You are on page 1of 149

Distributed Generation

Interconnection, Integration, and


Protection, Present and Future
Michael Ropp, Ph.D., P.E.
Northern Plains Power Technologies
Brookings, SD USA 57006-4716
A quick word about NPPT
• Engineering consulting firm based in Brookings, SD
• Founded in 2005
• Specializing in:
• Transient modeling of power systems
• Distributed generation interconnections
• Protection, control, design, analysis and characterization of low-inertia
systems
• Special deep expertise in:
• Islanding detection
• Temporary/transient overvoltage associated with DERs
• Protection of low-inertia and current-limited systems

Copyright 2015 Michael E. Ropp. All rights reserved.


Introduction
Module 0

Copyright 2015 Michael E. Ropp. All rights reserved.


Historical drivers for energy decisions
• Reliability
• Cost

Copyright 2015 Michael E. Ropp. All rights reserved.


Modern drivers for energy decisions

Incentives

Copyright 2015 Michael E. Ropp. All rights reserved.


Modern drivers for energy decisions

System resilience
Copyright 2015 Michael E. Ropp. All rights reserved.
Modern drivers for energy decisions

Sustainability

Copyright 2015 Michael E. Ropp. All rights reserved.


Key DER technologies
• Cogeneration
• Critically important because it’s often actually economic!
• Best when using a locally-derived fuel.
• Photovoltaics—presently the hot topic
• Small wind (each WT ≤ 1 MW, many < 250 kW)
• Small hydro
• Energy storage

Copyright 2015 Michael E. Ropp. All rights reserved.


Historic PV economics
Utility

Diesel
PV

PV

Copyright 2015 Michael E. Ropp. All rights reserved.


PV has come a long way!

Copyright 2015 Michael E. Ropp. All rights reserved.


PV module cost trend

Source: Presentation
Copyrightby Shalom
2015 MichaelGoffri
E. Ropp.to
AllAWEA WINDPOWER 2014
rights reserved.
PV system cost trend

Source: NREL/LBNL technical report DOE/GO-102012-3839, November 2012

Copyright 2015 Michael E. Ropp. All rights reserved.


PV deployment history
Projected
by
Bloomberg

Copyright 2015 Michael E. Ropp. All rights reserved.


Source: Adam Whitmore, http://theenergycollective.com/onclimatechangepolicy/2210026/extrapolating-deployment-trends-solar-pv
PV deployment projections

8% of total global generation

Copyright 2015 Michael E. Ropp. All rights reserved.


Source: Adam Whitmore, http://theenergycollective.com/onclimatechangepolicy/2210026/extrapolating-deployment-trends-solar-pv
PV system deployment trends
• Fastest-growing sector in MW terms is ground-mount utility scale
systems
• But residential and commercial:
• Are distribution connected
• Represent a larger number of systems
• Still represent more total aggregate generation
• We’re likely to see a LOT more PV on distribution feeders across the
country

Copyright 2015 Michael E. Ropp. All rights reserved.


But beware “grid parity”
• Grid parity is usually defined as the point at which the LCOE of PV
equals the LCOE of “grid electricity”.
• The key problem with this:
• The grid, and an on-grid PV plant without storage, do not provide the same
service (i.e. availability, capacity factor).


Copyright 2015 Michael E. Ropp. All rights reserved.
Types of DG
Module 1

Copyright 2015 Michael E. Ropp. All rights reserved.


Main types, according to grid interface
• Main classes
• Rotating machines
• Synchronous machines
• Induction machines
• Squirrel-cage
• Doubly-fed
• Inverter-interfaced
• In the transient sense, identity of prime mover is important only for:
• power limitations
• transient response characteristics

Copyright 2015 Michael E. Ropp. All rights reserved.


Rotating machines
• Rely on Lenz’ and Faraday’s Laws:

𝜕Φ
𝐸=−
𝜕𝑡
• All about conductors moving in magnetic fields
• Chief difference between them is in where the magnetic field comes
from

Copyright 2015 Michael E. Ropp. All rights reserved.


Synchronous generators

Key point here: modeled as a voltage behind an impedance (mostly reactance).

Copyright 2015 Michael E. Ropp. All rights reserved.


Self-excited induction generators

Examples:
• Type I WTs
• Some small hydro plants
Copyright 2015 Michael E. Ropp. All rights reserved.
Doubly-fed induction generator (Type III WT)

Copyright 2015 Michael E. Ropp. All rights reserved.


Rotating machine characteristics
• Rotating machine behavior is physics-dominated
• Means they can be generically modeled with very good accuracy
• Fault behaviors well known—transient, subsynchronous, synchronous periods
• Need to know the impedances, inertia, and protection
• Act like voltage sources behind impedance
• Real power closely linked to frequency (speed of the rotor); reactive power
closely linked to voltage (AVR)
• Power factor:
• Synchronous machines usually rated to 0.8 lagging (sourcing VArs); do NOT like to
sink Vars
• Induction machines are always leading (absorbing VArs), nearly always capacitively
compensated to ~ 0.9 pf
• Dynamics: relatively slow (cycles to seconds)
Copyright 2015 Michael E. Ropp. All rights reserved.
Inverters
Examples:
• PV
• Type IV WTs
• Batteries and most
other storage

Copyright 2015 Michael E. Ropp. All rights reserved.


Inverter characteristics
• Behavior much more software-determined than for rotating machines
• This provides enormous flexibility, but also means that generic
models will be highly approximate
• Act like constant-power current sources
• Very low fault current (~1.2 pu)
• Limited ability to drive overvoltages
• Phase/VArs more closely linked to frequency, not power
• Power factor: can operate at any power factor as long as current
limits not violated
• Dynamics: fast (subcycle to single-digit cycles)

Copyright 2015 Michael E. Ropp. All rights reserved.


