy in need of electrical expertise.
I have a motor-driven gear system that has a considerable about of backlash to be taken up before the motor+gear sees the load. When it sees the load, there is a large impact that can damage mechanical components in our system. The impact is due to the motor (3-phase, 1800 RPM) hitting the stationary load at full speed.

I need a way to softly and/or slowly run the motor+gear up against the load to reduce the impact effect. Keep in mind that the motor+gear is completely unloaded (except for the weights and inertias of their own components which is negligible) from the moment it is energized until the driven load appears. Also, once the load is engaged and we are past that point of impact, I need the motor to produce its full locked rotor torque (typically ~300% of Full Load Torque for the motors we use) in order to start driving the load.

Some specific questions:
1) Will a basic soft starter keep the current reduced when there is no load on the motor, such that when our unloaded motor suddenly encounters the load the electrical torque will still be reduced to the set value? For example, let's say I set the soft starter to ramp from 20% to 100% current over 8 seconds (10% per second), and let's say the motor will run unloaded for 1 second until it hits the driven load. Will the motor be at 20% current (and correspondingly reduced torque) when the load suddenly appears, or will the soft starter have already decided the motor is at full speed / no load and therefore bypassed the soft start function altogether?

2) How much torque will a typical VSD allow the motor to produce? I could use a VSD to slowly run the unit up against the driven load, but once the load is engaged I need the motor to produce high starting torque to get it moving. Does the VSD allow the motor to generate 150% torque, 200% torque, higher, lower? I would typically count on ~300% locked rotor torque to get the load moving, but if the VSD only allows %150-200 then I would need to increase the motor size accordingly.

Thanks in advance for any insight.

 

[mikekilroy]

How far the 5hp motor shaft turns is indeed important; that is why I said we needed to know the gear ratio to consider how the vfd would react to the sudden engagement; we still need that ratio.

Yes, worm boxes are typically less than 400:1 max; again, we need that ratio for further calcs. My direct coupled example was just to show that on a vfd with 1:1 ratio, the engagement of the clutch would have no detrimental effect what so ever since the speed at engagement would be only 1rpm different than commanded. Likewise, with the ratio, we can look at what speed it will engage; it may be well less than 60 hz and again cause the motor to not want anywhere near that 600% current. The current spike wanted can be shown once the ratio is given; then the vfd size can be properly determined instead of guessed at. As I said before, we need that gear ratio.

SkottyUK, thanks for providing the missing data showing WHY the sss is there; that is why I asked if the load shaft might need to go faster than this motor at some point. If that was said earlier & I missed it, I apologize to everyone for questioning the need for the sss in the first place.

Sosry110, what is the gear ratio? I believe if you share that with us we can show you how to totally eliminate that 600% current spike you have today - with a simple vfd drive that is not even oversized one bit.

http://www.KilroyWasHere<dot>com

 

[sry110]

@ mikekilroy:
Thanks for the input, and all valid questions. Here are some additional details:
There is a 20.5:1 gear reduction between the motor output and the clutch, so 20 degrees at the clutch refers as 20 deg. x 20.5 = 410 degrees at the motor, or 1.14 revolutions.

The term "breakaway" refers to the level of torque required to start the Customer's shaft train turning from rest. Typically the "bumping" effect of the impact that occurs at clutch engagement generates enough torque to achieve breakaway. Once the shaft train breaks away, the torque required to accelerate the shaft to full turning gear speed is greatly diminished so the load on the motor drops off proportionally.

Since we only need to generate the breakaway torque momentarily, we do not need a motor with full load rating equal to the required breakaway torque (referred to the motor). Instead, we choose a motor whose locked rotor torque exceeds the required breakaway torque referred to the motor. This allows us to minimize the motor size. Looking at this application specifically:

Breakaway Torque required (at output shaft) = 550 lb-ft
Gear Ratio = 20.5:1
Gear Efficiency = 85%
Torque required from motor = 550 lb-ft / 20.5 / 0.85 = 31.6 lb-ft

Assuming my motor will generate 300% locked rotor torque, and it is an 1800 RPM motor, I can solve for the power rating that is required:
Power required (hp) = 31.6 lb-ft / 3.0 / 3 lb-ft/hp = 3.5 hp
NOTE: the "3 lb-ft/hp" term comes from the equation relating power (in horsepower) to torque (in lb-ft) and speed (in RPM), specifically:
Power (hp) = Torque (lb-ft) x Speed (RPM) / 5252
Plugging in 1800 RPM for speed and rearranging the equation we can say that for an 1800 RPM motor we get 3 lb-ft of torque for each 1 hp worth of rating.

