Multi-speed DC brushless motors?

[RichLeimbach]

I am currently using a brush-type DC motor to drive a relatively high inertia flywheel from 0 to 15K RPM in approx. .5 sec with a DC source (battery).
The problem I am running into is that with a single, always-on gear ratio, we spend too much time in the high current, low speed, low efficiency range (< 1/2 the max motor speed) and I'm having durability and heat related problems with the current motor. Electrically modulating the applied voltage does not help, as it just increases the time the motor spends in this inefficient range.
Besides finding a way to mechanically shift gear ratios, is there any way to increase the low-end efficiency of a DC motor? I am thinking of something along the lines of 2 speed AC induction motors, but in DC form. Any ideas?

 

[cbarn24050]

Hi, how did you measure your input/output power?

 

[RichLeimbach]

cbarn24050 ... 2 ways. Theoretical calculation and experimental verification.
Experimental testing is done like this: Input power can be measured by current draw and voltage across terminals. P=V*I Output power in this timeframe can be monitored by checking flywheel RPM and calculating stored kinetic energy (since there is no applied torque during the ramp-up, all of the motor output energy has gone into kinetic energy in flywheel and rotor). KE = 1/2 J * w2. For the purposes of simplicity in this discussion, I averaged the instantaneuos values over the first 1000 RPM to get the numbers above. Actually at 0 RPM, Input is ~ 1100 Watts and Output is 0 Watts.

 

[RichLeimbach]

Follow-up clarification: There is some slight applied torque from the bearings, but compared to the inertial loading this is minimal. I've neglected this in the above method.

 

[ve7brz]

Hi Rich,

Ive been following this thread faithfully and it appears to me that there is no electrical solution to this problem unless we invent a remarkable new motor which posseses, or can be 'adjusted' in real time to operate at maximum efficiency over an extremely wide speed range!

A switchable winding DC motor such as you initially inquired about is obviously a step in this direction; unfortunately there seems to be no such beast.

The problem you face seems quite clear; you need to provide a (constant) torque for 500 msec in order to impart ~60 Watts of energy to the system by the most efficient method.

I would like a few more facts before tossing out some other ideas which I have. What is the duty cycle for this mechanism? How is the disk stopped? How critical are your size/weight restraints? Am I correct in assuming this is battery operated?

It's an interesting problem.

Bruce

 

[RichLeimbach]

Bruce,
You're right, this is exactly what I'm looking for (although I would rather impact approx 200 Watts for .5 s for a total of 100 Joules of energy).
Duty cycle is about 500 cycles per hour (possibly up to 1000 in future versions).
Flywheel is stopped during the engagement of the clutch and subsequent energy delivery (we store kenetic energy for .5s and deliver in a 5-10 millisecond burst).
Size and weight are both fairly critical. We can add more weight than we can size. Length of the motor is the most tightly controlled constraint (can't get longer than 3.25&quot. We could add .5 LB if needed and the can size could probably increase to 40mm or so.
You're right, we are battery operated (18V Sanyo Sub-C cells with ~ 3.5 milliohms resistance each).

Thanks for your help.

 

[ve7brz]

Hi Rich,

Thanks for the quick response. The reason I asked about duty cycle was to get a feel for the energy used (lost, wasted) while simply idling the flywheel in the approach I suggested earlier. Obviously using a flywheel to store energy is not a new concept in this design.

I realize this isn't an electric solution but my mind returns to the flywheel as a way to keep a much smaller motor operating within a fairly narrow (and efficient) RPM band. Given the time frame, the two flywheels would only need to be coupled for the 500 msec acceleration window, so the current consumption of the clutch may not be a major issue (at least compared to the existing losses). On the other hand the clutch needs to be firm enough to bring the second flywheel up to speed in the time alloted with a minimum of energy loss (heat due to slippage). A reduction of motor speed is not required.

The motor must bring the faster flywheel back up to speed in the time between machine cycles. This will be the limiting factor unless the mechanism is to become even more complex.

Another thought that came to mind would be to use two motors coupled in such a way as to both a low speed/high torque section and a high speed/low torque section. This is conceptually the same as your two speed motor, and you've probably considered it already.

The other thing that pops into my head is an air or CO2 motor for torque but that creates a whole new design approach. Besides it isn't electrical and definitely does not belong in this thread.

