How Does Regeneration Work

AC, DC, amps, volts and kilowatt. It's all discussed in here
EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

How Does Regeneration Work

Post by EVLearner » Tue, 10 Feb 2009, 02:02

As everything is starting to come together mentally, I am confused at how AC regeneration will work.

Assuming that I have a variable frequency drive, and the vehicle is moving and I want to slow down, is it simply a matter of slowing down the drive frequency so that the motor moves into generator mode?

If I brake gently, does the variable frequency decrease slightly and cause the power switches to back EMF power into the batteries?

If I brake hard, does the variable frequency come to almost zero and I really charge the battery with a heavy back EMF from the motor into the batteries?


User avatar
coulomb
Site Admin
Posts: 3780
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

How Does Regeneration Work

Post by coulomb » Tue, 10 Feb 2009, 02:42

EVLearner wrote: As everything is starting to come together mentally, I am confused at how AC regeneration will work.

Assuming that I have a variable frequency drive, and the vehicle is moving and I want to slow down, is it simply a matter of slowing down the drive frequency so that the motor moves into generator mode?
Yes, I believe so. I think you need to change the sign of the slip, which I believe means decrease the frequency of the drive below the actual speed of the rotor.

So for a 4-pole motor being driven at 50Hz, the slip might be 0.02 at a light load, so the actual rotor speed is 1500 * 0.98 = 1470 RPM. You could change the slip to say -0.01, by changing the frequency to 48.51 Hz, corresponding to 48.51*60/2 = 1455 RPM, so now the slip is (1455 - 1470)/1455 = -0.01. It's been a while, I hope I got the maths right, particularly regarding the slip s.
If I brake gently, does the variable frequency decrease slightly and cause the power switches to back EMF power into the batteries?
Yes, there are 6 switches, 3 bridges. Each bridge is bidirectional. At each instant, if the output voltage is greater than the input (bus) voltage, current and therefore power is transferred back to the bus. It all still works with AC as well. Bridges are very useful things.
If I brake hard, does the variable frequency come to almost zero and I really charge the battery with a heavy back EMF from the motor into the batteries?
I'll leave that to others. I think you can smoothly accelerate through zero RPM to power in reverse; obviously the controller has to do sensible things if that's not what the driver wants. I believe you only need a very small amount of slip to get a substantial torque (positive or negative) from the motor. So even for heavy braking, you won't be applying 30Hz when the rotor is near 50Hz, for example. If you do, you run the risk of exceeding breakdown torque in generation mode; the torque verses speed graph is quite symmetrical, so there is a maximum breaking torque that is similar, if not identical to, the motor breakdown torque. You don't want to lose the motor just as you are attempting a heavy braking force.

The controller will (if enabled) enable the brake resistor chopper if the bus voltage rises beyond a certain level, often 750v-780v. It's a good idea to have a resistor connected to that, in case you are braking heavily and the pack is suddenly at max voltage.

EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

How Does Regeneration Work

Post by EVLearner » Tue, 10 Feb 2009, 06:19

I'm almost there! Are you saying that by reducing the cyclic rate of the variable frequency drive to say 48.41 Hz, that the rotor in spinnning at say 1470 rpm, will try to slow down to 1455 rpm and produce a back emf (with current) that feeds backwards through the bypassing diodes to pull the rotor's angular velocity down?

I just don't get it (yet). I understand that the rotor spinning will cause a substantial EMF starting from the residual magnetism and building up towards saturation, but how does slowing the magnetic fields' rotation speed (angular velocity) create the back EMF to rechage the battery control the amount of braking?

antiscab
Senior Member
Posts: 2490
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

How Does Regeneration Work

Post by antiscab » Tue, 10 Feb 2009, 08:59

to get regen (torque in the direction opposite to rotation) you need the field direction of either the rotor or the stator to change direction (in the case of AC, this means to change direction in the opposite order).

for AC induction, the rotor field is changing at synchronous frequency.
the stator field is changing at the frequency of the supply (whatever frequency your vfd is set at).

when the supply frequency (and thus the stator frequency) goes below sync frequency, the current flow in the rotor reverses direction, relative to the stator.
thus you get opposing torque.

as with torque in the rotational direction, regen torque is controlled by motor voltage and frequency.

the motor back emf (voltage) is lower than battery voltage for *most* of the rpms that braking occurs in.
so to be able to regen down to a stop, regen enabled controllers have a 7th gate (in addition to the 6 which make up the main power stage), which allows the controller to use the motor inductance to form a boost converter. This allows motor voltage that is lower than battery voltage to charge the battery.

the boost converter approach is also used on DC regen systems including regen on series wound DC motors.
It isn't normally used on series DC motors though, due to other limitations.