Anyway....based on utilizing the motor's full locked rotor torque, theoretically I need a 3.5 hp motor. Of course this rating does not exist in NEMA world, so I would choose the next higher rating which is 5 hp. Now working through the math to figure out how much torque I could generate:

Max. allowable breakaway = 5 hp x 3 lb-ft/hp x 3.0 x 20.5 x 0.85 = 784 lb-ft

I only need 550 lb-ft, so my 5 hp motor could have a minimum of 211% LRT:
Min. breakaway = 5 hp x 3 lb-ft/hp x 2.11 x 20.5 x 0.85 = 552 lb-ft


So using my 5 hp, 1800 RPM motor with > 211% LRT, I need to be able to "creep" the clutch into engagement via the 20.5:1 gear reduction, and then once the clutch is engaged and the backlash is effectively removed from the system I need the motor to receive full line voltage so that it can provide its full Locked Rotor Torque and overcome the required breakaway torque of the Customer's shaft train.

 

[mikekilroy]

Since you bring up accelerating the load, can you tell us what the load inertia is? If you have never calculated it, can you tell us the size and materials or size and weights of the load? It sounds like it is just a long shaft of steel so should be easy for us to calculate the inertia?

Your added description helps; I will do my math when I get some free time later today, but in the meantime, I think what you consider the breakaway torque being the most significant and inertia to accel not very significant is maybe wrong. Have you ever measured this breakaway with a torque wrench? I suspect you will find it is only slightly higher than actual friction and your real current draw is accelerating that huge inertia in the 100msec that across the line motor starting causes. Remember that T=jw/t and your t time across the line starting is VERY short. Just the fact of putting a vfd on and increasing that t by orders of magnitude should solve you issue.

Just as last verification, you have no requirement to go from 0 to 1800rpm on this motor in a short period of time, right? It can be set to 60 seconds if you wished?

http://www.KilroyWasHere<dot>com

 

[waross]

I think that we may be trying to solve the wrong problem here.
We have a motor running unloaded at almost synchronous sped, Then we hit it with an instantaneous load.
The effect of the motor rotor inertia hitting a stationary load may far outweigh any concerns with peak torque and peak current.
The load hits the spinning motor and the gear train sees a very sharp torque transient. Probably the most severe torque effect of the starting cycle. Then the motor slows down. As the motor speed drops past rated speed the current will rise above rated current. But at this point the damage may have already been done by the initial torque transient. Given the method of engagement the issue may not be resolved by reducing back lash. The greatest backlash may well be in the clutch itself and that is fixed.
My understanding is that in most applications an SSS clutch is not intended to start a load, it is intended to allow a load to run above the speed of the clutch drive without over-driving the clutch and clutch drive mechanism.
As I understand it, the motor is up to speed unloaded and then an overload is almost instantly applied to the motor. A VFD or soft start won't do much good in that case. The damage may be done even before the motor drops below rated speed and before the current rises above rated full load current.
Possible solutions:
1: If this is for cool down duty, start the motor and engage the clutch while the load is still turning at above barring speed. Then the clutch and drive should take up the load gracefully and easily.
2: If this is for starting from rest, to establish the oil film, then engage the clutch while the motor is at rest. I understand that the clutch needs 20 degrees of rotation to engage but with a 20.5:1 reduction that is just over one revolution of an 1800 RPM motor. That should be no problem. Now the initial torque transient of engaging the clutch at speed has been avoided.
a> The motor may start easily now DOL.
b> The motor may still benefit from a ramped start with a VFD. I'll defer to jraef on the specifics of the VFD settings for this application.
DOL vs VFD:
If the backlash in the gearbox and in the drive from the motor to the gearbox is still an issue, You may benefit from the VFD.
Consider a design D motor. These motors are designed for impact loads such as shears and punch presses.
Look at the torque curve here:

http://www.baldor.com/pdf/manuals/PR2525.pdf

BUT WHATEVER.- GET THE CLUTCH ENGAGED IN THE FIRST TURN FROM REST OF THE DRIVE MOTOR SHAFT.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[mikekilroy]

BUT WHATEVER.- GET THE CLUTCH ENGAGED IN THE FIRST TURN FROM REST OF THE DRIVE MOTOR SHAFT.