Bruce

 

[RichLeimbach]

Bruce,
Thanks for the input. I'm going to take your suggeston for a 2 flywheel system and do some simulation and calculation with it. I'll post results when I get a chance to do it right.
We ARE looking at the 2 motor concept (packaging will be tough). Preliminary simulation shows we could get about the same benefit as with the 2 speed auto-shifting transmission. This was the original idea behind starting this thread (wanted to know if anyone had heard of getting two motors in the space for one).
I still feel that it is feasible for one motor to be produced with 2 different sets of windings that could be powered or unpowered depending on the current speed, giving 2 effective motors in the same package. Unfortunately, like you said, I have been unable to find a manufacturer that makes such a thing. Anybody have any leads on a manufactuer that would be willing to give this a shot or does something similar already?

 

[ve7brz]

I'm beginning to feel very guilty about proposing mechanical solutions in this forum. Perhaps this approach could be moved elsewhere?

In the meantime, I forgot one - engage a cog on a heavy flywheel (momentarily) to transfer an appropriate amount of energy to your load.

Bruce

 

[RichLeimbach]

added a thread for mechanical ideas here:

thread404-35775 Rich

 

[jbartos]

Suggestion/Comment to RichLeimbach (Mechanical) Nov 4, 2002
marked by ///\\Jbartos ... This is exactly my point. Increasing motor power does not give me any increase in efficiency.
///Disagree. See the Baldor motor 2hp has higher efficiency than Baldor motor 1.5 HP.\\ Since I need a certain total energy to come out of the motor and transfer to the flywheel, I will generate almost exactly the same amount of heat using any of these motors.
///Heat corresponds to RI**2. The larger the motor, the smaller the R and higher the I may be. Now, the motor designer can exaggerate the smallness of R to make the RI**2, and therefore the heat amount, extremely small. If it happens to be feasible, the superconductor would have the R negligible. In that case, the heat would be negligible.\\ The only difference is that with the more powerful motor, I am generating all of the heat energy in a shorter period of time.
///Very simplistic reasoning. See my comment above.\\\
Comparing this to the current situation as described above: Doubling the motor power would roughly cut my ramp-up time in half, but would mean I am putting 2000 watts of energy in and only getting 120 watts out to go from 0 to 1K RPM.
///This implies efficiency 100% x 120/2000 = 6% and 96% in losses, approximately.\\ I just doubled both sides of the equation. No gains, it just compressed the whole event.
I do agree that if the gear ratio was 1:1 instead of 1.59:1, our motor should be set at ~ 15K max.

 

[RichLeimbach]

Jbartos... Yes, I understand that a marginal increase in efficiency (which really has very little to do with rated power) would help me marginally. Moving from 75.5% to 78.5% would surely increase my overall efficiency by ~ 3% or so. This is the kind of marginal improvement I would expect to get by changing rotor bearings, reducing eddy current losses, or improving brush design slightly. It is not the kind of increase that will make a noticable impact on the issues we are discussing.

However, I agree that I have been overly simplistic (for the purposes of clarity and focus) so far in this post. Since the motor is not an isolated component in our actual system, but has to coexist and function efficiently with the battery described above, I should say that what I am most interested in is Total System Efficiency. For the purposes of this discussion I have tried to keep it confined to the motor.

All of the calculations, tests and simulations I have done show my statement &quot;doubling motor output power, just compresses heating into shorter period of time&quot; to be true. My calculations, based on the standard P=I^2*R, P=T*W, KE= 1/2J*w that we all know to define the situation, are a bit more lengthy than I would prefer to put in a post like this. If you want 'em, I might be able to find a way to post the simplistic version in Excel format.

Regarding the loss calculation above 120/2000 = 6%, this is true. We actually experience 94% loss as heat to get from 0 to 1K RPM (of course this goes down as the motor starts to move faster, but I have used this as an illustration). Taking this example, if I changed my overall efficiency upwards by 3%, I would expect to get 123.6 watts (120*1.03) out for the same 2000 watts in. The new efficiency moves to 6.18% for the same 0 to 1K RPM timeframe. This is of course an almost negligible gain at the low-end (the 3% won't really start helping much until the higher RPMs). I'm looking for a lot more.

 

[cbarn24050]

Hi, I still dont belive your 94% loss, perhaps you would be so kind as to explain HOW you MEASURED these values.

 

[RichLeimbach]

cbarn24050 ... As I have explained, the 94% loss is the loss from 0 to 1K RPM, there are of course different average efficiencies from 0 to 5K or from 14K to 15K (this efficiency is ~70% by the way).

The values were measured as explained above. I'll elaborate:

Input: Current and Voltage across terminals was measured. For this example (from 0 to 1K), avg current = 90A and average voltage = 11.1 (not 18V due to voltage drop across battery resistance). P=V*I = 11.1V * 90A= 1000 Watts.