Matt


Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells

User avatar
coulomb
Site Admin
Posts: 3780
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

How Does Regeneration Work

Post by coulomb » Tue, 10 Feb 2009, 16:13

EVLearner wrote: I'm almost there! Are you saying that by reducing the cyclic rate of the variable frequency drive to say 48.41 Hz, that the rotor in spinnning at say 1470 rpm, will try to slow down to 1455 rpm and produce a back emf (with current) that feeds backwards through the bypassing diodes to pull the rotor's angular velocity down?
Oops, I forgot to mention that voltage control (via the PWM duty cycle) is important here. Before regen, you will have been using a particular duty cycle, say 80%, so that 80% of the bus voltage is producing say 400vrms to supply the current load and also the magnetising current. When I say 80%, I mean 80% of the maximum mangled sine wave that you can apply to the motor; of course the instantaneous duty cycle is varying 50 or so times a second, and the three phases differ by 120 degrees. The back emf might be about 395vrms at this point; the 5v difference is used to draw current through the motor's internal impedance for the load and the magnetisation.

So as you change the frequency down to 48.41 Hz, you also adjust the duty cycle to say 78%. Now the voltage that the PWM is trying to produce is 390v; the back emf is still 395, so there is a difference of 5vrms in the opposite direction that the bridges will boost to more than bus voltage. Think of them as a buck boost DC stage, that is boosting from the instantaneous DC voltage of the motor to the bus voltage, but the instantaneous DC voltage of the motor is changing at some 48Hz, as is the duty cycle.

User avatar
weber
Site Admin
Posts: 2626
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

How Does Regeneration Work

Post by weber » Tue, 10 Feb 2009, 22:30

Hi EVLearner,

There are many things going on at once in the AC EV situation. To explain AC regenerative braking let's start with the simpler situation of a 2-pole 3-phase induction motor direct on line at say 415 V 50 Hz. Actually, let's simplify it a little more and give it a permanent magnet rotor instead of a squirrel cage rotor. One curved side of the rotor will be a permanent north pole and the opposite side a permanent south pole. This motor will run at exactly 3000 rpm (50 revolutions per second).

My apologies if I'm going too much back-to-basics on some parts of this, but maybe it will help someone else, who hasn't got even as far as you have.

The vector sum of the 3 fields from the 3 stator windings, being energised 120 degrees out of phase, results in a field of constant amplitude, whose direction rotates around the circle 50 times every second. This resulting field drags the opposite poles of the permanent-magnet rotor around with it. You can see an animation of this vector sum here:
http://en.wikipedia.org/wiki/3_phase#Re ... etic_field

Now imagine this motor mechanically connected to a turbine whose speed and torque (twisting force) can be varied. Imagine the turbine is controlled so it is running at 3000 rpm and there is zero torque between it and the motor. Everything is just coasting. There will be a small amount of line current but its sine wave will be pretty much 90 degrees out of phase with the line-voltage sine-wave, so it does not represent any real power. It is merely maintaining a magnetic field. Of course there will be a small amount of real power going into the motor to overcome friction and wire resistance and other losses, but we can ignore that for this exercise.

Now if you gradually try to make the turbine run slower than 3000 rpm, so it's producing a gradually-increasing dragging torque on the motor, the line-current sine-wave will gradually increase in amplitude and will move closer and closer to being in phase with the line-voltage sine-wave, which means the motor is drawing power from the line. This is like when an EV is going up a hill at constant speed.

If instead you gradually try to make the turbine spin _faster_ than 3000 rpm, so it is producing a gradually increasing _driving_ torque to the motor, the line-current sine-wave will gradually increase in amplitude and move closer to being 180 degrees out of phase with the line-voltage sine-wave, which means the motor is functioning as a generator and putting power back _into_ the line. This is like when an EV is going _down_ a hill at constant speed.

In this case, because we have a permanent magnet rotor, and we're on a fixed line frequency, the speed won't change at all. But changes in torque both positive and negative, will be reflected as changes in current both positive and negative (with respect to instantaneous line voltage). The system is entirely reversible as far as electrical to mechanical and mechanical to electrical energy conversion. No connections need to be reversed, nothing special needs to happen. Torque is simply reflected as current and vice versa.

There is no continuous slip _speed_ involved in the permanent magnet case, but the _angle_, by which the rotor poles lag behind the virtual poles of the rotating stator field, will increase with drag torque. And the angle, by which the rotor poles _lead_ the rotating stator field, increase with _driving_ torque. It is this lead or lag of the rotor poles that produces the torque and causes the increased current in the stator.

OK so far?

The situation for an induction motor is essentially the same, although it speeds up or slows down a little in response to torque. I'll attempt to explain why it's essentially the same and why it can't quite maintain synchronous speed.

So now we change the rotor from permanent magnet to a magnetically "soft" core, just like the stator. So the rotor has very little permanent magnetic field of its own and relies on a field being induced into it. We keep everything else the same. Direct on line, etc.

