Power system operation, SCADA and EMS

In short

Power system operation is the real-time business of keeping generation equal to demand, frequency and voltage inside limits, flows within line ratings, and the network secure against the next credible outage. SCADA (Supervisory Control and Data Acquisition) is the hardware-and-communication layer that brings field measurements to a control centre and sends commands back; EMS (Energy Management System) is the application layer on top of it that estimates the system state, runs power flow, contingency analysis, economic dispatch and automatic generation control. In India this is done through a three-tier statutory hierarchy - NLDC, five RLDCs and SLDCs, with sub-state control centres below - under the Electricity Act 2003 and the Indian Electricity Grid Code. It is examined in GATE EE (power system operation and control, frequency and voltage control, AGC), in university papers on "Power System Operation and Control", and is directly tested in interviews for POWERGRID, Grid Controller of India (POSOCO), state transcos, NTPC, Tata Power, Adani, Siemens, GE Vernova, Hitachi Energy and Schneider.

Explained

What operation means: objectives and time scales

• Objectives: (1) supply-demand balance, (2) frequency within band, (3) voltage profile within limits, (4) thermal/stability loading limits respected, (5) security against credible contingencies, (6) minimum cost and minimum emission, (7) quick restoration after disturbance.
• Operating states (Dy Liacco / Fink-Carlsen framework): normal secure (all limits met, survives credible contingency) → normal insecure / alert (limits met but a contingency would violate them) → emergency (limits violated, system intact) → in extremis (cascading, islanding, partial blackout) → restorative. Preventive control moves alert→secure; emergency control and load shedding act in emergency; restoration follows a black start.
• Time hierarchy: protection and inertial response (ms-s) → primary/governor frequency response (2-30 s) → secondary control/AGC (30 s-15 min) → tertiary control, reserve replacement, economic dispatch and intra-day market (15 min-hours) → unit commitment and day-ahead scheduling (hours-days) → maintenance and outage planning (weeks-years).
• Control areas: a control area is a part of the grid where all generators swing coherently and which answers for its own balance; the interconnected grid is a set of control areas linked by tie-lines.

SCADA: architecture and components

• Field level: instrument transformers (CT, VT/CVT) → transducers → RTU (Remote Terminal Unit) or modern substation automation system with bay-level IEDs; status from auxiliary contacts of breakers and isolators.
• Data types: analog (MW, MVAr, kV, A, Hz, tap position, temperature), status/digital (breaker open/close, relay flags, alarms), pulse/accumulator (energy, kWh), control/commands (breaker trip-close, tap raise-lower, setpoint raise-lower).
• Communication: fibre optic on OPGW (standard on Indian transmission lines), PLCC (power line carrier) on older links, microwave, leased lines, VSAT for remote sites; protocols IEC 60870-5-101/104 (serial/TCP, the Indian utility default for RTU-to-control-centre), DNP3 (common in distribution and in US-origin equipment), Modbus (device level), IEC 61850 (station bus GOOSE/MMS inside the substation), ICCP/TASE.2 (IEC 60870-6) for control-centre-to-control-centre exchange.
• Control centre level: front-end processors/communication servers, redundant SCADA servers, historical information system (HIS), real-time database, man-machine interface (MMI/HMI) with one-line diagrams, video projection system, DTS (dispatcher training simulator).
• Scan and refresh: cyclic polling or report-by-exception; typical SCADA scan 2-10 s for analogs, status changes reported immediately with time-tagging at the RTU for sequence-of-events.
• Control security: select-before-operate (SBO) - select point, check echo/feedback, then execute; interlocks, tagging, and operator authentication prevent wrong-point operation.
• Supervisory control functions: breaker/isolator operation, transformer tap change, capacitor/reactor switching, generator setpoints, auto-reclose enable, alarm annunciation and acknowledgement.

