1. Power System Operation
Janusz Bialek

2. Syllabus Main power system characteristics
Why power system looks the way it does
The future power system
Maintaining power balance
Illustration using historic blackouts

3. Breakdown of electricity bill (Ofgem, 2018)
17%, ?? in 2013
33% (58% in 2013)
Transmission increase – bootstraps and network investment. Check small supplier margin
17% (11% in 2013)
25% (20% in 2013 incl. 4% transmission)

4. Power System in 20th Century
Unidirectional flow of power from power stations to customers
Highly predictable and passive demand.
Limited number of large controllable power stations
Limited energy storage (pumped hydro)
Dispatch: generation follows demand
Legacy infrastructure

5. 20th Century power system planning and control
Planning target
Planning:
Infrastructure (G, T, D) planned to serve peak load + reserve
Very low utilization rates (50% generation, 30% transmission, even lower distribution)
Control:
Centralized, hierarchical
Deterministic
Preventive based on (N-1) reliability condition (not in Russia)
Very conservative - system expensive to build and operate
But very reliable!

6. Power System in 21st Century
Still legacy infrastructure but:
Bi-directional flow of power
Stochastic generation and demand
Large number of non-controllable stochastic power sources
Energy storage
Growth in ICT infrastructure (sensors, Internet, WAMS) enabling a change of paradigm of power system planning control
ICT: information and comms technology
Students are attracted: smart

7. Power system planning and control in 21st Century
from preventive (N-1) to corrective (just-in-time)
from deterministic to stochastic
from centralized to distributed
Storage changes dispatch – time linkages
No longer demand-driven generation, controllable loads
Result: reduced security margins, cheaper operation
Planning:
Changes in control reduce the need for infrastructure (G, T, D)
Improved utilization rates, deferred investment
But must maintain reliability!

8. Future Power Market
MCP2
MCP1
Market clearing price (MCP) is set by the marginal generator
Increasing load increases prices when the marginal cost curve is increasing
Sub-marginal generators receive rents (MCP – MC) to cover their fixed costs
Will it function when the marginal cost curve is (nearly) flat?
Source: PJM

9. Unique characteristics of the power system
Generation = demand + losses (including storage)
Power flows in the network according to the laws of physics (Ohm’s Law, Kirchhoff’s Current and Voltage Laws)
Power network is not a transportation network
Flows cannot be directly routed – a given line flow depends on generation/demand pattern in the whole system
There are no valves – network congestion may result in overheated circuits and damage
Congestion management requires coordinated redispatch
Compared to other networks. Laws of physics always win with the laws of economics.

10. Loop flows in a meshed network
In meshed networks power does not flow directly from A to B but through all transmission lines
Only 38% of F-I trade flows directly
Loop flows via the neighbours from wind in Northern Germany
Overload the neighbours networks if dispatch is not coordinated (it isn’t)
Phase-shifting transformers (PST) – direct (partial) control of power flows
When the laws of physics collide with the laws of economics, physics wins every time (George Loehr, after Jack Casazza). Power flows can be controlled only indirectly, by manipulating generation/demand pattern

11. Why interconnected?

12. Why interconnected?
Using the cheapest energy sources wherever they are
Taking advantage of differences in living (and loading) patterns: peak-shaving
Sharing generation reserve in case a power station is suddenly lost. The larger the area the greater the saving.
Assume 3 equal areas each requiring 2GW reserve when operating separately – total reserve 6 GW.
Connecting them gives the total reserve (assuming that loss of generation in each area is an independent event)
42% saving
Sharing reserve is the main historic reason for interconnection

13. Why so many voltage levels?
Transmission losses: Ploss = RI2
Power = voltage x current
Increasing voltage twice reduces current twice and therefore reduces transmission losses 4 times
But increasing voltage level increases the cost of insulation
Transmission lines – longer distances, higher voltage (275 kV, 400 kV)
Distribution lines – shorter distances, lower voltage (132 kV, 33 kV, 11 kV)

14. How many wires/circuits do you see?

15. Single Phase Synchronous Machine
Faraday’s Law: induced voltage = rate of change of flux
end of the arrow E
Head: pointy bit
max
direction of winding: head of the arrow
A rotating magnet will induce a sinusoidal voltage in the stator’s coil.