It is hard enough for some people just visualising the constant-speed rotating field of the stator. It is notoriously difficult to visualise what is going on in the _rotor_ of an induction motor. But let's face it. If it was easy, it wouldn't have taken a genius like Nikola Tesla to invent it. There are at least 3 different points-of-view that can be adopted: relative to the stator, relative to the rotor, and relative to the rotating stator field. Some animations would really help here, but I haven't found any online. (I haven't looked very hard either)

If we imagine the "coasting" case again, speeds synchronised, zero torque, and imagine ourselves shrunk down and walking around the surface of the rotor, impervious to centrifugal force and unable to see anything but the rotor surface and the magnetic fields in the air. If you're as old as me, you could put on the appropriate Jean Michel Jarre CD about now. Maybe it looks sort of like aurorae, but you probably should visualise it as the parallel field "lines" of the conventional diagrams.

Imagine that the stator is too far off in "space" to be seen. We would observe that the field lines are not moving relative to us, and when we send out expeditions around the rotor surface we find the field is entering a fixed region on one side of the rotor and exiting the opposite side, just as in the permanent magnet case. People on the inside of the the stator, looking at the rotor with powerful telescopes will say that this is because the rotor is spinning at the same speed as the rotating stator field they are creating. We say the rotor is spinning at synchronous speed.

But, I hear you ask, what's all this business about a squirrel cage? Why do we need those conducting aluminium bars embedded in the rotor iron and shorted together by aluminium rings at their ends?

First let's look at what would happen if we didn't have them. We try to slow down the turbine so it drags on the motor. In the permanent magnet case the rotor would "slip" just briefly until its poles were lagging by enough of an angle to give the required torque and cause the stator to draw the required current. With no permanent rotor poles and no conducting loops on the rotor, this cannot happen. The rotor poles will just follow the stator field no matter what the rotor does. The rotor can stop dead and the stator will draw no more current than before and there will never be any torque. It isn't a motor, it's a boat anchor.

[Note: I'm about to switch from talking about "field" to "flux". I don't want to go into the difference here as it would needlessly complicate things, but nor do I want to mis-state Faraday's law that follows. If you don't already know the difference, just take them to be the same thing for now.]

Now go back to the coasting case. This time our avatar on the surface of the rotor digs a canal, parallel to the shaft, along one side of the rotor, in the region midway between the induced poles, then he digs across one end (making a wiggle around the shaft) down the other side midway between the induced poles and across the other end, with another shaft-avoiding wiggle. Then he fills it with aluminium to make a short-circuited rectangular loop that completely encloses the magnetic flux that is passing sideways through the rotor. The plane of the loop is at right angles to the flux.

So far nothing different happens. I know this plot is slow-moving, but please bear with me.

Now we have to remember Faraday's law of electro-magnetic induction (so far we've only been talking about purely magnetic induction). This says that an EMF (electro-motive force) or voltage will be induced into a loop in proportion to the _rate_of_change_ of the quantity of magnetic flux enclosed in that loop. And if the loop is conductive, a current will flow according to Ohm's law.

You can think of the quantity of flux as the number of magnetic "lines" passing through the loop. (No, Virginia, there are no discrete magnetic lines in real life). Flux is measured in webers and a rate-of-change of one weber per second will produce an EMF of one volt. If the loop has a resistance of 0.01 ohm then 100 amps will flow around the loop.

In the coasting or synchronous case, the flux through the loop doesn't change, so no current flows in the loop. But lets see what happens _now_ when we try to slow down the turbine and hence the rotor. The stator field begins to rotate slowly relative to the rotor surface, and so flux begins moving _out_ of the rotor's conductive loop and thereby causes a current to flow in the loop. This (quite large) current makes its _own_ electromagnet that (according to Lenz's law) tends to _oppose_ the change that caused it. So its poles are in the _same_ direction as the original induced poles. This electromagnet will _not_ follow the rotating field of the stator but will be locked to the rotor, at right angles to the current loop that's causing it. Momentarily we will have the same situation as the permanent magnet rotor, with the rotor poles lagging behind the rotating stator field. Momentarily the same argument applies for the regenerative case, where we make the turbine run _faster_ than synchronous speed.

Why only momentarily?

Because as the loop rotates away from its original position, the rate of change of flux slows and so the current reduces, and eventually that electromagnet vanishes. Can you see what's coming now? How to fix that, so the offset electromagnet is permanent?

Our guy enlists some slaves to dig lots more canals around the planet, and eventually some bright spark says, "Hey, why are we digging new wiggles around the shaft every time and crossing over the previous canals. Why don't we just put a big fat ring around the shaft at each end.", and someone else says "Hey, if you were out in space this would look like a giant squirrel cage." And the other guys, on the stator with telescopes, aren't quite sure whether they are seeing canals or just atmospheric distortion. But I digress.

So now, just as the effect of one loop is fading, another one comes in to take its place. And in order for the rotor _poles_ to have the appropriate angular offset from the rotating virtual poles of the stator, the rotor _surface_ must continually rotate slowly relative to the stator field. This slow relative rotation is called "slip" and is usually measured as a (small) percentage of the stator field speed.

The more torque you apply, the more the rotor has to slip to maintain its own electromagnetic poles, which are offset from the stator field poles sufficiently to give the required torque and cause the stator to draw the required current. No different to the permanent magnet case in regard to regeneration, except for the slip. The slip will be positive or negative depending on whether the torque and current are positive or negative.