EMS application software

• Topology processor / network configurator: converts breaker and isolator status into the electrical bus-branch model; detects islands and energised/de-energised sections.
• State estimation (SE): uses redundant measurements (MW, MVAr, V, I, sometimes PMU angles) with weighted least squares to produce the best-fit complete, consistent state (all bus voltages and angles) and to detect/identify bad data. Needs observability; redundancy ratio = number of measurements ÷ number of states, typically 1.5-3.
• Load forecasting: very short term (minutes-hours) and short term (day ahead) using weather, day type, holidays; feeds unit commitment and scheduling.
• Dispatcher power flow / network sensitivity: operator runs "what-if" power flows on the estimated base case.
• Contingency analysis and security assessment: screens an N-1 list (and specified N-1-1) of line, transformer and generator outages; flags thermal overloads and voltage violations. Security-constrained economic dispatch/OPF then redispatches to remove violations at minimum cost; preventive (act before) versus corrective (act after, with a credible post-contingency action).
• AGC/LFC: closed-loop secondary control, described below.
• Scheduling and accounting: unit commitment, hydro-thermal coordination, reserve monitoring, generation and interchange scheduling, energy accounting for settlement.
• Other modules: short-circuit calculation, loss minimisation/VAr dispatch, outage scheduling, DTS, and in distribution utilities a DMS/ADMS (feeder reconfiguration, FLISR, Volt-VAr optimisation) plus OMS for outages.
• Architecture note: SCADA/EMS is the operational-technology (OT) system; a common information model (CIM, IEC 61968/61970) is used for data exchange between EMS, GIS, OMS and market systems.

Frequency control: primary, secondary, tertiary

• Physics: frequency falls when load exceeds generation. The swing equation gives the initial rate of change: df/dt = f₀·ΔP/(2H·S) for a lumped machine of inertia constant H.
• Primary (governor) control: each governor has a droop R% = (Δf/f_rated)/(ΔP/P_rated) × 100 %, typically 3-6 % on thermal sets in India (IEGC requires free governor/primary response capability; droop around 3-6 %). Droop gives proportional action, so frequency settles at a new value - steady-state error remains.
• Composite response: area frequency response characteristic β = 1/R_eq + D (MW/Hz), where D is the natural load damping (load falls as frequency falls).
• Secondary control (AGC / LFC): an integral controller at the control centre drives ACE (Area Control Error) = ΔP_tie + 10B·Δf to zero (B = frequency bias in MW/0.1 Hz, hence the factor 10 when f is in Hz), restoring frequency to nominal and tie-line flows to schedule. This is tie-line bias control; in flat frequency control only Δf is used and in flat tie-line control only ΔP_tie.
• Non-interactive property: with tie-line bias control, a load change in one area is corrected by that area alone - other areas' ACEs return to zero without changing their generation schedules.
• Tertiary control: reserve replacement, manual redispatch, start-up of fast units, and economic dispatch to restore reserves.
• Reserves: spinning reserve (synchronised, loadable), non-spinning/standby, primary response reserve, and in India Primary (PRAS), Secondary (SRAS) and Tertiary (TRAS) Reserve Ancillary Services procured under CERC's Ancillary Services Regulations (2022), which replaced the earlier RRAS mechanism.
• Emergency measures: under-frequency relays (UFR) and df/dt relays shed pre-identified load blocks; System Protection Schemes / defence mechanisms trip or island to arrest cascading; islanding schemes protect critical loads (for example at large industrial plants and some city networks).

Voltage and reactive power control

• Voltage is a local quantity controlled locally; frequency is a global one. Reactive power cannot be transported far without heavy losses, so VAr must be produced near the load.
• Means of control: generator excitation/AVR and the generator's reactive capability (P-Q) curve, on-load tap changers, switched shunt capacitors (to supply VAr at load/distribution ends) and shunt reactors (to absorb line-charging VAr at light load), series capacitors, synchronous condensers, and SVC/STATCOM.
• Voltage stability limits long-distance transfer; nose-point (P-V curve) and Q-V analysis are run in EMS as voltage-security assessment.
• Indian practice: IEGC specifies voltage bands by voltage class (approximately 0.95-1.05 pu around nominal for EHV, with narrower limits at 765 kV); users must not draw excess VAr at low voltage and a VAr-exchange charge applies for drawing/injecting reactive power at voltages outside the band.