16. 3 Phase Machine (i.e. with 3 sets of coils located at 1200 intervals around the stator)Y B
The three windings produce three ac voltage waveforms (red, yellow and blue) shifted by 1200 in phase wrt each other

17. joining the return wires givesY R Y B = Y Y Y B B B B N R Y B
Three-Phase Circuit = 3 single-phase circuits with common return (neutral)

18. Assuming balanced loading (ZR = ZY = ZB ), the return current is:
IN = IR + IY + IB = I sin(t) + I sin(t ) + I sin(t ) =
= I (sin(t) + sin(t ) + sin(t ) ) = 0
As the return current is zero, the wire can be removed resulting in significant savings!
Three single-phase systems require 6 wires connecting the source and the load. An equivalent three-phase system requires only 3 wires so the saving is 50% on all transmission and distribution lines!
In practice there is always some amount of load imbalance so a return wire is usually provided. But the return current is small so the return wire may be much thinner than the phase wires – see the photo.
If the load is balanced, we use a single-circuit representation of a 3-phase circuit.

19. Why AC rather than DC? Transmission transformer
Distribution transformer

20. Transformers Faraday’s Law:
Michael Faraday
Faraday’s Law:
Voltage induced (emf) is proportional to the rate of change of flux, E = - dΦ/dt
Transformer must be fed by ac voltage
Power electronic transformers are available but they are less efficient and cannot handle high power flows

21. The other reasons for AC
Induction motor: the workhorse of the industry
How to maintain a power balance without measuring all the generation and demand?

22. Frequency control: how to maintain a power balance (generation=demand) in an AC system?
In a time scale of minutes, the balance of power is maintained by power system operator instructing power plants to follow demand
How is an instantaneous power balance maintained when e.g Sizewell B trips?

23. What happens when say Sizewell B trips?
Immediately afterwards demand exceeds generation
But the lights stay on: where does the balance come from – no time for the dispatcher to give command to increase generation?
Inertia - kinetic energy stored in rotating masses
The mass keeps rotating even if the driving force is removed
Ek = I ω2/2 where I is the moment of inertia (kg m2), ω is the rotational velocity (rad/s) and ω = 2  f (for 2-pole machines)
That kinetic energy stored in the rotating plants provides a cushion for any momentary imbalance of power
Frequency drops indicating power imbalance and triggering control
Frequency is a universal power imbalance – the same everywhere in the system

24. Primary frequency control: restoring power balance after a large plant drops off
Stage 1 - inertial response:
power deficit is covered from inertia (kinetic energy) of all synchronous generators
If nothing else happens, frequency will keep dropping
Stage 2 – reaction by Turbine Governors:
frequency drop is sensed by Turbine Governors
Turbine Governors open more the turbine valves hence releasing more steam/water (increasing power output)
This activates adjustment of the steam generation process (in a thermal plant)
Increasing output power continues until power balance is restored: frequency drop is halted but it is less than 50 Hz
steady-state error typical for proportional (P) control

25. Primary frequency control
Primary frequency control is fully automatic – all generators must be equipped with Turbine Governors with a droop characteristic
Power stations have to operate derated to provide a headroom (frequency reserve)
speed
Actual output
power

26. Frequency trace following a large infeed loss
Initial slope of decline is determined by system inertia (or cumulative inertial response of all generators)
Increased output due to action of Turbine Governors
Frequency drop halted but at a lower level
Restoring power balance
Primary frequency control is fully autonomous and decentralised (turbine governors installed on all generators)
Primary Frequency
Control

27. Secondary frequency control: restoring the nominal frequency
Frequency is still less than 50 Hz
System Operator instructs power stations to increase generation and restore 50 Hz
Manual in GB: National Grid instructs most economic and flexible power stations to increase output
Automatic in Europe/USA: Automatic Generation Control – AGC. The required generation increase is allocated to individual power stations by a central controller – usually PI type.
Differences between primary and secondary frequency control
Primary: fully autonomous and all power stations contribute according to their droop characteristics
Secondary: only some (most economic and flexible) power stations contribute as directed by a central controller

28. Frequency trace following a large infeed loss
Initial slope of decline is determined by system inertia (or cumulative inertial response of all generators)
Increased output due to action of Turbine Governors
Initial frequency restored
Frequency drop halted
Restoring power balance
Restoring 50 Hz
Primary Freq.
Control
Secondary Freq. Control
Fully autonomous, decentralised
Centralised, manual or automatic