Can’t model one as the other


Copyright 2015 Michael E. Ropp. All rights reserved.
Impacts of PV on distribution
feeders, Part 1: voltage impacts
Module 2

Copyright 2015 Michael E. Ropp. All rights reserved.


Voltage rise
Module 2.1

Copyright 2015 Michael E. Ropp. All rights reserved.


Voltage rise

PV plants push current back upstream, through the feeder impedances. This
reduces the voltage drop, or if current direction reverses anywhere, leads to a
“negative voltage drop”.
Copyright 2015 Michael E. Ropp. All rights reserved.
Voltage rise
• Voltage rise strongly impacted by DG power factor
• Most DG are operated at unity PF, but a DG sourcing VArs can lead to much
larger voltage rise
• Setting a DG to absorb VArs can actually mitigate some problems—more on
this in a minute
• Voltage rise is very location-dependent
• Depends strongly on feeder impedances and load distribution along the
feeder

Copyright 2015 Michael E. Ropp. All rights reserved.


Voltage rise

Copyright 2015 Michael E. Ropp. All rights reserved.


Voltage rise

Copyright 2015 Michael E. Ropp. All rights reserved.


Issues with voltage rise
• Reduced lifetime of customer equipment and transformers
• Especially noticeable with light bulbs
• Increased power draw on the utility (opposite of CVR)
• DER plants tripping on overvoltage

Copyright 2015 Michael E. Ropp. All rights reserved.


Issues with voltage rise

Copyright 2015 Michael E. Ropp. All rights reserved.


Voltage rise mitigation
• Reconductoring
• Line regulators
• Limit the amount of DER on the circuit
• Inverters absorb some VArs (more later)
• Longer term: volt-VAr controls (more later)

Copyright 2015 Michael E. Ropp. All rights reserved.


Voltage modulation or flicker
Module 2.2

Copyright 2015 Michael E. Ropp. All rights reserved.


Flicker
• Older inverters used pulse-based anti-islanding in which VAr output
was pulsed and positive feedback applied.
• VERY effective in island detection.
• However, could cause flicker on higher-impedance circuits.
• For PQ reasons, pulse-based AI is fairly rare today. Most of today’s
commercial inverters will not cause flicker problems.
• Mitigation included adjusting gains, but must be cautious not to
violate UL 1741 listing.

Copyright 2015 Michael E. Ropp. All rights reserved.


Voltage ripple caused by (older) pulse-based AI
4
x 10

1.5 At substation
At PV

1.45

Phase A RMS voltage (V)


1.4

1.35

1.3

159 160 161 162 163 164 165 166 167 168
Time (sec)

Copyright 2015 Michael E. Ropp. All rights reserved.


Voltage modulation
• Close cousin to voltage rise
• Occurs when a high-impedance section of the system contains enough DER to
cause export
• Cloud passages, drops in wind, cause output fluctuations that modulate the
level of voltage rise
• Not strictly flicker because not periodic, but the impacts are similar

Copyright 2015 Michael E. Ropp. All rights reserved.


Harmonics
• Inverters are nonlinear switching devices, and as such harmonic
current injection and resulting voltage harmonic distortion are
sometimes mentioned as concerns.
• Modern grid-tie inverters produce VERY clean output currents.
• THD levels generally less than 3%
• IEEE 519 compliant
• For multiple inverters, harmonics tend to be uncorrelated, so there
isn’t much of a cumulative effect
• In general, harmonics are NOT a problem

Copyright 2015 Michael E. Ropp. All rights reserved.


FFT of output current of residential-scale US
inverter
THD < 2.5%

Copyright 2015 Michael E. Ropp. All rights reserved.


One thing to watch: resonance
• With the rise of transformerless inverters, we are starting to see
occasional problems with resonances.
• Inverter output capacitors can resonate with system inductances.
• Much more of a problem with transmission-level interconnections
• Identification: usually involves a transient study
• Mitigation: usually involves harmonic/trap filters

Copyright 2015 Michael E. Ropp. All rights reserved.


Impact on regulators
Module 2.3

Copyright 2015 Michael E. Ropp. All rights reserved.


Interference with voltage regulation
equipment
• Voltage regulation equipment includes:
• Substation LTCs
• Line regulators
• Switched capacitors
• DG can mess up the planning that went into the placement and
programming of voltage regulation equipment—PV has the added
challenge of variability

Copyright 2015 Michael E. Ropp. All rights reserved.


Clear morning, partly cloudy afternoon

Copyright 2015 Michael E. Ropp. All rights reserved.


P and V during cloud shadow passage

Copyright 2015 Michael E. Ropp. All rights reserved.


Line regulator response to previous slide

It’s easy to see that this kind of event would seriously


wear on line regulators.
Copyright 2015 Michael E. Ropp. All rights reserved.
Mitigation of voltage modulation
• Most popular method today: nonunity power factor operation
(leading/overexcited)
• Can completely eliminate V modulation if the pf is chosen correctly,
and X/R > 1
• Longer term: BIS performing output power averaging
• Can solve modulation problems but not steady-state voltage rise

Copyright 2015 Michael E. Ropp. All rights reserved.


Near-term voltage rise mitigation: fixed or
scheduled PF

P

Q
Copyright 2015 Michael E. Ropp. All rights reserved.
Fixed leading PF impact on voltage during
cloud passage

Copyright 2015 Michael E. Ropp. All rights reserved.