Waross, that is indeed what is happening now that we have the gear ratio; THAT was indeed my whole point - without ALL the data, we did not know that the sss engages the load at only 1.14rev of the motor - from a stop. So doing that DOL was BAD but doing that with any ol vfd is piece of cake and there will be NO drastic BANG or spike.

http://www.KilroyWasHere<dot>com

 

[LionelHutz]

ScottyUK - the motor would need a major pull-up torque dip to be able to "stall" it and maintain a slow speed for any significant time. Even then, I doubt 40%-50% speed reduction is what is really desired here. The point is there is no speed control ability.

mikekilroy - The current at clutch engagement with ATL operation really has no bearing on the current using a VFD. Using ATL operation, at the point of clutch engagement, the motor either stops or comes close to stopping and then it is drawing locked rotor current. With a VFD, the operational idea is to either hold the speed very low or run with a low torque limit until the clutch engages at which point you have the load connected and everything at 0 speed. Then, the VFD basically starts the motor without any torque limit. It will be operating the motor above the breakdown torque speed where the current will be much less then locked-rotor current. There is absolutely no need or point in forcing the equivalent of an ATL start when using a VFD.

Many VFD's have a programming capability where you can enter operation steps. It might be possible to program the VFD to run fast until right before the coupling and then slow down to softly couple before accelerating again.

 

[waross]

We're on the same page Mike.
I understood that the motor was run up to speed before the clutch was engaged. Starting from stop, a small starter with some resistors may be cheaper than a VFD and will allow the slack to be taken up at low torque before going DOL.
I would consider a design D motor.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[jraef]

An aspect of this is now revealed in the size. At that size, there is probably no economic benefit to using a soft starter, you will likely pay as much or more for it compared to a VFD now, it's just the way the market has gone. But see below... *

Given such a short engagement time, this is looking like a simple application of what is called an "S-curve" ramp profile in a VFD. Most good ones will provide this. A heavy duty Sensorless Vector Control (SVC) drive would be the ticket because of its ability to make the motor produce its peak torque at any point, and again given the size and relative minor difference in cost, over sizing it per LionelHutz's recommendations would be prudent.

* The only possible consideration for continuing to pursue a soft starter would be if the system is going to run for long periods of time and energy losses are of concern, but speed control has no other value. A VFD, when run at constant speed, withh have about 3-5% throughput losses (in spite of what some salesmen will claim). If you change speed in some other fashion, this is always less than the losses in the other speed change method, so it's not usually an issue. But if you are NEVER changing speed and thus have no associated losses, it becomes one. A soft starter is bypassed once it is at full voltage, eliminating any losses. There are soft starters that have a "slow speed" feature (essentially a cycloconverter) that can be incorporated into the ramp profile. So you would hit the start button, the soft starter would turn at something like 15% speed for a second or two to engage your clutch, then you can have it go Across-The-Line immediately after that. But like I said, it will likely cost the same or maybe more than an equivalent VFD at this HP, so only consider it if the energy losses are significant to your application. To be honest, at 3.5HP loaded I would not get excited about 3% additional losses... The VFD gives you more flexibility.

"Will work for (the memory of) salami"

 

[LionelHutz]

LOL, that's what happens when I get distracted and delay my response.

Only having 1.14 revolutions of the motor before clutch engagement should allow a soft-starter to work for the application. Your ratio at 20.5:1 is much less then I expected. With the description of the gear I was expecting more like 20-40 rotations of the motor.

A good soft-starter will have a dual ramp setup. You probably need to program a very soft ramp and a very aggressive ramp and then have a short delay timer operated off the start command which switches from the soft ramp to the aggressive ramp. The aggressive ramp would basically be "go to full voltage".

