21.03.2015 Views

Stability Analysis of an All-Electric Ship MVDC Power Distribution ...

Stability Analysis of an All-Electric Ship MVDC Power Distribution ...

Stability Analysis of an All-Electric Ship MVDC Power Distribution ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Stability</strong> <strong>Analysis</strong> <strong>of</strong> <strong>an</strong> <strong>All</strong>-<strong>Electric</strong> <strong>Ship</strong><br />

<strong>MVDC</strong> <strong>Power</strong> <strong>Distribution</strong> System<br />

Using a Novel Passivity-Based <strong>Stability</strong> Criterion<br />

Antonino Riccobono, Enrico S<strong>an</strong>ti<br />

Dept. <strong>of</strong> <strong>Electric</strong>al Engineering<br />

University <strong>of</strong> South Carolina<br />

Columbia, SC 29208, USA<br />

riccobon@cec.sc.edu, s<strong>an</strong>ti@cec.sc.edu<br />

Abstract—The present paper describes a recently proposed<br />

Passivity-Based <strong>Stability</strong> Criterion (PBSC) <strong>an</strong>d demonstrates its<br />

usefulness <strong>an</strong>d applicability to the stability <strong>an</strong>alysis <strong>of</strong> the<br />

<strong>MVDC</strong> <strong>Power</strong> <strong>Distribution</strong> System for the <strong>All</strong>-<strong>Electric</strong> <strong>Ship</strong>. The<br />

criterion is based on imposing passivity <strong>of</strong> the overall DC bus<br />

imped<strong>an</strong>ce. If passivity <strong>of</strong> the bus imped<strong>an</strong>ce is ensured, stability<br />

is guar<strong>an</strong>teed as well. The PBSC, in contrast with existing<br />

stability criteria, such as the Middlebrook criterion <strong>an</strong>d its<br />

extensions, which are based on the minor loop gain concept, i.e.<br />

<strong>an</strong> imped<strong>an</strong>ce ratio at a given interface, <strong>of</strong>fers several<br />

adv<strong>an</strong>tages: reduction <strong>of</strong> artificial design conservativeness,<br />

insensitivity to component grouping, applicability to multiconverter<br />

systems <strong>an</strong>d to systems in which the power flow<br />

direction ch<strong>an</strong>ges, for example as a result <strong>of</strong> system<br />

reconfiguration. Moreover, the criterion c<strong>an</strong> be used in<br />

conjunction with <strong>an</strong> active method to improve system stability,<br />

called the Positive Feed-Forward (PFF) control, for the design <strong>of</strong><br />

virtual damping networks. By designing the virtual imped<strong>an</strong>ce<br />

so that the bus imped<strong>an</strong>ce passivity condition is met, the<br />

approach results in greatly improved stability <strong>an</strong>d damping <strong>of</strong><br />

tr<strong>an</strong>sients on the DC bus voltage. Simulation validation is<br />

performed using a switching-model DC power distribution<br />

system.<br />

I. INTRODUCTION<br />

Recently, there has been <strong>an</strong> increased interest in using<br />

<strong>MVDC</strong> <strong>Power</strong> <strong>Distribution</strong> for the US Navy <strong>All</strong>-<strong>Electric</strong> <strong>Ship</strong><br />

[1-3] <strong>an</strong>d a new IEEE St<strong>an</strong>dard has been released [4]. The<br />

single-bus <strong>MVDC</strong> power distribution system depicted in Fig.<br />

1 is <strong>an</strong> example <strong>of</strong> a possible <strong>MVDC</strong> system topology<br />

described in the St<strong>an</strong>dard. As a well-known challenge, <strong>MVDC</strong><br />

distribution systems suffer from stability degradation caused<br />

by interactions among converters due to the presence <strong>of</strong><br />

const<strong>an</strong>t power loads (CPLs), such as feedback-controlled<br />

converters <strong>an</strong>d inverters. The CPLs exhibit negative<br />

incremental input imped<strong>an</strong>ce, which is the origin <strong>of</strong> potential<br />

instability [5-8].<br />

To address system-level stability issues several authors<br />

have studied the system by breaking it down into two<br />

subsystems: a source subsystem <strong>an</strong>d a load subsystem at <strong>an</strong><br />

arbitrary interface within the overall system. It is usually<br />

assumed that the two subsystems have been properly designed<br />

to be individually stable with good dynamic perform<strong>an</strong>ce. A<br />

fixed power flow direction is also assumed. To assess overall<br />

system stability, several stability criteria for DC systems based<br />

on the minor loop gain, i.e. the ratio <strong>of</strong> source <strong>an</strong>d load<br />

subsystem imped<strong>an</strong>ces, have been proposed in the literature,<br />

such as the Middlebrook Criterion [9], <strong>an</strong>d its various<br />

extensions, such as the Gain <strong>an</strong>d Phase Margin Criterion [10],<br />

the Opposing Argument Criterion [11-13], the ESAC Criterion<br />

[14, 15] <strong>an</strong>d its extension, the Root Exponential <strong>Stability</strong><br />

Criterion [16]. <strong>All</strong> these criteria have been reviewed by the<br />

authors in [17], which presents a discussion, for each criterion,<br />

<strong>of</strong> the artificial conservativeness <strong>of</strong> the criterion in the design<br />

<strong>of</strong> DC systems, <strong>an</strong>d the design specifications that ensure<br />

system stability. Shortcomings <strong>of</strong> all these criteria (also<br />

discussed in [17]) are that they lead to artificially conservative<br />

designs, encounter difficulties when applied to multi-converter<br />

systems (more th<strong>an</strong> two interconnected subsystems) especially<br />

in the case when power flow direction ch<strong>an</strong>ges, <strong>an</strong>d are<br />

sensitive to component grouping. Moreover, all these criteria,<br />

with the exception <strong>of</strong> the Middlebrook Criterion, are not<br />

This work was supported by the Office <strong>of</strong> Naval Research under gr<strong>an</strong>t<br />

N00014-08-1-0080.<br />

Figure 1: Proposed <strong>MVDC</strong> power distribution system for US<br />

Navy all-electric ship (simplified).<br />

978-1-4673-5245-1/13/$31.00 ©2013 IEEE 411


conducive to <strong>an</strong> easy design formulation. Another signific<strong>an</strong>t<br />

practical difficulty present with all the prior stability criteria is<br />

the minor loop gain online measurement [18]. It requires two<br />

separate measurements, source subsystem output imped<strong>an</strong>ce<br />

<strong>an</strong>d load subsystem input imped<strong>an</strong>ce, <strong>an</strong>d then some postprocessing.<br />

Due to the complexity in the calculation, this<br />

approach is not suitable for online stability monitoring. (Only<br />

in the work [19], a practical approach to measure the stability<br />

margins <strong>of</strong> the minor loop gain was proposed. However, such<br />

<strong>an</strong> approach, based on the Opposing Argument Criterion, fails<br />

when used with other less conservative criteria.)<br />

To tackle all these difficulties, a novel Passivity-Based<br />

<strong>Stability</strong> Criterion (PBSC) has been recently proposed [20].<br />

The method is based on the passivity <strong>of</strong> the overall DC bus<br />

imped<strong>an</strong>ce rather th<strong>an</strong> on the Nyquist Criterion applied to <strong>an</strong><br />

imped<strong>an</strong>ce ratio. If the bus imped<strong>an</strong>ce is passive, the stability<br />

<strong>of</strong> the system is guar<strong>an</strong>teed. In the present paper, the<br />

usefulness <strong>an</strong>d applicability <strong>of</strong> the PBSC to the stability<br />

<strong>an</strong>alysis <strong>of</strong> the <strong>MVDC</strong> <strong>Power</strong> <strong>Distribution</strong> System for the <strong>All</strong>-<br />

