Phasors vs. vectors and power calculations

Mar 10, 2005 11 Replies

Greetings:



This is just a reflection on some mathematical semantics and an interest in having a neat notation for calculating the average power dissipated in an AC circuit from rectangular forms of the voltage and current phasors. It arose out of my deciding to review in detail my understanding of AC power, so might be of some interest to students of EE. There is nothing here which is not found in any basic circuits text, but just my particular description of it.



To get started, my definition of a phasor would go like this:



"A phasor is a complex number that carries the information about the magnitude and phase of a sinusoidal time varying function of fixed frequency."



Thus, any old complex number is not a phasor, unless it is a complex number arising from the phasor transform applied to a sinusoidal time varying function of fixed frequency.



"Furthermore, a phasor is not a vector, though a phasor may be graphically represented as a vector in the complex plane."



"Nor do phasors or the vectors representing them "rotate" in the complex plane. There is nothing in the phasor transform that leads to the real and imaginary components of the phasor being time dependent functions. Indeed it's very point is to remove the time variation aspect from the voltage and current quantities. Thus, the components of the phasors are simply constants. It is only a demonstrative tool to illustrate how IF the vector is rotated in time, that for the case of starting with a cosine time domain function (and applying the respective phasor transform), then the real component of the phasor traces out the instantaneous magnitude of the time varying quantity. Likewise for the case of starting with a sine time domain function (and applying the respective phasor transform), then the imaginary component of the phasor traces out the instantaneous magnitude of the time varying quantity."



Average power dissipated in a linear AC circuit driven by a fixed frequency sinusoidal source can be determined by the dot product of the vectors representing the phasors for



V(t) = Vm cos(wt+phi_V) I(t) = Im cos(wt+phi_I)



Let's use the notation ~V and ~I to represent the phasor transforms of V(t) and I(t) respectively.



But since ~V and ~I are phasors we can't notate or even speak of their dot product as we would with vectors. Ie., the notation



~V . ~I



doesn't mean the dot product of the vectors, since the quantities ~V and ~I are phasors, not vectors.



Instead we would have to write out something ugly like:



Pave = [ Re{~V} Re{~I} + Im{~V} Im{~I} ]/2



which works, but isn't a concise mathematical notation such as the simple dot between two vectors.



Obviously, if we have the phasors in polar form, then the calculation is a more straightforward:



Pave = 0.5 Vm Im cos( phi_V - phi_I )



which of course is the definition of the dot product of the vectors representing the phasors.



Additionally, there is the defnition of complex power as



S = P + jQ , where



P = 0.5 Vm Im cos( phi_V - phi_I ) = Pave Q = 0.5 Vm Im sin( phi_V - phi_I ) = reactive power



In which case average power may be neatly expressed as:



Pave = Re{S}



And since S can be shown to be:



S = 1/2 ~V ~I* then



Pave = Re{ 1/2 ~V ~I* }



So perhaps that is it, huh? The real part of the product of a phasor with the conjugate of another gives the same effect as the dot product of the vectors representing those phasors.


Good day!


_______________________________________________________________________ Christopher R. Carlen Principal Laser/Optical Technologist Sandia National Laboratories CA USA crcarleRemoveThis@BOGUSsandia.gov NOTE, delete texts: "RemoveThis" and "BOGUS" from email address to reply.

Then I would have to say: A phasor is a stationary vector used to conveniently refer to a rotating vector.

--Larry Brasfield email: donotspam_larry_brasfield@hotmail.com Above views may belong only to me.

The phasor transform is found by taking only the non time-dependent factors of the argument to the real operator in:

V(t)=Vm cos(wt+phi_V)

=Re{Vm [cos(wt+phi_V) + j sin(wt+phi_V)]}

=Re{Vm exp[j(wt+phi_V)]}

=Re{Vm exp(jwt) exp(j phi_V)}

Thus:

~V = Vm exp(j phi_V)

~V = Vm cos phi_V + j sin phi_V

There is nothing time dependent here to allow for any conception of rotation. The phasor is a constant complex number. How can a constant rotate? The only difference between it and an ordinary constant complex number, such as an impedance, is mathematically no difference at all, but contextually that it is derived from the phasor transform of a time varying function.

That's why I said it is a semantics issue. Rotating the vector that graphically represents the phasor in the complex plane is a demonstrative tool. But it remains that there is nothing about the mathematical representation of the phasor that allows for rotation.

One last thing. A vector represents a magnitude and a direction in a coordinate system. It is based upon orthonormal basis vectors such as what we might consider to be r_hat and j_hat in the complex plane, analagous to x_hat, y_hat in the XY plane.