Farther term voltage rise mitigation: volt-VAr
controls

This function originated


with the IEC and has
been adopted by EPRI.
V1-V4 are adjustable;
“typical” values are
shown.

Copyright 2015 Michael E. Ropp. All rights reserved.


Does sinking VArs restrict real power output?
• Depends on the DC-AC ratio—follow the capability curve.

Copyright 2015 Michael E. Ropp. All rights reserved.


Other regulator problems to watch for
• Regulator chatter—some regulators operating to mitigate the
operation of other regulators.
• Can get worse with DERs because “upstream” and “downstream” may change
• Some regulators can’t handle reverse power flow. Mitigation:
replacement.
• Watch for this situation:
• regulator upstream from a branch point;
• DER downstream from the branch point on one branch only.
• In this case, voltage conditions can be significantly different on the two
branches and regulation may become ineffective.

Copyright 2015 Michael E. Ropp. All rights reserved.


Temporary/transient overvoltage
Module 2.4

Copyright 2015 Michael E. Ropp. All rights reserved.


The ITIC (CBEMA) curve

Need to keep Need to keep


the voltage in the voltage out
here. of here.

Copyright 2015 Michael E. Ropp. All rights reserved.


Definitions: Temporary vs. Transient
Overvoltage
Transient overvoltage
(TrOV)

Temporary
overvoltage (TOV)

Copyright 2015 Michael E. Ropp. All rights reserved.


Overvoltage mechanisms
• There are two that are of primary concern with DERs today:
• Ground fault overvoltage (GFOV)
• Load rejection overvoltage (LROV)
• Others do exist:
• Ground potential rise (GPR)
• Inductive coupling to fault current
• Interruption of inductive current

Copyright 2015 Michael E. Ropp. All rights reserved.


Perspective: ranking overvoltages

Copyright 2015 Michael E. Ropp. All rights reserved.


Ground Fault Overvoltage (GFOV)
• GFOV can happen if:
• an ungrounded voltage source continues to feed a four-wire (multigrounded
neutral) circuit,
• after that circuit has been cut off from the substation transformer (grounding
source).
• During GFOV, the phase-neutral voltage on the unfaulted phases can
rise to the phase-phase voltage, so any phase-neutral load sees a
173% overvoltage.

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV cont’d
Four-wire circuit; ungrounded
(three-wire) V-source DRs

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV voltage vectors

Unfaulted; Van ≈ nominal Faulted; Van ≈ Vac


 GFOV

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV in symmetrical components
For a fault on phase C in a circuit fed by an
ungrounded synchronous machine:

VBN
VAN

VCN
The VAN and VBN vectors are being held in place by the
generator, but VCN collapses and the result is a big
zero sequence component. This is GFOV.
Copyright 2015 Michael E. Ropp. All rights reserved.
Sequence
networks of a
distribution
feeder with PV,
pre-fault

This is the way this is


typically drawn, with
one exception: I’ve
dashed in the loads
(this is a balanced Z
load).

Copyright 2015 Michael E. Ropp. All rights reserved.


Sequence
networks of a
distribution
with PV and an
SPG fault

The three sequence


networks are
connected in series,
at the point of the
fault, through 3× the
fault Z (or R).

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV in the
sequence
domain—
inverter

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV in the sequence domain with inverters
• Notice that when using inverters without a grounding transformer, it’s
the loads and any other phase-neutral impedance that complete the
zero-sequence circuit.
• Normally loads are left out of sequence networks. OK for voltage
sources; wrong for current sources.
• Unbalanced loads will appear all along the feeder and will act to
couple the phases through different impedances. Gets complicated.

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV and synchronous machines

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV and synchronous machines
• Synchronous machines are well-modeled as a voltage behind a
variable Thevenin impedance.
• The controls (governor, etc.) are of secondary importance during the
initial phase of the fault; the machine physics predominate.
• When the SPG fault (previous slide) hits, VNn  Ecn. This is GFOV.
• Theoretically, the GFOV is 1.73 pu.

Copyright 2015 Michael E. Ropp. All rights reserved.


Example: GFOV with an ungrounded
synchronous generator in PQ control mode

Copyright 2015 Michael E. Ropp. All rights reserved.


NOTE: 59 AND 81O RELAYS TURNED OFF TO ENABLE VISUALIATION OF EFFECT
GFOV and inverters

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV and inverters
• Inverters are best modeled as a current source controlled to maintain
constant power. A Norton impedance can be added in some cases.
• Inverters are extremely fast; during the fault transient, inverter
control behavior is critical and must be represented.
• When the SPG fault hits (previous slide), VNn  IC × Z, which is the
voltage drop across the series impedance and by design is typically <
5% of the nominal V.
• This means that the normal GFOV mechanism associated with
synchronous generators does NOT exist with inverters.

Copyright 2015 Michael E. Ropp. All rights reserved.


BUT… that’s not the whole story
• If you have a transformer that breaks the zero-sequence path as seen
from the feeder side, you can get a GFOV, no matter what the DER is.
• If you have a purely ungrounded three-wire circuit with no Yg-
connected loads, you will see GFOV, but it’s not applied to any load.
• You can have GFOV and LROV at the same time: power rejection to
the unfaulted phases. But this can be mostly negative-sequence.
• Three-phase systems made up of three sets of single-phase inverters:
the jury is out. Stand by.

Copyright 2015 Michael E. Ropp. All rights reserved.


Watch out for backfeed of substation
transformers

Copyright 2015 Michael E. Ropp. All rights reserved.


GFOV with inverters: phase Vs during SPG
fault, 1:1 gen:load ratio, R load

Copyright 2015 Michael E. Ropp. All rights reserved.