<strong>Electric</strong> <strong>Ship</strong> is demonstrated using a me<strong>an</strong>ingful, somewhat<br />

simpler test system. Moreover, the PBSC c<strong>an</strong> be coupled with<br />

a recently proposed control strategy for switching converters<br />

called Positive Feed-Forward (PFF) control [21, 22], to design<br />

virtual damping imped<strong>an</strong>ces able to “passivate” <strong>an</strong>d, therefore,<br />

stabilize the DC bus.<br />

The digital network <strong>an</strong>alyzer technique [23] is the tool<br />

adopted in the present paper to perform nonparametric<br />

measurement <strong>of</strong> the bus imped<strong>an</strong>ce. From this measurement,<br />

online bus imped<strong>an</strong>ce passivity condition <strong>an</strong>d therefore system<br />

stability c<strong>an</strong> be monitored. This technique uses a switching<br />

converter to perturb the bus, which has a pseudo-r<strong>an</strong>dom<br />

binary sequence (PRBS) signal in addition to the duty cycle<br />

signal coming from its feedback controller. Since the PRBS is<br />

<strong>an</strong> approximation <strong>of</strong> white noise, all frequencies <strong>of</strong> interest<br />

c<strong>an</strong> be excited simult<strong>an</strong>eously. The perturbation at the bus<br />

interface is measured <strong>an</strong>d the cross-correlation technique is<br />

applied to construct the bus imped<strong>an</strong>ce <strong>of</strong> the system. Online<br />

measurement <strong>of</strong> the bus imped<strong>an</strong>ce c<strong>an</strong> be used for system<br />

health monitoring, fault detection <strong>an</strong>d localization, stability<br />

<strong>an</strong>d perform<strong>an</strong>ce monitoring, stability improvement using<br />

adaptive control, <strong>an</strong>d for distributed control.<br />

II. THE PASSIVITY-BASED STABILITY CRITERION<br />

To better underst<strong>an</strong>d the main difference between the<br />

PBSC <strong>an</strong>d all previous criteria, one c<strong>an</strong> examine Fig. 2. The<br />

single-bus DC power distribution system in Fig. 2a consists <strong>of</strong><br />

n source converters <strong>an</strong>d m load converters connected to the<br />

bus, a generalization <strong>of</strong> the <strong>MVDC</strong> power distribution system<br />

for <strong>All</strong>-<strong>Electric</strong> <strong>Ship</strong>s depicted in Fig. 1. By looking at the bus<br />

port, the given system c<strong>an</strong> be reduced to <strong>an</strong> equivalent<br />

interacting source subsystem <strong>an</strong>d load subsystem network<br />

(Fig. 2b) <strong>an</strong>d then to <strong>an</strong> equivalent 1-port network (Fig. 2c).<br />

While all previous criteria have stopped at the step shown in<br />

Fig. 2b, the proposed PBSC combines together the two<br />

subsystems. The resulting 1-port network shown in Fig. 2c<br />

seen from the DC bus port has <strong>an</strong> imped<strong>an</strong>ce<br />

Z bus (s)=V bus (s)/I inj (s), where I inj (s) is <strong>an</strong> injection current from<br />

<strong>an</strong> external bus-connected device used to perturb the bus. This<br />

imped<strong>an</strong>ce is clearly the parallel combination <strong>of</strong> all the<br />

Figure 2: (a) Typical <strong>MVDC</strong> power distribution system for <strong>Ship</strong>s<br />

with n+m converters, (b) equivalent interacting source subsystem <strong>an</strong>d<br />

load subsystem network, <strong>an</strong>d (c) equivalent 1-port network.<br />

converters’ input/output imped<strong>an</strong>ces, i.e.<br />

Z bus =Z S //Z in =Z 1 //...//Z n //Z n+1 //…//Z n+m . The resulting network is<br />

passive if <strong>an</strong>d only if:<br />

1) Z bus (s) has no right half pl<strong>an</strong>e (RHP) poles, <strong>an</strong>d<br />

2) Re{Z bus (jω)}≥0, ∀ ω .<br />

Condition 1) requires that the Nyquist contour <strong>of</strong> Z bus (jω)<br />

c<strong>an</strong>not enclose <strong>an</strong>y poles. Condition 2) is equivalent to<br />

-90°≤arg[Z bus (jω)]≤90°, <strong>an</strong>d corresponds to <strong>an</strong> imped<strong>an</strong>ce<br />

having positive real part at all frequencies. This also implies<br />

that the Nyquist contour <strong>of</strong> Z bus (jω) must lie wholly on the<br />

RHP.<br />

A passive network has the property <strong>of</strong> being stable [24].<br />

Therefore, the proposed Passivity-Based <strong>Stability</strong> Criterion<br />

(PBSC) for switching converter DC distribution systems (Fig.<br />

2a) states that:<br />

If the passivity condition (<strong>an</strong>d therefore the phase constraint)<br />

is satisfied for Z bus (s), then the overall system consisting <strong>of</strong> the<br />

parallel combination <strong>of</strong> all the converters’ input/output<br />

imped<strong>an</strong>ces (or equivalently <strong>of</strong> the two interacting<br />

subsystems) is stable.<br />

The PBSC has several adv<strong>an</strong>tages over the minor-loopgain-based<br />

stability criteria:<br />

• It c<strong>an</strong> easily h<strong>an</strong>dle multiple interconnected converters<br />

<strong>an</strong>d inversion <strong>of</strong> power flow direction because what<br />

matters is only the parallel combination <strong>of</strong> all input/output<br />

imped<strong>an</strong>ces. Notice that for this reason, the PBSC is also<br />

insensitive to component grouping – typically a problem<br />

for the more conservative prior criteria.<br />

• It reduces artificial design conservativeness typical <strong>of</strong> all<br />

prior stability criteria because the LHP <strong>of</strong> the Nyquist plot<br />

<strong>of</strong> Z bus (jω) is the “forbidden region”. One does not need to<br />

consider encirclements <strong>of</strong> the (-1,0) point.<br />

• Unlike the minor loop gain online measurement, the bus<br />

imped<strong>an</strong>ce online measurement is easy to implement,<br />

does not require complex post-processing, <strong>an</strong>d is suitable<br />

for system stability monitoring.<br />

• The criterion lends itself to the design <strong>of</strong> virtual damping<br />

imped<strong>an</strong>ces which c<strong>an</strong> be actively introduced in parallel<br />

at the bus load-side by a recently proposed control<br />

strategy for switching converters called Positive Feed-<br />

412


Forward (PFF) control. The PFF controller is designed<br />

based on imposing passivity <strong>of</strong> the overall DC bus<br />

imped<strong>an</strong>ce to provide a control design method that<br />

ensures system stability <strong>an</strong>d perform<strong>an</strong>ce.<br />

T<br />

T<br />

I<br />

Gi<br />

L c _ PICM<br />

= 1 +<br />

where T I =G cI·G iLd_OL .<br />

I<br />

(10)<br />

III. THE CONCEPT OF POSITIVE FEED-FORWARD CONTROL<br />

To underst<strong>an</strong>d the concept <strong>of</strong> PFF control, a complete<br />

small-signal model <strong>of</strong> a PFF-<strong>an</strong>d-FB-controlled switching<br />

converter using g-parameter representation [25] is given. The<br />

converter has <strong>an</strong> inner PI current loop <strong>an</strong>d two outer loops<br />

represented by a PI voltage control <strong>an</strong>d a PFF control as<br />

shown in Fig. 3. First, the open-loop PI current mode (PICM)<br />

model is given, <strong>an</strong>d then the closed-loop, both feed-forward<br />

<strong>an</strong>d feedback (FFFB), is derived. After the model is given, the<br />

role <strong>of</strong> PFF control in the passivation, <strong>an</strong>d therefore<br />

stabilization, <strong>of</strong> a <strong>MVDC</strong> bus system is described.<br />