WAMS, PMUs and cyber security

• PMU (phasor measurement unit): GPS/IRIG-B time-synchronised measurement of voltage and current phasors plus frequency and df/dt, reported 25-50 times per second (versus a 2-10 s SCADA scan), standard IEEE C37.118 (and IEC/IEEE 60255-118-1), aggregated by PDC into a WAMS.
• Uses: wide-area angle and oscillation monitoring (inter-area modes), event and post-disturbance analysis, model validation, linear state estimation, islanding detection, and the basis for future wide-area protection and control (WAMPAC). India's URTDSM (Unified Real Time Dynamic State Measurement) scheme, approved in 2012, deployed PMUs widely across the transmission grid; the National Open Access Registry sits alongside the SCADA/EMS stack.
• Cyber security: OT networks are segregated from IT with firewalls and DMZ, unidirectional gateways for data export, no direct internet path; standards IEC 62443 (industrial automation security) and IEC 62351 (power-system communication security). In India, CEA's Cyber Security in Power Sector Guidelines (2021) and the CERT-In/NCIIPC framework apply; control centres are protected/critical information infrastructure.
• Known risk examples are taught as cautionary cases (Ukraine 2015-16 grid attacks); the standard answer is defence-in-depth, no remote control paths from corporate networks, strict patching and periodic audit.

Indian institutional and regulatory framework

• Electricity Act 2003: sections 26-34 create the National Load Despatch Centre (NLDC), Regional Load Despatch Centres (RLDCs) and State Load Despatch Centres (SLDCs); the RLDC/NLDC is the apex body for integrated regional/national operation and its directions are binding on all users.
• Hierarchy: NLDC → 5 RLDCs (Northern, Western, Southern, Eastern, North-Eastern) → SLDCs (one per state) → sub-LDCs/distribution control centres. Operation of NLDC/RLDCs is with Grid Controller of India Ltd (formerly POSOCO), a government company carved out of POWERGRID; SLDCs are with state transcos or separate companies.
• One nation, one grid: the Southern grid was synchronised with the rest on 31 December 2013, giving a single synchronous national grid at 50 Hz.
• IEGC (Indian Electricity Grid Code): CERC regulation governing planning, connectivity, scheduling and despatch, system security, and metering. The IEGC 2023 regulations replaced the 2010 code; details such as band limits and timelines change by amendment, so the current notification should be checked.
• Frequency band: operating band 49.90-50.05 Hz with the IEGC target of 50 Hz; deviation settlement charges rise in steps as frequency departs from this band. Earlier regimes used 49.5-50.2 Hz and then 49.9-50.05 Hz - the exact band and the price vector have been revised several times, so quote the current DSM regulation.
• ABT and DSM: the Availability Based Tariff (capacity charge + energy charge + unscheduled interchange/deviation charge) made every drawee pay for deviating from schedule; the Deviation Settlement Mechanism (DSM) regulations now price deviations against frequency and against the volume of deviation, with tighter limits for renewable-rich entities.
• Scheduling: day-ahead schedules revised in time blocks of 15 minutes (96 blocks a day), with intra-day revisions; the Indian grid also operates Security Constrained Economic Despatch (SCED) at the national level to redispatch ISGS thermal units for economy, and market products on IEX/PXIL including real-time market and the green day-ahead market.
• Metering: Special Energy Meters (SEM) at interchange points with 15-minute integration for regional energy accounting; CEA (Installation and Operation of Meters) Regulations govern accuracy classes.