 

[mikekilroy]

Finally made a few minutes spare time, so here is my conclusion that states any old vfd drive rated 5hp will do the job as long a reasonable accel RAMP is acceptable.

Using d=.5at^2 & v=at, I ran 4 scenarios and chart them here for the point where the sss engages. RAMP is time from 0-1800rpm programmed into the vfd.

RAMP(sec)....d(rev)....t(sec)....v(rpm)
ATL(DOL).....1.14......0.100.....1800
1.0..........1.14......0.276......500
10.0.........1.14......0.871......156
100.0........1.14......2.760.......50

We can assume worse case that on sss engagement, the motor will instantly stall to 0rpm. So ATL has 60hz going into it yet runs 0rpm momemtarily and hence smacks the engagement with full locked rotor torque and amps as proven empirically so far. Not good.

sry110, if you set the vfd to 1 sec ramp, it will be commanded to run only 500rpm with 16.6hz & 127v going to it. I need to get back to work so cannot think thru the rest right now, but Bill & Scotty sure could pick up and show you how to calculate, from your speed-torque-amps curve, the reduced current that will happen here. I would not recommend setting the ramp to 1 sec though.

10 sec prob is ok but look at the numbers for 100 sec ramp. 50rpm engagement speed - this is at normal slip freq of the motor! And so the motor will try to draw normal nameplate rated current to plow through it - IF you leave it a nice cheap skalar (non vector) drive! You will get a nice soft torque build up to your required 30-40#-ft, no bang, no current spike AT ALL, and it will soft start that sss engagement in my opinion. This is why I asked is you could stand a 100sec accel ramp earlier, before I ran the numbers to confirm this. No oversized drive required at all. Hope I did not go out into left field and miss something - I am sure the others here will let all know immediately if so

http://www.KilroyWasHere<dot>com

 

[waross]

sry110;
I am still not completely sure;
Do you engage the clutch at zero speed or at full speed???

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[waross]

The design D motor has maximum torque at zero speed and the torque curve from zero speed to full speed is not too many percent from linear. They accept and recover from shock loading very well. They accelerate difficult and high inertia loads well.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[ScottyUK]

Anecdotal because I no longer have access to the actual data, but from memory:

A gas turbine generator used an 11kW DOL motor to drive the barring gear. This was through the cyclodrive arrangement I previously mentioned. In the event of mains failure the system was designed to transfer to an alternate supply which was arranged as follows: the 110V station battery fed an inverter which produced 415V 3-ph, which in turn fed the input of a Brush 'Falcon' scalar VFD. The output of the VFD was connected to the barring motor.

The installation never once managed to break the shaft away on the inverter supply, regardless of what we tried with the drive configuration. We actually blew one drive up in the process of trying, during a black start trial witnessed by the TSO: Brush had a included marvellous option to disable all the on-board protection, which my former boss instructed us to do because we were going to fail the witnessed trial and he was under a lot of pressure to pass. You can guess the rest. The dead drive was later replaced with a half-decent Eurotherm 584S, and that too failed to break the shaft away, although from memory the drive would have had a better chance if the inverter feeding it had been more capable.

In the end the system was abandoned and replaced by a DC motor controlled by a resistor timestarter, and that scheme never failed to break the shaft away, except for the time when the bearing lift oil failed and the 60T rotor was left sitting on the bearing metal.

I am sceptical about a VFD solution for this application, but wait with interest.

 

[mikekilroy]

sry110: With Scotty's experience that breakaway may be WAY more than you thought, maybe you need to do my suggestion of:

Have you ever measured this breakaway with a torque wrench?

It could be your 'measured' 30-40#-ft breakaway is in fact 300-400#-ft as Scotty's experience suggests! Keep in mind with your DOL present method you are ASSUMING that locked rotor torque is what is doing the breakaway! It may be the KE=.5mv^2 kenetic energy in the rotating rotor that is supplying 300#-ft and the locked rotor only adding a measely 40#-ft more! You GOTTA get a torque wrench on that sss input or output shaft and MEASURE real life stiction!

Obviously if the stiction ('breakaway') torque is way high then that changes everything.

http://www.KilroyWasHere<dot>com

 

[mikekilroy]

Being a mech engr, you should be able to measure the striction with a torque wrench.