29. Frequency is a system-wide power balance indicator
No need for telecommunication – frequency can be measured anywhere
Another big advantage of AC power system: frequency provides a power balance signal the controllers can act upon.
How would you do it in a large DC system?
Load-frequency mechanism works only in AC system: DC links separate the systems (see GB-France)
AC interconnected system is the largest machine ever built (US National Academy of Engineering, 2003)

30. The bigger the system, the higher inertia and smaller frequency excursions
Example of large frequency variations in a small system: Ireland, August 5th 2005
Large power systems experience smaller frequency excursions

31. The effect of reduced inertia due to renewable penetration
EirGrid, SONI: “All Island TSO Facilitation of Renewables Studies”.
A limit on the amount of instantaneous wind generation+imports
Remedies: synthetic inertia, market for inertial response

32. Electricity Demand – TV Pickup Royal wedding 29th April 2011
Detailed Demand & Frequency Overview
People drop whatever they are doing to watch.
Frequency did not vary much as SO expected it and too action – they have TVs in control room 32

33. England vs Argentina: 30 June 1998
Put a kettle on. Wonderkid Michael Owen scoring a wondergoal, Beckham kicking an opponent and being sent out
Any Scots? Why am I not showing the effects of Scotland/Wales/Northern Ireland matches? I am not English.
Would NG be able to control frequency if England played Scotland in WC Final? No, because they plan only for credible events.
Every tournament – lower demand change

34. Demand may also rapidly decrease: effect of the total solar eclipse on National Grid demand 11th August 1999
36500
Tue 10/08/1999
3000MW increase in demand
Wed 11/08/1999
36000
2200MW drop in demand
Totality
35500
End of the eclipse
Start of the eclipse
35000
Demand MW
34500
Overall demand lower – many people developed sudden sickness and decided to say home. 2.2 GW drop – 2 large power plants
34000
33500
33000
09:00
09:10
09:20
09:30
09:40
09:50
10:00
10:10
10:20
10:30
10:40
10:50
11:00
11:10
11:20
11:30
11:40
11:50
12:00
12:10
12:20
12:30
12:40
12:50
13:00
Time

35. TV pick-ups happen every day: demand curve for Britain's Got Talent final 05/06/2010
Each act caused mini pick-up

36. Earth Hour (World Wildlife Fund), 26 March 2011
Danger of a “frequency collapse” if a frequency drop is too deep
Luckily, no accelerated fall in demand at 8.30 pm or pick-up at 9.30 pm
Just as well!
The roads to hell are paved with good intentions. Earth Hour is a worldwide event organized by the World Wildlife Fund (WWF) and held on the last Saturday of March annually, encouraging households and businesses to turn off their non-essential lights for one hour to raise awareness about the need to take action on climate change. The event, conceived by WWF and Leo Burnett, first took place in 2007, when 2.2 million residents of Sydney participated by turning off all non-essential lights.[1] Following Sydney's lead, many other cities around the world adopted the event in 2008.[2][3] Earth Hour 2012 took place on 31 March 2012 from 8:30 p.m. to 9:30 p.m., at participants' local time.

37. What happens when there is not enough spinning reserve to cover an unexpectedly high loss of generation?
Power system security is maintained by keeping spinning reserve (i.e. not loading all the generators up to the maximum) should a power plant be unexpectedly lost
(N-1 contingency)
The normal reserve is the loss of the largest infeed (a nuclear reactor of 1320 MW at Sizewell B)
Keeping a larger reserve (N-2) is costly while it is highly unlikely that 2 or more units will be lost simultaneously
… but “unlikely” does not mean impossible
How to stop a declining frequency (i.e. prevent frequency collapse) when there is not enough generation reserve to restore a power balance?

38. Two plants were lost. The total loss 1582 MW > 1320 MW planned
Frequency
Two plants were lost. The total loss 1582 MW > 1320 MW planned
The resulting frequency drop caused further loss of wind generation
National Grid reduced voltage to reduce demand by 1200 MW (P = V2 /R for resistive loads)
Automatic load shedding of 581 MW (580,000 customers), triggered by a frequency drop to below 48.8 Hz, restored the generation/demand balance
More generation was connected and the supply was restored within 1 hour
Longannet and Sizewell B