Key points

• SCADA - data acquisition, supervisory control, alarms and archiving; it is the plumbing, not the intelligence.
• EMS - the application suite (topology, state estimation, power flow, contingency analysis, OPF, AGC, scheduling) that uses SCADA data to operate the grid economically and securely.
• RTU/IED - field device that digitises analogs, status and energy pulses and talks IEC 60870-5-101/104 or DNP3 to the control centre.
• State estimation - weighted least-squares fit of redundant measurements giving a complete, consistent, bad-data-free state; needs observability and redundancy of about 1.5-3.
• Topology processor - turns breaker status into the bus-branch model and finds islands; runs before state estimation.
• Contingency analysis (N-1) - checks every credible single outage for thermal and voltage violations; defines the secure operating region.
• Operating states - normal secure, alert, emergency, in extremis, restorative (Dy Liacco diagram).
• Droop control (primary) - proportional, fast, leaves a steady-state frequency error; R ≈ 3-6 % in Indian thermal units.
• AGC (secondary) - integral control on ACE = ΔP_tie + 10B·Δf; zero steady-state error in both frequency and tie-line flow.
• Tie-line bias control - non-interactive: the area that caused the imbalance corrects it.
• AFRC β = 1/R + D (MW/Hz) - stiffness of an area; Δf_ss = −ΔP_L/β for an isolated area.
• Reserves - spinning, non-spinning and ancillary (PRAS/SRAS/TRAS in India); UFR load shedding is the last automatic defence.
• Voltage control is local (excitation, taps, capacitors/reactors, SVC/STATCOM); frequency control is global.
• PMU/WAMS - 25-50 samples/s synchronised phasors (IEEE C37.118) for angle and oscillation monitoring; India's URTDSM.
• Indian hierarchy - NLDC → 5 RLDCs → SLDCs, statutory under the Electricity Act 2003; RLDC directions are binding.
• IEGC + DSM + ABT - the code and commercial mechanisms that enforce schedule discipline; band around 49.90-50.05 Hz, 15-minute blocks.
• Cyber security - IEC 62443/62351, OT-IT segregation, CEA Cyber Security Guidelines 2021.
• Single national grid - all regions synchronous since 31 December 2013 at 50 Hz.

Formulas and examples

Example 1 - droop in MW/Hz and load sharing. A 500 MW unit has 4 % droop on a 50 Hz base.

• Full droop range: Δf = 4 % × 50 = 2 Hz for 500 MW → R = 2/500 = 0.004 Hz/MW, i.e. gain 1/R = 250 MW/Hz.
• Frequency falls from 50.00 to 49.90 Hz: ΔP = Δf/R = 0.10/0.004 = +25 MW (unit picks up 25 MW, 5 % of rating).
• Now add a 200 MW unit with 5 % droop: Δf range = 2.5 Hz → 1/R = 200/2.5 = 80 MW/Hz; for the same 0.1 Hz dip it gives +8 MW. Combined 1/R_eq = 330 MW/Hz → total primary response 33 MW.

Example 2 - steady-state frequency after a load increase. An isolated area: total capacity 6,000 MW all on governors with average 5 % droop on a 50 Hz base; load 5,000 MW with damping k = 1 %/%. A 300 MW load is suddenly added, with no secondary control.

• 1/R_eq = 6,000/(0.05 × 50) = 6,000/2.5 = 2,400 MW/Hz.
• D = 1 % × 5,000 MW per 1 % of 50 Hz = 50/0.5 → 50 MW per 0.5 Hz = 100 MW/Hz.
• β = 2,400 + 100 = 2,500 MW/Hz.
• Δf = −300/2,500 = −0.12 Hz → frequency settles at 49.88 Hz.
• Governors supply 2,400 × 0.12 = 288 MW; the load itself gives up 100 × 0.12 = 12 MW; total 300 MW. ✔
• With AGC, the integral action pushes generation up by the full 300 MW and frequency returns to 50.00 Hz.

Example 3 - df/dt and inertia. System of 40,000 MVA connected capacity, average H = 4 s. A 1,600 MW generation block trips.

• df/dt = −50 × 1,600/(2 × 4 × 40,000) = −80,000/320,000 = −0.25 Hz/s initially.
• Time to fall 0.5 Hz if nothing responded: 0.5/0.25 = 2 s - which is why primary response must act within seconds and why falling inertia from inverter-based generation is a concern.
• If a df/dt relay is set at 0.2 Hz/s with 0.3 s delay, it would operate here; a 0.3 Hz/s setting would not - set points must be chosen against the credible largest infeed loss.