The small-signal model <strong>of</strong> the open-loop converter under<br />

duty cycle control (OL subscript) based on g-parameter<br />

representation c<strong>an</strong> be found in [26-28] for the case <strong>of</strong> DC-DC<br />

converters <strong>an</strong>d in [21, 22] for the case <strong>of</strong> a three-phase DC-AC<br />

converter.<br />

A. Open-loop model<br />

The small-signal open-loop (G cFB (s)=0 <strong>an</strong>d G cFF (s)=0)<br />

PICM converter model is the following<br />

⎡<br />

⎡iˆ<br />

⎤<br />

in ⎢<br />

⎢ ⎥ Z<br />

vˆ<br />

= ⎢<br />

⎢ ⎥ ⎢G<br />

⎢⎣<br />

iˆ<br />

⎥<br />

L ⎦ ⎢<br />

⎣Gi<br />

in<br />

vg<br />

Lg<br />

1<br />

⎤<br />

Gii<br />

_ PICM<br />

Gic<br />

_ PICM ⎥⎡<br />

vˆ<br />

g ⎤<br />

_ PICM<br />

⎥⎢<br />

⎥<br />

−<br />

⎥⎢<br />

iˆ<br />

(1)<br />

load<br />

_ PICM<br />

Zout<br />

_ PICM<br />

Gvc<br />

_ PICM ⎥<br />

⎥⎢⎣<br />

iˆ<br />

⎥<br />

c ⎦<br />

_ PICM<br />

Gi<br />

i PICM<br />

G<br />

L _<br />

iLc<br />

_ PICM ⎦<br />

The tr<strong>an</strong>sfer functions for (1) are given in (2)-(10)<br />

Z<br />

G<br />

G<br />

G<br />

Z<br />

G<br />

G<br />

G<br />

1<br />

in _ PICM<br />

ii _ PICM<br />

ic _ PICM<br />

vg _ PICM<br />

out _ PICM<br />

vc _ PICM<br />

1<br />

=<br />

Z<br />

= G<br />

G<br />

=<br />

G<br />

= G<br />

in _ OL<br />

ii _ OL<br />

G<br />

−<br />

id _ OL<br />

G<br />

G<br />

Gid<br />

_ OLG<br />

−<br />

G<br />

G<br />

−<br />

iL<br />

g _ OL<br />

iL<br />

d _ OL<br />

iLd<br />

_ OL<br />

iLi<br />

_ OL<br />

G<br />

TI<br />

1+<br />

T<br />

TI<br />

1+<br />

T<br />

id _ OL id _ OL iL<br />

g _ OL I<br />

i d OL<br />

G<br />

L _<br />

iL<br />

d _ OL<br />

1+<br />

vg _ OL<br />

1<br />

=<br />

Z<br />

G<br />

=<br />

G<br />

in _ OL<br />

iL<br />

g<br />

iL g _ PICM<br />

+<br />

G<br />

−<br />

G<br />

+<br />

vd _ OL I<br />

iL d _ OL<br />

1+<br />

G<br />

_<br />

= 1 T<br />

G<br />

_<br />

= 1 T<br />

iLi<br />

iL i _ PICM<br />

+<br />

I<br />

OL<br />

I<br />

OL<br />

vd _ OL<br />

G<br />

vd _ OL<br />

T<br />

T<br />

G<br />

I<br />

G<br />

iL<br />

d _ OL<br />

iL<br />

g _ OL<br />

G<br />

iL<br />

d _ OL<br />

iLi<br />

_ OL<br />

I<br />

I<br />

T<br />

T<br />

TI<br />

1+<br />

T<br />

I<br />

TI<br />

1+<br />

T<br />

I<br />

I<br />

(2)<br />

(3)<br />

(4)<br />

(5)<br />

(6)<br />

(7)<br />

(8)<br />

(9)<br />

B. Closed-loop model<br />

The closed-loop FFFB converter model (11) is obtained<br />

from the open-loop PICM converter model (1) by imposing a<br />

control current iˆ<br />

= iˆ<br />

−iˆ<br />

= G ⋅ vˆ<br />

− G ⋅ vˆ<br />

vˆgs<br />

c<br />

vˆg<br />

î load<br />

G cI<br />

î<br />

c<br />

î FF<br />

G CFF<br />

+<br />

î in<br />

dˆ<br />

vˆg<br />

+<br />

‐<br />

FF<br />

î FB<br />

FB<br />

⎡ 1<br />

⎤<br />

⎡iˆ<br />

⎤ ⎢<br />

G<br />

in<br />

ii _ FFFB ⎥⎡<br />

vˆ<br />

⎤<br />

⎢ ⎥ = Z<br />

g<br />

⎢ in _ FFFB<br />

⎥⎢<br />

⎥<br />

(11)<br />

⎣ vˆ<br />

⎦ ⎢<br />

⎥⎣<br />

iˆ<br />

load ⎦<br />

⎣<br />

Gvg<br />

_ FFFB − Zout<br />

_ FFFB ⎦<br />

î FF<br />

î<br />

c<br />

dˆ<br />

(a)<br />

î FB<br />

î<br />

L<br />

vˆo<br />

î load<br />

1<br />

î<br />

+ in<br />

Z<br />

_<br />

( s)<br />

+<br />

in OL +<br />

G<br />

_<br />

( s)<br />

+ vˆ<br />

vg OL -<br />

+<br />

+<br />

G iL g _ OL<br />

( s)<br />

î L<br />

+<br />

+<br />

G ii _ OL<br />

( s)<br />

Z out _ OL<br />

( s)<br />

G iL i _ OL<br />

( s)<br />

G id _ OL<br />

( s)<br />

G vd _ OL<br />

( s)<br />

G iL<br />

( s)<br />

G cFB<br />

d _ OL<br />

Converter<br />

T I<br />

‐ PICM Converter<br />

(b)<br />

Figure 3: Switching converter with inner PI current loop, outer voltage<br />

loop, <strong>an</strong>d PFF control: (a) circuital representation, <strong>an</strong>d (b) small-signal<br />

block diagram representation.<br />

cFF<br />

g<br />

cFB<br />

T FB<br />

413


î in<br />

ˆ =0 i load<br />

100<br />

Bode plot <strong>of</strong> Z<br />

Z bus_FB<br />

Z S<br />

Z S<br />

vˆg<br />

vg _ FB g<br />

Z<br />

in _ FB<br />

G<br />

ii _ FB<br />

iˆ<br />

G<br />

load<br />

vˆ<br />

Z<br />

out<br />

_<br />

FB<br />

vˆo<br />

Magnitude (dB)<br />

50<br />

0<br />

-50<br />

-100<br />

90<br />

Wres<br />

R<strong>an</strong>ge <strong>of</strong> frequencies where<br />

Passivity is violated<br />

Z in_FB<br />

(a)<br />

ˆ =0 i load<br />

Phase (deg)<br />

0<br />

-90<br />

-180<br />

Z S<br />

vˆg<br />

î in<br />

Z<br />

in _ FB<br />

Z damp<br />

G<br />

ii<br />

_<br />

FB<br />

G<br />

G<br />

iˆ<br />

vd_<br />

PICM<br />

id_<br />

PICM<br />

load<br />

G<br />

vˆ<br />

Z<br />

vg_<br />

g<br />

damp<br />

FB<br />

vˆ<br />

g<br />

Z<br />

out _ FB<br />

vˆo<br />

-270<br />

-360<br />

10 -1 10 0 10 1 10 2 10 3 10 4 10 5 10 6<br />

250<br />

Frequency (Hz)<br />

(a)<br />

Nyquist plot <strong>of</strong> Z<br />

Z bus _ FB<br />

200<br />

Z S<br />

Passivity Condition Violated<br />

150<br />

(b)<br />

Figure 4: Source imped<strong>an</strong>ce Z S connected to the equivalent input port<br />