NOTE: 59 AND 81O RELAYS TURNED OFF TO ENABLE VISUALIATION OF EFFECT
GFOV with inverters—sequence voltages
during SPG fault, 1:1 gen:load ratio

Copyright 2015 Michael E. Ropp. All rights reserved.


NOTE: 59 AND 81O RELAYS TURNED OFF TO ENABLE VISUALIATION OF EFFECT
Average per-phase power during the
SPG/island event

Copyright 2015 Michael E. Ropp. All rights reserved.


NOTE: 59 AND 81O RELAYS TURNED OFF TO ENABLE VISUALIATION OF EFFECT
Mitigating GFOV
• For rotating machines: use effective grounding. A grounding
transformer will do the job, and standard design methods apply. See
IEEE 142 (the Green Book).
• If the generator is already grounded, you might need a grounding
reactor to moderate fault currents, but you should not need a
grounding transformer.

Copyright 2015 Michael E. Ropp. All rights reserved.


For inverters: why not just add a grounding
transformer?
• There are a lot of reasons.
• Proliferation of grounding transformers would make feeder protection
difficult.
• Adding grounding transformers, plus the protection of the grounding
transformer itself (and shunt-tripping of the PV), adds a lot of cost to projects.
• There is some evidence that including grounding transformers could modestly
increase the THD in inverter output currents.

Copyright 2015 Michael E. Ropp. All rights reserved.


Do you need a grounding transformer with
inverters?
• No, you do not, if the PV GSU transformer is Yg-Yg. This approach
solves the problem at lower cost and without interfering with existing
protection.
• Yes, you do, if the PV plant GSU transformer does not have zero-
sequence continuity. (Then it must be on the MV side of the GSU
transformer.)

Copyright 2015 Michael E. Ropp. All rights reserved.


Sequence
networks:
SPG fault,
inverter, and a
grounding
transformer

Copyright 2015 Michael E. Ropp. All rights reserved.


Sizing an inverter grounding transformer
• If you do need a grounding transformer, you need to specify six
electrical parameters:
• Terminal voltage (known)
• Zero-sequence fault current, Ig
• Duration of zero-sequence fault current (usually taken to be 2 s)
• Continuous zero-sequence circulating current, Ig’
• Zero-sequence resistance Rg and reactance Xg
• Note that kVA rating isn’t on here. Grounding transformers are not
load-serving, so the kVA rating is not particularly meaningful.

Copyright 2015 Michael E. Ropp. All rights reserved.


Sizing an inverter grounding transformer
Finding Rg and Xg:
2
𝑉𝑃𝑉 𝑋𝑔
𝑍𝑏𝑎𝑠𝑒,𝑃𝑉 = 𝑋𝑔 = 0.6 × 𝑍𝑏𝑎𝑠𝑒,𝑃𝑉 ≥4
𝑆𝑃𝑉 𝑅𝑔

Finding the circulating Finding the 2-s fault


current Ig’: current Ig:

Copyright 2015 Michael E. Ropp. All rights reserved.


Protecting a grounding transformer
• Grounding transformers must have overcurrent protection.
• The foregoing procedure will rate the grounding transformer acceptably for
normal steady-state imbalance and for faults.
• BUT, there can be abnormal imbalance (ex.: tripping of large single-phase
lateral) that can persist for much more than 2 s. In this case, the grounding
transformer should trip offline.
• Grounding transformers should have thermal overload protection.
• If the grounding transformer trips, the PV plant must trip also.

Copyright 2015 Michael E. Ropp. All rights reserved.


Load Rejection Overvoltage (LROV)
• LROV can happen if:
• a section of a power system is isolated from the larger grid,
• with current-source DERs,
• During a time of heavy export.
• The isolated section may include any combination of phases (i.e.,
single-phase opens).
• LROV can occur on three- or four-wire circuits, and on two-phase or
single-phase laterals.
• LROV could in theory reach well over 2 pu; laboratory examples of
2.25 pu have been observed for 100% load rejection.
Copyright 2015 Michael E. Ropp. All rights reserved.
LROV cont’d
Exporting circuit; I-source DRs
P

Copyright 2015 Michael E. Ropp. All rights reserved.


LROV in the sequence domain
• LROV can have positive, negative and zero-sequence components.
• Three-phase breaker with balanced load: + only.
• Single or two-phase interruption: LROV on the open phases; +/-/0.
• Unbalanced loads: current source DERs will produce unbalanced voltages;
LROV can be +/-/0.
• Both LROV and GFOV can include zero sequence voltage, but they’re
separate physical mechanisms; you can have an LROV without a fault,
or on a three-wire or single-phase circuit.

Copyright 2015 Michael E. Ropp. All rights reserved.


Unbalanced LROV during islanding

The inverter produces positive


sequence current that flows
into the unbalanced load R.
Ohm’s Law gives a voltage
that’s as unbalanced as the
load R is. If the load is R + jX,
the angles will also change.

Copyright 2015 Michael E. Ropp. All rights reserved.


LROV and inverters
If the generation:load
ratio is called GLR, a very
conservative estimate of
LROV would be:
∆𝑉 ∝ 𝐺𝐿𝑅
This relationship is
shown at the right. But
this neglects some
effects that will lower
∆V, so this is still too
high.

Copyright 2015 Michael E. Ropp. All rights reserved.


LROV and inverters
This is more typical of
what we see: LROV
tends to “roll off”
because of inverter
power stage gains and
overmodulation.

Copyright 2015 Michael E. Ropp. All rights reserved.


Example: inverter-driven LROV at 5:1
generation:load ratio, RLC load

Copyright 2015 Michael E. Ropp. All rights reserved.