Example 4 - ACE and tie-line bias control. Two areas. Area A: 1/R = 1,200 MW/Hz, D = 100 MW/Hz; Area B: 1/R = 800 MW/Hz, D = 60 MW/Hz. Scheduled tie flow A→B = 200 MW. Load in Area B rises by 150 MW.

• β_total = (1,200 + 100) + (800 + 60) = 1,300 + 860 = 2,160 MW/Hz → Δf = −150/2,160 = −0.0694 Hz (49.931 Hz).
• Area A's response = 1,300 × 0.0694 = 90.3 MW, so tie flow rises to 200 + 90.3 = 290.3 MW; ΔP_tie(A) = +90.3 MW.
• Bias is set to match the area characteristic: B_A = β_A/10 = 130 MW/0.1 Hz, so ACE_A = +90.3 + 10 × 130 × (−0.0694) = 90.3 − 90.3 = 0 — Area A's AGC correctly stays idle; a bias set below β_A/10 would leave a small spurious ACE in A.
• Area B: ΔP_tie(B) = −90.3 MW; with B_B = β_B/10 = 86 MW/0.1 Hz, ACE_B = −90.3 + 10 × 86 × (−0.0694) = −90.3 − 59.7 = −150 MW → exactly the load it must pick up. AGC in B raises generation by 150 MW; tie flow and frequency return to schedule and 50 Hz, Area A's schedule untouched.

Example 5 - state estimation redundancy and observability. A 14-bus network with 20 branches has 60 telemetered measurements (bus injections and branch flows, both MW and MVAr) plus 6 voltage magnitudes.

• States to estimate: 2n − 1 = 2 × 14 − 1 = 27 (14 magnitudes + 13 angles, one bus as angle reference).
• Measurements m = 66 → redundancy ratio = 66/27 ≈ 2.4, comfortably observable and able to identify bad data.
• If 20 measurements are lost by a communication failure, m = 46, ratio ≈ 1.7 - still observable only if the surviving set spans the network; a whole substation losing telemetry can create an unobservable island whatever the ratio, filled by pseudo-measurements from load forecast.

Where it is used

• Load despatch centres: Grid Controller of India (NLDC and the 5 RLDCs) and every SLDC run SCADA/EMS for real-time despatch, scheduling, DSM accounting and SCED; roles are shift charge engineer, real-time operator, system analyst.
• Transmission utilities: POWERGRID and state transcos (MSETCL, GETCO, TANTRANSCO) for substation automation, OPGW/fibre SCADA networks, URTDSM PMU deployments.
• Generating stations: plant DCS interfaced to the RLDC for AGC setpoints, free-governor operation, ABT metering - NTPC, NHPC, Adani Power, Tata Power and IPPs.
• Discoms: SCADA/DMS control centres under earlier R-APDRP/IPDS and the RDSS for urban feeders, outage management, AMI integration (Tata Power Delhi Distribution, BESCOM, Adani Electricity Mumbai, CESC).
• Vendors and system integrators: Hitachi Energy (Network Manager), Siemens (Spectrum Power), GE Vernova (e-terra), Schneider (EcoStruxure ADMS / Geo SCADA), ABB/Larsen & Toubro, Kalkitech and Easun for RTUs/gateways - jobs in SCADA engineering, FAT/SAT, protocol testing, database and display building.
• Industrial and railway systems: captive plants with islanding schemes, Indian Railways traction SCADA (TPC remote control of traction substations), refinery and steel-plant power management systems.