It is the definition of the basis vectors that gives mathematical consistency to a statement such as:

A_vec . B_vec = (A_x x_hat + A_y y_hat).(B_x x_hat + B_y y_hat) = A_x B_x x_hat.x_hat + 2 A_x B_y x_hat.y_hat + A_y B_y y_hat.y_hat = A_x B_x + A_y B_y

since x_hat.x_hat=1, y_hat.y_hat=1, and x_hat.y_hat=0

But there are no basis vectors in a complex number or phasor. Thus, the phasor doesn't contain information about "direction" in an orthonormal coordinate system. That is why I carefully say that a phasor may be

*graphically represented* as a vector in a complex coordinate plane characterized by orthonormal basis vectors r_hat and j_hat such that

r_hat.r_hat=1 and j_hat.j_hat=1 and r_hat.j_hat=0

So that we can represent the phasors as a vector by letting:

V_vec = Re{~V} r_hat + Im{~V} j_hat I_vec = Re{~I} r_hat + Im{~I} j_hat

Then the real average power:

Pave = 0.5 V_vec.I_vec works.

But the expression:

~V.~I is meaningless.

Good day!

_______________________________________________________________________ Christopher R. Carlen Principal Laser/Optical Technologist Sandia National Laboratories CA USA crcarleRemoveThis@BOGUSsandia.gov NOTE, delete texts: "RemoveThis" and "BOGUS" from email address to reply.

Something is rotating here; I seem to be getting dizzy.

Luhan Monat (luhanis 'at' yahoo 'dot' com) "The future is not what it used to be..." http://members.cox.net/berniekm

Either this, or a weapon of mass destruction.

Keith

And the coordinate plane it is drawn on and the viewer of that plane are rotating at the same frequency. ;-)

John Popelish

Terry, dropping the exp(jwt) is the *definition* of the phasor transform.

For references, try here:

Ulaby, Fawwaz. "Fundamentals of Applied Electromagnetics 2001 media ed." pg. 25

And here:

Nilsson, James W. and Susan Riedel. "Electric Circuits 6th ed." pg. 418

And there must be a pile of other texts which explain this definition.

Good day!

_______________________________________________________________________ Christopher R. Carlen Principal Laser/Optical Technologist Sandia National Laboratories CA USA crcarleRemoveThis@BOGUSsandia.gov NOTE, delete texts: "RemoveThis" and "BOGUS" from email address to reply.

I think that I would have to say a phasor is a rotating vector.

d

Pearce Consulting

formatting link

OK - I'll try again. A phasor is a representation of the instantaneous angular difference between subject and reference vectors, with the absolute magnitude of the subject vector.

d

Pearce Consulting

formatting link

what about exp(jwt) that you carefully dropped off? that is the rotating bit.

Indeed. If machines are your thing, Retter, Krause, Kron, Vas etc. have written plenty on the subject.

I think this is a semantic issue. Phasors are, by definition, rotating vectors. The trick with phasor transforms is to rotate synchronously with some reference phasor, so all synchronous phasors appear stationary, but with angular displacement.

Mathematically this is achieved by dropping the exp(jwt) term.

If you build a synchronously rotating reference frame controller for electrical machinery, this "phasor transform" is normally implemented as follows:

1) Choose a phasor on which to orient your rotating reference frame - eg AC line voltage. By definition this transformed phasor is purely real (or purely complex).

2) Calculate the transformed real and imaginary parts by taking the original phasor and multiplying by exp(j*theta) [in a 3-phase system we typically assume some symmetry and perform a 3-phase to 2-phase transformation, resulting in a single complex phasor.

3) use the imaginary (or real) part of the transformed phasor as the feedback signal for a PI controller to calculate the actual w. The setpoint for this PI controller will be zero, again by definition

4) the output of the PI controller is angular speed, w. Usually the PI controller has the desired w (eg 2pi*50Hz) added to its output, so it only corrects the phase error, but its not strictly necessary - the PI controller will eventually get the right w.

5) integrate the estimated w, giving theta = wt

6) use this theta in the phasor transform of step 2.

Ive implemented quite a few of these, and it works very well, for a wide range of PI gains. The magnitude of the error (ie |Eq|) can be used to detect synchronisation. Once synchronised, all them pesky rotating things suddenly look DC. If your are controlling an induction motor, the rotor vectors all spin at the slip speed.

And of course its quite feasible to transform the synchronous PI controller into a stationary reference frame, it just becomes a resonant controller (DG Holmes, P Mattiavelli etc have covered these in detail) but thats boring power electronics control stuff.

Interestingly enough, the electrical model for a machine in the stationary reference frame implicitly has time-varying inductances (hardly surprising, flux linkage being a function of rotor angle), making the analysis a right bastard. But through a phasor transformation, the time-varying nature of the inductance "disappears", making the resultant analysis straightforward. Its been around in power systems engineering since the 1920s, often referred to as Clarke (or is it Park, one is the 3:2 phase transform, the other is the stationary-rotating) transforms.

This is another form of feedback linearisation based nonlinear control BTW

Conversely whenever I analyse a 50Hz AC circuit, I implicitly use phasor transforms by simply ignoring the rotating aspect, and pretending the relevant vectors are all stationary.

Cheers Terry

Are we allowed to Park here?

:-)

Then there's duct tape ... (Garrison Keillor)

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required