Then we need to consider the variable frequency drive. But I need to do something else for a while.

I will just note the terminology is that there are 3 _half_bridges in a 3 phase bridge, not 3 bridges. A half-bridge consists of two switching devices (called hi and lo) driven in anti-phase, in series across the DC bus (i.e. across the battery), with their midpoint being an AC (or "variable DC") line input/output. This only requires 6 switches (typically IGBTs). There is no need for a 7th switch to do AC regeneration, except to dump excess power into a resistor when the batteries can't take it all.

-- weber
(coulomb and weber are working together on an EV)

User avatar
coulomb
Site Admin
Posts: 3780
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

How Does Regeneration Work

Post by coulomb » Wed, 11 Feb 2009, 08:42

EVLearner wrote:If I brake hard, does the variable frequency come to almost zero and I really charge the battery with a heavy back EMF from the motor into the batteries?

If you brake hard or not, my understanding is that the back emf is about the same initially; it depends only on speed.

Think of a DC permanent magnet motor running at 3000 RPM, off a 100v supply, drawing 1000W. Let's say its resistance is 0.1 ohms. We neglect friction and windage for simplicity. We model the motor as an 0.1 ohm resistance in series with a 99v battery; the 99v battery represents the back emf. The 10A across the resistor drops the remaining volt.

Note that if you put this motor across the 100v supply when it's stalled, it will draw 1000A. A real DC traction motor would have a lower resistance than that, and hence a larger starting current; it's one of the reasons you really need a controller to slowly ramp up the voltage across the motor.

Now we brake hard, by replacing the 100v supply with a 99A constant current sink. Initially the sink looks like an 0.9 ohm resistor, so we have a 99v back EMF running into a total of 1.0 ohms (0.9+0.1), for a current of 100A, and a regenerative power of initially 99^2 * 0.9 = 8820W. This takes energy from the load, so the motor slows down to say 2800RPM, so then the back emf becomes 2800/3000 * 99 = 92.4v, but now the constant current sink has become a resistance of 0.833 ohms to maintain the 99A of regeneration current. The regeneration power has reduced to 99^2 * .833 = 8164W. This will smoothly reduce the speed of the motor down to about 300RPM, at which point the back emf is only 9.9v, and the constant current sink is a short circuit. At this point, the regeneration current will ramp down to zero.

Using the constant current sink roughly equates to the controller extracting a constant regenerative torque from the motor. If the motor is permanent magnet and the controller is a half bridge (thanks, Weber!), it can do this merely by changing the duty cycle of the switching devices.

Let's compare that with a gentle braking, where we replace the 99A constant current sink with a 9.9A constant current sink. Initially, the sink is equivalent to a 9.9 ohm resistor, making the total resistance 10 ohms. The regeneration power is 9.9^2 * 9.9 = 970W. At 2800 rpm (and we take a lot longer to get there), the back emf is again 2800/3000 * 99 = 92.4, and the constant current sink is equivalent to a resistance of 9.23 ohms, so the regenerative power is 9.9^2 * 9.23 = 905W. The speed will slow down to just 30 RPM before the constant current sink is a short circuit.

So when you brake hard, the current increases, because the effective load is a lower resistance, but the back EMF is the same at any particular speed. I assume that induction motors behave in a similar way (please correct me if I'm wrong here).

Edit: Also, just like you wouldn't put almost a short circuit across the DC motor, you wouldn't immediately apply almost zero RPM to the induction motor. You would reduce the applied voltage by 10-100RPM below where the rotor speed currently is (or where you predict it to be if you don't have a shaft encoder). Just as in the DC case, you would adjust the controller's duty cycle so that the effective "output" voltage from the controller is 0.1 to perhaps a few volts below where the back emf currently is.

Of course, you will eventually get to almost zero RPM for the induction motor, or almost zero "output" voltage for the DC motor.
Last edited by coulomb on Tue, 10 Feb 2009, 21:53, edited 1 time in total.

EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

How Does Regeneration Work

Post by EVLearner » Wed, 11 Feb 2009, 17:28

Thanks, I think that I get it "Of course, you will eventually get to almost zero RPM for the induction motor, or almost zero "output" voltage for the DC motor."

It seems to me that there is a virtually linear correlation (in the AC case) between drive frequency and shaft revs (frequency) - providing the back EMF is large enough to sustain braking current flow ie above say 300 rpm.

If this is the case then the ratio of the drive frequency compared to the shaft frequency could be looked at like a gear teeth "deadspot" where in the drive case the drive frequency is greater than the shaft frequency, but in the braking case the drive frequency is lower than the shaft frequency. Is this ratio substantially constant (with allowance for load variation)??

This is going to take some fancy closed loop feedback analysis to master, and the last time I did this stuff was about 1971! Don't know if I still have the books - but I know that I was taught this stuff for a purpose other than passing exams!