<strong>of</strong> the switching converter: (a) unstable, (b) stable cases.<br />

The tr<strong>an</strong>sfer functions <strong>of</strong> model (11) are given in (12)-(15).<br />

Notice that for the case <strong>of</strong> a three-phase DC-AC converter, the<br />

model is developed in the dq synchronous reference frame <strong>an</strong>d<br />

therefore the elements G ii_FFFB , G vg_FFFB , <strong>an</strong>d Z out_FFFB <strong>of</strong> the<br />

matrix in (1) are 2x2 sub-matrices [21, 22].<br />

Z<br />

G<br />

G<br />

Z<br />

1<br />

in _ FFFB<br />

⎪⎧<br />

1<br />

= ⎨<br />

⎪⎩ Zin<br />

_<br />

⎪⎧<br />

1<br />

= ⎨<br />

⎪⎩ Zin<br />

_<br />

PICM<br />

FB<br />

1<br />

1+<br />

T<br />

FB<br />

⎪⎫<br />

⎪⎧<br />

1<br />

⎬ + ⎨<br />

⎪⎭ ⎪⎩ Z<br />

G<br />

damp<br />

+<br />

Z<br />

⎪⎫<br />

⎬<br />

⎪⎭<br />

1<br />

N _ vc _ OL<br />

Z<br />

TFB<br />

1+<br />

T<br />

FB<br />

⎪⎫<br />

⎧ TFF<br />

⎬ + ⎨<br />

⎪⎭ ⎩1+<br />

T<br />

FB<br />

⎫<br />

⎬<br />

⎭<br />

(12)<br />

ic _ PICM out _ PICM FB<br />

ii _ FFFB<br />

= Gii<br />

_ PICM<br />

+<br />

⋅ = G (13)<br />

vg _ FB<br />

Gvc<br />

_ PICM<br />

1+<br />

TFB<br />

vg _ FFFB<br />

⎧Gvg<br />

_<br />

= ⎨<br />

⎩ 1+<br />

T<br />

=<br />

{ G }<br />

vg _ FB<br />

Z<br />

PICM<br />

FB<br />

⎫<br />

⎬ +<br />

⎭<br />

G<br />

+<br />

G<br />

G<br />

G<br />

vc _ PICM<br />

ic _ PICM<br />

⎧ TFF<br />

⋅ ⎨<br />

⎩1+<br />

T<br />

⎪⎧<br />

1 ⎪⎫<br />

⋅ ⎨ ⎬<br />

⎪⎩ Zdamp<br />

⎪⎭<br />

vc _ PICM<br />

ic _ PICM<br />

FB<br />

⎫<br />

⎬<br />

⎭<br />

T<br />

(14)<br />

out _ PICM<br />

out _ FFFB<br />

= = Z<br />

(15)<br />

out _ FB<br />

1+<br />

TFB<br />

where<br />

Z<br />

1<br />

1<br />

G<br />

ic _ PICM vg _ PICM<br />

= −<br />

(16)<br />

N _ vc _ PICM<br />

Zin<br />

_ PICM<br />

Gvc<br />

_ PICM<br />

T FB =G cFB·G vc_PICM <strong>an</strong>d T FF =G cFF·G ic_PICM are the FB <strong>an</strong>d FF<br />

loop gains, respectively. Notice that T FF has dimensions <strong>of</strong><br />

admitt<strong>an</strong>ce [21, 22, 26-28]. Also, the special cases <strong>of</strong> FB<br />

control only <strong>an</strong>d PFF control only c<strong>an</strong> be found from the more<br />

G<br />

Imaginary Axis<br />

100<br />

50<br />

0<br />

-50<br />

-100<br />

-150<br />

-200<br />

-250<br />

-200 -100 0 100 200 300 400 500<br />

Real Axis<br />

(b)<br />

Figure 5: Bode plot (a) <strong>an</strong>d Nyquist plot (b) <strong>of</strong> the imped<strong>an</strong>ces Z bus_FB.<br />

The blue arrows represent the desired passivation <strong>of</strong> the bus<br />

imped<strong>an</strong>ce.<br />

general FFFB model (11) <strong>an</strong>d (12)-(15) by imposing G cFF =0<br />

(T FF =0) <strong>an</strong>d G cFB =0 (T FB =0), respectively.<br />

C. Effect <strong>of</strong> PFF Control <strong>an</strong>d Control Trade<strong>of</strong>f<br />

The expression for the input imped<strong>an</strong>ce (12) shows that<br />

the PFF control provides a way to control the converter input<br />

imped<strong>an</strong>ce through the last term on the right h<strong>an</strong>d side <strong>of</strong> (12).<br />

In particular, the PFF control has the effect <strong>of</strong> introducing<br />

damping imped<strong>an</strong>ce Z damp in parallel to the already existing<br />

Z in_FB with the goal <strong>of</strong> stabilizing the system. Given a desired<br />

stabilizing imped<strong>an</strong>ce Z damp , designed so that the bus<br />

imped<strong>an</strong>ce satisfies the PBSC, the tr<strong>an</strong>sfer function <strong>of</strong> the PFF<br />

controller is easily found from the last term on the right h<strong>an</strong>d<br />

side <strong>of</strong> (12) <strong>an</strong>d from the expression <strong>of</strong> the FF loop gain:<br />

G<br />

cFF<br />

T<br />

=<br />

G<br />

FF<br />

ic _ PICM<br />

=<br />

G<br />

1 + TFB<br />

(17)<br />

ic _ PICM<br />

⋅ Z<br />

damp<br />

Even though a switching converter is independently<br />

designed to be stable, its dynamics in a large DC distribution<br />

system, such as <strong>an</strong> <strong>All</strong>-<strong>Electric</strong> <strong>Ship</strong> <strong>MVDC</strong> <strong>Power</strong><br />

<strong>Distribution</strong> System, c<strong>an</strong> be affected by the subsystem it is<br />

connected to. Therefore, two types <strong>of</strong> interactions c<strong>an</strong> be<br />

414


identified: the source subsystem interaction, <strong>an</strong>d the load<br />

subsystem interaction. In this paper, the focus is on the source<br />

subsystem interaction problem, i.e., a degradation <strong>of</strong> the<br />

closed-loop stability <strong>of</strong> a switching converter when it is fed by<br />

a DC voltage source subsystem with finite Thévenin<br />

imped<strong>an</strong>ce Z S . Fig. 4a shows the two-port network terminal<br />

model [29, 30], directly drawn from model (11), for the case<br />

<strong>of</strong> FB only (G cFF (s)=0) with a non-ideal source imped<strong>an</strong>ce Z S .<br />

In terms <strong>of</strong> passivity concept, passivity condition for<br />

Z bus_FB is usually violated around ω res , i.e. the frequency where<br />

Z bus_FB exhibits reson<strong>an</strong>ce, as Fig. 5 shows. Imped<strong>an</strong>ce Z bus_FB<br />

follows the output imped<strong>an</strong>ce <strong>of</strong> the source subsystem Z S<br />

everywhere except around the r<strong>an</strong>ge <strong>of</strong> frequencies where it<br />

exhibits reson<strong>an</strong>ce (in the parallel combination the smaller<br />

imped<strong>an</strong>ce dominates). To solve the passivity condition<br />

violation, i.e. to bring Z bus to the RHP so that it exhibits<br />

positive real part at frequencies around ω res , the PFF control is<br />

added to the conventional FB control according Fig. 3a. If<br />

Z damp is designed so that (18) is satisfied, the passivity<br />

violation c<strong>an</strong> be solved.<br />

Z<br />

1<br />

bus _ FFFB<br />

1<br />

=<br />

Z<br />

S<br />

⎧ 1<br />

⎪<br />

⎪ Z<br />

S<br />

⎪ 1<br />

≈ ⎨<br />

⎪Z<br />

⎪ 1<br />

⎪<br />

⎩ Z<br />

S<br />

1<br />

+<br />

Z<br />

damp<br />

in _ FB<br />

1<br />

+<br />

Z<br />

at low frequencies<br />

at high<br />

damp<br />

atω<br />

= ω<br />

res<br />

frequencies<br />

(18)<br />

Z bus_FFFB is the bus imped<strong>an</strong>ce resulting from the addition <strong>of</strong><br />

the PFF control. In other words, (18) shows that if Z damp is<br />

designed so that it dominates at ω res (note, again, that in the<br />

parallel combination the smallest imped<strong>an</strong>ce dominates), it<br />

c<strong>an</strong> provide the desired passivation effect (see blue arrows in<br />