NOTE: 59 AND 81O RELAYS TURNED OFF TO ENABLE VISUALIATION OF EFFECT
Mitigating LROV: rotating machines
• For rotating machines, LROV should not be a significant issue,
because they act as voltages behind impedance.
• Short term transient LROV may occur due to the long time constants
of the system, but relaying should catch this:
• Overfrequency
• Standard per-phase OV
• Negative sequence V

Copyright 2015 Michael E. Ropp. All rights reserved.


Mitigating LROV: inverters
The standard inverter topology
includes an antiparallel
rectifier, to enable out-of-
phase current flow.

If the AC side voltage rises


above the DC side voltage, this
rectifier will act as a clamp,
clamping the AC peaks to the
DC voltage level.

Copyright 2015 Michael E. Ropp. All rights reserved.


Mitigating LROV
• The picture is not entirely clear.
• On the one hand, we may not need any mitigation of LROV with inverters
because of the DC clamping and self-protection mechanisms.
• NPPT modeling and NREL testing support this hypothesis.
• However, inverter DC voltages are becoming quite high, and this may change
the picture.
• Further modeling and testing is ongoing; stay tuned.
• Additional mitigation means:
• Limit the PV penetration.

Copyright 2015 Michael E. Ropp. All rights reserved.


Impacts of PV on distribution
feeders, Part 2: protection
impacts
Module 2

Copyright 2015 Michael E. Ropp. All rights reserved.


Impact on coordination: inverters
• The impact on coordination is all about fault current contribution.
• Inverters themselves contribute little to fault current (typical
assumption is 1.2 pu; IEEE 1547.8 recommendations are based on 1.6
pu). Inverters connected to feeders via Yg-Yg GSU transformers
should have essentially no impact on coordination.
• If they are not Yg-Yg, inverter plant transformers may cause significant
changes in asymmetrical fault currents, mostly through impact on
zero-sequence Z.
• Yg-∆ transformers (Yg MV) will need grounding impedances
• Grounding transformers need to have sufficiently high impedance to avoid
excessive coordination impacts

Copyright 2015 Michael E. Ropp. All rights reserved.


Impact on coordination: inverters
• There can be some protection impacts if there are a LOT of inverters.
(a large number of 1.2 pu contributions add up)
• TCCs may need to be adjusted on feeders with many inverters, but if
that is insufficient different protection approaches may be needed.
• Generally can’t use fuses; increases cost
• Negative-sequence relaying can work
• Impedance-based schemes sometimes work
• Communications may be needed—variants of differential protection usually
are effective but costly
• Novel methods—stay tuned!

Copyright 2015 Michael E. Ropp. All rights reserved.


Impact on coordination: rotating machines
• Rotating machines contribute significantly to fault current, and this
protection needs to be adjusted.
• Potential mitigation strategies:
• Fast DG tripping (DTT, high-speed negative sequence relaying)
• Limiting allowable DG rating
• Protection system reconfiguration—distance relaying, directional relaying,
added reclosers/zones
• Adaptive protection (make sure this is modeled and tested!)

Copyright 2015 Michael E. Ropp. All rights reserved.


Impact on FLISR
• The situation here is similar: because inverters contribute little to
fault current, they should have little impact on FLISR, but inverter
transformers can have significant impacts for asymmetrical faults.
• Zero sequence impedances need to be made high enough to minimize
impacts. IEEE 1547.8 recommendations should be OK.
• Because rotating machines can contribute large fault currents, FLISR is
strongly impacted by these. Usually in this case you need:
• Adaptive protection
• Modeling and study to set it up properly

Copyright 2015 Michael E. Ropp. All rights reserved.


Impacts of PV on distribution
feeders, Part 3: unintentional
islanding
Module 2

Copyright 2015 Michael E. Ropp. All rights reserved.


Definition of islanding

Copyright 2015 Michael E. Ropp. All rights reserved.


Loss of mains detection (LOMD) solutions
Type of method Description Effective- Cost Compatibility
ness with grid
support
functions
Passive Monitors terminal voltage Fair to $ Good to
for trip condition, but poor, if excellent
does not adjust inverter high speed
output to cause change required
Active Actively changes inverter Excellent $$ Poor to very
output to cause change, in single poor
usually uses positive inverter
feedback cases
Communications- Relies on communications Excellent $$$$ Excellent
based for system awareness (?)

Copyright 2015 Michael E. Ropp. All rights reserved.


Under what conditions is islanding a concern?
• In general, if there are no VAr sources in the island, the likelihood of
an unintentional island is nearly zero.
• No single manufacturer’s product is more than 2/3 of the total DG
• Mixture of rotating machines and inverter-based DG
• New market entrants may use unproven detection methods
• Risk of islanding also rises with grid support functions

Copyright 2015 Michael E. Ropp. All rights reserved.


Run-on time for non-problem case

Copyright 2015 Michael E. Ropp. All rights reserved.


Run-on time for problem case

Copyright 2015 Michael E. Ropp. All rights reserved.


How anti-islanding is dealt with today
• Nearly all commercial inverters use active anti-islanding that
destabilizes the island frequency when an island forms.
• UL-1741 listing includes testing of anti-islanding functionality with a
resonant RLC circuit, Q = 1
• If a circumstance arises in which active AI is insufficient, usually resort
to DTT.

Copyright 2015 Michael E. Ropp. All rights reserved.


Why islanding is reemerging as a concern
• Much higher DG deployment levels means more of the problem cases
mentioned on a previous slide
• Also, as DG penetration increases, need DG to act more as a
generation asset—need grid support functions. These conflict with
today’s active anti-islanding.
• Eventually, new anti-islanding means are needed for the future.