Interesting that the whole system collapses below say 300 rpm. Does this mean that below these revs (shaft frequency) the pulse width of the drive frequency needs to be narrowed so that the pulse width does not cause the motors magnetic circuits to go into (or towards) saturation, and the inductance then falls away causing far greater current peaks?

User avatar
coulomb
Site Admin
Posts: 3780
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

How Does Regeneration Work

Post by coulomb » Wed, 11 Feb 2009, 19:57

EVLearner wrote:It seems to me that there is a virtually linear correlation (in the AC case) between drive frequency and shaft revs (frequency) - providing the back EMF is large enough to sustain braking current flow ie above say 300 rpm.
One minus the ratio of (speed/drive frequency) is called the slip s. This is the main thing that you manipulate to adjust the torque (positive (motor) or negative (regen)).
Is this ratio substantially constant (with allowance for load variation)??
For a fixed torque demand (positive or negative), I believe so, yes.
Interesting that the whole system collapses below say 300 rpm.
Not really a collapse, just that the motor resistance is such that the demand torque can't be met. It can still regenerate, just at a lower rate (lower regen/braking torque).
Does this mean that below these revs (shaft frequency) the pulse width of the drive frequency needs to be narrowed so that the pulse width does not cause the motors magnetic circuits to go into (or towards) saturation, and the inductance then falls away causing far greater current peaks?
No, nothing to do with saturation, you just make the duty cycle such that it tries to multiply the back emf by as large a factor as you can manage, effectively short circuiting the stator. (I guess this would mean close to constant 50% duty cycle, with very little AC modulation, I'm not sure). The current will be limited by the internal resistance of the motor.

Remember that I chose a rather high internal resistance (a tenth of an ohm) for my example, and a rather high (99A) regen current (for a motor with such resistance). In a real vehicle, you'd get good regen down to almost zero speed, where it doesn't matter much.

EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

How Does Regeneration Work

Post by EVLearner » Sat, 14 Feb 2009, 15:59

So, mentally, I am making a 3 phase controller, using IGBTs from a nominal 200 V battery, runnning a 15 kW squirrel cage motor!

I can see how to accelerate (above the critical slip frequency) by increasing the drive frequency (and probably using an FPLA/EPROM/uPROC to get the 3 phases).

By putting the foot on the Go pedal, the drive freqency increases and the motor spins faster - but with an internal slew rate limitiation dependent on the pressure on the Go pedal and the difference between the current speeed and the predicted Go pedal speed.

Take the foot off the Go pedal and the drive frequency slowly decreases with an exponential decay.

Put the foot on the Stop pedal, and the drive frequency decreases at a faster exponential decay - that is somewhat dependent on the pressure on the Stop pedal and and the difference between the current speeed and the predicted Stop pedal speed. (then the friction brakes come in)

Apart from the 'deadband' frequency gap (of say 5% and -5% total) between accelerating and decelarating (and yes, I understand fairly well how slip works in Squirrel Cage motors - thanks anyway Weber) but I am a bit concerned on how the voltages pan out with the stator voltages being bigger than the supply voltage, and the current apparently going in reverse through the IGBTs - or does the current go through the reverse protection diodes (parallel with the IGBTs) and if that is the case then how do the IGBTs have control to limit the amount of braking!

If the IGBTs are switching on/off slower than the generated frequency caused by the spinning cage, then the switching is essentially 'negative phase shifted' but the currents are reversed as the spinning rotor causes the motor to be a generator, and I just can't picture how the IGBTs can be 'reverse switched' when there are reverse diodes already there (and I believe the diodes would become conductive at a slightly lower voltage than the IGBTs - if indeed the IGBTs conduct in reverse)!

The pennies have not dropped yet, and before I start making an EV, I need to get this part of the homework right, as braking regeneration is absolutely essential, because then, a homemade yet professional production run (proof of concept) of cheap, reliable and useful AC controllers that are engineered to match the EV mechanics will be the watershed for excellent EVs.


User avatar
Electrocycle
Senior Member
Posts: 985
Joined: Sun, 19 Oct 2008, 20:23
Real Name: Andrew
Location: Sydney
MSN: dumhed@dumhed.com
Contact:

How Does Regeneration Work

Post by Electrocycle » Sat, 14 Feb 2009, 21:48

you don't have to have IGBTs conducting in reverse, because it's in a bridge configuration, so you can connect each phase to the battery pack in either polarity.
The Engine Whisperer - fixer of things

EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

How Does Regeneration Work

Post by EVLearner » Sun, 15 Feb 2009, 00:18

Yeah, but when I think about it: if the battery is to be charged by the stator then the stator voltage will be larger than the battery voltage, and so the current flow will be in reverse to it being used as a motor. This being the case, the voltage across the IGBT will be reversed as the current flow is reversed (providing the current is in phase with the voltage)!

Now assume that the stator is slightly inductive, so the current slightly lags the voltage, and if we are switching the IGBTs (sinusoidally flattened PWM) at a slower frequency than the rotor is spinning (plus the slip percentage), then the rotor is being pulled down from its faster rotational speed, so how does the extra reverse current get back into the batteries?