Fig. 5a), i.e. -90°≤arg[Z bus_FFFB (jω res )]≤90° while leaving the<br />

bus imped<strong>an</strong>ce unch<strong>an</strong>ged everywhere else. This is also<br />

equivalent to moving the Nyquist plot <strong>of</strong> the bus imped<strong>an</strong>ce<br />

from the LHP to the RHP, as the blue arrow in Fig. 5b<br />

indicates.<br />

Fig. 4 shows <strong>an</strong> equivalent model for the system under FB<br />

control only (a) <strong>an</strong>d under FFFB control (b). The feed-forward<br />

action introduces two additional elements, imped<strong>an</strong>ce Z damp on<br />

the source side <strong>an</strong>d a controlled voltage source on the output<br />

side. The imped<strong>an</strong>ce Z damp stabilizes the bus, but at a cost: the<br />

output side voltage source has a negative effect on audio<br />

susceptibility. This represents a trade<strong>of</strong>f between stability<br />

improvement <strong>an</strong>d output perform<strong>an</strong>ce degradation. In fact, as<br />

Fig. 5 also depicts, on the one h<strong>an</strong>d the addition <strong>of</strong> Z damp small<br />

in magnitude at ω res (so that it dominates to solve the passivity<br />

violation problem) is desired for stability improvement<br />

purposes, on the other h<strong>an</strong>d it negatively affects the audiosusceptibility<br />

tr<strong>an</strong>sfer function (14). For this reason, Z damp<br />

should be carefully chosen. Notice that Z damp c<strong>an</strong> be chosen to<br />

be either passive or active imped<strong>an</strong>ce. What it import<strong>an</strong>t is<br />

that it should be chosen so that ||Z damp (jω res )|| be not too small,<br />

but small enough to dominate the bus imped<strong>an</strong>ce at ω res . A<br />

PGM‐M PGM‐M PGM‐M PGM‐M<br />

PCM‐A<br />

PDM‐B<br />

PCM‐B<br />

PCM‐B<br />

PCM‐B<br />

PMM<br />

PDM‐A<br />

In‐Zone <strong>Distribution</strong><br />

PMM<br />

PCM‐B<br />

PCM‐B<br />

PCM‐B<br />

PCM‐A<br />

Figure 6: Traditional 3-PDM IPS Architecture, with the two zones<br />

that will be simulated.<br />

good rule <strong>of</strong> the thumb is to choose Z damp so that<br />

||Z bus_FB (jω res )|| dB is 20-25 dB larger th<strong>an</strong> ||Z damp (jω res )|| dB . Of<br />

course, condition (18) must be respected as well. In particular,<br />

it is desired that arg[Z bus_FFFB (jω res )] lies as close as possible to<br />

0° so that the system has a good passivity margin (it is far<br />

away from the passivity boundary).<br />

IV. SIMULATION RESULTS<br />

In this section, the PBSC is validated by using a switching<br />

model <strong>of</strong> a <strong>MVDC</strong> bus system built in Simulink. The<br />

simulated system is a power down-scaled portion <strong>of</strong> one <strong>of</strong> the<br />

possible Integrated <strong>Power</strong> System (IPS) architectures that<br />

have been proposed in [31]. The architecture taken into<br />

consideration in this paper is the traditional three Propulsion<br />

<strong>Distribution</strong> Module (PDM) ISP architecture, shown in Fig 6.<br />

PDM-A is a HVDC bus (10 KV), PGMs are high frequency<br />

<strong>Power</strong> Generation Modules with rectifier units, PMMs are<br />

Propulsion Motor Modules, PDM-B is a <strong>MVDC</strong> bus (1000 V),<br />

<strong>an</strong>d the <strong>Power</strong> Conversion Modules (PCMs) are switching<br />

converters. Two portions <strong>of</strong> the system in Fig. 6 will be<br />

simulated, represented by areas within the red <strong>an</strong>d blue dashed<br />

lines. Fig. 7 shows the system switching model built in<br />

Simulink. The area within the red dashed line is a cascade <strong>of</strong> a<br />

PICM-FB-controlled buck converter <strong>an</strong>d a PICM-FFFBcontrolled<br />

three-phase voltage source inverter (VSI). The<br />

controller <strong>of</strong> the VSI has a PFF controller in addition to the<br />

conventional FB controller as shown in Fig. 8. The area within<br />

the blue dashed line is the same system with the addition <strong>of</strong> a<br />

high b<strong>an</strong>dwidth CPL, representative <strong>of</strong> <strong>an</strong>other switching<br />

converter or a pulsed load, like a radar or rail-gun for future<br />

naval combat<strong>an</strong>ts. Specifications for both buck converter <strong>an</strong>d<br />

VSI are also reported in Fig. 7. The PFF control was designed<br />

to actively introduce the following damping imped<strong>an</strong>ce at the<br />

VSI input port.<br />

1<br />

Zdamp<br />

= Rb<br />

+ sLb<br />

+<br />

(19)<br />

sCb<br />

where R b =21.34Ω, L b =17.88mH, <strong>an</strong>d C b =157.19µF. The PFF<br />

controller tr<strong>an</strong>sfer function is then calculated from (17).<br />

PDM‐B<br />

415


Figure 7: Cascade <strong>of</strong> a PICM-FB controlled buck converter <strong>an</strong>d a PICM-FFFB-controlled VSI. Another buck converter is used to perturb the bus <strong>an</strong>d<br />

extract bus imped<strong>an</strong>ce. A high b<strong>an</strong>dwidth CPL added to the bus has the effect <strong>of</strong> destabilizing the bus.<br />

Viewed from the bus port, the entire system c<strong>an</strong> be lumped<br />

into a 1-port network, as previously described in Section II.<br />

The digital network <strong>an</strong>alyzer technique [23] is the tool used to<br />

measure the bus imped<strong>an</strong>ce <strong>an</strong>d address system level stability<br />

issues in a <strong>MVDC</strong> power distribution system. This technique<br />

uses a switching converter as a perturbation source, <strong>an</strong>d its<br />

controller as a signal <strong>an</strong>alyzer to measure small-signal tr<strong>an</strong>sfer<br />

functions <strong>an</strong>d imped<strong>an</strong>ces <strong>of</strong> interest. Referring to Fig. 9, a<br />

pseudo-r<strong>an</strong>dom binary sequence (PRBS) test signal is added to<br />

the duty cycle signal from the feedback controller. Applying<br />

the cross-correlation technique to the appropriate measured<br />

qu<strong>an</strong>tities allows online monitoring <strong>of</strong> the bus imped<strong>an</strong>ce<br />

Z bus (s)=V bus (s)/I inj (s).<br />

A. Cascade <strong>of</strong> Buck Converter <strong>an</strong>d VSI<br />

For this case, the system with only FB control is<br />

marginally stable, while the same system with FFFB control is<br />

highly stabilized. Fig. 10 shows the Bode plot <strong>of</strong> the bus<br />

imped<strong>an</strong>ce, which is the parallel combination <strong>of</strong> the output<br />

Figure 8: Control <strong>of</strong> the VSI. PICM-FB loop in the middle <strong>an</strong>d FF loop at the top.<br />