Copyright 2015 Michael E. Ropp. All rights reserved.


New anti-islanding means
• Communications-based methods
• Synchrophasor-based methods
• Power line carrier permissive
• New passive methods
• Signal processing based methods
• Statistically-based methods
• Advanced decision trees and combinations of methods

Copyright 2015 Michael E. Ropp. All rights reserved.


DG and resiliency: microgrids
Module 3

Copyright 2015 Michael E. Ropp. All rights reserved.


What is a microgrid?
• The IEEE’s definition: the closest we have come is in 1547.4-2011,
where we define a “distributed resources (DR) island system”.

“DR island systems are parts of electric power systems (EPSs) that
have DR and load, have the ability to disconnect from and parallel
with the EPS, include the local EPS and may include portions of the
area EPS, and are intentional and planned.”

Copyright 2015 Michael E. Ropp. All rights reserved.


Key concern today: system
hardness against major
disasters (storms, sabotage)

Copyright 2015 Michael E. Ropp. All rights reserved.


Key concern today: system
hardness against major
disasters (storms, sabotage)

Copyright 2015 Michael E. Ropp. All rights reserved.


Improving disaster readiness/hardness
• Need to improve reliability indices
• xAIDI: Average Interruption Duration Index
• xAIFI: Average Interruption Frequency Index
• SAIDI/SAIFI = system level; CAIDI/CAIFI = customer level
• There are limits to the traditional centralized approach
• Moving to UG improves xAIFI but degrades xAIDI, and can be expensive
• FLISR will help a lot for local situations, but will help minimally for major
events
• DG will be a key part of the solution
• Ideally, these distributed resource islands could interact with each other

Copyright 2015 Michael E. Ropp. All rights reserved.


Microgrid design—what isn’t new

Copyright 2015 Michael E. Ropp. All rights reserved.


Microgrid design—what isn’t new
• Standby/backup power supplies have been around forever—they are
a key part of improving reliability to critical loads (reducing
CAIFI/CAIDI).
• Mostly small engine-gensets with automatic or manual transfer switches.
• Typical automatic switchover time is 20-40 s. If you need* seamless
transitions, you add a battery.
• For the most part, design of backup power systems assumes:
• Single dominant energy/power source (i.e., all sources connected at the same
bus)
• Grounding per IEEE 142 (the Green Book)
• Grid interactivity totally controlled by the local system operator

* “need” really means want to pay for or can economically justify


Copyright 2015 Michael E. Ropp. All rights reserved.
Microgrid design—what IS new
We’re violating the assumptions.
• More complicated system
topologies.
• Multiple sources with widely
varying characteristics.
• Higher power quality and
reliability demands (seamless
transitions?).
• New economic demands, on-
and off-grid.

System integration and controls design are critical.


Copyright 2015 Michael E. Ropp. All rights reserved.
Making it all work
• Transient simulation is extremely valuable.
• Design stage: enables better, more optimized, less expensive design of
controls and protection and improved understanding of expected
performance.
• Operations stage: enables rapid and accurate system diagnostics and
corrective action.
• HIL and PHIL simulation is extremely valuable.
• Can test actual hardware quickly, inexpensively, and nondestructively.

Bottom line: modeling enables more reliable and better-performing


systems, better suited to customer needs, in less time and at lower
cost.
Copyright 2015 Michael E. Ropp. All rights reserved.
Energy storage and microgrids
• A BIS can, if controlled properly:
• Enable seamless on-off grid transitions.
• Allow higher PV penetrations to be accommodated.
• Improve reliability.
• Improve off-grid system dynamics.
• But many microgrids do not include a BIS.
• They’re expensive, especially over the life cycle.
• Energy and power sizing are often oversimplified and not done properly.
• Controls are often not well designed or implemented.
• Bottom line: it’s easy to spend a lot of money on a BIS and not get what you
wanted.

Copyright 2015 Michael E. Ropp. All rights reserved.


Case study #1

Case study #1: High Reliability Zone


• Goal: from an existing distribution feeder with ~5 MW existing diesel
generation, using a new Battery-Inverter System (BIS), create a
microgrid that:
• Can “seamlessly” connect to and disconnect from the main grid
• Increases reliability of electric service to critical loads
• Can participate in grid support in either autonomous or commanded modes
• NPPT’s role:
• System-level modeling
• Design of controls for the BIS and optimization of generator controls
• Support for design of load shedding and system reconfiguration
• Other engineering support as needed (mostly test plan designs)

Copyright 2015 Michael E. Ropp. All rights reserved.


Case study #1
First step: development of system
performance goals
• In this case, a “seamless” transition is defined as one in which:
• The voltage remains within the CBEMA/ITI curve limits at all times; and
• The frequency remains between 59.3 Hz and 60.5 Hz
• The key is speed of response.
• The BIS and its controls have to detect the condition and respond to it quickly
enough to keep parameters within spec.
• The BIS must be large enough in power to carry the entire microgrid load, and
large enough in energy to carry that load until the diesels can start, sync, and
take over.
• The control path delays must be carefully managed.

Copyright 2015 Michael E. Ropp. All rights reserved.


Case study #1

HRZ system modeling DG sites

• 5 MW of standby/backup engine-
gensets (diesel)
• Added a 5 MW, 1.25 MWh BIS at
the substation
• Desire to add PV later, perhaps as
much as 25% penetration

Substation with BIS


Copyright 2015 Michael E. Ropp. All rights reserved.
Case study #1

Controls solution designed


• It became clear early on that the main obstacle to achieving seamless
transitions was the BIS mode transition between standard grid-tied
current control and off-grid V-f regulation control.
• NPPT’s solution was to design a controller that was in droop mode at
all times, even while grid-tied. Proper design of adaptive droops and
a secondary controller are thus critical.
• Detailed computer modeling enabled determination of the required
parameters.