Also, back to inertia; as I asked before, if you tell us what the load inertia is, or the mechanics of the load so we can calculate it, we can then calculate the % that KE involvement is helping breakaway. With both these last pieces of data, you can be comfortable with trying a vfd or softstart I think.

Based on Scotty's experience in similar breakaway case, I would not be comfortable going the vfd route yet since the above 2 facts are MIA.

http://www.KilroyWasHere<dot>com

 

[LionelHutz]

With your calculation of the motor being required to develop 211% torque minimum you run a very high risk of failure using any old vfd drive rated for 5hp.

My previous VFD sizing recommendation stands. There are 2 ways to do it.

1. Use a VFD with sequencing or step logic and program it to run slow long enough to engage the clutch before accelerating to full speed.
2. Use a timer to hold a slow speed on the VFD long enough for the clutch to engage before "releasing" the VFD to run at full speed.

You clearly posted the numbers and information for the mechanical side so taking your numbers the above is what you need to use so the motor will start how you asked.

Sadly, it's reached the point in this thread that the background noise made me seriously considered not even bothering to respond....

 

[ozmosis]

The problem still seems to be the inerrant characteristics of the SSS clutch, and it's inclusion.
Coming from an electrical perspective and if someone had an application whereby a motor going DOL was causing issues with the high current demand and the same question was posed "soft start of VFD?", then you would have a similar discussion to above. However, if the request was to retain the contactor between soft start/VFD, then the general recommendation would be to remove the contactor, it adds no value and only adds problems. This is looking from an electrical analogy.
The 'solution', could be a change of design rather than a modification to what you already have, if the core issue is the impact when the clutch engages.
However, this is what happens when you ask electricals for a solution to a mech problem...

 

[sry110]

I apologize for the radio silence, gentlemen. I was out of the office at our assembly shop all day and did not want to put myself through the pain of typing anything out on my phone. And I have to say a big THANK YOU to everyone chiming in...this has turned out to be a very educational discussion (for me, at least).

Here are some additional data / comments regarding the application:

* On this particular project, the inertia (WR^2) of the drive train referred to the Turning Gear output shaft is 152 lb-ft^2. Referring this to the motor output shaft via the 20.5:1 speed reduction gives us 0.36 lb-ft^2 at the motor output shaft.

* This particular application has 20.5:1 speed reduction between the motor and the driven equipment, but this is actually toward the lower end of our typical turning gear ratios. The vast majority of our turning gear systems have total gear ratio falling between 76:1 and 300:1, and have either 1500RPM (for 50Hz jobs) or 1800RPM (for 60Hz jobs) motors. That being said, I'm using this project as a means to learn about VFD's and/or soft starters as they apply to our system, but I would like the option to also employ this method on higher ratio (lower speed) turning gears. So assuming my 300:1 ratio is the lowest speed turning gear I would design, and using an 1800RPM motor, this would be 16.7 revolutions of the motor shaft before the clutch completes its 20 degrees of free rotation to engage.

* Regarding verification of required breakaway torque by manual turning test (torque wrench): unfortunately we never have the opportunity to get this information on a new project because the turning gear is being designed and manufactured in parallel with the Customer's drivetrain. Therefore we do not have the opportunity to get a manual torque check of the specific drive train before designing the turning gear....in fact, in most cases the turning gear is installed very early in the drive train assembly process so that it can be used as the manual turning (indexing) tool!
That being said, for every design we do, we get the rotor weights, bearing diameters, bearing types, rotor inertias, etc., from the customer so that we can calculate the required breakaway torque for the drive train. Of course this is just an estimate, but we build in plenty of margin by assuming artificially high friction coefficients for the bearings....like 0.37 for a journal bearing that realistically is in the 0.15-0.22 range.

* Bill (waross) asked for clarification of the speed at clutch engagement. The scenario is this: The customer's shaft, to which the Output Component of the clutch is connected, is at rest. Therefore the clutch output component is also at rest. The turning gear motor is energized, which quickly gets the motor+gearbox+clutch input component up to full speed (I'm assuming that a nearly unloaded 4-pole AC motor will get up to near synchronous speed within a fraction of a revolution). So as my chief engineer would say, the motor is sitting there "twiddling its thumbs" at full speed as the clutch input component is sliding into mesh with the output component and then BANG, the clutch is engage and the motor+gear sees static friction torque and inertia of the customer's drive train.
So to answer your question maybe more concisely, when the clutch engages the input is at full turning gear speed, and the output is at Zero.