Figure 9: Conceptual block diagram showing injection <strong>of</strong> a test<br />

signal for system bus imped<strong>an</strong>ce measurements.<br />

imped<strong>an</strong>ce <strong>of</strong> the buck converter <strong>an</strong>d the input imped<strong>an</strong>ce <strong>of</strong><br />

the VSI (Eq. (20). The <strong>an</strong>alytical tr<strong>an</strong>sfer functions, given in<br />

(20), are compared with the nonparametric results given by the<br />

digital network <strong>an</strong>alyzer in simulation.<br />

Z<br />

⎛ 1 1 ⎞<br />

_<br />

= ⎜<br />

+ ⎟<br />

(20)<br />

bus CL<br />

⎝ Zout<br />

_ CL(<br />

Buck )<br />

Zin<br />

_ CL(<br />

VSI ) ⎠<br />

where CL denotes either FB or FFFB.<br />

−1<br />

416


240<br />

230<br />

220<br />

210<br />

200<br />

190<br />

180<br />

170<br />

160<br />

150<br />

Magnitude [dB]<br />

Phase [deg]<br />

60<br />

40<br />

20<br />

0<br />

-20<br />

100<br />

50<br />

-50<br />

-100<br />

Magnitude <strong>of</strong> Z bus<br />

10 1 10 2 10 3 10 4<br />

0<br />

Phase <strong>of</strong> Z bus<br />

10 1 10 2 10 3 10 4<br />

Frequency [Hz]<br />

Estimation with FFFB<br />

Analytical TF with FFFB<br />

Analytical TF with FB only<br />

Estimation with FB only<br />

Figure 10: Z bus estimation through PRBS injection <strong>an</strong>d comparison with<br />

<strong>an</strong>alytical tr<strong>an</strong>sfer functions for the cascade <strong>of</strong> buck converter <strong>an</strong>d VSI.<br />

R = 10 ohm -> 20 ohm<br />

Vbus<br />

R = 20 ohm -> 10 ohm<br />

Vabc<br />

FB only<br />

FFFB<br />

arg[Z bus_FB (jω res )]≈±90°. Notice that for both cases the bus<br />

imped<strong>an</strong>ce satisfies the passivity condition, i.e.<br />

-90°≤arg[Z bus (jω)]≤90°, ∀ ω . Fig. 11 depicts the tr<strong>an</strong>sient<br />

responses <strong>of</strong> the bus voltage <strong>an</strong>d three-phase output voltage in<br />

correspondence <strong>of</strong> (a) bal<strong>an</strong>ced three-phase load step <strong>an</strong>d (b)<br />

voltage reference step, starting from a steady-state condition.<br />

Notice the presence <strong>of</strong> sustained oscillations for the FB case<br />

only, <strong>an</strong>d the stabilization <strong>of</strong> the bus voltage for the FFFB<br />

case. Moreover, the three-phase output voltage is somehow<br />

more distorted during the tr<strong>an</strong>sient for the FFFB case. This is<br />

due to the fact that the PFF control alters the audiosusceptibility<br />

<strong>of</strong> the VSI, as discussed in Section III. However,<br />

Fig. 11 shows that a good trade-<strong>of</strong>f between stability<br />

improvement <strong>an</strong>d FB control b<strong>an</strong>dwidth reduction has been<br />

achieved.<br />

B. Addition <strong>of</strong> a High B<strong>an</strong>dwidth CPL<br />

The addition <strong>of</strong> a high b<strong>an</strong>dwidth CPL (P CPL =300W)<br />

further deteriorates the passivity condition <strong>an</strong>d the stability <strong>of</strong><br />

the system. The system with only FB control is now unstable,<br />

while the same system with FFFB control is again highly<br />

stabilized. Fig. 12 shows the Bode plot <strong>of</strong> the bus imped<strong>an</strong>ce.<br />

The <strong>an</strong>alytical tr<strong>an</strong>sfer function for the bus imped<strong>an</strong>ce, given<br />

in (21), now includes the CPL load input imped<strong>an</strong>ce in<br />

parallel. The <strong>an</strong>alytical results are compared with the<br />

nonparametric results given by the digital network <strong>an</strong>alyzer<br />

simulation.<br />

100<br />

50<br />

0<br />

-50<br />

-100<br />

-150<br />

300<br />

280<br />

260<br />

240<br />

220<br />

200<br />

180<br />

160<br />

140<br />

120<br />

150<br />

100<br />

50<br />

0<br />

-50<br />

R = 10 ohm -> 20 ohm<br />

R = 20 ohm -> 10 ohm<br />

0.1 0.15 0.2 0.25<br />

(a)<br />

Vbus<br />

Vref = 90 Vpk -> 45 Vpk<br />

Vref = 45 Vpk -> 90 Vpk<br />

Vabc<br />

FB only<br />

FFFB<br />

Z ⎛ 1 1 1 ⎞<br />

_<br />

= ⎜<br />

+ + ⎟<br />

(21)<br />

bus CL<br />

⎝ Zout<br />

_ CL(<br />

Buck )<br />

Zin<br />

_ CL(<br />

VSI )<br />

Zin<br />

_ CPL ⎠<br />

where CL denotes either FB or FFFB, <strong>an</strong>d Z in_CPL =-V 2 bus /P CPL .<br />

As shown in Fig. 12, in the stable case the bus imped<strong>an</strong>ce<br />

well satisfies the passivity condition because<br />

arg[Z bus_FFFB (jω res )]≈0°, while in the unstable stable case the<br />

passivity condition is not met because arg[Z bus_FB (jω)] goes<br />

beyond ±90° for some frequencies near the reson<strong>an</strong>t<br />

frequency. Note that at low frequency the phase is -270°,<br />

which is the same as +90°, <strong>an</strong>d at high frequency the phase is<br />

-90°; however around the reson<strong>an</strong>t frequency the phase<br />

exceeds the ±90° r<strong>an</strong>ge. Fig. 13 depicts the tr<strong>an</strong>sient responses<br />

<strong>of</strong> the bus voltage, three-phase output voltage, in<br />

correspondence <strong>of</strong> (a) symmetric three-phase load step, (b)<br />

voltage reference step, <strong>an</strong>d (c) CPL power step, starting from a<br />

steady-state condition. Notice the instability for the FB case<br />

only, <strong>an</strong>d the stabilization <strong>of</strong> the bus voltage for the FFFB<br />

case.<br />

−1<br />

-100<br />

-150<br />

Vref = 90 Vpk -> 45 Vpk<br />

Vref = 45 Vpk -> 90 Vpk<br />

0.1 0.15 0.2 0.25<br />

(b)<br />

Figure 11: Time-domain results for the cascade <strong>of</strong> buck converter <strong>an</strong>d VSI.<br />

Comparison <strong>of</strong> FB only <strong>an</strong>d FFFB in correspondence to (a) bal<strong>an</strong>ced resistive<br />

step applied to the VSI <strong>an</strong>d (b) voltage reference step applied to the VSI.<br />

As shown in Fig. 10, in the stable case the bus imped<strong>an</strong>ce<br />

well satisfies the passivity condition at the reson<strong>an</strong>t frequency,<br />

because arg[Z bus_FFFB (jω res )]≈0°, while in the marginally<br />

stable case the passivity condition is only weakly met, because<br />

CONCLUSIONS<br />

The PBSC has been proposed as a new stability criterion<br />

based on the passivity <strong>of</strong> the bus imped<strong>an</strong>ce. If that imped<strong>an</strong>ce<br />

is passive then the entire system is stable. Adv<strong>an</strong>tages <strong>of</strong> the<br />