Output: Flywheel RPM is recorded vs. time during one event (at the same time the above current and voltage readings are being measured). At 1000 RPM, Flywheel has a known and measured kinetic energy = .73J. Average power out of the motor is then P avg=Energy input / time. In this instance, it took the motor .012s to get from 0 to 1K RPM, so average power transfer from motor to flywheel= .73J / .012s = 60 watts.

60/1000 = 6% efficiency in this range. The reasons for this should be intuitively obvious if you look at the 0 to 1K RPM efficiency of any 15K RPM max motor (or the lowest 1/15th of any other motor chart). This is the efficiency that ANY standard motor would get in this range under these conditions.

 

[cbarn24050]

Hi again, still a bit mysterious! however i'll asume that you used a storage oscilloscope to measure voltage/current waveforms,it goes without saying that using a meter would be madness but i'll say it anyway.Your figures indicate an armature resistance of 100mohm, which would normaly be way too high for a motor with a emf of only 1.2v/1000rpm.Did you say that this was a PM motor? if so this sort of current is going to destroy the magnets sooner rather that later, I shan't tell you what it's doing to the battery I'll let you guess.Did you say that DC motors are inefficient at low revs? I don't think so, DC motors are used extensively because they are very efficient througout their operating range.Your problem is that to get high speed you need a weak field, but to get high torque you need a strong field, or to put it another way you need a motor with a field winding and a field control circuit of some sort.

 

[jbartos]

Comment to RichLeimbach (Mechanical) Nov 5, 2002 marked ///\\cbarn24050 ... As I have explained, the 94% loss is the loss from 0 to 1K RPM,
///If I understand you correctly, your intent is to either avoid the 94% loss or convert/store this loss in some energy storage.\\ there are of course different average efficiencies from 0 to 5K or from 14K to 15K (this efficiency is ~70% by the way).
///Naturally, the avoidance of the loss or the &quot;would be energy loss&quot; storage will wary.\\\

 

[RichLeimbach]

Jbartos.... Yes. This is exactly what I am looking to do. Trying to get more for less. Either more low-end output for the same power or the same output for less power.

 

[RichLeimbach]

cbarn24050 ... Now you've got it. The whole purpose of this post was to find a good way to get multiple fields at different RPMs. Still looking for any good suggestions / ideas on how to get this done.

Incidentally, I agree with your statement &quot;DC motors are used extensively because they are very efficient througout their operating range.&quot; However, we need an operating range that is much wider than required in most applications. DC motors are not efficient through MY operating range. Trying to find a way to change this.

 

[jbartos]

Suggestion: Some motors have designed their characteristics to the motor anticipated loads, e.g. the squirrel cage induction motors have A, B, C, D, and E designs. D design has substantially different characteristics from others. It is good for motor operated valves. What about if the dc motor has its characteristic designed to your peculiar needs?

 

[ve7brz]

Hi Rich,

I've come back to some of the basic principles of this application. The ideal solution for a 500 mSec acceleration would be to apply a constant torque to your flywheel for that period. Since motor torque is a function of motor current, in theory a constant current source would provide a constant torque to the system.

It also reasons that since individual motors have a current range in which they operate most efficiently, you would look for a motor which exhibits its best efficiency at about 17 A from your 12 V supply (using the numbers you've given).

Perhaps I have those two paragraphs out of order, but I'm sure you'll follow the reasoning nontheless. The RPM of the motor is irrelevant in regard to the quest for best motor efficiency. As long as your motor stays within allowed RPM you gear the motor up or down for your 15K flywheel RPM. At high motor RPM the motor armature itself becomes a more significant portion of your 'stored' energy but that can be corrected at the flywheel.

I've not followed the reasoning any further than this yet, but I'll continue to try to find flaws in it. By the way, although it doesn't apply to this concept, I don't know if you're aware that controllers are available for motors of this size which are not load sensitive. They are purely a speed control without regard to load, similar to large VFDs.

Bruce

 

[RichLeimbach]

Bruce,
This is just like current limiting as discussed above. The problem is that the output power is P= T*w, so you still have poor efficiency at the start, you have just constrained your current flow. In this example, output power would then rise linearly when graphed vs. RPM (from 0 to say 250 watts) while input power would stay constant (at say 300 watts) in all RPM ranges.
Clipping the current draw will help the battery efficiency and heating, but will not help the motor much since it will increase the motors residence time in the low-efficiency area.