Do we have to have a much narrower IGBT conduction arc (angle) so that the reversing diodes can switch in and absorb the extra power back into the Battery supply - then if so when (at what conduction angles) does this happen?

User avatar
Electrocycle
Senior Member
Posts: 985
Joined: Sun, 19 Oct 2008, 20:23
Real Name: Andrew
Location: Sydney
MSN: dumhed@dumhed.com
Contact:

How Does Regeneration Work

Post by Electrocycle » Sun, 15 Feb 2009, 01:39

on a DC motor and controller doing regen the controller is basically shorting out the motor briefly to build current, then releasing it - which gives a voltage spike when the magnetic field collapses.
That spike goes to the battery pack via the freewheel diodes, and different duty cycles can give more "boost" for more braking by letting the current build up higher before letting the field collapse.

I'm guessing that the AC motors are essentially doing the same thing but in the middle of a sine wave :)

Induction motors work well as generators, as long as you can get the field to start - which is easy with a VFD, so I guess the controller would pump some current into the motor, then clamp the coil, release it, and collect the spike - all synchronised in the sine wave.

So the controller becomes a PWMed sine wave load on the "generator" instead of being a sine wave power supply.
The Engine Whisperer - fixer of things

EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

How Does Regeneration Work

Post by EVLearner » Sun, 15 Feb 2009, 02:41

Yes - but I don't guess!

There is a very big void here that needs methodical analysis so that the regeneration and braking can work in concert with maximum efficiency.

As our illustrious PM keeps saying again, and again, and again....
"It's going to take some time"
(And that was 3 phases of it!)

User avatar
weber
Site Admin
Posts: 2626
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

How Does Regeneration Work

Post by weber » Sun, 15 Feb 2009, 03:00

Electrocycle wrote: on a DC motor and controller doing regen the controller is basically shorting out the motor briefly to build current, then releasing it - which gives a voltage spike when the magnetic field collapses.
That spike goes to the battery pack via the freewheel diodes, and different duty cycles can give more "boost" for more braking by letting the current build up higher before letting the field collapse.

I'm guessing that the AC motors are essentially doing the same thing but in the middle of a sine wave :)

Essentially right. But what everyone seems to be forgetting is that this is AC, so even when you're not regenerating the current is reversing itself many times per second anyway. Regeneration just means that the current sine-wave is close to 180 degrees out of phase with the voltage sine-wave.

A pulse-width-modulated (PWMed) half-bridge with an inductor is an inherently reversible thing. No, IGBTs don't conduct backwards, but the diodes across them work just fine. And yes, as far as the PWM frequency is concerned, a 150 Hz sine-wave is just slowly varying (and reversing) DC.
Electrocycle wrote: Induction motors work well as generators, as long as you can get the field to start - which is easy with a VFD, so I guess the controller would pump some current into the motor, then clamp the coil, release it, and collect the spike - all synchronised in the sine wave.
No need to pump any current "in" (which I'll take to mean "supply current _in_phase_ with the voltage") during regen. When slip is _negative_, current of _opposite_phase_ to the voltage is just what's needed to excite the rotor field. Nothing special needs to be done in this regard.

Same goes for DC motors that are separately excited, shunt excited or permanent magnet excited. It is only the _series_ excited DC motor that is weird in this regard, and needs special treatment.

A series excited DC motor does not have a neutral speed for _any_ given voltage (except zero). Torque doesn't go to zero except at rpm approaching infinity (i.e. motor destruction). This is because, as the back-EMF (induced voltage) approaches the applied voltage, the current drops, and so the field weakens, and so the speed needed to achieve sufficient back-EMF increases. So without special treatment the induced voltage (back-EMF) of a series excited DC motor can never exceed the applied voltage, which is what's needed to make it become a generator.
Electrocycle wrote: So the controller becomes a PWMed sine wave load on the "generator" instead of being a sine wave power supply.

Absolutely right.

-- weber

wombat6926
Noobie
Posts: 2
Joined: Mon, 09 Feb 2009, 13:13
Real Name: Mark Milligan
Location: VancouverCanada
MSN: wombat6926@hotmail.com
Contact:

How Does Regeneration Work

Post by wombat6926 » Sun, 15 Feb 2009, 17:09

Thanks for all the info Guys.

I am starting to really understand how this regen works on an ACIM. It has taken me a while to find our how it works. I really didn't have a clue but it seems so simple now. This forum has been my best source of information. I originally thought I would have to build some fancy complicated circuit to get regen Breaking.

It seem to me that programing the VFD controller to keep the slip negative and adjust it to how much breaking torque you need is the whole exercise. For Heavy breaking you want maximum negative slip (before breakaway) and for lighter breaking you can use a minimum neg slip.