Copyright 2015 Michael E. Ropp. All rights reserved.


Case study #1
Use case #1
Use case 1: seamless separation of the
HRZ from the grid because of a voltage
depression to 0.7 pu.

Top plot: BIS picks up load power with


minimal disruption.

Bottom plot: frequency remains within


bounds.

Copyright 2015 Michael E. Ropp. All rights reserved.


Case study #1

Use cases #1 and #2


Use cases:
#1: seamless separation of the HRZ
from the grid because of a voltage
depression to 0.7 pu.
#2: seamless separation of the HRZ
from the grid due to a fast frequency
drop to 59.5 Hz (UFLS).

The plot shows that in terms of voltage,


the on- to off-grid transition is seamless
for both Use Cases.

Copyright 2015 Michael E. Ropp. All rights reserved.


Case study #1

Use case #3 • On- to off-grid transition starting from nominal


voltage, at 3 s
• BIS hands V/f regulation to diesels, at 6.6 s
• BIS transitions from discharging to charging
mode form 15 to 25 s, under droop control
• Step doubling of microgrid load, at 25 s (using
motor-containing dynamic load model)
• Step halving of microgrid load, at 35 s

The frequency remains within bounds throughout.

Copyright 2015 Michael E. Ropp. All rights reserved.


Case study #1

Use case #3
Use Case #3
• On- to off-grid transition at 3 s
• BIS hands V/f regulation to diesels at
6.6 s
• BIS transitions from discharging to
charging mode form 15 to 25 s
• Step doubling of microgrid load, at 25
s (using motor-containing dynamic
load model)
• Step halving of microgrid load, at 35 s

The microgrid voltage almost remains within bounds during the entire sequence. The one exception is during the
stepwise doubling of the load. Working with the utility client, it was determined that such a transition was probably
unlikely in the field, so it was decided that this was not a concern.

Copyright 2015 Michael E. Ropp. All rights reserved.


Case study #2

Impact of control path delay

Essentially, too much control path delay renders the BIS


ineffective, and eventually causes instability.
Copyright 2015 Michael E. Ropp. All rights reserved.
Case study #1

Key lessons learned


• It is possible to achieve seamless on/off grid transitions by eliminating
the mode transition and having the BIS in droop mode all the time.
• It is possible to seamlessly hand off control from the BIS to the diesels
and transition the BIS into recharge mode using the secondary
controller.
• Keeping the BIS in droop mode while recharging has significant
benefit.
• PV can be accommodated (more on this in a minute).
• BIS control path delays are absolutely critical and must be kept short
(well under 50 ms).

Copyright 2015 Michael E. Ropp. All rights reserved.


The role of PV in emergency power
• PV is often already available because of on-grid economic considerations.
• PV reduces fuel usage.
• Improves off-grid run-time without refueling (if running from local stores).
• Can reduce costs in some cases.
• Can be VERY helpful in recovery from major devastating events
• PV is noiseless and often has lower zoning/permitting barriers than engine-
gensets.
• PV technology itself is very reliable.
• However, adding PV to a microgrid can reduce reliability if not integrated
properly.

Copyright 2015 Michael E. Ropp. All rights reserved.


What happens as renewable penetration level
(P) rises in a low-inertia system
P Definitions of “mains” and “DG” questionable; generators = backup
for PV? Must have storage above 60% PV.
 50% • PV must act as a system asset
• Storage, curtailment, coordination, grid support,
ramp rate controls all important

 30% • PV variability impacts mainline gens, but problems can generally still be solved by
gen control adjustments OR LOAD CONTROLS
• Mass tripping events start to become a problem—need FRT in PV inverters (H and L!)
• Minimum diesel loading starts to become a constraint
 10%
PV variability disappears into load variability

System impacts
Copyright 2015 Michael E. Ropp. All rights reserved.
Mitigating PV impacts on low-inertia systems
• Use the right inverters!
• Otherwise, should limit PV penetration—actual limit without mitigation is
system-specific, usually ≈ 30%
• Widen frequency trip setpoints in the PV inverters—usually ≈ ± 5% of
nominal frequency
• Carefully tune the generator controls
• The conventional “tune and watch” approach is insufficient
• Consider smaller, modular engine-gensets
• Use of grid support functions in inverters can help—more shortly
• For all of these—need quantitative guidance to set all parameters

Copyright 2015 Michael E. Ropp. All rights reserved.


Frequency-watt function
Power

Pnom

Frequency
flo1 fhi1 fhi2
The PV inverters curtail their output above a setpoint fhi1 and cut off
completely at fhi2 and at flo1.
Copyright 2015 Michael E. Ropp. All rights reserved.
Can you do PV only (w/o storage)?
• Some customers want to take advantage of incentives available for PV
and want to use that PV as some kind of backup, but they do not
want the expense of storage.
• To do this, one would have to do the following:
• You must have the right inverters—must be able to operate in isochronous
mode.
• You need very fast-acting BEMS/load control. You’d have to make sure that
your facility load is below the available irradiance at all times.
• You will not have power at night, of course.
• For most customers, this doesn’t pencil out.

Copyright 2015 Michael E. Ropp. All rights reserved.