PBSC over prior stability criteria, based on the minor loop<br />

gain concept, have been discussed. Combined with the PFF<br />

control, it is possible to design stabilizing virtual damping<br />

imped<strong>an</strong>ces so that the PBSC is satisfied. The usefulness <strong>of</strong><br />

the PBSC in system stability online monitoring <strong>an</strong>d design so<br />

that the entire system is stable <strong>an</strong>d well-behaved has been<br />

proved by the switching model simulation <strong>of</strong> a portion <strong>of</strong> the<br />

417


three-PDM ISP architecture proposed for the <strong>MVDC</strong> <strong>Power</strong><br />

<strong>Distribution</strong> System for the <strong>All</strong>-<strong>Electric</strong> <strong>Ship</strong>. Frequency <strong>an</strong>d<br />

time domain results have validated the proposed method.<br />

350<br />

300<br />

250<br />

Vbus<br />

Magnitude <strong>of</strong> Z bus<br />

60<br />

Estimation with FFFB<br />

200<br />

Analytical TF with FFFB<br />

Magnitude [dB]<br />

40<br />

20<br />

0<br />

Analytical TF with FB only<br />

Estimation with FB only<br />

150<br />

100<br />

150<br />

100<br />

R = 10 ohm -> 20 ohm<br />

R = 20 ohm -> 10 ohm<br />

Vabc<br />

FB only<br />

FFFB<br />

50<br />

-20<br />

10 1 10 2 10 3 10 4<br />

0<br />

Phase <strong>of</strong> Z bus<br />

-50<br />

100<br />

-100<br />

0<br />

-150<br />

R = 10 ohm -> 20 ohm<br />

R = 20 ohm -> 10 ohm<br />

0.1 0.15 0.2 0.25<br />

Phase [deg]<br />

-100<br />

(a)<br />

-200<br />

350<br />

Vbus<br />

-300<br />

10 1 10 2 10 3 10 4<br />

Frequency [Hz]<br />

Figure 12: Z bus estimation through PRBS injection <strong>an</strong>d comparison with<br />

<strong>an</strong>alytical tr<strong>an</strong>sfer functions for the cascade <strong>of</strong> buck converter <strong>an</strong>d VSI <strong>an</strong>d the<br />

addition <strong>of</strong> a 300W CPL.<br />

300<br />

250<br />

200<br />

150<br />

100<br />

REFERENCES<br />

[1] Parker, D.S.; Hodge, C.G., "The electric warship [electric propulsion],"<br />

<strong>Power</strong> Engineering Journal, Volume: 12 Issue: 1 , Feb. 1998, Page(s): 5<br />

-13.<br />

[2] Clayton, D.A.; Sudh<strong>of</strong>f, S.D.; Grater, G.F., "<strong>Electric</strong> ship drive <strong>an</strong>d<br />

power system," <strong>Power</strong> Modulator Symposium, 2000. Conference<br />

Record <strong>of</strong> the 2000 Twenty-Fourth International, 2000, Page(s): 85 -88.<br />

[3] N. Doerry, Next Generation Integrated <strong>Power</strong> Systems, NGIPS Master<br />

Pl<strong>an</strong>, Whashington, DC: Naval Sea Systems Comm<strong>an</strong>d, 2007.<br />

[4] IEEE Std 1709-2010, "IEEE Recommended Practice for 1 kV to 35 kV<br />

Medium-Voltage DC <strong>Power</strong> Systems on <strong>Ship</strong>s," IEEE Std 1709-2010,<br />

vol., no., pp.1-54, Nov. 2 2010.<br />

[5] Emadi, A. Khaligh, C. H. Rivetta, <strong>an</strong>d G. A. Williamson, "Const<strong>an</strong>t<br />

power loads <strong>an</strong>d negative imped<strong>an</strong>ce instability in automotive systems:<br />

definition, modeling, stability, <strong>an</strong>d control <strong>of</strong> power electronic<br />

converters <strong>an</strong>d motor drives," IEEE Tr<strong>an</strong>s. Vehicular Technology, vol.<br />

55, pp. 1112-1125, Jul. 2006.<br />

[6] Y. J<strong>an</strong>g <strong>an</strong>d R. W. Erickson, "Physical origins <strong>of</strong> input filter<br />

oscillations in current programmed converters," IEEE Tr<strong>an</strong>s. <strong>Power</strong><br />

Electronics, vol. 7, no. 4, pp. 725-733, Jul. 1992.<br />

[7] Emadi, A.; Ehs<strong>an</strong>i, A.; , "Dynamics <strong>an</strong>d control <strong>of</strong> multi-converter DC<br />

power electronic systems ," <strong>Power</strong> Electronics Specialists Conference,<br />

2001. PESC. 2001 IEEE 32nd Annual , vol.1, no., pp.248-253 vol. 1,<br />

2001.<br />

[8] A. Emadi, B. Fahimi, <strong>an</strong>d M. Ehs<strong>an</strong>i, "On the concept <strong>of</strong> negative<br />

imped<strong>an</strong>ce instability in adv<strong>an</strong>ced aircraft power systems with const<strong>an</strong>t<br />

power loads," Society <strong>of</strong> Automotive Engineers (SAE) Journal, Paper<br />

No. 1999-01-2545, 1999.<br />

[9] Middlebrook, R.D., "Input Filter Considerations in Design <strong>an</strong>d<br />

Application <strong>of</strong> Switching Regulators," IEEE IAS Annual Meeting,<br />

1976.<br />

[10] Wildrick, C.M.; Lee, F.C.; Cho, B.H.; Choi, B.; , "A method <strong>of</strong><br />

defining the load imped<strong>an</strong>ce specification for a stable distributed power<br />

system," <strong>Power</strong> Electronics, IEEE Tr<strong>an</strong>sactions on , vol.10, no.3,<br />

pp.280-285, May 1995.<br />

[11] Xiaog<strong>an</strong>g Feng; Zhihong Ye; Kun Xing; Lee, F.C.; Borojevic, D.; ,<br />

"Imped<strong>an</strong>ce specification <strong>an</strong>d imped<strong>an</strong>ce improvement for DC<br />

distributed power system," <strong>Power</strong> Electronics Specialists Conference,<br />

1999. PESC 99. 30th Annual IEEE , vol.2, no., pp.889-894 vol.2, 1999.<br />

50<br />

150<br />

100<br />

50<br />

0<br />

-50<br />

-100<br />

-150<br />

240<br />

220<br />

200<br />

180<br />

160<br />

150<br />

100<br />

50<br />

0<br />

-50<br />

-100<br />

-150<br />

300<br />

250<br />

200<br />

150<br />

100<br />

50<br />

0<br />

Vref = 90 Vpk -> 45 Vpk<br />

Vref = 90 Vpk -> 45 Vpk<br />

Vref = 45 Vpk -> 90 Vpk<br />

0.1 0.15 0.2 0.25<br />

Vabc<br />

Vref = 45 Vpk -> 90 Vpk<br />

(b)<br />

Vbus<br />

Vabc<br />

P_CPL<br />

0.05 0.1 0.15 0.2 0.25<br />

(c)<br />

Figure 13: Time-domain results for the cascade <strong>of</strong> buck converter <strong>an</strong>d VSI<br />

with the addition <strong>of</strong> a 300W CPL connected to the bus. Comparison <strong>of</strong> FB<br />

only <strong>an</strong>d FFFB in correspondence to (a) bal<strong>an</strong>ced resistive step applied to the<br />

VSI, (b) voltage reference step applied to the VSI, <strong>an</strong>d (c) CPL power step.<br />

FB only<br />

FFFB<br />

FB only<br />

FFFB<br />

418


[12] Xiaog<strong>an</strong>g Feng; Zhihong Ye; Kun Xing; Lee, F.C.; Borojevic, D.; ,<br />

"Individual load imped<strong>an</strong>ce specification for a stable DC distributed<br />

power system," Applied <strong>Power</strong> Electronics Conference <strong>an</strong>d Exposition,<br />