39. Dangers of interconnection: blackouts caused by transmission faults
When a line trips, power flows redistribute almost instantaneously potentially leading to a cascading blackout
That’s why (N-1) reliability criterion – power system must be secure not only for the assumed operating conditions but also when “something” happens (so-called a credible event)
UK system is planned to (N-D) standard
Urban networks often operate to (N-2) standard
Blackouts happen usually due to (N-2) (or more) events – but (-2) is not always a network fault

40. US/Canada blackout 2003
Statue of Liberty the only lighted object against New York skyline

41. Where it all began: Ohio and surrounding areas (USA and Canada)
Complicated situation:
Initiation of blackout involved 2 control areas (FE and AEP) and their respective reliability coordinators Midwest ISO (MISO) and PJM
FE consists of 7 utilities
AEP is both control area operator and transmission owner
Midwest ISO (MISO) is reliability coordinator for 37 control areas in four reliability regions.
PJM is AEP’s reliability coordinator
18 control-area-to-control area interfaces across PJM and ISO
Five RTOs/ISOs affected by blackout: MISO, PJM, NYISO, ISO New-England, Ontario Independent Market Operator (IMO)
Source: US/Canada Power System Outage Force

42. How it all started: a tree flashover at 3.05 pm
Source: US/Canada Power System Outage Force

43. Is (N-1) criterion good enough?
Blackouts are always due to a combination of unusual factors
Midwest ISO (MISO) state estimator was off-line during most of the afternoon
Alarm and logging system in FirstEnergy (FE) control room failed 1 hour before the cascade started
Not only it failed, but control room engineers did not know about it
Unknown unknowns: they were blind but they did not know about that they were blind
No real-time information sharing and collaboration between TSOs
Donald Rumsfeld

44. Three 345 kV lines trip between 3.05 and 3.41
Source: NERC

45. Effect of a line trip: increased loading on other lines potentially leading to a cascade
Source: US/Canada Power System Outage Force

46. Effect of a line trip: depressed voltage
Source: US/Canada Power System Outage Force

47. Source: US/Canada Power System Outage Force
Three lines trip on tree flashovers supplying Northern Ohio. Around lake Erie, Michigan, Ontario, New York, Pennsylvania. Cities: Cleveland, Detroit. Voltages depressed in Cleveland
Further trips separate northern Ohio from the supplies from the south
Full cascade starts. Separation spreads to Michigan. Northern Ohio supplied from Michigan. Increased transmission distance causes voltages to be severely depressed in Michigan
Michigan in voltage collapse. Separation mid Michigan. The only route to northern Ohio and Michigan is via Ontario and New York.
Source: US/Canada Power System Outage Force

48. Source: US/Canada Power System Outage Force

49. Speed of cascading
Source: US/Canada Power System Outage Force

50. Dangers of interconnection: Europe 2006bb
The Times, 6 Nov 2006
But it did. EON got it wrong

51. UCTE disturbance on 4 November 2006: a problem in Northern Germany caused cascaded tripping of lines and a split of the system into 3 separate areas
Initial disturbance
10 GW surplus
51.4 Hz
0.8 GW deficit
8.9 GW deficit
49.7 Hz
49 Hz
Source: UCTE

52. The system was re-synchronised after 38 minutes
Source: UCTE

53. What happens when load shedding is not quick enough
What happens when load shedding is not quick enough? Italian blackout 28 Sept. 2003

54. Italian blackout 28 September 2003
Restored power balance
Automatic load shedding and switching off pump storage relieved the power deficit but voltage problems caused power stations to trip as well and the total blackout

55. How to avoid being dragged down by the neighbours?
UK and Scandinavia are connected by submarine DC links with mainland Europe
The main reason is technical –AC vs DC transmission (see later)
The second reason: DC connections separate the systems as the load-frequency mechanism requires AC connection (generators must operate in synchronism)
UCTE disturbance did not affect UK or Scandinavia

56. US transmission system

57. Take-home messages
Power system of the (not-too-distant) future will be distributed, stochastic, low inertia, near-zero marginal cost and with storage changing the dispatch paradigm
Power flows cannot be routed – hence need for a coordinated dispatch
Why are power systems AC, three-phase?
Frequency is a global power balance indicator
Power balance following a disturbance is restoring by primary (droop) frequency control which is autonomous and distributed
The nominal 50 Hz (or 60 Hz) frequency is restored by centralised secondary frequency control
Low inertia can create problems with frequency control
DC links separate the systems electrically
Under-frequency load-shedding is the ultimate power balance restoration tool
Blackouts are usually caused by (N-2) events and spread out quickly