You will have to put a Potentiometer on the Break pedal (same as the Gas?)to encode how much breaking force you want. And adjust the Hydraulic breaks to kick in a little later. So if you just touch the break pedal lightly the controller would kick into regen mode and bring you to a nice gentle stop. Heavier breaking would use more neg slip and the hydraulic brakes as well.

your controller could/would be your cruise control as well if you have cruise control set to on then it would maintain a constant speed, if off then it would switch to light regen/breaking if you took your foot off the gas.

My 2 cents
Mark.


User avatar
coulomb
Site Admin
Posts: 3780
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

How Does Regeneration Work

Post by coulomb » Sun, 15 Feb 2009, 17:58

wombat6926 wrote: You will have to put a Potentiometer on the Brake pedal (same as the Gas?) to encode how much breaking force you want. And adjust the Hydraulic brakes to kick in a little later. So if you just touch the break pedal lightly the controller would kick into regen mode and bring you to a nice gentle stop. Heavier breaking would use more neg slip and the hydraulic brakes as well.
Spot on. However, the concensus seems to be that for very heavy braking, you actually want to remove the regen, as it may interfere with front to back braking ratios (especially on a RWD vehicle), and of course ABS won't do its job if you have other braking happening.

So I think this might work:
no brake pedal: very light regen, similar to engine compression
light pedal: light regen
more pedal: medium regen, possibly light mechanical brakes
more pedal: light regen, medium mech brakes
heavy pedal: no regen, heavy mech brakes

To make the transition smooth, I think you'd really need an analogue sensor on the brake hydraulic pressure. But somehow done in a way that guarantees safety.

Or maybe this: use the existing brake lights sensor coming on (no danger of a leak causing brake failure) to indicate to a micro somewhere that the mech brakes have started coming on. (You will need some logic here, as you want anything more than very light regen to cause the brake lights to come on). The micro remembers where this pedal position is, and uses it to judge how much mech braking there is. As the mech brakes drift in pedal position, the transition is still smooth, as the micro adjusts where the regen starts and ends (pedal position wise) based on where the last known mech brake start was.

Or maybe only use the brake light sensor coming on to start reducing the regen, so the last known position isn't needed at all (the last known position may be invalid or corrupted if e.g. there was a power failure or spike).

User avatar
Electrocycle
Senior Member
Posts: 985
Joined: Sun, 19 Oct 2008, 20:23
Real Name: Andrew
Location: Sydney
MSN: dumhed@dumhed.com
Contact:

How Does Regeneration Work

Post by Electrocycle » Sun, 15 Feb 2009, 19:32

the brake light sensor is just a switch on the pedal itself, and usually comes on a bit before any braking is actually done.
It'd work well for enabling light regen because you can just touch the pedal to trigger it without using the hydraulic brakes.
The Engine Whisperer - fixer of things

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

How Does Regeneration Work

Post by bga » Sun, 15 Feb 2009, 19:47

Hi coulomb,

Spot on, my thoughts exactly.

I was going to do the sensing using
a) activation of the brake light - probably not necessary because of:
b) A link to a potentiomenter off the brake pedal. I thought this to be better than brake line pressure (0 to 4 MPA approx) because the initial travel of the brake pedal can be used before the wasteful friction brakes start.

I would like to try an experiment with a dashboard mounted potentiometer that controls the amount of throttle backoff regen, simulating a range of motors and gearings. It'd be good to be able to coast easily (low) or achieve efective throttle pedal speed control in creeping traffic (high).
Also high level of regen probably should light the brake lamps.

Last edited by bga on Sun, 15 Feb 2009, 08:57, edited 1 time in total.

User avatar
coulomb
Site Admin
Posts: 3780
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

How Does Regeneration Work

Post by coulomb » Sun, 15 Feb 2009, 20:37

bga wrote:A link to a potentiometer off the brake pedal. I thought this to be better than brake line pressure (0 to 4 MPA approx) ...

Yes, I still wanted to use a potentiometer off the brake pedal, but I wanted to use the brake light switch (I thought they were hydraulc) to allow for brake pad wear, which causes the point where brake effect starts (on the brake pedal pot) to vary. Otherwise, there might be a dead spot, or a sudden grab spot, both of which are undesirable Image.

But if it's just a switch on the pedal, then that's no help for brake wear. A pity.

HeadsUp
Groupie
Posts: 265
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

How Does Regeneration Work

Post by HeadsUp » Thu, 05 Mar 2009, 10:09

something else to consider is ;

a) keep a standard foot operated brake ... ( no issue with NCOP )

b) have a steering column fitted regen operating lever like what trucks use for exhaust brakes ( like a hand throttle next to the indicator stalk lever as used on some offroad 4x4 )
( except trucks only have an on/off switch on the lever for exhaust brakes , whereas we need a spring return on the lever , and potentiometer )

if you have driven a truck with exhaust braking , you know you get used to it very quickly and it doesnt alter your ability to incorporate both systems safely.
but you would still want adjustability in rate of speed retardation somehow


another option is to put a pressure sensor on the brake pedal , fit a small spring under the brake pedal to give progressive feel in regen braking in the first 40 mm of brake pedal travel ( which is normal ) , then program overlap in the controller so regen tapers off when brake hydraulic pressure increases ( with a timer built in to the regen so it doesnt come back on as soon as you lift your foot off the brake pedal ) , dont want any pogo-ing in regen brakes


i think my personal choice would be the steering column fitted, fingertip , pressure sensitive regen lever fitted with a return spring