* No doubt there is a kinetic energy component at play here. The problem is that it's a tricky animal to calculate, especially when each drive train is different from the one before (we are talking about one-off, custom designs here, not mass-produced commodities). So what I'm trying to do is dial out the kinetic energy aspect altogether by eliminating the impact effect at clutch engagement and then relying on the motor's electrical torque, multiplied by the gear ratio and efficiency, to overcome the calculated required breakaway torque at the turning gear output shaft.

* I agree the problem is inherent of the SSS clutch. But the company I work for was built around, and has become very successful at, the implementation of SSS clutches for automatic turning gear systems. So changing to a different type of clutch is not an option, and I need to find a way to cope with its natural behavior.

* Bill suggested that I get the clutch engaged in the first turn from rest of the drive motor shaft....and I agree! That's exactly what I need to do, but I need a way to do it that either creeps it into engagement at a slow speed to dial out the kinetic energy impact effect (VFD), or engages the clutch at full speed but at a very low electrical torque (Soft start).
In my estimation, neither the kinetic energy "flywheel" effect nor the electrical torque output of the motor independently can cause damage to the mechanical components in our system. I believe it is the combination of the full speed impact and full breakdown torque of the motor occuring in that moment of impact that will exceed the strength of the components.

* Someone mentioned starting the turning gear when the customer's shaft is already turning (catch-on-the-fly) to gently engage the clutch. I agree, this is a beautiful functionality of the clutch, but I am tasked with designing a system that will not only catch-on-the-fly (during drive train coast down) but must also be able to start the train from rest (breakaway). This is non-negotiable.


General comment: In my perfect world, yet still assuming I must use an SSS clutch, I would size my motor and all mechanical components downstream such that the system could achieve breakaway using only the motor's electrical torque and the mechanical advantage of the gear ratio. I would employ a millwright to manually turn the input shaft of turning gear to pre-engage the SSS clutch before each turning gear start-up. I would use a DOL starter with my appropriately-sized motor to achieve breakaway and accelerate to full turning speed.
BUT....since I am in the business of supplying "automatic" turning gear systems, I can't require the customer to provide manual labor to pre-engage the clutch each time they want to use the turning gear.


Thanks again for all the input. If I have left any questions unanswered please let me know and I will post up any more info necessary to further the discussion.

 

[sry110]

I can't figure out how to edit my previous post, but I want to clarify one point. I wrote:
"* I agree the problem is inherent of the SSS clutch. But the company I work for was built around, and has become very successful at, the implementation of SSS clutches for automatic turning gear systems. So changing to a different type of clutch is not an option, and I need to find a way to cope with its natural behavior."

You probably ask "well if the company is so successful at this, why are you having so much trouble figuring this out???"

The issue here is that this particular turning gear is required to run at a much higher output speed than our typical design, complicated by the fact that we need to have a torque-release coupling on the output shaft to protect us from the customer's shaft trying to back-drive the turning gear (just trust me on this one, and like the SSS clutch it's non-negotiable), so this makes our system very sensitive to how much torque is being transmitted through it and I'm worried about the impact torque due to engaging at such a high relative speed. Not enough torque, we fail to break away the customer's shaft train. Too much torque, we trip our torque limiting coupling and fail to break away the customer's shaft train.
So instead of brute-forcing it like we normally do, i.e. just increasing the mechanical component size to the point where it's strong enough to absorb the impact torque spike without breaking, I want to learn how to be more finesse about it and use a readily available electronic device to do the job "right" and minimize my mechanical component sizes.

 

[mikekilroy]

LionelHutz says: Sadly, it's reached the point in this thread that the background noise made me seriously considered not even bothering to respond....

sry110, I apologize for making useless background noise. Good luck with your investigation; you were close to separating the details into proper columns so you can do more than guess in the future. But at least big L gave you the solution.

http://www.KilroyWasHere<dot>com