Interview questions

• Q: What is the difference between SCADA and EMS? A: SCADA acquires data and issues supervisory controls; EMS is the application layer that uses that data for state estimation, contingency analysis, optimal power flow, AGC and scheduling. SCADA can exist without EMS, but EMS cannot work without SCADA data.
• Q: Why is state estimation needed when SCADA already gives measurements? A: SCADA data is incomplete, non-simultaneous and contains errors; state estimation uses redundancy to produce one complete, consistent, bad-data-filtered network state on which power flow and contingency analysis can be trusted.
• Q: Why does primary (droop) control leave a frequency error while AGC does not? A: Droop is proportional, so a non-zero frequency deviation is needed to hold the extra output; AGC adds integral action on ACE, which keeps correcting until the error is exactly zero.
• Q: What is ACE and why the frequency bias term? A: ACE = ΔP_tie + 10B·Δf. The bias term makes the control non-interactive: only the area whose load changed sees a non-zero ACE, so the deviating area corrects its own imbalance without disturbing others.
• Q: If frequency is 49.9 Hz and your area's tie-line export is above schedule, what is happening? A: Another area is short of generation and your area's governors are supporting it through the tie-line. With the bias set to match your area characteristic, your own ACE stays close to zero, so your AGC does not act; the extra export is withdrawn automatically as the deficient area's AGC picks up its own load. A sustained condition means that area's secondary control or reserve is deficient.
• Q: Why is voltage controlled locally and frequency globally? A: Frequency is common to the whole synchronous system, so any generator anywhere can correct it; reactive power cannot be transmitted over distance without large losses and voltage drop, so VAr support must be near the load.
• Q: What does N-1 security mean? A: The system must survive the outage of any single credible element without violating thermal, voltage or stability limits, and without load shedding - checked continuously by contingency analysis.
• Q: How does a PMU differ from a conventional SCADA measurement? A: A PMU gives GPS-time-synchronised voltage and current phasors (magnitude and angle) at 25-50 samples per second, so angles across the grid can be compared directly; SCADA gives unsynchronised magnitudes every few seconds.
• Q: What is select-before-operate and why does it matter? A: A two-step control: the operator selects a point, the device echoes the selection back, then execute is sent. It prevents a corrupted or misdirected command from tripping the wrong breaker.
• Q: What protocol is used between an Indian substation RTU and the SLDC, and what inside the substation? A: IEC 60870-5-104 (TCP/IP) - earlier -101 serial - between RTU/gateway and control centre; IEC 61850 MMS and GOOSE on the station bus between IEDs.
• Q: What is ABT and how did it change grid discipline in India? A: Availability Based Tariff splits central-station tariff into capacity charge, energy charge and a frequency-linked charge for unscheduled interchange, so deviating from schedule costs money; it sharply tightened frequency around 50 Hz and has since evolved into the Deviation Settlement Mechanism.
• Q: Why is falling system inertia a concern for operation? A: Inverter-based solar and wind contribute little rotating inertia, so df/dt after a large infeed loss becomes steeper, giving primary response and UFR schemes less time; remedies include fast frequency response, synthetic inertia, synchronous condensers and tighter reserve requirements.

Common mistakes

• Treating SCADA and EMS as the same thing, or claiming SCADA "controls" frequency - AGC in the EMS does that.
• Dropping the factor 10 in ACE = ΔP_tie + 10B·Δf, or mixing up the sign convention (export above schedule is positive ΔP_tie).
• Using percentage droop directly as Hz/MW: R% must be converted using f_rated and P_rated before use.
• Quoting the frequency bias B in MW/Hz when the ACE formula expects MW/0.1 Hz - a factor-of-ten error that makes AGC look wrong.
• Forgetting load damping D when computing steady-state Δf - it is a real part of β and makes the answer smaller.
• Assuming high measurement redundancy guarantees observability; a substation losing all telemetry creates an unobservable island regardless of the global ratio.
• Confusing state estimation with load flow: load flow needs specified injections and gives an exact solution; state estimation fits over-determined noisy measurements.
• Quoting the old 49.5-50.2 Hz band or old RRAS/UI numbers as current - the IEGC 2023 and DSM regulations supersede them and figures keep changing by amendment.
• In interviews, saying N-1 means only line outages - generator, transformer, bus and ICT outages are all in the contingency list.

Related topics

• Load flow and fault analysis · Economic dispatch and unit commitment · Power system stability · Power system protection: relays and circuit breakers · Smart grid and renewable integration · Electricity markets and regulation in India