( it complies with the KISS principle )
Last edited by HeadsUp on Wed, 04 Mar 2009, 23:16, edited 1 time in total.

fuzzy-hair-man
Groupie
Posts: 118
Joined: Wed, 12 Nov 2008, 16:40
Location: Canberra

How Does Regeneration Work

Post by fuzzy-hair-man » Thu, 05 Mar 2009, 18:51

coulomb wrote: Yes, I still wanted to use a potentiometer off the brake pedal, but I wanted to use the brake light switch (I thought they were hydraulc) to allow for brake pad wear, which causes the point where brake effect starts (on the brake pedal pot) to vary. Otherwise, there might be a dead spot, or a sudden grab spot, both of which are undesirable Image.

But if it's just a switch on the pedal, then that's no help for brake wear. A pity.
I'm not quite with you here coulomb my impression was as your pads wear down the pads are pushed further out towards the disks so even if your pads are worn down they still require the same pedal pressure and travel to move the pads out onto the disks, if you've had your brake fluid reseviour overflow whilst changing brake pads it's because you are moving all the brake fluid that was taking up the wear in your old pads back to the reseviour and it no longer has room.

So AFAIK the brake pedal free play should remain roughly constant regardless of wear, in the cases where the foot goes straight to the floor there's other problems I think like you've boiled the brake fluid or you have air in the lines. Having to 'chase' the brake pedal because you're pads are worn would be obviously bad.

Regarding really heavy regen perhaps you could make it so that it cannot be engaged using the brake pedal, ie really heavy regen (above what might be ICE engine braking) must be selected via a dash mounted pot, then if you need to emergency brake the preselected regen (heavy or not) can stay in place, I think if you started braking hard then all of a sudden find you've lost some of your braking effort because hard regen has turned off it would come as a nasty surprise and potentially cause an accident, I know if I miss a gear changing down into a turn I'm often surprised about how much extra pedal force is required to make up for the lack of engine braking with hard regen you are talking more braking force than ICE engine braking would deliver right?

ABS normally plays nicely with the engine braking developed by ICE vehicles right? doesn't it monitor how much the wheel turns relative to the other wheels to detect the wheel that locks? if hard RWD regen is in place it will find it doesn't have to brake very hard with the rear wheels relative to the front wheels, there is the possibility of regen being hard enough that the usual front to back balance of braking isn't there but say I'm going downhill in 2nd gear @ 70km/hr in my RWD ICE there's considerable ICE engine braking if I touch the pedals it will bring in the front brakes but the balance of overall braking won't be the same as if I had been in 4th gear and applied the same slowing effect with just the brakes. The thing in the 2nd gear ICE example that imbalance existed before the person touched the brakes, so the additional braking is applied in the usual front to back balance... if the regen applied because you hit the brake pedal is not all that severe then I'd guess there's not all that much to worry about. If you wanted to regain the balance and you had drum rear brakes you could do something like reduce the rear wheel cylinder size to make the rear brakes less effective relative to the fronts...

I might be wrong here though...

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

How Does Regeneration Work

Post by Johny » Thu, 05 Mar 2009, 19:07

The topic of controlling regen braking has been discussed a lot over the past few months.
My feeling is to go with natural regen braking similar to a 4 speed manual in third gear. Then a spring loaded lever on the steering column that can increase it. Brake lights come on when the lever is used.

The idea being that the car is safe for anyone to drive.

This of course assumes that real-time torque or current control during generator mode is possible with the controller/VFD in use.

User avatar
acmotor
Senior Member
Posts: 3603
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

How Does Regeneration Work

Post by acmotor » Thu, 05 Mar 2009, 19:12

As a matter of interest, the amount of regen used does depend on the battery recharge rate accepted.
I find that regen of max 10kW on a 1200kg vehicle is like ICE engine braking in second gear. i.e. a fair bit. Just that it continues down to zero speed.
10kW regen on 600V+ pack is around 15A. This is already a higher rate than most batteries like for recharge.

The point is that regen at a level more than "engine braking" (that is driven off the accelerator pedal anyway) may not be usefull without ultracaps or something.

Prius run regen off the top end of brake pedal as per HeadsUp's thinking.

I do like the idea of jake brake sound effects from the fingertip regen control though ! Image
iMiEV MY12     110,230km in pure Electric and loving it !

User avatar
acmotor
Senior Member
Posts: 3603
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

How Does Regeneration Work

Post by acmotor » Thu, 05 Mar 2009, 19:14

Oh yes, and motor controller operates brake lights as it knows what is going on.
iMiEV MY12     110,230km in pure Electric and loving it !

Post Reply