1999. APEC '99. Fourteenth Annual , vol.2, no., pp.923-929 vol.2, 14-<br />

18 Mar 1999.<br />

[13] Xiaog<strong>an</strong>g Feng; Jinjun Liu; Lee, F.C.; , "Imped<strong>an</strong>ce specifications for<br />

stable DC distributed power systems ," <strong>Power</strong> Electronics, IEEE<br />

Tr<strong>an</strong>sactions on , vol.17, no.2, pp.157-162, Mar 2002.<br />

[14] Sudh<strong>of</strong>f, S.D.; Glover, S.F.; Lamm, P.T.; Schmucker, D.H.; Delisle,<br />

D.E.; , "Admitt<strong>an</strong>ce space stability <strong>an</strong>alysis <strong>of</strong> power electronic<br />

systems," Aerospace <strong>an</strong>d Electronic Systems, IEEE Tr<strong>an</strong>sactions on ,<br />

vol.36, no.3, pp. 965- 973, Jul 2000.<br />

[15] Sudh<strong>of</strong>f, S.D.; Glover, S.F.; , "Three-dimensional stability <strong>an</strong>alysis <strong>of</strong><br />

DC power electronics based systems," <strong>Power</strong> Electronics Specialists<br />

Conference, 2000. PESC 00. 2000 IEEE 31st Annual , vol.1, no.,<br />

pp.101-106 vol.1, 2000.<br />

[16] Sudh<strong>of</strong>f, S.D.; Crider, J.M.; , "Adv<strong>an</strong>cements in generalized<br />

immitt<strong>an</strong>ce based stability <strong>an</strong>alysis <strong>of</strong> DC power electronics based<br />

distribution systems," <strong>Electric</strong> <strong>Ship</strong> Technologies Symposium (ESTS),<br />

2011 IEEE , vol., no., pp.207-212, 10-13 April 2011.<br />

[17] Riccobono, A.; S<strong>an</strong>ti, E.; , "Comprehensive review <strong>of</strong> stability criteria<br />

for DC distribution systems," Energy Conversion Congress <strong>an</strong>d<br />

Exposition (ECCE), 2012 IEEE , vol., no., pp.3917-3925, 15-20 Sept.<br />

2012.<br />

[18] P<strong>an</strong>ov, Y.; Jov<strong>an</strong>ovic, M.; , "Practical issues <strong>of</strong> input/output imped<strong>an</strong>ce<br />

measurements in switching power supplies <strong>an</strong>d application <strong>of</strong> measured<br />

data to stability <strong>an</strong>alysis," Applied <strong>Power</strong> Electronics Conference <strong>an</strong>d<br />

Exposition, 2005. APEC 2005. Twentieth Annual IEEE , vol.2, no.,<br />

pp.1339-1345 Vol. 2, 6-10 March 2005.<br />

[19] Xiaog<strong>an</strong>g Feng; Lee, F.C.; , "On-line measurement on stability margin<br />

<strong>of</strong> DC distributed power system," Applied <strong>Power</strong> Electronics<br />

Conference <strong>an</strong>d Exposition, 2000. APEC 2000. Fifteenth Annual IEEE ,<br />

vol.2, no., pp.1190-1196 vol.2, 2000.<br />

[20] Riccobono, A.; S<strong>an</strong>ti, E.; , "A novel Passivity-Based <strong>Stability</strong> Criterion<br />

(PBSC) for switching converter DC distribution systems," Applied<br />

<strong>Power</strong> Electronics Conference <strong>an</strong>d Exposition (APEC), 2012 Twenty-<br />

Seventh Annual IEEE , vol., no., pp.2560-2567, 5-9 Feb. 2012.<br />

[21] Riccobono, A.; S<strong>an</strong>ti, E.; , "Positive Feedforward Control <strong>of</strong> Three-<br />

Phase Voltage Source Inverter for DC Input Bus Stabilization With<br />

Experimental Validation," Industry Applications, IEEE Tr<strong>an</strong>sactions on<br />

, vol.49, no.1, pp.168-177, J<strong>an</strong>.-Feb. 2013.<br />

[22] Riccobono, A.; S<strong>an</strong>ti, E.; , "Positive Feed-Forward control <strong>of</strong> threephase<br />

voltage source inverter for DC input bus stabilization," Applied<br />

<strong>Power</strong> Electronics Conference <strong>an</strong>d Exposition (APEC), 2011 Twenty-<br />

Sixth Annual IEEE , vol., no., pp.741-748, 6-11 March 2011.<br />

[23] Barkley, A.; S<strong>an</strong>ti, E.; , "Improved Online Identification <strong>of</strong> a DC–DC<br />

Converter <strong>an</strong>d Its Control Loop Gain Using Cross-Correlation<br />

Methods," <strong>Power</strong> Electronics, IEEE Tr<strong>an</strong>sactions on , vol.24, no.8,<br />

pp.2021-2031, Aug. 2009.<br />

[24] J. E. Colgate; “The control <strong>of</strong> dynamically interacting systems”, S.M.<br />

thesis, M.I.T. 1998, Pages 52–53, 70-76.<br />

[25] S. Hiti, D. Boroyevich, <strong>an</strong>d C. Cuadros, "Small-signal modeling <strong>an</strong>d<br />

control <strong>of</strong> three-phase PWM converters," in Proc. IEEE Industry<br />

Applications Society Annual Meeting, vol. 2, Oct. 1994, pp. 1143-<br />

1150.<br />

[26] Cho, H.Y.; S<strong>an</strong>ti, E.; , "Modeling <strong>an</strong>d <strong>Stability</strong> <strong>Analysis</strong> <strong>of</strong> Cascaded<br />

Multi-Converter Systems Including Feedforward <strong>an</strong>d Feedback<br />

Control," Industry Applications Society Annual Meeting, 2008. IAS<br />

'08. IEEE , vol., no., pp.1-8, 5-9 Oct. 2008.<br />

[27] H. Y. Cho <strong>an</strong>d E. S<strong>an</strong>ti, "Modeling <strong>an</strong>d stability <strong>an</strong>alysis in multiconverter<br />

systems including positive feedforward control," in Proc.<br />

IEEE 34th Annual Conference IECON08, Nov. 2008, pp. 839-844.<br />

[28] Ahmadi, R.; Paschedag, D.; Ferdowsi, M.; , "Closed-loop input <strong>an</strong>d<br />

output imped<strong>an</strong>ces <strong>of</strong> DC-DC switching converters operating in voltage<br />

<strong>an</strong>d current mode control," IECON 2010 - 36th Annual Conference on<br />

IEEE Industrial Electronics Society , vol., no., pp.2311-2316, 7-10<br />

Nov. 2010.<br />

[29] Cvetkovic, I.; Boroyevich, D.; Mattavelli, P.; Lee, F. C.; Dong, D.; ,<br />

"Unterminated Small-Signal Behavioral Model <strong>of</strong> DC–DC Converters,"<br />

<strong>Power</strong> Electronics, IEEE Tr<strong>an</strong>sactions on , vol.28, no.4, pp.1870-1879,<br />

April 2013.<br />

[30] Arnedo, L.; Burgos, R.; Boroyevich, D.; W<strong>an</strong>g, F.; , "System-Level<br />

Black-Box Dc-to-Dc Converter Models," Applied <strong>Power</strong> Electronics<br />

Conference <strong>an</strong>d Exposition, 2009. APEC 2009. Twenty-Fourth Annual<br />

IEEE , vol., no., pp.1476-1481, 15-19 Feb. 2009.<br />

[31] CAPT Norbert Doerry, " Open Architecture Approach for the Next<br />

Generation Integrated <strong>Power</strong> System ," Presented at the ASNE<br />

Automation <strong>an</strong>d Controls Symposium 2007, Biloxi, MS, Dec 10-11,<br />

2007.<br />

419

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!