Can you do PV only (w/storage)?
• Absolutely. Many off-grid plants that have PV only as the primary
energy source with battery backup.
• These are GREAT in storm recovery.
• No fuel delivery requirements
• PV is storm-hard; proven in blizzards, hurricanes, ice storms
• If you go this route, two key things to bear in mind.
• The PV buys you extended off-grid time, but the storage itself buys you
reliability during shorter outages. PV may need to be cost-justified by on-grid
considerations.
• Load control and system management when off-grid are essential to avoid
dissatisfaction.

Copyright 2015 Michael E. Ropp. All rights reserved.


Cost vs. availability for a PV-only system
(w/storage)
5

Relative PV system cost 4


3
2
1
0
0.5 0.6 0.7 0.8 0.9 1
Design availability
Copyright 2015 Michael E. Ropp. All rights reserved. 134
Don’t forget about protection!
• Protection includes:
• Fault current interruption
• Avoiding inadvertent energization of anything
• Avoidance of unintentional islands of DG
• In more complex microgrids, traditional protection schemes no longer
provide good results:
• Systems with multiple sources
• No single dominant source
• Current-limited sources (i.e. inverters)
• Multiple reconfigurable systems
• FLISR schemes
• If your system falls into any of these categories, pay special attention to
protection aspects

Copyright 2015 Michael E. Ropp. All rights reserved.


Modeling of DG and low-inertia
systems
Module 4

Copyright 2015 Michael E. Ropp. All rights reserved.


Modeling can have very high value
• Pros
Reduces risk
Improves fundamental understanding
Shortens development/commissioning times
Allows testing that can’t be done experimentally
Can significantly decrease overall costs
• Cons
Can be expensive
Need a lot of data
Validation is critical
Need someone who knows what they’re doing—easy to build a lying model

Copyright 2015 Michael E. Ropp. All rights reserved.


When modeling is most useful
• Complex systems—multiple sources, complex protection, networked
systems, multiple source types, weird loads…
• Highly critical situations in which field adjustment is not possible
(example: sizing a grounding transformer such that feeder conditions
do not lead to excessive false trips)
• Situations with mixtures of source types (both inverters and rotating
machines)
• Inverter-dominated systems
• Systems with energy storage

Copyright 2015 Michael E. Ropp. All rights reserved.


Must use the right type of model
• Calculation automators—for standardized calculations
• CYME
• Aspen One-Line
• Electromechanical (“slow dynamic”) simulators—for machine dynamics and
longer-scale transients on large systems (not power electronics)
• PSS/E
• PSLF
• Full transient (“fast dynamic”) simulators—power electronics, fast transients, full
details (minimum assumptions, but maximum data input requirements)
• EMTP variants—EMTP-RV, PSCAD, ATP
• MATLAB/Simulink
• DigSilent PowerFactory
• Steady-state simulators—for time series steady-state snapshots (“8760 sims”)
• OpenDSS

Copyright 2015 Michael E. Ropp. All rights reserved.


Why doesn’t everybody do simulation?
Perceived value
Cost
Understanding gained

0% Laboratory Controls and Controls HIL 100%


(actual HW on scale HW power HIL (pure simulation)
actual system) Copyright 2015 Michael E. Ropp. All rights reserved.
Example: modeling GFOV with rotating
generators
• In reality, the GFOV never reaches 1.73 pu; transient modeling is
needed to really get it right.
• Standard low-order machine models work reasonably well for this;
don’t need full-blown sixth-order models.
• AVR/exciter should be represented.
• Governor should be represented, but a constant-torque or constant-
power model may be an acceptable approximation over short time
intervals.

Copyright 2015 Michael E. Ropp. All rights reserved.


Example: modeling GFOV with inverters
• Must properly represent inverter controls
• Need sufficiently short time steps
• Represent saturation/overmodulation—very important with switch-averaged models
• Must properly represent inverter power stage gain
• Constant-current models do NOT work
• To get initial TrOV right, must have inverter output filter components
modeled
• All transformers need to be modeled
• Power-limited nature of DC source may be important
• Ideal DC voltage source models may not work

Copyright 2015 Michael E. Ropp. All rights reserved.


Switched vs. averaged models of inverters
• Constant-current representations do NOT work for GFOV, but do
averaged models work?
• Our simulations suggest that averaged models CAN be acceptable for
GFOV work, as long as the controls are represented in sufficient
detail.
• Averaged models do tend to slightly underestimate the zero sequence
voltage components vs. switched models, but the difference is small.

Copyright 2015 Michael E. Ropp. All rights reserved.


Example inverter model

Copyright 2015 Michael E. Ropp. All rights reserved.


Validation Simulated and experimental inverter terminal voltages, no neutral, with parasitic Z
800
Sim A
results 600 Sim B
Sim C
Exp A
400
Exp B
Exp C
200

Voltage (V)
0

-200

-400

-600

-800
6.98 6.99 7 7.01 7.02 7.03 7.04
Time
Copyright 2015 Michael E. Ropp. All rights reserved.
GFOV test feeder

Copyright 2015 Michael E. Ropp. All rights reserved.


Low-inertia systems and modeling
• We highly recommend partnering with an expert in transient
modeling of controls and protection if you are implementing:
• A microgrid
• A storage solution for reliability enhancement
• Any system that can operate off-grid and that has multiple distributed sources
• What the modeling does
• “Virtual laboratory”
• Dramatically improves controls design
• Allows design of effective protection
• Enables faster commissioning
• Facilitates later system diagnosis

Copyright 2015 Michael E. Ropp. All rights reserved.


Thank you!

Copyright 2015 Michael E. Ropp. All rights reserved.


Questions?
michael.ropp@northernplainspower.com

Copyright 2015 Michael E. Ropp. All rights reserved.

You might also like